text
stringlengths 1
1.05M
| meta
dict |
|---|---|
Introduction {#Sec1}
============
Bone is one of the most common sites of metastasis for various primary tumors including prostate, breast, lung, and kidney cancers^[@CR1],[@CR2]^. Although bone metastasis is associated with increased morbidity and mortality, promising therapy to prevent bone metastasis is currently unavailable. This deficiency emphasizes the need for new therapeutic approaches targeting molecular mechanisms that regulate bone metastasis and for new models to study this disease phenomenon.
Murine models of bone metastasis using intracardiac (IC) and intratibial injections have been instrumental in revealing molecular mechanisms underlying metastatic processes and translational studies for drug development^[@CR3],[@CR4]^. During the past two decades, IC injection has been the gold standard to develop bone metastasis in mice^[@CR5]--[@CR9]^ by injecting cancer cells into the left ventricle to disseminate them to the whole body including bone marrow tissue via the arterial bloodstream, which eventually develop into metastatic colonies in the bone and other organs^[@CR10]^. Unlike intratibial injection that severely damages the tibia, IC injection recapitulates the bone metastasis process, including survival of cancer cells in the bloodstream, extravasation, micro-colony formation, and metastatic progression in the intact bone marrow, and thus provides more relevant information for drug development. IC injection, however, is insufficient for rapid studies in this field, mainly owing to its requirement for high technical proficiency to exactly insert a syringe needle into the left ventricle of a mouse, causing severe cardiac stresses^[@CR3],[@CR4]^. This limits the number of cancer cells that can be injected at one time, leading to limited delivery of cancer cells to the bone. Thereby analysis with IC model may bias toward cancer cell lines with relatively high metastatic ability. Furthermore, cancer cells are preferably delivered to organs other than bone, such as the lungs and liver, and often develop into lethal cancers in other organs, hampering or even terminating studies of bone metastasis with cell lines with relatively slow metastasis development. New models overcoming such limitations would accelerate basic studies and drug development for bone metastasis.
Here, we present the establishment of a new murine model that predominantly develops bone metastasis in the hind limbs at high frequency. In this model, cancer cells are injected via the caudal artery (CA) in the tail, and the technique is as easy as tail vein injection. CA injection rarely causes acute death and facilitates the injection of a large number of cancer cells, thereby greatly increasing the frequency of bone metastasis for various types of cancer cells. Therefore, CA injection provides an easy-to-use murine model to develop overt bone metastasis in a short time and could greatly facilitate studies to understand bone metastasis and to prevent them.
Results {#Sec2}
=======
CA as a new route for injection {#Sec3}
-------------------------------
To develop a novel murine bone metastasis model, we searched for an alternative arterial route to deliver cancer cells to bone marrow in mice. The CA was the most easily accessible route to inject cancer cells without any surgical procedures (Fig. [1a](#Fig1){ref-type="fig"}). Although cell distribution after IC injection has been well studied, no study has assessed CA-injection route. Therefore, to examine whether this route could be practically used for injection, we injected fluorescent nanoparticles emitting near-infrared II (NIR-II) fluorescence (maximum emission at 1530 nm)^[@CR11],[@CR12]^. Because the nanoparticles injected via CA were thought to eventually travel to the tail vein, we compared their distributions after CA and intravenous (IV) injection by video-rate fluorescence imaging. Surprisingly, CA-injection exhibited totally different routes from IV injection: Injecting nanoparticles into the CA quickly illuminated the capillary bed in the lower body of mice, whereas nanoparticles injected via the tail vein resulted in slow and modest illumination (Fig. [1b](#Fig1){ref-type="fig"} and Supplementary Movies [1](#MOESM3){ref-type="media"} and [2](#MOESM4){ref-type="media"}). This result implied that the CA can be a practical injection rout and may be suitable for delivery of cancer cells to the bone of hind limbs. To track the fate of cancer cells after CA injection, we used murine lung carcinoma LLC cells constitutively expressing firefly luciferase (LLC/luc). In vivo bioluminescence (BL) imaging revealed predominant delivery of LLC/luc cells to the lower body by CA injection that is technically as easy as tail vein injection (Fig. [1c](#Fig1){ref-type="fig"} and Supplementary Movie [3](#MOESM5){ref-type="media"}). CA injection delivered cancer cells three-fold more efficiently to hind-limb bone marrow than IC injection, as revealed by luciferase activity in the bone marrow 30 min after injection of LLC/luc cells (Fig. [1d](#Fig1){ref-type="fig"}). This result was well correlated to the one of ex vivo bone imaging acquired just after dissection (Supplementary Fig. [1](#MOESM1){ref-type="media"}). In addition, ex vivo BL imaging of representative organs confirmed dominant delivery of cancer cells to organs of the lower body after CA injection; in contrast, IC injection resulted in dissemination of cancer cells to various tissues (Fig. [1e](#Fig1){ref-type="fig"} and Supplementary Fig. [2](#MOESM1){ref-type="media"}). These results indicated that CA injection provide a preferable model to study bone metastasis.Fig. 1CA injection efficiently delivered cancer cells to bone marrow of a hind limb. **a** Location of caudal artery in a mouse (left) and a schematic of cross section of mouse tail (right). A yellow arrow indicates the caudal artery along the tail. **b** Comparison of fluorescence images after intra-caudal artery (CA) or tail vein (IV) injection of near-infrared II nanoparticles. **c** Representative BL images at 30 min after injecting LLC/luc cells through caudal artery (CA) or left ventricle (IC). **d** BL intensity of LLC/luc cells harvested from bone marrow of hind limbs at 30 min after CA or IC injection. *n* = 8, \**P* \< 0.05 (two-side student's *t*-test). Error bars indicate s.e.m. **e** Biodistribution of LLC/luc cells after CA or IC injection. Major organs were removed at 30 min after injecting LLC/luc cells and ex vivo BL imaging was performed. BL intensity of each organ was quantitatively analyzed and its relative BL intensity to the one in hind limb is shown. B brain, H heart, Lv liver, S spleen, K kidney, P pancreas, T testis, Lg lung, SI stomach and intestine, V vesicular gland. *n* = 3, \**P* \< 0.05 (two-side student's *t*-test)
Cancer cell distribution after CA injection {#Sec4}
-------------------------------------------
We next compared the efficacy of bone metastasis development in mice, between CA and IC injections, by comparing luciferase activity over time after transplanting LLC/luc cells. BL intensities in the hind limbs were significantly higher after CA injection than after IC injection (Fig. [2a](#Fig2){ref-type="fig"}). Over time, growth rates were similar between these models throughout the development of bone metastasis (Fig. [2a](#Fig2){ref-type="fig"} and Supplementary Fig. [3](#MOESM1){ref-type="media"}), indicating that differences in stress between the methods, during injection and dissemination, did not affect cell proliferation in the bone marrow. Histological analysis confirmed an increase in the number of bone metastatic lesions in CA-injected mice 7 days after the injection of LLC/luc cells (Fig. [2b](#Fig2){ref-type="fig"} and Supplementary Fig. [4](#MOESM1){ref-type="media"}). In addition, significantly larger tumors were observed in the hind-limb bones of CA-injected mice at 14 days after LLC/mKO2-Rluc cell injection (Fig. [2c](#Fig2){ref-type="fig"} and Supplementary Fig. [5](#MOESM1){ref-type="media"}). X-ray micro-computed tomography (CT) imaging revealed a decrease in bone mass in CA-injected mice compared to that in IC-injected animals (Fig. [2d](#Fig2){ref-type="fig"} and Supplementary Fig. [6](#MOESM1){ref-type="media"}), confirming enhanced bone metastasis in CA-injected mice. CA injection enabled observation for more than 32 days after cancer cell injection. In contrast, IC-injected mice became weak and could not be observed after day 25 due to death (Fig. [2e](#Fig2){ref-type="fig"}). Ex vivo BL imaging of representative organs at 32 days after CA injection of LLC/luc cells revealed the development of metastasis predominantly in the hind-limb bones and some nonlethal micrometastases in vesicular glands (Fig. [2f](#Fig2){ref-type="fig"}). Most importantly, we obtained essentially similar BL imaging results in all CA-injected mice without failure.Fig. 2CA injection accelerated development of bone metastasis in a hind limb. **a** Representative BL images (left) and BL intensity in region of interest (ROI) (right) at indicated days after LLC/luc cells injection by CA (*n* = 16) or IC injection (*n* = 12). Red circles in the images of 0.5 h indicate ROI. \**P* \< 0.05 (two-side student's *t*-test). Error bars indicate s.e.m. **b** Metastatic lesions in hind limb at 7 days after CA or IC injection of LLC/luc cells. The average number of a metastatic lesion per bone section was shown above the bar graph. *n* = 23 (CA), *n* = 28 (IC), \**P* \< 0.05 (two-side student's *t*-test). **c** Representative fluorescence images (left) and quantitative analysis of bone metastasis in femur bones at 14 days after injecting LLC/mKO2-Rluc8.6 cells via CA or IC. Metastasis and bone were visualized with mKO2 fluorescence (red) and tissue autofluorescence (cyan), respectively. A scale bar is 500 μm. *n* = 8 (CA), *n* = 6 (IC). \**P* \< 0.05 (two-side student's *t*-test). **d** X-ray micro CT imaging (left) and quantitative analysis (right) of bone volume in the femur at 14 days after CA or IC injection of LLC/mKO2-Rluc8.6 cells. Representative images are transverse plane of femurs shown in **c**. *n* = 8 (CA), *n* = 6 (IC). \**P* \< 0.05 (two-side student's *t*-test). **e** Representative BL images of mice at 25 and 32 days after CA or IC injection of LLC/luc cells. The representative image of mice at 32 days after IC injection are not shown because all IC-injected mice died before the day. **f** Representative ex vivo BL images of organs harvested from a mouse at 32 days after CA injection of LLC/luc cells. A scale bar is 10 mm. V vesicular gland, B brain, Lg lung, T testis, K kidney, S spleen, Lv liver, SI stomach and intestine
Improved bone metastasis development by CA injection {#Sec5}
----------------------------------------------------
Furthermore, we were able to inject a larger number of cancer cells (1 × 10^6^ cells) via CA injection without any acute death, overcoming one of the limitations of IC injection model (Supplementary Table [1](#MOESM1){ref-type="media"}). Cancer cells delivered to bone marrow in the femur increased as the number of cancer cells injected through the CA was increased (Fig. [3a](#Fig3){ref-type="fig"}). The delivery efficiency directly reflected the efficiency of bone metastasis development (Fig. [3b](#Fig3){ref-type="fig"}).Fig. 3Injection of larger number of cancer cells enhances bone metastasis development. **a** Fluorescence microscopic analysis of cancer cells in bone marrow of the femur. Representative fluorescence images of the femur (left) and fluorescent intensities of cancer cells (right) at 30 min after CA injection of green-fluorescently labeled LLC/luc cells (2.0 × 10^5^ or 1.0 × 10^6^ cells). Cells were stained with Hoechst dyes (blue). White-dashed lines indicate growth plates. A scar bar is 200 μm. *n* = 4, \**P* \< 0.05 (two-side student's *t*-test). **b** Development of bone metastasis after injection of different number of LLC/luc cells. Representative BL images (left) and BL intensity in hind limbs (right) at 14 days after CA injection of LLC/luc cells (2.0 × 10^5^ or 1.0 × 10^6^ cells). *n* = 8, \**P* \< 0.05 (two-side student's *t*-test). Error bars indicate s.e.m.
These results motivated us to examine other cancer cell lines, because cancer cell lines applicable to the IC bone metastasis model has been limited. Notably, CA-injected MCF7, which has been recognized as a non-metastatic human breast cancer cell line^[@CR13],[@CR14]^, developed bone metastasis (Fig. [4a](#Fig4){ref-type="fig"} and Supplementary Fig. [7a](#MOESM1){ref-type="media"}). We further applied CA injection of several human cancer cell lines including breast (MDA-MB-231), prostate (PC-3), and kidney (786-O) cancers as well as osteosarcoma (143B). These cancers often metastasize to the bone in patients. In addition, syngeneic mouse models were examined using three cell lines including breast carcinoma E0771 that is first described here in a bone metastasis model. In vivo BL imaging confirmed that these cell lines developed bone metastasis after CA injection (Fig. [4b](#Fig4){ref-type="fig"} and Supplementary Fig. [7b](#MOESM1){ref-type="media"}). It is noteworthy that we could detect bone metastasis by 5--12 days after CA injection of all the cell line examined. Overall, the results demonstrated that CA injection provides a reliable method to develop bone metastasis by increasing the delivery efficiency of a wide variety of cancer cell lines to the bone marrow of the hind limbs in mice.Fig. 4Efficient bone metastasis development with various cancer cell lines by CA injection. **a** Representative BL images at indicated days after CA injection of MCF7 constitutively expressing firefly luciferase. **b** Representative BL images at indicated days after CA injection of cancer cell lines (MDA-MB-231, PC-3, 786-O, 143B, E0771, and 4T1) constitutively express firefly luciferase
Discussion {#Sec6}
==========
Here, we present a new method to establish a murine model of bone metastasis by caudal artery injection of cancer cells. The reliability (almost no failures) and easiness of this method minimize animal use and distress. The reliability is the result of the following features. (1) Because the caudal artery is visible on the body surface, injection into the caudal artery is as easy as tail vein injection. (2) Failure of CA injection is immediately detectable by low resistance upon pushing the plunger or leakage at the needle tip, and reinjection is possible at different points of the caudal artery closer to the body. (3) Successful injection can be confirmed by BL in the lower body of the mouse immediately after CA injection (Fig. [1c](#Fig1){ref-type="fig"}).
Although we first thought that the cells injected into the CA would eventually go to the tail vein, they were delivered to the bone marrow of hind limbs via a route very different from IV injection (Fig. [1b](#Fig1){ref-type="fig"}). In the CA injection model, injected cancer cells were forced to move upstream against the flow of the CA by strongly pushing them in a short time (0.1 mL per 3 s), allowing them to enter the femoral artery through the common iliac artery (see Supplementary Fig. [8](#MOESM1){ref-type="media"})^[@CR15]^. Then they rarely move further upstream to reach the interior mesenteric artery because a high BL signal was not detected in the liver, stomach, or spleen after CA injection (Fig. [1e](#Fig1){ref-type="fig"} and Supplementary Fig. [2](#MOESM1){ref-type="media"}). During these processes, cancer cells might be exposed to severe shear stresses. The in vivo growth rates, however, were similar between CA-injected and IC-injected cells (Fig. [2a](#Fig2){ref-type="fig"} and Supplementary Fig. [3](#MOESM1){ref-type="media"}). These data suggest that the cellular stress by CA-injection was, if any, similar to the one by IC injection and that the injection methods did not influence the growth of cancer cell after their delivery to the bone marrow. In addition, CA injection barely caused acute death of mice even at injection of a large number of cancer cells, allowing more cancer cells to be delivered to the bone marrow (Supplementary Table [1](#MOESM1){ref-type="media"} and Fig. [3a](#Fig3){ref-type="fig"}). Increased delivery efficiency by CA injection increases the chance of successful colonization of cancer cells. For example, MCF7, which has been recognized as a non-metastatic cell line^[@CR13],[@CR14]^, successfully developed bone metastasis as soon as 8 days after CA-injection, suggesting that the non-metastatic phenotype of these cells might be due to low homing efficiency. Further analysis will reveals the effect of hormone pre-treatment, which makes MCF7 metastatic^[@CR16]^, on bone metastasis. This result suggests that our model might open up new avenues for exploring key events in bone metastasis by enabling analyses that were previously impossible using current models.
The fact that CA-injection delivers cancer cells to the organs downstream of the common iliac artery (Supplementary Fig. [8](#MOESM1){ref-type="media"}) limits the investigation for systemic effect of injected cancer cells or the development of metastasis in the bones located upper sites of the body such as skull. In contrast, this model enables the investigation of bone metastasis for a much longer time than that with the IC-injection model because of reduced incidence of lethal metastasis in other organs (Fig. [2f](#Fig2){ref-type="fig"}). This represents a great advantage for investigating cancer cell dormancy, which are key issues for bone metastasis.
Recently, intra-illiac artery (IIA) injection was described to selectively deliver cancer cells to bone marrow of a hind limb in mice, allowing efficient immunohistological analysis of metastatic colonies at the early stage of bone metastasis^[@CR17]^. Although delivery efficiency to the bone might be higher by IIA injection than CA injection, the requirement of a surgical procedure and microscopic observation of IIA injection is not suitable for experiments with large numbers of mice^[@CR18]^. In addition, inflammation at the site of surgery may influence the analysis of bone metastasis.
CA injection might facilitate a particular scope of studies to understand bone metastatic processes. For example, interactions between cancer cells and platelets might be a key event in bone metastasis and remain to be fully understood^[@CR19],[@CR20]^. A previous study raised the issue that IC injection often causes death of transgenic mice with platelet dysfunction, probably due to lasting bleeding from the left ventricle^[@CR21]^. The CA injection model may solve this problem by direct hemostasis at the injection site, enabling investigation of bone metastasis in mice with platelet dysfunction or depletion.
In the CA injection model using LLC/luc cells, metastatic formation was often detected in vesicular glands, probably because of the efficient delivery of cancer cells (Figs. [1e](#Fig1){ref-type="fig"}, [2f](#Fig2){ref-type="fig"}), but this metastasis was not lethal. Although further studies will need to reveal organotropism of various cell lines injected via the CA, our new model could represent a powerful tool for the development of drugs that target bone metastasis, in addition to facilitating research on the molecular mechanisms regulating the initiation and progression of bone metastasis.
Methods {#Sec7}
=======
Cell lines {#Sec8}
----------
The murine lung carcinoma cell LLC, human breast cancer cell MDA-MB-231 and MCF7, human prostate cancer cell PC-3, human renal cell adenocarcinoma 786-O, human osteosarcoma 143B and murine breast cancer cell 4T1 were obtained from ATCC (Rockville, MD, USA). Murine breast cancer cell E0771 was purchased from C3H BioSystems (Buffalo, NY, USA). Isolation of LLC/luc and MDA-MB-231/luc were described previously^[@CR9],[@CR22]^. Similarly, PC-3/luc, 786-O/luc, 143B/luc, and 4T1/luc were isolated after transfection with plasmid pEF/luc by calcium phosphate method^[@CR23]^. E0771 and MCF7 were stably transduced with luc2 fused with monomer KusabiraOrange2 (mKO2) using *Sleeping Beauty* transposon system^[@CR24]^. To utilize the system, first pT2/CMV-MSC-SVNeo. was constructed by amplifying a fragment of CMV promoter-multi cloning site (MSC)-poly A using pcDNA3.1 plasmids (Invitrogen, Carlsbad, CA, USA) as a template and inserting the fragemt into Addgene plasmid \#26553. Then pcDNA/mKO2-luc2 was constructed by inserting a luc2-mKO2 fused cDNA fragment amplified from pGL4.32 (Promega, Madison, WI, USA) and Addgene plasmid \#67661 as templates, respectively and inserting the fragment into pT2/CMV-SMC-SVNeo using In-Fusion HD Cloning Kit (Clontech, Palo Alt, CA, USA). E0771 and MFC7 were co-transfected with pT2/CMV-mKO2-luc2-SVNeo and pCMV(CAT)T7-SB100 (Addgene plasmid \#34879) using NEPA21 electroporator (NEPA gene, Chiba, Japan). The fluorescence/bioluminescence dual reporter-introduced cells E0771/mKO2-luc2 and MCF7/mKO2-lu2 were established from a single colony after antibiotic selection. LLC/mKO2-Rluc8.6 was established using the *Sleeping Beauty* transposon system. To construct pT2/mKO2-Rluc8.6, Rluc8.6 cDNA was amplified from pGEX/PTD-ODD-Rluc8.6 as described previously^[@CR25]^. LLC/luc, LLC/mKO2-Rluc8.6, MDA-MB-231/luc, 786-O/luc, and 143B/luc were maintained at 37 °C in 5% FCS-DMEM (Nacalai Tesque, Kyoto, Japan) supplemented with penicillin (100 U/mL) and streptomycin (100 μg/mL). PC-3/luc, E0771/mKO2-luc2, and 4T1/luc were maintained at 37 °C in 10% FCS-RPMI (Nacalai Tesque, Kyoto, Japan) supplemented with penicillin (100 U/mL) and streptomycin (100 μg/mL). The cells were regularly checked for mycoplasma contamination by a mycoplasma check kit (Lonza, Basel, Switzerland) and were independently stored and recovered from the original stock every time for each experiment.
Mice {#Sec9}
----
C57B/6 mice (male), C57B/6 albino mice (male), NOD-SCID mice (female), SCID mice (male and female), BALB/c mice (female), and BALB/c-nu nude mice (male) were obtained from Charles River Laboratory Japan (Yokohama, Japan). All mice used were provided access to food and water ad libitum, and were housed in the animal facilities at Tokyo Institute of Technology. All the experimental procedures using mice were approved by the Animal Experiment Committees of Tokyo Institute of Technology (authorization number 2010006--3 and 2014005) and carried out in accordance with relevant national and international guidelines.
NIR-II fluorescence imaging {#Sec10}
---------------------------
NIR-II fluorescence images were acquired with SAI-1000 (SHIMADZU, Kyoto, Japan). Mice were injected with 50 μL OTN ceramic probe Y (Katayama Chemical Industries, Osaka, Japan) via cannulate line connected to caudal artery or tail vein, respectively. NIR-II fluorescence images were obtained using following settings: excitation/emission wavelength = 980 nm/1530 nm, laser power = 10.5 mW, and camera exposure time = 500 ms.
In vitro analysis of cancer cells from the bone marrow {#Sec11}
------------------------------------------------------
Hind-limb bones were harvested and crushed in a mortar with PBS. The bone marrow extract was centrifuged to obtain the cell pellet and the pellet was suspended into 100 μL of Passive Lysis Buffer (Promega). Fifty μL of cell lysate was mixed with an equal volume of Luciferase Assay Reagent (Promega) in a 96-well plate and then BL imaging was performed using IVIS Spectrum (PerkinElmer, Boston, MA, USA). The BL images were analyzed by Living Image 4.3 software (PerkinElmer) specialized for IVIS.
Bone metastasis models {#Sec12}
----------------------
For CA injection, LLC/luc (2 × 10^5^ or 1 × 10^6^ cells to male C57B/6), MDA-MB-231/luc (5 × 10^5^ cells to female NOD-SCID), PC-3/luc (1 × 10^6^ cells to male SCID), 786-O/luc (1 × 10^6^ cells to male BALB/c-nu), 143B/luc (5 × 10^5^ cells to male BALB/c-nu), E0771/mKO-luc2 (2 × 10^5^ cells to female C57B/6 albino), 4T1/luc (1 × 10^3^ cells to female BALB/c), or MCF7 (1.5 × 10^6^ cells to female SCID) suspended in 100 μL PBS was injected into the caudal artery of anesthetized mice using 29 G syringe needle in a short time (\<3 s). IC injection was performed as described previously^[@CR4],[@CR5]^. LLC/luc (2 × 10^5^ cells) suspended in 100 μL PBS and was injected into left cardiac ventricle of 5-week-old-male C57B/6 mice. IC injection is well established and CA injection is technically as easy as IV injection. Therefore, \>8 samples are adequate sample size for evaluation of metastasis growth in each experiment. Randomization and blind tests were not performed during the experiments.
In vivo bioluminescence imaging {#Sec13}
-------------------------------
BL images of tumor bearing mice were acquired with IVIS Spectrum at 15 min after intraperitoneal injection of [d]{.smallcaps}-luciferin (50 mg/kg). The following conditions were used for image acquisition: open emission filter, exposure time = 60 s, binning = medium: 8, field of view = 12.9 × 12.9 cm, and f/stop = 1. The BL images were analyzed by Living Image 4.3 software (PerkinElmer) specialized for IVIS.
Ex vivo bioluminescence imaging {#Sec14}
-------------------------------
A mouse was scarified immediately after in vivo BL imaging and major organs were removed. BL images of the organs were obtained with following conditions: open emission filter, exposure time = 30 s, binning = medium: 8, field of view = 12.9 × 12.9 cm, and f/stop = 1. The BL images were analyzed by Living Image 4.3 software (PerkinElmer) specialized for IVIS.
Histological analysis {#Sec15}
---------------------
The isolated bone of a hind limb was fixed in 70% ethanol for 48 h, decalcified in 10% EDTA for 2 weeks, processed, and embedded in paraffin. Sectioned bones (10 μm thickness) were then stained with hematoxylin--eosin. To detect bone metastasis of mKO2-expressing cancer cells, isolated bones were embedded in OCT compound (Leica Microsystems, Wetzlar, Germany), and cryosections of the embedded bones were prepared by using CM3050 Cryostat (Leica BIOSYSTEMS), stained with Hoechst and observed under a confocal fluorescence microscope (LSM700, Carl Zeiss, Oberkochen, Germany). To quantitatively evaluate the formation of bone metastatic lesions, multiple sections were prepared from eight bones dissected from CA and IC-injected mice and metastatic lesions in each section were counted. The size of bone metastasis was quantifies by measuring area with mKO2 fluorescence with ImageJ software.
X-ray micro CT imaging {#Sec16}
----------------------
X-ray micro CT imaging was performed using CosmoScan GX II system (Rigalu Corp., Tokyo, Japan). Dissected hind-limb bones were imaged using following parameters: 90 KV of X-ray tube voltage, 88 μA of X-ray tube current. For quantitative analysis of bone mass, X-ray CT images of transverse plane at 0.7 mm from the growth plate were analyzed using ImageJ softeware.
Imaging of cancer cells in the bone marrow {#Sec17}
------------------------------------------
Femur bones were harvested from mice at 30 min after CA injection of LLC/luc cells labeled with CellTracker Green CMFDA Dye (Thermo Fisher Scientific, San Jose, CA, USA). Also, mice were injected with Hoechst 33342 dyes (50 mg/kg) (Wako, Tokyo, Japan) at 5 min before isolation of bones. The isolated bone was then embedded into OCT compound to freeze it at --80 °C. The bone marrow of a femur bone was exposed using CM3050 Cryostat (Leica BIOSYSTEMS, Wetzlar, Germany) and mounted to confocal microscope (Carl Zeiss, Oberkochen, Germany) for fluorescence imaging of cancer cells in the bone marrow. Total intensity of green fluorescence in 0.5 mm from the growth plate was quantified using ImageJ 1.47.
Statistical analysis {#Sec18}
--------------------
Data sets with similar variance were statistically analyzed. Data are presented as means ± standard error of the mean (s.e.m.) and were statistically analyzed with a two-side student's *t*-test. *P* values of \<0.05 were considered statistically significant.
Data availability {#Sec19}
-----------------
The data that support the findings of this study are available from the corresponding author upon request.
Electronic supplementary material
=================================
{#Sec21}
Supplementary Information Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3
**Electronic supplementary material**
**Supplementary Information** accompanies this paper at 10.1038/s41467-018-05366-3.
**Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We are grateful to Shigeaki Watanabe and Machiko Horiuchi (Summit Pharmaceuticals International Corporation, Tokyo, Japan) for discussion and technical support of IVIS, SAI-1000, and CosmoScan GX II system. We thank Doreen Chan (California Institute of Technology) for assistance in animal imaging. We also thank Biomaterials Analysis Division, Technical Department of Tokyo Institute of Technology for DNA sequencing analysis. This research was supported by a Grant-in-Aid for Scientific Research on Innovative Areas "Integrative Research on Cancer Microenvironment Network" from the "Ministry of Education, Culture, Sports, Science and Technology of Japan" (S.K.-K.), Grant-in-Aid for Young Scientist (B) (T.Ku.).
T.Ku. and K.N. conceived CA injection. S.K.-K. designed and managed the overall project. T.Ku., N.K., K.N., T.I., H.M., M.M., and T.Ka. performed experiments. T.Ku. and N.K. analyzed data. T.Ku. and S.K.-K. wrote the manuscript.
Competing interests {#FPar1}
===================
The authors declare no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .14}
{#sp2 .15}
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Bladder cancer, the fourth most common tumor in men and the eighth in women, remains a huge concern for the medical community because of its incidence and prevalence rates, as well as high percentage of recurrence and progression \[[@B1]--[@B4]\]. Mortality rates in muscle-invasive disease are still very high, despite the growing efforts on earlier diagnosis and aggressive and multidisciplinary treatments \[[@B4], [@B5]\].
In this context, preventive strategies are crucial for the management of bladder cancer, but they still demand a better elucidation of the carcinogenetic process. Exogenous factors, such as cigarette smoking, which accounts for a huge percentage of cases, as well as occupational carcinogens, such as aromatic amines and polycyclic aromatic hydrocarbons, are important determinants of the disease appearance \[[@B6], [@B7]\]. However, apart from the genetic features already characterized \[[@B8], [@B9]\], the cellular and molecular mechanisms might involve inflammatory, proliferative, and oxidative stress phenomena that deserve further elucidation. In fact, the identification of promising drugs depends on continuous research concerning the molecular/cellular mechanisms underlying cancer appearance and progression.
The experimental model of rat bladder cancer induced by *N*-butyl-*N*-(4-hydroxybutyl) nitrosamine (BBN) is an appropriate and validated model to study human cancer development. In fact, due to the histological similarities with the human bladder cancer, it has been the most used model for the study of tumor pathophysiology, as well as for the evaluation the efficacy of therapeutic strategies \[[@B10]--[@B12]\]. The urothelial carcinogenesis is a continuous and slow process that goes through molecular and morphological changes, from benign to aggressive lesions, including initial dysplastic and proliferative epithelial abnormalities, preneoplastic changes, and malignant lesions (papilloma and carcinoma) \[[@B12]--[@B14]\]. Thus, an early treatment targeting these pathways could hypothetically prevent bladder cancer development and growth. Previous reports have demonstrated beneficial effects of preventive strategies, including our own studies using anti-inflammatory and anti-proliferative agents \[[@B15]--[@B19]\].
Polyunsaturated fatty acids (PUFAs) have many physiological roles in the body, including acting as sources of cellular energy, regulators of protein synthesis, building blocks of phospholipids and glycolipids required for cell membrane structure, and components of membranes that regulate the fluidity, permeability, and dynamics of cell membranes, as well as precursors for many hormones, inter- and intracellular messengers as well as their receptors \[[@B20], [@B21]\]. PUFAs nomenclature is based on the position of the first carbon-carbon double bond within the long hydrocarbon chain; two of the four families cannot be synthesized from the body\'s carbohydrate stores and are required from the diet and are, thus, named as essential fatty acids (EFAs) and include the omega-3 (*ω*-3) and omega-6 (*ω*-6) oils. Considering their various biological activities, such as antioxidant and anti-inflammatory properties, *ω*-3 fatty acids have been viewed as useful weapons against several conditions, including those of cardiovascular/atherogenic, neuronal/degenerative, and inflammatory and neoplastic nature \[[@B22]--[@B26]\].
There is a wide range of chronic inflammatory, oxidative, and proliferative conditions that make cells susceptible to neoplastic transformation. The omega-3 fatty acids, due to their antioxidant and anti-inflammatory properties have been tested as chemopreventive and/or therapeutic agents, *per se* and in combination with drugs and/or radiotherapy, in several cancer types \[[@B27]--[@B30]\], but the putative efficacy on bladder cancer remains to be elucidated. The purpose of this study was, thus, to evaluate the chemopreventive efficacy of *ω*-3 fatty acids, in particular eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), in a rat model of bladder cancer induced by a nitrosamine.
2. Material and Methods {#sec2}
=======================
2.1. Animals and Groups {#sec2.1}
-----------------------
Forty-four male Wistar rats obtained from Charles River Lab. Inc. (Barcelona, Spain), weighting around 250 g, were maintained in an air-conditioned room, subjected to 12 hours dark/light cycles, and given standard laboratory rat chow (SAFE A-04, Augy, France) and free access to tap water. Animal experiments were conducted according to the European Communities Council Directives on Animal Care. The animals were divided in four groups: control group (*n* = 8)---vehicle (orange juice); carcinogen (BBN) group (*n* = 20)---0.05% of *N*-butyl-*N*-(4-hydroxybutyl) nitrosamine (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan); *ω*-3 group (*n* = 8)---600 mg/kg BW/day of DHA + EPA (240 + 360, resp.); and *ω*-3 + BBN group (*n* = 8)---in the same conditions described for the previous treatments. The animals underwent a two-phase protocol: a first period of 8 weeks for tumor induction and pharmacological treatment (orange juice, BBN and/or *ω*-3) and a second one of 12 weeks for cancer expression and putative prevention. BBN was given in drinking water and *ω*-3 and vehicle were administered by gavage, using an esophageal cannula. All the animals completed the 20-week study protocol. Body weight (BW) was measured weekly and drink beverage was monitored during the experimental period at intervals of two days.
2.2. Sample Collection and Preparation {#sec2.2}
--------------------------------------
Blood: at the end of treatment, animals were subjected to a blood collection procedure: first, the rats were injected with intraperitoneal anesthesia with 2 mg/Kg BW of a 2 : 1 (v : v) 50 mg/mL Ketamine (Ketalar, Parke-Davis, Pfizer Lab., Seixal, Portugal) solution in 2.5% chlorpromazine (Largactil, Rhône-Poulenc Rorer, Vitória lab., Amadora, Portugal). Blood samples were immediately collected by venipuncture from the jugular vein in needles with no anticoagulant (for serum samples collection). Tissues: the rats were sacrificed by cervical dislocation, and the lungs, stomach, liver, kidneys, and intestines were immediately removed, weighted, and placed in formaldehyde for histological evaluation. Before removal, bladders were intraluminally injected with a buffered formaldehyde solution as prefixation for histological analysis.
2.3. Tumor Chemoprevention Analysis {#sec2.3}
-----------------------------------
### 2.3.1. Quantitative Analysis (Number and Tumor Volume) {#sec2.3.1}
Each bladder pre-fixated in formaldehyde was carefully open; the lumen was inspected for grossly visible lesions and the number of tumors per rat and the tumor volume were reported in order to calculate the % of tumor per group and the mean volume per rat and per tumor.
### 2.3.2. Qualitative Analysis (Bladder Histology) {#sec2.3.2}
The bladder was immersion fixed in 4% buffered formaldehyde and processed for paraffin sectioning. Three slices from each bladder were embedded. Three micrometer thick sections were stained with haematoxylin & eosin (H&E) and examined histologically by clinical personnel with expertise.
2.4. Proliferation, Inflammation, and Redox Status Markers {#sec2.4}
----------------------------------------------------------
Serum levels of transforming growth factor beta-1 (TGF-β1) were measured through an ultrasensitive Quantikine ELISA kits (R&D Systems, Minneapolis, USA). Serum C-reactive protein (CRP) was measured by using an ELISA kit from Helica Biosystems, Inc. (Fullerton, CA, USA). Serum redox status was assessed by the thiobarbituric acid reactive species (TBARs) assay, measuring lipid peroxidation via malondialdehyde (MDA) content, and by the total antioxidant status (TAS) quantification, through ferric reducing antioxidant potential (FRAP) assay, as previously described \[[@B16]\].
2.5. Bladder Cancer CD31 Immunohistochemistry {#sec2.5}
---------------------------------------------
Immunohistochemical (IHC) staining of CD31 in bladder cancer tissue was performed in paraffin-embedded tissue, which was cut into 4 mm sections and mounted on polisinated slides, using standard staining procedures, as described previously \[[@B18]\]. Representative slides were selected for staining and histologic evaluation. Briefly, slides were deparaffinized and hydrated with water. Antigen enhancement was performed by pretreating with microwave heating in a citrate buffer, pH 6.00 (for three pulses of 5 min each at 250 W). The slides were washed three times, 2 min each, and then incubated with blocking serum for 10 min to block the nonspecific binding, and the excess of blocking serum was removed. Staining was performed using a primary monoclonal antibody. A negative control was obtained by omitting the primary antibody. Diaminobenzidine was used as chromogen. Standard procedures were used for visualization and the intensity and/or percentage of positive staining in the dominant pattern within the tumor was graded on a semiquantitative scale (0--3), in which 0 is very low intensity and \<25% of staining; 1 is mild and between 25 and 50%; 2 is moderate and between 50 and 75%; and 3 is severe and \>75% of staining. All slides were reviewed by expert investigators in a blinded data manner.
2.6. Safety Profile {#sec2.6}
-------------------
### 2.6.1. Biochemical Assays {#sec2.6.1}
The following serum parameters were assessed on a Hitachi 717 analyser (Roche Diagnostics Inc., MA, USA) using standard methods: glucose, creatinine, aspartate and alanine aminotransferase (AST and ALT), total cholesterol (Total-c), and triglycerides (TGs).
### 2.6.2. Haematological Data {#sec2.6.2}
Red blood cell (RBC) count, haematocrit (Hct), haemoglobin (Hb) concentration, platelet (PLT) count, and plateletcrit (PCT) were assessed in an automatic Coulter counter (Beckman Coulter Inc., CA, USA).
2.7. Statistical Analysis {#sec2.7}
-------------------------
For statistical analysis, we used the GraphPad Prism version 5.00 software from GraphPad Software (San Diego, CA, USA). Results are means ± standard error of means (s.e.m). Comparisons between groups were performed using ANOVA with Bonferroni post hoc test. Significance was accepted at *P* less than 0.05.
3. Results {#sec3}
==========
3.1. Tumor Chemoprevention Data {#sec3.1}
-------------------------------
### 3.1.1. Body Weight and Beverage Consumption {#sec3.1.1}
No relevant changes were obtained between the groups during the study concerning BW and beverage consumption, despite a trend to lower BW in the two groups under *ω*-3 treatment (data not shown).
### 3.1.2. Macroscopic Evaluation {#sec3.1.2}
All formaldehyde pre-fixated bladders were opened and microscopically analysed for wall (urothelium) texture, thickness, and vascularization ([Figure 1](#fig1){ref-type="fig"}). The percentage of rats with tumor in each group, the number of tumors per rat with tumor, and the mean tumor volume per rat and per tumor were then evaluated ([Table 1](#tab1){ref-type="table"}). All the bladders from the control animals have revealed a pattern of normality, with absence of any type of malignity. The wall texture, thickness, and vascularization were normal ([Figure 1(a)](#fig1){ref-type="fig"}). Similar profile was found for the *ω*-3 group, with limpid, translucent, and tiny bladders, without presence of any abnormal mass or vascularization ([Figure 1(b)](#fig1){ref-type="fig"}). In the BBN group, however, 65.0% (13 in 20) of the rats had bladder tumors, almost all easily seen by macroscopic analysis. The bladder walls were thicker, with new or enlarged small vessels, suggesting neoangiogenesis, and there was unequivocal formation of tumor, some of them of relevant volume, also visible in the whole bladder ([Figure 1(c)](#fig1){ref-type="fig"}). In the *ω*-3 + BBN group, the percentage of mice with tumor was 62.5%. In these, the bladder walls were thicker, with some visible small vessels, suggesting neo-angiogenesis and displaying preneoplasic lesions from hyperplasia but without high-grade dysplasia ([Figure 1(d)](#fig1){ref-type="fig"}).
### 3.1.3. Quantitative Evaluation {#sec3.1.3}
In the control group, as expected, no rat developed bladder cancer. The same profile was found for the *ω*-3-treated animals. In the BBN group (*n* = 20 rats), the percentage of rats with bladder cancer was 65.0% (13 in 20), with a mean of 1.2 ± 0.1 tumors per rat with tumor. Furthermore, the mean tumor volume per rat with tumor (in 13 rats) was 138.5 ± 7.5 mm^3^ and the mean tumor volume per tumor (in 16 tumors) was 112.5 ± 6.4 mm^3^ ([Table 1](#tab1){ref-type="table"}). In the *ω*-3 + BBN group, there was no preventive action of this type of drug concerning the incidence of tumor appearance, since the percentage of mice with tumor was 62.5% (5 in 8), with an average of 1.2 ± 0.2 tumors per rat (6 tumors in 5 rats). The mean tumor volume per rat was 1.1 ± 0.1 mm^3^ and the mean tumor volume per tumor was 0.9 ± 0.1 mm^3^ ([Table 1](#tab1){ref-type="table"}).
### 3.1.4. Qualitative Evaluation {#sec3.1.4}
Concerning the microscopic analysis of urothelium layer and urothelial tumors, the control and *ω*-3 rats have shown normal patterns (Figures [2(a)](#fig2){ref-type="fig"} and [2(b)](#fig2){ref-type="fig"}, resp.). The bladder from control animals had no signs of pre-neoplasic lesions (neither hyperplasia nor dysplasia), as well as those from the *ω*-3 group ([Table 1](#tab1){ref-type="table"}). In the carcinogen (BBN) group, there was evident malignant transformation, including hyperplasia (100%) and dysplasia (100%), present in all the animals, including those without tumor formation ([Table 1](#tab1){ref-type="table"} and [Figure 2(c1)](#fig2){ref-type="fig"}). Furthermore, there were also malignant lesions, papillary, infiltrative, and carcinoma *in situ* (*Cis*) ([Table 1](#tab1){ref-type="table"} and [Figure 2(c2)](#fig2){ref-type="fig"}). Omega-3 treatment has not reduced the incidence of cancer appearance (62.5%), but high-grade dysplasia was absent and there were no infiltrative tumor and no *Cis*([Table 1](#tab1){ref-type="table"} and Figures [2(d1)](#fig2){ref-type="fig"} and [2(d2)](#fig2){ref-type="fig"}). The lungs, stomach, liver, kidneys, and intestines were normal on gross inspection and on histological examination for all the rats of all the groups.
3.2. Proliferation, Inflammation, and Redox Status Profiles {#sec3.2}
-----------------------------------------------------------
In the BBN group, there was a significant increment in serum TGF-β1 (*P* \< 0.01) and CRP (*P* \< 0.001) contents, when compared with the control group (Figures [3(a)](#fig3){ref-type="fig"} and [3(b)](#fig3){ref-type="fig"}, resp.). In the *ω*-3 group, TGF-β1 and CRP serum concentrations remained unchanged when compared with control rats. In the *ω*-3 + BBN group, significant decrease of serum TGF-1 (*P* \< 0.05) and CRP (*P* \< 0.001) were obtained, versus the BBN group.
Serum MDA concentration showed a trend to higher values in the BBN group when compared with control rats. This value was accompanied by a significant increase (*P* \< 0.05) in serum total antioxidant status (TAS). In agreement, serum MDA/TAS ratio, a redox status marker, was unchanged, compared with the control rats (Figures [4(a)](#fig4){ref-type="fig"}, [4(b)](#fig4){ref-type="fig"}, and [4(c)](#fig4){ref-type="fig"}). In the *ω*-3-treated animals, there was a reduction in MDA and TAS levels, in relation to the control group, which was most significant in the case of MDA (*P* \< 0.001). Thus, the MDA/TAS ratio, in the *ω*-3 group, was also significantly lower, showing a clear antioxidant sketch. Similar antioxidant profile was found for the *ω*-3 + BBN group (*P* \< 0.05), when compared with the BBN one (Figures [4(a)](#fig4){ref-type="fig"}, [4(b)](#fig4){ref-type="fig"} and [4(c)](#fig4){ref-type="fig"}).
3.3. Bladder Cancer CD31 Immunohistochemistry {#sec3.3}
---------------------------------------------
The bladders from control ([Figure 5(a)](#fig5){ref-type="fig"}) and Atorva (data not shown) treated rats presented low CD31 immunostaining intensity (grade 0). However, in the BBN group, both the hyperplastic ([Figure 5(b)](#fig5){ref-type="fig"}) and tumoral ([Figure 5(c)](#fig5){ref-type="fig"}) lesions showed elevated expression of CD1 (grade 1 and 2 intensity, resp.). In the *ω*-3 + BBN rats, all the bladders with tumor ([Figure 5(d)](#fig5){ref-type="fig"}) showed a reduced CD31 expression (grade 1) when compared with those of the BBN group with carcinoma and identical to the intensity found in the nontumoral hyperplastic regions of the BBN.
3.4. Safety Profile: Biochemical and Haematological Data {#sec3.4}
--------------------------------------------------------
In the BBN group, serum glucose, creatinine, and lipidic profile data (Total-c and TGs) were unchanged; however, AST was significantly (*P* \< 0.001) increased when compared with the control group ([Table 2](#tab2){ref-type="table"}). Concerning the *ω*-3 group, there was a significantly decrease in serum creatinine (*P* \< 0.001) and TGs (*P* \< 0.01) levels when compared to the control group, while the other parameters were unchanged. Similarly, creatinine (*P* \< 0.001) and TGs (*P* \< 0.01) contents were significantly reduced in the *ω*-3 + BBN group when compared to both the control and the BBN group, while the other markers remained unchanged when compared with the BBN-treated animals ([Table 2](#tab2){ref-type="table"}). Concerning the hematological data, no significant changes were encountered for all the parameters, except a reduction of PCT for the groups under *ω*-3 treatment ([Table 2](#tab2){ref-type="table"}).
4. Discussion {#sec4}
=============
The bladder carcinoma is one of the most prevalent in the Western world. Although 75--80% are not invasive and exhibit high survival rates, they have a high risk of relapse and a significant proportion progresses to more invasive forms \[[@B31]\]. This recurring nature, together with the high morbidity and mortality of the invasive forms that justifies adjuvant therapeutics and aggressive multimodal treatment, makes it one of the cancers with higher socioeconomic costs. Effective methods in preventing relapses would have a role in addressing this neoplasm and become increasingly necessary.
Several types of new and old drugs have been tested during the last years as putative chemopreventive agents for bladder cancer \[[@B32]\]. Preventive and/or treatment options have been based on the possibility of targeting the mechanisms behind the carcinogenesis and tumor growth, including anti-inflammatory and antioxidant agents \[[@B16]--[@B19], [@B33], [@B34]\]. PUFAs, namely, omega-3 fatty acids, might meet these criteria, since they have demonstrated anti-inflammatory and antioxidant properties, among others, and are usually viewed as safe compounds. Furthermore, *ω*-3 fatty acids, besides their use in cardiovascular diseases, have already been tested in several types of tumors, including prostate, colorectal, and breast cancer \[[@B27]--[@B30], [@B35]--[@B37]\], but the data already available are insufficient and sometimes discrepant \[[@B38], [@B39]\]. Concerning the bladder cancer, in particular, the information available is very scarce. The study of Molinari et al. (2011) reported the ability of DHA to induce immunogenic cell death in a cell line of bladder cancer \[[@B40]\], while Mackie et al. (2006) have previously described the *in vitro* synergic effect between a soluble formulation of *ω*-3 fatty acids and epirubicin and mitomycin, used as intrabladder instillation adjuvants \[[@B41]\]. In rodents, the chronic deficiency of essential fatty acids induces transitional cell hyperplasia and augments the possibility of urinary tract carcinogenesis \[[@B42]\], which is in agreement with previous data showing lower serum levels in patients with bladder cancer \[[@B43]\]. In humans, a recent case control study reported that consumption of linolenic acid, an omega-3 fatty acid found in vegetable oils, may have a protective role against bladder carcinoma \[[@B44]\]. Despite some previous information on other types of carcinomas, sometimes divergent, the putative efficacy on bladder cancer remains to be elucidated and more experimental research is justified. The rat model of bladder cancer induced by a nitrosamine (BBN) has been used as a good tool to evaluate chemopreventive strategies, as it has histological similarities with the human bladder carcinoma \[[@B10]--[@B12]\]. Our results are in agreement with previous data from this model, concerning both % incidence of tumors and the type of lesions in the BBN rats \[[@B45], [@B46]\], confirming the credibility and value of this model to study preventive efficacy of drugs. Considering that omega-3 fatty acids are recognized as safe compounds and even present a positive impact on cardiovascular parameters, preventive strategies might be considered for this type of agents. However, in humans the exposure to carcinogens is chronic and we should recognize that even with preventive aims it is unexpected that a treatment could precede the exposure to carcinogens, which was the reason why we adopt the present protocol.
It is largely accepted that the mechanisms underlying cancer appearance and progression are multifactorial. Chronic inflammation and oxidative stress, through their mediators (cytokines, growth factors, and reactive oxygen and nitrogen species), seem to be great contributors for cancer growth \[[@B47]\], and justify the research using anti-inflammatory and antioxidant agents. Among them, omega-3 fatty acids could be viewed as good alternatives \[[@B20]--[@B30]\]. In our study, *ω*-3 fatty acids were unable to promote a significant reduction of incidence of bladder cancer, but the most aggressive premalignant and malignant lesions, such as high-grade dysplasia or *Cis* were absent and no infiltrative cancers were seen, which is of imperative interest. The remarkable reduction of tumor volume also deserves a special note. In fact, in the group under BBN beverage and *ω*-3 fatty acids treatment, the mean volume of tumor per rat was reduced from 138.5 ± 7.5 mm^3^ in the BBN group to 1.1 ± 0.1 mm^3^ in the *ω*-3 + BBN rats, accompanied by a similar outstanding reduction in the mean volume of tumor (from 112.5 ± 6.4 mm^3^ in the BBN rats to 0.9 ± 0.1 mm^3^ in the *ω*-3 + BBN ones). The inhibition of tumor growth is in agreement with other experimental *in vivo* and *in vitro* studies in hepatocarcinoma, breast, colon, and prostate cancer cells or animal models \[[@B48]--[@B51]\].
Given the potential anticarcinogenic mechanisms assigned to these compounds, which include the modulation of eicosanoid production, inflammation, angiogenesis, and susceptibility to apoptosis, similar to the pleiotropic activity of other compounds, such as statins, which have previously demonstrated a beneficial effect on this animal model of BBN-induced rat bladder cancer \[[@B18]\], there was the expectation of different results concerning tumor incidence, reinforced by other data from animal models \[[@B52]\]. The experimental knowledge concerning the preventive anticancer capacity of *ω*-3 fatty acids comes mainly from models of colorectal carcinogenesis. Studies in rats receiving diet supplements reported important reductions of 20 to 50% in the incidence of carcinoma and 30 to 70% in tumor multiplicity \[[@B36], [@B53], [@B54]\]. According to our view, the unchanged incidence of bladder cancer in the *ω*-3 + BBN group, when compared with the carcinogen group, might be due to two particular aspects. Firstly, the dose used might be viewed as a low/moderate *ω*-3 dose and higher DHA and EPA amounts might eventually be able to promote better impact on the cancer appearance. The doses tested vary between different studies of different carcinomas and there is no clear definition yet of the most effective dose, as well as the duration of treatment required to promote a significant effect on cancer prevention \[[@B36]\]. In our study, a dose of 600 mg/kg/day (240 of DHA and 360 of EPA) was used, which might be viewed as a low/moderate dose in terms of an experimental study in a rat model. Secondly, the most effective effect at this concentration of DHA and EPA was the remarkable beneficial impact on redox status, with a reduced MDA/TAS ratio, demonstrative of an antioxidant activity. Since the BBN group was unable to promote a significant impact on MDA/TAS ratio, the redox imbalance (oxidative stress) might not be the "fuel" for cancer appearance but eventually for cancer progression. In agreement, the outstanding positive effect of *ω*-3 on this aspect might have an importance for the reduction of tumor growth, but without impact on the percentage of tumor appearance, to which other mechanisms (genetic and nongenetic) should be better contributors and in which the *ω*-3 treatments are most probably less effective. However, and even more importantly, there was a remarkable inhibition of the development of more aggressive forms of tumor, such as high-grade dysplasia, *Cis*, and infiltrative cancers, which is of crucial interest.
Several molecular mechanisms have been proposed to explain the putative anticancer activity of *ω*-3 fatty acids \[[@B36], [@B55]--[@B57]\]. In our study, omega-3-treated rats were able to prevent the BBN-induced increment of CRP and of TGF-*β*1, suggesting that apart from a clear antioxidant action, there was an anti-inflammatory and anti-proliferative activities both of which might contribute to the inhibition of tumor growth. In fact, previous studies have suggested that the chemopreventive effect of *ω*-3 fatty acids on cancer growth might rely on anti-inflammatory, anti-proliferative, and anti-angiogenic activities \[[@B36], [@B55]--[@B58]\], and our preliminary data of VEGF, bcl2 and P53 expression in bladder cancer tissue of omega-3 treated rats (data not shown) also suggest those properties, recommending additional research. In addition, we found a reduced CD31 expression in the bladders of the omega-3-treated rats when compared with the BBN samples. CD31 is a 130 kDa integral membrane protein, also known as platelet endothelial cell adhesion molecule-1 (PECAM-1), which mediates cell-to-cell adhesion, and is used as a marker of vascularization/angiogenesis. CD31 in bladder carcinoma has been correlated with the tumor grade and stage \[[@B59]\]. Further studies should also address the effect on COX-2 expression, which has been described as an important contributor for carcinogenesis \[[@B60]\] and able to be modulated by *ω*-3 fatty acids \[[@B61], [@B62]\].
One of the aspects of concern in the anticancer agents is the safety profile, often jeopardizing some drugs with good chemopreventive efficacy in cancer growth. Omega-3 fatty acids are recognized as safe compounds and even present a positive impact on cardiovascular parameters. In our study, *ω*-3 therapy was associated with a safe profile on biochemical and haematological data, with a reduction of TGs and unchanged values of renal and liver markers, which is in agreement with what would be expected for this type of compounds. Considering this feature, further studies should address the chemopreventive efficacy in bladder cancer with higher doses of *ω*-3.
5. Conclusions {#sec5}
==============
Omega-3 fatty acids therapy was able to inhibit tumor growth in a rat model of bladder cancer, which might be due to anti-inflammatory, antioxidant, anti-proliferative, and anti-angiogenic properties. In addition, the development of more aggressive pre-malignant and malignant lesions was also inhibited. Given the safe profile demonstrated, and the benefits described in other inflammatory and cardiovascular conditions, omega-3 fatty acids might be viewed as potential interesting natural compounds for the prevention and/or therapy of carcinoma growth, eventually as adjuvants for other drugs, deserving further attention in this area of knowledge.
The authors declare that they have no conflict of interests.
Belmiro Parada, Flávio Reis, Alfredo Mota, Arnaldo Figueiredo, and Frederico Teixeira conceived and designed the study protocol. Belmiro Parada, Flávio Reis, Raquel Cerejo, Patrícia Garrido, José Sereno, Maria Xavier-Cunha, and Paula Neto performed experiments. Belmiro Parada, Flávio Reis, Raquel Cerejo, Arnaldo Figueiredo, and Frederico Teixeira analysed the data and prepared the paper. All authors have read and approved the paper. Belmiro Parada, Flávio Reis and Raquel Cerejo equally contribute to the work.
The authors acknowledge Associação Portuguesa de Urologia (APU) for the research grant.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
{#fig5}
######
Omega-3 effects on bladder cancer chemoprevention: macroscopic (quantitative) and microscopic (qualitative) evaluation of urothelial lesions.
--------------------------------------------------------------------------------------------------------------------------
Macroscopy (quantitative) & microscopy (qualitative) Control\ Omega-3\ BBN\ Omega-3 + BBN\
(*n* = 8) (*n* = 8) (*n* = 20) (*n* = 8)
------------------------------------------------------ ----------- ----------- ---------------------- --------------------
Tumor number
Rats with tumor \[% (*n*/*n*)\] 0 0 65.0 (13 in 20) 62.5 (5 in 8)
Tumors/rat \[mean (*n*/*n*)\] 0 0 1.2 ± 0.1 (16 in 13) 1.2 ± 0.2 (6 in 5)
Tumor volume
Mean/rat \[mm^3^ (*n*)\] 0 0 138.5 ± 7.5 (in 13) 1.1 ± 0.1 (in 5)
Mean/tumor \[mm^3^ (*n*)\] 0 0 112.5 ± 6.4 (in 16) 0.9 ± 0.1 (in 6)
Preneoplasic lesions \[% (*n*/*n*)\]
Hyperplasia 0 0 100 (20 in 20) 87.5 (7 in 8)
High-grade dysplasia 0 0 100 (20 in 20) 0 (0 in 8)
Neoplasic lesions \[% (*n*/*n*)\]
Papillary tumor 0 0 100 (13 in 13) 100 (5 in 5)
Infiltrative tumor 0 0 15 (2 in 13) 0 (0 in 5)
Carcinoma*in situ* 0 0 31 (4 in 13) 0 (0 in 5)
--------------------------------------------------------------------------------------------------------------------------
######
Safety profile: biochemical and hematological data.
--------------------------------------------------------------------------------------------------------
Parameters Control\ Omega-3\ BBN\ Omega-3 + BBN\
(*n* = 8) (*n* = 8) (*n* = 20) (*n* = 8)
----------------------- ----------------- -------------------- ------------------ ----------------------
Biochemical data
Glucose (mmol/L) 9.90 ± 0.54 9.52 ± 0.48 10.22 ± 0.61 9.84 ± 0.52
Creatinine (mmol/L) 51.53 ± 1.68 35.36 ± 1.84^aaa^ 54.81 ± 1.72 32.90 ± 1.64^aaabbb^
Total-c (mmol/L) 1.39 ± 0.06 1.21 ± 0.07 1.31 ± 0.05 1.43 ± 0.06
TGs (mmol/L) 1.73 ± 0.1 0.93 ± 0.03^aa^ 1.47 ± 0.1 0.96 ± 0.11^aabb^
AST (IU/L) 51.57 ± 1.09 62.00 ± 1.08 77.18 ± 4.9^aaa^ 73.86 ± 3.45^aaa^
ALT (IU/L) 30.86 ± 1.75 23.00 ± 0.91 36.17 ± 2.32 38.63 ± 1.86
Hematological data
RBC count (10^12^/L) 7.58 ± 0.29 7.79 ± 0.58 8.07 ± 0.11 7.47 ± 0.15
HCT (pp of 1.0) 0.40 ± 0.02 0.40 ± 0.03 0.42 ± 0.004 0.38 ± 0.009
HGB (g/L) 141.5 ± 5.24 139.5 ± 8.63 146 ± 1.47 137.3 ± 3.09
PLT (×10^9^/L) 990.67 ± 46.21 760.8 ± 77.07 1008.13 ± 46.96 811.00 ± 37.13^b^
PCT (pp of 1.0) 0.0057 ± 0.0002 0.004 ± 0.0004^aa^ 0.005 ± 0.0002 0.0047 ± 0.0002^ab^
--------------------------------------------------------------------------------------------------------
Values are mean ± SEM. ^aa^ *P* \< 0.01 and ^aaa^ *P* \< 0.001 versus control; ^b^ *P* \< 0.05, ^bb^ *P* \< 0.01 and ^bbb^ *P* \< 0.001 versus BBN.
[^1]: Academic Editor: Achille Cittadini
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .70}
{#sp2 .71}
{#sp3 .72}
{#sp4 .73}
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Calcitonin (CT) is a 32-amino acid peptide secreted from the parafollicular cells (C-cells) of the thyroid gland and belongs to a family of peptides including CT gene-related peptide, amylin, adrenomedullin, intermedin and CT receptor (CTR)-stimulating peptide \[[@b1]--[@b4]\]. CT is a hypocalcaemic hormone that induces Ca^2+^ deposition in bones and stimulates calcium excretion into urine \[[@b5]--[@b7]\]. Clinically, CT has been used for treatment of osteoporosis, Paget's disease and hypercalcemia. CT functions by binding to its receptor, CTR, which is a class II G-protein-coupled receptor predominantly expressed in osteoclasts. CTR in osteoclasts has been previously shown to be coupled to Gα and Gα~q~ proteins that links the receptor to both adenylate cyclase--cAMP--protein kinase A (PKA) and Ca^2+^-protein kinase C (PKC)-dependent pathways \[[@b8]\].
High levels of CT lead to diarrhoea. For example, diarrhoea has been reported to occur in 28--39% of patients with medullary thyroid carcinoma (MTC) associated with elevated levels of CT \[[@b9]--[@b11]\]. The diarrhoea in patients with MTC is usually severe, watery and lacks specific treatment and therefore, the mortality is high. Diarrhoea is also observed in other cases such as infusion of CT for hypercalcemia, CT-secreting pancreatic micro-tumours and small cell lung tumours \[[@b12]--[@b16]\]. Previous perfusion studies have shown that CT infusion in healthy humans and rabbits inhibited active sodium absorption and induced active chloride secretion \[[@b17], [@b18]\]. However, the detailed molecular mechanisms of CT-induced chloride secretion in intestinal epithelial cells are poorly understood.
Intestinal chloride secretion plays an important role in body fluid homeostasis and diarrhoea. Several transport processes present on the basolateral and the apical membranes of intestinal cells are involved in driving chloride secretion into the intestinal epithelial lumen. The main chloride channel expressed in small intestine and colon is the cystic fibrosis transmembrane conductance regulator (CFTR). CFTR is expressed in many epithelial tissues, where it has been found to have multiple putative functions, the major one being that of a chloride channel. Activation of CFTR as a Cl^−^ channel requires cyclic AMP, PKA and ATP \[[@b19]\]. Mutations in the gene encoding CFTR leading to a decrease in chloride channel function, is the primary defect in cystic fibrosis, a disease that affects approximately 30,000 patients in the United States alone \[[@b20]\]. On the other hand, activation of CFTR by cholera toxin leads to a massive secretory diarrhoea and life-threatening dehydration \[[@b21]\]. Whether the activation of CTR by CT also stimulates chloride secretion *via* CFTR is not known.
Therefore, current studies were undertaken to investigate the expression of CTR in intestinal epithelial cells and to examine the effect of CT on electrolyte secretion in colonic T84 cell line. Our data showed that CT induced chloride secretion *via* CFTR in a Ca^2+^- and cAMP-dependent manner. Our current studies provide novel insights into the molecular basis of CT-induced chloride secretion that may unravel potential targets for better therapy of diarrhoea associated with high levels of CT.
Materials and methods
=====================
Cell culture
------------
Experiments were performed with the human intestinal T84 cell lines as previously described \[[@b22]\]. DMEM/F12 with 6% calf serum was used for T84 cells. For simultaneous measurements of \[Ca^2+^\] and increased short circuit current (*I*~SC~), cells were seeded onto Snapwell membranes (Costar, Corning, NY, USA) with 0.4 μm pore diameter (culture area 0.1 cm^2^). Cells reached confluency after 7 to 8 days, with a resistance greater than 500 -Ω- cm^2^, and then mounted in Ussing chambers (Physiologic Instruments, San Diego, CA, USA) for electrical measurements.
Real time quantitative RT-PCR analysis
--------------------------------------
RNA was extracted from T84 cells using Qiagen RNeasy kits (Qiagen, Valencia, CA, USA). Equal amounts of RNA from T84 cells were reverse transcribed and amplified in one-step reaction for β-actin and CTR by using Brilliant SYBR Green quantitative RT-PCR (QRT-PCR) Master Mix kit (Stratagene, Santa Clara, CA, USA). Real-time QRT-PCR was performed by using Mx3000P (Stratagene). Human CTR was amplified with gene-specific primers (sense primer: 5′-GCAGGAAGATGTATGCTTTGA-3′; antisense primer: 5′-CTTTACAACAGCTAGGTCCTG-3′) \[[@b23]\]. Human β-actin was amplified as an internal control by using gene-specific primers (sense primer: 5′-CATGTTTGAGACCTTCAACAC-3′; antisense primer: 5′-CCAGGAAGGAAGGCTGGAA-3′) \[[@b24]\].
Immunoblotting
--------------
For immunoblotting studies, briefly, cell lysates were prepared from T84 cells using radio immunoprecipitation assay (RIPA) buffer. A total of 100 μg protein from each of the T84 cells lysates was solubilized in Laemmli sample buffer (2% SDS, 100 mM dithiothreitol, 60 mM Tris, pH 6.8, 0.01% bromophenol blue) and was separated on 8% Tris/glycine SDS-PAGE. For CFTR, 75 μg protein from T84 cell lysate was used. The blot was then probed with primary rabbit anti-CTR antibodies (1:500, Abcam, Cambridge, MA, USA) or rabbit anti-CTR antibody (1:1000) from SantaCruz (Santa Cruz, CA, USA) or rabbit anti-actin antibody (1:5000) from Sigma (Saint Louis, MO, USA) for loading control. Goat anti-rabbit antibody (1:2000, SantaCruz) was used as secondary antibody. The bands were visualized by enhanced chemiluminescence according to the manufacturer's instructions (Amersham, Piscataway, NJ, USA).
Measurement of intracellular Ca^2+^ and cAMP levels in T84 cells
----------------------------------------------------------------
Fluo calcium indicator, Fluo-4,AM (Invitrogen, Carlsbad, CA, USA) was used for measuring intracellular \[Ca^2+^\] changes after CT treatment according to company's suggested protocol. Briefly, transwell cultured T84 cells were incubated with Fluo-4,AM (5 μM) in cell culture incubator for 45 min. After washing with 1 × phosphate-buffered saline (PBS), cells were incubated in 1 × PBS for 30 min. to allow complete de-esterification of intracellular AM esters. The cells were then mounted on a Carl Zeiss LSM 510 laser (Carl Zeiss, Jena, Germany) scanning confocal microscope for live calcium imaging. Beam of 488 from a UV laser was used for excitation. CT was added to basolateral side at a concentration of 10 nM. Images were captured every 5 sec. for 5 min. Intracellular cAMP levels were determined using the Amersham Direct Biotrak EIA kit. On the day of assay, cells were harvested and assayed according to company's suggested protocol. Each assay point was performed at least in triplicate.
Measurements of short-circuit current
-------------------------------------
Agonist-induced anion secretion was measured in T84 monolayer as described \[[@b25], [@b26]\]. Briefly, cells grown on a Snapwell membrane were incubated with bicarbonate-buffered Krebs-Henseleit. The bicarbonate-buffered Krebs-Henseleit solution contained (in mM or mmoles/l) NaCl, 117; NaHCO~3~, 25; KCl, 4.7; MgSO~4~, 1.2; KH~2~PO~4~, 1.2; CaCl~2~, 2.5 and D-glucose, 11, pH 7.4, when bubbled with 5% CO~2~, 95% O~2~. Cl^−^-free solution was prepared by isosmotically replacing NaCl and KCl with sodium gluconate and potassium gluconate, respectively; CaCl~2~ was replaced with 11 mM calcium gluconate to counteract the chelating effect of gluconate anion. The potential difference was clamped to 0 mV, and *I*~SC~ was simultaneously measured using a voltage-clamp amplifier. Both signals were digitized and recorded. For antagonist experiments, cells were pre-incubated with the antagonist 45 min. before CT was added.
*I*~SC~ in nystatin-permeabilized T84 monolayers
------------------------------------------------
T84 cell monolayers were mounted in the Ussing chamber and bathed in normal Krebs-Henseleit solution while the *I*~SC~ was measured as described above. Apical membrane Cl^−^ currents, defined as *I*~(ap)~, were measured in cells permeabilized basolaterally with 300 μg/ml nystatin in the presence of asymmetrical buffers that imposed an apical to basolateral Cl^−^ gradient. Basolateral NaCl was replaced by equimolar sodium gluconate. Nystatin was added to the basolateral membrane 30 min. before the addition of drugs. Under these asymmetrical conditions, activation of the apical membrane Cl^−^ conductance would cause a rapid downward current deflection. Basolateral membrane K^+^ currents, defined as *I*~(bl)~, were measured in cells permeabilized apically with 300 μg/ml nystatin for 30 min. in the presence of asymmetrical buffers that imposed an apical-to-basolateral K^+^ gradient. The apical NaCl was replaced by equimolar K-gluconate, whereas basolateral NaCl was substituted with equimolar sodium gluconate. In all gluconate-containing solutions, CaCl~2~ was increased to 11 mM to compensate for the Ca^2+^ chelating effect of the gluconate anion \[[@b27]\].
Cell surface biotinylation
--------------------------
Cell surface biotinylation studies were performed in T84 monolayers utilizing Sulfo-NH-SS-Biotin (1.5 mg/ml; Pierce, Rockford, IL, USA) in borate buffer (in mM: 154 NaCl, 7.2 KCl, 1.8 CaCl~2~, 10 H3BO~3~, pH 9.0) as described previously \[[@b28]\]. Labelling was allowed to proceed at 4°C to prevent endocytosis and internalization of antigens for 60 min. The biotinylated antigens were immunoprecipitated utilizing streptavidin agarose beads, and the biotinylated proteins were released by boiling in Laemmli buffer containing 100 mM dithiothreitol. Proteins were subjected to SDS-PAGE and transferred to nitrocellulose membranes for immunoblotting.
Statistical analysis
--------------------
In all stances, data shown are the mean ± S.E. of three to six independent experiments. Difference between control *versus* treated was analysed using t-test and ANOVA. Differences were considered significant at *P* \< 0.05.
Results
=======
Calcitonin receptor is expressed in intestinal epithelial cells
---------------------------------------------------------------
First set of studies were undertaken to determine the expression of CTR in human intestinal epithelial cells. As shown in [Figure 1A](#fig01){ref-type="fig"}, real-time PCR analysis demonstrated the expression of CTR mRNA in human colonic T84 cell line. Western blot analysis utilizing CTR-specific antibodies also showed the presence of a ∼56 kD protein band representing CTR protein ([Fig. 1B](#fig01){ref-type="fig"}). These results indicate that T84 cell line represents suitable *in vitro* cellular model for investigating the effects of CT on intestinal electrolyte transport.
{#fig01}
Calcitonin induces short circuit current in T84 cells
-----------------------------------------------------
Human colonic T84 cells are widely used as an *in vitro* model to study the secretory processes across colonic epithelial cells by the measurement of *I*~SC~\[[@b29], [@b30]\]. Therefore, to examine the effect of CT on *I*~SC~, we utilized T84 monolayers mounted in Ussing chambers. Exposure of cells to 10 nM CT added to the basolateral compartment caused a sharp increase in *I*~SC~ ([Fig. 2C](#fig02){ref-type="fig"}). The addition of CT from the apical side had no effect suggesting that CTR is expressed on the basolateral membrane of T84 cells. The increase in *I*~SC~ began within 20 sec. of addition of CT, peaked at 5--10 min. and started to return to baseline slowly by 45--60 min. When compared to the actions of carbachol (CCH, 100 μM) ([Fig. 2A](#fig02){ref-type="fig"}) or Forskolin (FSK (10 μM) ([Fig. 2B](#fig02){ref-type="fig"}), CT-induced *I*~SC~ was much longer in duration than CCH, but shorter than FSK; it was slower than CCH but more rapid than FSK to reach its peak. CT-induced *I*~SC~ in the T84 monolayer was concentration dependent, *i.e.* 10 nM produced higher effects than 1 nM, and interestingly, 100 nM had less effect than 10 nM ([Fig. 2D](#fig02){ref-type="fig"}). Pre-incubation (30 min.) of the T84 monolayer with increasing concentrations of CTR-specific antagonist, CT~8--32~ resulted in a dose-dependent blockage of CT-induced *I*~SC~, with 1 μM of CT~8--32~ almost completely abolishing CT- (10 nM) induced *I*~SC~ (3.1 ± 1.1% of control) ([Fig. 2E](#fig02){ref-type="fig"}). Inhibition only occurred when CT~8--32~ was added to the basolateral not the apical side, further indicating a functional basolateral distribution of CTR in this cell line.
{#fig02}
CT-induced *I*~SC~ in T84 cell monolayer is Cl^−^ dependent
-----------------------------------------------------------
To test whether CT-induced *I*~SC~ in T84 cells results from Cl^−^ secretion, T84 monolayer was mounted in Ussing chambers and Cl^−^ was replaced with equimolar gluconate in the bathing solutions. As shown in [Figure 3A](#fig03){ref-type="fig"}, Cl^−^-free conditions produced a significant reduction of *I*~SC~ (12.5 ± 6.0% of Cl^−^ containing medium, *P* \< 0.001) at 1 min., and was abolished (5.4 ± 1.3% of Cl^−^ containing medium, *P* \< 0.001) at 5 min. after basolateral application of CT. To further determine the nature of the CT-induced *I*~SC~, we tested the effects of inhibitors of transport pathways involved in chloride secretion. As depicted in [Figure 3B and C](#fig03){ref-type="fig"}, the addition of 100 μM bumetanide, a specific blocker of sodium, potassium, chloride co-transporter (NKCC)1, to the basolateral side caused a significant decrease (12.5 ± 6.0%, compared to 80 ± 7.6% of peak *I*~SC~ with vehicle alone). Barium chloride (5 mM), a general inhibitor for K^+^ channels, also resulted in significant decrease in *I*~SC~ induced by CT (13.3 ± 3.3%, compared to 80 ± 7.6% of peak *I*~SC~ with vehicle alone) when added from the basolateral side ([Fig. 3D and F](#fig03){ref-type="fig"}). Furthermore, the addition of 10 μM of CFTR~inh~172 (a specific inhibitor for CFTR), significantly reduced CT-induced *I*~SC~ when added to the apical side as shown in [Figure 3E and F](#fig03){ref-type="fig"} (*I*~SC~ induced by CT decreased to 11.7 ± 4.4%, compared to 80 ± 7.6% of peak *I*~SC~ with vehicle alone). These data suggested that the CT-induced *I*~SC~ in T84 cells is contributed by K^+^ channels, CFTR and NKCC.
{#fig03}
PKA- and Ca^2+^-dependent pathways are involved in CT-induced Cl^−^ secretion
-----------------------------------------------------------------------------
Activation of CTR in osteoclasts has been previously shown to stimulate both adenylate cyclase/cAMP/PKA and Ca^2+^/PKC intracellular signalling pathways \[[@b8]\]. In order to determine if CT-induced chloride secretion in T84 cells involves Ca^2+^-dependent pathways, we examined if there were changes in intracellular Ca^2+^ in these cells in response to CT. The results depicted in [Figure 4A](#fig04){ref-type="fig"} showed that intracellular Ca^2+^ signal increased almost immediately after addition of 10 nM of CT to the basolateral side, and kept elevated for more than 5 min. We also measured intracellular cAMP levels in T84 cells to determine if CT-induced chloride secretion involves the cAMP signalling mechanism. Intracellular cAMP levels increased from baseline 93 ± 12 fmol/μg protein to 9000 ± 346 fmol/μg protein after 20 min. incubation with 10 nM CT ([Fig. 4B](#fig04){ref-type="fig"}). The CT effect was blocked (9000 ± 346 fmol/μg protein *versus* 157 ± 19 fmol//μg protein) by co-incubation with 1000 nM of the CT antagonist CT8--32, while CT8--32 alone had no effect on intracellular cAMP level (93 ± 12 fmol/well *versus* 125 ± 25 fmol/μg protein). FSK (10 μM), used as a positive control, similarly increased cAMP level (17105 ± 2421 fmol/μg protein).
{#fig04}
We next examined the effect of H-89, a PKA inhibitor, and RpcAMP, a cAMP antagonist to further confirm the involvement of the PKA-cAMP pathway in CT-induced chloride secretion. As shown in [Figure 4C](#fig04){ref-type="fig"}, pre-treatment of T84 cells with the H-89 (10 μM) or RpcAMP (25 μM) for 45 min., reduced CT-mediated *I*~SC~ to 45.8 ± 8.2% and 68.5 ± 3.9% of control, respectively.
In order to assess the role of Ca^2+^ or PKC pathways in CT-induced chloride secretion in T84 cells, BAPTA-AM (Ca^2+^ chelator) and bisindolylmaleimide (BIM) (general PKC inhibitor) \[[@b31]\] were used. Pre-incubation of T84 cells with 20 μM BAPTA-AM for 45 min. resulted in 42.4 ± 13.2% of CT-induced *I*~SC~, whereas BIM (5 μM) had no effect ([Fig. 4C](#fig04){ref-type="fig"}). These results indicate that cAMP and Ca^2+^-dependent but not PKC-dependent pathways are involved in the effects of CT. Interestingly, pre-incubation with both 20 μM BAPTA-AM and 10 μM H-89 produced an additive effects (decreased to 25.2 ± 2.3% of control, [Fig. 4C](#fig04){ref-type="fig"}) suggesting each of the cAMP and Ca^2+^-pathways has separate effects on CT-induced *I*~SC~.
CT stimulates *I*~SC~ across the apical and basolateral membranes of intestinal epithelial cells
------------------------------------------------------------------------------------------------
To further delineate the involvement of apical conductance and/or basolateral conductance in the T84 monolayer, the pore-forming antibiotic nystatin was used to selectively permeabilize either cellular membrane \[[@b25]\]. The appropriate transepithelial ion gradients were also established to measure the apical *I*~(ap)~ and basolateral *I*~(bl)~ currents. CT was added to basolateral side in both conditions. *I*~(ap)~ was measured after nystatin permeabilization of the basolateral membrane. The CT-induced *I*~(ap)~ was significantly blocked (28.4 ± 8.7%, [Fig. 5A and C](#fig05){ref-type="fig"}) by H89 (10 μM, 45 min. pre-incubation), while BAPTA-AM (30 μM, 45 min. pre-incubation) alone had no effect in this setting (86.1 ± 13.5%, [Fig. 5C](#fig05){ref-type="fig"}). This current was also sensitive to CFTR~inh172~, which represented a CFTR current ([Fig. 5A](#fig05){ref-type="fig"}). After nystatin permeabilization of the apical membrane, application of CT generated a small *I*~bl~, which was significantly blocked by 45 min. pre-incubation of 20 μM of BAPTA-AM and 10 μM of H89; however, H89 had much less effect (21.2 ± 7.3% of control for BAPTA-AM and 50.0 ± 11.2% of control for H89, *P* \< 0.01, [Fig. 5B and D](#fig05){ref-type="fig"}). This current was not found to be sensitive to NKCC inhibitor, bumetanide, but sensitive to K^+^ channel blocker, barium chloride, which represented a K^+^ channel current ([Fig. 5B](#fig05){ref-type="fig"}). These results suggest that CT evokes both *I*~SC~ across the apical membrane *via* CFTR, and *I*~SC~ current across the basolateral membrane *via* K^+^ channels. These findings also suggest that cAMP signalling component of CT-induced *I*~SC~ involves both apical and basolateral processes, while Ca^2+^ component mostly involves basolateral process induced by CT.
![CT-induced *I*~SC~ involved Ca^2+^- and PKA-dependent pathways in Nystatin-permeabilized preparations. (A) T84 cell monolayer was pre-incubated either with vehicle (control), BAPTA-AM (20 μM) or H89 (10 μM) from both the basolateral and apical sides for 45 min. Then the regular basolateral Krebs-Henseleit solution was replaced with a lower \[Cl^−^\] Krebs-Henseleit medium, so that a chloride gradient was maintained from the apical to basolateral. After stabilizing the baseline, Nystatin (300 μg/ml) was added from the basolateral side. CT (10 nM) was then added from the basolateral side after the baseline was stabilized. CFTR~inh~-172 was added from the apical compartment after the peak of CT-induced *I*~SC~. (B) T84 cell monolayer was pre-incubated either with control = vehicle, BAPTA-AM (20 μM) or H89 (10 μM) from both the basolateral and apical sides for 45 min. Then the regular basolateral Krebs-Henseleit solution was replaced with a lower \[K^+^\] Krebs-Henseleit medium, so that a \[K^+^\] gradient was achieved from the apical to basolateral. After stabilizing the baseline, Nystatin (300 μg/ml) was added from the apical compartment. CT (10 nM) was then added from the basolateral side after the baseline was stabilized. Bumetanide (100 μM) and barium chloride (5 mM) were separately added from the basolateral side after the peak of CT-induced *I*~SC~. (C) Data shown here are the peak *I*~SC~ induced by CT from different pre-treatments: control = vehicle, BAPTA-AM and H89 in (A), which were calculated as percentage of control. (D) Data shown here are the peak *I*~SC~ induced by CT from different pre-treatments: control (vehicle), BAPTA-AM and H89 in (B), which were calculated as percentage of control. Results represent mean ± S.E.M. of three or more independent experiments in all above figures. \* *P* \< 0.01, \*\**P* \< 0.001.](jcmm0015-2697-f5){#fig05}
CT increased CFTR levels on the apical membrane
-----------------------------------------------
A mechanism of activation of CFTR is an increase in its trafficking from the intracellular compartments to the membrane. Aberrations in trafficking are among the causes underlying the manifestation of the disease in CF patients. To examine whether CT increases *I*~SC~ *via* altering the membrane levels of CFTR, cell surface biotinylation studies were performed. Results showed that CT treatment (20 min.) of T84 cells significantly increased the surface level of CFTR compared to control ([Fig. 6](#fig06){ref-type="fig"}).
{#fig06}
Discussion
==========
Previous studies have shown that CT stimulates active Cl^−^ secretion in adult rat ileum *in vitro* \[[@b32]\], rabbit *in vivo* \[[@b17], [@b33]\] and healthy humans *in vivo* \[[@b18]\]. The current studies provide more in depth understanding of the mechanisms of intestinal chloride secretion induced by CT. The major findings of current studies are: (*i*) CTR is expressed in human colonic epithelial cells; (*ii*) direct activation of CTR by CT in intestinal epithelial cells; (*iii*) CT-induced Cl^−^ secretion involves both Ca^2+^- and cAMP-dependent pathways; (*iv*) CT-induced Cl^−^ secretion occurs *via* CFTR chloride channel.
Earlier studies demonstrated the expression of CTR in several cell types including cells in the central nervous system, renal epithelial cells, breast, prostate cells and abundantly in mature osteoclasts \[[@b34]\]. CT was shown to inhibit sodium, hydrogen exchanger (NHE) activity and the Na, K-ATPase in the kidney cell line LLC-PK1 \[[@b29]\]. CT was also found to inhibit proton extrusion in resorbing rat osteoclasts *via* PKA \[[@b30]\]. Although, previous studies showed that CT-induced chloride secretion was observed in human beings and rabbits, no information was available regarding the mechanisms of CT-induced chloride secretion, including CTR expression and its direct activation in intestinal epithelial cells. Our findings provided novel evidence showing the expression of CTR in colonic T84 cell line both at the mRNA and protein levels. These findings provided a compelling evidence for the suitability of these intestinal cell lines to be used as an *in vitro* cellular model to investigate the molecular mechanisms underlying electrolyte transport alterations by CT.
In the T84 cell model, when CT was added to the basolateral side of T84 monolayer, an *I*~SC~ was generated almost immediately, which was blocked by CTR-specific antagonist CT8--32 in a dose-dependent manner, while no current was produced when CT was added from the apical compartment. These findings indicate that this *I*~SC~ was induced by CTR located on the basolateral membrane of the T84 cells, which is consistent with the fact that CT is a blood borne hormone. It is interesting that CT at 10 nM concentration induced chloride secretion more than at 100 nM concentration. As 10 and 100 nM CT-induced cAMP levels to the same extent (data not shown), we believe that this observation is likely due to a receptor de-sensitization. Indeed, diarrhoea occurs in about 28--39% of patients with MTC associated with high CT. These individual differences may reflect different levels of CTR expression in the intestinal epithelial cells or may indicate individual variations in the mechanisms involved in receptor desensitization.
Further experiments confirmed that CT-induced *I*~SC~ in T84 monolayers was indeed due to increases in chloride secretion involving apical CFTR and basolateral NKCC and potassium channels ([Figs 2](#fig02){ref-type="fig"} and [3](#fig03){ref-type="fig"}). Further, biotinylation studies showed that CT increased CFTR levels on the apical membrane. Our findings also demonstrated that after activation of CTR by CT, two intracellular signalling pathways are triggered in T84 cells: cAMP- and Ca^2+^-dependent pathways. The earlier studies in rats, rabbits and humans, measured only the total amount of electrolyte movements across of the intestinal epithelia. There are no data delineating the signalling pathways in intestinal epithelial cells. Our data showed that the direct activation of CTR by CT in T84 monolayers, significantly increased intracellular cAMP levels, which then either directly activated CFTR or stimulated basolateral membrane K^+^ channel activity. The basolateral K^+^ channels, promote potassium recycling across the basolateral membrane by extruding K^+^ entering either *via* the NKCC cotransporter or the Na, K-ATPase pump, and thereby increasing the electrochemical gradient for Cl^−^ entering *via* NKCC to exit the cell *via* apical membrane chloride channels ([Figs 4](#fig04){ref-type="fig"} and [5](#fig05){ref-type="fig"}). Our findings also demonstrated that CT caused an immediate increase in Ca^2+^ and the BAPTA data showed that the role of Ca^2+^ in CT evoked chloride secretion was through K^+^ channels on the basolateral membrane ([Figs 4](#fig04){ref-type="fig"} and [5](#fig05){ref-type="fig"}). Although activating distinct pathways, Ca^2+^and cyclic nucleotides often modulate each other's activities and the fact that CT activates both provides some intriguing possibilities for exploring the cross-talk between the two cascades. Responses to cyclic nucleotide-mediated agonists are sustained, whereas those to Ca^2+^-mediated agonists are transient even though levels of intracellular Ca^2+^ can remain elevated after the secretory response has resolved \[[@b35]\]. This may explain the fact that CT-induced chloride secretion has a unique feature: quick response that could be due to activation of Ca^2+^-dependent pathway like CCH and longer duration that may result from activation of cAMP-dependent pathway like FSK ([Fig. 2](#fig02){ref-type="fig"}).
Based on our findings, we propose the following cell signalling pathways of CTR mediated CFTR activation ([Fig. 7](#fig07){ref-type="fig"}). CT binds and activates CTR located at the basolateral membrane of T84 cells, which then evokes cAMP- and Ca^2+^-dependent pathways stimulating CFTR and secreting chloride. In cAMP-dependent pathway, cAMP not only affects CFTR directly, either by activating CFTR channels or by increasing CFTR levels on the apical membrane, but also activates K^+^ channel on the basolateral membrane. In Ca^2+^-dependent pathways, Ca^2+^ activates the K^+^ channel at the basolateral side providing the driving force for chloride influx *via* NKCC.
{#fig07}
The current findings of CTR expression in intestinal epithelial cells provide a possible target for managing CT-induced severe diarrhoea in MTC patients, especially patients with liver metastasis. Most patients have more than 12 bowel movements a day, thus the mortality is high due to dehydration. Currently, specific treatment for such a condition is not available. CT8--32, a peptide lacking the critical N-terminal 7 amino acids, showed significant blocking of the effects of CT-induced chloride secretion. Therefore, targeting at CTR may help to develop some novel therapeutic modality to treat CT-induced diarrhoea in MTC patients and other clinical situations where severe diarrhoea is developed in CT-secreting pancreatic micro-tumour, and small cell lung tumours, etc.
In summary, the current study showed for the first time the expression of CTR in intestinal epithelial cells. Our data further demonstrated that CT, a hormone that regulates calcium homeostasis, directly activates CTR and induces CFTR-mediated chloride secretion in intestinal epithelial cells *via* cAMP- and Ca^2+^-dependent signalling pathways. Our results define the molecular mechanisms underlying CT effects on intestinal ion transport and may enhance the possibility of developing novel therapeutic approaches for the treatment CT-induced diarrhoea.
Conflict of interest
====================
The authors confirm that there are no conflicts of interest.
These studies were supported by the Department of Veterans Affairs (P.K.D. and W.A.A.) and the NIDDK grants: DK54016 (P.K.D.), DK81858 (P.K.D.), DK71596 (W.A.A.), DK74458 (R.K.G.) and the Program Project Grant DK67887 (P.K.D. and M.C.R.) and the training grant T32DK007788 (H.L.).
[^1]: These co-senior authors contributed equally.
| {
"pile_set_name": "PubMed Central"
} |
According to a 2010 orthopaedic residency program directors\' forum, the top concerns among programs include potential compromises to resident learning experience caused by work-hour restrictions and the need to identify a body of core knowledge with specific goals and expectations that all residents should meet.^[@R1]^ Similar challenges in Canada and Europe have resulted in some programs transitioning from a time-based structure to a competency-based medical education model, where residents progress at an independent pace by demonstrating the required knowledge and skills.^[@R2][@R3][@R4]^ It remains unclear whether US programs will eventually transition to a competency-based medical education. Canadian and European examples demonstrate the challenge in the lack of reliable evaluation tools or standardized curriculum to cover the breadth of orthopaedic education.^[@R5]^
The Accreditation Council for Graduate Medical Education (ACGME) provides guidelines indicating graduating resident case log minimums for certain core competency procedures. National summative case log data are available publicly (Figure [1](#F1){ref-type="fig"}), and individual programs also receive their specific data with which they can compare with national averages. By way of comparison, the ACGME provides much less guidance about the amount of time residents do and should spend on different rotations throughout residency. Currently, the ACGME and American Board of Orthopaedic Surgery (ABOS) provide only broad guidelines indicating the number of months postgraduate year (PGY) 2 to 5 residents should spend on general orthopaedic services^[@R6],[@R7]^ (Table [1](#T1){ref-type="table"}). However, there are no guidelines outlining the amount of time residents should spend on subspecialty services such as hand, arthroplasty, foot and ankle, sport, spine, and oncology. Furthermore, orthopaedic residency programs are increasingly offering elective and research rotations, but little is known about how variable these practices currently are.^[@R8]^
{#F1}
######
ACGME and ABOS Requirements for Curriculum Organization in Orthopaedic Surgery Residency Programs: Postgraduate Years 2 to 5
The purpose of this study is to summarize the subspecialty rotation exposure across ACGME-accredited orthopaedic residency programs and to examine associations between rotation schedule structure and available program-specific factors. Characterizing variation in resident rotation exposure is useful to residency program administrators because they compare their practices with other ACGME-accredited programs and work toward developing new and potentially more standardized orthopaedic residency training structure. Applicants to orthopaedic surgery residencies may also benefit from knowledge of this variability as they compare programs.
Methods {#s1}
=======
The ACGME Accreditation Data System website was used to generate a list of the 165 ACGME-accredited US orthopaedic surgery residency programs (updated August 2017) and their program coordinators\' contact information. Programs for D.O. medical graduates only were excluded (n = 3). Programs were included if the rotation schedules for PGY 2 to 5 showing the number of weeks spent on distinct rotations could be obtained from the residency program coordinator, from the program\'s website, or from faculty and residents at the program. Regardless of the method of collection, programs were excluded if the obtained schedule did not clearly outline the amount of time residents spend on each rotation throughout PGY 2 to 5. PGY-1 schedules were excluded from the analysis, given the time spent on nonorthopaedic specialty services during the intern year.
Schedules were reviewed to categorize the number of months that PGY-2, PGY-3, PGY-4, and PGY-5 residents spent on each of the following rotations: general orthopaedics, trauma, pediatrics, hand, sport, foot and ankle, arthroplasty, oncology, spine, research, and elective. The percentage of residency spent in each category was then calculated as number of months divided by 48 months. Rotations that were unable to be placed into one of the aforementioned 11 categories were noted. These included rotations that were listed by hospital name in their program schedules for which it was not possible to determine the types of cases residents were specifically involved with at these locations. Programs were excluded from analysis if \>3 months of the rotation schedule across the 4 years could not adequately be categorized with the rotation schedule information.
A total of 115 rotation schedules were obtained, representing 70.1% of all ACGME-accredited orthopaedic surgery residency programs. Forty-eight (42%) schedules were obtained from program coordinators, 53 (46%) were obtained from the program website, and 14 (12%) were obtained from program faculty or residents. Of the 115 programs, 9 programs were excluded because of the inability to accurately categorize \>3 months of time in their rotation schedules. Therefore, a total of 106 rotation schedules (65.4% of accredited programs) were included in the analysis. Of the 106 included programs, 15 (14.2%) had between 1 and 3 months of time in their rotation schedules across the 4 years that were unable to be adequately categorized.
The total percent of time spent on subspecialty rotations (including hand, pediatrics, arthroplasty, sport, foot and ankle, spine, and oncology) was then compared for the following program-specific variables: program size and presence of orthopaedic subspecialty fellowships at the program\'s institution.
Programs were divided into three groups based on size: group 1, 1 to 3 residents/yr; group 2, 4 to 5 residents/yr; and group 3, ≥6 residents/yr based on the interquartile range. A Kruskal--Wallis test was performed to compare the percentage of residency spent on subspecialty rotations between the three groups. A Mann--Whitney *U* test was performed to compare the percentage of residency spent on subspecialty rotations in programs with orthopaedic subspecialty fellowships at their institution with programs without any subspecialty fellowships.
Results {#s2}
=======
Subspecialty Rotation Exposure {#s2-1}
------------------------------
The average percentage of residency and the number of months spent on each rotation are presented in Figures [2](#F2){ref-type="fig"} and [3](#F3){ref-type="fig"}. The greatest percentage of residency spent was in the following categories: trauma (16.6%; mean 8.0 months), general orthopaedics (13.7%, mean 6.6 months); and pediatrics (12.5%, mean 6.0 months). Rotations with the highest variation between programs included the following: general orthopaedics (SD 5.8 months; range 0 to 30 months), sport (SD 2.5 months, range 0 to 15 months), and arthroplasty (SD 2.3 months, range 0 to 11.8 months). Sixty-seven programs (63.2%) had dedicated blocks for research (mean 1.65 months, range 0 to 6.25 months), and 25 programs (23.6%) had dedicated blocks for electives (mean 0.60 months, range 0 to 5 months).
{#F2}
{#F3}
Relationship Between Subspecialty Rotation Exposure and Program-specific Factors {#s2-2}
--------------------------------------------------------------------------------
Of the 106 programs, 45.3% (n = 48) had 4 to 5 residents/yr, 30.1% (n = 32) had ≥6 residents/yr, and 24.5% (n = 26) had 1 to 3 residents/yr (Figure [4](#F4){ref-type="fig"}). There were no significant differences in the percentage of residency spent on subspecialty rotations based on program size (1 to 3 residents, mean 62.3%; 4 to 5 residents, mean 63.9%; and ≥6 residents, mean 62.9%; *P* = 0.80). Programs with subspecialty fellowships at their institution spent an average percentage of time on subspecialty rotations of 63.9% compared with programs without subspecialty fellowships in which 61.4% of time was spent on subspecialty rotations. This difference was not statistically significant (*P* = 0.77).
{#F4}
Discussion {#s3}
==========
The purpose of this study was to summarize the current subspecialty rotation exposure between ACGME-accredited orthopaedic surgery programs. We found that residents on average spent the most time on trauma rotations, followed by general orthopaedics and pediatrics. Subspecialty rotation exposure among programs was highly variably; as such, we found no notable correlations between subspecialty rotation exposure and program size or presence of fellowship training at the corresponding institution. Our study also demonstrated that most programs have dedicated time for research (63.2%) and about a quarter of the programs offer residents dedicated elective time.
Our finding that 63.2% of programs offer dedicated research time is consistent with that of Williams et al.,^[@R8]^ who reported that 69.6% of ACGME-accredited orthopaedic programs offer residents protected research time. This percentage is likely higher than ours because Williams et al.^[@R8]^ included programs with longitudinal research time across the 5 years, which was not represented on the block rotation schedules that we analyzed. Previous literature has not characterized how common it is among orthopaedic programs to offer elective time for residents. In the landscape of increasing competition for orthopaedic fellowships and specialization by practicing orthopaedic surgeons, these trends may increase moving forward.
Dedicated research and elective time provide residents with more autonomy and may help to position them more favorably for fellowship applications. Increased autonomy is also known to increase motivation, which may increase general resident experience and well-being. Protected research time has been associated with an increased number of publications in residency,^[@R8]^ and electives offer residents an opportunity to gain additional experience in research or their subspecialty interests. Our study showed that both research and elective rotations are variable between programs. Program administrators may take these data in to consideration because they modify and enhance their curricula.
This study has several limitations. Our findings are only as strong as was our ability to accurately categorize each rotation listed on the rotation schedule. Several programs\' rotation schedules categorized rotations by hospital name instead of the nature of the clinical experience. As a result, we excluded programs in which we were unable to adequately categorize over 3 months of time in the rotation schedule over the course of the 4 years examined. However, for included programs where between 1 and 3 months of time was unable to be categorized, it is possible that the time spent on subspecialty rotations was underestimated, unaccounted for, or inappropriately represented. Another limitation was the heterogeneity in means by which rotation schedules were obtained and reported and the fact that program curriculum may change from year to year. Although sources such as websites may not reflect the most current program practices, a recent study showed that orthopaedic surgery residency program websites have markedly increased in comprehensiveness over the past 10 years.^[@R9]^
We also acknowledge some limitations in appropriate categorization of trauma. Although the ABOS requires a minimum of 12 months, our study found an average of 8 months across programs. This discrepancy may be linked to the fact that trauma is incorporated into many general orthopaedics rotations and as such, residents are exposed to additional trauma on these rotations. The sum of the average number of months spent on general orthopaedics and trauma in our study equals 14.4 (closer to the 12-month ABOS requirement), suggesting that a large overlap exists between general orthopaedics and trauma. Because of this overlap, we decided to exclude the time spent on trauma from the total time spent on subspecialty services in our analyses examining the association between subspecialty rotation exposure and program size/fellowship availability. In support of these methodological decisions, our data closely parallel the ABOS requirement for months spent on pediatric orthopaedics (6.0 months), and this subspecialty rotation also had less variation between programs. This consistency indicates two important points. First, this suggests generalizability of accuracy in our ability to categorize the other subspecialty services, and second, less variation exists in the categories in which the ABOS provides guidelines.
It is important to underscore that the time spent on subspecialty rotations does not equate with resident surgical or educational experience on the rotation. The aim of this study was to provide summative information regarding the amount of time residents are being exposed to different subspecialty rotations between programs where these data are not publicly available from the ACGME. Our study found a relatively smaller variation in exposure to pediatric orthopaedic surgery, likely because that it is the only subspecialty in our analysis for which the ACGME provides guidance to programs regarding resident exposure. To better understand the educational impact of subspecialty rotation exposure, it would be important to study how the amount of time spent on a rotation correlates with the number and type of surgical cases residents performed or performance on standardized tests of knowledge. Unfortunately, although the ACGME provides national average case log data, program-specific information is not publicly available and was therefore not feasible in the present study of 115 residency programs. Furthermore, as demonstrated in Figure [1](#F1){ref-type="fig"}, ACGME\'s representation of the case log data does not allow one to understand where the case exposure is coming from for the different rotations.
Finally, there are likely many more factors contributing to the current variation in subspecialty exposure between programs than those measureable in this study, including the specific hospital and patient care environment, teaching faculty, and the available volume and diversity of clinical opportunities between programs.
Conclusion {#s4}
==========
Our study found a high variability in the subspecialty rotation exposure among ACGME-accredited orthopaedic surgery residency programs. Additional research should be conducted to investigate the impact of this variation on resident experience, competency, and satisfaction. It is important for graduate medical education leadership to understand and appreciate this variation as they consider alternative training structures. Program directors receive data from the ACGME, detailing how their program case volumes compare with national averages. The ACGME does not currently provide summative national program data regarding rotation structure. In combination with available case log data, program directors can use the information presented in this study to compare their rotation structure with other ACGME-accredited programs to inform changes to their curriculum. Furthermore, applicants to orthopaedic surgery residencies may also benefit from knowledge of this variability as they compare programs.
None of the following authors or any immediate family member has received anything of value from or has stock or stock options held in a commercial company or institution related directly or indirectly to the subject of this article: Ms. Sacks, Dr. Stepan, Dr. Felix, Dr. Albert, Dr. Bostrom, and Dr. Fufa.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Despite improvement in the treatments and techniques for peritoneal dialysis (PD) patients, long-term PD leads to the decline of residual kidney function (RKF) and peritoneal membrane function (PMF), as a result of membrane or ultrafiltration (UF) failure^[@CR1],[@CR2]^. Existing epidemiological studies have illustrated that RKF deteriorations over time in PD patients compromising patient survival as well as overall health-related quality of life (HRQOL)^[@CR3]--[@CR5]^. The CANUSA (Canada-United States Peritoneal Dialysis) study, the landmark multicenter prospective cohort of incident PD patients, showed 12% and 36% reductions in the risk of death for each 5 L/week/1.73 m^2^ increment in estimated glomerular filtration rate (eGFR) and each 250 mL increase in urine volume, respectively^[@CR3]^. Likewise, the risk of UF failure has increased 3--5% in the first year and 30--50% after three years of PD^[@CR6]--[@CR8]^. There is increasing evidence on the inter-relationship between the RKF and PMF^[@CR9]^. Alterations in RKF and peritoneal characteristics over time are important determinants of patients' technique survival and mortality^[@CR9],[@CR10]^. Subsequently, treatment strategies for maintaining RKF in conjunction with PMF are crucial.
Blockade of the renin-angiotensin aldosterone systems (RAAS) with angiotensin-converting enzyme inhibitors (ACEIs), angiotensin II receptor blockers (ARBs), and mineralocorticoid receptor antagonists (MRAs) in PD patients are likely to preserve residual glomerular filtration rate (rGFR) along with residual urine volume until the PD patients reach anuria that may improve survival in these population^[@CR11]--[@CR15]^. Several studies revealed that blockade of RAAS positively effects the peritoneal membrane by reducing morphologic changes and preserving peritoneal membrane integrity^[@CR16]--[@CR18]^. Therefore, inhibitions of RAAS could potentially improve technique survival and allow patients to be sustained on PD programs for longer periods.
However, the role of RAAS blockade in PD patients has not been fully elucidated. Some studies have revealed the protective properties^[@CR11]--[@CR15],[@CR19]^, whereas others have not^[@CR20]--[@CR26]^. Previous systematic reviews have shown that ACEIs/ARBs substantially benefit in preserving rGFR in PD patients, while a lack of evidence exist regarding the relative efficacy of MRAs and direct renin inhibitors (DRIs)^[@CR27]--[@CR29]^. Moreover, existing pairwise meta-analyses have focused mainly on RKF rather than other clinically relevant and related outcomes such as PMF and adverse events^[@CR27]--[@CR29]^. Recently, treatment with an ACEIs or ARBs has been recommended by the International Society for Peritoneal Dialysis (ISPD)^[@CR30]^ for PD patients with significant RKF, although the comparative effectiveness of specific RAAS blockade classes remains unknown.
To address this knowledge gap, we conducted a systematic review and network meta-analysis (NMA) of randomised controlled trials (RCTs) and non-randomised studies in PD patients to evaluate the effects of specific RAAS blockade classes on RKF and PMF as determined by five key parameters: rGFR, urine volume, incidence of anuria, dialysate-to-plasma creatinine ratio (D/P Cr), and acceptability of treatment.
Results {#Sec2}
=======
Search strategy and characteristics of included studies {#Sec3}
-------------------------------------------------------
The systematic search details are described in Fig. [1](#Fig1){ref-type="fig"}. After screening of titles/abstracts, 101 full-text articles of potentially relevant studies were acquired. After appraising these articles against study inclusion/exclusion criteria (Supplementary, Table [S1](#MOESM1){ref-type="media"}), we included 10 RCTs^[@CR12]--[@CR14],[@CR25],[@CR31]--[@CR36]^ and 10 non-randomised studies^[@CR11],[@CR20]--[@CR24],[@CR26],[@CR37]--[@CR39]^ that compared RAAS blockade classes with active control (Table [1](#Tab1){ref-type="table"}). Four RAAS blockade classes were compared with active control---ACEIs, ARBs MRAs, and mixed ACEIs/ARBs, however, no data for DRIs in any outcome of interest. Two studies provided direct comparisons of ACEIs and ARBs^[@CR31],[@CR36]^. Network diagrams presenting the available evidence for primary and secondary outcomes are illustrated in Figs. [2](#Fig2){ref-type="fig"} and [S1](#MOESM1){ref-type="media"}. Detailed methods of measurement and definition of outcomes are described in Supplementary Table [S2](#MOESM1){ref-type="media"}. A total of 3,789 PD participants were enrolled in the set of included studies; the majority of these patients received continuous ambulatory peritoneal dialysis (CAPD). The baseline mean age and rGFR ranged from 40.2--66.8 years and 0.6--8.4 mL/min/1.73 m^2^, respectively. The follow-up periods ranged from 7 days to 66.3 months, and 12 (60%) studies encompassed participants from Asia. The study- and participant-characteristics are illustrated in Table [1](#Tab1){ref-type="table"} and Supplementary, Table [S3](#MOESM1){ref-type="media"}.Figure 1Selection of studies. Abbreviations: RCTs, randomised-controlled trials.Table 1Description of included studies: RCTs and non-randomised studies.Author, YearDesignCountry EnrollmentSample SizeInterventionControlMean Age ± SD, YearFemale, N (%)Mean rGFR ± SD, mL/minMean Urine Volume ± SD, mL/dayPD ModalityFollow-Up Period, Mean ± SDRisk of Bias^a^Favazza et al., 1992^[@CR32]^RCT: open label, crossover studyItaly9Enalapril (40 mg/day)Nifedipine (60 mg/day), Clonidine (0.45 mg/day)64.0 ± 5.43 (33.3)3.9 ± 0.8NRCAPD14 days1/8Moist et al., 2000^[@CR11]^Non-randomised studies: prospective cohort studyUSA1,032ACEI userNon-ACEI users55.5 ± 14.6490 (47.5)7.5 ± 2.7^b^NRCAPD, APD11.9 ± 1.7 months7/9Johnson et al., 2003^[@CR37]^Non-randomised studies: prospective cohort studyAustralia146ACEI usersNon-ACEI users54.8 ± 16.383 (56.8)4.9 ± 2.3^b^NRCAPD, APD20.5 ± 14.8 months7/9Li et al., 2003^[@CR12]^RCT: open-label, parallel studyHong Kong60Ramipril (5 mg/day)Active control^c^58.6 ± 12.122 (36.7)3.6 ± 2.0^b^NRCAPD12 months3/8Phakdee-kitcharoen et al., 2004^d [@CR31]^RCT: open label, crossover studyThailand21Candesartan (8 mg/day)Enalapril (10 mg/day)44.8 ± 10.17 (33.3)2.0 ± 2.4NRCAPD1 months1/8Suzuki et al., 2004^[@CR13]^RCT: open-label, parallel studyJapan34Valsartan (40--80 mg/day)Active control^c^63.5 ± 3.516 (47.0)4.3 ± 1.7^b^1045.0 ± 220.6CAPD24 months3/8Rojas-Campos et al., 2005^[@CR20]^Non-randomised studies: quasi experimental (crossover) studyMexico20Losartan (50--200 mg/day)Prazosin (2--6 mg/day), verapamil (80--240 mg/day)42.9 ± 16.64 (20.0)NRNRCAPD7 days1/8Wang et al., 2005^[@CR33]^RCT: open-label, parallel studyChina32Valsartan (40--80 mg/day)Active control^c^42.0 ± 11.512 (35.3)4.9 ± 2.2^b^1085 ± 696.3CAPD28 ± 13 months1/8Furuya et al., 2006^[@CR21]^Non-randomised studies: quasi experimental (crossover) studyJapan8Candesartan (8 mg/day)Active control^c^66.8 ± 8.84 (50.0)NR1035 ± 383.5CAPD, APD3 months1/8Jearnsujitwimol et al., 2006^[@CR39]^Non-randomised studies: quasi experimental (crossover) studyThailand7Candesartan (8--16 mg/day)Active control^c^62.0 ± 3.62 (28.6)0.6 ± 0.416.9 ± 8.2CAPD12 weeks for treatment,6 week for control1/8Zhong et al., 2007^[@CR34]^RCT: open-label, parallel studyChina44Irbesartan (300 mg/day)Active control^c^44.0 ± 14.614 (31.8)4.5 ± 2.7^b^1255 ± 425.1CAPD12 months1/8Wontanatawatot et al., 2009^[@CR35]^RCT: open-label, parallel studyThailand46Enalapril (40 mg/day)Active control^c^48.1 ± 12.025 (54.3)NRNRCAPD6 months1/8Jing et al., 2010^[@CR22]^Non-randomised studies: retrospective cohort studyChina66ACEI/ARB usersNon-ACEI/ARB users52.5 ± 12.224 (36.4)4.6 ± 2.7NRCAPD12 months6/9Kolesnyk et al., 2011^[@CR23]^Non-randomised studies: prospective cohort studyNetherland452ACEI/ARB usersNon-ACEI/ARB users50.8 ± 10.6154 (34.1)4.9 ± 2.4^b^NRNot specified3 years8/9Basturk et al., 2012^[@CR24]^Non-randomised studies: prospective cohort studyTurkey43ACEI usersNon-ACEI users40.2 ± 18.719 (44.2)NR332 ± 476.3CAPD6 months6/9Reyes-Marín et al., 2012^[@CR36]^RCT: open-label, parallel studyMexico60Enalapril (10 mg/day)Losartan (50 mg/day)45.8 ± 19.024 (40.0)3.9 ± 1.8^b^NRAPD12 months1/8Ito et al., 2014^e [@CR14]^RCT: open-label, parallel studyJapan158Spironolactone (25 mg/day)Active control^c^56.5 ± 13.445 (28.5)NR1009.2 ± 762.2Not specified24 months3/8Szeto et al., 2015^[@CR38]^Non-randomised studies: retrospective cohort studyHong Kong645ACEI/ARB usersNon-ACEI/ARB users57.2 ± 12.7286 (44.3)3.7 ± 2.3^b^NRCAPD66.3 ± 34.7 months7/9Yongsiri et al., 2015^[@CR25]^RCT: double-blind, crossover studyThailand20Spironolactone (25 mg/day)Placebo^f^52.4 ± 12.412 (60.0)NR895.0 ± 582.0CAPD1 months3/8Shen et al., 2017^[@CR26]^Non-randomised studies: retrospective cohort studyUSA886ACEI/ARB usersNon-ACEI/ARB users65.5 ± 13.6390 (44.0)8.4 ± 4.8^b^991.6 ± 648.8CAPD, CCPD12.0 ± 10.8 months8/9^a^For RCTs, and quasi-experimental study, the risk of bias was assessed based on the Cochrane Collaboration's tool and expressed as the number of low risk-risk judgments (ranging 0--8), while the Newcastle-Ottawa Scale (NOS) was applied for cohort study and summary scores ranging from 0--9 points.^b^Adjusted for body surface area.^c^Trial did not use a placebo.^d^Data were based on nonanuric and anuric patients at baseline.^e^All participants in both arm received ACEI or ARB treatment for at least 3 months.^f^Antihypertensive agents were allowed except for ACEIs or ARBs treatment.Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; APD, automated peritoneal dialysis; ARB, angiotensin II receptor blocker; CAPD, continuous ambulatory peritoneal dialysis; CCPD, continuous cyclic peritoneal dialysis; rGFR, residual glomerular filtration rate; NR, not reported; PD, peritoneal dialysis; RCTs, randomised-controlled trials; SD, standard deviation; USA, the United States of America.Figure 2Network plot of eligible comparisons for primary outcomes. Notes: The circles (nodes) represent the available treatments and the lines (edges) represent the available comparisons. Size nodes and width of edges indicate weighting according to the numbers of studies involved for each treatment and comparison, respectively. Abbreviations: ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; D/P Cr, dialysate-to-plasma creatinine; MRAs, mineralocorticoid receptor antagonists; rGFR, residual glomerular filtration rate.
Risk of bias of included studies {#Sec4}
--------------------------------
The summary results of the risk of bias assessment are shown in Table [1](#Tab1){ref-type="table"}. Overall, the included studies had low methodological quality and high risk of bias (Supplementary Table [S4](#MOESM1){ref-type="media"}). Among 10 RCTs and 3 quasi-experimental included studies, the risk of bias was high or unclear for sequence generation (61.5%); allocation concealment (69.2%); blinding of participants (92.3%), blinding of personnel (92.3%), and blinding of outcomes assessors (92.3%); completeness of outcome reporting (53.8%); and selective reporting of outcomes (92.3%). According to the Newcastle-Ottawa Scale (NOS), the summary scores of included 7 cohort studies ranged from 6--8 points, 5 cohort studies were considered to be at high-quality (≥7 points).
Residual kidney function {#Sec5}
------------------------
Findings from pairwise meta-analysis and NMA for RKF outcomes were consistent, except for the effect of ACEIs on change in rGFR (Table [2](#Tab2){ref-type="table"}). The overall results are summarised in Fig. [3](#Fig3){ref-type="fig"}; and Supplementary, Tables [S5](#MOESM1){ref-type="media"} and [S6](#MOESM1){ref-type="media"}. For pairwise comparison, heterogeneity was low to moderate (*I*^2^ \< 75%), however, *I*^2^ values higher than 75% was identified for the comparisons of ARBs versus active control on change in urine volume (*I*^2^ = 87%). Rankogram and cumulative probability curves of RAAS blockade against active control are provided in Supplementary, Figs. [S2](#MOESM1){ref-type="media"} and [S3](#MOESM1){ref-type="media"}, respectively.Table 2Summary of findings versus active control and the strength of evidence from pairwise meta-analysis and NMA.Treatment Comparison^a^Findings from RCTsFindings from RCTs and non-randomised studiesStrength of EvidenceNo. of Studies Included^b^ (N)Pairwise Meta-AnalysisNetwork Meta-Analysis^c^No. of Studies Included^b^ (N)Pairwise Meta-AnalysisNetwork Meta-Analysis^d^Effect Estimate (95% CI)*I*^2^ (*P*-Value)*τ*^2^Effect Estimate (95% CI)Effect Estimate (95% CI)*I*^2^ (*P*-Value)*τ*^2^Effect Estimate (95% CI)**Change in rGFR, mL/min/1.73 m**^**2**^ACEIs2 (62)0.17(-0.80 to 1.15)70% (0.069)0.353SMD 0.52(−0.07 to 1.11)2 (62)0.17(−0.80 to 1.15)70% (0.069)0.353SMD 0.55(0.06 to 1.04)LowARBs3 (104)0.82(0.17 to 1.47)59% (0.086)0.195SMD 0.62(0.10 to 1.14)3 (104)0.82(0.17 to 1.47)59% (0.086)0.195SMD 0.62 (0.19 to 1.04)LowMixed ACEIs/ARBsNANANANANA2 (711)0.41 (0.25 to 0.57)0% (0.620)\<0.001SMD 0.45 (0.03 to 0.86)Insufficient**Change in Urine Volume, mL/day**ACEIsNANANANANA1 (43)SMD 0.20(−0.45 to 0.86)NANASMD 0.20(−2.39 to 2.80)InsufficientARBs3 (112)SMD 1.38(−0.07 to 2.82)91% (\<0.001)1.466SMD 1.39(−0.29 to 3.08)4 (120)SMD 1.07(−0.07 to 2.21)87% (\<0.001)1.164SMD 1.08(−0.25 to 2.41)LowMRAs1 (20)SMD −0.24(−0.86 to 0.39)NANASMD −0.24(−3.11 to 2.63)1 (20)SMD −0.24(−0.86 to 0.39)NANASMD −0.24(−2.83 to 2.35)Insufficient**Incidence of Anuria**ACEIs1 (60)OR 0.58(0.36 to 0.94)NANAOR 0.62(0.41 to 0.95)4 (1,265)OR 0.69(0.57 to 0.83)0.0% (0.436)\<0.001OR 0.69(0.57 to 0.83)LowARBs2 (76)OR 0.89(0.45 to 1.73)0.0% (0.903)\<0.001OR 0.77(0.46 to 1.29)2 (76)OR 0.89(0.45 to 1.73)0.0% (0.724)\<0.001OR 0.81(0.51 to 1.31)LowMixed ACEIs/ARBsNANANANANA2 (1,338)OR 0.88(0.75 to 1.03)0.0% (0.409)\<0.001OR 0.88(0.75 to 1.03)Insufficient**Change in D/P Cr Ratio**ARBsNANANANANA2 (28)SMD 0.04(−0.48 to 0.57)0.0% (0.957)\<0.001SMD 0.04(−0.48 to 0.57)InsufficientMixed ACEIs/ARBsNANANANANA1 (66)SMD −1.60(−2.16 to −1.04)NANASMD −1.60(−2.16 to −1.04)Insufficient**Acceptability of Treatment**ACEIs3 (131)OR 1.57(0.52 to 4.71)26.4% (0.257)0.266OR 1.49(0.59 to 3.80)4 (185)OR 0.93(0.26 to 3.26)59.0% (0.062)0.897OR 0.93(0.37 to 2.37)LowARBs3 (116)OR 1.12(0.23 to 5.43)0.0% (0.831)\<0.001OR 1.21(0.29 to 5.09)6 (151)OR 1.08(0.29 to 3.99)0.0% (0.996)\<0.001OR 1.06(0.29 to 3.84)LowMRAs2 (178)OR 1.46(0.75 to 2.84)0.0% (0.850)\<0.001OR 1.45(0.59 to 3.57)2 (178)OR 1.46(0.75 to 2.84)0.0% (0.850)\<0.001OR 1.42(0.40 to 5.05)Low^a^Summary of treatment effects compared with active control.^b^Number of studies with direct comparison.^c^The *τ*^2^ values in the network analyses from RCTs were: change in rGFR, 0.153 (moderate heterogeneity); change in urine volume, 2.043 (high heterogeneity); incidence of anuria, \< 0.001 (low heterogeneity); change in D/P Cr ratio (NA); acceptability of treatment, 0.104 (moderate heterogeneity).^d^The *τ*^2^ values in the network analyses from RCTs and non-randomised studies were: change in rGFR, 0.061 (moderate heterogeneity); change in urine volume, 1.647 (high heterogeneity); incidence of anuria, \< 0.001 (low heterogeneity); change in D/P Cr ratio (NA); acceptability of treatment, 0.340 (high heterogeneity).Abbreviations: ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; CI, confidence interval; D/P Cr, dialysate-to-plasma creatinine; MRAs, mineralocorticoid receptor antagonists; NA, not applicable; NMA, network meta-analysis; OR, odds ratio; rGFR, residual glomerular filtration rate; RCTs, randomised-controlled trials; SMD, standardised mean difference.Figure 3Network meta-analysis of RAAS blockade compared with active control for primary outcomes. Abbreviations: ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; CI, confidence interval; D/P Cr, dialysate-to-plasma creatinine; MRAs, mineralocorticoid receptor antagonists; NA, not applicable; OR, odd ratio; RAAS, renin-angiotensin-aldosterone system blockade; RCTs, randomised-controlled trials; rGFR, residual glomerular filtration rate; SMD, standardised mean difference.
For change in rGFR, ACEIs or ARBs was substantially more efficacious than active control with medium treatment effects (standardised mean difference \[SMD\] range 0.45 to 0.55; Table [2](#Tab2){ref-type="table"} and Fig. [3](#Fig3){ref-type="fig"}). Interestingly, the NMA resulted that the protective effect of RAAS blockade on preservation of rGFR was found if duration of treatment was ≥12 months (SMD range 0.41 to 0.76; Fig. [4](#Fig4){ref-type="fig"}). No significant difference among treatment comparisons for change in urine volume was observed. Moreover, only ACEIs was associated with a significantly decreased risk of anuria compared with active controls. The odd ratios (ORs) from NMA of RCTs alone and RCTs/non-randomised studies together were 0.62 (95% confidence interval \[CI\], 0.41 to 0.95) and 0.69 (95% CI, 0.57 to 0.83), respectively (Table [2](#Tab2){ref-type="table"} and Fig. [3](#Fig3){ref-type="fig"}).Figure 4Mean change in rGFR by duration of treatment: evidence from NMA (RCTs and non-randomised studies). Note: Bold values indicate statistical significance. For study duration \<6 or \<12 months, SMDs \>0 indicate that the treatment specified in the row is more efficacious than that in the column. For study duration ≥6 to \<12 or ≥12 months, SMDs \>0 indicate that the treatment specified in the column is more efficacious than that in the row column. To obtain SMDs for comparisons in the opposite direction, positive values should be converted into negative values, and vice versa. Abbreviations: ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; CIs, confidence intervals; NMA, network meta-analysis; RCTs, randomised-controlled trials; rGFR, residual glomerular filtration rate; SMDs, standardised mean differences.
Owing to limited data, a dose-response relationship and several preplanned subgroup analyses could not be performed. However, our subgroup analyses showed no effect of RAAS blockade on change in rGFR from baseline among study size ≤ 50 participants or non-Asian countries (Supplementary, Tables [S7](#MOESM1){ref-type="media"} and [S8](#MOESM1){ref-type="media"}). Meanwhile, there was no effect of study size or location on the main findings of change in urine volume and incidence of anuria.
Peritoneal membrane function and ultrafiltration volume {#Sec6}
-------------------------------------------------------
Overall results of PMF and UF volume did not differ among pairwise meta-analyses (Supplementary, Tables [S5](#MOESM1){ref-type="media"} and [S6](#MOESM1){ref-type="media"}) and NMAs (Supplementary, Tables [S9](#MOESM1){ref-type="media"} and [S10](#MOESM1){ref-type="media"}). Nonetheless, effect estimates were mainly drawn from non-randomised studies, and very few data were available. Thus, it was impossible to establish dose- and duration-response effects or subgroup analyses. Compared with active controls, mixed ACEIs/ARBs were associated with a statistically significant decrease in D/P Cr ratio (SMD −1.60; 95% CI \[−2.16 to −1.04\]; Table [2](#Tab2){ref-type="table"} and Fig. [3](#Fig3){ref-type="fig"}), whereas increase the 4-hour UF by the peritoneal equilibration test (SMD 1.36; 95% CI \[0.02 to 2.70\]; Supplementary, Table [S9](#MOESM1){ref-type="media"}).
Acceptability of treatment and safety outcomes {#Sec7}
----------------------------------------------
Both pairwise meta-analysis and NMA of primary analysis (Table [2](#Tab2){ref-type="table"} and Fig. [3](#Fig3){ref-type="fig"}; Supplementary, Tables [S5](#MOESM1){ref-type="media"} and [S6](#MOESM1){ref-type="media"}) and subgroup analyses (Supplementary, Tables [S7](#MOESM1){ref-type="media"} and [S8](#MOESM1){ref-type="media"}) did not differ between treatment comparisons. Likewise, no associations with the occurrence of adverse events were found (Supplementary, Tables [S9](#MOESM1){ref-type="media"} and [S10](#MOESM1){ref-type="media"}). Nevertheless, only one study^[@CR14]^ revealed that MRAs with spironolactone increased the risk of gynaecomastia (OR 6.40 \[1.37--29.92\], Supplementary, Table [S10](#MOESM1){ref-type="media"}), while no study reported safety outcome on angioedema/oedema or HRQOL.
Sensitivity analyses {#Sec8}
--------------------
Post-hoc analysis by adding conference abstracts could not be performed due to lack of additional relevant studies. The results for incidence of anuria, daily UF volume, and acceptability of treatment were robust and did not change substantially in sensitivity analyses (Supplementary, Tables [S11](#MOESM1){ref-type="media"}--[S13](#MOESM1){ref-type="media"}).
For change in rGFR, there was no protective effect of treatment comparisons after the analysis was restricted to studies from non-mainland China, as well as when studies with mixed ACEIs/ARBs were removed, except for ARBs which was not different when excluding mixed treatment effects (Supplementary, Table [S14](#MOESM1){ref-type="media"}). Notably, the positive association between ARBs and change in urine volume was found when restricted analysis by excluding studies from mainland China was performed (SMD 3.19 \[2.51--3.87\], Supplementary, Table [S12](#MOESM1){ref-type="media"}). Exclusion of crossover studies did not result in substantial differences for change in D/P Cr ratio and 4-hour UF volume, while removal of these studies resulted in ACEIs being associated with a significantly increased risk of dry cough (Supplementary, Table [S13](#MOESM1){ref-type="media"}).
Assessment of heterogeneity, inconsistency, transitivity, and publication bias {#Sec9}
------------------------------------------------------------------------------
For NMAs, most outcomes were associated with low to moderate statistically heterogeneity, except for change in urine volume, acceptability of treatment, and dry cough which were associated with high heterogeneity. Network meta-regression analyses could not be performed due to limited data availability. However, univariate meta-regression of pairwise analyses revealed that the heterogeneity of studies was not accounted by any of the study- or baseline participant-level characteristics for change in rGFR, urine volume, incidence of anuria or acceptability of treatment (Supplementary, Table [S15](#MOESM1){ref-type="media"}).
There was no evidence of inconsistency between direct and indirect evidence based upon findings from the loop-specific and node-splitting methods (Supplementary, Tables [S16](#MOESM1){ref-type="media"} and [S17](#MOESM1){ref-type="media"}), however, inconsistency for change in D/P Cr ratio and 4-hour UF volume were identified by the design-by-treatment interaction approach (Supplementary, Table [S18](#MOESM1){ref-type="media"}). Analysis of comparison-adjusted funnel plots indicated no evidence of asymmetry, except for change in urine volume (Supplementary, Fig. [S4](#MOESM1){ref-type="media"}).
Strength of the body of evidence {#Sec10}
--------------------------------
We graded the strength of evidence of ACEIs or ARBs for the preservation of RKF (rGFR and anuria) as low, whereas several outcomes were graded as insufficient due to study limitations, inconsistency, and imprecision of effect estimates (Table [2](#Tab2){ref-type="table"}). We therefore ranked RAAS blockade with regard to rGFR and incidence of anuria as two dimensions (Fig. [5](#Fig5){ref-type="fig"}). The comparative effects of RAAS blockade according to the efficacy on preserve RKF and acceptability of treatment are provided in Supplementary, Fig. [S5](#MOESM1){ref-type="media"}. However, effects of RAAS blockade on PMF could not be established due to limited evidence. Details of evidence synthesis by the Grading of Recommended Assessment, Development and Evaluation (GRADE) system are described in Supplementary, Table [S19](#MOESM1){ref-type="media"}.Figure 5Two-dimension rank plot of effect estimates according to efficacy on preservation of rGFR and incidence of anuria. Abbreviations: ACEIs, angiotensin-converting enzyme inhibitors; ARBs, angiotensin II receptor blockers; rGFR, residual glomerular filtration rate; SUCRA, surface under the cumulative ranking curve.
Discussion {#Sec11}
==========
Our NMA sheds unified hierarchies of evidence for all RAAS inhibitors, only ACEIs and ARBs treatments were efficacious than active control and ranked as the highest level of efficacy for the prevention of anuria and preservation of rGFR, respectively. Effects of RAAS blockade on PMF were inconclusive likely based on few studies examining PMF and its constituents as an outcome. No specific RAAS blockade classes were superior to other treatments with regard to adverse outcomes. However, it should be noted that our findings are limited by the strength of evidence according to the GRADE system revealed low or insufficient quality evidence.
Compared with existing reviews, our study has important methodological differences. This review incorporated of both RCTs and non-randomised studies to address some limitations of RCT, such as small sample size and short follow-up period, allow assessment of treatment comparisons simultaneously, and generalisable evidence^[@CR40]^. A review by Akbari *et al*.^[@CR29]^, demonstrated a protective effect of treatment with ACEIs/ARBs on preservation of rGFR at 12 months (mean difference \[MD\] 0.91 \[0.14--1.68\] mL/min/1.73 m^2^, and in line Zhang *et al*.^[@CR28]^, (MD 1.11 \[0.38--1.83\] mL/min/1.73 m^2^ for treatment with ARBs). Given our findings, the superiority of ACEIs or ARBs is similar to the pattern seen in pairwise meta-analyses. Long-term use (≥12 months) of ACEIs and ARBs shown substantial benefit for change in rGFR (SMD 0.76 \[0.34--1.18\] and 0.70 \[0.34--1.06\] mL/min/1.73 m2, respectively), and prove more clinically significant benefit compared with treatment \<12 months.
From the methodological viewpoint and based on the natural disease progression, the length of follow-up time is important to provide adequate statistical power. Besides a relatively small number of patients, it can be postulated that the short duration of treatment \<12 months was ultimately underpowered to observe the intended effects of the RAAS blockade. Additionally, although the mechanisms of action of the RAAS blockade effects have not been clearly described in PD patients, delayed therapeutic effects of the use of RAAS blockade has been apparently illustrated in the certain population such as diabetic nephropathy and those with dietary protein restriction---presumably because of the hemodynamic effect of the inhibition of the RAAS activation^[@CR41]--[@CR43]^. In this circumstance, it may further explain that no significant association of the RAAS blockade on the rGFR was observed when a duration of treatment was \<12 months. However, adequate powered controlled trials are warranted to confirm our findings in terms of duration-response effects.
Indeed, the urine volume is variable among the dialysis population, ranging from oliguria to normal or even above normal levels. These results are related to the circumstance that the urine output is influenced not only by the GFR, but also by the difference between the GFR and the rate of tubular transport mechanisms. In the PD population, urine volume accounted for only one half of the variance in GFR (r^2^ = 0.55, *P* \< 0.001)^[@CR37]^. Thereby, we hypothesised that the effect of RAAS blockade on the preserve of urine volume does not always in line with the rGFR. However, it is unclear whether or not the use of RAAS blockade among PD patients may have an additional effect on a slower rate of decline of urine volume. Owing to the statistical power was limited and the imprecision of this measure, including the errors and reliability of urine collection. Moreover, the pooled SMDs in our analyses were calculated using the median, range and/or interquartile range, which could be limited by the sample size of the included studies. Thus, caution must be considered in an interpretation of the findings.
Long-term PD treatment can lead to peritoneum injury by hypertonic PD solutions. Theoretically, RAAS blockade may have important role in "peritoneal protection", as RAAS may reduce membrane fibrosis and neoangiogenesis, and decreased peritoneal protein loss that occurs over time^[@CR16]--[@CR18],[@CR24]^. However, we found that the data for these outcomes were substantially inconsistent, which suggest less reliability of the findings. We conclude that the effects of RAAS blockade on peritoneal membrane integrity remain unclear and warrant further clarification. Both pairwise and NMA had similar adverse outcomes across all other treatments, however, we underscore that the results were surrounded by uncertainty of effects estimated.
To our knowledge, this systematic review and NMA consisted of unpublished data, and more comprehensive than previous reviews as we did not impose any language restrictions. Our study was conducted by retrieving evidence from both RCTs and non-randomised studies, which reflect real-world practices.
However, several potential limitations of our study are worth noting. Firstly, even sophisticated adjusted cohort studies may be subjects to unmeasured confounders^[@CR40]^. Moreover, the limited interconnections in the evidence networks studies should be stated that an indirect estimate is susceptible to confounding. Thus, findings should be interpreted with caution since it does not always agree with respect to direct evidences.
Secondly, heterogeneity of outcomes definition is an important concern because of measurement and definition of endpoint were not standardised and defined poorly across included studies. Given the low rate of reporting, measurement of urine volume as well as the incidence of anuria may have been under-ascertained if patients stopped collecting their urine when anuria is near or reached, particularly in the non-randomised studies. We therefore suggest that future trials with high methodological quality are needed.
Thirdly, the interpretation of our subgroup analyses was limited by small sample size and number of studies. Moreover, details of clinically relevant characteristic such as peritoneal equilibration test status, UF, and history of peritonitis were not fully reported across included studies. As such, several preplanned subgroup analyses cannot be performed to explore treatment effects in different subpopulations.
Fourthly, based on crossover design, carry-over effect may contribute to effect estimates of included studies which could have biased results towards a neutral effect on some outcomes, and against specific treatment comparisons. No further significant association was observed between treatment comparisons and change in D/P Cr ratio and 4-hour UF volume, whereas risk of dry cough substantially increases among ACEI users (OR 17.83 \[1.22--261.44\], Supplementary, Table [S13](#MOESM1){ref-type="media"}) after removing the crossover design.
Fifthly, the majority of included studies were conducted in Asian countries and restricted largely to CAPD patients. This might affect our findings by limiting the generalisability to other ethnic/racial groups and individuals PD who treated with other PD modalities.
Lastly, because of primary data for the effects of MRAs and DRIs on RKF and PMF were scant, our review was unable to make any meaning conclusions regarding the comparative effectiveness and safety of these drugs among PD patients.
Our findings provide the evidence to support the use of ACEIs or ARBs among PD patients, which in line with the ISPD guidelines for assessment and management of RKF as a cardiovascular risk factor^[@CR30]^. Besides cardiovascular benefits, we therefore advise the long-term use (≥12 months) of ACEIs or ARBs as an essential treatment compared with other antihypertensive, and this may improve mortality in these population as well. Thus, it is reasonable to suggest that clinician should prescribe ACEIs or ARBs in adult PD patients with significant RKF.
Currently, no recent controlled trial provides direct comparison between MRAs and ACEIs or ARBs treatment among PD patients, thus future trials with high methodological quality are needed. To understand the complex relationship between peritoneum and the kidney, clarifying the effects of each other is required. Besides systemic RAAS blockade, additional experimental study is required to determine effective strategies to decrease local peritoneal RAAS activation. In addition, dose- and duration response relationship of RAAS blockade for these outcomes also needs further investigation.
In summary, our analysis reveals that ACEIs or ARBs treatment is the most effective strategies for preserving RKF, and has similar clinical adverse outcomes across all other treatments. However, little evidence is available for effects of RAAS blockade on PMF in adults PD patients.
Methods {#Sec12}
=======
This study was performed in accordance with guidelines for comparative effectiveness reviews produced by the Agency for Healthcare Research and Quality^[@CR44]^, and has been reported in line with guidance from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension statement for NMA (Supplementary, Appendix-[I](#MOESM1){ref-type="media"})^[@CR45]^. The pre-specified protocol was registered in PROSPERO (CRD42018083525). Details of the methodology used for this review are described in Supplementary, [eMethods](#MOESM1){ref-type="media"}.
Data sources and search strategy {#Sec13}
--------------------------------
In brief, we searched electronic databases, including PubMed, Embase, Scopus, Web of Science, CINAHL, the Cochrane Library, and grey literature from inception to May 31, 2019, without language restrictions (search strategies are described in Supplementary, Appendix-[II](#MOESM1){ref-type="media"}). Two independent reviewers screened eligible titles/abstracts and relevant full-text studies. Any disagreement was resolved through a team discussion.
Study selection and outcomes {#Sec14}
----------------------------
We included both RCTs (parallel and cross-over trials) and non-randomised studies (quasi-experimental and retrospective/prospective cohort studies with a control group) that: (i) included adult participants aged ≥18 years; (ii) addressed the effects of the RAAS blockade with ACEIs, ARBs, DRIs, and MRAs and reported at least one of the outcomes of interest; and (iii) compared RAAS blockade against placebo/active control with follow-up of one week onward. We excluded studies that: (i) were N-of-one, case-control, cross-sectional, case series/case reports, and phase I or II study design; (ii) used a dual ACEIs and ARBs treatment; (iii) compared intraperitoneal administered of intervention/control groups; and (iv) included participants that received both PD and haemodialysis treatments.
The primary outcomes of interest were key parameters on RKF and PMF, including (i) rGFR, on the basis of 24-hour collection and calculated using the valid equation proposed (mL/min or mL/min/1.73 m^2^); (ii) urine volume, using a 24-hour timed urine collection or self-reported (mL/day); (iii) the incidence of anuria (defined according to each study, commonly \<200 mL/day); (iv) ratio of D/P Cr by the peritoneal equilibration test; and (v) acceptability of treatment (defined to be study dropout for any reason).
The secondary outcomes of interest were: (i) 4-hour UF volume by the peritoneal equilibration test (mL/4-hour); (ii) daily UF volume (mL/day); (iii) adverse events, including hyperkalaemia (as defined by study authors, commonly \>5.5 mmol/L), dry cough, hypotension, dizziness, angioedema/oedema, gynaecomastia, and peritonitis; (iv) HRQOL; and (v) healthcare expenditure (e.g. costs, hospitalisation).
Data extraction and risk of bias assessment {#Sec15}
-------------------------------------------
Two reviewers independently extracted the following data using a standardised extraction form: study- and participant-characteristics, interventions (specific RAAS blockade classes \[for studies that cannot be specified the class of RAAS blockade, we reported separately as mixed treatment\]), control groups, dose of drug, concomitant medications, and predefined outcomes. Data extraction was cross-checked and any discrepancies were resolved through discussion with a third reviewer. For studies with missing or incomplete data, the study authors were contacted by email for clarification. Two reviewers independently performed a critical appraisal of the risk of bias using the Cochrane risk of bias assessment tool for RCTs/quasi-experimental^[@CR46]^ and the NOS for cohort studies^[@CR47]^.
Data synthesis and analysis {#Sec16}
---------------------------
Only full-text articles were considered in the primary analysis. We employed a two-step approach to the performance of both pairwise meta-analysis and NMA. For the first step, only RCTs data were used. Subsequently, data from both RCTs and non-randomised studies were incorporated. For any continuous outcome parameters, the summary results from individual studies were captured as the mean difference in the changes with the formula: ∆value~change~ = value~endpoint~ -- value~baseline~, in which SD^2^~change~ = \[SD^2^~baseline~ + SD^2^~endpoint~ -- (2 × ρ × SD~baseline~ × SD~endpoint~)\], where ρ stands for the correlation coefficient. We anticipated ρ value of 0.75 between the baseline and endpoint values and equal variances during the RAAS blockade and control groups. Additionally, an estimated value of 0.56 was considered in a sensitivity analysis as described by Szeto *et al*.^[@CR38]^. For meaningful interpretation of treatment effects, a mean difference of 0.2, 0.5, and 0.8 are considered to be small, medium, and large, respectively^[@CR48]^.
A random-effects NMA was performed to compare indirect treatment effects for each outcome using a frequentist model. As a measurement varied across studies, the findings from NMAs were expressed in terms of SMDs for continuous outcomes and ORs for dichotomous outcomes, along with 95% CIs. The surface under the cumulative ranking curve, a relative ranking probability, was also estimated for each RAAS blockade class, and rankograms were also prepared^[@CR49]^.
We evaluated characteristics of populations and study designs across all included studies to evaluate the assumption^[@CR49],[@CR50]^. The network heterogeneity was assumed as the *τ*^2^ statistics, a between-study variance^[@CR51]--[@CR53]^. Assessment for consistency of direct and indirect effects was performed using three different approaches: the loop-specific approach, the node-splitting approach, and the design-by-treatment interaction model^[@CR49],[@CR50],[@CR54]^. Comparison-adjusted funnel plots were used to investigate for reporting bias and potential effects of small studies^[@CR55]^.
Preplanned subgroup analyses and meta-regressions were carried out to address potential sources of heterogeneity, including study sample size (≤50 vs. \>50 participants), geographical region (Asian vs. non-Asian countries), age (\<65 vs 65 years), sex (as reflected by % female), baseline rGFR, urine volume, D/P Cr, blood pressure, history of diabetes, and PD modality.
A series of sensitivity analyses were performed to assess the robustness of primary findings. These included: (i) assuming the correlation coefficient of 0.56 for estimating the SD~change~^[@CR38]^; (ii) excluding studies from mainland China to avoid a systematic bias related to the trustworthiness as described elsewhere^[@CR56],[@CR57]^; (iii) excluding crossover studies to avoid carry-over effects^[@CR58],[@CR59]^; (iv) excluding studies which classified as mixed treatment intervention; and (v) post-hoc analysis by adding unpublished conference abstracts.
Statistical significance for all tests was 2-tailed, with *P* \< 0.05. All analyses were performed using STATA version 14.0 (StataCorp LP).
Grading the strength of evidence {#Sec17}
--------------------------------
To assess the strength of evidence for each outcome, the GRADE approach was performed and classified into insufficient-, low-, moderate-, or high-quality evidence^[@CR60],[@CR61]^. The confidence in estimates could be downgraded or upgraded according to the risk of bias, imprecision, inconsistency, and indirectness. Any disputes were resolved by a third reviewer.
Supplementary information
=========================
{#Sec18}
Supplementary Information
**Publisher's note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
These authors contributed equally: Surapon Nochaiwong and Chidchanok Ruengorn.
Supplementary information
=========================
is available for this paper at 10.1038/s41598-019-55561-5.
This work was supported by grant from the Faculty of Pharmacy, Chiang Mai University. No funding source had any role in the study concept and design, data collection, data analysis and interpretation, writing the report, or in the decision to submit for publication. The authors thank Pajaree Mongkhon, School of Pharmaceutical Sciences, University of Phayao, Phayao, Thailand., who helped develop search strategies.
Study concept and design: S Phatthanasobhon, S.N. and C.R.; Acquisition, analysis or interpretation of data: All authors; Drafting of the manuscript: S Phatthanasobhon, S.N., K.T. and C.R.; Critical revision of the manuscript for important intellectual content: S Panyathong, Y.S., B.H., M.S. and K.N.; Statistical analysis: S Phatthanasobhon, S.N. and C.R.; Administrative, technical, or material support: S.N. and K.T.; Study supervision: S.N. and C.R. Each author contributed important intellectual content during manuscript drafting or revision and take responsibility for the integrity of the data and the accuracy of the data analysis.
All data generated and analysed in this review were drawn from the existing article and reported along with the Supplementary Materials.
The authors declare no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Pulmonary arterial hypertension (PAH) is a progressive disease characterized by an elevation of pulmonary artery pressure and pulmonary vascular resistance, leading to right ventricular failure and death. Idiopathic PAH (IPAH; formerly termed primary pulmonary hypertension) occurs in the absence of known causes. Estimates of the incidence of IPAH and familial PAH (FPAH) range from 1-2 cases per million people in the general population, with at least 6% of these patients having FPAH. Although the incidence of PAH in patients with other illnesses is not known with certainty, from various reports it appears that 2-4% of patients with portal hypertension and 0.1-0.6% of HIV patients have PAH. The incidence of PAH that occurs in patients with connective tissue disease is extremely variable; prevalence ranges from 2 to 35% in patients with the scleroderma spectrum of disease and may reach as high as 50% of patients with limited scleroderma. PAH has also been reported to occur in 10-45% of patients with mixed connective tissue disease and in 1-14% of cases with systemic lupus erythematosus. The incidence of PAH associated with anorexigens is cyclical in nature and varies depending on the availability of specific appetite suppressants. The link was first identified in the 1960s when an epidemic of PAH occurred in Switzerland, Austria and Germany that was linked to the anorexigen aminorex fumarate. Use of the anorexigens fenfluramine and dexfenfluramine have also been linked with an increased risk for PAH.
Prior to the development of disease-specific targeted PAH therapies, the median survival for subjects diagnosed with IPAH was approximately 2.8 years. However, 2.8 years likely underestimates current survival as the course of the disease has been favorably altered by therapeutic advances since that report from the 1980s. Prognosis is also dependent on the underlying etiology of the disease. The prognosis for patients with PAH associated with connective tissue disease appears to be worse than for those with IPAH. Estimates for 2-year survival in scleroderma patients with associated PAH are 40% compared with 48% for 3-year survival in patients with IPAH. Survival in patients with HIV-associated PAH is similar to patients with IPAH. With current HIV therapies, most of the deaths in patients with HIV and associated PAH are now attributed to PAH.
Although Ernst von Romberg, a German physician, described an autopsy in 1891 as 'pulmonary vascular sclerosis,' it is only since 1995 with the introduction of intravenous epoprostenol that disease-specific targeted medical therapies for PAH have become available. In addition, significant advances in the treatment of PAH have occurred during the past decade, with six medical therapies now having received regulatory approval worldwide targeting the prostacyclin pathway, the nitric oxide pathway and the endothelin pathway \[[Figure 1](#F0001){ref-type="fig"}\]. Furthermore, ongoing clinical trials are evaluating novel therapeutic approaches based on scientific insights gleaned over the past decade in the pathobiology of PAH \[[Figure 2](#F0002){ref-type="fig"}\].
{#F0001}
{#F0002}
From a therapeutic standpoint, why had it taken from 1891 until 1995 to develop a safe and efficacious therapeutic modality for the treatment of PAH? \[[Figure 3](#F0003){ref-type="fig"}\] Although several reports of young women dying of right heart failure without a diagnosis were published in 1940, it was not until pulmonary artery pressures could be recorded directly with the introduction of right heart catheterization that the physiology of the pulmonary circulation could be studied. In 1951, Dresdale tested the acute effects of tolzoline in a young woman with IPAH; the tolzoline caused a sudden decrease in pulmonary artery pressure and pulmonary vascular resistance without significant systemic effects. Unfortunately, no drugs were available at that time for chronic treatment. However, despite this, there remained little interest in PAH until the epidemic of the aminorex-induced PAH became apparent in the late 1960s. Prompted by the aminorex-induced PAH epidemic in 1973, the World Health Organization (WHO) held its first meeting in Geneva to assess what was known about IPAH and what remained unknown. In 1981, the National Heart, Lung and Blood Institute of the National Institutes of Health supported a national registry of patients with IPAH, which resulted in several reports over the next decade describing clinical features of IPAH and its natural history. Interestingly, despite the fact that IPAH was an orphan disease, significant interest from the scientific community rapidly ensued. Advances in the understanding of the mechanisms involved in the pathobiology of IPAH and PAH associated with other conditions have focused on molecular biology, developmental biology and genetics. Together with epidemiological and natural history studies, collaborative efforts between the scientific community and industry have led to a surge in clinical trials over the past decade: since the approval of intravenous epoprostenol for the treatment of IPAH in 1995, the prostacyclin analogue treprostinil has been approved for continuous subcutaneous infusion in 2002 and for continuous intravenous infusion in 2004. In addition, the prostacyclin analogue iloprost was approved in 2004 via inhalation. In 2001, bosentan, an endothelin ET ~A~/ET~B~ receptor antagonist, was the first oral therapy approved for the treatment of PAH; and sildenafil citrate, an oral phosphodiesterase type 5 inhibitor, was approved in 2005. In 2007, the oral ET~A~ selective ERA ambrisentan was approved, and the oral ET~A~ selective ERA sitaxsentan was approved in the EU.
{#F0003}
Prompted by the scientific insights from the 1990s, in 1998 the second WHO meeting was held on the 25^th^ anniversary of the original meeting; and with the dramatic advances over the next 5 years, the 3^rd^ WHO Symposium on PAH was held in 2003 and the 4^th^ World Symposium on PAH in 2008. Based on the clinical trials to date, current consensus evidence-based guidelines for the treatment of PAH are shown in \[[Figure 4](#F0004){ref-type="fig"}\]. What have we been able to achieve? The disease-specific PAH therapies, currently available in conjunction with anticoagulant, diuretic, digitalis and oxygen therapy, have improved exercise capacity, functional capacity, time to clinical worsening, hemodynamic parameters, overall quality of life and survival. However, PAH remains a devastating, life-threatening disorder. In more than 50% of patients, exercise capacity remains significantly limited, approximately 50% of patients remain WHO functional class III or IV, PAH patients continue to have frequent hospitalizations for PAH, right heart function remains significantly impaired in most patients, quality of life is suboptimal and despite an increase in survival for functional class III and IV patients with IPAH from a predicted survival of 33% (based on the NIH Registry) to 63% with our current therapeutic modalities, the outlook is far from ideal; we need to continue to aggressively pursue furthering our understanding of PAH if we ever hope to give these patients a near-normal life.
{#F0004}
We believe that future developments in vascular biology will improve our understanding of the pathobiology of PAH and provide rationale and 'proof of concept' for more disease-specific targeted therapies. With the advent of genomic technologies and methods, the necessary tools are now becoming available to begin pinpointing the genes that contribute to disease susceptibility and progression. Candidate gene discovery, that is, gene analysis using microarrays, can identify genes that may provide valuable insight into disease biology and may represent an initial step towards the identification of genetic polymorphisms that may help predict efficacy or lack thereof, with various disease-specific targeted PAH therapeutic modalities. By identifying the genes and gene variants that determine individual disease susceptibility, we might be able to one day identify patients in pre-clinical stages of disease as well as allow for individualized therapies that are most efficacious and least likely to cause side effects. Furthermore, although right ventricular function appears to be the most significant prognostic parameter in PAH, comparatively little attention has been devoted to how right ventricular function and dysfunction can be detected and measured, what specific molecular and cellular mechanisms contribute to maintenance or failure of right ventricular function, how right ventricular dysfunction evolves structurally and functionally or what interventions might best preserve right ventricular function. In addition, right ventricular-left ventricular interaction and right ventricular-pulmonary arterial coupling have largely been overlooked as potential targets for investigation and therapy. Whether cell-based or gene therapy, in addition to new drugs or new combinations of existing drugs targeting right heart failure, in conjunction with PAH-specific vasodilator and antiproliferative drugs will improve outcomes in PAH will require further study. Ultimately, as these novel therapeutic options are developed, individualized treatment regimens will evolve. However, many questions remain regarding the treatment of patients with PAH, e.g., identification of patient populations who will most benefit from a specific therapy, determining when treatment should be initiated and establishing optimal drug sequencing and combinations. We hope that by further increasing our understanding of the pathobiology of PAH, we will one day be able to prevent and cure this disease.
However, in the interim, it is imperative that we base our treatment regimens on evidence-based studies. As stated by Hippocrates in Precepts (∼440 B.C.E.), "In Medicine one must pay attention not to plausible theorizing but to experience and reason together... I agree that theorizing is to be approved, provided that it is based on facts and systematically makes its deductions from what is observed... But conclusions drawn from unaided reason can hardly be serviceable; only those drawn from observed fact."
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .280}
{#sp2 .281}
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction
===============
The National Institute of Standards and Technology (NIST), Systems and Software Division, sponsored a Users' Forum on the Application Portability Profile (APP) and Open System Environment (OSE) at NIST in May. This forum was the fifteenth in a continuing semiannual series on the NIST APP and its application to OSE. The APP Users' Forums are designed to provide users and providers with the opportunity to exchange information and respond to NIST proposals regarding the evaluation and adoption of an integrated set of standards to support the APP and OSE.
The forum offered the customary presentation of standards and activities in the APP, OSE, Institute of Electrical and Electronic Engineers (IEEE), and Joint Technical Committee 1 (JTC1-international activities). A workshop on Automated Testing Technologies was featured on the second day with extensive discussions concerning participants' case studies, current activities, plans and lessons learned. A tutorial for beginners with little or no experience with the APP and OSE was held on the morning of the first day. The tutorial presented basic OSE concepts and the reference model.
The next APP/OSE Users' Forums will be held May 7 and 8, 1996 at NIST.
The APP/OSE Users' Forum has been developed to assist federal agencies with information technology (IT) issues. Central to this assistance is publication and maintenance of a technical guidance document, the Application Portability Profile (APP), facilitating the migration to open systems. An Open System Environment encompasses the functionality needed to provide interoperability, portability, and scalability of computerized applications across networks of heterogeneous, multi-vendor hardware/software/communications platforms. The APP integrates industry, federal, national, international, and other specifications into a Federal application profile to provide the functionality necessary to accommodate a broad range of Federal information technology requirements. The Application Portability Profile (APP), The U.S. Government's Open System Environment Profile OSE/1 Version 3.0 provides recommendations on a variety of specifications that will generally fit the requirements of U.S. Government systems. A specific organization will not necessarily require all of the recommended specifications in the APP. As the U.S. Government's OSE profile, this guidance is provided to assist Federal agencies in making informed choices regarding the selection and use of OSE specifications, and in the development of more selective application profiles based on the APP. It is directed toward managers and project leaders who have the responsibilities of acquiring, developing, and maintaining information systems supported by heterogeneous application platform environments.
2. Standards Status
===================
Fritz Schulz, NIST, presented the following updates on the OSE standards activities of IEEE, JTC1 and the Computer Systems Laboratory (CSL) of NIST.
The IEEE Portable Application Steering Committee (PASC), which sponsors the Portable Operating System Interface (POSIX) projects has reorganized. Previously each standard activity had been individually numbered, and now activities are grouped into seven areas. These areas are system services, shells and utilities, system administration, language bindings, security, profiles, and test methods. In addition to lowering overhead and increasing efficiency, this reorganization will make it easier to progress approved standards to the international arena.
The OSE guide developed by P1003.0 has been approved and will be published very soon as a technical guide. The POSIX OSE guide describes an OSE Reference Model (OSE/RM) that is closely aligned with the APP and that provides a framework for describing open system concepts and defining a lexicon of terms that can be agreed upon generally by all interested parties. The same document is in ballot as a draft technical report (DTR) 14252 within working group (WG)15 of subcommittee (SC)22 of JTC1. The DTR is also expected to be approved very soon. The status of individual programs within the POSIX project were distributed in a handout.
Technical Report 10000-3: "Information Technology---Framework and Taxonomy of International Standardized Profiles---Part 3: Principles and Taxonomy for Open System Environment Profiles," produced by the JTC1 Special Group on Functional Standardization (SGFS) has been approved and will be published very soon. TR 10000, part 3 provides a context for functional standardization in support of Open System Environments (OSE). It outlines the basic OSE objectives and concepts, and defines an approach to the taxonomy and format for OSE Profiles specified by International Standardized Profiles. The technical report gives guidance on the nature and content of International Standardized Profiles (ISPs) documents to organizations proposing Draft OSE ISPs.
2.1 Application Portability Profile (APP) Version 3
---------------------------------------------------
Gary Fisher, NIST, made the presentation on the new version of the APP.
A selected suite of specifications that defines the interfaces, services, protocols, and data formats for a particular class or domain of applications is called a profile. The Application Portability Profile (APP) integrates industry, Federal, national, international, and other specifications into a Federal application profile to provide the functionality necessary to accommodate a broad range of Federal information technology requirements.
The APP is *not* a standard and is not designed to cover every case. In some instances, the selection of one specification recommended in the APP will obviate the need for other specifications that are also recommended (i.e., select one or the other, but not both.) There is some overlap in functionality covered in different specifications. There are also gaps in functionality. In areas where the APP does not meet all of a user's requirements, the user must augment the recommended specifications to ensure that proposed systems built on these specifications meet organizational requirements. The APP is designed to help users determine which specifications to use.
Not only is the U.S. Government involved in the development of profiles, but industry, national, and international organizations are preparing specifications that encompass numerous types of profiles. Corporations such as American Airlines, Boeing, DuPont, General Electric, Kodak, McDonnell Douglas, Merck, Motorola, Northrop, and Unilever are developing profiles for use within their own organizations and in many cases have based these profiles on the APP. The Institute of Electrical and Electronics Engineers, the International Organization for Standardization, and other standards-making organizations are in the process of developing profiles for specific types of application domains. U.S. Government organizations that are engaging the concepts of organizational profiles include the U.S. Army Sustaining Base Information Services, the U.S. Bureau of the Census, the Internal Revenue Service, the Defense Information Systems Agency, and many others.
Many specifications were reviewed and evaluated before the final recommended specifications were selected. If there are other specifications that should be considered in the APP and that meet a broad range of U.S. Government application requirements, users, vendors, and other interested parties should formally recommend them for evaluation using the same evaluation criteria applied to the selected specifications. This is one of the ways in which the APP will continue to evolve as technology evolves.
The initial version of the APP was published by the National Institute of Standards and Technology (NIST) in April 1991 as Special Publication 500-187. Version 2 of the APP Guide, NIST Special Publication 500-210, was published in June 1993. The changes in this third revision reflect the evolutionary developments that have occurred in the standards arena. Examples of the types of changes in this version include the following: The introductory material incorporates work done by the Institute of Electrical and Electronics Engineers (IEEE) POSIX Working Group 1003.0 on the Open System Environment Reference Model (OSE/RM).The evaluation criterion, *de facto usage*, has been removed and others have been reworded to provide more usable definitions.A new *bindings* information item has been added to individual specifications where appropriate.All of the recommended specifications have been updated and many new ones have been added. Areas that have seen the most change are those that encompass data interchange and communications where numerous new specifications have been added.
Specific changes between Version 2 and Version 3 recommended specifications include the following: Operating System ServicesIEEE 1003.2-1992 POSIX Shells and Utilities is now FIPS 189.IEEE 1003.4 Realtime is now IEEE 1003.1b.IEEE 1003.6 Security is now IEEE 1003.1e and IEEE 1003.2c.IEEE P1387.2, .3, and .4 are new.Human/computer Interface ServicesProposed FIPS 158-1 X Window System is now officially FIPS 158-1.IEEE P1295 X Window Toolkit is now IEEE 1295.1.Software Engineering ServicesFIPS 119 Ada is now FIPS 119-1 Ada.FIPS 21-3 COBOL is now FIPS 21-4 COBOL.FIPS 119 Pascal has been deleted due to very limited interest in this specification.ECMA PCTE has been replaced by ISEE Repository ISO/IEC 13719-1.Data Management ServicesFIPS 127-1 SQL is now FIPS 127-2.FIPS 193 SQL Environments is new.Data Interchange ServicesODA/ODIF/ODL ISO 8613 has been deleted due to lack of implementations.Draft Portable Document Delivery Format (PDDF) is new.SPDL ISO 10180 has been deleted and replaced by PDDF.Standard Data Elements ISO 11179 Parts 3, 4, and 5 are new.FIPS 194 Raster is new.JPEG is new.MPEG is new.STEP ISO 10303 has been replaced by the planned FIPS on STEP.FIPS 173 SDTS is now FIPS 173-1.Graphics ServicesFIPS 153 PHIGS is now FIPS 153-1.Network ServicesPII API P1003.12 has been renamed P1003.1g.IEEE 1238.1 FTAM has been deleted. (This specification is part of FIPS 146-2.)FIPS 146-1 GOSIP is now FIPS 146-2 POSIT.ISDN is now FIPS 182 ISDN.IEEE 1003.8 TFA has been deleted. (This specification is part of FIPS 146-2.)CORBA is new.FIPS 179 GNMP has been deleted and replaced with OMNIPoint.FIPS 192 GILS is new.NISO Z39.50 is new.FIPS 46-2 DES is new.FIPS 186 DSS is new.
The universe of OSE is continually evolving and the APP Guide will strive to reflect this evolution. The Computer Systems Laboratory (CSL) welcomes any recommendations for changes to the APP.
2.2 Profiles for Open System Internetworking Technology (POSIT)
---------------------------------------------------------------
Tassos Nakassis, NIST, reported that the Secretary of Commerce recently approved two revised standards: FIPS 146-2, Profiles for Open Systems Internetworking Technologies (POSIT), and FIPS 179-1, Government Network Management Profile (GNMP). Effective immediately, FIPS 146-2 removes the requirement that federal agencies specify Government Open Systems Interconnection Profile (GOSIP) protocols when they acquire networking products and services and communications systems and services. FIPS 179-1 provides implementations for network management based on the service and protocol standards issued by the International Organization for Standardization (ISO). These revised standards promote the interoperability of computers and systems that are acquired from different manufacturers in an open systems environment.
2.3 Document Management Services
--------------------------------
Mike Rubinfeld, NIST presented the status of three standards used in document management, Joint Photographic Experts Group (JPEG), Moving Pictures Experts Group (MPEG) and Portable Document Delivery Format (PDDF). JPEG is being developed under the auspices of ISO/IEC JTC1/SC2 Working Group 10. The current standard, IS 10918:1992, specifies the digital compression and coding of continuous-tone still images. These images can be either grayscale or color. The standard uses 24 bit compression and consists of three elements, an encoder, a decoder, and the interchange format. ISO 10918:1992 uses other standards as well. They are SGML Z39.50, MPEG, Huffman Encoding and ISO/IEC IS9660.
ISO/IEC JTC1/SC2 Working Group 11 is the sponsor of the IS 11172:MPEG-1 standard. The standard is for video compression for multimedia applications. It addresses compression of video signals up to 1.5 Mbits/s. MPEG audio compresses the audio signal at rates of 64, 128 and 192 kbits/s. MPEG is used in conjunction with mass media such as hard drives, CD-ROM and other optical storage, writable CD, DAT tape, and network servers.
MPEG utilizes two techniques, blocked-based motion compression---reduction of temporal redundancy and transform domain-based compression---reduction of spacial redundancy (DCT).
PDDF is based on a blue ribbon panel's recommendations and a set of basic requirements for a standard. Final Form Portable Document Delivery Format consists of encoded representation on electronic medium in presentation quality final form. The current situation requires the use of proprietary formats resulting in conversion nightmares that often require resorting to ASCII as a common denominator. The PDDF project goals are: Identify Needs within the GovernmentDevelop a set of RequirementsAssess the Current TechnologyDescribe a PDDF that meets the RequirementsDevelop a Conformance Test Suite Based on the PDDFDraft a FIPS for the Preferred PDDF
To meet these goals, government user and vendor workshops were held with NIST serving as an overall catalyst, coordinator and initiator of cooperative research and development agreements (CRADAs) with vendors. NIST will also provide documents from workshops, develop a conformance test plan and consider PDDF as a future FIPS.
PDDF provides new way to preserve documents that will alleviate costs associated with conversion and use of unnecessary software. This will make the use of electronic medium for document exchange much easier. Storage cost and paper cost savings will be significant.
The baseline set of requirements for choosing a format was developed in the Open System Environment Implementors Workshop (OIW) from contributions by vendors and the Blue Ribbon Panel. A set of 19 requirements was established to serve as a guide for selecting a PDDF. The project will also address the following recommendations from the Blue Ribbon Panel: Conformance Verification---Provide for software conformance to the format specification via a conformance test plan and associated test suite. Provide a registry of conformant software products.Organize a Users PDDF Forum comprised of government users and industry developers.
2.4 SQL Standards and FIPS 127-2
--------------------------------
Joan Sullivan, NIST, gave the presentation on SQL and the associated FIPS. First introduced in 1986, FIPS 127 (SQL-86) addressed only basic functionality. In 1989, integrity enhancement was added resulting in FIPS 127-1 (SQL-89). 1992 saw the issuance of FIPS 127-2 (SQL-92), with a four level structure. The levels are entry, transitional, intermediate, and full. The next revision of FIPS 127-2 will be based on SQL-9x, which will consist of six major parts.
The SQL conformance testing began in 1988 with 191 tests growing to 384 by December of 1989. In April of 1990, NIST started a SQL trial testing service, and issued registered validation summary reports. The trial period ended in 1992, and testing certificates started being issued in 1993. Currently tests exists for FIPS 127-1 and two levels (entry and transitional) of FIPS 127-2. Additional levels of FIPS 127-2 will be available in 1996. The FIPS 127-1 validated product list contains 12 companies, offering 14 products. The list for FIPS 127-2 has six companies, offering 12 products. There is worldwide interest in SQL testing. The NIST test suite is licensed internationally in Australia, Belgium, Canada, China, France, Germany, Italy, Greece, Japan, Korea, Sweden, United Kingdom, and the USA.
To ensure portability of SQL programs, a simple FIPS 127-2 strategy is necessary. Users should specify FIPS 127-2 conformance in request for proposals (RFPs) and require a test certificate. On existing database products users should upgrade to validated products. Most importantly, they should educate development staff on standard SQL and enforce its use in application development.
2.5 Digital Encryption Standard (DES) and Digital Signature Standard (DSS)
--------------------------------------------------------------------------
Lisa Carnahan, NIST, presented the status of FIPS 46-2, DES, and FIPS 186, DSS. FIPS 46 was first issued in 1977 to protect unclassified information from unauthorized disclosure or modification. NIST reviews the standard every 5 years, and has reaffirmed it at its last review in 1993. As a result of that review, use of software implementations is now allowed in addition to hardware implementations. DES is documented and is validated in accordance with NIST SP 500-20. The validation test entails using a NIST supplied key and 64 bit input and then performing 8 million encryptions and 4 million decryptions.
FIPS 186, DSS, was issued in May of 1994. The standard contains an algorithm to use in designing and implementing public-key based signature systems. A companion FIPS 180-1 for a secure hash standard (SHS) was issued in April of 1995 for use when computing a condensed representation of a message or data file. Any change in the message will, with a high degree of probability, result in a different result. DSS conformance tests are modular, consisting of signature generation, signature verification, primality tests, global parameter generation (p,q,g), key generation (x,y), and per message parameter generation (k). All implementations must generate k (per message parameter) and sign or verify.
2.6 Standard for the Exchange of Product Model Data
---------------------------------------------------
A FIPS has been proposed for the Standard for the Exchange of Product Model Data (STEP) that will adopt the voluntary industry specification International Organization for Standardization (ISO) Product Data Representation and Exchange, ISO 10303:1994. STEP defines and describes all product data used during the manufacturing life-cycle of a product, the production steps needed to make and product, and the order in which they occur. Comments on this proposed standard are welcomed. The proposed FIPS is available from the CSL Office.
2.7 Standard Generalized Markup Language (SGML)
-----------------------------------------------
Ron Wilson, NIST reported on a task initiated by the CALS Project Office to organize an SGML Conformance Testing Service. The NIST SGML Conformance Testing Program will certify that SGML parsers meet the requirements of the Federal Information Processing Standard (FIPS) 152. The Computer Systems Laboratory of the National Institute of Standards and Technology (NIST) is responsible for establishing conformance testing programs for Federal Information Processing Standards (FIPS). In carrying out this responsibility, CSL specifies the necessary conformance test specifications, test methods (i.e., test suites, test tools, and technical procedures), validation procedures, and testing laboratories for testing product compliance to FIPS.
NISTIR 5538, SGML Parser Validation Procedures, establishes operating policy and procedures for the Computer Systems Laboratory's (CSL) validation program for Federal Information Processing Standards (FIPS) 152, Standard Generalized Markup Language (SGML) parsers. The testing methodology is based on ANSI X3.190-1992, Text and Office Systems---Conformance Testing for Standard Generalized Markup Language Systems. This document contains operating policy for a Standard Generalized Markup Language Conformance Testing Service and is not intended to explain the detailed procedures that can be found in the documentation associated with the SGML parser validation system, commonly called the SGML Test Suite.
2.8 POSIX.2 Shells and Utilities
--------------------------------
Sheila Frankel, NIST Portable Operating System Interface (POSIX)---Part 2: Shells and Utilities provides a command language interpreter (shell) and a set of utility programs that promotes user and application portability. It is used for directory/file/data creation and manipulation, interaction with the operating system, and automation of repetitive tasks. POSIX part 2, shells and utilities is the subject of FIPS 189, and is based on ISO/IEC standard 9945-2:1993. This standard is also known as ANSI/IEEE Standard 1003.2-1993 or POSIX.2. The effective date of FIPS 189 is April 3, 1995. FIPS 189 is required for operating systems and/or applications development where POSIX shell and utility interfaces are required. FIPS 189 adopts the POSIX.2 Standard, but omits obsolescent features, or violation of the general syntactic guidelines of POSIX.2 that may be deleted from POSIX.2 at a future date. POSIX.2 requires that these features not be used by strictly conforming applications. Most obsolescent features have equivalent, nonobsolescent counterparts in POSIX.2. A FIPS 189 Testing Program is being developed by NIST. It will be similar to the conformance testing program that was established for FIPS 151-2. The testing program will use accredited Labs from the National Voluntary Laboratory Accreditation Program (NVLAP) to perform the testing. Certificates of Validation will be issued by NIST and an accredited products list maintained. The FIPS 189 Conformance Testing Program will adopt an existing test suite. A call for available test technology was published in the Commerce Business Daily (CBD) was published on October 20, 1994. Test suites submitted in answer to that announcement have been evaluated and an in-house report has been issued. An advisory board has been convened that will publish the testing model, test suite criteria, and select test suite(s). A follow-on activity, POSIX 2003.2 is under way. Upon approval the Test Methods for POSIX.2 standard will result in a test suite(s) update.
For further information contact Sheila Frankel at (301) 975-3297 or <[email protected]>.
2.9 Open System Environment Implementors' Workshop (OIW)
--------------------------------------------------------
Joe Hungate, OIW chairman, made the presentation on the OIW. The Open Systems Environment Implementors Workshop is a public international technical forum for the timely development of implementation agreements based on emerging international standards and public specifications. Its purpose is to broaden the utilization of Open Systems Environment (OSE)-based technologies and to speed their development. The workshop intent is to support the advancement of a technically efficient and compatible technology base for emerging Open Systems on a nationwide basis. NIST chairs the OIW, which meets quarterly at NIST. Workshop organizational components include the Plenary (now conducted electronically), two standing committees---the Technical Liaison Committee (TLC) and the Open System Environment Technical Committee (OSETC), and Special Interest Groups (SIGs) that perform the technical work. The Plenary reviews and ratifies SIGs technical programs of work. SIGs may also have subsidiary working and study groups to address specific issues. The workshop also consists of various Working Groups and Special Project Teams created to deal with emerging technologies and issues.
For additional information on the OIW, please contact Joe Hungate at (301) 975-3368, or <[email protected]>.
3. Automated Testing Technologies Workshop
==========================================
The workshop presented with the assistance of Clemson University, hosted a second workshop on automated testing technologies. The goals of this workshop, like the first, included reviewing existing and emerging technologies for automated specification and development of test methods, exploring the relationship between automated testing and standards development, establishing a forum for the continuing exchange of information between experts working in this area, and proposing an agenda for action which will support and accelerate efforts in automated testing.
The three focus areas of this workshop were the ADL Project, presentations on university experiences and formal methods, and industry experiences.
3.1 The ADL Project
-------------------
### 3.1.1 Update on the Assertions Definition Language (ADL) Project
Shane McCarron, Testing Research Manager, X/Open Company Ltd., presented an overview of the Assertion Definition Language Project. This overview provided a brief history of the project, described in some detail the activities over the last year, and presented a high level view of the activities planned for the coming year. The presentation also included a description of last year's efforts to use ADL to generate a test suite for the CORBA (Common Object Request Broker Architecture specification) 1.2 specification and a test suite for TET (a public domain, joint industry developed test harness).
McCarron described ADL as "a (semi) formal language in which it is possible to describe the behavior of interfaces. The ADLTranslation System is a collection of tools and additional 'languages' that permit the specification and generation of natural language interface specifications, test specifications, and tests based upon the ADL interface specification." The goals of ADL are to improve test coverage, reduce costs of test development, speed up the process of test suite generation, reduce lifecycle costs and improve the reliability of testing.
### 3.1.2 ADLT in Practice: Experiences and Anecdotes
Roger Hayes of Sun Microsystems Laboratories described the ADLT as a freely-available system for test software generation, developed by Sun Laboratories with the support of X/Open and Japan's Ministry of International Trade and Industry (MITI/IPA). Hayes described, briefly, some of the large-scale testing projects that have adopted ADLT, i.e., OpenDoc, OpenGL, PIKS, OMG CORBA, and TET, and lessons that have been learned in the course of supporting those projects. The lessons learned included that ADLT works, that there is a high degree of re-use, that it is useful in clarifying specifications, and that it can be used to develop portable tests.
3.2. University Experiences and Formal Methods
----------------------------------------------
### 3.2.1 Exploration of Easy-to-Use Formalisms for Software Specification
The project results of a 1 year collaborative effort between Sun Microsystems Laboratories and Clemson University were reported in this presentation by Kathy Liburdy of Clemson University. The goal of this effort was to provide insights on the ease of learning and using a software specification language such as the Assertion Definition Language (ADL) developed by Sun Microsystems Laboratories.
A classroom experiment involving senior students in Clemson University's Computer Engineering Program was designed to provide both a unique learning experience for the students as well as to achieve the desired goals of the collaborative effort.
The classroom experiment consisted of two distinct phases: tutorial development and specification development. Phase one required the students to develop a tutorial for ADL based on the ADL Language Reference Manual. This effort served the dual purpose of familiarizing the students with ADL and identifying difficulties with the Language Reference Manual. There were three teams involved in the experiment, and three very different tutorials resulted. An overview of the tutorials as well as observations regarding the Language Reference Manual were reported.
The second phase of the experiment required the students to develop specifications using ADL for a relatively simple problem such as a symbol table manager. After the specifications were developed, the students exchanged specifications and were asked to implement software satisfying another team's specification. As the final component of the experiment, implementations were returned to the specification developers for evaluation. Students' comments on the ease of learning ADL, using ADL, and implementing software from ADL specifications based on this experience were reported. Suggestions for supplemental material for teaching ADL concluded the presentation.
### 3.2.2 Automated Test Methods for POSIX.5
This presentation by Jim Leathrum of Clemson University, described the development and transfer of an automated testing technology in the open systems standards arena which resulted from a government and university alliance. The Clemson Automated Testing System (CATS) was initiated in response to the U. S. Navy's request for conformance testing technology. The vision during the effort was a system which would provide life cycle support for conformance testing (i.e., assertion writing, test generation, test execution, and test results analysis). The system design was based upon the traditional compiler paradigm in that assertions about the required behavior of an implementation under test are translated into executable tests.
In 1994, the Defense Information Systems Agency (DISA) sponsored the application of the CATS technology in the development of test methods for the Ada Language binding to POSIX. The development of assertions for POSIX.5 revealed significant realizations which can be directly attributed to the technology, such as the value of providing rapid feedback from the CATS environment during test development. Additionally, Leathrum disclosed that it became apparent that testing issues were addressed much earlier in the process than in traditional approaches. A discussion of lessons learned from this experience concluded the presentation.
### 3.2.3 Symbolic Execution and Constraint Satisfaction in Automatic Test Case Generation
Steven Zeil from Old Dominion University submitted a paper which described the following.
For over 20 years, researchers have noted that symbolic execution offered a conceptually elegant approach to the automatic generation of tests for structural and other implementation-based testing criteria.
A number of symbolic execution systems have been built, typically for older programming languages and offering limited facilities for constraint satisfaction. The inability of these systems to deal with abstract data and more modern programming languages has raised questions about the viability of symbolic execution in general.
The ARIES symbolic executor attempted to modernize symbolic execution, allowing symbolic execution of Ada programs. Although its runtime performance was disappointing, it offered some significant advances in design, including: Preservation of abstraction in expressions involving ADT'sSeparation of constraint satisfaction from the execution engine
Advances in constraint satisfaction techniques also hold new promise. Test case generation has unusual characteristics that place a strain on constraint solvers. The constraint systems seldom fall into the neat classifications on which most solvers are designed, but tend to mix many constraint theories. On the other hand, some recent projects have indicated that the vast majority of constraint systems that arise during testing tend to be easily solvable, despite being ill-formed for conventional techniques. Some preliminary experience with a MTCSS (Multi-Theory Constraint Satisfaction System) suggests directions for future work.
3.3. Industry Experiences
-------------------------
### 3.3.1 Automatic Efficient Test Generator (AETG) System
According to David Cohen, Bellcore, software testing is expensive, tedious and time consuming. By some estimates, testing accounts for 30 % to 50 % of development costs. Making testing more efficient has long been a goal of software engineering research. Cohen described the Automatic Efficient Test Generator (AETG) System that is a new tool developed by Bellcore that automatically generates test sets from high-level test requirements. It uses new algorithms from combinatorial design theory to generate test sets that efficiently cover the test requirements. The AETG system has been used in Bellcore and a major telecommunications manufacturer for feature and protocol conformance testing, for inter-operability testing, and for testing user interface software. Cohen presented data from these experiences.
### 3.3.2 TGGS: A Flexible System for Generating Efficient Test Case Generators
Ronald F. Guilmette, RG Consulting, described the Test Generator Generator System, TGGS. TGGS is a simple yet flexible system for generating highly efficient automated random test case generators. The random test case generator programs generated by TGGS may themselves used to generate randomized test cases for a variety of programs, most notably compilers. TGGS is based upon a specification language (SL), very similar to the input language accepted by YACC, in which the user may express both the syntactic and semantic constraints of the input language for the program to be tested. The SL language was described in detail. A description of the SL compiler, GTG, and of the associated SL runtime system was also provided, and the application of TGGS to some example testing problems described.
### 3.3.3 A Solution for Automated Testing to Ensure Product Interoperability
John Reardon from Midnight Networks Inc. described his views on testing networking products. He said that thorough testing is a requirement for network products to interoperate acceptably in customer networks. However, it is not possible to perform such testing solely via manual methods and test net operation, as such approaches are costly, slow and ineffective.
Automated testing allows thorough, rigorous testing to be performed, while cutting the time it requires. By using automated testing with positive conformance tests, negative tests, and stress tests, interoperability may be achieved. Reardon described the design and architecture of a system for automated network testing that has been implemented by Midnight Networks. He also described experiences and case histories that show its benefits in cost, cycle time and quality to organizations that make use of it.
### 3.3.4 Automated Test Generation with TestMaster
The presentation by Larry Apfelbaum, Teradyne Software & Systems Test, described a new approach that has been successful for automating the test generation phase for software based systems. This solution uses a model of an application's desired behavior as the basis for a flexible, automated test generator. The presentation covered the process of modeling and how a path generation engine operating with that model can efficiently generate tests of a known quality. Included were a description of the elements of a model and the role they play as the basis of a testable specification. The process a path generator uses to build tests was also explained with some examples. Samples of the results obtained by some of the existing production applications of this technology were given.
### 3.3.5 STEP Conformance Testing
The NIST (National Institute of Standards and Technology) and ITI (Industrial Technology Institute) program on STEP Conformance Testing is entering its fourth year. Bob Matthews from ITI presented a review and update of the status of this program. To date, this program has developed a variety of tools, tests, and services enabling STEP product conformance testing. The testing technology is currently being used by many vendors, users, test laboratories, and standards developers, and is accelerating STEP product realization.
The goals of the fourth year of the program are to extend the features of several prototypes, complete integration of tools into a CASE-style environment, launch a prototype conformance test service, apply conformance test technology to support interoperability testing, adapt tools and services to enable testing of related standards, and validate test technology and methods.
Several tools which are being extended were described including the: Coverage Analyzer (CA), ARM/AIM Browser/Editor (AABE), and Verdict Criteria Generator (VCGEN). The CA serves two principal purposes: verification of conformance test suite completeness and quantitative analysis of interoperability tests. The AABE tool provides application-domain views of STEP exchange data. The VCGEN tool produces sets of detailed verdict criteria that are used to efficiently evaluate testing outputs.
Matthews described the STEP Test System (STS) which integrates the complete set of STEP conformance test tools. The STS provides an object-oriented testing paradigm to enable users to quickly learn and manipulate the many artifacts involved in conducting tests. The core of the STS is a testing "harness" which enables simplified plug-in capability for standard-specific test tools and data.
Matthews also described a "beta" STEP conformance testing service which was recently launched. Using the conformance test technology developed under the NIST-ITI program, early STEP vendors engage the service to formally evaluate and demonstrate to users the conformance, and therefore interoperability potential, of their products. This beta service is expected to evolve into a replicable NIST-NVLAP accredited STEP testing process.
Matthews summarized that to date, most of the testing program efforts have focused on working with one principal STEP standard, AP 203. Dozens of other STEP standards will be coming on-line, and require similar testing support. The NIST-ITI tools, designed to serve the many standards of STEP, are now being configured, extended, and applied to these other STEP standards.
4. Tutorial for Novices
=======================
The forum began with an introductory tutorial on the Open System Environment (OSE). The OSE forms an extensible framework that allows services, interfaces, protocols, and supporting data formats to be defined in terms of nonproprietary specifications that evolve through open (public), consensus-based forums. A selected suite of specifications that defines these interfaces, services, protocols, and data formats for a particular class or domain of applications is called a profile. Fritz Schulz presented OSE general concepts and the reference model.
4.1 Open System Environment (OSE) Reference Model (RM)
------------------------------------------------------
The Institute of Electrical and Electronics Engineers (IEEE) POSIX Working Group 1003.0 defines an OSE Reference Model (OSE/RM) that provides a framework for describing open system concepts and defining a lexicon of terms that can be agreed upon generally by all interested parties. The OSE/RM is also identified at the international level in Joint Technical Committee 1 (JTC1) Technical Report (TR) 14250. [Figure 1](#f1-j16ce-hun){ref-type="fig"} illustrates the OSE/RM.
Two types of elements are used in the model: entities consisting of the application software, application platform, and platform external environment; and interfaces including the application program interface and external environment interface.
The three classes of OSE reference model entities are described as follows: *Application Software*---Within the context of the OSE Reference Model, the application software includes data, documentation, and training, as well as programs.*Application Platform*---The application platform is composed of the collection of hardware and software components that provide the generic application and system services.*Platform External Environment*---The platform external environment consists of those system elements that are external to the application software and the application platform (e.g., services provided by other platforms or peripheral devices).
There are two classes of interfaces in the OSE reference model: the application program interface and the external environment interface. *Application Program Interface (API)*---The API is the interface between the application software and the application platform. Its primary function is to support portability of application software. An API is categorized in accordance with the types of service accessible via that API. There are four types of API services in the OSE/RM: Human/computer interface servicesInformation interchange servicesCommunication servicesInternal system services*External Environment Interface (EEI)*---The EEI is the interface that supports information transfer between the application platform and the external environment, and between applications executing on the same platform. Consisting chiefly of protocols and supporting data formats, the EEI supports interoperability to a large extent. An EEI is categorized in accordance with the type of information transfer services provided. There are three types of information transfer services. These are transfer services to and from: Human usersExternal data storesOther application platforms
In its simplest form, the OSE/RM illustrates a straightforward user-supplier relationship: the application software is the user of services and the application platform/external environment entities are the suppliers. The API and EEI define the services that are provided.
4.2 OSE Profile and the APP
---------------------------
A profile consists of a selected list of standards and other specifications that define a complement of services made available to applications in a specific domain. Examples of domains might include a workstation environment, an embedded process control environment, a distributed environment, a transaction processing environment, or an office automation environment, to name a few. Each of these environments has a different cross-section of service requirements that can be specified independently from the others. Each service, however, is defined in a standard form across all environments.
An OSE profile is composed of a selected list of open (public), consensus-based standards and specifications that define services in the OSE/RM. Restricting a profile to a specific domain or group of domains that are of interest to an individual organization results in the definition of an organizational profile. The Application Portability Profile (APP) is an OSE profile designed for use by the U.S. Government. It covers a broad range of application software domains of interest to many Federal agencies, but it does not include every domain within the U.S. Government's application inventory. The individual standards and specifications in the APP define data formats, interfaces, protocols, or a mix of these elements.
4.3 APP Service Areas
---------------------
The services defined in the APP tend to fall into broad *service areas*. These service areas are: operating system services (OS)human/computer interface services (HCI)data management services (DM)data interchange services (DI)software engineering services (SWE)graphics services (GS)network services (NS)
Each service area is defined in the following sections. [Figure 2](#f2-j16ce-hun){ref-type="fig"} illustrates where each of these services areas relates to the OSE/RM. (Assume that software engineering services are applicable in all areas.)
Each of the APP service areas addresses specific components around which interface, data format, or protocol specifications have been or will be defined. Security and management services are common to all of the service areas and pervade these areas in one or more forms.
*Security* as applied to both stand-alone and distributed systems takes a holistic approach. Each component provides different elements of functionality and security service. Security services are provided to support the secure distribution and integrity of information and to protect the computing infrastructure from unauthorized access. Security policy, authority, domains, and interactions among these domains are specifically defined in IEEE P1003.22 *Draft Guide for POSIX Open Systems Environment---A Security Framework*. Security is a cross-category service and part of the overall context in which information systems must operate. It is of relevance within all system functions, for example system services, communications services, and data management services.
Currently, specifications for security can be recommended in operating system services, network services, and access control and integrity constraints for data management services. Specifications for security in the other service areas are not sufficiently advanced to warrant inclusion at this time.
*Distributed system management* is coming to be regarded as the integration of distinct, supporting management areas. Among these areas are system administration, communication (network) management, information management, and human/computer interface management. Management services provide the mechanisms to monitor and control the operation of individual applications, databases, systems, platforms, networks, and user interactions with these components. Management services also enable users and systems to become more efficient in performing required work.
These services are just now being addressed by standards development organizations (SDO) and user consortia, particularly for heterogeneous systems. The disparate mechanisms necessary for competent management of distributed systems require an integrated approach to assure consistency. Standardization is being developed by many committees in various SDOs, workshops, and consortia. Recent attempts by these committees has led to closer coordination. True integration among them, however, requires significant additional effort. As specifications for management services mature and stabilize, they will be reviewed and appropriate ones may be selected for use in the APP.
5. General References
=====================
Gary Fisher, Application Portability Profile (APP), The U.S. Government's Open System Environment Profile OSE/1 Version 2.0, NIST Special Publication 500-210, June 1993.
{#f1-j16ce-hun}
{#f2-j16ce-hun}
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION {#sec1-1}
============
Periodontitis is an immuno-inflammatory disease process resulting from the interaction of a bacterial attack and host inflammatory response, causing inflammation of the supporting tissues of the teeth leading to tissue destruction and tooth loss. Arrays of molecules are considered to mediate the inflammatory response at one time or another, among these is free radicals (FRs) and reactive oxygen species (ROS) like superoxide anion radicals, hydrogen peroxide, hydroxyl radicals and hypochlorous acid. All these molecules are capable of damaging either cell membranes or associated bio-molecules. Periodontal pathogens can induce ROS overproduction and thus may cause collagen and periodontal cell breakdown. When ROS are scavenged by antioxidants, there is a reduction of collagen degradation.\[[@ref1]\] Oxidative stress arises within tissues when the normal balance between ROS generation and antioxidant defense shifts in favor of the former, a situation arising from either an excess of ROS and/or a depletion of antioxidants.
There has been a tremendous expansion in dental research concerned with free radicals, ROS and anti-oxidant defense mechanisms. These are essential to many normal biological processes and low doses of certain radicals or radical-derived species can stimulate the growth of fibroblasts and epithelial cells in culture. A low FR/ROS often behaves as an inductor stimulus, whereas higher levels may result in injury. It may be necessary to deliver anti-oxidants selectively to specific cell types and to define the concentrations suitable for blocking inappropriate cell responses but leaving the unimpaired physiological levels of FR/ROS activity necessary for normal cell function.\[[@ref2]\]
Coenzyme Q10 was discovered in beef heart mitochondria at the University of Wisconsin.\[[@ref3]\] coenzyme Q10 is also known as ubiquinone because of its ubiquitous presence in nature and its quinone structure (similar to that of vitamin K).\[[@ref4]\] It is also called as "coenzyme" because of its unique ability to participate in chemical reactions but remain at steady-state levels in the cell, and plays a central role in energy metabolism. It has a positive inotropic effect.\[[@ref5]\]
The effects and mechanisms of action of CoQ10 include stabilization of calcium-dependent channels, inhibition of intracellular phospholipases, prostaglandin metabolism, free-radical scavenging and direct membrane stabilization.\[[@ref6]\]
CoQ10 is also known to play a crucial role in the generation of adenosine triphosphate (ATP) and cellular respiration. It exists in two molecular forms, ubiquinone, the oxidized form, and ubiquinole, the reduced form, which are the basis for its antioxidant properties.\[[@ref7]\] Co-Q10 functions as an intercellular antioxidant by acting as a primary scavenger of FRs and ROS. It serves as an endogenous antioxidant, and its increased concentration in the diseased gingiva effectively suppresses advanced periodontal inflammation.
A deficiency of coenzyme Q10 in the gingival tissue may exist independently of and/or because of periodontal disease. If a deficiency of coenzyme Q10 existed in the gingival tissue for nutritional causes and independently of periodontal disease, then the advent of periodontal disease could enhance the gingival deficiency of coenzyme Q10. In such patients, oral dental treatment and oral hygiene procedures can remove the local factors only but cannot correct the deficiency of CoQ10 due to systemic cause. Thus, mechanical periodontal therapy along with the adjunctive use of CoQ10 can be included for an overall improvement of the gingival health in periodontal disease.\[[@ref8]\] The present study was designed with the aim to evaluate the efficacy of coenzyme Q-10 gel application in the treatment of chronic periodontitis.
MATERIALS AND METHODS {#sec1-2}
=====================
This was a randomized, controlled, clinical trial with a split-mouth design. A total of 18 patients were selected from the Outpatient Department of Periodontology. Ethical clearance was obtained from the institutional ethical committee. Systemically, healthy patients in the age group of 20-55 years of both the genders (mean age 33.8 years) who were diagnosed with chronic periodontitis by their clinical and radiographic findings were included in the study. Written and verbal consent was obtained from the sample recruited for the study.
Patients suffering from chronic periodontitis and having a probing pocket depth of ≥5 mm in different quadrants (having a minimum of six permanent teeth in each quadrant) of the mouth with radiographic evidence of bone loss were included in the study \[Figures [1](#F1){ref-type="fig"}--[3](#F3){ref-type="fig"}\].
{#F1}
{#F2}
{#F3}
Patients with a history of any systemic diseases, any apparent oral infection like herpes or candida, patients who had taken antibiotic therapy in the past 3 months or undergone any periodontal therapy in the past 6 months, smokers, pregnant women and lactating mothers were excluded from the study.
Perio Q gel (Perio Q™) is a mixture of CoQ10 in a vegetable oil base in ratio of 1:9 and is supplied as a pack of gel \[[Figure 4](#F4){ref-type="fig"}\], and was stored at a temperature between 4 and 8°C to maintain its shelf-life.
{#F4}
In this study, three quadrants were assigned randomly in each patient: Group I: Scaling and root planning only (Control group); Group II: Scaling and root planing and topical application of Perio Q gel (Test group A); Group III: Scaling and root planing and intrapocket Perio Q gel application (Test Group B) \[Figures [5](#F5){ref-type="fig"} and [6](#F6){ref-type="fig"}\].
{#F5}
{#F6}
All the clinical parameters, i.e. plaque index,\[[@ref9]\] gingival index,\[[@ref10]\] modified sulcular bleeding index and probing pocket depth were recorded at baseline and at the 2^nd^ week and 4^th^ week after treatment.
At the baseline, scaling and root planing were performed in all the groups and in Group II, the topical application of gel was performed with the tip of the applicator completely soaked in gel. Intrapocket application was performed in Group III using special needles designed to deliver gel in the pocket. Patients were recalled every alternate day for the application of gel for 1 week. Eating, spitting and drinking were restricted for 1 h after application. Patients were recalled at the 2^nd^ and 4^th^ weeks after treatment to record all the clinical parameters \[[Figure 7](#F7){ref-type="fig"}\].
{#F7}
RESULTS {#sec1-3}
=======
The ANOVA test was employed for plaque index, gingival index, gingival bleeding index and probing pocket depth. A comparison of the mean plaque index, gingival index, gingival bleeding index and probing pocket depth among all the three groups at the observational period of 2^nd^ and 4^th^ weeks showed statistically significant results (*P* \< 0.01) \[Tables [1](#T1){ref-type="table"}--[4](#T4){ref-type="table"}\].
######
Plaque index (Silness and Loe 1964)
######
Gingival index (Loe and Silness 1963)
######
Gingival bleeding index (Ainamo and Bay 1975)
######
Probing pocket depth
Plaque index {#sec2-1}
------------
There was a statistically significant improvement in the plaque index from baseline to the 4^th^ week revaluation in Groups I-III \[[Table 1](#T1){ref-type="table"}\]. For Group I, the plaque index was reduced from the baseline value of 1.72 ± 0.492 to 0.96 ± 0.274 (*P* \< 0.0001\*), and for Groups II and III, it was reduced from 1.93 ± 0.468 to 0.61 ± 0.230 and 1.69 ± 0.424 to 0.48 ± 0.148 (*P* \< 0.0001\*), respectively.
Gingival index {#sec2-2}
--------------
The gingival index scores were significantly improved from baseline to the 4^th^ week revaluation \[[Table 2](#T2){ref-type="table"}\]. For Group I, the gingival index scores was reduced from 1.64 ± 0.422 to 0.63 ± 0.366 (*P* \< 0.0001\*), and for Groups II and III, it was reduced from 1.82 ± 0.391 to 0.5 ± 0.227 and 1.96 ± 0.57 to 0.57 ± 0.397 (*P* \< 0.0001\*), respectively.
Gingival bleeding index {#sec2-3}
-----------------------
Similarly, for Group I, the bleeding scores were improved significantly from 1.93 ± 0.329 to 1.06 ± 0.107 (*P* \< 0.0001\*), and for Groups II and III, it was reduced from 1.88 ± 0.261 to 1.01 ± 0.059 and 1.82 ± 0.341 to 1 ± 0 (*P* \< 0.0001\*), respectively \[[Table 3](#T3){ref-type="table"}\].
Probing pocket depth {#sec2-4}
--------------------
Improvement in probing pocket depth from baseline to the 4^th^ week was also significant. For Group I \[[Table 4](#T4){ref-type="table"}\], it was reduced from 5 ± 0.84 to 3.66 ± 0.686 (*P* \<0.0001\*), and for Groups II and III, it was reduced from 5.72 ± 0.57 to 3.72 ± 0.826 and 6.33 ± 1.085 to 3.72 ± 0.826 (*P* \< 0.0001\*), respectively.
Thus, there was a more significant improvement in all the clinical parameters among Groups II and III as compared with Group I at the 2^nd^ - and 4^th^ -week revaluations.
DISCUSSION {#sec1-4}
==========
Periodontal disease affects 60% of the young adults and 90% of individuals over the age of 65 years. Healing and repair of periodontal tissue requires efficient energy production, and the metabolic functions of the periodontal tissues depend on an adequate supply of CoQ10.\[[@ref3]\] Gingival biopsies revealed subnormal tissue level of CoQ10 in 60--96% patients with periodontal disease, indicating that periodontal disease is frequently associated with CoQ10 deficiency.\[[@ref3]\]
Physiologically, CoQ10 plays four major roles. It has an essential role in mitochondrial energy (ATP) production through redox activity in the respiratory chain, transporting electrons between enzymes. Second, it plays a role in extra-mitochondrial redox activity in the cell membrane and endo-membranes. CoQ10 also functions as an antioxidant, inhibiting lipid peroxidation and scavenging FRs. Finally, it plays an important role in membrane stabilization and fluidity.\[[@ref4]\]
The antioxidant nature of CoQ10 is derived from its energy carrier function. As an energy carrier, the CoQ10 molecule is continuously going through an oxidation--reduction cycle. CoQ10 inhibits lipid peroxidation by preventing the production of lipid peroxyl radicals. In addition, the reduced form of CoQ10 effectively regenerates vitamin E from the α-tocopheroxyl radical. Furthermore, during oxidative stress, interaction of H~2~ O~2~ with metal ions bound to DNA generates hydroxyl radicals and CoQ10 efficiently prevents the oxidation of bases, particularly in mitochondrial DNA.\[[@ref8]\]
Data show that a high respiration rate is positively correlated with an increased amount of mitochondrial CoQ10, suggesting that its supplementation can play an important role in diseases related to CoQ10 deficiency. The mitochondrial respiratory chain organization plays an important role in the increase of respiratory activity.\[[@ref11][@ref12]\] Currently, two models have been proposed: The random collision model and a super complex organization called respirasome.\[[@ref13]\]
In the first model, electron transfer through the respiratory chain is assured by free diffusion of each component within the inner mitochondrial membrane. In this scenario, CoQ10 forms a pool used by all the CoQ-dependent respiratory complexes (mainly Complex I, II and III). On the other hand, the respirasome requires a solid state organization in which only bound CoQ10 is involved in electron transfer. This last hypothesis seems to be in contrast with a dose-dependent effect of CoQ10 addition on the respiratory rate. However, it may be possible that the bound ubiquinone should be in equilibrium with the pool. This hypothesis explains the beneficial effect of exogenous CoQ10 supplementation.\[[@ref14]\]
Cells and tissues that play a role in immune function are highly energy dependent and therefore require an adequate supply of CoQ10 for optimal function. Several studies have demonstrated the immune-enhancing effects of CoQ10 or its analogues. These effects included the increased phagocytic activities of macrophages, increasing the proliferation of granulocytes in response to infection.\[[@ref8]\] Patients with a decreased level of serum CoQ10 were at an increased risk of disease progression,\[[@ref15][@ref16]\] and application of CoQ10 would improve the clinical parameters.
Our study has shown a significant decrease in the plaque index, gingival index and gingival bleeding index in Groups II and III (test group) as compared with Group I (control group), thereby corroborating the added advantage of coenzyme Q10 gel. On comparison between Groups II and III, there was a more significant improvement in the plaque index, gingival index and gingival bleeding index in Group III as compared with Group II. There was a significant decrease in the probing pocket depth in Groups II and III (test groups) at the end of the 4^th^ week as compared with Group I (control group). The *P*-value, 0.001, was statistically significant at the 4^th^ week for the test groups. All the parameters as well as the clinical indices indicate that coenzyme Q10 (Perio Q gel) when used with scaling and root planning gave an added advantage as compared with sites treated with scaling and root planing alone. The results were more significant in the group treated with intra-pocket gel delivery along with scaling and root planing.
One problem encountered during the study was the substantivity of the gel, as it was neither a sustained release nor a controlled release formulation. The bioavailability of the gel was not known. During the study, some patients gave the statement of improved subjective feeling of the disease condition and less bleeding from the gums, indicating the effectiveness of the gel. No adverse reactions were reported.
Our research was focused on the adjunctive use of Q gel in patients with chronic periodontitis. On intragroup comparison, more significant results were seen in the sites treated with Scaling and root planing and Perio Q gel combination as compared with the sites treated with SRP alone. However, in this study, because of the lack of further follow-up of 3--6 months, the long-term effect of the Perio Q gel cannot be commented upon. Thus, this study necessitates further detailed, long-term evaluation of the effect of the Perio Q gel in a bigger population with chronic periodontitis, along with follow-up of up to 3 months.
Our study has shown decreases in the clinical parameters scores, which are consistent with the earlier studies performed by Hans *et al*.,\[[@ref3]\] Wilkinson *et al*.\[[@ref15]\] and Chatterjee *et al*.\[[@ref17]\] A study conducted by Matthews *et al*. in 2007 showed that coenzyme Q10 with vitamin E has a beneficial effect on the periodontal tissue.\[[@ref18]\] A split-mouth trial conducted by Hanioka demonstrated improved periodontal scores along with gingival scores when Co-Q10 was applied alone or as an adjunct to scaling and root planing.\[[@ref19]\] A study was conducted by Figuero (2006) to evaluate the potential oxidant/antioxidant interactions of nicotine with antioxidant coenzyme Q10 in smokers who were diagnosed with periodontitis, suggesting that the catabolic effects of nicotine could be reversed by the addition of antioxidants such as CoQ10.\[[@ref20]\]
A study reported deficiency of CoQ10 in patients with periodontal disease by conducting a series of trials on succinate dehydrogenase--CoQ10 reductase enzyme, which is found in the mitochondrial complex II of the cell.\[[@ref5]\] This study was later supported by the works of Nakamura\[[@ref8]\] and Matsumura.\[[@ref21]\]
Clinical trials showed a positive relation between Co-Q10 administration and improved periodontal health and immune response.\[[@ref22]\] Other than antioxidant action, it has also been shown in the literature that CoQ10 acts as an immune enhancer and also accelerates tissue healing.\[[@ref23][@ref24]\] Thus, the result of our study could probably be due to the cumulative effects of Perio Q gel as an antioxidant, immune enhancer and tissue healer.
CONCLUSION {#sec1-5}
==========
Intrapocket and topical application of Perio Q gel along with mechanical debridement has improved the clinical parameters. The results of the research were encouraging and suggested the possibility to use the gel as a topical agent to support standard treatment procedures in periodontitis. In our research, the clinical parameters significantly improved in the phase of periodontal treatment, indicating that CoQ10 opens new treatment options by improving the host response to disease activity. The results were more significant and encouraging in the group treated with intrapocket gel application. Thus, this study necessitates further long-term clinical trials of Perio Q gel in larger sample sizes with long-term revaluation, along with other antioxidants agents with various doses and durations for designing a strategy for their use in routine supportive periodontal therapy.
**Source of Support:** Nil
**Conflict of Interest:** None declared.
| {
"pile_set_name": "PubMed Central"
} |
Since the discovery of graphene, two-dimensional (2D) materials with atomic thickness have drawn increased attentions in recent years that hold promise for developing next-generation high-performance electronics, optoelectronics and spintronics[@b1][@b2][@b3][@b4][@b5][@b6][@b7][@b8]. Many exotic physical and chemical phenomena emerge with reducing dimensions owing to the change of electronic behaviors governed by quantum confinement effects within the 2D layer. Although excellent electrical and thermal transport properties make graphene a promising candidate for applications in transparent conductor and high mobility devices, its zero bandgap nature limits its technological applications in digital electronic devices, which inspires researches for materials with finite bandgap, such as transition metal dichalcogenides (TMDCs), (MX~2~, M a transition metal atom, such as Mo, W, Re, Pt, Sn etc. and X a chalcogen atom, such S, Se, or Te)[@b9][@b10][@b11] and black phosphorus[@b12] (BP) that were recently extensively studied. More recently, layered tin (II) oxide (SnO) was found to show excellent semiconducting performances in which bipolar conductivity can be easily realized[@b13][@b14][@b15][@b16]. It is a rare example of layered oxide semiconductors that holds promise for a wide variety of technological applications[@b14][@b17][@b18], which makes it quickly become the subject of significant theoretical and experimental investigations[@b1][@b13][@b14][@b18][@b19]. Compared to other TMDC-based layered structure, oxide-based materials are expected to be more stable in air since they are reluctant to oxidation.
Bulk SnO has a tetragonal unit cell (litharge crystal structure with space group: *P4/nmm*) with lattice constants of *a = b* = 3.8 Å and *c* = 4.84 Å[@b20][@b21]. A lone pair model was suggested for the electronic interaction of SnO[@b22], in which the lone pair states result from the crystal structure of SnO. In SnO, each Sn atom loses two 5*p* electrons to adjacent O atoms leaving an electronic configuration of 4*d*^*10*^5 *s*^*2*^5*p*^*0*^ in which the two Sn 5*s* electrons constitute a lone pair pointing towards the interlayer spacing. The inter-layer lone-pair interaction was found to be crucial for understanding the electronic structure of SnO[@b14]. In analogy with other TMDCs-based layered structure, Sn--O--Sn slabs in SnO are stacked along \[001\] direction with weak dipole-dipole van der Waals interlayer interaction[@b1][@b14][@b15]. In its monolayer counterpart, the lack of dipole-dipole lone pair interaction shows great impact on the electronic structure and widens the band gap which bring additional potential applications. To design a SnO-based material with required electronic structure, it is essential to explore the relationship between the engineering methods and their effects on the electronic and magnetic modifications. The unique lone pair electronic states around Sn atoms could also be modified by adatoms which needs to be understood.
In order to explore the potential applications of 2D materials, many strategies could be used to tune their electronic and magnetic properties, such as strain[@b15], defects and substitutional doping[@b23]. It is well-known that the exfoliation or growth processes can introduce defects and impurities in 2D materials which can dramatically alter their electronic, thermal and mechanical properties. Vice versa, a deliberate introduction of defects can be a feasible approach to modify the properties of the pristine materials. For instance, vacancies and Stone-Wales (SW) defects introduced in graphene by ion or electron irradiation brought new functionalities for graphene-based devices[@b24][@b25]. Defects in MoS~2~ formed during the growth process also play significant roles on their electronic behaviors as well as device performance[@b26][@b27][@b28]. Besides intrinsic defects, extrinsic defects such as adatoms are also shown to be important for 2D materials based devices with dedicated properties[@b29]. For real applications, modifications and engineering are generally applied to enhance the materials/devices functionalities. Especially for new materials, great efforts are needed to explore its function enhancement in all manner of possibilities. Adsorption of foreign atoms is another effective and promising strategy for tuning the electronic and magnetic properties of 2D materials[@b23][@b30]. Considering that the scientific investigations on the properties of SnO has just started[@b14][@b15][@b16][@b18][@b19][@b31][@b32][@b33][@b34], the role of extrinsic point-defects and the effects of adatoms on SnO need to be explored to widen the range of its applications.
In this work, we systematically investigate the stable crystal structures and electronic and magnetic properties of nonmetal atoms (B, C, N, O, and F) adsorbed SnO monolayers. We will calculate their geometric structure and electronic properties to gain insights in the adsorption mechanism. Compared with the magnetic moment from *d*-electrons of transitional-metal atoms, the magnetism from *sp* electrons of nonmetal elements could have stronger long-range exchange coupling interactions and avoid cluster formation of magnetic ions[@b35]. We will show that the electronic behavior of C and O is totally different from that of B, N and F adatoms, and the adsorption mechanism is dominated by the electronegativity of the adatom.
Results and Discussion
======================
Band structure property
-----------------------
SnO possesses a layered structure with each Sn atom bonded with four adjacent O atoms and vice versa, see the ball-and-stick model in [Fig. 1(a)](#f1){ref-type="fig"}. In [Fig. 1](#f1){ref-type="fig"}, the band structure at the valence band maximum (VBM) and conduction band minimum (CBM) for both bulk and monolayer (ML) SnO are given. It is found that both bulk and ML SnO are indirect band gap semiconductor, in which VBM is situated among Γ-M line. However, the CBM is located at M point for bulk and shifted to Γ point for ML which is in excellent agreement with previous theoretical reports[@b13][@b14][@b15]. The involvement of spin-orbital coupling (SOC) is found to have no effect on the band structure, *i.e.* no band splitting at CBM and VBM ([Fig. S1](#S1){ref-type="supplementary-material"}). The evolution of band structure implies indirect-direct-indirect band bap conversion may occur during the layer-thinning process, which is in strong contrast to other 2D layered materials, such as TMDCs. For the bulk SnO, the band gap is 0.38 eV while it increases dramatically to 2.94 eV for ML SnO at GGA level, see [Fig. 1(b,c)](#f1){ref-type="fig"}. The huge change of the band gap highlights the importance of interlay coupling which affects significantly the lone pair states that could also be modified by adsorption of foreign atoms. As is well known, the band gap is underestimated by the standard DFT calculation due to the poor description of the correlation interaction between electrons. To overcome this drawback, GGA+U and HSE06 methods were used to correct this. The results show that the correction of band gap do not have noticeable effect on the band structure behaviour ([Figs S2--S3](#S1){ref-type="supplementary-material"}). Although interaction in the \[001\] direction is governed by weak van der Waals, the calculated band structure for bulk SnO \[see [Fig. 1(b)](#f1){ref-type="fig"}\] exhibits significant band dispersion along the Γ-Z direction at VBM. However, [Fig. 1(c)](#f1){ref-type="fig"} shows that the band dispersion along *z* directions become flat for ML SnO which implies the band dispersion is originated from the strong inter-layer interaction. This is a very different behavior from the transition-metal dichalcogenides[@b9]. The orbital contributions to the electronic structure of pristine bulk SnO is obtained from partial density of states (PDOSs) which shows the VBM is derived from the in-layer hybridization of Sn 5*sp* orbitals and the O 2*p* orbitals consistent with lone pair model[@b22], while the CBM is contributed primarily from the hybridization of Sn 5*p* and O 2*s* antibonding orbitals and associated with the interlayer Sn^2+^--Sn^2+^ bonds. Since the band gap evolution is dominated by the orbitals at CBM, it is the interlayer lone pair interaction between Sn atoms that dominates the electronic structure and the band gap of SnO. Our calculated results are in excellent agreement with the previous theoretical reports[@b14][@b36].
Geometry stabilization
----------------------
In SnO, each Sn atom shares its two 5*p* electrons with the neighboring O atoms while the remaining 5*s* electrons do not participate in the bonding process and constitute a lone pair. As mentioned above, the band structure of SnO is determined by the lone pair interaction, the above model gives a strategy for the electronic structure design by adatoms which bond with Sn atoms via the lone pair electrons. To do so, the adsorption of a series of adatoms of B, C, N, O and F were studied, which provides an interesting variation in the number of valence electrons with a \[He\] core and electronegativity. After adsorption, the geometric structure was changed as shown in [Fig. 2(a--e)](#f2){ref-type="fig"}. In the equilibrium configurations, the preferred binding sites are between two surface Sn atoms. B, C, N and F adatoms prefer the bridge site whereas O adatom prefers the top site. On the other hand, B and C adatoms appear to form bonds with two native Sn atoms at the surface in their equilibrium configurations. It is noted from [Fig. 2](#f2){ref-type="fig"} that the surface reconstruction is pronounced in B- and C-adsorbed SnO monolayers. The native Sn atoms along *x* direction are attracted toward the adatoms while repelled along *y* direction resulting in lattice distortions with a positive Poisson's ratio. It is noticed that the adsorption of additional O atoms has little effect on the lattice distortion, see [Fig. 2(e)](#f2){ref-type="fig"}. SnO bond length in the pristine 2D lattice is 2.257 Å. For all adatoms expect B, the Sn-X (X = C, N, O and F) distance is shorter than that of pristine Sn-O bond suggesting strong bonding. The binding energy (*E*~*B*~) of the surface adatom is defined as
where *E*~*total*~ is the total energy of SnO with surface adatom, *E*~*pristine*~ is the energy of pristine SnO, *E*~*atom*~ is the energy of a single adatom calculated using a 10 × 10 × 10 Å^3^ unit cell. According to the definition of binding energy, the negative value of *E*~*B*~ implies that the adsorption process is exothermic, and the larger is the absolute value of *E*~*B*~, the stronger is the interaction between the adatom and SnO monolayer. As shown in [Table 1](#t1){ref-type="table"}, the calculated binding energies of B, C, N, O, and F adatoms on SnO are −1.765, −3.495, −4.243, −5.419, and −3.820 eV, respectively, which are stronger than they adsorbed on graphene[@b37][@b38][@b39]. The binding strength order is also in consistent with their bonding length. Since B, C, N, O and F atoms are more electronegative than the host Sn atoms, these adatoms tend to attract electrons from the lattice Sn atoms.
Magnetic property
-----------------
Having established the stability of X (X = B, C, N, O and F) defects on 2D SnO, we start to look at the changes in the electronic and magnetic properties of the host SnO by an X atom. In [Table 1](#t1){ref-type="table"}, it can be found that C and O atoms do not introduce magnetic moments to the system as B, N and F do. The magnetic energy (*ΔE*~*m*~), which represents the energy gained from the spin polarization, is calculated as:
for the decorated SnO monolayer, where *E*~*NM*~ and *E*~*Mag*~ are the total energy for non-magnetic and magnetic system, respectively. The magnetic energies of O-, C-, B-, N- and F-adsorbed SnO monolayers are 1.3, 14.3, 280.8, 471.7 and 134.8 meV, respectively. The obtained *ΔE*~*m*~ for B-, N- and F-adsorbed SnO are higher than thermal fluctuation of the surroundings. Therefore, the B-, N- and F-adsorbed SnO monolayers exhibit stable magnetic ground states with total magnetic moment of 0.41, 0.63 and 0.79 *μ*~*B*~ per adsorption, respectively. The O- and C-adsorbed SnO monolayers are nonmagnetic because their magnetic moment are zero.
In order to explore the mechanism of the difference in forming magnetic moment for different adatoms, we first present the spin-resolved density of states (DOS) for O and C adsorbed cases in [Fig. 3](#f3){ref-type="fig"}. The adsorption gives rise to highly localized states within the band gap, as shown in the DOS plots in [Fig. 3](#f3){ref-type="fig"}. From the spin resolved DOS spectra, the additional states introduced by O atom sit just on top of VBM and below CBM while C adsorption introduces three deeper mid-gap states. However, in both cases, the up-spin and down-spin are symmetrical which cancel each other resulting in nonmagnetic ground states. Although the O adsorption has a small effect on the electronic properties of ML SnO within the gap, \[[Fig. 3(a)](#f3){ref-type="fig"}\], the new states it introduced reduce the band gap by 0.44 eV. In comparison with the O adsorption, the C adatom brings a deeper *n*-type and *p*-type mid-gap states with negligible effect on the matrix band gap \[see [Fig. 3(b,c)](#f3){ref-type="fig"}\].
Since all B, N and F adsorptions bring in magnetic moments to the system, it is neccesarry to explore their nature. [Figure 4](#f4){ref-type="fig"} displays spin and atom resolved DOS of the adsorbed systems. All B, N, and F atoms induce mid-gap states in the band gap of pristine SnO. Different from B and F, adsorption of N introduces a state just below CBM thus reduces the band gap by 0.24 eV. Adsorption of B, N, or F also results in the spin polarized DOS \[[Fig. 4(a--c)](#f4){ref-type="fig"}\] leading to noticeable magnetic moments less than 1 μ~B~. The spin polarized charge density is found to be mainly localized on the adatoms and nearby Sn atoms, see [Figs 4(d--f)](#f4){ref-type="fig"} and [5](#f5){ref-type="fig"}. In B adsorption case, all the mid-gap states are originated from B 2*p*~*y*~ and *2p*~*z*~ orbitals, while its 2*p*~*x*~ orbitals are mainly spread within conduction bands, see [Fig. 4(d)](#f4){ref-type="fig"}. In addition, the spin splitting for B 2*p*~*z*~ orbitals is less than that of 2*p*~*y*~ orbitals, which is in agreement with the spin charge density distribution as illustrated in [Fig. 5(a)](#f5){ref-type="fig"}. Although the adsorption site of C atom is similar to that of B atom, C atom has one more valence electron than B atom, which could possibly pair up with the electrons associated with Sn atom resulting in zero net magnetic moment. On the other hand, the mid-gap states are originated from 2*p*~*y*~, *2p*~*x*~ and *2p*~*z*~ orbitals of N atom \[[Fig. 4(e)](#f4){ref-type="fig"}\] and 2*p*~*x*~ orbitals of F atom \[[Fig. 4(f)](#f4){ref-type="fig"}\], respectively. In all these three cases, the spin splitting is mainly originated from the in-plane *p* orbitals, i.e. *p*~*x*~ and/or *p*~*y*~ orbitals and the hybridization of adatoms is formed with Sn 5*p* orbitals \[see [Fig. 4(g--i)](#f4){ref-type="fig"}\]. The strong hybridization between the adatoms and matrix Sn atom demonstrates the formation of covalent bond.
We note that all B, N and F atoms have odd number of 2*p* electrons while C and O has even number of 2*p* electrons. Therefore, the introduced magnetic moments are originated from the unpaired electrons for B, N and F adsorption cases. All B, N and F adatoms will attract the electron from the lone pair of Sn leaving an unpaired electron on the 2*p* orbitals of the native Sn atom, which contributes to the calculated magnetic moment. For example, B has *s*^2^*p*^1^ electron configuration attracting the lone pair of Sn atom, possibly forming Sn-B bond with a larger bond length of 2.518 Å. N has *s*^2^*p*^3^ configuration and will form bond with two native Sn atoms. F has *s*^2^*p*^5^ valence electron configuration and will attract one electron of the lone pair forming bond with one native Sn atom. Their bonding strength can be quantified by the binding energies as given in [Table 1](#t1){ref-type="table"} which is in line with their bonding lengths. In addition, the total magnetic moments they introduced monotonically increase for B-, N-, and F-SnO which is consistent with trend of their electronegativity values.
The spin charge density of B, N and F adsorbed cases is presented in [Fig. 5](#f5){ref-type="fig"}. The yellow region represents accumulation of spin density and the isovalue is 0.002 e/Å^3^. It is shown that the spatial extensions of spin density changes with atomic numbers: the higher the atomic number of the adsorbate, the less spin density distribution on the adatoms and more on matrix Sn atoms. This suggests that the magnetic moment is mainly contributed from the adatoms and their proximate Sn atoms.
Summary and conclusion
----------------------
In summary, we present a systematic study on the stable configurations, electronic structures, and magnetic properties of B-, C-, N-, O- and F-adsorbed SnO monolayers. It is found that all these adatoms can be chemically adsorbed on the surface of SnO substrate. The electronic structure of SnO monolayer can be widely tuned by the adsorbed adatoms. The B-, N- and F-adsorbed SnO monolayers exhibit magnetic ground states, and the contributions of B or N to the total magnetic moments are weaker than that of F. Our studies suggest that nonmetal atoms adsorption is an effective approach to tune the electronic and magnetic properties of SnO monolayer, which opens an alternative way for future optoelectronic and spintronic applications.
Calculations details
--------------------
Our calculations are performed using the Vienna Ab-initio Simulation Package (VASP)[@b40][@b41][@b42]. The valence and core interactions are treated by the projected augmented wave (PAW) method[@b43] and the exchange correlation energy was described by the generalized gradient approximation (GGA)[@b44]. The Kohn-Sham orbitals were expanded in a plane-wave basis with a cutoff energy of 500 eV and the exchange-correlation functional is treated by Perdew-Burke-Ernzerhof form generalized gradient approximation (GGA-PBE)[@b44]. The Brillouin zone (BZ) was sampled by 3 × 3 × 1 Monkhorst-Pack *k*-point grids. The convergence threshold for self-consistent field (SCF) energy is set at 10^−6^ eV, and all the atomic positions are fully optimized until the Hellman-Feynman forces are smaller than 0.01 eV/Å.
Additional Information
======================
**How to cite this article:** Tao, J. and Guan, L. Tailoring the electronic and magnetic properties of monolayer SnO by B, C, N, O and F adatoms. *Sci. Rep.* **7**, 44568; doi: 10.1038/srep44568 (2017).
**Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Material {#S1}
======================
###### Supplementary Information
This work was supported by Natural Science Foundation of Tianjin City (15JCYBJC52800), Technology Foundation for Selected Overseas Chinese Scholar (CG2015008001) and Natural Science Foundation of Hebei Province of China (E2016202023).
The authors declare no competing financial interests.
**Author Contributions** J.T. performed the calculations. L.G. proposed the idea and advised the calculation. All authors wrote the manuscript together.
{#f1}
{#f2}
{#f3}
{#f4}
{#f5}
###### Adsorption of adatoms with a \[He\] core on SnO monolayer. The bonding length (*R* ~Sn-adatom~ Å), binding energy (*E* ~B~, eV), magnetic stabilization energy (∆*E* ~M~, meV), total magnetic moment (μ~B~), the magnetic moment of adatom (μ~B~) and their surrounding Sn atoms (μ~B~).
Defects *R*~Sn-adatom~ (Å) *E*~B~ (eV) ∆*E*~M~ (meV) Total magnetic moment (μ~B~) Magnetic moment of adatom (μ~B~) Magnetic moment of surrounding Sn (μ~B~)
--------------------- -------------------- ------------- --------------- ------------------------------ ---------------------------------- ------------------------------------------
Nonmagnetic defects
O 1.864 −5.419 1.3 0.0 0.0 0.0
C 2.160 −3.495 14.3 0.0 0.0 0.0
Magnetic defects
N 2.006 −4.243 471.7 0.63 0.61 0.01
B 2.518 −1.765 280.8 0.41 0.26 0.09
F 2.080 −3.820 134.8 0.79 0.12 0.34
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Most protein sequences do not have an experimentally determined structure and at least 40% do not even have a sequence homolog with a known structure [@pcbi.1002175-Bork1]. Nevertheless, the current Protein Data Bank (PDB) [@pcbi.1002175-Berman1] is thought to represent structure space nearly exhaustively [@pcbi.1002175-Kihara1]--[@pcbi.1002175-Zhang2]. Therefore, for most proteins, a structural homolog that can serve as a "template" for modeling at least part of its structure is likely to exist. However, the degree of sequence similarity will generally be too low to allow a template to be detected or for an accurate sequence alignment to be found [@pcbi.1002175-Rost1]. A central problem is that current alignment methods based on dynamic programming (DP) [@pcbi.1002175-Waterman1] generate the unique "optimal" alignment (the alignment producing the highest score based on a residue-residue similarity score and a gap penalty), while the "correct" alignment (producing the most accurate model) is not guaranteed to be optimal in terms of this score at low sequence identity ranges.
Numerous variations of both the residue-residue similarity score and gap penalty have been developed to address these issues. Individual residue-based scoring functions have been replaced with more complex profile-profile [@pcbi.1002175-Edgar1]--[@pcbi.1002175-Tang1] and environment-dependent methods [@pcbi.1002175-Bowie1]--[@pcbi.1002175-Shi1]. Recognizing that affine gap penalties typically over-penalize long gaps, several studies have described the probability of a gap as a function of its length or location in the structure with the goal of penalizing it appropriately [@pcbi.1002175-Barton1]--[@pcbi.1002175-Goonesekere1]. Threading methods [@pcbi.1002175-Jones1], [@pcbi.1002175-Panchenko1] incorporate an energy term into the alignment procedure, but they face the drawback of not being compatible with the traditional DP algorithm [@pcbi.1002175-Madej1].
Even with these more sophisticated approaches, there are still many issues that will confound the generation of an accurate alignment. Moreover, it is generally necessary to consider an ensemble of alternative alignments in order to produce an accurate model at low sequence identity ranges. Such ensembles are frequently called "suboptimal" since by necessity they have lower scores than the optimal alignment produced by DP. A variety of suboptimal sequence alignment schemes have been reported. Waterman [@pcbi.1002175-Waterman2] produced an ensemble of alternative alignments by changing the dynamic programming algorithm to return all alignments with scores within a small difference, δ, from that of the optimal alignment. However, the difference between the DP scores of the correct alignment and the optimal sequence alignment can be significant, especially for remote homologues. Increasing δ until it encompasses the correct alignment often produces an unmanageably large ensemble. Keeping δ small returns a more reasonable number, but the alignments tend to deviate negligibly from the optimal alignment.
Saqi and Sternberg [@pcbi.1002175-Saqi1] adapted this approach to return a more diverse ensemble by penalizing an alignment that is similar to one previously determined. John and Sali [@pcbi.1002175-John1] used genetic algorithm operators to splice and re-combine alignments in order to achieve the same goal. Chivian and Baker [@pcbi.1002175-Chivian1] produced alternative alignments by systematically varying the parameters in their optimal alignment method. Each alignment in their returned ensemble was therefore "optimal" (*ie.* highest-scoring) under a different set of conditions. One problem faced by all suboptimal methods is how to adequately sample the gigantic space of possibilities. Jaroszewski *et al* [@pcbi.1002175-Jaroszewski1] sought to explore the size of alignment space by examining pairs of small and medium-sized proteins (seven or fewer template secondary structures). Even though only "significantly different" alignments were enumerated by disallowing gaps in template secondary structures and ignoring alignment variations in loop regions, tens of millions of alternative alignments were required in some cases to generate the correct one.
We describe here a new method to generate suboptimal alignments, S4 (**S**ampling **S**hifts in **S**econdary **S**tructures), that takes an approach that is fundamentally different from the standard dynamic programming algorithm. In validation tests that we describe below, we show that S4 is highly effective at producing an accurate alignment within a set of 100 top-ranked alternatives and can almost always produce such an alignment within a set of 1000 alternative alignments. The utility of the S4 approach is most evident when the query/template sequence identities are low, but S4 also improves accuracy when the homology is clear. Our results are shown to constitute a significant improvement over DP-based alternative alignment methods, which we show is due to unique features of the algorithm, in particular to the effective use of the 3-dimensional structure of the template. The ability to generate a small set of alignments likely to contain the correct one suggests that S4 offers the possibility of significantly improving the accuracy of homology models, extending the number of sequences that can currently be modeled based on existing structures in the PDB.
Results {#s2}
=======
A flowchart for the S4 algorithm is shown in [Figure 1](#pcbi-1002175-g001){ref-type="fig"}. The method starts by searching the DP matrix for a set of short, ungapped alignments bounded by individual template secondary structure elements (SSEs). The rationale is that whatever sequence similarity may exist between query and template will more likely be in SSEs than loop regions. To generate a global alignment, pairs from a high-scoring set of "primary" fragments are connected with lower-scoring "secondary" fragments. This is a crucial feature of S4. In particular, we find that correctly aligned fragments can generally be identified within a very small set of primary fragments, significantly reducing the combinatorial complexity of the alignment problem. This characteristic, combined with the requirement that alignments containing the fragments be structurally plausible (see [Materials and Methods](#s4){ref-type="sec"}), improves accuracy in regions where the relationship between the query and template is less clear. The constraints also allow S4 to remove many alignments from consideration through the application of filters that identify geometrically or energetically unreasonable alignments based on knowledge of the template structure. Filters are also applied to check for redundancy in order to ensure that the alignments represent unique regions of alignment space. (A detailed description of each step of the S4 algorithm and the filters applied is provided in the [Materials and Methods](#s4){ref-type="sec"}.)
{ref-type="sec"}); or simply being high-scoring (Score). (3) Starting at the N-terminus, the algorithm enumerates all connections to downstream primary and secondary fragments, resulting in a large ensemble of "fragment alignments". Alignment rules are tested (see [Materials and Methods](#s4){ref-type="sec"}) whenever any fragment is added to an alignment. (4) The number of fragment alignments is reduced by filtering with thresholds based on statistical energies, core contacts and a redundancy measure (see [Materials and Methods](#s4){ref-type="sec"}). (5) To generate a final global alignment from a set of fragments (e.g. the green line, a boundary is defined around each remaining fragment alignment (dashed lines) within which the traditional a DP-based suboptimal algorithm is used to find an ensemble of full alignments. DFIRE then selects the alignment with the lowest/best energy to represent the set of fragments. (6) The process continues until it has returned the top N alignments, ranked by their residue similarity score.](pcbi.1002175.g001){#pcbi-1002175-g001}
Improvement in alignment accuracy {#s2a}
---------------------------------
S4 was tested on a set of target sequences from the CASP [@pcbi.1002175-Moult1] experiments (T0129--T0359). Potential templates for each target/query sequence were identified by structurally aligning the native structure to other proteins in the PDB using the ska program [@pcbi.1002175-Petrey1], [@pcbi.1002175-Yang1]. Templates were then selected based on a set of criteria (see [Materials and Methods](#s4){ref-type="sec"}) to ensure that an alignment existed between the template and query structure that would produce a model with a TM-score [@pcbi.1002175-Zhang3] \>0.5, and also that S4 would not be run on sequences longer than 350 residues. The resulting test set contained 3,342 query sequence/template pairs and was heavily populated by those with low sequence identity: over 90% of all pairs had less than 20% identity and more than 60% had less than 10% identity. Overall, there were 137 queries with an average of 24 templates each that satisfied all the criteria. The queries represented at least 65 different SCOP folds (some targets are not classified in SCOP).
We define the correct alignment to be the structure-based sequence alignment between the query and template and evaluate the performance of S4 by comparing it to three DP-based approaches, HMAP [@pcbi.1002175-Tang1], hhalign [@pcbi.1002175-Sding1], and SP3 [@pcbi.1002175-Zhou1]. We also compare against a DP-based suboptimal alignment method [@pcbi.1002175-Waterman2]. We calculate the accuracy of an alignment in different ways. While an alignment algorithm should ideally be able to reproduce the structure-based sequence alignment residue-by-residue, several issues make this an overly sensitive measure of success. For example, consider a situation in which a template contains a helix with an axis that is at an angle with respect to that of the topologically equivalent helix in the query. Because of such differences between the template and query structures, no alignment in this region can be considered strictly correct even though there may be residues in the query and template that occupy roughly equivalent positions in space. The same difficulty occurs in the alignment of loop regions and also in β-strands, where β-bulges can affect alignment accuracy. However, it is clearly desirable for an alignment algorithm to pair residues in topologically equivalent SSEs, even if this pairing does not exactly correspond to the structure-based sequence alignment because of conformational differences.
Because of these issues, we use three measures that reflect a variety of characteristics. We first use a measure called "inter-alignment distance" (IAD). As described in [Materials and Methods](#s4){ref-type="sec"}, IAD corresponds to the average deviation of the position of residues in a given alignment from their correct position in the structure-based alignment. An IAD of 2 implies that, on average, each residue is shifted by two away from its position in the correct alignment, but also implies that topologically equivalent SSEs in the template and query have been correctly paired. Thus IAD is a measure of overall alignment quality. To calculate how well the correct alignment is generated on a residue-by-residue level, we use a measure that we call FDS2, adapted from the FD measure of Sauder *et al.* [@pcbi.1002175-Sauder1]. This measure is simply the percentage of residues that are within 2 from their position in the correct alignment, with the restriction that this is calculated only in regions corresponding to template SSEs. This restriction results in a more informative alignment metric, since measuring accuracy in the structurally equivalent---but conformationally dissimilar---loop regions of remote homologs imposes a correspondence of residues that is not necessarily meaningful. Finally, to determine whether the models produced from the alignments are actually useful, we directly compare models to the native structure using the TM-score [@pcbi.1002175-Zhang3].
[Figure 2](#pcbi-1002175-g002){ref-type="fig"} plots IAD for the best alignment generated by S4 and the single optimal alignment produced by HMAP, hhalign, and SP3. Points in the figure represent individual query/template pairs and are ordered according to the IAD of the optimal alignment generated by the different methods (i.e., moving left-to-right in the graph corresponds roughly with query/template pairs that range from higher to lower sequence identity). [Figure 2](#pcbi-1002175-g002){ref-type="fig"} illustrates a central difficulty with most DP-based alignment methods. That is, at the higher range of sequence identities, most methods produce a reasonably accurate alignment, but there appears to be a threshold beyond which an accurate alignment becomes impossible when considering a single, optimal alignment. On the other hand, S4 generates an alignment with improved accuracy at all sequence identity levels and the improvement is quite dramatic at lower identities when the optimal alignment is severely flawed.
{#pcbi-1002175-g002}
[Table 1](#pcbi-1002175-t001){ref-type="table"} presents this explicitly, showing IAD values for the different methods averaged over all pairs in several ranges of sequence identity. In the 0--5% identity range, the average IAD for the optimal alignment is over 13, implying that many topologically equivalent SSEs are not correctly paired. In contrast the average IAD for the best S4 alignment found in the top 1000 is 2.3, indicating that S4 is able to find good alignments even in the low identity regime. We note that this is true whether or not the template is identified as a significant hit by the individual methods (E-value\<10 for HMAP, E-value\<0.001 for hhalign and Z-score\<−0.5 for SP3). In [Figure 2](#pcbi-1002175-g002){ref-type="fig"}, the IAD\'s for the best S4 alignments are colored in light or dark green if the template for that case was identified as significant by the corresponding alignment method, and in red for those templates that are not considered significant.
10.1371/journal.pcbi.1002175.t001
###### Accuracy of S4 alignments at different levels of sequence identity.
{#pcbi-1002175-t001-1}
Average IAD
-------- ------ ------------- ----- ------
0--5 567 2.3 4.3 13.6
5--10 1460 1.5 2.3 6.2
10--15 585 0.9 1.0 1.9
15--20 275 0.5 0.5 0.7
20--30 167 0.4 0.4 0.4
30--50 84 0.3 0.3 0.3
ID Range (%) \# Pairs S4 1000 S4 100 hhalign
-------------- ---------- --------- -------- ---------
0--5 507 2.2 4.1 11.8
5--10 1273 1.5 2.1 6.2
10--15 629 0.9 1.0 2.6
15--20 314 0.5 0.5 1.2
20--30 160 0.3 0.4 0.6
30--50 76 0.3 0.3 0.3
ID Range (%) \# Pairs S4 1000 S4 100 SP3
-------------- ---------- --------- -------- ------
0--5 271 2.4 4.8 15.7
5--10 701 1.5 2.3 7.0
10--15 370 0.9 1.1 2.1
15--20 183 0.5 0.5 0.6
20--30 96 0.3 0.3 0.3
30--50 36 0.3 0.3 0.3
The average accuracy of S4 alignments compared to the DP-based optimal alignment programs HMAP, hhalign, and SP3, measured using inter-alignment distance (IAD). The IAD for S4 is based on the best available in an ensemble of either 1,000 or 100 alignments.
Of course, there is an inherent difficulty in comparing the performance of a method which generates an ensemble to a method which generates a single alignment. In fact, the optimal alignment is the most accurate in many cases (about 30% of the time for hhalign and 21% for SP3) and is more often than not in the top 5% in a set of 1,000 alignments ranked by IAD. The average rank is ∼200 however, so there is generally room for improvement, and our main point here is not that the optimal alignment shouldn\'t be used, but that an ensemble is necessary to generate an alignment that makes an accurate model, especially for highly remote/query template pairs. In practice, the optimal alignment would always be part of such an ensemble.
To determine the extent to which the improvement in alignment quality of S4 relative to the optimal alignment is due simply to the increased number of alignments generated, we also compared S4 to two versions of the conventional DP-based Waterman algorithm for generating alternative alignments [@pcbi.1002175-Waterman2], which have been implemented in-house as part of HMAP [@pcbi.1002175-Tang1]. [Figure 3](#pcbi-1002175-g003){ref-type="fig"} shows the results. As discussed above, while the IAD is effective at measuring overall alignment accuracy, it does not define the fraction of residues that are within a specified distance from their position in the correct alignment, and thus in [Figure 3](#pcbi-1002175-g003){ref-type="fig"} we use the FDS2 measure. We also compare to two versions of the DP-based suboptimal alignment algorithm. A problem with the strict implementation of this algorithm is that alternate alignments can be generated that are not meaningfully different because variations in loop regions produce essentially equivalent models. Thus, we also implemented a modified algorithm which ignores such alignment variations. In the Figure the standard implementation of the algorithm is referred to as "unconstrained Waterman" and the modified version is referred to as "constrained Waterman" (see [Materials and Methods](#s4){ref-type="sec"} and Figure S6 in [Text S1](#pcbi.1002175.s001){ref-type="supplementary-material"} for more detail).
{ref-type="sec"}). In the first panel, the query/template pairs are grouped based on the FDS2 of the optimal alignment (0.0--0.1, 0.1--0.2 ... 0.9--1.0), and in the second panel the groupings reflect the sequence identity of the pair (0--5%, 5--10% ... 30--40%, 40--50%). In these graphs, a higher FDS2 correlates with a more accurate model. The data points represent the average FDS2 over all pairs in each group, plotted as a function of the IAD of the optimal alignment. The averages are based on the best alignment in the ensemble generated by each method for a query/template pair. For S4, we examined the different ensemble sizes given in the inset legend and used an ensemble of 1,000 for the Waterman based approaches.](pcbi.1002175.g003){#pcbi-1002175-g003}
[Figure 3](#pcbi-1002175-g003){ref-type="fig"} depicts the best FDS2 in the ensemble from each method as a function of the FDS2 of the optimal alignment. The vertical distance above the dotted line represents the improvement over optimal for the best alternative alignment generated. S4 is seen to significantly outperform the DP-based optimal and suboptimal algorithms, particularly when the optimal alignment is flawed. Even the best alignment out of the top 100 S4 alignments is significantly better than the best out of 1000 from the other DP-based methods. A further improvement in accuracy can be obtained by modeling the ensemble of 1000 alignments and using the pG score [@pcbi.1002175-Mirkovic1], [@pcbi.1002175-Sanchez1], to select the top 100 alignments based on the quality of the models they produce. [Figure 3](#pcbi-1002175-g003){ref-type="fig"} also shows the same data as a function of sequence identity. Again, we see that S4 offers a significant improvement compared to all DP-based methods for aligning remote homologs, even when using an ensemble one-tenth as large.
Evaluation of models from S4 alignments {#s2b}
---------------------------------------
The results shown in [Figure 2](#pcbi-1002175-g002){ref-type="fig"} suggest that S4 generates alignments that are much improved over DP-based optimal methods, but since the IADs of the best S4 alignments are not 0 (i.e., the S4 alignments are not identical to the correct alignment) an important question is whether these improved alignments produce improved 3-dimensional models. To examine this, we made models from the optimal alignment, the correct, structure-based alignment and all alignments in each S4 ensemble for each pair in the data set. The models were then compared to the native structure using the TM-score [@pcbi.1002175-Zhang3] with results shown in [Figure 4](#pcbi-1002175-g004){ref-type="fig"}. It is evident from the figure that many of the models produced by S4 constitute a significant improvement over the one produced by dynamic programming. The improvement in model quality is most dramatic when the model produced by the optimal alignment is inaccurate. Notably, the best models from S4 are often quite close to the accuracy of the model from the correct alignment. The line labeled "S4 90%" represents the 90^th^ percentile cutoff within each segment, indicating that S4 produced a model for 10% of the pairs that was as accurate as possible, i.e., as good as the model produced by the correct, structure-based alignment.
{#pcbi-1002175-g004}
[Figure 4](#pcbi-1002175-g004){ref-type="fig"} also shows that evaluating models can significantly reduce the number of models that need to be considered. "S4 100 (pG)" represents the best model of the 100 top-ranked models in the ensemble as determined by the pG score. The proximity of this line to "S4 1000" demonstrates that the pG score consistently ranks the best model from the entire ensemble in the top 100. It is important to be able to reduce the ensemble size in this manner without removing the best models, if further processing of the models is to be carried out (i.e., refinement, minimization, etc.)
Sampling of alignment space {#s2c}
---------------------------
Since they use the same scoring function, the improved performance of S4 compared to HMAP seen in [Figure 3](#pcbi-1002175-g003){ref-type="fig"} is not due to better scoring, but to a broader sampling of alignment space while also avoiding regions that would produce poor alignments. The latter feature is achieved with the rules and filters discussed in [Materials and Methods](#s4){ref-type="sec"}. The ability of S4 to sample broadly should manifest itself in greater sampling at both the residue and whole alignment levels. Indeed, in [Figure 5A](#pcbi-1002175-g005){ref-type="fig"}, we see that S4 samples 3--5 times as many different query residues at each template position compared to the DP-based methods with the same ensemble size.
{ref-type="fig"}. The different ensemble sizes used for S4 are shown in the inset legend and an ensemble size of 1,000 was used for the Waterman based approaches.](pcbi.1002175.g005){#pcbi-1002175-g005}
In [Figure 5B](#pcbi-1002175-g005){ref-type="fig"}, we choose the structure-based sequence alignment as a reference and report the standard deviation of the IAD for all alignments in an ensemble. A low standard deviation indicates that many of the alignments in the ensemble are clustered around a particular distance from the correct alignment, which implies that they are in a narrow region of alignment space. For DP-based methods that region will be centered on the optimal alignment (see [Discussion](#s3){ref-type="sec"} below). We see in [Figure 5B](#pcbi-1002175-g005){ref-type="fig"} that S4 samples broadly within its small ensemble, but can still return an alignment closer to the correct alignment than the DP-based methods (see [Figure 3](#pcbi-1002175-g003){ref-type="fig"}).
Discussion {#s3}
==========
A specific example illustrates S4\'s approach to sampling alignment space. [Figure 6](#pcbi-1002175-g006){ref-type="fig"} depicts a query/template structure alignment along with a listing of their respective SSEs and several ways they are matched in the alignments produced by different methods. The query is the N-terminal domain of KaiA, a non-enzyme circadian clock protein [@pcbi.1002175-Williams1] and the template is a single domain of DXR, which is a reductoisomerase [@pcbi.1002175-MacSweeney1]. The two proteins are classified as belonging to different folds in SCOP [@pcbi.1002175-Murzin1] and have less than 2% sequence identity.
{#pcbi-1002175-g006}
Despite being classified as different folds, these two proteins have high overall structural similarity and thus an alignment exists that would generate an accurate model. The structural alignment for this pair describes the proper correspondence of all eight of the SSEs that are common between the template and query, as depicted in the first two rows of the alignment shown in [Figure 6](#pcbi-1002175-g006){ref-type="fig"}. The DP-based optimal alignment contains major flaws and only four out of eight SSEs are in proper correspondence. The poor performance of the DP-based approach is due more to issues with sampling alignment space than to the absence of a detectible sequence relationship between the two proteins. In fact, the eight fragments representing correct correspondences of query and template SSEs were all highly-ranked fragments (or "primary" fragments, in the terminology used in [Materials and Methods](#s4){ref-type="sec"}) as determined by the same HMAP scoring function. All eight correct fragments were chosen within the first 58 (out of a total of 122 used). This local similarity between the profiles is consistent with other local structural, functional and sequence similarities that have recently been described between proteins that have significantly different topologies [@pcbi.1002175-Petrey2], [@pcbi.1002175-Friedberg1].
Overall, out of an ensemble of 1,000 alignments, the best alignment from S4 has an IAD of 0.56 and an FDS2 of 97% compared to the correct alignment and the TM-score of the corresponding model is 0.50 (compared to a TM-score of 0.57 for the model built from the structure-based alignment). In contrast, the best alignment generated by the constrained Waterman approach (out of an ensemble of 1,000) had an IAD of 15.4. That the improvement in accuracy of S4 is due to differences in sampling can clearly be seen by calculating average IADs of the alignments in each ensemble, but here with respect to the DP-based optimal alignment instead of the correct alignment. The constrained Waterman approach is "trapped" near this incorrect alignment (average IAD of 0.3 and standard deviation of 0.8). Even though S4 samples the DP-based optimal alignment, it also searches far from this alignment (average IAD of 9.4, standard deviation of 4.6, and a maximum IAD of 25.4).
Though we have shown that S4 generates accurate alignments to almost every template appropriate for a given query sequence, we have not discussed how to identify these templates or how to select the correct alignment from the S4 ensemble. However, the results shown in [Figure 2](#pcbi-1002175-g002){ref-type="fig"} suggest that S4 can be a valuable component of currently used homology modeling strategies. That is, based on the results in [Figure 2](#pcbi-1002175-g002){ref-type="fig"}, most of the appropriate templates that we identify based only on structural similarity to the native structure are recognized as significant using the scoring function associated with the different methods we compare to in the figure. But for a significant majority of these templates an accurate alignment is not possible, at least considering a single alignment generated based on the techniques and information used in the different alignment strategies. This severely limits the number of templates which can be considered useful even if they are recognized.
By building models from templates selected by other methods, but based on alignments generated by S4, these templates can be exploited assuming an accurate model evaluation procedure can be applied. There is a wide array of such tools that range from measures of the suitability for residues to be in a given environment (e.g., Verify3D [@pcbi.1002175-Eisenberg1]), to statistical potentials such as D-FIRE, Prosa, or Anolea [@pcbi.1002175-Melo1], to all-atom molecular dynamic simulations to estimate the thermodynamic stability of the model (GROMOS [@pcbi.1002175-Scott1]). The choice of the best method of evaluation is a complicated one and goes beyond the scope of the current paper where we have focused on S4 as an alignment tool. Nevertheless, for a third of the cases used in our benchmarking, the model with the lowest pG score differs negligibly from the best possible model available from the ensemble (i.e., the best model and the model selected based on pG have TM-scores with respect to the native structure that are within 0.05 of each other). Further, it has been shown that construction of 3D models followed by evaluation using a statistical potential can be used to distinguish true from false homologs when the sequence relationship is ambiguous [@pcbi.1002175-Mirkovic1], [@pcbi.1002175-Lee1]. These results suggest that more accurate alignments obtained using S4 should significantly expand the number of good templates and models that can be found.
Since S4 produces accurate alignments in nearly every case where there is a structural similarity that leads to an accurate model, this suggests that, using a model-building and evaluation procedure, templates with scores that are outside the range of what is usually considered significant for a particular method could also be identified. Using the widely used tool PSI-BLAST as an example, about half of the templates in our data set were identified as significant (where we define this loosely as E-value\<10). As shown in [Table 2](#pcbi-1002175-t002){ref-type="table"}, in these cases S4 can generate more accurate alignments, in terms of the FDS2 score, than PSI-BLAST. Even for those templates with E-values that are not typically considered useful, (10^−3^\<E-value\<10), S4 is able to find an alignment that is more than twice as accurate and S4\'s performance decreases only slightly among the pairs that are not detected at all by PSI-BLAST, which comprise over half the benchmark set. The results shown in [Figure 2](#pcbi-1002175-g002){ref-type="fig"} indicate that the same conclusion holds no matter what the method used to identify templates. Moreover, preliminary work using a protocol in which templates are selected by PSI-BLAST, models are built from every alignment in the S4 ensemble and evaluated using the pG score as well as other criteria suggests that good templates in this E-value range can be identified with high precision.
10.1371/journal.pcbi.1002175.t002
###### Accuracy of S4 alignments for templates identified by PSI-BLAST.
{#pcbi-1002175-t002-2}
Psi-Blast E-value \# Pairs S4 FDS2 (1000) S4 FDS2 (100) Psi FDS2
-------------------- ---------- ---------------- --------------- ----------
EV\<10^−6^ 902 95.6 94.9 83.4
10^−6^\<EV\<10^−3^ 164 92.1 91.4 62.1
10^−3^\<EV\<10 316 88.8 86.1 41.9
No Hit 1516 82.0 76.0 N/A
Alignment pairs have been separated into three regions: clear homology (E-value\<10^−6^), intermediate homology (10^−6^\<E-value\<10^−3^), remote homology (10^−3^\<E-value\<10) and undetectable homology (template not among PSI-BLAST hits). A default maximum E-value of 10 was used in PSI-BLAST for all queries.
As shown in [Figure 5](#pcbi-1002175-g005){ref-type="fig"}, the primary difference between S4 and other alternative alignment methods is the manner in which alignment space is sampled. The central advantages of S4\'s sampling are that it generates enough diversity in a small ensemble so that an accurate alignment can be found, while limiting on the number of *potential* alignments that need to be considered (\<10 million, see [Materials and Methods](#s4){ref-type="sec"}). In contrast, as we show in [Figure 5](#pcbi-1002175-g005){ref-type="fig"}, DP-based sampling is highly local as a result of the fact that DP must start with the optimal alignment and successively generate other alignments in decreasing order based on their score. This severely limits the amount of diversity that DP can generate and ensures that many more alignments would need to be considered (at least an order of magnitude and probably more) when the DP-based score of the correct alignment is far below the optimal. A low DP score is typical for the more remote query/template pairs in our benchmark, since the correct alignments frequently require long indels or pass through low-scoring regions of the alignment matrix. Moreover, application of the structural filters used in S4 would not be expected to improve this situation, since there are a significant number of inaccurate alignments that satisfy them. Again, if an inaccurate alignment had a better DP-score than the correct one, a DP-generated ensemble would be trapped near the inaccurate alignment, since the local sampling inherent in DP would most likely not generate alignments that break the structural rules in any manageably small ensemble.
While it appears necessary based on our results to consider an ensemble in order to find an accurate alignment, especially for highly remote query/template pairs, it is clearly beneficial to consider the optimal alignment as well. As mentioned above, the optimal DP-based alignment is the most accurate (in an ensemble of 1,000 S4 alignments and 1 optimal alignment) for many cases in our benchmark. An ideal modeling strategy then would be one that generates an ensemble with S4 and simply adds the optimal alignment to that ensemble. This would ensure the best of both worlds at no increase in computational cost. Moreover, the S4 algorithm is independent of the underlying residue-residue scoring function employed. In the work presented here, the HMAP profile-profile method was used, but the sampling algorithm used in S4 could be applied using any other residue-residue scoring function. Therefore, if better scoring functions are available or if future improvements to scoring functions are able to raise the level of accuracy of the DP-based methods, S4\'s performance using the same scoring function should improve as well.
Materials and Methods {#s4}
=====================
Template selection for CASP targets {#s4a}
-----------------------------------
To ensure that a meaningful structural relationship existed within each query-template pair, several conditions had to be met: the protein structural distance (PSD) [@pcbi.1002175-Yang1] could not exceed 0.5 (corresponding to a maximum RMSD of about 3.5 for aligned residues); the sequence identity was less than 50%; and a "pseudomodel" of the query built from the aligned portions of the structure-based sequence alignment and based on the template structure had to return a TM-score [@pcbi.1002175-Zhang3] of 0.5 or greater compared against the native query structure. A pseudomodel is constructed by simply copying the backbone and Cβ coordinates of residues of the template mutated to the identities of the corresponding aligned residues in the query (unaligned residues are ignored). Also, proteins of length greater than 350 residues were not considered.
The Algorithm: Overview {#s4b}
-----------------------
The S4 algorithm is described in detail as six distinct steps below (see [Figure 1](#pcbi-1002175-g001){ref-type="fig"}). Overall, the algorithm proceeds as follows. First (Steps 1 and 2), short ungapped alignments entirely contained within template SSEs ('fragments') are selected based on their sequence similarity. Any subset of fragments, listed in order from N to C-terminal, is called a fragment alignment. Next, all fragment alignments are exhaustively enumerated and those that pass a set of tests for modelability, are kept. Finally, full alignments are constructed from fragment alignments. The full alignments are generated by standard dynamic programming with the constraint that DP is applied only to a narrow region (defined by the fragment alignment) of the dynamic program scoring matrix. A schematic for the different steps in the process is provided in [Text S1](#pcbi.1002175.s001){ref-type="supplementary-material"}, as well as a specific example of how S4\'s features lead to improvement in alignment accuracy.
The Algorithm: Selecting fragments (Steps 1 & 2) {#s4c}
------------------------------------------------
[Figure 1](#pcbi-1002175-g001){ref-type="fig"} shows a typical dynamic programming matrix with the query sequence along the side and the template sequence across the top. The template sequence is divided into columns defined by its secondary structure elements. A diagonal contained within a column is called a "fragment" and represents a short ungapped alignment of the query to the template. To start the alignment process, an initial set of "primary" fragments is identified as follows. Each fragment, (i.e., every diagonal in every column) is examined and is assigned a score that is the sum of the residue-residue similarity scores of the aligned pairs it contains, calculated based on the HMAP profiles [@pcbi.1002175-Tang1] of the query and template sequences. The fragment from each column with the highest normalized score (the profile-profile similarity score divided by the length of the fragment) is added to the list of "primary" fragments (black lines in [Figure 1](#pcbi-1002175-g001){ref-type="fig"}). Each template SSE will contain at least one primary fragment and usually several more.
For every pair of primary fragments we perform a recursive search for "secondary" fragments to fill in the region defined by the fragment endpoints, if the fragments in the pair belong to non-consecutive SSE\'s. For example, in [Figure 1](#pcbi-1002175-g001){ref-type="fig"}, two secondary fragments are chosen for being the highest scoring secondary fragments that are "adjacent" to primary fragments PF1 and PF2. (An adjacent fragment is contained in a neighboring SSE and is on the same or a nearby diagonal.) Other secondary fragments are chosen by virtue of being high-scoring or in an SSE whose deletion would violate the alignment rules (e.g., a missing core strand, see below). This process continues recursively until all regions between non-consecutive fragments in a subset have been filled in.
The Algorithm: Enumerating and filtering fragment alignments (Steps 3 & 4) {#s4d}
--------------------------------------------------------------------------
A "fragment alignment" is a list of primary and secondary fragments in order from the N- to C-terminal. Two examples of fragment alignments are shown in [Figure 1](#pcbi-1002175-g001){ref-type="fig"}. The blue and green lines both run alongside two sets of four fragments (which share a common first member). Fragment alignments such as these will later form the basis of full alignments (constructed as described below).
To enumerate all fragment alignments that are possible within our set of primary and secondary fragments, S4 connects the N-terminal pseudo-fragment (upper-left corner of [Figure 1](#pcbi-1002175-g001){ref-type="fig"}) to each downstream primary fragment (either directly or through subalignments of secondary fragments). This process progresses to further downstream fragments until all alignments end at the C-terminal pseudo-fragment (bottom-right corner of [Figure 1](#pcbi-1002175-g001){ref-type="fig"}). After any connection between fragments is established, a set of conditions must be met. If an alignment fails to meet one of these conditions (described below), the enumeration process is discontinued for that particular path. (Some conditions can only be applied when the C-terminal is reached). It should be noted the total number of possible fragment alignment can be calculated efficiently during the above process, and no new fragments are added once the total number of alignments exceeds 10 million.
Some of the conditions placed on the fragment alignments are based on the properties of the alignment itself and some are based on a 3D pseudomodel of the query. The conditions that must be met by each alignment/pseudomodel are described below.
### Coverage {#s4d1}
We are generally not interested in alignments that pair a very small number of residues. Therefore, only alignments where at least 10% of the shorter sequence is aligned to the longer sequence are retained. Since only residues in template SSEs are counted in S4, this fraction represents a somewhat more stringent condition than it may initially appear.
### Contact order {#s4d2}
The contact order for a pseudomodel is defined here as the percentage of its SSE residues containing a Cβ that lie within 6 Å of a Cβ from a residue in a different SSE. Fragment alignments whose pseudomodel has a contact order less than 65% of the contact order of the template itself are rejected. Making this threshold relative to the template ensures that "extended" models will not be built from compact templates, but if the template itself is extended, the fragment alignment will be kept.
### Strand pairing {#s4d3}
There are two general rules governing the pairing of beta strands in homologous proteins that can be used to eliminate bad alignments [@pcbi.1002175-Russell1]. First, a paired strand in the template should not become unpaired in the pseudomodel. Second, a core strand of a beta sheet in the template must be present in the pseudomodel if its flanking strands are also present.
### Loop lengths {#s4d4}
Fragment alignments are rejected if there are not enough residues in the query sequence to bridge the gap between any two consecutive fragments. Specifically, we require that , where *q~p~* is the index of the final query residue of the fragment preceding the loop, *q~f~* is the index of the first query residue following the loop, and *d(t~p~,t~f~)* denotes the distance (in Å) in the template structure between the Cα atoms of the corresponding, aligned template residues. The factor of 3.3 was determined by studying a database of several hundred high-resolution structures. It was found that the maximum distance traversed by a loop was slightly over 3 Å per residue, which is, of course, roughly the length of an individual amino acid. We used 3.3 to allow our algorithm to keep some pairings of fragments whose loops would normally be 'over-stretched'. The purpose of this test was to remove only blatantly incorrect fragment pairings, since loops that were just slightly over-stretched may be fixed when the full alignment is found in Step 5 of the algorithm.
Three other measures were used to eliminate fragment alignments that are unlikely to produce good models: preserved core contacts, query energy and template energy. For an alignment to be kept, all three of these measures must have values above the 66^th^-percentile for each measure and one of these three values had to surpass the 90^th^-percentile. The measures are listed below.
### Preserved core contacts {#s4d5}
A pair of residues in the template structure is considered to be a "core contact" if both residues in the pair are buried (60% or more of surface area inaccessible), have Cβ atoms that are within 6 Å and are both hydrophobic (amino acid types A, F, G, I, L, M, P, W, V and Y). An alignment that pairs hydrophobic amino acids in the query with template residues in a core contact generates a preserved core contact.
### Statistical energy of query residues {#s4d6}
An implementation of the DFIRE statistical potential [@pcbi.1002175-Zhou2] was used to evaluate each alignment by using the Cα and Cβ positions from the template with the amino acid types of the aligned query residues. A pseudo-Cβ position was determined for glycine residues based on the Cβ position in alanine. Loop residues were not considered in either the calculation of the statistical energies or in the tabulation of inter-residue distances that form the basis of this implementation of DFIRE. The value thus calculated, called the "query energy", and the proximity of the alignment to the correct one were found to be highly correlated.
### Statistical energy of template residues {#s4d7}
Similar to evaluating the statistical energy of the pseudomodel, we calculate the energy of the aligned template residues, which we term the "template energy". In effect, this is the statistical energy of a subset of the template structure. The motivation behind this is to recognize and remove alignments that pair query residues with an unlikely combination of template SSEs. This often occurs when the template is a multi-domain protein and the query is a single domain. In these cases, the highest scoring fragments may be spread out across multiple domains of the template in a structure that does not resemble a folded protein. Calculating this value allows S4 to eliminate many such alignments.
### Redundancy {#s4d8}
Lastly, to decrease the redundancy of the final results, some fragment alignments are removed due to their similarity to a higher-scoring alignment. Fragment alignments are considered redundant if they align to the same template SSEs, have all corresponding fragments within a shift of 4, and an inter-alignment distance (IAD) of less than 1.
The Algorithm: Constructing full alignments (Steps 5 & 6) {#s4e}
---------------------------------------------------------
At this stage in the process, no full alignments in the normal sense have been created, only fragment alignments, which are just lists of fragments. A round of alignment sampling using the full sequences of the query and template is used to generate a final alignment from each fragment alignment. In this final step, alignments are restricted to a specific region of the dynamic program matrix. The boundaries of the region extend 3 residues above and below the fragments in each fragment alignment. The loop regions are constrained only by the boundaries of the surrounding fragments (dashed lines in [Figure 1](#pcbi-1002175-g001){ref-type="fig"}). Alignment sampling is carried out using the constrained Waterman approach. That is, we apply this algorithm in regions of alignment space that we expect to be unique based on the structure of the template. Again, a pseudomodel is constructed for each alignment which is scored with DFIRE [@pcbi.1002175-Zhou2] as described above. The alignment with the best/lowest energy is selected to represent the original fragment alignment.
The S4 algorithm typically generates thousands of fragment alignments. A single, full alignment is generated for each one, starting with the highest-scoring, until N unique alignments have been found, where N is the ensemble size chosen by the user. The score of an alignment is simply the sum of the similarity scores of the paired residues in the original fragment alignment minus a flat penalty for each inserted residue. The insertion penalty only applies to residues inserted between template residues and is therefore used to encourage insertions at the termini. Deletions are not penalized since we found that structural considerations enabled us to disallow unreasonable gaps without an explicit penalty. A worked example illustrating each step is provided in [Text S1](#pcbi.1002175.s001){ref-type="supplementary-material"}.
Inter-alignment distance (IAD) {#s4f}
------------------------------
We calculate the distance between any two alignments using a measure similar to the gALD measure developed by Chen and Kihara [@pcbi.1002175-Chen1]. If we plot two alignments of the same two sequences on the dynamic programming matrix ([Figure 1](#pcbi-1002175-g001){ref-type="fig"}, blue and green lines) there is a region between them for which we can calculate the area. Dividing this area by the length of the template yields an average height of this region, which can be interpreted as the average distance that a query residue in one alignment is shifted from its position in the other. This average distance we have termed the IAD and it should be considered to have units of residues. This measure is quick to calculate and useful for determining if two alignments occupy the same region of alignment space.
The standard DP-based alternative alignment methods {#s4g}
---------------------------------------------------
The unconstrained Waterman and constrained Waterman in [Figures 3](#pcbi-1002175-g003){ref-type="fig"} and [5](#pcbi-1002175-g005){ref-type="fig"} are implementations of the method described by Waterman. [@pcbi.1002175-Waterman2]. The "unconstrained Waterman" approach is an unmodified version of that algorithm that that use the HMAP scoring function and gap penalty and generates alternate alignments by allowing the DP procedure to branch to an alternate path at any point in the DP matrix where doing so will lead to an alignment with a score within δ of optimal. However, in the constrained Waterman approach, branching to alternate paths is allowed only when moving between SSE and loop regions (see Figure S6 in [Text S1](#pcbi.1002175.s001){ref-type="supplementary-material"} for more details). For both methods, it is impossible to know which value of δ will generate an ensemble of a desired size. To generate the alignments for comparison, we started with very small values for δ and increased it until the ensemble size exceeded 1000. We then sorted the alignments by their DP-based score and kept only the top 1000.
Model building and model accuracy {#s4h}
---------------------------------
Models were built with the program Nest [@pcbi.1002175-Petrey1] for all S4 alignments, the optimal HMAP alignment and the correct/structure-based alignment. TM-score [@pcbi.1002175-Zhang3] was used to evaluate the accuracy of the model compared to the native query structure.
Supporting Information {#s5}
======================
######
Supplemental information for "Using structure to explore the sequence alignment space of remote homologs".
(DOC)
######
Click here for additional data file.
The authors have declared that no competing interests exist.
This research was funded through grants from NIH ([www.nih.gov](http://www.nih.gov)): GM008281 GM030518. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
[^1]: Conceived and designed the experiments: AK BH DP. Performed the experiments: AK. Analyzed the data: AK BH DP. Contributed reagents/materials/analysis tools: AK DP. Wrote the paper: AK BH DP.
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the manuscript and its Supporting Information files.
Introduction {#sec001}
============
Effective humoral immunity is contingent upon the phenomenal diversity of antibodies. In mammals, this is derived via genetic recombination of numerous variable (V), diversity (D) and joining (J) gene segments localised to heavy, kappa and lambda immunoglobulin loci. In recent years, the capacity to clone and express antibodies from single B cells has proved a powerful tool to study antibody repertoires in a variety of infectious disease settings in humans \[[@pone.0233794.ref001]--[@pone.0233794.ref004]\], and important animal models such as mice \[[@pone.0233794.ref005], [@pone.0233794.ref006]\] and non-human primates \[[@pone.0233794.ref007], [@pone.0233794.ref008]\]. These approaches have subsequently been extended using next-generation sequencing platforms (reviewed in \[[@pone.0233794.ref009], [@pone.0233794.ref010]\]), allowing unprecedented depth in the characterisation of anti-pathogen antibody responses.
The domestic ferret *(Mustela putorius furo)* is a critical mammalian model to study pathogenesis and evaluate vaccines against a variety of human respiratory pathogens (reviewed in \[[@pone.0233794.ref011]\]), most critically influenza. However, the majority of influenza research using ferrets is focused upon viral transmission and/or pathogenesis, with in-depth immunological studies limited by a limited understanding of the ferret immune system (reviewed in \[[@pone.0233794.ref012]\]). A key knowledge gap surrounds the immunogenetics of ferret immunoglobulins. While the ferret genome was recently sequenced \[[@pone.0233794.ref013]\] accurate annotation of germline immunoglobulin genes is currently incomplete. This has hindered the ability to sequence ferret B cell receptors and/or allow the recovery of ferret monoclonal antibodies, limiting detailed interrogation of ferret serological responses that informs current influenza vaccine strain selection efforts.
Here we sought to increase the utility of ferrets for studying humoral immunity. Ferret heavy, kappa and lambda immunoglobulin loci were annotated using available genomic sequences, allowing the design of a novel set of multiplex PCR primers flanking recombined ferret immunoglobulin genes. Recombined B cell receptor sequences were recovered from single sorted ferret B cells, partially confirming our initial gene segment annotation and allowing identification of potential novel germlines. Ferret immunoglobulin constant gene sequences were confirmed using *de-novo* assembly of RNA-seq transcripts, allowing the design of expression plasmids and the recombinant production of ferret IgG monoclonal antibodies. In summary, we present a single-cell, RT-PCR based approach for recovery of B cell receptor immunoglobulins from ferret B cells and the recombinant production of ferret monoclonal antibodies in vitro, analogous to methodologies in widespread use in rodents and primates.
Materials and methods {#sec002}
=====================
Annotation of ferret immunoglobulin loci {#sec003}
----------------------------------------
Ferret genomic contigs containing potential immunoglobulin genes were retrieved from *e*!*Ensembl* ([http://www.ensembl.org](http://www.ensembl.org/)). (Immunoglobulin heavy loci---GL897360.1, GL897427.1, GL897453.1, GL897498.1, GL897556.1, GL897558.1, GL897564.1, GL897795.1, GL898421.1; kappa loci---GL896905.1; lambda loci---GL897406.1, AEYP01111698.1, GL896906.1, AEYP011112098.1, GL897285.1, GL897406.1, GL897344.1, GL897565.1, GL897418.1, AEYP01110728.1, AEYP01108526.1, GL897638.1 GL897285.1, GL897484.1, GL897019.1, GL897400.1). Iterative BLAST searches using human, and then ferret immunoglobulin gene segments were used to identify and annotate putative germline genes. Ferret immunoglobulin gene sequences were analysed with reference to human, mouse or canine databases using IMGT/V-Quest \[[@pone.0233794.ref014]\] and assigned to mammalian clans based upon phylogenetic analyses. Sequences with nonsense mutations and/or non-functional regulatory elements were considered pseudogenes. Phylogenetic relationships of functional V genes were determined based on the Jukes-Cantor model. Consensus phylogenetic trees were built using the Neighbour-Joining method with no outgroups and resampled by bootstrapping using Geneious tree builder (10.1.3). Ferret V, D, J and constant gene sequences have been uploaded to Genbank.
Flow cytometric sorting of single ferret B-lymphocytes {#sec004}
------------------------------------------------------
Ferret studies and related experimental procedures were approved and conducted in accordance to the University of Melbourne Animal Care and Use Standards by the relevant ethics committee (\#CT-FER-17-05). Single cell suspensions were prepared from the spleen of immunologically naïve ferrets. PBMCs were purified using 95% Ficoll-Paque Plus and cryopreserved in heat-inactivated fetal calf serum (FCS) containing 10% dimethylsulfoxide (DMSO). Cryopreserved ferret PBMCs were thawed, stained with Live/Dead Fixable Aqua (Thermo Fisher), surface stains anti-CD11b-BV510 (Biolegend: clone M1/70), anti-CD8-BV450 (Thermo Fisher: clone OKT8) and anti-ferret IgA/IgM/IgG-FITC (Rockland Immunochemicals cat.618-102-130). Stained cells were resuspended in OptiMEM (Thermo Fisher) before single, live, surface Immunoglobulin positive B cells were sorted into 96-well PCR plates and stored at -20°C.
For the recovery of antigen-specific ferret B cells, a single ferret was infected with 1000 TCID~50~ of H1N1 A/California/04/2009 and a single cell suspension of parapharyngeal lymph node cells (pLN) was prepared at 28 days post-infection and cryopreserved in heat-inactivated FCS containing 10% DMSO. Cells were subsequently thawed and stained with Live/Dead Fixable Aqua (Thermo Fisher), surface stains anti-CD11b-BV510 (Biolegend: clone M1/70), anti-ferret IgA/IgM/IgG-FITC (Rockland Immunochemicals cat.618-102-130), anti-CD8 eFluor450 (eBioscience Clone OKT8) and a prototype IgD Mab conjugated to APC-Cy7,. Biotinylated recombinant full length A/California/04/2009 hemagglutinin (HA) probes \[[@pone.0233794.ref015]\] conjugated to streptavidin-PE or streptavidin-APC (Invitrogen) were used to sort single HA-specific B cells into 96-well plates and stored at -20°C.
Ferret B cell receptor (BCR) sequencing {#sec005}
---------------------------------------
A B cell receptor (BCR) sequencing protocol was developed based upon the multiplex nested RT-PCR approaches previous described for humans \[[@pone.0233794.ref001]\] and mice \[[@pone.0233794.ref005]\]. Reverse transcription of total cellular RNA from single sorted ferret B cells was performed in 25μL reaction volumes in the sort plate using 450 ng random hexamers (Bioline), 50 U Superscript III (Thermo Fisher), 1X First Strand buffer (Thermo Fisher), 8 U RNAsin (Promega), 0.125pmol Dithiothreitol (DTT) (Thermo Fisher), 0.8% v/v IGEPAL CA-630 (Sigma Aldrich) and 0.8mM deoxynucleotide triphosphate (dNTP) (Bioline). Cycling conditions for cDNA synthesis were: 42°C for 10 min, 25°C for 10mins, 50°C for 60 mins and 94°C for 5min. 3μL of cDNA was used as template in multiplex nested PCR reactions to amplify paired recombined ferret heavy and light chain (IgK, IgL) sequences. Primary reactions were carried out in 50μL volumes using Hotstart Taq plus polymerase (Qiagen), 1X reaction buffer, 2.0 mM MgCl~2,~ 0.1mM dNTP (Bioline) and 5 nmol each of primary forward and reverse primer pools ([S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}). 4μL of primary PCR product was used as template in a secondary, nested PCR (50 μL volume) containing 1 X reaction buffer, 1.5mM MgCl~2~, 1 x solution Q and 5 nmol secondary forward and reverse primer pools ([S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}). For recovery of antigen-specific B cells from an infected ferret, the protocol was amended by substituting primary and secondary heavy chain reverse primer pools with 5 nmol reverse primer binding in the heavy chain joining gene (IGHJ) ([S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}). Cycling conditions for both heavy and light chain amplification are as follows: Primary PCR: 95°C for 5 mins, 50 cycles of 95°C for 30s, 55°C for 30s and 72°C for 55s. Final extension for 7 mins at 72°C. Secondary PCR: 95°C for 5 mins, 50 cycles of 95°C for 30s, 50°C for 30s and 72°C for 60s. Final extension for 7 mins at 72°C. PCR products were visualised by agarose gel electrophoresis prior to standard purification and standard sanger sequencing using the reverse constant chain primer (IgM, IgK, IgL) from the secondary amplification steps or the IGHJ reverse primer.
Immunoglobulin gene sequence analysis {#sec006}
-------------------------------------
Full length V~H~, V~κ~ and V~λ~ sequences recovered from single sorted ferret B-cells were analysed using IMGT/High V-Quest. As well-validated germline ferret immunoglobulin sequences are not yet available, sequences were compared against dog (*Canis lupus familiaris*) germline immunoglobulin sequences to identify V(D)J segments with the highest identity. CDR-H3 sequences were determined by counting amino acid residues immediately after framework region 3 starting from the conserved cysteine (C) residue and ending with a conserved tryptophan (W) or phenylalanine (F).
De-novo transcriptome assembly by RNA-Seq {#sec007}
-----------------------------------------
RNA was extracted from five million cryopreserved PBMCs derived from a single ferret spleen using RNeasy Plus micro Kit (Qiagen). mRNA libraries were prepared using Illumina Truseq Stranded mRNA kit and 100bp single end reads were obtained using an Illumina HiSeq 3000. Analysis was performed using Galaxy ([https://usegalaxy.org](https://usegalaxy.org/)) \[[@pone.0233794.ref016]\]. Sequences were filtered (Q\>30) and trimmed (Trim) to remove illumina adapter sequences. Contigs were assembled *de-novo* with \> 40 bp minimum read overlap for path extension using Trinity \[[@pone.0233794.ref017]\]. Contigs were aligned to ferret genomic sequences or alternatively, canine immunoglobulin constant region homologues, using MAGIC-BLAST \[[@pone.0233794.ref018]\] to identify putative constant genes. Ferret constant region gene sequences were then aligned to human and canine homologues using Geneious 10.1.3.
Cloning and expression of chimeric human/ferret and fully ferret monoclonal antibodies {#sec008}
--------------------------------------------------------------------------------------
Validated ferret IgG, and kappa and lambda constant domain sequences were synthesized (Geneart) and cloned into established human IgG1 and light chain eukaryotic expression vectors \[[@pone.0233794.ref001]\] using SalI/BamHI (IgG1), XhoI/BamHI (Lambda) or BsiWI/BamHI (Kappa) restriction digests, agarose gel purification and T4 DNA ligation (NEB:New England Biolabs).
Recombined heavy chain (VDJ) and lambda chain (VJ) sequences of the human anti-HA antibody CR9114 \[[@pone.0233794.ref019]\] or of recovered HA-specific ferret antibodies were synthesized (Geneart) and cloned into ferret IgG1 and lambda chain expression vectors using flanking AgeI/SalI (IgG) and AgeI/XhoI (lambda) restriction enzymes (NEB). Plasmid DNA was prepared using standard MIDI or MEGA kits (Qiagen).
For recombinant expression, heavy and lamdba chain expression plasmids were co-transfected into Expi293F cells using Expifectamine (Thermo Fisher). Briefly Expi293F cells were grown to a density of 2.5 x 10^6^ cells/mL and transfected with 100 μg each of heavy and light chain expression plasmids complexed with 270 μL of Expifectamine. After addition of transfection enhancers 16 hours post-transfection, culture supernatants were harvested 5 days post-transfection, clarified by high speed centrifugation (5000 g, 4°C, 15 min) and filtration (0.22μm). Antibodies were purified using Pierce Protein A agarose (Thermo Fisher). Briefly, supernatants were equilibrated with Protein A IgG binding buffer and loaded into purification columns packed with 0.2mL of equilibrated Protein A agarose. Bound antibody was washed using Protein A IgG binding buffer (Thermo Fisher), and eluted with 15mL IgG elution buffer (Thermo Fisher) and neutralised with 1.5mL 1M TE buffer pH8.0 (Merck). Antibodies were concentrated via centrifugation (100 kDa Amicon; Merck Millipore at 3000g, 4°C, 15 min), resuspended in PBS and analysed using SDS-PAGE. Samples were denatured in 1x SDS loading buffer containing 5% v/v β-mercaptoethanol at 95°C for 15 min, resolved using 16.5% precast polyacrylamide gel (Biorad) in 1 x SDS buffer (Biorad), fixed (40% Ethanol; 10% acetic acid) and stained with colloidal Coomassie blue (Biorad).
Influenza hemagglutinin (HA) ELISA {#sec009}
----------------------------------
96-well ELISA plates were coated with 400ng per well of full length recombinant CA09 HA at 4°C overnight. Wells were then blocked with 2.5% BSA/PBS for 1h at room temperature and washed with PBS-Tween (0.05%) five times. Four-fold serial dilutions of recombinant monoclonal antibodies (starting at 100 μg/mL), or ferret serum samples (starting at 1 in 100 dilution) added at room temperature for 1h. Detection was performed by sequential staining with donkey anti-ferret IgG (Rockland cat. 618-101-012) and goat anti-donkey-HRP (Abcam cat. Ab6988) at 1:2000 and 1:1000 dilutions respectively for 30 min each. OD~450~ readings were obtained after addition of Sureblue TMB peroxidase substrate (Seracare) and TMB BlueSTOP Solution (Seracare).
Viral escape asssay {#sec010}
-------------------
The generation of viral escape mutants was based upon previously described protocols \[[@pone.0233794.ref020]\]. Briefly, 24-well plates were seeded with 2.5 x 10^5^ Madin Darby Canine Kidney (MDCK) cells per well to form confluent monolayers. The next day, serial dilutions of recombinant ferret mAbs were incubated with A/California/04/2009 virus for one hour at 37°C in Flu-media (Dulbecco\'s Modified Eagle\'s Medium (DMEM) with 0.8% bovine serum albumin (BSA), 1% penicillin/streptomycin and 0.1% L-1-Tosylamide-2-phenylethyl chloromethyl ketone (TPCK)-treated trypsin), before adding the virus-antibody mixture to MDCK cells. After 2 to 3 days culture media supernatants from wells with visible cytopathic effect were collected and used to infect a fresh monolayer of MDCK cells in the presence of increasing concentrations of mAb. After serial passaging, culture supernatants were harvested, viral RNA was extracted and the HA gene sequenced. Putative mutant viruses were identified based upon sequence comparison to similarly passaged media-only controls.
Results {#sec011}
=======
Annotation of ferret germline variable, diversity and joining genes {#sec012}
-------------------------------------------------------------------
The initial publication of the ferret genome was reported in 2014 \[[@pone.0233794.ref013]\]. However, the assembly and annotation of immunoglobulin loci is currently incomplete. We therefore identified and annotated genomic contigs containing potential heavy chain variable (IGHV) genes using iterative rounds of BLAST. A total of 27 unique IGHV-like genes were initially identified, including potential pseudogenes and/or non-functional segments. 20 IGHV genes retaining an open reading frame (ORF), downstream recombination signal sequence (RSS; heptamer and nonomer) and critical amino acid residues (eg. Cys 74) were considered potentially functional, although 4 of these lacked identifiable functional leader exons. Based on DNA sequence homology, three groupings of immunoglobulin genes, corresponding to the three vertebrate "clans" \[[@pone.0233794.ref021]\] could be delineated ([Fig 1A](#pone.0233794.g001){ref-type="fig"}): clan I (3 genes), clan II (1 gene) and clan III (16 genes). In line with reports from other *Carnivora* such as dogs \[[@pone.0233794.ref022]\] and cats \[[@pone.0233794.ref023]\], the majority of IGHV gene diversity in ferrets lies within Clan III (human IGHV3-21 or 3--69 homology), with sequence variance concentrated within the CDR-H1 and CDR-H2 regions ([Fig 1B](#pone.0233794.g001){ref-type="fig"}). Based upon the conserved arrangement of RSS sequences, we identified 7 putative D gene segments (numbered IGHD1-7) ([Fig 1C](#pone.0233794.g001){ref-type="fig"}), of which 3 appear orthologous to canine, murine and/or human variants. Similarly, 5 putative germline J gene segments (IGHJ1-5) were identified ([Fig 1D](#pone.0233794.g001){ref-type="fig"}) including 2 conserved orthologues.
{#pone.0233794.g001}
The 0.7Mb kappa chain locus is contained within a single contig (contig [GL896905.1]{.ul}) allowing identification of 48 potentially functional germline IGKV genes ([Fig 2A](#pone.0233794.g002){ref-type="fig"}) using similar criteria as IGHV genes, in good agreement to previous computational approaches \[[@pone.0233794.ref024]\]. Based upon sequence homology with human and mouse genes, these could be divided among three clans; II (44 genes), I (2 genes) and III (2 genes) ([Fig 2B](#pone.0233794.g002){ref-type="fig"}).
{#pone.0233794.g002}
Partial contigs bridging the lambda chain locus were extracted and analysed for lambda variable gene segments ([Fig 2C](#pone.0233794.g002){ref-type="fig"}). 32 potentially functional IGLV gene segments were identified, spanning four probable clans; I (2 genes), II (22 genes), IV (2 genes) and V (6 genes). Five potential kappa joining gene segments (IGKJ) and four lambda joining gene segments (IGLJ) were identified based on sequence homology to canine variants. Functionality was inferred based upon the presence of canonical junction F/W-G-X-G motifs, RSS elements and 5' donor splice sites (GTRAGT) in the adjacent intronic sequences. All putative heavy, kappa and lambda gene segments are provided in [S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}.
Single cell RT-PCR for recovery of ferret immunoglobulin heavy chain sequences {#sec013}
------------------------------------------------------------------------------
To investigate the expressed repertoire of germline immunoglobulin gene segments, we used the above sequence information to design multiplex PCR primers ([S2 Table](#pone.0233794.s002){ref-type="supplementary-material"}) targeting conserved leader and constant regions of recombined heavy, kappa and lambda mRNA transcripts. Using an approach analogous to other mammals \[[@pone.0233794.ref001], [@pone.0233794.ref005]\], single ferret B cells were sorted from cryopreserved splenocyte preparations by flow cytometry using a simple antibody panel and gating scheme ([Fig 3A](#pone.0233794.g003){ref-type="fig"}). cDNA was generated from each cell and immunoglobulin sequences recovered by nested multiplex PCR. A total of 121 functional recombined heavy chain sequences (all using a mu constant chain) were recovered from 480 sorted ferret B cells from three genetically outbred ferrets (\~25% recovery). In line with the frequency of germline gene segments, the majority of immunoglobulins were derived from III clan genes, with only two sequences recovered from clan II and a single example of a sequence from clan I ([Fig 3B](#pone.0233794.g003){ref-type="fig"}).
{#pone.0233794.g003}
An accurate assessment of germline IGHV gene utilisation is difficult due to (a) the current inability to segregate naïve versus B cells that are somatically mutated (memory) during the sort and (b) the limited genomic information surrounding the ferret IGHV locus and (c) a poor understanding of any allotypic variation within these outbred animals. Nevertheless, we recovered clusters of IGHV gene sequences that were highly conserved (\>99% homology) to predicted germlines including examples from two of the three clans (II-HV1, III-HV7, III-HV8, III-HV9, III-HV12, III-HV16). A single clan I sequence was recovered, albeit with more limited homology to the putative germline gene (97.6%). The 18--20 IGHV genes we found within genomic contigs is less than observed in felines (24 genes) and canines (38 genes) \[[@pone.0233794.ref024]\], suggesting additional variable germlines may remain undiscovered. Supporting this, we repeatedly recovered multiple heavy chain immunoglobulin sequences that shared identical IGHV gene sequences, but recombined with different D and J genes, strongly suggestive of a common germline progenitor (III-HV17, III-HV18, III-HV19 and III-HV20, III-HV21). The existence and identity of these additional putative germlines (listed in [S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}) will be clarified as genomic sequencing of ferrets continues. Utilisation of all five predicted IGHJ gene segments and 6 of 7 IGHD gene segments (not IGHD6) was evident within recombined BCR sequences. CDR-H3 regions, often a critical determinant for antigen recognition, ranged from 5 to 25 amino acids (mean 13.2) in length ([Fig 3C](#pone.0233794.g003){ref-type="fig"}); broadly comparable to canines \[[@pone.0233794.ref025]\] and potentially shorter on average than observed in humans \[[@pone.0233794.ref026]\].
Recovery of ferret immunoglobulin light chain sequences {#sec014}
-------------------------------------------------------
For ferret immunoglobulin light chains, 99/480 (20.6%) functional and recombined kappa sequences were recovered, all of which were derived from variable gene segments belonging to clan II and recombined with 4 of 5 predicted IGKJ gene segments (not IGKJ2) ([Fig 4A](#pone.0233794.g004){ref-type="fig"}). Germlines II-KV13, II-KV18, II-KV19 II-KV20, II-KV21, II-KV22. II-KV23, II-KV29, II-KV36, II-KV40, II-KV41, II-KV42 and II-KV43 were matched to the sequences (\>99% identity) found *in silico*. Four additional novel germlines were evident within recovered sequences (II-KV45-48) ([S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}). Given the kappa locus is fully annotated, these additional germlines most likely represent allelic variants of clan II germlines that differ between the outbred ferrets used for genomic sequencing versus B cell sorting.
{#pone.0233794.g004}
For lambda chain immunoglobulins, 144/384 (37.5%) productive and recombined sequences were recovered, most homology to annotated clan II and clan I sequences (II-LVI1, II-LV2, II-LV4, II-LV5, II-LV6, II-LV7, II-LV8, II-LV11, II-LV12, II-LV14, II-LV15, II-LV16, II-LV17, II-LV18, II-LV20, I-LV1; [Fig 4A](#pone.0233794.g004){ref-type="fig"}). Additional potential germline genes were also recovered (II-LV22, II-LV23) ([S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}) as were sequences putatively derived from clan V but with limited homology to the annotated genes.
The lengths of the CDR-L3 ranged from 5 to 11 for the kappa locus and 8--13 for the lambda ([Fig 4B](#pone.0233794.g004){ref-type="fig"}), both comparable to that of humans and other mammals. Overall, this initial pilot allowed about 12.5% (60/480) recovery efficiency of functional heavy and light chain pairs from single sorted ferret B cells, with 31/60 pairs utilizing Kappa and 29/60 pairs utilizing Lambda chains. Further improvements in the flow cytometric panel for ferret B-cells and annotation of immunoglobulin genes will enhance the recovery of antibody sequences in the future.
Sequence validation of ferret constant gene segments {#sec015}
----------------------------------------------------
Using a next-generation sequencing approach analogous to previous reports \[[@pone.0233794.ref027]\], we recovered cDNA sequences of ferret constant regions and investigated potential immunoglobulin subclasses. RNA was extracted from ferret splenocytes and subject to RNA-Seq. Putative mRNA transcripts were assembled *de novo* and 5 heavy chain isotypes (IgM, IgG, IgE, IgD and IgA) and two light chain (IgK and IgL) constant genes identified (sequences in [S1 Table](#pone.0233794.s001){ref-type="supplementary-material"}). Overall homology to both human and canine sequences was high, with the exception of IgD where the CH1 and hinge domain shows high sequence divergence, consistent with past reports \[[@pone.0233794.ref028]\]. Minimal variation was observed to artificially spliced genomic sequences, with the exception of some single amino-acid substitutions which could indicate allelic variation within outbred ferrets. Notably, we were unable to identify any IgG subclass variants using this approach, which with four distinct subclasses identified in other carnivores \[[@pone.0233794.ref029]\], might reflect low transcript abundance within our samples precluding sufficient subclass cDNA recovery.
Recombinant expression of a chimeric human/ferret monoclonal antibody {#sec016}
---------------------------------------------------------------------
We next developed the capacity to recombinantly express and purify ferret IgG. Constant genes for ferret IgG, kappa and lambda chains were synthesised and cloned into mammalian expression vectors. Recombined human VDJ (heavy) and VJ (lambda) genes from influenza-specific human antibody CR9114 \[[@pone.0233794.ref019]\] were joined to ferret constant regions to create chimeric IgG. Transfection of heavy and lambda chain plasmids into a mammalian expression system enabled the purification of chimeric ferret/human IgG using standard protein-A purification ([Fig 5A](#pone.0233794.g005){ref-type="fig"}), with the resultant antibody retaining HA-specificity ([Fig 5B](#pone.0233794.g005){ref-type="fig"}).
{#pone.0233794.g005}
Recovery of hemagglutinin-specific monoclonal antibodies from an influenza-infected ferret {#sec017}
------------------------------------------------------------------------------------------
We next sought to recover monoclonal antibodies from influenza immune ferrets. From a single ferret infected with H1N1 A/California/04/2009, we obtained cells from parapharyngeal lymph nodes 28 days post-infection, stained with a panel of ferret B cell surface markers and sorted single cells that bound to recombinant HA probes \[[@pone.0233794.ref015]\] labelled in two alternate fluorophores (PE and APC) ([Fig 6A](#pone.0233794.g006){ref-type="fig"}). Due to an inability to identify ferret IgG subclasses, we designed additional primers ([S2 Table](#pone.0233794.s002){ref-type="supplementary-material"}) targeting IGHJ-gene segments for multiplex PCR amplification of ferret immunoglobulin genes. From 960 sorted HA specific B-cells, we recovered 263 productive, recombined heavy chain immunoglobulin sequences including significant clonal expansions ([Fig 6B](#pone.0233794.g006){ref-type="fig"}). Representative examples from various antibody lineages were cloned into ferret expression vectors, expressed in mammalian cell culture and screened by ELISA. Two antibodies (3B03 and 4A06), derived from a common lineage, were found to both bind to full length A/California/04/2009 HA by ELISA ([Fig 6C](#pone.0233794.g006){ref-type="fig"}). Both antibodies (sequences in [Fig 6D](#pone.0233794.g006){ref-type="fig"}) mediated potent hemagglutinin inhibition activity against A/California/04/2009 (data not shown) and similarly when tested using an in vitro neutralization assay, both 3B03 and 4A06 were able to prevent A/California/04/2009 infection of MDCK cells down to an effective concentration of 0.4mg/ml and 0.08mg/ml respectively. In order to better understand the epitope on HA recognised by 3B03 and 4A06, we generated escape viruses by serial passaging of A/California/04/2009 in the presence of ferret mAbs followed by sequencing of the full-length HA gene using standard techniques \[[@pone.0233794.ref020]\]. Neutralisation resistant virus displayed a single K163E mutation ([Fig 7](#pone.0233794.g007){ref-type="fig"}) localised within the canonical Sa epitope \[[@pone.0233794.ref030]\] and previously associated in human populations with escape from pandemic H1N1 serum neutralising activity \[[@pone.0233794.ref031], [@pone.0233794.ref032]\].
{#pone.0233794.g006}
{#pone.0233794.g007}
Discussion {#sec018}
==========
We sought to improve the utility of ferrets as an immunological model by deriving techniques for the analysis of B cell immunity. Annotation of genomic immunoglobulin gene segments enabled the design of a multiplex PCR approach to recover ferret BCR sequences. A broad cross-section of predicted germline segments from both heavy and light chain loci were amplified, including several potentially novel germline genes currently absent from available genomic contigs. We noted highly biased V gene utilisation for both heavy and light chain naïve repertoires, with a majority of recovered sequences derived from the most numerous V gene clans (which we termed HV-III, KV-II and LV-II respectively). This observation mirrors the gene distribution in other carnivores, where majority of recovered heavy chain sequences are similarly biased \[[@pone.0233794.ref022], [@pone.0233794.ref023], [@pone.0233794.ref033], [@pone.0233794.ref034]\]. Light chain bias has been reported for mice (kappa) \[[@pone.0233794.ref035]\] and both dogs and cats (lambda) \[[@pone.0233794.ref036]\]. However, humans display more balanced usage of both the kappa and lambda chains \[[@pone.0233794.ref034], [@pone.0233794.ref037]\], as did ferrets in the current study. The distribution of ferret CDR-H3 (mode 13--14 AA) and lambda and kappa CDR-L3 lengths (mode 11 AA and 8--9 AA respectively) were broadly similar to reports from cats \[[@pone.0233794.ref023]\], dogs \[[@pone.0233794.ref033]\] and humans \[[@pone.0233794.ref038]\].
Using the new information on ferret Ig sequences, we were able to generate a chimeric ferret/human mAb that retained specificity for influenza HA, and also two novel fully ferret mAbs that exhibited potent binding and neutralisation against A/California/04/2009. These reagents and similar mAbs could be used in repeated immunotherapy studies in ferrets with less concern about the generation of ferret-anti-human responses to human mAbs. We note that while we were successful in the recombinant expression of chimeric human/ferret or fully ferret lambda chain utilising antibodies, further work is needed to generate analogous kappa chain utilising mAbs.
The utility of BCR repertoire analysis to study antigen specific B-cell responses in ferrets will increase as key knowledge gaps are bridged. Firstly, novel ferret-specific flow cytometry reagents are required to enable the resolution of different ferret B-cell populations, in particular, validated pan-B cell lineage surface markers (such as CD19 or CD20) and markers such CD27 and IgD to accurately distinguish memory versus naïve B cell populations. Further development of reagents to identify ferret memory B cells will enhance the resolution of antigen-specific B cell staining and improve recovery of HA-specific ferret mAbs.
Secondly, increased genomic information, particularly, the confirmation of ferret immunoglobulin germline genes and allelic variation among outbred ferrets will allow the development of comprehensive gene databases such as those maintained by International ImMunogeneTics (IMGT) \[[@pone.0233794.ref014]\]. While we employed a multiplex PCR approach to amplify immunoglobulin genes, the use of template switching based methods such as 5'RACE has been applied to other species such as dogs \[[@pone.0233794.ref039]\] and may reduce the potential for primer bias driving preferentially recovery of specific immunoglobulin gene segzments. Alternatively, next generation high throughput approaches and tools for the analysis of large RNA-seq data sets such as VDJPuzzle \[[@pone.0233794.ref040]\], BraCer \[[@pone.0233794.ref041]\], BALDR \[[@pone.0233794.ref042]\] and BASIC \[[@pone.0233794.ref043]\] have been deployed for analysis of antibody repertoires from cats \[[@pone.0233794.ref044]\] and rhesus macaques \[[@pone.0233794.ref045]\]. While the entire ferret kappa loci is assembled in the current copy of the ferret genome, the lambda and heavy chain loci were distributed across a number of contigs and the gene arrangement of these two loci remains unclear. Recent advances in long-read NGS approaches such as Oxford Nanopore \[[@pone.0233794.ref046]\] technologies could enable high resolution mapping of these loci in ferrets, as recently demonstrated by the assembly of reference gene loci in rhesus macaques \[[@pone.0233794.ref045]\] which facilitates the in depth analysis of antigen-specific immunoglobulin gene repertoires and the characterisation of antigenic epitopes \[[@pone.0233794.ref047]\]. Additional germline inference methods such as IgDiscover \[[@pone.0233794.ref048]\] and TIgER \[[@pone.0233794.ref049]\] could also be deployed in ferrets to more accurately assess and categorise allelic variation within these outbred animals.
Further work clarifying the range of ferret IgG subtypes and the engagement with cellular Fc-receptors (FcR) is required. While we failed to identify IgG subclasses in the current study, the presence of three or four different isotypes in closely related species such as minks \[[@pone.0233794.ref050]\], cats \[[@pone.0233794.ref051]\] and dogs \[[@pone.0233794.ref029]\] suggests these may still exist in ferrets. Recent studies have proposed a critical role for antibody effector functions for protection against viral pathogens such as influenza \[[@pone.0233794.ref052], [@pone.0233794.ref053]\]. As such, further characterisation of ferret IgG, FcR and capacity to mediate ADCC and other Fc-dependent antibody responses is needed.
The HA reactivity of immune ferret sera is a critically important issue since it guides the selection of human influenza strains for inclusion in seasonal influenza vaccines each year \[[@pone.0233794.ref054]\]. We identified two ferret mAbs, derived from a single clonally expanded family, which displayed anti-HA reactivity and mediated virus neutralisation in vitro. Interestingly, a single mutation (K163E) was able to mediate complete escape from neutralising activity, and similar mutations have been described in both circulating viruses in human populations \[[@pone.0233794.ref055]\] and from viral escape mutants selected under pressure from human serum \[[@pone.0233794.ref031], [@pone.0233794.ref032]\] or mAbs \[[@pone.0233794.ref056]\]. Futher cataloguing differential recognition of HA at the mAb level would complement serological studies, since serum samples from humans, guinea pigs, mice and ferrets have been shown to exhibit differential immunodominance hierarchies within polyclonal responses targeting the canonical epitopes of HA \[[@pone.0233794.ref057]\] and drive different patterns of viral escape \[[@pone.0233794.ref058]\]. The tools reported in the current study may also be informative for other emerging human respiratory viruses, most notably the current SARS-CoV2 pandemic. Ferrets can be productively infected with SARS-CoV2 \[[@pone.0233794.ref059], [@pone.0233794.ref060]\] and will serve as a critical model for testing therapeutic options and vaccines against the virus. The capacity to characterise immune repertoires and recover ferret mAbs from SARS-CoV2 infected or immunised ferrets might further support global efforts to develop effective countermeasures against COVID-19.
In summary, we report a methodology to sequence antigen specific B-cells in ferrets, allowing expression of chimeric ferret/human IgG monoclonal antibodies. Further in-depth studies of ferret B-cell repertoires will significantly advance the utility of ferrets as immunological models for critical human diseases such as influenza and SARS-CoV2.
Supporting information {#sec019}
======================
######
(XLSX)
######
Click here for additional data file.
######
(XLSX)
######
Click here for additional data file.
###### Uncropped gel image.
(PDF)
######
Click here for additional data file.
The Melbourne WHO Collaborating Centre for Reference and Research on Influenza is supported by the Australian Government Department of Health. JW is supported by a Melbourne International Research Scholarship and Melbourne International Fee Remission Scholarship.
10.1371/journal.pone.0233794.r001
Decision Letter 0
Tompkins
Stephen Mark
Academic Editor
© 2020 Stephen Mark Tompkins
2020
Stephen Mark Tompkins
This is an open access article distributed under the terms of the
Creative Commons Attribution License
, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
6 Apr 2020
PONE-D-20-05643
Sequencing B cell receptors from ferrets (Mustela putorius furo)
PLOS ONE
Dear Dr Wheatley,
Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE's publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.
Please consider the points raised by reviewers one and two. While minor, they are important to address. Please provide details on the frequencies of immunoglobulin chains and discuss the coverage as noted by reviewer one. Similarly Reviewer two asks for additional information on the sequences as a figure or table. Also, the constant gene segments are missing from Sup Table 1. And the recommended discussion of other approaches is important to include. A brief discussion on applications to the current pandemic would be interesting to consider. Perhaps utilize this opportunity to better justify figure 5 as reviewer one notes it is somewhat tangential. Finally, while reviewer one recommends publication, please keep in mind that the reviewers do not make decisions on acceptance of manuscripts.
We would appreciate receiving your revised manuscript by May 21 2020 11:59PM. When you are ready to submit your revision, log on to <https://www.editorialmanager.com/pone/> and select the \'Submissions Needing Revision\' folder to locate your manuscript file.
If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.
To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: <http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols>
Please include the following items when submitting your revised manuscript:
A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled \'Response to Reviewers\'.A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled \'Revised Manuscript with Track Changes\'.An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled \'Manuscript\'.
Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.
We look forward to receiving your revised manuscript.
Kind regards,
Stephen Mark Tompkins
Academic Editor
PLOS ONE
Journal Requirements:
When submitting your revision, we need you to address these additional requirements:
1\. Please ensure that your manuscript meets PLOS ONE\'s style requirements, including those for file naming. The PLOS ONE style templates can be found at <http://www.plosone.org/attachments/PLOSOne_formatting_sample_main_body.pdf> and <http://www.plosone.org/attachments/PLOSOne_formatting_sample_title_authors_affiliations.pdf>
2\. To comply with PLOS ONE submissions requirements, please provide methods of sacrifice in the Methods section of your manuscript.
3\. PLOS ONE now requires that authors provide the original uncropped and unadjusted images underlying all blot or gel results reported in a submission's figures or Supporting Information files. This policy and the journal's other requirements for blot/gel reporting and figure preparation are described in detail at <https://journals.plos.org/plosone/s/figures#loc-blot-and-gel-reporting-requirements> and <https://journals.plos.org/plosone/s/figures#loc-preparing-figures-from-image-files>. When you submit your revised manuscript, please ensure that your figures adhere fully to these guidelines and provide the original underlying images for all blot or gel data reported in your submission. See the following link for instructions on providing the original image data: <https://journals.plos.org/plosone/s/figures#loc-original-images-for-blots-and-gels>.
In your cover letter, please note whether your blot/gel image data are in Supporting Information or posted at a public data repository, provide the repository URL if relevant, and provide specific details as to which raw blot/gel images, if any, are not available. Email us at <[email protected]> if you have any questions.
4\. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to 'Update my Information' (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: <https://www.youtube.com/watch?v=_xcclfuvtxQ>
5\. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: [http://journals.plos.org/plosone/s/supporting-information](about:blank).
\[Note: HTML markup is below. Please do not edit.\]
Reviewers\' comments:
Reviewer\'s Responses to Questions
**Comments to the Author**
1\. Is the manuscript technically sound, and do the data support the conclusions?
The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.
Reviewer \#1: Yes
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
2\. Has the statistical analysis been performed appropriately and rigorously?
Reviewer \#1: N/A
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
3\. Have the authors made all data underlying the findings in their manuscript fully available?
The [PLOS Data policy](http://www.plosone.org/static/policies.action#sharing) requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data---e.g. participant privacy or use of data from a third party---those must be specified.
Reviewer \#1: Yes
Reviewer \#2: No
\*\*\*\*\*\*\*\*\*\*
4\. Is the manuscript presented in an intelligible fashion and written in standard English?
PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.
Reviewer \#1: Yes
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
5\. Review Comments to the Author
Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)
Reviewer \#1: The manuscript "Sequencing B cell receptors from ferrets (Mustela putorius furo)" describes the development of a multiplex PCR approach to single cell-sequence ferret IgG antibodies. This study builds the framework for more in-depth immunological studies in an extremely important animal model for respiratory infectious diseases. The manuscript is scientifically soloid, well planned, written and relatively easy to follow by anyone in the field. The authors efforts to assemble the ferret immunoglobulin germline repertoire has been long needed, particularly in the Influenza research world, and will certainly pioneer similar studies in other respiratory infectious diseases.
Overall this manuscript is of extreme value for the scientific community and I strongly recommend its publication in PLOSone. Nonetheless I do have some small concerns:
1\. The authors report the recovery of IgGH, IgGV� and IgGV� as 121/480, 99/480 and 144/384. Furthermore, it is then stated that 60/480 heavy/light chain pairs were recovered, but it would be important to know if these are equally distributed between � and � light chains. Plate based single cell sorting for BCR amplification is known to be considerably inefficient mostly due to low RNA input or inefficient diversity coverage during amplification. The ratio of successful heavy and light chain pairs to the number of individual heavy or light chain sequences recovered might be a good indicator good diversity coverage.
2\. In figure 5 the authors report the expression and purification of a Ferret/human chimeric antibody. In spite of the technological ingenuity of such experiment, it would be much more interesting to actually express and purify a full ferret monoclonal antibody. If using influenza infected animals, it would be easy to prove the specificity of this antibody by ELISA. Standing alone this experiment adds very little to the manuscript and actually distracts the reader from its main message.
Reviewer \#2: The manuscript describes the development of multiplex PCR for obtaining ferret Ig genes and generating recombinant monoclonal antibodies. This is an extremely important development to gain a better understanding of immune responses in ferrets, which have been a valuable model for the study of human respiratory infections. The manuscript is clearly written with well-designed experiments.
Some minor comments are:
Using bulk RNA-Seq to obtain constant region genes with multiple sub-classes can be problematic. It is mentioned that the sequences from RNA-Seq showed minimal variation compared to the genomic sequences. It will be helpful if this information is made available in some Figure or Table.
Sequences of Constant gene segments are not provided in the Supplementary Table 1
Line 117 -- Should it be supplementary table 2 instead of 1?
The sequence data should be deposited in Genbank
This is a very good start but significant more work is required to obtain a complete repertoire of germline Ig genes. Some discussion on how previous studies using template-switching methods such as BASIC, BALDR, BraCeR and VDJPuzle2 may be implemented for ferrets will be useful. In addition, long read sequencing has helped significantly improve our understanding of Ig loci in rhesus macaques (Ramesh et al, Front. Immunol. 2017 and Cirelli et al, Cell 2019). Discussion on using long-read sequencing and germline inference methods such as IgDiscover, partis and Tigger will also be helpful with respect to future directions in this field. Lastly, this work is extremely relevant with respect to the current pandemic. Discussing how ferrets maybe useful for studying viral infections such as COVID-19 will highlight the importance of this work.
\*\*\*\*\*\*\*\*\*\*
6\. PLOS authors have the option to publish the peer review history of their article ([what does this mean?](https://journals.plos.org/plosone/s/editorial-and-peer-review-process#loc-peer-review-history)). If published, this will include your full peer review and any attached files.
If you choose "no", your identity will remain anonymous but your review may still be made public.
**Do you want your identity to be public for this peer review?** For information about this choice, including consent withdrawal, please see our [Privacy Policy](https://www.plos.org/privacy-policy).
Reviewer \#1: Yes: Rodrigo Abreu
Reviewer \#2: Yes: Steven Bosinger & Amity Upadhyay
\[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link \"View Attachments\". If this link does not appear, there are no attachment files to be viewed.\]
While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, <https://pacev2.apexcovantage.com/>. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at <[email protected]>. Please note that Supporting Information files do not need this step.
10.1371/journal.pone.0233794.r002
Author response to Decision Letter 0
28 Apr 2020
Response to Reviewers
We thank the reviewers for their positive comments and suggestions which have significantly strengthened the manuscript. We address each reviewers' comments below in point-by-point form.
Reviewer \#1: The manuscript "Sequencing B cell receptors from ferrets (Mustela putorius furo)" describes the development of a multiplex PCR approach to single cell-sequence ferret IgG antibodies. This study builds the framework for more in-depth immunological studies in an extremely important animal model for respiratory infectious diseases. The manuscript is scientifically soloid, well planned, written and relatively easy to follow by anyone in the field. The authors efforts to assemble the ferret immunoglobulin germline repertoire has been long needed, particularly in the Influenza research world, and will certainly pioneer similar studies in other respiratory infectious diseases.
Overall this manuscript is of extreme value for the scientific community and I strongly recommend its publication in PLOSone. Nonetheless I do have some small concerns:
1\. The authors report the recovery of IgGH, IgGV� and IgGV� as 121/480, 99/480 and 144/384. Furthermore, it is then stated that 60/480 heavy/light chain pairs were recovered, but it would be important to know if these are equally distributed between � and � light chains. Plate based single cell sorting for BCR amplification is known to be considerably inefficient mostly due to low RNA input or inefficient diversity coverage during amplification. The ratio of successful heavy and light chain pairs to the number of individual heavy or light chain sequences recovered might be a good indicator good diversity coverage.
We have updated the manuscript (Lines 296-297) to reflect the relative proportions of kappa versus lambda light chain pairs recovered. This now reads:
"Overall, this initial pilot allowed about 12.5% (60/480) recovery efficiency of functional heavy and light chain pairs from single sorted ferret B cells, with 31/60 pairs utilizing Kappa and 29/60 pairs utilizing Lambda chains."
2\. In figure 5 the authors report the expression and purification of a Ferret/human chimeric antibody. In spite of the technological ingenuity of such experiment, it would be much more interesting to actually express and purify a full ferret monoclonal antibody. If using influenza infected animals, it would be easy to prove the specificity of this antibody by ELISA. Standing alone this experiment adds very little to the manuscript and actually distracts the reader from its main message.
We agree and have undertaken the recovery of a fully-ferret monoclonal antibody and included both ELISA and viral escape data in the revised manuscript. We have updated the Methods with the following:
"For the recovery of antigen-specific ferret B cells, a single ferret was infected with 1000 TCID50 of H1N1 A/California/04/2009 and a single cell suspension of parapharyngeal lymph node cells (pLN) was prepared at 28 days post-infection and cryopreserved in heat-inactivated FCS containing 10% DMSO. Cells were subsequently thawed and stained with Live/Dead Fixable Aqua (Thermo Fisher), surface stains anti-CD11b-BV510 (Biolegend: clone M1/70), anti-ferret IgA/IgM/IgG-FITC (Rockland Immunochemicals cat.618-102-130), anti-CD8 eFluor450 (eBioscience Clone OKT8), anti-ferret CD4 (Layton et al., 2017) conjugated to APC-Cy-7. Biotinylated recombinant full length A/California/04/2009 hemagglutinin (HA) probes (Whittle et al., 2014) conjugated to streptavidin-PE or streptavidin-APC (Invitrogen) were used to sort single HA-specific B cells into 96-well plates and stored at -20oC."
And:
"Viral Escape Asssay
The generation of viral escape mutants was based upon previously described protocols (Leon et al. 2017). Briefly, 24-well plates were seeded with 2.5 x 105 Madin Darby Canine Kidney (MDCK) cells per well to form confluent monolayers. The next day, serial dilutions of recombinant ferret mAbs were incubated with A/California/04/20090 virus for one hour at 37oC in Flu-media (Dulbecco\'s Modified Eagle\'s Medium (DMEM) with 0.8% bovine serum albumin (BSA), 1% penicillin/streptomycin and 0.1% L-1-Tosylamide-2-phenylethyl chloromethyl ketone (TPCK)-treated trypsin), before adding the virus-antibody mixture to MDCK cells. After 2 to 3 days culture media supernatants from wells with visible cytopathic effect were collected and used to infect a fresh monolayer of MDCK cells in the presence of increasing concentrations of mAb. After serial passaging, culture supernatants were harvested, viral RNA was extracted and the HA gene sequenced. Putative mutant viruses were identified based upon sequence comparison to similarly passaged media-only controls."
We have updated the Results with the following:
"Recovery of hemagglutinin-specific monoclonal antibodies from an influenza-infected ferret
We next sought to recover monoclonal antibodies from influenza immune ferrets. From a single ferret infected with H1N1 A/California/04/2009, we cells from parapharyngeal lymph nodes 28 days post-infection, stained with a panel of ferret B cell surface markers and sorted single cells that bound to recombinant HA probes (Whittle et al., 2014) labelled in two alternate fluorophores (PE and APC) (Figure 6A). Due to an inability to identify ferret IgG subclasses, we designed additional primers (Supplementary table 2) targeting IGHJ-gene segments for multiplex PCR amplification of ferret immunoglobulin genes. From 376 sorted HA specific B-cells, we recovered 263 productive, recombined heavy chain immunoglobulin sequences including significant clonal expansions (Figure 6B). Representative examples from various antibody lineages were cloned into ferret expression vectors, expressed in mammalian cell culture and screened by ELISA. Two antibodies (3B03 and 4A06), derived from a common lineage, were found to both bind to full length A/California/04/2009 HA by ELISA (Figure 6C). Both antibodies (sequences in Figure 6D) mediated potent hemagglutinin inhibition activity against A/California/04/2009 (data not shown) and similarly when tested using an in vitro neutralization assay, both 3B03 and 4A06 were able to prevent A/California/04/2009 infection of MDCK cells down to an effective concentration of 0.4mg/ml and 0.08mg/ml respectively. In order to better understand the epitope on HA recognised by 3B03 and 4A06, we generated escape viruses by serial passaging of A/California/04/2009 in the presence of ferret mAbs followed by sequencing of the full-length HA gene using standard techniques (Leon et al 2017). Neutralisation resistant virus displayed a single K163E mutation (Figure 7) localised within the canonical Sa epitope (Caton et al. 1982) and previously associated in human populations with escape from pandemic H1N1 serum neutralising activity (Chengjun et al. 2016; Davis et al. 2018)."
And we have included a new Figures 6 and 7:
And additional text has been incorporated throughout the discussion.
Reviewer \#2: The manuscript describes the development of multiplex PCR for obtaining ferret Ig genes and generating recombinant monoclonal antibodies. This is an extremely important development to gain a better understanding of immune responses in ferrets, which have been a valuable model for the study of human respiratory infections. The manuscript is clearly written with well-designed experiments.
Some minor comments are:
Using bulk RNA-Seq to obtain constant region genes with multiple sub-classes can be problematic. It is mentioned that the sequences from RNA-Seq showed minimal variation compared to the genomic sequences. It will be helpful if this information is made available in some Figure or Table.
Sequences of Constant gene segments are not provided in the Supplementary Table 1
Line 117 -- Should it be supplementary table 2 instead of 1?
Sequences for the ferret constant gene segments have been included within Supplementary Table 1 (Tab -- "constant")
The sequence data should be deposited in Genbank
Relevant sequence data have been deposited to Genbank (submission ID:2331742)
This is a very good start but significant more work is required to obtain a complete repertoire of germline Ig genes. Some discussion on how previous studies using template-switching methods such as BASIC, BALDR, BraCeR and VDJPuzle2 may be implemented for ferrets will be useful. In addition, long read sequencing has helped significantly improve our understanding of Ig loci in rhesus macaques (Ramesh et al, Front. Immunol. 2017 and Cirelli et al, Cell 2019). Discussion on using long-read sequencing and germline inference methods such as IgDiscover, partis and Tigger will also be helpful with respect to future directions in this field.
We have expanded the discussion to cover these important sequencing technologies. It now includes the following section:
"While we employed a multiplex PCR approach to amplify immunoglobulin genes, the use of template switching based methods such as 5'RACE has been applied to other species such as dogs (39) and may reduce the potential for primer bias driving preferentially recovery of specific immunoglobulin gene segzments. Alternatively, next generation high throughput approaches and tools for the analysis of large RNA-seq data sets such as VDJPuzzle(40), BraCer(41), BALDR(42) and BASIC(43) have been deployed for analysis of antibody repertoires from cats (44) and rhesus macaques (45). While the entire ferret kappa loci is assembled in the current copy of the ferret genome, the lambda and heavy chain loci were distributed across a number of contigs and the gene arrangement of these two loci remains unclear. Recent advances in long-read NGS approaches such as Oxford Nanopore (46) technologies could enable high resolution mapping of these loci in ferrets, as recently demonstrated by the assembly of reference gene loci in rhesus macaques (45) which facilitates the in depth analysis of antigen-specific immunoglobulin gene repertoires and the characterisation of antigenic epitopes (47). Additional germline inference methods such as IgDiscover (48) and TIgER (49) could also be deployed in ferrets to more accurately assess and categorise allelic variation within these outbred animals."
Lastly, this work is extremely relevant with respect to the current pandemic. Discussing how ferrets maybe useful for studying viral infections such as COVID-19 will highlight the importance of this work.
We have expanded the discussion to cover the importance of ferrets for coronavirus research. It now includes the following section:
"The tools reported in the current study may also be informative for other emerging human respiratory viruses, most notably the current SARS-CoV2 pandemic. Ferrets can be productively infected with SARS-CoV2 (59, 60) and will serve as a critical model for testing therapeutic options and vaccines against the virus. The capacity to characterise immune repertoires and recover ferret mAbs from SARS-CoV2 infected or immunised ferrets might further support global efforts to develop effective countermeasures against COVID-19."
######
Submitted filename: PLOS_rebuttal_Final.docx
######
Click here for additional data file.
10.1371/journal.pone.0233794.r003
Decision Letter 1
Tompkins
Stephen Mark
Academic Editor
© 2020 Stephen Mark Tompkins
2020
Stephen Mark Tompkins
This is an open access article distributed under the terms of the
Creative Commons Attribution License
, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
13 May 2020
Sequencing B cell receptors from ferrets (Mustela putorius furo)
PONE-D-20-05643R1
Dear Dr. Wheatley,
We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.
Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.
Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at <https://www.editorialmanager.com/pone/>, click the \"Update My Information\" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at <[email protected]>.
If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact <[email protected]>.
With kind regards,
Stephen Mark Tompkins
Academic Editor
PLOS ONE
Additional Editor Comments (optional):
Reviewers\' comments:
10.1371/journal.pone.0233794.r004
Acceptance letter
Tompkins
Stephen Mark
Academic Editor
© 2020 Stephen Mark Tompkins
2020
Stephen Mark Tompkins
This is an open access article distributed under the terms of the
Creative Commons Attribution License
, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
18 May 2020
PONE-D-20-05643R1
Sequencing B cell receptors from ferrets (*Mustela putorius furo*)
Dear Dr. Wheatley:
I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.
If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact <[email protected]>.
For any other questions or concerns, please email <[email protected]>.
Thank you for submitting your work to PLOS ONE.
With kind regards,
PLOS ONE Editorial Office Staff
on behalf of
Dr. Stephen Mark Tompkins
Academic Editor
PLOS ONE
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
| {
"pile_set_name": "PubMed Central"
} |
Coronavirus disease 2019 (COVID-19), caused by the newly identified strain of the coronavirus family severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has rapidly evolved into a worldwide pandemic and caused a public health emergency of major international concern.[@bib1], [@bib2] As a result, a profound reorganisation of hospital wards and clinical activities is happening worldwide to deal with the increasing number of COVID-19-positive patients who require hospitalisation and intensive care support.[@bib3] This comprehensive reallocation of health resources is of particular concern in patients such as those with underlying chronic diseases, including cancer.
The prioritisation of health support towards patients with COVID-19 is raising apprehension within the medical oncology community, in which physicians are increasingly being forced to select which patients should receive anticancer therapy on the basis of who is most likely to have a positive outcome.[@bib4] In this context, the threat of COVID-19 infection might also factor into decision making---a role which could possibly be lessened by knowledge of the COVID-19 status of patients suitable for anticancer therapy.[@bib4] This already dismal scenario seems to be even more severe for patients with lung cancer because of the high risk of interference of COVID-19 with their effective diagnostic and therapeutic management by treating physicians.
Clinical manifestations of COVID-19 range from asymptomatic, to mild symptoms (such as cold, fever, cough, or other non-specific signs), to severe pneumonia leading to acute respiratory distress syndrome, which occurs in 17--29% of infected individuals.[@bib2] Mortality due to COVID-19 has been reported in about 3% of COVID-19-positive patients in the Chinese population,[@bib5] while higher mortality rates are being reported in Italy,[@bib6] which is, after the USA, currently the country with the second highest number of confirmed COVID-19 cases worldwide.[@bib7]
In the early phase of COVID-19-induced pneumonia, the main CT findings include multifocal peripheral and basal ground-glass opacities, crazy paving patterns, traction bronchiectasis, and air bronchogram signs. A progressive transition to consolidation, together with pleural effusion, extensive small lung nodules, irregular interlobular or septal thickening, and adenopathies, characterise the more advanced phase of the disease.[@bib8], [@bib9] These radiological manifestations can overlap with CT findings that are often found in patients with lung cancer upon disease progression or onset of concomitant pneumonia due to overlapping opportunistic infections. Regarding clinical manifestations, the worsening of pulmonary symptoms during lung cancer progression can be similar to that typical of COVID-19, adding further complexity to the thorough assessment of the course of disease in lung cancer patients. Together, these similarities can pose a major challenge to clinicians in distinguishing lung cancer evolution from a potential COVID-19 super-infection on the basis of radiological and clinical evidence, and, importantly, these specific conditions require very different therapeutic approaches.
Adding further complexity to this scenario, pneumonitis can also be induced by immune checkpoint inhibitor therapy, an effective and widely used standard-of-care treatment for lung cancer in various treatment lines and settings.[@bib10] Immune checkpoint inhibitor-related pneumonitis has been reported in about 2% of cancer patients,[@bib11] with a seemingly higher incidence in patients with lung cancer.[@bib12] Similar to COVID-19 infection, the clinical symptoms of immune checkpoint inhibitor-induced pneumonitis are often not specific, consisting mainly of cough (or its worsening), chest pain, dyspnoea, and fever. Additionally, CT assessment of immune checkpoint inhibitor-related pneumonitis shows radiological findings similar to those typical of COVID-19-induced pneumonia ([figure](#fig1){ref-type="fig"} ), thus hindering discrimination between the two clinical entities. Similarly, tyrosine kinase inhibitors can induce radiological patterns of interstitial-like pneumonitis, which develops in 4% of patients with epidermal growth factor receptor-mutant lung cancer treated with osimertinib.[@bib13] FigureCT scans of pneumonia due to COVID-19 and immune checkpoint inhibitor therapy(A) Axial lung image (without intravenous contrast) of 49-year-old man with COVID-19, showing two sub-solid areas in the upper right lobe (arrows). (B) Axial lung image (without intravenous contrast) of an immune checkpoint inhibitor-treated 76-year-old man with metastatic melanoma, showing a sub-solid area and ground-glass opacities with a rounded morphology in the upper right lobe (arrows). COVID-19=coronavirus disease 2019.
In this scenario, standard chemotherapy does not seem to represent a suitable or potentially safer alternative to immune checkpoint inhibitor therapy---neither for treating physicians who want to avoid the overlapping immune checkpoint inhibitor-related and COVID-19-related radiological and clinical changes, or for patients who are unsuitable for immune checkpoint inhibitor therapy. First, combinations of chemotherapies and immunotherapies have shown the best efficacy and represent the standard of care in a large group of patients without oncogene-driven lung cancer and without high PD-L1 expression in tumour cells. Second, the development of chemotherapy-associated pneumonitis is known to occur in up to 16% of treated patients,[@bib14] and cytotoxic chemotherapy has immunosuppressive activity.[@bib15] Notably, administration of chemotherapy within the month preceding COVID-19 diagnosis has been shown to be associated with a higher risk of severe infection-related complications.[@bib16]
The clinical and biological aggressiveness of lung malignancies clearly does not allow for anticancer therapy to be withheld or postponed. Thus, while awaiting specific evidence-based guidelines, the comprehensive management of patients with lung cancer during the COVID-19 pandemic should involve specific and careful attention to their clinical and radiological pulmonary signs, more so than for patients with other types of tumour. From a practical viewpoint, it seems reasonable to suggest that patients with lung cancer undergo systematic testing for SARS-CoV-2 at the beginning of treatment and whenever it is deemed necessary by the treating physician in the course of therapy. This strategy might become more feasible with the increasing availability and progressive use of real-time PCR assays that can provide COVID-19 status results within an hour.[@bib17] Furthermore, the availability of laboratory IgM or IgG testing to evaluate the exposure and immunity to SARS-CoV-2 infection will be helpful when the COVID-19 pandemic begins to decline. Allocating resources for these methodological approaches to patients with lung cancer should facilitate the most appropriate clinical management by multidisciplinary lung cancer care teams.
© 2020 Dr P Marazzi/SPL2020Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company\'s public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active.
LC has served as consultant or advisor to Bristol-Myers Squibb and Merck Sharp and Dohme. SP has received education grants, provided consultation, attended advisory boards, or provided lectures for AbbVie, Amgen, AstraZeneca, Bayer, Biocartis, Bioinvent, Blueprint Medicines, Boehringer-Ingelheim, Bristol-Myers Squibb, Clovis, Daiichi Sankyo, Debiopharm, Eli Lilly, F Hoffmann-La Roche, Foundation Medicine, Illumina, Janssen, Merck Sharp and Dohme, Merck Serono, Merrimack, Novartis, Pharma Mar, Pfizer, Regeneron, Sanofi, Seattle Genetics, Takeda, and Vaccibody, from whom she has received honoraria (all fees to institution). J-CS was employee for AstraZeneca from September, 2017, to December, 2019, and has shares in Gritstone. AMDG has served as a consultant or advisor to Incyte, Pierre Fabre, GlaxoSmithKline and Bristol-Myers Squibb, Merck Sharp Dohme, and Sanofi. FB has served as a consultant, advisor, or lecturer for AstraZeneca, Bayer, Bristol-Myers Squibb, Boehringer-Ingelheim, Eli Lilly Oncology, F Hoffmann-La Roche, Novartis, Merck, Merck Sharp and Dohme, Pierre Fabre, Pfizer, and Takeda. CG has received honoraria as consultant or advisor or speaker bureau member for AstraZeneca, Bristol-Myers Squibb, F Hoffmann-La Roche, and Merck Sharp and Dohme, and has received funds (to organisation) from Merck Sharp and Dohme. MR has served as a consultant and provided lectures for Amgen, AbbVie, AstraZeneca, Bristol-Myers Squibb, Boehringer-Ingelheim, Celgene, Merck, Merck Sharp and Dohme, Novartis, Pfizer, Roche, Samsung. MM has served as a consultant or advisor to Roche, Bristol-Myers Squibb, Merck Sharp and Dohme, Incyte, AstraZeneca, Amgen, Pierre Fabre, Eli Lilly, GlaxoSmithKline, Sciclone, Sanofi, Alfasigma, and Merck Serono. AC, MA, and VV declare no competing interests. We thanks Nicholas Landini of the Diagnostic Radiology Department of Ca\' Foncello Regional Hospital (Treviso, Italy), for providing the CT scan images of a COVID-19-positive patient with pneumonia. NR declares no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
From an architectural point of view, bacterial cells are among the simplest forms of life on the planet. Their cytoplasm is typically devoid of membrane bound organelles, and their cellular morphology relies upon a semi-rigid peptidoglycan (PG) cell wall that imposes its shape on the malleable cell membrane(s). Bacterial cells are inflated by their high internal turgor pressure, which pushes the membranes against the cell wall, causing the PG to stretch and the cell to adopt its appropriate shape. Despite the apparent simplicity, studies in the past few decades have demonstrated that bacterial cellular architecture is far more complex than previously thought, in terms of both its ultrastructure and dynamic capabilities ([@bib17]; [@bib52]; [@bib53]).
Endospore formation in *Bacillus subtilis* represents a striking example of the dynamic capabilities of bacterial cells, as it entails dramatic changes in cellular architecture. First, the division site shifts to polar position, generating a sporangium comprised of two cells: a larger mother cell and a smaller forespore ([Figure 1A](#fig1){ref-type="fig"}; [@bib13]; [@bib18]; [@bib46]). The polar septum traps the forespore chromosome, which is subsequently transported to the forespore by SpoIIIE, a membrane-anchored ATPase that assembles a translocation complex at septal midpoint ([@bib3]; [@bib58]; [@bib55]; [@bib56]). Chromosome translocation increases the turgor pressure in the forespore, causing it to inflate and expand into the mother cell ([@bib25]). Simultaneously, the mother cell engulfs the forespore in a process that resembles eukaryotic phagocytosis ([Figure 1A](#fig1){ref-type="fig"}). After engulfment, the forespore is fully enclosed within the cytoplasm of the mother cell, where it matures in a process that involves the synthesis of protective layers of cortex and coat, and the partial dehydration of the forespore cytoplasm. Finally, the mother cell lyses and the mature spore is released to the environment, where it remains dormant until the conditions are appropriate for germination.
![Visualizing the 3D architecture of engulfment during sporulation in *B. subtilis*.\
(**A**) Schematic illustrating the process of polar septation, chromosome translocation, engulfment and spore maturation. Membranes (red), PG (gray), chromosomes (blue), SpoIIIE (orange) and coat proteins (shades of brown) are highlighted. Outward arrows in the stage II~ii~ forespore indicate increased turgor pressure due to chromosome translocation. (**B**) Revised engulfment model ([@bib38]): new PG (purple) is synthesized ahead of the leading edge of the engulfing membrane by forespore-associated PG biosynthetic enzymes (blue) and is subsequently degraded by DMP (yellow pacman), making room for the engulfing membrane to advance. The coordinated synthesis and degradation of PG at the leading edge of the engulfing membrane can move the junction between the septum (pink) and the lateral cell wall (gray) around the forespore. (**C**) Schematic illustrating cryo-FIB-ET methodology for bacterial samples (see Materials and methods). (**D--I**) Slices through cryo-electron tomograms representing different stages of engulfment: (**D**) flat polar septum (Stage II~i~), (**F**) curved septum (Stage II~ii~) and (**H**) engulfing septum (Stage II~iii~). Scale bar for (D,F,I): 200 nm. The corresponding forespore membrane (green) and the mother cell membrane (yellow) are annotated in (**E,G,I**) respectively. n indicates the number of tomograms acquired for each cell type. Scale bars have been omitted for (E,G,I) as cells are shown in perspective views . (**J--N**) Zoomed-in slices through cryo-electron tomograms showing (**J**) a large ellipsoidal complex adjacent to the forespore membrane (black arrow), (**K**) a putative SpoIIIE channel (brown arrow) and (**L**) another putative channel (brown arrow), (**L,M**) coat filaments (green arrows), (**M**) basement coat layer (maroon arrow) and (**N**) amorphous coat proteins (purple arrow). Scale bar for (J-N): 50 nm.](elife-45257-fig1){#fig1}
Engulfment represents a major rearrangement of the sporangium, from two cells that lie side by side to a cell within a cell. Such rearrangement likely involves a profound remodeling of the PG cell wall around the forespore, which would otherwise constrain the movement of the engulfing membrane. At the onset of engulfment, the engulfing mother cell membrane must circumvent the physical barrier posed by the septal PG in order to migrate around the forespore. This has led to the logical proposal that engulfment must entail the complete dissolution of the septal PG, a process often referred to as 'septal thinning' ([@bib9]; [@bib20]; [@bib39]). This proposal was supported by early electron microscopy studies of fixed and stained sporangia showing that engulfment-defective mutants had thicker septa than wild type sporangia ([@bib19]; [@bib20]; [@bib42]). Further studies showed that engulfment requires three mother cell proteins: SpoIID, SpoIIM and SpoIIP, which form a complex (DMP) with PG degradation activity ([@bib1]; [@bib2]; [@bib9]; [@bib35]). In principle, DMP could mediate the complete dissolution of the septal PG to remove the steric block to the movement of the mother cell membrane around the forespore.
The idea that septal PG is completely degraded has been more recently challenged by cryo-electron tomography (cryo-ET) images showing that a thin PG layer is present between the forespore and the mother cell membranes throughout engulfment ([@bib48]). It has also been shown that DMP-mediated PG degradation is required and rate-limiting for membrane migration even after the septal barrier has been bypassed, suggesting that DMP plays a role separate from the dissolution of septal PG ([@bib1]; [@bib16]). In addition, the movement of the mother cell membrane also requires PG synthesis at the leading edge of the engulfing membrane ([@bib34]; [@bib38]). Based on these observations, we proposed a revised model for engulfment membrane migration in which coordinated PG synthesis and degradation at the leading edge of the engulfing mother cell membrane moves the junction between the septum and the lateral cell wall around the forespore, making room for the engulfing membrane to advance ([Figure 1B](#fig1){ref-type="fig"}; [@bib38]). This model eliminates the need for complete dissolution of the septal PG and predicts that PG synthesis happens ahead of the leading edge of the engulfing membrane. Then, the mother cell DMP degrades this new PG to mediate membrane migration. However, due to the limited resolution of optical microscopy, conclusive evidence that PG synthesis occurs ahead of PG degradation is lacking.
Cryo-ET allows the visualization of three-dimensional (3D) architecture of bacterial membranes and cell wall in a hydrated near-native state that cannot be achieved by methods reliant on chemical fixation and staining ([@bib5]; [@bib31]; [@bib37]). However, a limitation of cryo-ET is that the samples must be less than \~500 nm thick to obtain high-resolution tomograms, constraining its application to only a handful of bacteria that are either naturally slender, or, as in the case of *B. subtilis*, for which slender mutant strains are available ([@bib48]). But the latter typically contain mutations in genes involved in PG metabolism and may not be ideal to study cell wall remodeling. Recent application of cryo-focused ion beam (cryo-FIB) milling has produced artifact-free thin sample sections of \~100--300 nm, which allows the acquisition of high-resolution tomograms of sections of wild type cells ([Figure 1C](#fig1){ref-type="fig"}; [@bib27]; [@bib41]; [@bib51]). Cryo-FIB milling coupled with cryo-ET (or cryo-FIB-ET) is therefore becoming the method of choice for studies of cellular architecture of both eukaryotic and prokaryotic cells ([@bib8]; [@bib12]; [@bib25]; [@bib26]).
Here, we have used cryo-FIB-ET to study sporulation in *B. subtilis*, revealing the different stages of engulfment with a resolution that has not been achieved previously. We have analyzed wild type sporangia, engulfment mutants, and sporangia treated with PG synthesis inhibitors to obtain new mechanistic insights into the PG transformations that occur during engulfment. First, we provide evidence that septal PG is not degraded completely at the onset of engulfment. Second, we show that during membrane migration, the newly synthesized PG deforms the forespore membrane ahead of the leading edge of the engulfing mother cell membrane, indicating that PG synthesis precedes PG degradation. Third, we observe that the mother cell membrane migrates around the forespore by forming tiny finger-like projections, the formation of which depends on DMP complexes tethering to and degrading the PG ahead of the engulfing membrane. The methodology, images and analyses presented here will provide valuable resources for future studies of spore assembly and other fundamental cell biological processes.
Results {#s2}
=======
Visualizing sporulation in wild type *B. subtilis* at molecular resolution {#s2-1}
--------------------------------------------------------------------------
Recently, we used cryo-FIB-ET to illustrate the role of DNA translocation in inflating the forespore ([@bib25]). These data confirmed the presence of a thin layer of PG between the forespore and the mother cell membranes in the wild type strain, as previously visualized by cryo-ET of a slender *ponA* mutant of *B. subtilis* ([@bib48]). We expanded our cryo-FIB-ET studies to investigate the architecture of *B. subtilis* cells during different stages of sporulation ([Figure 1C](#fig1){ref-type="fig"}; see Materials and methods). We acquired high-quality data of wild type cells during engulfment (Stage II, [Figure 1D--I](#fig1){ref-type="fig"}; [Figure 1---figure supplement 1](#fig1s1){ref-type="fig"}) and during later stages of sporulation, when cortex and coat were being assembled (Stages III-VI, [Figure 1---figure supplement 2](#fig1s2){ref-type="fig"}). Data of very late stages of sporulation failed to provide high-resolution information inside the forespore, likely due to the dehydration of this cell, which increases sensitivity of cryo-ET samples to the electron beam ([@bib54]).
Our data showed that the external cell wall of sporulating cells was \~20--30 nm thick, consistent with other EM studies ([@bib31]; [@bib48]). The polar septum formed close (within 500 nm) to one cell pole, and was initially flat ([Figure 1D,E](#fig1){ref-type="fig"}, [Figure 1---figure supplement 1A--D](#fig1s1){ref-type="fig"}, [Video 1](#video1){ref-type="video"}). Eventually, the septum bent smoothly into the mother cell ([Figure 1F,G](#fig1){ref-type="fig"}, [Figure 1---figure supplement 1E--H](#fig1s1){ref-type="fig"}, [Video 2](#video2){ref-type="video"}) and the mother cell membrane moved forward to engulf the forespore, at which stage the forespore was roughly rounded ([Figure 1H,I](#fig1){ref-type="fig"}, [Figure 1---figure supplement 1I--L](#fig1s1){ref-type="fig"}, [Video 3](#video3){ref-type="video"}).
###### Movie showing slices through a cryo-electron tomogram of *B. subtilis* wild type sporangium (flat septum, Stage II~i~) shown in [Figure 1D](#fig1){ref-type="fig"}.
10.7554/eLife.45257.007
###### Movie showing slices through a cryo-electron tomogram of *B. subtilis* wild type sporangium (curved septum, Stage II~ii~) shown in [Figure 1F](#fig1){ref-type="fig"}.
10.7554/eLife.45257.008
###### Movie showing slices through a cryo-electron tomogram of *B. subtilis* wild type sporangium (engulfing septum, Stage II~iii~) shown in [Figure 1H](#fig1){ref-type="fig"}.
10.7554/eLife.45257.009
Visual inspection of tomograms also revealed several structures that have not been characterized previously. Immediately after polar septation, we observed ellipsoidal complexes that were present only in the forespore, adjacent to the membrane and often close to the intersection between the septum and the lateral cell wall ([Figure 1J](#fig1){ref-type="fig"}, [Figure 1---figure supplement 3A--C](#fig1s3){ref-type="fig"}). These structures were observed in 2 out of 7 tomograms of wild type sporangia with flat septa, with \~4 ellipsoidal complexes observed in \~200 nm slices. The 3D reconstruction of these complexes revealed that the ellipsoidal structures had a mean radius of \~45 nm ([Figure 1---figure supplement 3A--C](#fig1s3){ref-type="fig"}, see Materials and methods). The molecular identity of these structures remains undetermined. We also identified a region approximately in the center of a flat septum where the two membranes are closer together than elsewhere on the septum (14 nm vs. 23 nm, [Figure 1K](#fig1){ref-type="fig"}, [Figure 1---figure supplement 3D,E](#fig1s3){ref-type="fig"}). This constriction may correspond to paired hexameric SpoIIIE channels ([@bib14]; [@bib24]; [@bib58]). We also observed structures that appear to be channels crossing the septum ([Figure 1L](#fig1){ref-type="fig"}) that may correspond to SpoIIIA-SpoIIQ complexes ([@bib6]; [@bib23]; [@bib36]; [@bib60]). We observed this thin region in the polar septum and channel-like features ([Figure 1K,L](#fig1){ref-type="fig"}) in 2 out of 15 tomograms of wild type sporangia with flat and curved septa. The low frequency of observation could be attributed to the fact that we are imaging only \~150--250 nm slices of cells that are over 1 μm wide. So, many tomograms would not include low abundance structures such as SpoIIIE, which is comprised of two adjacent channels. Next, we observed a basement coat layer adjacent to the outer forespore membrane that is likely comprised of SpoVM and/or SpoIVA ([Figure 1M](#fig1){ref-type="fig"}; [@bib40]). This layer is visible as an array of dots spaced \~4--6 nm apart, similar to that observed previously in *Acetonema longum* sporulating cells ([@bib47]). Moving outward, we observed a dense amorphous layer and a filamentous layer ([Figure 1L--N](#fig1){ref-type="fig"}, [Figure 1---figure supplement 3F,G](#fig1s3){ref-type="fig"}) that may contain CotE, SpoIVA and other coat proteins that are recruited to the outer forespore membrane during engulfment ([@bib33]). Further studies are required to determine the molecular identity of these structures unambiguously.
Septal PG is not completely degraded at the onset of engulfment {#s2-2}
---------------------------------------------------------------
The cryo-FIB-ET images provided high-resolution details of the septum and the engulfing membrane. Hence, we focused on those details to obtain mechanistic insights about engulfment. The complete degradation of the septal PG during septal thinning has been traditionally considered a prerequisite for engulfment ([@bib1]; [@bib9]; [@bib11]; [@bib39]). However, we observed a thin PG layer between the mother cell and the forespore membranes throughout engulfment in both wild type ([Figure 2A](#fig2){ref-type="fig"}; [@bib25]) and a slender *ponA* mutant ([@bib48]), suggesting that the septal PG is either not completely degraded or is quickly resynthesized after degradation.
{#fig2}
To probe this, we imaged engulfing sporangia after treatment with antibiotics that block PG synthesis. When PG synthesis is inhibited, engulfment membrane migration does not continue, although the septum still stretches and curves into the mother cell ([@bib38]). We reasoned that, if the septal PG was completely degraded, we would observe sporangia lacking PG between the mother cell and the forespore membranes after antibiotic treatment. However, if the septal PG was not completely degraded, the sporangia would show a layer of PG around the forespore, independent of antibiotic treatment.
To inhibit PG synthesis, we treated the cells with bacitracin, cephalexin, penicillin V or a combination of penicillin V and cephalexin ([Figure 2](#fig2){ref-type="fig"}). We previously assessed the extent of PG inhibition after antibiotic treatment by determining the frequency of division events, which rely on the synthesis of new PG ([@bib38]). As expected, our results showed that untreated cells continued to grow and divide under the experimental conditions ([Figure 2---figure supplement 1A](#fig2s1){ref-type="fig"}) but when treated with cephalexin and penicillin V, cell division was completely blocked. These results also indicated that PG synthesis was inhibited within a few minutes of antibiotic treatment, because cells that were already undergoing septation were unable to complete septum formation ([Figure 2---figure supplement 1B--D](#fig2s1){ref-type="fig"}, top panels). To complement this analysis, we tested if these drugs also inhibited elongation, which also depends on PG synthesis ([@bib45]). To do so, we measured the elongation of vegetative cells present in sporulating cultures over a period of one hour after antibiotic treatment ([Figure 2---figure supplement 1E](#fig2s1){ref-type="fig"}, see Materials and methods). The length of wild type untreated vegetative cells increased by \~35% in an hour, but when treated with cephalexin or penicillin V alone, they elongated only \~10%. Furthermore, when treated with a combination of cephalexin and penicillin V, the cells elongated negligibly, indicating complete blockage of both septation and elongation ([Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}, [Videos 4](#video4){ref-type="video"}--[7](#video7){ref-type="video"}).
###### Time-lapse microscopy of untreated sporulating *B. subtilis* cells stained with the membrane dye FM4-64.
Pictures were taken every 5 minutes for 2 hours (related to [Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}).
10.7554/eLife.45257.013
###### Time-lapse microscopy of penicillin V-treated (500 μg/ml) sporulating *B. subtilis* cells stained with the membrane dye FM4-64.
Pictures were taken every 5 minutes for 2 hours (related to [Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}).
10.7554/eLife.45257.014
###### Time-lapse microscopy of cephalexin-treated (50 μg/ml) sporulating *B. subtilis* cells stained with the membrane dye FM4-64.
Pictures were taken every 5 minutes for 2 hours (related to [Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}).
10.7554/eLife.45257.015
###### Time-lapse microscopy of cephalexin- (50 μg/ml) and penicillin V- (500 μg/ml) treated sporulating *B. subtilis* cells stained with the membrane dye FM4-64.
Pictures were taken every 5 minutes for 2 hours (related to [Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}).
10.7554/eLife.45257.016
For cryo-FIB-ET, we added the antibiotics two hours after inducing sporulation, when \~ 40--50% of the cells have undergone polar septation ([@bib38]), and plunge froze the samples either one (for bacitracin) or two hours (for cephalexin, penicillin V, and combination of cephalexin and penicillin V) later. Indeed, we observed a thin PG layer in both untreated and antibiotic-treated sporangia ([Figure 2](#fig2){ref-type="fig"}, [Figure 2---figure supplement 2](#fig2s2){ref-type="fig"}), suggesting that septal PG is not completely degraded at the onset of engulfment.
Septal thickness decreases slightly and uniformly across the entire septum during engulfment {#s2-3}
--------------------------------------------------------------------------------------------
The above observation prompted us to re-evaluate the process of septal thinning. The current model for septal thinning proposes that DMP initially localizes to the septal midpoint, where it starts degrading the septal PG as it moves towards the edge of the septal disk ([@bib1]; [@bib9]; [@bib20]; [@bib34]). This enzymatic septal thinning model predicts that, during the transition from flat to curved septa, the septum should be thinner in the middle than at the edges ([Figure 3A](#fig3){ref-type="fig"}). To test this, we measured the distance between the forespore and the mother cell membranes across the length of the septum (see Materials and methods) for cells with flat (Stage II~i~), curved (Stage II~ii~) and engulfing (Stage II~iii~) septa ([Figure 3B--D](#fig3){ref-type="fig"}, [Figure 3---figure supplement 1](#fig3s1){ref-type="fig"}). Sporangia with flat septa had an average septal thickness of \~23 nm±3.3 nm ([Figure 3B,E,F](#fig3){ref-type="fig"}) with 3 out of 5 septa being thicker at the middle (\~28 nm) than at the edges (\~22 nm) ([Figure 3B,E](#fig3){ref-type="fig"}, [Figure 3---figure supplement 1A](#fig3s1){ref-type="fig"}), contrary to what is proposed by the enzymatic septal thinning model. The septal thickness decreased by \~25% to\~18 nm during later stages of engulfment ([Figure 3C--F](#fig3){ref-type="fig"}, [Figure 3---figure supplement 1B,C](#fig3s1){ref-type="fig"}) and a thin layer of PG was observed in all septa ([Figure 2A](#fig2){ref-type="fig"}, [Figure 2---figure supplement 2A--C](#fig2s2){ref-type="fig"}). Importantly, septal thickness was uniform across the entire septum during later stages ([Figure 3C,D,E](#fig3){ref-type="fig"}, [Figure 3---figure supplement 1B,C](#fig3s1){ref-type="fig"}) and no septum was thinner in the middle than in the edges. These results show that the transition from thick to thin septum is homogenous, contrary to the prediction of the enzymatic septal thinning model, and consistent with the model that DNA translocation dependent forespore growth stretches septal PG.
{ref-type="fig"}.](elife-45257-fig3){#fig3}
SpoIIDMP is essential to maintain a thin, flexible septum {#s2-4}
---------------------------------------------------------
Next, we tested whether DMP was required to mediate the slight thinning observed during the transition from flat to curved septum. To address this question, we measured septal thickness in DMP mutants. In single mutants lacking D, M or P, engulfment is blocked but the septum bulges towards the mother cell, which complicates the analysis ([Figure 3---figure supplement 2](#fig3s2){ref-type="fig"}). However, bulge formation is largely abolished in a triple mutant lacking functional versions of D, M and P simultaneously ([@bib11]). Therefore, we imaged the DMP triple mutant by cryo-FIB-ET and measured its septal thickness ([Figure 3G--I](#fig3){ref-type="fig"}, [Figure 3---figure supplement 3](#fig3s3){ref-type="fig"}).
Most sporulation septa of the DMP triple mutant were either flat or slightly curved into the mother cell ([Figure 3G--I](#fig3){ref-type="fig"}, [Figure 3---figure supplement 3A--D](#fig3s3){ref-type="fig"}). In some cells, we observed membrane accumulation in the mother cell ([Figure 3H,I](#fig3){ref-type="fig"}, [Figure 3---figure supplement 3A,B](#fig3s3){ref-type="fig"}) and small bulges approximately in the middle of the septum ([Figure 3---figure supplement 3C,D](#fig3s3){ref-type="fig"}, [Video 8](#video8){ref-type="video"}). Septal thickness ranged from \~25 nm to \~45 nm, with an average thickness of 28 nm ±2.09 nm ([Figure 3E--G](#fig3){ref-type="fig"}, [Figure 3---figure supplement 3E](#fig3s3){ref-type="fig"}), which is \~5 nm greater than that of wild type sporangia with flat septa. Importantly, there were no significant differences in septal thickness between flat and curved septa in DMP mutant sporangia ([Figure 3G](#fig3){ref-type="fig"}, [Figure 3---figure supplement 3E](#fig3s3){ref-type="fig"}), indicating that DMP is in fact necessary for the slight thinning of the septum observed in wild type sporangia.
###### Movie showing slices through a cryo-electron tomogram of *spoIIDMP* mutant sporangium shown in [Figure 3---figure supplement 3C](#fig3s3){ref-type="fig"}.
10.7554/eLife.45257.023
Surprisingly, the thickness of individual DMP mutant septum was irregular across the septal length, with thicker regions of more than 45 nm, that were not observed in wild type cells ([Figure 3G](#fig3){ref-type="fig"}, [Figure 3---figure supplement 3E](#fig3s3){ref-type="fig"}). One possible explanation for this finding is that in the absence of DMP, proteins involved in PG synthesis persist at the septum, and their continued activity leads to thicker and less flexible septal regions ([Figure 3J](#fig3){ref-type="fig"}). To test this model, we stained wild type and DMP mutant sporangia with BOCILLIN-FL, a fluorescent-derivative of penicillin V with affinity for multiple penicillin-binding proteins (PBPs) ([@bib61], see Materials and methods). We observed continuous fluorescent signal around the septa in both strains, but the signal was brightest at the leading edge in wild type sporangia ([@bib38]) and at different positions across the septum in DMP mutant sporangia ([Figure 3---figure supplement 4](#fig3s4){ref-type="fig"}). This mislocalization might allow ongoing synthesis of septal PG, leading to abnormally thick sporulation septa.
PG is synthesized ahead of the leading edge of the engulfing membrane {#s2-5}
---------------------------------------------------------------------
Once the septum curves, the mother cell membrane starts to migrate around the forespore. Since the PG is not completely degraded, it will represent a major obstacle for the advancement of the engulfing membrane. To explain how cells overcome this hurdle, we previously proposed a conceptually new model for engulfment in which coordinated PG synthesis and degradation at the leading edge of the engulfing membrane moves the junction between the septum and the lateral cell wall around the forespore, making room for the engulfing mother cell membrane to advance ([Figure 1B](#fig1){ref-type="fig"}; [@bib38]). In this 'make before break' model, new PG would be synthesized ahead of the engulfing membrane by forespore-associated PG biosynthetic complexes and subsequently degraded by DMP. To test this model, we first focused on the shape of the forespore membrane opposing the leading edge of the engulfing mother cell membrane. In wild type sporangia, the forespore membrane was rounded immediately ahead of the engulfing membrane ([Figure 4A](#fig4){ref-type="fig"}). This could be due to accumulation of additional PG at this site, which might push and deform the forespore membrane, introducing a broader curve. To confirm this, we analyzed the shape of the forespore membrane of sporangia in which PG synthesis was blocked with either bacitracin or cephalexin ([Figure 4B](#fig4){ref-type="fig"}, [Figure 4---figure supplement 1](#fig4s1){ref-type="fig"}, see Materials and methods). The forespore membrane was less rounded and had a sharp corner with a radius of curvature that was four times smaller than that of untreated cells (\~30 nm vs. \~120 nm, [Figure 4C](#fig4){ref-type="fig"}). These results indicate that new PG is indeed synthesized ahead of the leading edge of the engulfing membrane, and that it deforms the forespore membrane at the junction between the septum and the lateral cell wall.
{ref-type="fig"}.](elife-45257-fig4){#fig4}
To exclude the possibility that new PG synthesis also happened behind the leading edge of the engulfing membrane, we analyzed septal thickness in sporangia treated with cephalexin and a combination of cephalexin and penicillin V using cryo-FIB-ET. If the septum is thinner in the presence of antibiotics, it would suggest that additional septal PG is synthesized after the DMP complex advances and degrades the PG ahead of the leading edge of the engulfing membrane. However, if septal thickness is independent of the presence of antibiotics, it would suggest that PG is not normally synthesized behind the DMP complex. Our data showed that the septal thickness of cells treated with cephalexin was comparable to that of untreated cells. Surprisingly, the septal thickness of cells treated with a combination of cephalexin and penicillin V was \~3 nm greater than that of untreated cells ([Figure 4---figure supplement 2](#fig4s2){ref-type="fig"}). This modest increase in septal thickness is of a magnitude that is consistent with recent molecular dynamics simulations which show that relaxed PG is thicker than stretched PG ([@bib4]). It is possible that treatment with antibiotics blocking PG synthesis inhibits the stretching of septal PG due to the absence of membrane migration which could pull the septal PG around the forespore. Alternatively, the residual DMP might partially cleave septal PG, releasing the tension and leading to slightly thicker septa. Taken together, in both cases of antibiotic treatment the septal thickness did not decrease compared to untreated sporangia, suggesting that PG is not synthesized behind the DMP complex.
3D architecture of the leading edge of the engulfing membrane {#s2-6}
-------------------------------------------------------------
The new PG synthesized ahead of the leading edge of the engulfing membrane might interfere with the movement of the engulfing membrane, until it is degraded by DMP. To obtain insights about the movement of the engulfing membrane, we annotated the forespore and the mother cell membranes in our tomograms to visualize the architecture of the leading edge in 3D. Our data showed that the leading edge of the engulfing membrane in wild type sporulating cells moved around the forespore in tiny finger-like projections ([Figure 4D--G](#fig4){ref-type="fig"}, [Figure 4---figure supplement 3](#fig4s3){ref-type="fig"}). The projections were \~10--30 nm wide and \~5--20 nm long, with significant variations from cell to cell. We hypothesized that those projections could be due to the uneven degradation of PG ahead of the leading edge of the engulfing membrane. To test this possibility, we imaged *spoIIP* mutant sporangia, which lack a functional DMP complex. As expected, the septum bulged towards the mother cell cytoplasm, but the mother cell membrane did not move forward in these cells. No membrane projections anywhere were observed in the mother cell membrane in any of the tomograms that were annotated ([Figure 4H--K](#fig4){ref-type="fig"}, [Figure 4---figure supplement 4](#fig4s4){ref-type="fig"}, [Video 9](#video9){ref-type="video"}). Also, no projections were observed in the DMP triple mutant, most of which did not form bulges ([Figure 4---figure supplement 5](#fig4s5){ref-type="fig"}). These findings suggest that SpoIIDMP is necessary for the formation of finger-like projections in the engulfing mother cell membrane.
###### Movie showing slices through a cryo-electron tomogram of *spoIIP* mutant sporangium shown in [Figure 4H](#fig4){ref-type="fig"}.
10.7554/eLife.45257.031
We next tested whether PG synthesis was also required for the formation of finger-like projections. To study this possibility, we focused on sporangia in which PG synthesis was blocked using cephalexin ([Figure 4L--O](#fig4){ref-type="fig"}, [Video 10](#video10){ref-type="video"}). When treated with antibiotics that block PG synthesis, membrane migration is blocked although the forespore continues to grow into the mother cell ([@bib38]). The case of cephalexin-treated cells is more complicated than other antibiotics, because after the septum curves into the mother cell, the leading edge sometimes retracts on one side while advancing slightly on the other ([Video 10](#video10){ref-type="video"}; [@bib38]). This appears to consist of rotation of the 'cup' formed by the bulging septum relative to the lateral cell wall, rather than membrane migration, because the distance between the leading edges does not decrease during this process. Thus, rotation of the septal cup does not reflect the degree to which the forespore is engulfed. Cephalexin inhibits the earliest stages of cell division ([@bib10]; [@bib21]), and therefore we speculate that it might be required to tether the extending septum to the lateral cell wall. In the absence of these bridges, the septum might be free to rotate according to Brownian motion, perhaps anchored by the Q-AH ratchet that can also mediate engulfment in the absence of the cell wall ([@bib7]). We used cryo-FIB-ET to compare the architecture of both sides of the engulfing membrane in cephalexin-treated sporangia ([Figure 4L--O](#fig4){ref-type="fig"}). We did not observe finger-like projections in the side of the membrane that retracts ([Figure 4M](#fig4){ref-type="fig"}) but observed a few projections in the opposite side ([Figure 4N](#fig4){ref-type="fig"}), which might remain tethered to PG ahead of the engulfing membrane. Also, when cells were treated with penicillin V, we observed fewer projections that were shorter compared to untreated cells ([Figure 4---figure supplement 6](#fig4s6){ref-type="fig"}). Taken together, these results suggest that the finger-like projections at the leading edge of the engulfing membrane might be caused by tethering of the engulfing membrane to the PG via DMP ([Figure 4P](#fig4){ref-type="fig"}).
###### Movie showing slices through a cryo-electron tomogram of wildtype *B. subtilis* sporangium treated with cephalexin shown in [Figure 4L](#fig4){ref-type="fig"}.
10.7554/eLife.45257.032
Discussion {#s3}
==========
In this study, we have visualized the developmental process of sporulation in *Bacillus subtilis* using cryo-FIB-ET. We have obtained images of cells during different stages of sporulation at a resolution of a few nanometers, revealing new details about the architecture of spore assembly, as well as several hitherto unknown structures inside and around the developing spore ([Figure 1](#fig1){ref-type="fig"}). Our results also provide mechanistic insights into engulfment, including the early step of septal thinning ([Figures 2](#fig2){ref-type="fig"} and [3](#fig3){ref-type="fig"}) and membrane migration ([Figure 4](#fig4){ref-type="fig"}) which are captured in the model presented in [Figure 5](#fig5){ref-type="fig"}.
{#fig5}
We provide evidence that the septal PG is not completely degraded at the onset of engulfment. Instead, the septum gets slightly (\~25%) thinner as it curves into the mother cell, with PG continuously present between the mother cell and the forespore membranes. In addition, the whole septum transitions from thick to thin homogenously, contrary to previous results that suggested that septal thinning started in the middle and progressed towards the edges. The homogenous thinning of the septum can be explained by our previous finding that septal PG is stretched as the forespore grows towards the mother cell due to increased turgor pressure caused by SpoIIIE-mediated chromosome translocation ([@bib25]). Thus, septal thinning could simply represent a change in the conformation of the septal PG, from a relaxed to a stretched state, triggered by the increased turgor pressure in the forespore. This is consistent with recent molecular dynamics simulations on Gram-positive cell walls, which show that relaxed PG fragments are \~25% thicker than those in a strained conformation ([@bib4]).
Our results also indicate that DMP is required to maintain a flexible septum that can curve as the forespore grows into the mother cell ([Figure 3](#fig3){ref-type="fig"}). We found that the septa of DMP triple mutants have irregular thickness and are on average thicker than wild type septa ([Figure 3E--G](#fig3){ref-type="fig"}). Since DMP is produced after polar septation, the septa of DMP mutant sporangia must have thickened after they have been formed. Thus, it is possible that DMP prevents septal thickening by clearing PG synthases from the septum, where they would accumulate after polar septation. This model is consistent with the increased localization of PBPs throughout the septum in DMP mutants ([Figure 3---figure supplement 4](#fig3s4){ref-type="fig"}). Nevertheless, we cannot rule out the possibility that DMP degrades the septal PG partially to generate a flexible septum. If this was the case, the partial degradation should happen simultaneously across the whole septum to enable the homogenous transition from thick to thin. However, since DMP is rate limiting for engulfment, it seems unlikely that there may be enough DMP complexes to mediate the homogeneous thinning of the septum. We therefore favor the model that septal thinning is primarily driven by stretching of septal peptidoglycan.
Our data also provide insights into the mechanism of membrane migration during engulfment. Using fluorescence microscopy, we previously observed that new PG is synthesized at the leading edge of the engulfing membrane ([@bib38]; [@bib48]). We also showed that many forespore penicillin binding proteins (PBPs) can track the leading edge of the engulfing mother cell membrane ([@bib38]), suggesting that PG synthesis at the leading edge of the engulfing membrane is carried out by forespore PBPs. The cryo-FIB-ET images presented here provide further support to this idea: the comparison of the architecture of the leading edge of the engulfing membrane between native sporangia and sporangia in which PG synthesis is blocked by antibiotics shows that new PG deforms and rounds the forespore membrane ahead of the leading edge of the engulfing membrane ([Figure 4A--C](#fig4){ref-type="fig"}), indicating that new PG is synthesized by forespore PBPs immediately ahead of the leading edge of the engulfing mother cell membrane. We propose that DMP complexes at the leading edge of the engulfing membrane target this new PG for degradation, making room for the engulfing membrane to advance ([Figures 1B](#fig1){ref-type="fig"} and [5C](#fig5){ref-type="fig"}). This relatively simple model suggests that engulfment could have evolved by developing new mechanisms to spatially and temporally regulate the conserved protein machineries that synthesize and degrade peptidoglycan.
The 3D reconstruction of the leading edge of the engulfing mother cell membrane shows the presence of finger-like projections that resemble the filopodia of eukaryotic cells ([@bib32]). In eukaryotic cells, these membrane projections are produced by cytoskeletal proteins, and the projections are typically a few micrometers wide. In contrast, no cytoskeletal elements contributing to engulfment have been described so far (or visualized in our tomograms) and the finger-like projections at the leading edge of the engulfing membrane are only a few nanometers wide. Instead, our results demonstrate that DMP is required for the formation of these finger-like projections. The simplest model to explain these data is that DMP tethers the engulfing membrane to the PG ahead of the leading edge of the engulfing membrane, as it degrades the new PG to make room for the engulfing membrane to expand. DMP activity is rate limiting for membrane migration ([@bib1]; [@bib34]), suggesting that there is a discrete number of DMP complexes to remove the steric barrier posed by the newly synthesized PG ahead of the leading edge. The limited number of DMP complexes might cause the PG to be degraded irregularly, generating an uneven membrane front in the form of finger-like projections. The distance between the tips of the finger-like projections is \~5--20 nm, which would require that some DMP complexes are ahead of others by 2 to 8 glycan strands ([@bib50]). Our observations are consistent with the hypothesis that the finger-like projections are produced via the tethering of the DMP complex to existing peptidoglycan, since short fingers are observed when PG synthesis is inhibited, and that these fingers are either stabilized or enhanced by binding newly synthesized peptidoglycan, since longer fingers are observed in the presence of ongoing PG synthesis. This is supported by prior studies of the two enzymes that degrade peptidoglycan (D and P), which have shown that these comprise a processive enzyme complex in which P binds and cleaves PG first, allowing D to bind and cleave PG ([@bib35]). Our cell biological data showed that localization of P to the leading edge of the engulfing membrane is decreased but not completely absent when PG synthesis is inhibited ([@bib38]), suggesting that P can bind to existing PG, but that high affinity binding requires newly synthesized peptidoglycan. This increased binding of P would likely allow the formation of longer finger-like projections. From a functional perspective, we speculate that the finger-like projections could be compared to 'friction ridges', the minutely raised ridges of the epidermis that provide a grasping surface on our fingers ([@bib49]). During engulfment, these membrane projections may provide the engulfing membrane a tighter lateral grip while moving around the forespore, hence serving as a ratchet that prevents backward movement of the membrane.
The application of cryo-FIB-ET has been instrumental in allowing us to visualize and accurately measure structural details of engulfing sporangia, which transpire at a scale of a few nanometers. Our studies support a mechanistic model for the enigmatic process of engulfment, while revealing novel architectural details about engulfment and spore assembly, including intriguing new molecular structures that will require further study to unambiguously identify them. Our findings reveal details about sporulation at an unprecedented resolution and further illustrate the potential of cryo-FIB-ET to reveal critical new information about dynamic biological processes.
Materials and methods {#s4}
=====================
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Reagent\ Designation Source or reference Identifiers Additional\
type (species)\ information
or resource
------------------------------------------------------ --------------------------------------------------------------------------- ------------------------------------------------------------------------- ------------------------------------------------------------------ -----------------------
Strain,\ PY79 ([@bib59]) Tax. ID:1415167 Wild type
strain background (*Bacillus subtilis* PY79)
Strain, strain background (*Bacillus subtilis* PY79) *ΔspoIIP::tet* ([@bib15]) KP513
Strain, strain background (*Bacillus subtilis* PY79) *spoIID298, spoIIM-mls, ΔspoIIP::tet* ([@bib7]) KP4188
Chemical compound, drug FM4-64 Thermo Fisher Scientific Cat\#T13320
Chemical compound, drug Bacitracin MilliporeSigma Cat\#B0125 Conc. used: 50 μg/ml
Chemical compound, drug Penicillin V MilliporeSigma Cat\#1504489 Conc. used: 500 μg/ml
Chemical compound, drug Cephalexin MilliporeSigma Cat\#C4895 Conc. used: 50 μg/ml
Software, algorithm JFilament ([@bib44]) <http://athena.physics.lehigh.edu/jfilament/>
Software, algorithm IMOD ([@bib29]) <http://bio3d.colorado.edu/imod/>;\
RRID: [SCR_003297](https://scicrunch.org/resolver/SCR_003297)
Software, algorithm TomoSegMemTV ([@bib28]) <https://sites.google.com/site/3demimageprocessing/tomosegmemtv>
Software, algorithm Amira Commercial software by Thermo Scientific (formerly FEI) <https://www.fei.com/software/amira-3d-for-life-sciences/>;\
RRID: [SCR_014305](https://scicrunch.org/resolver/SCR_014305)
Software, algorithm SerialEM ([@bib30]) <http://bio3d.colorado.edu/SerialEM/>
Software, algorithm Matlab code to calculate septal distances of sporangia This paper; [Source code 1](#scode1){ref-type="supplementary-material"}
Software, algorithm Matlab code to calculate cell length using fluorescence microscopy images This paper;\
[Source code 2](#scode2){ref-type="supplementary-material"}
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Strains and culture conditions {#s4-1}
------------------------------
We used *Bacillus subtilis* PY79 background for all data acquisition. The strains (see *Key Resources Table*) were grown in LB plates at 30°C. The bacteria were first grown in ¼ diluted LB to OD~600~ \~0.5--0.7. Sporulation was then induced by resuspension in A + B media at 37°C. For cryo-FIB-ET, we collected wild type sporulating cells 1.5--3 hr after sporulation induction. For cells treated with antibiotics, 50 μg/ml of bacitracin, 50 μg/ml of cephalexin and 500 μg/ml of penicillin V were used. Antibiotics were added two hours after induction of sporulation and samples for cryo-FIB-ET were collected either one (for bacitracin) or two hours (for cephalexin, penicillin V and a combination of cephalexin and penicillin V) later. For *spoIIP* and *spoIIDMP* mutant sporangia, cells were collected 2.5 hr after induction of sporulation for cryo-ET.
Cryo-FIB-ET and image processing {#s4-2}
--------------------------------
7 μl of diluted liquid culture was deposited onto glow-discharged (using Pelco easyGlow) holey carbon coated QUANTIFOIL R 2/1 200 mesh copper grids. Manual blotting was performed using Whatman No. one filter paper from the reverse side to remove excess resuspension media such that cells form a monolayer on the grids. They were then immediately plunge-frozen into a liquid ethane/propane mixture cooled by liquid nitrogen using a custom-built vitrification device (Max Planck Institute for Biochemistry, Munich). These grids were then stored in storage boxes in liquid nitrogen until further use.
Vitrified bacterial samples forming a monolayer on the grids were mounted into modified autogrids (Max Planck Institute for Biochemistry) for milling inside a Thermo Scientific Scios DualBeam (cryo-FIB/SEM) (Materials and methods). 100--250 nm thin sections or lamellae (12--15 μm in width) were then prepared using rectangular milling patterns and beam current of 0.1 nA or 0.3 nA for rough milling and 10 pA or 30 pA for fine milling as described in [@bib8]. Tilt series were collected in a 300-keV Tecnai G2 Polara (Thermo Scientific) equipped with a K2 Summit direct detection camera (Gatan) and a post-column Quantum energy filter (Gatan). The samples were usually tilted from −66° to +66° (the range per tilt series depending on the quality of the lamellae) with an increment of 1.5° or 2°. The tilt series were acquired using SerialEM ([@bib30]). The images were recorded at a defocus of −5 to −6 μm at nominal magnifications of 34,000 (pixel size: 0.61 nm) or 41,000 (pixel size: 0.49 nm) with a cumulative dose of \~60--130 e^-^/Å^2^. Images for later stages of sporulation ([Figure 1---figure supplement 2B](#fig1s2){ref-type="fig"}) were acquired at nominal magnification of 22,500 (pixel size: 0.92 nm) as these samples were highly sensitive to radiation damage.
The patch-tracking feature of IMOD was used to reconstruct the tilt-series ([@bib22]). TomosegmemTV ([@bib28]) was used for semi-automatic segmentation of membranes followed by manual refinement in the Amira software package.
Measuring mean radius of ellipsoidal complexes {#s4-3}
----------------------------------------------
The mean radius (r) of an ellipse is given by:$$r = \sqrt{\frac{a^{2} + b^{2}}{2}}$$where *a* is the semi-major axis of the ellipse and *b* is the semi-minor axis of the ellipse. For ellipsoidal complexes observed in our tomograms ([Figure 1J](#fig1){ref-type="fig"}, [Figure 1---figure supplement 3A--C](#fig1s3){ref-type="fig"}), average value of *a* is \~45 nm and that of *b* is \~42 nm.
Calculating septal distances {#s4-4}
----------------------------
'Surface distance' feature of Amira was used to calculate the septal distances between the forespore and mother cell membranes. The septal-disc was color coded according to these values ([Figure 3B--D,G](#fig3){ref-type="fig"}, [Figure 3---figure supplement 2E](#fig3s2){ref-type="fig"}). Five tomograms each of wild type cells representing different stages of engulfment (flat, curved and engulfing septa) were used for analysis. For engulfment mutants, five *spoIIP* sporangia and seven *spoIIDMP* sporangia were analyzed. For antibiotic treated cells, six cephalexin-treated sporangia and five cephalexin- and penicillin V- combination treated sporangia were analyzed. To get the linear profiles of the distances, the data was grouped into smaller bins of approximately equal area for each tomogram. Then the average of the distance within a bin was used to represent the distance between the membranes at that location of the bin (see [Source code 1](#scode1){ref-type="supplementary-material"}). [Figure 3E](#fig3){ref-type="fig"} represents an average of all the profiles for each cell type in [Figure 3---figure supplement 1](#fig3s1){ref-type="fig"} and [Figure 3---figure supplement 3E](#fig3s3){ref-type="fig"}.
Fluorescence microscopy {#s4-5}
-----------------------
Cells were visualized on an Applied Precision DV Elite optical sectioning microscope equipped with a Photometrics CoolSNAP-HQ2 camera and deconvolved using SoftWoRx v5.5.1 (Applied Precision). For experiment outlined in [Figure 3---figure supplement 4](#fig3s4){ref-type="fig"}, the median focal plane of the image is shown. Membranes were stained with 0.5 μg/ml of FM4-64 (Thermo Fisher Scientific) that was added directly to 1.2% agarose pads prepared using sporulation resuspension medium. 10 μg/μl of BOCILLIN-FL was added to 1 ml of culture aliquoted \~2.5 hr after sporulation induction (at 37°C) and washed with sporulation resuspension medium three times. 12 μl of washed cells were then transferred to agarose pads for imaging.
For time-lapse microscopy, sporulation was induced at 30°C. 0.5 μg/ml FM4-64 was added to the cultures \~1.5 hr after sporulation induction and incubation continued for another hour. Composition of agarose pads for time-lapse microscopy is as follows: 2/3 vol of supernatant from the sporulation culture, 1/3 vol 3.6% agarose in fresh A + B sporulation medium, 0.17 μg/ml FM4-64, supplemented with antibiotics according to concentrations mentioned above in '*Strains and culture conditions*'. 12 μl samples were taken 3 hr after resuspension and transferred to the agarose pads. Pads were covered with a glass slide and sealed with petroleum jelly to avoid dehydration during time-lapse imaging. Pictures were taken in an environmental chamber at 30°C every 5 min for \~5 hr. Excitation/emission filters were TRITC/CY5 for membrane imaging. Excitation light transmission was set to 5% to minimize phototoxicity and exposure time was set to 0.1 s.
Calculating cell length using fluorescence microscopy images {#s4-6}
------------------------------------------------------------
To determine the length of vegetative cells over time ([Figure 2---figure supplement 1E](#fig2s1){ref-type="fig"}), membrane contours were extracted from microscopy images for each time frame (up to 60 min post treatment with antibiotics) using semi-automated active contour software JFilament, available as a Fiji plugin ([@bib43]; [@bib44]). The cell length was then calculated by measuring along the long axis of the contours using a custom-built MATLAB script (see [Source code 2](#scode2){ref-type="supplementary-material"}). To plot the average cell length, the data was normalized to the initial cell length for each of the cases.
Calculating radius of curvature {#s4-7}
-------------------------------
To calculate radius of curvature ([Figure 4A--C](#fig4){ref-type="fig"}, [Figure 4---figure supplement 1](#fig4s1){ref-type="fig"}), a slice was taken approximately from the center of the z-stack for each of the tomograms. 'Measure spline curvature' feature of sabl_mpl ([@bib57]) was then used to plot radii of curvatures around the forespore membranes just ahead of the leading edge for seven native sporangia and five antibiotic-treated sporangia wherein membrane migration appears to be blocked.
Funding Information
===================
This paper was supported by the following grants:
- http://dx.doi.org/10.13039/100000002National Institutes of Health National Institute of Health\'s Director\'s New Innovator Award (1DP2GM123494) to Elizabeth Villa.
- http://dx.doi.org/10.13039/100000002National Institutes of Health RO1-GM057045 to Kit Pogliano, Elizabeth Villa.
This work was supported by National Institutes of Health Director's New Innovator Award 1DP2GM123494 (EV), and the National Institutes of Health R01-GM057045 (EV and KP). We used the UC San Diego Cryo-EM Facility (partially supported by a gift from the Agouron Institute to Tim Baker), and the San Diego Nanotechnology Infrastructure of UC San Diego (supported by the NSF grant ECCS-1542148). The authors would like to thank Antonio Martínez-Sánchez for useful insights into data analysis.
Additional information {#s5}
======================
No competing interests declared.
Conception and design, Data acquisition, Data analysis, Writing---original draft.
Conception and design, Data acquisition, Data analysis, Writing---original draft.
Data analysis.
Assistance with cryo-EM.
Data analysis.
Data analysis.
Conception and design, Data analysis, Writing---review and editing.
Conception and design, Data analysis, Writing---review and editing.
Additional files {#s6}
================
10.7554/eLife.45257.034
###### Matlab code to calculate septal distances of sporangia.
10.7554/eLife.45257.035
###### Matlab code to calculate cell length using fluorescence microscopy images.
10.7554/eLife.45257.036
Data availability {#s7}
=================
The authors have created a library of *B. subtilis* tomograms accessible at: <http://villalab.ucsd.edu/research/engulfment>. The authors have also deposited three representative tilt-series to Electron Microscopy Data Bank (EMDB) in the form of 4x binned tomograms. The IDs are EMD-20335, EMD-20336, EMD-20337 for Figure 1D,F,H respectively.
The following datasets were generated:
KhannaKLopez-GarridoJZhaoZWatanabeRYuanYSugieJPoglianoKVillaE2019Representative tilt-seriesElectron Microscopy Data BankEMD-20335
KhannaKLopez-GarridoJZhaoZWatanabeRYuanYSugieJPoglianoKVillaE2019Representative tilt-seriesElectron Microscopy Data BankEMD-20336
KhannaKLopez-GarridoJZhaoZWatanabeRYuanYSugieJPoglianoKVillaE2019Representative tilt-seriesElectron Microscopy Data BankEMD-20337
10.7554/eLife.45257.044
Decision letter
Mignot
Tâm
Reviewing Editor
CNRS-Aix Marseille University
France
Morlot
Cecile
Reviewer
Institut de Biologie Structurale
France
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for sending your article entitled \"The molecular architecture of engulfment during *Bacillus subtilis* sporulation\" for peer review at *eLife*. Your article is being evaluated by two peer reviewers, and the evaluation is being overseen by a Reviewing Editor and Gisela Storz as the Senior Editor.
Engulfment is a phagocytic-like process that is essential for bacterial sporulation. During engulfment, the mother cell membrane migrates around the developing spore, generating a cell within a cell. Membrane migration involves cell wall synthesis and remodeling events that remain incompletely resolved. In this study, authors use state-of-the-art cryo-FIB-ET to investigate these unresolved aspects in *Bacillus subtilis*. Their data confirm earlier observations (obtained by ECT, Tocheva et al., 2013) that a layer of PG remains in the intermembrane space separating the mother cell and forespore. In addition, the unprecedented resolution of their images provides new insights into the thickness of this septal PG layer, as well as the morphology of the engulfing membranes at different stages of the engulfment process and upon impaired PG degradation or synthesis.
Essential revisions:
While their observations represent a significant contribution to the field, there are two major concerns regarding their conclusions, both based on the fact that the extend of PG inhibition in the presence of antibiotics is not clear:
\- Authors observe finger-like projections associated with migrating membranes. They provide convincing evidence that these projections are associated with the DMP complex and they propose that they result from \"tethering of the engulfing membrane to newly synthesized PG via DMP\". However, these finger-like projections are still observed at the leading side of the engulfing membrane, although being less prominent, when PG synthesis is inhibited by cephalexin (Figure 4N and subsection "3D architecture of the leading edge of the engulfing membrane", second paragraph).
Do authors really consider that PG synthesis is fully inhibited in the presence of cephalexin? If so, remnant finger-like projections and membrane migration still occurs when PG synthesis is inhibited. They might result from tethering of the engulfing membrane to old PG surrounding the forespore via DMP. In that case, finger-like projections may be stabilized/enhanced (rather than \"caused\") by tethering of the engulfing membrane to newly synthesized PG. To resolve this problem, the authors need to test the extent of PG synthesis inhibition by Cephalexin.
\- Figure 4B shows a reduced rounding of the forespore membrane in the presence of antibiotics, providing evidence that PG is synthesized at the leading edge of the engulfing membrane. Based on this observation and a previous study (Ojkic et al., 2016), authors propose a model (Figure 5) in which PG is exclusively synthesized ahead of the leading edge of the engulfing membrane. However, these data do not exclude the possibility that PG synthesis also happens behind the leading edge of the engulfing membrane, or in other words that PG is resynthesized after degradation by the DMP complex. Indeed, as pointed in the previous point, the fact that membrane migration still happens in the presence of antibiotics again questions whether PG synthesis is fully inhibited. In cells displaying incomplete or asymmetrical engulfment (such as those shown in Figure 2), PG synthesis might have continued for a while and stopped, resulting in septal disks containing newly synthesized PG and leading edges showing impaired PG synthesis. Analysis of septum thickness (as performed in Figure 3B-E) in the presence of antibiotics might shed light on this issue: a thinner septum (but not absent if DMP does not totally degrades septal PG) in the presence of antibiotics would suggest that PG is synthesized behind the DMP complex. If septum thickness is independent of the presence of antibiotics, it suggests that either PG is not synthesized behind the DMP complex or that PG is not fully inhibited in the presence of antibiotics. This uncertainty impacts the hypothesis proposed by the authors because PG synthesis happening behind the DMP complex might explain why a slight and homogenous septal thinning is observed (Figure 3); it might also contribute to the residual PG layer observed in the intermembrane space
10.7554/eLife.45257.045
Author response
This revised version includes additional experiments and analysis to address the reviewers' comments. Here is a summary of the main points: We did additional experiments to more precisely test the extent of peptidoglycan (PG) synthesis inhibition upon antibiotic treatment to ensure that our results are representative. In addition to using septation as a readout for PG synthesis, we determined the rate of elongation of vegetative cells when treated with different antibiotics and tested two additional drug treatments -- penicillin V and a combination of cephalexin and penicillin V. We found that a combination of cephalexin and penicillin V was most effective to ensure complete block of both septation and elongation (Figure 2---figure supplement 1, Videos 7-10). We also acquired cryo-electron tomography data of wild type sporangia treated with penicillin V and a combination of cephalexin and penicillin V (Figure 2, Figure 2---figure supplement 2), measured the septal thickness of sporangia treated with cephalexin and a combination of cephalexin and penicillin V, and compared it to untreated sporangia (Figure 4---figure supplement 2). The data showed that the septal thickness did not decrease upon antibiotic treatment which suggested that PG synthesis did not precede the mother cell DMP complex. In addition, we have modified the text wherever necessary to explain the impact of antibiotic treatment on septal architecture and engulfment in a more detailed and clear manner.
> Essential revisions:
>
> While their observations represent a significant contribution to the field, there are two major concerns regarding their conclusions, both based on the fact that the extend of PG inhibition in the presence of antibiotics is not clear:
We agree that determining the extent of PG inhibition in the presence of antibiotics is key for the interpretation of our results. We explain below the experiments we have already done to determine this, and propose additional experiments to more precisely test the extent of PG synthesis inhibition. We will also collect additional cryo-electron tomograms of sporulating cells treated with additional PG inhibitors to ensure that our results are representative.
We anticipate being able to submit a revised manuscript including the new experiments in less than two months. We expect that the cryo-EM experiments can be completed within 2-3 weeks. However, based on our experience with cryo-EM techniques that are not routine and require more than one high-end microscope to be online, we have learned to be conservative when making predictions about our timelines. We are confident that two months will suffice to perform the described experiments, regardless of potential instrument downtime.
> \- Authors observe finger-like projections associated with migrating membranes. They provide convincing evidence that these projections are associated with the DMP complex and they propose that they result from \"tethering of the engulfing membrane to newly synthesized PG via DMP\". However, these finger-like projections are still observed at the leading side of the engulfing membrane, although being less prominent, when PG synthesis is inhibited by cephalexin (Figure 4N and subsection "3D architecture of the leading edge of the engulfing membrane", second paragraph).
>
> Do authors really consider that PG synthesis is fully inhibited in the presence of cephalexin? If so, remnant finger-like projections and membrane migration still occurs when PG synthesis is inhibited. They might result from tethering of the engulfing membrane to old PG surrounding the forespore via DMP. In that case, finger-like projections may be stabilized/enhanced (rather than \"caused\") by tethering of the engulfing membrane to newly synthesized PG. To resolve this problem, the authors need to test the extent of PG synthesis inhibition by Cephalexin.
Thank you for this comment, which made us realize that some of our descriptions were imprecise.
First, we did not completely explain the impact of antibiotic treatment on septal architecture and engulfment. We previously showed that engulfment membrane migration does not continue when PG synthesis is inhibited (see Ojkic et al., 2016, Figure 1 and its supplements), although forespore growth causes the septum to stretch and curve into the mother cell, making it appear that membrane migration has occurred. Measurements of individual cells during timelapse microscopy shows that the distance between the leading edges of the engulfing membrane decreases during engulfment in untreated cells. However, after antibiotic treatment, this distance does not decrease, indicating that the mother cell membrane does not migrate around the forespore. The case of cephalexin-treated cells is more complicated than other antibiotics, because after the septum curves into the mother cell, the leading edge sometimes retracts on one side while advancing slightly on the other. This appears to consist of rotation of the 'cup' formed by the bulging septum relative to the lateral cell wall, rather than membrane migration, because the distance between the leading edges does not decrease during this process. Thus, rotation of the septal cup does not increase the degree to which the forespore is engulfed. We are uncertain as to the mechanistic basis for this rotation, but our timelapse experiments indicate that it is a consequence of cephalexin treatment, although it probably also happens after treatment with some penicillins (Ojkic et al. Figure 1---figure supplement 1). Cephalexin inhibits the earliest stages of cell division (Adam et al., 1997; Eberhardt et al., 2003; Kocaoglu et al., 2012), and therefore we speculate that it might be required to tether the extending septum to the lateral cell wall. In the absence of these bridges, the septum might be free to rotate according to Brownian motion, perhaps anchored by the Q-AH ratchet that can also mediate engulfment in the absence of the cell wall (Broder and Pogliano, 2006). We will describe the phenotypes produced by antibiotic treatment more precisely in the text. One area where this will clearly be necessary is where we refer to the side of the membrane that has retracted as the "lagging side" and the other one as the "leading side", which may have created the misunderstanding that membrane migration happens in cephalexin-treated cells. A revision of this section, and expanded description of the phenotypes, will greatly improve our revised submission.
Second, the reviewers correctly point out that our observations are consistent with the hypothesis that the finger like projections are produced via the tethering of the DMP complex to existing peptidoglycan, since short fingers are observed when PG synthesis is inhibited, and that these fingers are either stabilized or enhanced by binding newly synthesized peptidoglycan, since longer fingers are observed in the presence of ongoing PG synthesis. This is supported by prior studies of the two enzymes that degrade peptidoglycan (D and P), which have shown that these comprise a processive enzyme complex in which P binds and cleaves PG first, allowing D to bind and cleave PG (Morlot and Rudner, 2010). Our cell biological data showed that localization of P to the leading edge of the engulfing membrane is decreased but not completely absent when PG synthesis is inhibited (Ojkic Figure 2F-G and associated supplementary figure), suggesting that P can bind to existing PG, but that high affinity binding requires newly synthesized peptidoglycan. This increased binding of P would likely allow the formation of longer finger like projections. Thus, we thank the reviewers for this suggestion, and will revise the manuscript accordingly.
We agree that it is critical to determine the extent of PG inhibition upon antibiotic treatment. The experiments in this manuscript were based on previously published results of time-lapse microscopy in which we assessed if new division events were blocked after antibiotic treatment, thereby using cell division as a read out for PG synthesis (Ojkic et al., Figure 1---figure supplement 2). These studies showed that cell division was completely blocked by cephalexin, indicating that septal PG synthesis was blocked, and that PG synthesis was inhibited within a few minutes of cephalexin treatment, because cells that were already undergoing septation were unable to complete septum formation. To complement this analysis, we will determine the elongation rate of the vegetative cells that are present in sporulating cultures under our experimental conditions, in order to identify an antibiotic or combination of antibiotics that inhibit both septation and elongation, and we will then collect additional cryo-FIB-ET of cells treated with this antibiotic(s). This will allow us to address the question of the extent of PG synthesis inhibition, and the second essential revision (see below). Our preferred antibiotic candidate is Penicillin V, which has a broad affinity for many *B. subtilis* PBPs (Lakaye et al., 1994; Zhao et al., 1999; Kocaoglu, 2012), and which our preliminary data indicates severely affects both septation and cell elongation.
> \- Figure 4B shows a reduced rounding of the forespore membrane in the presence of antibiotics, providing evidence that PG is synthesized at the leading edge of the engulfing membrane. Based on this observation and a previous study (Ojkic et al., 2016), authors propose a model (Figure 5) in which PG is exclusively synthesized ahead of the leading edge of the engulfing membrane. However, these data do not exclude the possibility that PG synthesis also happens behind the leading edge of the engulfing membrane, or in other words that PG is resynthesized after degradation by the DMP complex. Indeed, as pointed in the previous point, the fact that membrane migration still happens in the presence of antibiotics again questions whether PG synthesis is fully inhibited. In cells displaying incomplete or asymmetrical engulfment (such as those shown in Figure 2), PG synthesis might have continued for a while and stopped, resulting in septal disks containing newly synthesized PG and leading edges showing impaired PG synthesis. Analysis of septum thickness (as performed in Figure 3B-E) in the presence of antibiotics might shed light on this issue: a thinner septum (but not absent if DMP does not totally degrades septal PG) in the presence of antibiotics would suggest that PG is synthesized behind the DMP complex. If septum thickness is independent of the presence of antibiotics, it suggests that either PG is not synthesized behind the DMP complex or that PG is not fully inhibited in the presence of antibiotics. This uncertainty impacts the hypothesis proposed by the authors because PG synthesis happening behind the DMP complex might explain why a slight and homogenous septal thinning is observed (Figure 3); it might also contribute to the residual PG layer observed in the intermembrane space.
Thank you for raising this important point. As proposed by the reviewers, we have analyzed septal thickness in cephalexin-treated sporangia. As seen in [Author response image 1](#respfig1){ref-type="fig"}, the septal thickness for untreated and cephalexin-treated wild type sporangia is comparable (17.46 ± 1.17 nm for untreated and 18.72 ± 1.13 nm for cephalexin-treated sporangia). This therefore suggests that either PG is not synthesized behind the DMP complex or that PG is not fully inhibited in the presence of antibiotics.
{#respfig1}
However, our published results (Ojkic et al., 2016) indicate that septal PG synthesis is completely inhibited in cephalexin treated cells, suggesting that PG synthesis does not happen behind DMP. To provide additional evidence on this important point, we will perform the additional experiments explained above to determine if both elongation and septal PG synthesis is completely inhibited by the antibiotics used in our experiments, and if not, to identify an antibiotic treatment or combinatorial treatment that blocks both processes. We will also collect cryo-electron tomograms of sporangia treated with at least one additional antibiotic (probably Penicillin V or a combination of Penicillin V and cephalexin) and perform a similar analysis to further test this point.
| {
"pile_set_name": "PubMed Central"
} |
{.253}
{.254}
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .362}
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-antioxidants-09-00293}
===============
The family Lamiaceae has 236 genera and about 6900 to 7200 species. *Salvia* (\~900 species) is one of the largest genera of Lamiaceae \[[@B1-antioxidants-09-00293]\]. The name derives from the Latin word "salveo", that means "to save, to heal" \[[@B2-antioxidants-09-00293]\]. *Salvia* species have been used as tea since ancient times to prevent colds, coughs, nervous exhaustion, stomach and abdominal pain, pharyngitis, inflammation of the mouth, inflammation of the gums, excessive sweating, and increased lactation \[[@B3-antioxidants-09-00293],[@B4-antioxidants-09-00293],[@B5-antioxidants-09-00293]\]. Many studies on *Salvia* species have shown that plants, extracts, and essential oils possess biological activities such as antiseptic, antifungal, antibacterial, antiviral, analgesic, antispasmodic, antioxidant, astringent, hallucinogenic, central nervous system depressant, anticancer, cardiovascular, antidiabetic, and insecticidal activities \[[@B6-antioxidants-09-00293]\]. As a result of the phytochemical studies, it has been learned that these plants are rich in flavonoids, phenolic compounds, as well as diterpenes and triterpenes \[[@B7-antioxidants-09-00293],[@B8-antioxidants-09-00293],[@B9-antioxidants-09-00293]\]. These compounds show a natural antioxidant property by stopping or inhibiting the reactions caused by free radicals \[[@B10-antioxidants-09-00293],[@B11-antioxidants-09-00293]\].
*Salvia aramiensis* Rech. f. grown in *Pinus brutia* woodlands, rocky places, and limestones in Hatay (Turkey) province, are perennial, evergreen, and subshrub with woody stems \[[@B12-antioxidants-09-00293]\]. Its flowers and leafy branches are used as stomachic herbal tea \[[@B13-antioxidants-09-00293]\]. The literature has focused on *S. aramiensis* essential oil composition and its antioxidant and antimicrobial activity \[[@B14-antioxidants-09-00293]\]. According to literature, *S. aramiensis* essential oil is a potent antimicrobial and antioxidant agent \[[@B2-antioxidants-09-00293],[@B6-antioxidants-09-00293],[@B15-antioxidants-09-00293]\]. To date, the biological activity and chemical composition of *S. aramiensis* have not been investigated.
Despite extensive research into the discovery of new collagenase, elastase, and hyaluronidase enzyme-inhibiting compounds of both synthetic and natural origin, it is still a major basis for recent inhibitors of these enzymes due to the side effects or low efficacy of existing enzymes. Also, the current number of these enzyme inhibitors is very limited, and recent inhibitors are in demand mainly in the cosmetic and pharmaceutical industry (wound healing) \[[@B16-antioxidants-09-00293]\]. The dermis, the middle layer of the skin, consists of elastin and collagen, the main component of the connective tissue. These proteins are responsible for the resistance and elasticity of the skin and are destroyed as a result of the formation of free radicals and the induction of elastase and collagenase enzymes \[[@B17-antioxidants-09-00293]\]. Collagen and elastin also play an important role in the wound-healing process. Inhibition of collagenase activity retards the progression of formation of pre-collagen fibers \[[@B18-antioxidants-09-00293]\]. Overproduced elastase enzyme accelerates the degradation of the surrounding healthy tissue around the wound by catalyzing this protein \[[@B18-antioxidants-09-00293],[@B19-antioxidants-09-00293]\]. Reactive oxygen species (ROS) are one of the factors that trigger skin aging and that delay the wound healing process by causing oxidative damage of skin lipids, proteins, and DNA \[[@B20-antioxidants-09-00293],[@B21-antioxidants-09-00293]\]. To increase the antiaging effect, to prevent loss of skin elasticity, and to accelerate wound healing, it is essential to find inhibitors of elastase and collagenase enzymes, which have a radical scavenging feature.
Encapsulation technologies have been used to increase the effectiveness of the active compounds using drug delivery systems in situations where water solubility is low and to improve long-term stability \[[@B22-antioxidants-09-00293]\]. Of them, nanoliposomes (LPs) are spherical, single or multi-layered vesicles that can be micro- or nanosized but can trap both hydrophobic and hydrophilic compounds \[[@B23-antioxidants-09-00293]\]. LPs are known as systems that provide and enhance the passage of active compounds both in the epidermis and in the deeper layers of the skin due to the similarity to cell membrane structure. Also, LPs are biologically compatible, biodegradable, non-immunogenic, and nontoxic systems that are widely used in cosmeceuticals \[[@B24-antioxidants-09-00293],[@B25-antioxidants-09-00293]\].
For hundreds of years, natural ingredients have been used mainly for antioxidant, antimicrobial, and enzyme inhibitory activities for skin care and wound healing \[[@B26-antioxidants-09-00293]\]. For this purpose, we studied (1) the antioxidant activities with radical scavenging assays and inhibition of β-carotene/linoleic acid co-oxidation of *S. aramiensis* extracts, (2) inhibitory effects of the most antioxidant active extracts (70% methanol) on elastase and collagenase enzymes, (3) the chemical composition of the 70% methanol extract, (4) the preparation and characterization of novel liposomal formulation containing 70% methanol extract, and (5) toxicity of 70% methanol extract and liposomal formulation on L929 fibroblast cell line.
2. Materials and Methods {#sec2-antioxidants-09-00293}
========================
2.1. Plant Sample and Reagents {#sec2dot1-antioxidants-09-00293}
------------------------------
*Salvia aramiensis* which was used for experimental studies was collected on 01.06.2016 from Hatay (Belen), Turkey. Herbarium sample of the plant is stored in Mustafa Kemal University Science and Literature Faculty Herbarium (Plant Collector no: 1864). All reagents and chromatographic standards were obtained from Sigma Chemical Company (St. Louis, MO, USA).
2.2. Extraction Procedure {#sec2dot2-antioxidants-09-00293}
-------------------------
The dried whole aerial part of the plant material (stem, leaves, and flower) was powdered and divided into three parts. The first part (50 g) was extracted with 70% methanol (MeOH) and the second part (50 g) was extracted with 70% ethanol (EtOH) three times in a shaking water bath at 37 °C for 24 h. The resulting extracts were combined separately and concentrated under vacuum with a rotary evaporator (37--38 °C). Since traditional use is in the form of tea, 2 g of aerial plant material is brewed with 100 mL of boiling water and left to rest for 10 min. This 2% form of infusion was then lyophilized. Until the analysis time, lyophilized extracts were stored at −20 °C.
2.3. Total Phenolic (TFC) and Flavonoid Content (TPC) {#sec2dot3-antioxidants-09-00293}
-----------------------------------------------------
TFC was calculated as gallic acid equivalents (GAE) in mg/g dry weight of plant material \[[@B27-antioxidants-09-00293]\]. TPC content was calculated as catechin (CA) equivalents in mg/g dry weight of plant material through the aluminum chloride colorimetric assay \[[@B28-antioxidants-09-00293]\].
2.4. Antioxidant Activity {#sec2dot4-antioxidants-09-00293}
-------------------------
### 2.4.1. 1,1-Diphenyl-2-Picrylhydrazyl Radical (DPPH^●^) Scavenging Activity {#sec2dot4dot1-antioxidants-09-00293}
The determination of DPPH^●^ radical scavenging was carried out by the method of Braca et al. \[[@B29-antioxidants-09-00293]\]: 500 µL of sample (0.025, 0.05, 0.1, 0.15, 0.2, 0.3, 0.4, 0.6, 0.8, 1, 1.5, and 2 mg/mL) and standard rosmarinic acid (5, 10, 15, 20, 25, 30, 40, 50, 75, and 100 µg/mL) were added to 1.5 mL of a 0.1 mmol/L MeOH solution of DPPH. Percent inhibition was calculated using Equation (1) after reading the absorbance at 517 nm.
### 2.4.2. 2,2'-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS^+●^) Radical Scavenging Activity {#sec2dot4dot2-antioxidants-09-00293}
The determination of ABTS^+●^ radical scavenging was performed as declared by Huang et al. \[[@B30-antioxidants-09-00293]\]. The ABTS^●+^ radical (7 mM) and potassium persulfate (2.45 mM) were dissolved in water to a final concentration and left for 16 h in the dark at room temperature. Extracts and the standard (rosmarinic acid) were prepared at 50 and 100 µg/mL concentrations. Trolox was used as a reference standard; 0.1 mL of each sample was mixed with 3.9 mL ABTS^●+^ solution. After addition of each extract and standards, absorbances were read for 30 min at 734 nm each minute. The results are expressed as the Trolox equivalent antioxidant capacity (TEAC, mmol/L Trolox).
### 2.4.3. Determination of Inhibition of β-Carotene/Linoleic Acid Co-Oxidation {#sec2dot4dot3-antioxidants-09-00293}
Antioxidant activity of *S. aramiensis* extracts was evaluated according to the *β*-carotene bleaching method of Velioğlu et al. \[[@B31-antioxidants-09-00293]\]. *β*-carotene (5 mg) was dissolved in chloroform (5 mL CHCl~3~), and then *β*-carotene solution (1.2 mL) was added to the flask containing linoleic acid and Tween 20. CHCl~3~ was removed by evaporation from the flask. Distilled water (300 mL) was quietly dropped to the mixture for emulsifying. The control solution was prepared by the same method without adding *β*-carotene. For autoxidation, the samples were set into a 50 °C water bath for 120 min and the blenching level was recorded every 15 min. Measurements were made at 470 nm for extracts, standards, and controls. Rosmarinic acid was used as a positive control. Antioxidant activity was calculated according to Equation (2).
2.5. Enzyme Inhibitory Activity {#sec2dot5-antioxidants-09-00293}
-------------------------------
### Anti-Collagenase (MMP-1) and Anti-Elastase Activities
The inhibitory effects of 70% MeOH extract of *S. aramiensis* on the collagenase and elastase enzymes, which cause collagen and elastin degradation and is induced by the formation of free radicals, were measured. The measurements were performed according to the kit procedures of Collagenase and Elastase Elisa kits (Sun Red Bio, China); 70% MeOH extracts were studied at 50, 100, and 200 µg/mL, and positive control rosmarinic acid was studied at 25 µg/mL concentration \[[@B17-antioxidants-09-00293]\].
2.6. Qualitative and Quantitative Chromatographic Analysis {#sec2dot6-antioxidants-09-00293}
----------------------------------------------------------
### 2.6.1. Analysis with Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) High Performance Liquid Chromatography (HPLC) Systems {#sec2dot6dot1-antioxidants-09-00293}
LC-MS/MS analysis was performed with Absciex 3200 Q trap MS/MS detector. Experiments were carried out using Shimadzu 20A HPLC system coupled to an Applied Biosystems 3200 Q-Trap LC-MS/MS instrument equipped with an electrospray ionization (ESI) source operating in negative ion mode. A GL Science Intersil octadecyl silica gel (250 × 4.6 mm, i.d., 5 µm particle size) analytical column has been used for the chromatographic separation (at 40 °C). The flow rate of the solvent was set to 0.3 mL/min. Photo Diode Array (PDA) detector was used for detection. The elution gradient consists of mobile phases A and B and contains acetonitrile:water:formic acid. The ratios are 10:89:1--89:10:1, v/v/v, respectively. The composition of B was increased from 10% to 100% in 40 min. Collected data were handled by Analyst 1.6 software.
### 2.6.2. Analysis with High Performance Liquid Chromatography (HPLC) Systems {#sec2dot6dot2-antioxidants-09-00293}
HPLC experiments were performed with Agilent HP1100. Separations were performed reverse-phase Mediterranean-C18 analytical column (250 × 4.6 mm i.d., 5 µm particle size). At room temperature, flow rate was set to 1 mL/ min. Detection was carried out with PDA detector between the wavelengths of 200 and 550 nm. Three sets of solvent systems---methanol/water/acetic acid (10:88:2, v/v/v) (solvent A), methanol/water/acetic acid (90:8:2, v/v/v) (solvent B), and methanol (solvent C)---were used for elution. Injections were brought to equilibrium for 10 min. A complete confirmation of assay consisting of linearity, the lower limit of detection and quantitation (LOD and LOQ), intraday and interday accuracy, and precision of the method were performed.
2.7. Liposomal Formulation Studies {#sec2dot7-antioxidants-09-00293}
----------------------------------
### 2.7.1. Preparation of Nanoliposomes {#sec2dot7dot1-antioxidants-09-00293}
Dry film hydration method was used for preparing LPs; Dipalmitoylphosphatidylcholine (DPPC) and Cholesterol (Ch) were added in a 1:1 molar ratio to the flask. They were dissolved with methanol:chloroform (3:1 v/v) and evaporated under reduced pressure at \~ 42--44 °C using Heidolph Rotavapor. The dry film was hydrated by 1 mg/mL 70% MeOH extract of *S. aramiensis* and vortexed. After ultracentrifugation of LP suspension at 15,000 rpm at 4 °C for 60 min, the supernatant and LPs were separated \[[@B32-antioxidants-09-00293]\].
### 2.7.2. Determination of Nanoliposomes Characteristics {#sec2dot7dot2-antioxidants-09-00293}
Zetasizer Nano ZS-Malvern was used to determine characterization parameters of LPs: Particle size (PS), zeta potential (ZP), and polydispersity index (PDI) were determined. After ultracentrifugation, the supernatant phase was analyzed by HPLC for determining the 70% MeOH extract contents of LPs. The types of LPs were monitored using scanning electron microscopy (SEM).
### 2.7.3. In vitro Release Study {#sec2dot7dot3-antioxidants-09-00293}
Release study was performed using Franz diffusion cells with a dialysis membrane (12,000 Dalton pore size) for 24 h at 37 °C. The donor compartment of the diffusion cells was filled with 70% MeOH extract-loaded LP suspension (2 mL); 2 mL phosphate buffer (pH 7.4) was added to the receptor compartment. At the end of 24 h, the released 70% MeOH extract amounts were analyzed by HPLC as described above.
2.8. Toxicity Assessment on L929 Cell Line {#sec2dot8-antioxidants-09-00293}
------------------------------------------
The L929 cell line (mouse fibroblast) was obtained from *American Type Culture Collection* (ATCC^®^ CCL-1™) Manassas, VA, USA). For L929 cells, 10% solution of Eagle's Minimum Essential Medium (EMEM), inactivated horse serum, medium containing the antibiotic mixture, and L-glutamine were used.
The toxicity of the 70% MeOH extract and 70% MeOH extract-loaded liposomes on L929 cell line was determined by the Sulforhodamine B colorimetric (SRB) assay. Cells were counted and seeded in 96-well plates (1 × 10^4^ in 100 μL cells/well). The plates were then kept for 24 h before adding the samples. Stock solution extract was prepared at the concentration of 2 mg/mL in cell culture medium containing 0.5% DMSO (Dimethyl sulfoxide) and diluted to 7.81, 15.625, 31.25, 62.5, 125, 250, 500, and 1000 μg/mL concentrations. The control group was prepared with media containing 0.5% DMSO. After addition of diluted extracts and 70% MeOH extract-loaded liposomes (suspended formulation in 1mL cell culture medium), plates were incubated for 24 h at the same conditions; 10 mm unbuffered Tris-Base (100 μL) was added to solubilize the bound SRB, and the color density was measured at 540 nm with ELISA (Biotek Synergy HT). Results are shown as mean values of six measurements \[[@B33-antioxidants-09-00293]\].
2.9. Statistical Analysis {#sec2dot9-antioxidants-09-00293}
-------------------------
The experimental data are shown as mean ± SD. SPSS software version 12.0 was used for statistical analysis. Analysis of variance was performed using ANOVA procedures. Significant differences were determined by *Tukey's pairwise* and *Games--Howell* comparison test (*p* \< 0.05). The *Levene* test was used to analyze the variance homogeneity of the groups. EC 50 values were calculated using nonlinear regression curves (Sigma Plot 2001 version 7.0, SPSS Inc., Chicago IL, USA).
3. Results and Discussion {#sec3-antioxidants-09-00293}
=========================
In this study, in vitro antioxidant activities of *S. aramiensis* extracts prepared with 70% methanol, 70% ethanol, and 2% infusion were evaluated, and enzyme inhibitor activities of the active 70% methanol extract were investigated using various *in vitro* methods. In the study, 70% methanol extract, which was found active as a result of antioxidant activity tests, was loaded into the nanoliposome formulation, and the release of the extract was examined. Also, phytochemical examination of this active extract was determined using chromatographic techniques (LC-MS/MS, HPLC).
*S. aramiensis* extracts were found to be rich in TPC (198.07 ± 2.77) like other *Salvia* species \[[@B34-antioxidants-09-00293],[@B35-antioxidants-09-00293]\]. Also, TFC content (141.76 ± 1.64) was found to be higher than *S. virgata* \[[@B36-antioxidants-09-00293]\]. Results are given in [Table 1](#antioxidants-09-00293-t001){ref-type="table"}.
The nitrogen-centered, stable radical DPPH^●^ was scavenged by the extracts at physiological pH, and the results were referred as the EC50 (µg/mL) value. According to [Table 2](#antioxidants-09-00293-t002){ref-type="table"}, the EC50 value of 70% MeOH extract was found as 28.4 μg/mL. The activity of rosmarinic acid was statistically significant (*p* \< 0.05) compared to the activities of the *S. aramiensis* extracts. Based on the results given in [Table 2](#antioxidants-09-00293-t002){ref-type="table"}, all extracts and rosmarinic acid succeed in scavenging the ABTS^●+^ radical and no significant difference (*p* \> 0.05) was found between the activity of 70% MeOH extract at the concentration of 100 µg/mL and the activity of rosmarinic acid at 50 µg/mL concentration. According to Tukey test results, the activity of rosmarinic acid was statistically significant (*p* \< 0.05) compared to the 70% EtOH extract and the infusion. *Salvia* species have been proven by studies that they have strong antioxidant activity \[[@B34-antioxidants-09-00293],[@B35-antioxidants-09-00293],[@B36-antioxidants-09-00293],[@B37-antioxidants-09-00293]\]. Therefore, our results have been found to be compatible with the literature.
Oxidation of linoleic acid generates free radicals because the hydrogen atom is removed from the diallylic methylene groups of linoleic acid. These free radicals oxidize unsaturated β-carotene and degradation of the orange-colored β-carotene monitored by spectrophotometer \[[@B38-antioxidants-09-00293]\]. However, the presence of *S. aramiensis* extracts showed an anti-bleaching effect of β- carotene. The AAC values of all extracts are given in [Figure 1](#antioxidants-09-00293-f001){ref-type="fig"}. According to the Levene statistic, no assumption was made for the variable (*p* \< 0.001). According to the results of the Games--Howell test, the comments are as follows: The activities of rosmarinic acid and 70% MeOH extract are statistically the same (*p* \> 0.05) and are statistically significant when compared with the activities of the remaining extracts (*p* \< 0.05). In light of the results of three different antioxidant activity studies, it was decided to continue with 70% MeOH extract in the next part of the study.
The results of the anti-collagenase and anti-elastase activities of the *S. aramiensis* 70% MeOH extract are shown in [Table 3](#antioxidants-09-00293-t003){ref-type="table"}. According to the results presented in [Table 3](#antioxidants-09-00293-t003){ref-type="table"}, there was no statistically significant difference between 100 and 200 µg/mL concentrated *S. aramiensis* extracts and rosmarinic acid which was used as the positive control (*p* \> 0.05). The collagenase enzyme inhibition of the different concentrations of the *S. aramiensis* extract was found between 66.64%--72.66%. The anti-elastase activity of the extract obtained with different concentrations was statistically significant between each other and standard rosmarinic acid (*p* \< 0.05). As can be seen in [Table 3](#antioxidants-09-00293-t003){ref-type="table"}, 200 µg/mL concentrated *S. aramiensis* extract was found to be more active than rosmarinic acid and the inhibition percentage was found at 86.9%. The activities of 50 and 100 µg/mL concentrated *S. aramiensis* extracts and rosmarinic acid was found statistically the same (*p* \> 0.05). Although there are studies on the wound-healing effect of *Salvia* species, the extracts do not have formulations prepared in nano-carrier \[[@B39-antioxidants-09-00293],[@B40-antioxidants-09-00293],[@B41-antioxidants-09-00293]\].
The qualitative-quantitative analyses of the extracts were carried out using LC-MS/MS and HPLC systems and the results are presented in [Table 4](#antioxidants-09-00293-t004){ref-type="table"} and [Table 5](#antioxidants-09-00293-t005){ref-type="table"}, respectively. In the present work, phenolic lignan, phenolic acids, and flavonoids were detected in negative ionization mode ([Figure 2](#antioxidants-09-00293-f002){ref-type="fig"}). At the beginning of the analysis, \[M − H\]^−^ ions at m/z 387 as well as fragment ions at m/z 207 and m/z 163 revealed the compound to be the phenolic lignin Icariside b5/medioresinol. The peak at Rt 8.7 showed an ion \[M − H\]^−^ at m/z 179 and a fragment peak at m/z 135. This compound was identified as caffeic acid. After the loss of a glucose unit (−162 Amu) from \[M − H\]^−^ ion at m/z 463, fragment ion \[quercetin-H\]-- at m/z 301 was obtained, and this was identified as quercetin glucoside. At 10.7, a peak with m/z 447 fragmented to m/z 285 agreed with the fragmentation pattern of luteolin glucoside. With the presence of m/z 315 fragment ions obtained from the loss of 162 amu glucose moiety and the presence of fragment ion m/z 300 obtained by the loss of the methyl group from m/z 315, the peak (m/z 477) at 11.2 min was determined as the isorhamnetin glucoside. The peak at Rt 12.3 min had a molecular ion \[\[M − H\]^−^ at m/z 717 and the base peak at m/z 519 with loss of 198 amu danshensu. The ions at m/z 537 \[M-H-caffeic acid\]−, m/z 475 \[M − H-danshensu-CO~2~\]^−^, and m/z 321 \[M − H-danshensu-danshensu\]− were seen. Based on the loss of danshensu and caffeic acid moieties, this compound was assigned as salvianolic acid E. The presence of salvianolic acid E before rosmarinic acid in the reverse phase column complies with the literature. The peak at 13.3 min was identified as rosmarinic acid. Rosmarinic acid showed a molecular ion \[M − H\]^−^ at m/z 359 and fragment ions m/z 197, 179, and 161. Medioresinol-O-glucuronide was observed at 14.9 min with a molecular ion at m/z 563. Medioresinol-O-glucuronide peak yielded a fragment ion at m/z 387 corresponding to the medioresinol moiety after the loss of glucuronide moiety. The peak at 17.3 min was tentatively determined as luteolin with m/z 285 and fragmented to m/z 151 and 133 which was consistent with the fragmentation pattern of luteolin.
The detection limit (LOD) and the quantitation limit (LOQ) were found as presented in [Table 5](#antioxidants-09-00293-t005){ref-type="table"} for the HPLC method when 10 μL was used as the injection sample volume. Comparing the retention times and UV spectra of the standards, caffeic acid, luteolin-7-O-glycoside, rosmarinic acid, and luteolin were determined ([Figure 3](#antioxidants-09-00293-f003){ref-type="fig"}). The peak increase of rosmarinic acid at 320 nm was linked to the maximum UV absorption at 329 nm of rosmarinic acid, a phenolic acid. The calibration curves of the standards are used to calculate the quantitative data. For *S. aramiensis*, the data obtained by both UV spectrophotometry and HPLC analysis showed that the 70% MeOH extract had a high phenolic content. Rosmarinic acid was the main compound, and its concentration was found as 122.32 ± 0.21 mg/g. The luteolin concentration of the extract was 4.06 ± 0.03 mg/g.
*S. aramiensis* 70% MeOH extract-loaded LPs were prepared using dry film hydration. Nano drug delivery systems are systems that transmit the low-stability active substance to the lower layers of the skin; moreover, it offers the opportunity to work with a lower dose for a longer time. Therefore, liposomes were preferred. In the characterization studies, PS, ZP, and PDI of liposomes were found as 0.395 ± 0.004 µm, −35.2 ± 2.1 mV, and mono dispersed due to low PDI values (below 0.2) ([Table 6](#antioxidants-09-00293-t006){ref-type="table"}). The encapsulating efficiency of the LPs was found as 13.83% ± 1.25%. In the field of pharmaceutical nanosystems, researchers have collected some information about size, stability, and morphology through the microscopic techniques. Of them, SEM was used in our study and the type and size of nanoliposomes that we developed in our study were determined by SEM. The liposomes, which are developed as seen in [Figure 4](#antioxidants-09-00293-f004){ref-type="fig"}, have a narrow particle size range and a smooth surface. In vitro release experiment from LPs was carried out for 24 h, and cumulative released amount of (RA) was found to be 60% ± 0.85% (83.7 µg/mL extract) with HPLC analyses ([Figure 5](#antioxidants-09-00293-f005){ref-type="fig"}).
In literature, some studies with nanocarrier systems such as liposome, nanoparticles, etc. were developed with *Salvia* extracts. In the previous study, anti-neurotoxic efficacy of nanoparticles derived from *Salvia officinalis* extract against methylmercury-induced rats was evaluated. Sixty % ethanol extract of *Salvia officinalis* was associated with poly(lactic-co-glycolic acid) (PLGA) nanoparticles to enhance the antioxidant activity against mercury-induced neurotoxicity \[[@B47-antioxidants-09-00293]\]. In another study, liposome oral liquid containing *Cordyceps sinensis*, *Astragali* L., and *Salvia miltiorrhiza*, which improved liver function, was prepared for the treatment of Hepatitis B and its efficacy was evaluated \[[@B48-antioxidants-09-00293]\]. Aisha et al. \[[@B49-antioxidants-09-00293]\] studied LPs which were prepared to contain *Orthosiphon stamineus* Benth. (Lamiaceae) (OS) ethanolic extract. Liposomes were prepared by the conventional film method with particle size and zeta potential of 152.5 ± 1.1 nm and −49.8 ± 1.0 mV, respectively. Release studies showed 94% ± 0.1% cumulative release in non-formulated extract and 62.4% ± 0.1% after 24 h at 37 °C in OS liposomes \[[@B49-antioxidants-09-00293]\]. For these reasons, our characterization parameters seem to be compatible with the literature and successful when compared with OS-loaded LPs. In that study, the released extract showed improvement in DPPH^●^ scavenging effect. The results showed strong cleavage of DPPH^●^ both on formulated and unformulated extracts. However, a stronger effect with unformulated extract (EC50 = 32.4 ± 0.5 μg/mL) in connection with the high rate of release was found compared to OS LPs (EC50 = 23.5 ± 1.1 μg/mL) \[[@B49-antioxidants-09-00293]\].
The amount of released extract compared to the activity experiments in the present study seems to be sufficient in all the experiments; 83.7 µg/mL of extract released from the LPs is well above the DPPH radical scavenging EC50 value. When we examine the results of the concentrations that we have tested in enzyme inhibition assays, it is clear that the extract released from the LPs will inhibit collagenase and elastase enzymes by more than 66.64% and 76.68%, respectively.
Fibroblast cells are commonly found in tendon, ligament, and skin tissues. They are generally described as collagen-producing cells and are known as the primary source of many extracellular matrix components \[[@B50-antioxidants-09-00293]\]. L929 mouse fibroblast cells are frequently used to determine cytotoxic concentrations of many samples. It is also shown by international standards (ISO 10993 part 5, 1999; ISO 7405, 1997) as a reference cell for cytotoxicity testing \[[@B51-antioxidants-09-00293]\]. The data acquired by the SRB cell viability method are given in [Figure 6](#antioxidants-09-00293-f006){ref-type="fig"}. Significance according to Levene statistics, homogeneity of variance, was found at *p* \< 0.05 and Games--Howell test was used for statistical analysis. Between concentrations of 7.81--250 μg/mL, the 70% MeOH extract did not reduce the viability compared to the control group. The effect of 70% MeOH extract on cell viability at a concentration of 1000 μg/mL was 69%, and the influence of nanoformulation on cell viability was found to be 99.66%. Extract amounts obtained as a result of both encapsulation efficiency and release study are between 82.98 and 138.3 μg/mL. In the toxicity test, both concentrations were not toxic. The lack of toxicity of the LPs on the cell line is the first proof that it is reliable; also, it will shed light on future studies. This study is important in terms of proving the effectiveness of the nanoformulation containing extract rather than classical activity studies and proves the necessity of handling traditional plants with a modern understanding.
4. Conclusions {#sec4-antioxidants-09-00293}
==============
All data obtained from the study revealed the strong antioxidant effect of *S. aramiensis* as well as its potential wound-healing effect. The formulation study showed innovations in many respects as well as the potential of the extracts from plants to turn into products. Although the antioxidant activity of plant extracts is frequently studied, it is very important to show antioxidant capacities of formulated extracts. Therefore, in this study, *S. aramiensis* extract was first formulated with liposome, which is a nano-carrier system that attracts attention and is at the forefront with its advantages. With the successful demonstration of *in vivo* activity in future studies, a pharmaceutical product obtained from *S. aramiensis* will be safely functionalized.
Conceptualization: E.K.A. and G.Ş.K.; methodology: G.Ş.K., Ç.Y., and E.K.A.; validation: G.Ş.K.; formal analysis: G.Ş.K., Ç.Y., and F.G.; investigation: G.Ş.K. and Ç.Y.; data curation: G.Ş.K. and E.K.A.; writing---original draft preparation: G.Ş.K., E.K.A., and E.S.-S.; writing---review and editing: E.K.A. All authors have read and agreed to the published version of the manuscript.
This research was supported with Project no: TSA-2019-8514 from Erciyes University Research Council.
The authors declare that they do not have any conflict of interest.
{#antioxidants-09-00293-f001}
{#antioxidants-09-00293-f002}
{#antioxidants-09-00293-f003}
{#antioxidants-09-00293-f004}
{#antioxidants-09-00293-f005}
{#antioxidants-09-00293-f006}
antioxidants-09-00293-t001_Table 1
######
Total phenol and flavonoid content of *S. aramiensis* extracts.
-----------------------------------------------------------
Extracts Total Phenol\ Total Flavonoids\
(mgGAE/gextract) ^a^ (mgCA/gextract) ^b^
-------------- ---------------------- ---------------------
S.a 70% MeOH 198.07 ± 2.77 141.76 ± 1.64
S.a 70% EtOH 153.24 ± 0.94 93.28 ± 9.53
S.a inf 124.75 ± 1.58 118.85 ± 4.47
-----------------------------------------------------------
^a^ mgGAE/gextract: Total phenols expressed as gallic acid equivs milligrams of gallic acid per gram (dry weight) of extract. ^b^ mgCA/gextract: Total flavonoids expressed as rutin per gram (dry weight) of extract. Values are expressed as means ± standard error (*n* = 3). S.a 70% MeOH; *S. aramiensis* 70% methanol extract, S.a 70% EtOH; *S. aramiensis* 70% ethanol extract, S.a inf; *S. aramiensis* 2% infusion.
antioxidants-09-00293-t002_Table 2
######
DPPH^●^ and ABTS^●+^ radical scavenging activities of *S. aramiensis* extracts.
Extracts/Standard DPPH● EC50 (µg/mL) ABTS●+ (mmol/ L /Trolox)
------------------- -------------------- -------------------------- --------------------
S.a 70% MeOH 28.4 ± 0.002 ^b^ 1.15 ± 0.101 ^a^ 1.77 ± 0.09 ^b^
S.a 70% EtOH 36.6 ± 0.004 ^c^ 1.59 ± 0.091 ^c^ 1.65 ± 0.05 ^b,c^
S.a inf 32.6 ± 0.001 ^b^ 1.24 ± 0.113 ^a^ 1.37 ± 0.071 ^a,d^
Rosmarinic acid 4.4 ± 0.00 ^a^ 2.16 ± 0.044 ^b,e^ 2.56 ± 0.001 ^e^
Values = mean ± SD, statistical analyses by Tukey comparison test; values with the same lowercase letters (a--d) are not significantly (*p* \> 0.05) different, *n* = 3. S.a 70% MeOH; *S. aramiensis* 70% methanol extract, S.a 70% EtOH; *S. aramiensis* 70% ethanol extract, S.a inf; *S. aramiensis* 2% infusion.
antioxidants-09-00293-t003_Table 3
######
Collagenase and elastase enzyme inhibition activities of *S. aramiensis* 70% MeOH extract and rosmarinic acid.
S.a 70% MeOH/Standard Collagenase Enzyme Inhibition (%) Elastase Enzyme Inhibition (%)
----------------------- ----------------------------------- --------------------------------
50 µg/mL 66.64 ± 1.53 ^b^ 76.68 ± 1.21 ^a^
100 µg/mL 69.54 ± 1.49 ^a,b^ 83.70 ± 1.72 ^b^
200 µg/mL 72.66 ± 0.70 ^a^ 86.91 ± 0.81 ^c^
RA 25 µg/mL 72.89 ± 0.99 ^a^ 80.58 ± 0.89 ^a,b^
Values = mean ± SD, statistical analyses by Tukey comparison test; values with the same lowercase letters (a--c) are not significantly (*p* \> 0.05) different, *n* = 3. S.a 70% MeOH; *S. aramiensis* 70% methanol extract, RA: rosmarinic acid.
antioxidants-09-00293-t004_Table 4
######
LC-MS/MS results of 70% MeOH extract of *S. aramiensis*.
Rt \[M − H\]^−^ Fragments Identified Compounds Ref
------ -------------- -------------------- ----------------------------------------------- ----------------------------------
7.2 387 207, 163 Icariside b5/medioresinol \[[@B42-antioxidants-09-00293]\]
8.7 179 135 Caffeic acid \[[@B43-antioxidants-09-00293]\]
8.9 463 300, 271 Quercetin glucoside/hydroxyluteolin glucoside \[[@B43-antioxidants-09-00293]\]
10.7 447 285 Luteolin glucoside \[[@B43-antioxidants-09-00293]\]
11.2 477 315, 300, 285 Isorhamnetin hexoside \[[@B44-antioxidants-09-00293]\]
12.3 717 537, 519, 493, 321 Salvianolic acid E \[[@B45-antioxidants-09-00293]\]
13.3 359 197, 179, 161, Rosmarinic acid \[[@B43-antioxidants-09-00293]\]
14.9 563 387, 207 Icariside b5/Medioresinol O-glucoronide \[[@B46-antioxidants-09-00293]\]
17.3 285 151, 133 Luteolin \[[@B43-antioxidants-09-00293]\]
Rt: Retention time; \[M − H\]^−^ molecular ion.
antioxidants-09-00293-t005_Table 5
######
HPLC results of *S. aramiensis* extract.
Compounds Extract \* Standards
-------------------------- --------------- ----------------------------------- -------- --------
Caffeic acid 1.84 ± 0.09 y = 44.283x + 40.553 r^2^ = 0.998 0.0067 0.0204
Luteolin-7-*o*-glucoside 3.70 ± 0.22 y = 39.422x + 78.434 r^2^ = 0.997 0.0085 0.0257
Rosmarinic acid 122.32 ± 0.21 y = 23.777x − 11.797 r^2^ = 0.997 0.2595 0.7865
Luteolin 4.06 ± 0.03 y = 41.981x − 70.752 r^2^ = 0.997 0.1407 0.4263
\* Sa 70% MeOH: 70% MeOH extract. ^&^ mg/g~extract~, mean ± SD.
antioxidants-09-00293-t006_Table 6
######
Characterization of liposomes (*n* = 3).
Formulation PS ± SD (µm) ZP ± SD (mV) PDI ± SD EE ± SD (%) RA ± SD (%)
------------- --------------- -------------- --------------- -------------- -------------
Liposome 0.395 ± 0.004 −35.2 ± 2.1 0.107 ± 0.017 13.83 ± 1.25 60 ± 0.85
PS: particle size; ZP: zeta potential; PDI: polydispersity index; EE: encapsulation efficiency; RA: released amount of extract. Values are expressed as mean ± SD (*n* = 3).
| {
"pile_set_name": "PubMed Central"
} |
###### Strengths and limitations of this study
- Health workers from health facilities of all three levels in China were sampled to draw a whole picture of prevalence, correlates and reasons for workplace violence (WPV) against health workers in this study.
- This study achieved a larger sample size (4862) and a higher valid response rate (79.8%) than past studies related to WPV against health workers.
- The convenience sampling strategy of this study may compromise the representativeness of the sample.
- A single Yi Nao incident usually affects a larger number of health workers in the facility, and thus, there might be potential impacts of Yi Nao at organisational level that could not be identified in the current study.
Introduction {#s1}
============
Workplace violence (WPV) against health workers has become a major global health problem.[@R1] It finds its expression in verbal abuse, bullying, threats, sexual harassment, physical assault and homicide.[@R2] WPV has threatened health workers' well-being, safety, health and even their lives, and has also resulted in a tendency to defensive medicine, poorer health service quality and productivity, resignation of health personnel and difficulties in recruitment.[@R2] This study focused on client-initiated WPV (ie, assaults by patients, caregivers or relatives of the patient) because it is the most common type of WPV in the health sector.[@R4]
China is known to have a high prevalence of WPV in the health sector.[@R5] A survey organised by the Chinese Hospital Association which drew data from 316 hospitals showed that the proportion of hospitals experiencing WPV increased from 90% in 2008 to 96% in 2012.[@R8] A study conducted by the Chinese Medical Doctor Association in 2014 revealed that nearly three quarters of physicians ever experienced physical injuries or verbal abuse at work.[@R9] China issued an Amendment of Criminal Law which included regulations against WPV in the health sector in 2015.[@R10] This amendment might explain the results of a widely circulated study conducted from 2013 to 2017 which showed that WPV in the health sector reached a peak in 2015 and then began to decrease.[@R11] Nevertheless, a 2016 survey with 1024 Chinese nurses showed 75.4% of physicians had experienced at least one form of WPV,[@R12] and a 2017 study indicated that 66% of participants had experienced WPV in the past year.[@R13] The WPV against health workers remains widespread even after the law legislations, and medical professions are considered by many to be unsafe working environments.[@R7]
The prominent problem of WPV in China is believed to be related to systemic and historical factors.[@R14] Insufficient public funding into the public health sector, commercialisation of public health facilities, and high out-of-pocket payment contributed to patient dissatisfaction with public health services over time, leading to a distrust of healthcare professionals and poor doctor-patient relationships.[@R16] Legal channels for reporting malpractice or resolving medical disputes are perceived to be time-consuming, unaffordable, ineffective and unfair.[@R8] Biassed media reports may also exacerbate the tension.[@R15] The Chinese government started a massive healthcare reform in 2009, rolled out universal health insurance coverage and zero-profit drug policy aiming to improve accessibility and affordability of health services.[@R19] However, with health resources remaining insufficient and the structural imbalance among the three tiers of healthcare facilities,[@R22] namely tertiary hospitals, secondary hospitals and primary care facilities, patients in China still flood into major hospitals and deal with long waiting times[@R15] and inadequate communication from physicians.[@R24] The accumulation of these patients' poor experiences can instigate more tension between health providers and patients.[@R25] Previous researches conducted in tertiary hospitals and/or together with secondary hospitals found that there is a high prevalence of WPV against health workers.[@R6] Seldom have these studies compared rates of WPV within the three-tier health system structure in China, and WPV in a rapidly changing health system merits more research.
Former studies have identified different kinds of WPV. One large sample survey of WPV in China from 2016 recognised seven types of WPV,[@R27] which include some special types such as 'smear reputation' and 'mobbing behaviour'.[@R28] Most Chinese studies only described WPV in the most general terms or divided into physical violence, verbal violence and sexual harassment.[@R13] In all these types of WPV, the prevalence of physical violence and its correlates have been frequently studied. However, little is known about specific forms of physical violence in China. On the other hand, threats of physical violence are a form of psychological violence.[@R2] Threats are associated with a higher risk of actual physical violence.[@R30] The self-rated degree of seriousness of the threat incident outcomes was similar to that of actual physical violence, and threats can cause consequences comparable to actual physical violence.[@R31] Specific data about threats and their associated factors in healthcare settings are needed. In addition, Yi Nao is a unique form of WPV in China, and has been defined as an extreme and organised form of WPV in which patients, relatives and even hired gangs attack health workers, damage hospital facilities and disturb the running of hospitals in order to gain monetary compensation from the hospital.[@R18] Many health workers have had individual exposure to Yi Nao by either witnessing or falling victim to it, but little is known about its correlates in the Chinese setting.
Furthermore, how employees perceive organisational policies and practices to prevent aggression is important because it affects victims' willingness to report violent cases to relevant parties.[@R31] It would be useful to conduct more research on the relationship between physicians' perception of organisational policies to report WPV and the rates at which they experienced incidents of WPV. In this study, we aimed to (1) Examine the magnitude of physical violence, threats, Yi Nao and their correlates, at the three levels of healthcare facilities in China. (2) Examine specific forms of physical violence and the aftermath. (3) Perception of organisational policies to report WPV and its association with physical violence, threats and Yi Nao.
Methods {#s2}
=======
Study population and design {#s2-1}
---------------------------
The study was a cross-sectional survey conducted from July 2016 to July 2017 in Zhejiang Province, which has a population of 56 million people including half a million health sector workers.[@R32] The target population was all health workers who had patient contact. The sampling strategy was divided into two steps:
First, we stratified the sampling by facility type which has been built according to political administration levels. Specifically, primary care facilities provide primary healthcare to the population of a district or a town, while tertiary and secondary hospitals are usually located in cities and counties, aiming to provide referral care to the population residing in their administrative areas. We purposefully sampled tertiary and secondary hospitals across different cities and counties. We included general hospitals and specialty hospitals. The latter included women and children hospitals, and traditional Chinese medicine hospitals.
The selected health facilities included tertiary hospitals (mostly in the developed capital city of Hangzhou and prefecture-level cities), secondary hospitals (mostly in counties) and primary care facilities including urban community health centres and rural township health centres. Specifically, the selected health facilities at each tier included: (1) Five tertiary hospitals: one general hospital was selected in Hangzhou, a developed prefecture-level city Jiaxing, and a less developed city Quzhou, respectively; one women and children hospital and one traditional Chinese medicine hospital in Hangzhou were selected. (2) Eight secondary hospitals: one general hospital and one traditional Chinese medicine hospital were sampled from the developed counties Ninghai and Shenzhou, and the less developed counties Jiangshan and Kaihua, respectively. (3) Thirty-two primary care facilities including 16 urban community health centres in Hangzhou and Ningbo (8 in each city) and 16 rural township health centres in Ninghai and Jiangshan (8 in each county). A total of 46 health facilities were selected for the study.
Second, we used convenience sampling to recruit participants. We first obtained permission from hospital managers or primary care facility directors to conduct the survey. Afterwards, an administrator was designated to help us contact each unit or contact health workers. We invited health workers on duty on surveying days to fill in the questionnaire, including doctors, nurses, laboratory technicians (who need to dispense sample containers and collect biological samples from patients) and administrative staff who had patient contact in their daily practices. The purpose of the survey was fully explained, and verbal consent was obtained. The front page of the questionnaire also included a statement of consent. Anonymity and confidentiality were assured. We informed participants that the survey was fully voluntary.
Questionnaire {#s2-2}
-------------
The questionnaire was designed based on WHO's instrumental survey tools which have previously been used in China.[@R2] The original questionnaire was developed by the joint programme on WPV in the health sector among the International Labour Office, International Council of Nurses, WHO and Public Services International in 2003. It consists of five sections: personal and workplace data, physical WPV, psychological WPV, health sector employer, and opinions on WPV. In this study, we adapted our questionnaire based on the former two and the last sections with a focus on physical violence. To fit the study aims and reflect the Chinese setting, additional items were added, including two items about experience of threats and Yi Nao, four items about specific forms of physical violence and how victims respond, and six items about the aftermath.
The final questionnaire comprised five sections: (1) Sociodemographic information and professional background. (2) Experience of physical violence, threats of physical violence (defined in the questionnaire as threats of physical violence against targeted individuals, with or without weapons, resulting in fear of physical harm or other negative impacts on the targeted individuals), and some form of exposure to, that is witnessing or experiencing, Yi Nao in the last year before the survey. (3) Specific forms of physical violence, victims responses and consequences to the perpetrators. (4) Aftermath of physical violence and threats. (5) Perception of organisational policies to report WPV.
Data analysis {#s2-3}
-------------
Descriptive analyses were conducted. Pearson's chi-square test and Fisher's exact test were used to test the differences in WPV between subgroups. Multivariable logistic regression examined the association between perceived organisational encouragement of reporting WPV and physical violence, threats and Yi Nao after controlling for age, sex, level of facility, professional ranking and type of health worker. A P value of \<0.05 was considered statistically significant. Statistical analysis was conducted using SPSS V.25.
Patient and public involvement {#s2-4}
------------------------------
No patients were involved in the design, recruitment and conduct of the study. The study participants and their facilities were offered a summary of the study's results.
Results {#s3}
=======
Sample characteristics and perception of organisational reporting policy (n=4862) {#s3-1}
---------------------------------------------------------------------------------
We distributed 6089 questionnaires, of which 5145 were returned. Among them, 283 were considered invalid due to missing values of key variables, resulting in a valid response rate of 79.8% (4862/6089). Sample characteristics are shown in [table 1](#T1){ref-type="table"}. We sampled similar numbers of respondents across the three levels of facilities: tertiary (1873, 38.5%), secondary (1509, 31.0%) and primary (1480, 30.4%). The mean age of the respondents was 33.7 years (SD=8.6), and 3551 respondents (73.0%) were female. Out of all respondents, 1954 (40.2%) were doctors, 1911 (39.3%) were nurses, 692 (14.3%) were laboratory staff and 299 (6.2%) were administrators. A total of 2205 (45.5%) were junior staff and 1571 (32.4) were intermediate. In the past year, 224 (4.6%) of the respondents were physically attacked, 848 (17.4%) experienced threats and 1956 (40.4%) had some exposure (witnessing or experiencing) to a Yi Nao incident in their facility. Over half of the respondents (56.6%) reported organisational encouragement to report WPV, but the rest either did not perceive such encouragement (487, 10%) or were unaware of such an organisational atmosphere (1609, 33.2%).
######
Demographics and professional characteristics of health workers who reported physical violence, threats or Yi Nao in the past year in Zhejiang, China
--------------------------------------------------------------------------------------------------------------------------------------------------------------
Total\*\ Physical violence (%) (n=224) Threats of physical violence (%) (n=848) Yi Nao (%) (n=1956)
(%) (n=4862)
---------------------------------------------- -------------- ------------------------------- ------------------------------------------ ---------------------
Age
Mean (SD) 33.7 (8.6) 34.9 (8.5) 34.5 (7.9) 33.9 (8.1)
Sex
Male 1295 (26.7) 99 (44.6) 277 (32.7) 625 (32.1)
Female 3551 (73.3) 123 (55.4) 570 (67.3) 1325 (67.9)
Level of facility
Tertiary hospitals 1873 (38.5) 67 (29.9) 315 (37.1) 765 (39.1)
Secondary hospitals 1509 (31.0) 120 (53.6) 323 (38.1) 765 (39.1)
Primary care facilities 1480 (30.4) 37 (16.5) 210 (24.8) 426 (21.8)
Type of health workers
Doctors 1954 (40.2) 118 (52.7) 398 (47.0) 898 (46.0)
Nurses 1911 (39.4) 78 (34.8) 328 (38.7) 755 (38.7)
Laboratory 692 (14.3) 20 (8.9) 95 (11.2) 221 (11.3)
Administrators 299 (6.2) 8 (3.6) 26 (3.1) 78 (4.0)
Professional ranking
Junior 2205 (45.5) 94 (42) 344 (40.6) 821 (42.1)
Intermediate 1571 (32.4) 84 (37.5) 320 (37.8) 678 (34.7)
Senior 579 (11.9) 33 (14.7) 140 (16.5) 303 (15.5)
None 496 (10.2) 13 (5.8) 43 (5.1) 150 (7.7)
Encourage reporting violence in the facility
No 487 (10.0) 62 (27.7) 141 (16.7) 282 (14.4)
Yes 2752 (56.8) 101 (45.1) 423 (50.1) 1058 (54.1)
Don\'t know 1609 (33.2) 61 (27.2) 281 (33.3) 615 (31.5)
--------------------------------------------------------------------------------------------------------------------------------------------------------------
%: valid percentages.
\*Column frequencies do not add up to the total due to missing values.
Characteristics of physical violence (n=224) {#s3-2}
--------------------------------------------
The characteristics of physical violence are indicated in [table 2](#T2){ref-type="table"}. Among 224 health workers who experienced physical violence in the past year, 107 (47.8%) were pushed, 32 (14.3%) were beaten up, 28 (12.5%) were attacked by perpetrators using implements, 23 (10.3%) were kicked and 1 (0.4%) was stabbed. Nineteen respondents (8.5%) reported slapping, 20 (8.9%) punching, 14 (6.3%) strangling, 12 (5.4%) pulling hair, 11 (4.9%) scratching and 3 (1.3%) stamping. A hundred and sixty (71.7%) of the perpetrators were the patient's relatives or friends, and 12 (5.4%) involved Yi Nao events. Among all, 64 (28.6%) reported physical injuries due to the violence. Among 12 cases involving Yi Nao, 9 (75.0%) caused physical injuries to victims, compared with 55 out of 205 (26.8%) without Yi Nao involvement.
######
Characteristics of physical violence incidents against health workers in Zhejiang, China (n=224)
Characteristics N (%) Characteristics N (%)
--------------------------------------------- ------------ ----------------------------------------------- ------------
How did the perpetrator(s) attack you? How did you respond after it happened?
Pushing 107 (47.8) Reported to police 64 (29.1)
Beating up 32 (14.3) Reported to senior doctors/hospital 122 (55.5)
Kicking 23 (10.3) Shared with my friends/family 47 (21.4)
Stamping 3 (1.3) Shared with my colleague 90 (40.9)
Slapping 19 (8.5) Took no action 34 (15.5)
Pulling hair 12 (5.4) Reported to medical association 1 (0.4)
Strangling 14 (6.3) Why did not you report to police? (n=153)
Punching 20 (8.9) Felt useless 90 (58.8)
Biting 7 (3.1) Not important 14 (9.2)
Scratching 11 (4.9) Felt ashamed 7 (4.6)
Stabbing 1 (0.4) Cumbersome 28 (18.3)
Other damaging behaviours using implements 28 (12.5) Fear of negative consequences or retaliation 26 (17.0)
Others 62 (27.7) Didn't know who to report to 32 (20.9)
Who attacked you? Victim was physically injured 64 (28.6)
Patient 102 (45.7) The consequences of the perpetrator(s)
Relatives or friends of patient 160 (71.7) None 157 (72.4)
Yi Nao 12 (5.4) Verbal warning and apology 27 (12.4)
Others 3 (1.3) Fined 2 (0.9)
How did you respond when it happened? Short-term police detention 19 (8.8)
Swearing 7 (3.2) Prosecuted with unknown results 5 (2.3)
Fight back 19 (8.6) Sentenced 3 (1.4)
Run away 52 (23.5) Other 13 (6.0)
Shout for help 32 (14.5) Case investigated
Protected myself, not fight back 63 (28.5) Yes 56 (25.7)
Asked attacker to stop 45 (20.4) No 119 (54.6)
Try to calm perpetrator(s) down 109 (49.3) Don't know 43 (19.7)
Took no action 17 (7.7) Spread on social media\* 10 (4.5)
Other 13 (5.9) Reported in wider news media† 6 (2.7)
Satisfied with how the incident was handled 11 (4.9)
%: valid percentages.
\*Common social media platforms included Weibo (microblog similar to Twitter), QQ and WeChat (instant messaging and social networking mobile phone applications which are similar to Facebook).
†This included news websites or traditional news media like TV and newspapers.
The immediate responses of the victims included 7 (3.2%) swearing, 19 (8.6%) fighting back and 109 (49.3%) trying to calm down the perpetrators. After the incident, 64 (29.1%) reported their experiences to the police; 122 (55.5%) reported to senior doctors/the hospital, only 1 (0.4%) reported to medical association and 34 (15.5%) took no action at all. Among the 153 respondents who explained their reasons for not reporting their experiences to the police, 90 (58.8%) felt it useless to do so, 32 (20.9%) did not know who to report to and 28 (18.3%) thought it was too much trouble. Totally, 56 (25.7%) of the physical violence cases were investigated and most (37/56, 66.1%) were conducted by hospitals. A total of 157 (72.4%) of the attackers received no punishment, and these included the stabbing case causing physical injury, 9/14 (64.3%) who strangled, 12/19 (63.2%) who punched, 14/23 (60.9%) who kicked and 21/27 (77.8%) who had used implements. Only 60 (27.6%) received some form of punishment, including 27 (12.4%) being verbally warned, 19 (8.8%) detained for a short time in prison and 3 (1.4%) sentenced. Only 7/12 (58.3%) physical violence cases involving Yi Nao were reported to the police and 10/12 (83.3%) of these perpetrators were not punished. Only 11/224 (4.9%) were satisfied with how the incident was handled.
Impact of physical violence or threats (n=1072) {#s3-3}
-----------------------------------------------
[Table 3](#T3){ref-type="table"} demonstrates that among 1072 health workers who experienced physical violence or threats, 835 (79.9%) had repeated, disturbing memories, thoughts or images of the attack; 653 (62.8%) avoided thinking about or talking about the attack; 1000 (95.5%) were 'super-alert' or more vigilant; 873 (83.5%) considered how to protect themselves while seeing patients; 826 (79.0%) trusted patients less; 791 (76.0%) were fearful of dealing with urgent or severe cases; and 618 (59.4%) wanted to quit their current post. Nearly a fifth had to take some time off work due to the incident. There were no significant differences in these impacts between those who experienced physical violence and those who encountered threats.
######
Impact of physical violence or threats on victims in the health sector in Zhejiang, China
-----------------------------------------------------------------------------------------------------------------------
Total (%)\ Physical violence (%)\ Threats (%)\
(n=1072) (n=224) (n=848)
----------------------------------------------------------------- ------------- ------------------------ --------------
Repeated, disturbing memories, thoughts or images of the attack 835 (79.9) 169 (77.9) 666 (80.4)
Made you trust less in patients as a whole 826 (79.0) 161 (75.2) 665 (79.9)
Avoiding thinking about or talking about the attack 653 (62.8) 125 (58.7) 528 (63.9)
Being 'super-alert' or watchful and on guard 1000 (95.5) 201 (93.1) 799 (96.1)
Fearful of dealing with urgent or severe cases 791 (76.0) 163 (76.5) 628 (75.8)
Often consider how to protect yourself while seeing patients 873 (83.5) 189 (87.5) 684 (82.5)
Wanting to quit the current post 618 (59.4) 131 (61.5) 487 (58.9)
Took some time off from work after being attacked 185 (18.0) 46 (21.0) 139 (17.1)
-----------------------------------------------------------------------------------------------------------------------
%: valid percentages.
Correlates of physical violence, threats and exposure to Yi Nao {#s3-4}
---------------------------------------------------------------
The correlates of physical violence, threats and exposure to Yi Nao are presented in [table 4](#T4){ref-type="table"}. Male health workers were more likely than female to report that they had experienced physical violence (adjusted OR, or AOR=2.18, 95% CI 1.53 to 3.12), threats (AOR=1.33, 95% CI 1.10 to 1.62) and exposure to Yi Nao (AOR=1.42, 95% CI 1.22 to 1.66). Health workers in secondary hospitals had a higher risk of experiencing physical violence (AOR=3.29, 95% CI 2.21 to 4.89), threats (AOR=1.61, 95% CI 1.32 to 1.98) and Yi Nao (AOR=2.47, 95% CI 2.10 to 2.91) compared with those in primary care facilities. Health workers in tertiary hospitals were more likely than primary care health workers to report threats (AOR=1.28, 95% CI 1.04 to 1.57) and Yi Nao exposure (AOR=1.68, 95% CI 1.44 to 1.97). Health workers who did not perceive organisational policies to report WPV were more likely to have experienced physical violence (AOR=3.64, 95% CI 2.57 to 5.18), threats (AOR=2.21, 95% CI 1.76 to 2.78) and Yi Nao (AOR=2.26, 95% CI 1.84 to 2.78).
######
Correlates of physical violence, threats and exposure to Yi Nao among health workers in China
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Physical violence Threats of physical violence Exposure to Yi Nao
---------------------------------------------- ---------------------- ------------------------------ ---------------------- --------- ---------------------- --------- ---------------------- --------- ---------------------- --------- ---------------------- ---------
Age 1.02\ 0.034 1.00\ 0.689 1.01\ 0.002 0.99\ 0.074 1.01\ 0.129 0.98\ \<0.001
(1.00 to 1.03)\* (0.98 to 1.03) (1.00 to 1.02)\*\* (0.97 to 1.00) (1.00 to 1.01) (0.97 to 0.99)\*\*\*
Sex
Female Ref
Male 2.31\ \<0.001 2.18\ \<0.001 1.42\ \<0.001 1.33\ 0.004 1.57\ \<0.001 1.42\ \<0.001
(1.76 to 3.03)\*\*\* (1.53 to 3.12)\*\*\* (1.21 to 1.67)\*\*\* (1.10 to 1.62)\*\* (1.38 to 1.78)\*\*\* (1.22 to 1.66)\*\*\*
Level of facility
Primary care Ref
Secondary hospitals 3.37\ \<0.001 3.29\ \<0.001 1.65\ \<0.001 1.61\ \<0.001 2.57\ \<0.001 2.47\ \<0.001
(2.31 to 4.91)\*\*\* (2.21 to 4.89)\*\*\* (1.36 to 1.99)\*\*\* (1.32 to 1.98)\*\*\* (2.21 to 2.99)\*\*\* (2.10 to 2.91)\*\*\*
Tertiary hospitals 1.45\ 1.447 1.47\ 0.082 1.22\ 0.038 1.28\ 0.018 1.71\ \<0.001 1.68\ \<0.001
(0.96 to 2.17) (0.95 to 2.28) (1.01 to 1.48)\* (1.04 to 1.57)\* (1.48 to 1.98)\*\*\* (1.44 to 1.97)\*\*\*
Types of health workers
Nurses Ref
Doctors 1.51\ 0.006 1.02\ 0.908 1.23\ 0.011 1.07\ 0.536 1.30\ \<0.001 1.17\ 0.053
(1.13 to 2.03)\*\* (0.69 to 1.51) (1.05 to 1.45)\* (0.87 to 1.30) (1.15 to 1.48)\*\*\* (1.00 to 1.37)
Laboratory 0.70\ 0.160 0.57\ 0.044 0.77\ 0.036 0.78\ 0.064 0.72\ 0.001 0.74\ 0.004
(0.42 to 1.15) (0.33 to 0.98)\* (0.60 to 0.98)\* (0.59 to 1.02) (0.60 to 0.87)\*\* (0.60 to 0.91)\*\*
Administrators 0.65\ 0.246 0.69\ 0.362 0.46\ \<0.001 0.62\ 0.035 0.54\ \<0.001 0.69\ 0.022
(0.31 to 1.35) (0.30 to 1.58) (0.30 to 0.70)\*\*\* (0.40 to 0.96)\* (0.41 to 0.71)\*\*\* (0.51 to 0.94)\*
Professional ranking
Junior Ref
Intermediate 1.27\ 0.122 1.14\ 0.482 1.38\ \<0.001 1.46\ \<0.001 1.28\ \<0.001 1.48\ \<0.001
(0.94 to 1.72) (0.79 to 1.65) (1.17 to 1.64)\*\*\* (1.19 to 1.78)\*\*\* (1.12 to 1.46)\*\*\* (1.25 to 1.74)\*\*\*
Senior 1.36\ 0.142 0.92\ 0.780 1.73\ \<0.001 1.86\ \<0.001 1.84\ \<0.001 2.04\ \<0.001
(0.90 to 2.04) (0.51 to 1.64) (1.38 to 2.15)\*\*\* (1.35 to 2.56)\*\*\* (1.53 to 2.22)\*\*\* (1.57 to 2.65)\*\*\*
None 0.60\ 0.119 0.75\ 0.351 0.51\ \<0.001 0.53\ \<0.001 0.73\ 0.022 0.75\ 0.045
(0.34 to 1.09) (0.41 to 1.40) (0.37 to 0.72)\*\*\* (0.37 to 0.75)\*\*\* (0.59 to 0.91)\*\* (0.60 to 0.94)\*
Encourage reporting violence in the facility
Yes Ref
No 3.83\ \<0.001 3.64\ \<0.001 2.24\ \<0.001 2.21\ \<0.001 2.22\ \<0.001 2.26\ \<0.001
(2.75 to 5.34)\*\*\* (2.57 to 5.18)\*\*\* (1.80 to 2.80)\*\*\* (1.76 to 2.78)\*\*\* (1.82 to 2.70)\*\*\* (1.84 to 2.78)\*\*\*
Don\'t know 1.03\ 0.838 1.07\ 0.709 1.17\ 0.070 1.23\ 0.020 0.99\ 0.847 1.02\ 0.825
(0.75 to 1.43) (0.76 to 1.49) (0.99 to 1.37) (1.03 to 1.46)\* (0.87 to 1.12) (0.89 to 1.16)
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
\*\<0.05; \*\*\<0.01; \*\*\*\<0.001.
\*Adjusted for age, sex, level of facility, type of health workers, professional ranking and encouragement perceived to report violence in the facility.
AOR, adjusted OR.
Discussion {#s4}
==========
High prevalence of WPV in the health sector represents a major public health challenge in China. We examined physical violence, threats and Yi Nao against health workers at three tiers of health facilities in Zhejiang province. The study extends the literature by demonstrating that health workers at secondary and tertiary hospitals were at a higher risk of physical violence, threats and exposure to Yi Nao, and widespread impunity of perpetrators. In addition, lack of organisational encouragement to report WPV was associated with a higher risk of physical violence, threats and Yi Nao.
The prevalence of physical violence in this study was 4.6%, which was lower than that of previous studies in China[@R6] but still higher than other countries.[@R37] Our study also found high rates of physical violence threats and exposure to Yi Nao. There were 17.4% of participants who encountered threats while 40.4% of them witnessed or experienced an incident of Yi Nao in the past year. Besides, the overwhelming majority of the victims of physical violence and threats became vigilant in the clinical environment and worried about personal safety while seeing patients. We found that health workers in secondary hospitals were more likely to have directly encountered physical violence with the prevalence of 8.0% than those who worked in primary care facilities with a prevalence of 2.5%. And compared with health workers in primary care facilities, those in secondary and tertiary hospitals were at a higher risk of receiving threats and being exposed to Yi Nao. There are two potential explanations for such a trend. First, there was a greater chance of dealing with patients with more severe illnesses, and mismatching patient expectations and consequent patient frustration put higher-level facilities at higher risk than primary care facilities. A study conducted in Thailand reported that the severity of a patient's illness is an important contributing factor to physical violence initiated by the patient.[@R38] Unmet expectations were identified as a major reason for assaults.[@R39] Resolving the mismatch between patient expectations and care delivered is an important strategy to prevent WPV.[@R40] Unreasonably high patient expectations, however, were also reported by a previous Chinese study,[@R41] indicating that, in addition to providing adequate explanations to patients, proper management of unrealistic expectations may be necessary to reduce WPV.
Second, health workers with better education and advanced medical equipment are concentrated in top tier hospitals.[@R42] Primary care facilities are relatively underused due to less competitivity and patient trust compared with higher level facilities.[@R43] Many patients sought care even for simple common conditions at outpatient services of tertiary and secondary hospitals. This shortens the consultation time for each patient, which may compromise the quality of care and patients' satisfaction.[@R44] Improving primary care quality and increasing its utilisation may be an essential, long-term strategy to tackle the underlying factors of WPV from the health system perspective.
Our study also found that incidents of physical violence are severely under-reported to relevant authorities or organisations, despite the fact that reporting experiences of violence is an essential part of a successful WPV prevention programme.[@R46] Less than 30% of victims reported their experiences to police and only 25.7% of the cases were investigated in some way. Impunity to perpetrators of physical violence was also widespread. In physical violence cases, more than 70% of the perpetrators were left unpunished, including those who committed severe forms of physical violence such as stabbing, strangling, kicking and damaging behaviours using implements. As for those cases involving Yi Nao, over 80% of the perpetrators were not punished. Overall, only 4.9% of victims suffering physical violence were satisfied with how their cases were handled.
Major reasons for under-reporting and impunity may be related to poor law enforcement against physical violence in the health sector and inadequate organisational support or advocacy to reporting WPV in their health facilities. Among the victims who did not report the incident to the police, more than a fifth did not know who to report to and nearly 60% felt useless to do so. Health workers are generally believed to be more powerful due to the information asymmetry,[@R8] and patients are deemed as more vulnerable. Patient violence against health workers is often regarded by the public as an understandable manifestation of grievance or self-protection.[@R47] To meet socially desirable expectations, police authorities and hospital leaders often adopt tolerant attitudes towards patient perpetrators. Law enforcement professionals also tend to regard WPV against health workers as a simple doctor-patient relationship problem rather than a problem of criminal behaviour. A survey with policemen showed that when asked by hospitals for assistance to resolve patient-initiated WPV, 55% refused to go to the scene or reacted too slowly. Policemen who arrived at hospitals in response to reported WPV were often reluctant to intervene in the dispute and sometimes just stood to observe.[@R48] Joint efforts by multiple sectors are needed to create an environment where health workers feel safe to work.
Furthermore, 43.2% of respondents either did not perceive an organisational encouragement to report WPV or were unaware of such a policy. Management models and organisational cultures of public hospitals in China usually emphasise internal and centralised control.[@R49] The hospital occupies a far more dominant position in both resources and authority than its staffs. This means it is hard that doctors' feedback about their experiences in the workplace will produce concrete results; the hospital makes the rules and doctors are supposed to follow them.[@R50] The bureaucratic system enables facilities to be adept at restricting employees' behaviour and encouraging them to tolerate WPV, and leads to their not keen on effectively protecting their employees from WPV or encouraging them to report WPV.[@R8] In order to avoid patient-doctor disputes, some hospital administrators even set guidelines which disproportionately punish their own employees in the event of a WPV incident, no matter whether victims are at fault.[@R10] It is understandable that staff feels helpless and were ignorant of the organisational encouragement to report.[@R10] Consequently, under-reporting and tolerance to WPV may lead to more violent incidents. These findings further indicate inadequate leadership responsibility, lack of training on WPV prevention measures for employees and poor management of aftermaths by health facilities. Campaigns aiming to improve the awareness of WPV and to establish preventive and protective measures at organisational level are urgently needed in China.
Another interesting finding from our study is that victims of WPV rarely reported the incident to medical associations. Only one respondent in our study has chosen these associations as somewhere to report his/her encounter of WPV to. Relevant medical and labour associations are important WPV prevention resources. They play an important role in supporting victims and combatting WPV in health sectors of other countries.[@R3] Most unions designed policies and procedures to deal with violence, and these materials can be used by employers to train their staff on how to prevent WPV.[@R54] Nevertheless, such a role of medical and labour unions in the Chinese setting is largely missing. In 2013, the Chinese Medical Doctor association, Chinese Medical Association, Chinese Hospital Association, and China Health Law Society took their first step to announce a joint call for 'zero-tolerance to violence in health sector'.[@R57] However, follow-up prevention measures and implementation strategies led by these associations are absent. We recommend that professional associations and unions in China take up more accountability and build up systematic capacity to prevent and handle WPV in the health sector.
The study had several limitations. First, this study only investigated hospitals in a highly developed eastern Chinese province, where health services are more accessible and affordable than in less developed regions. It may be partly why we found a lower prevalence of physical violence compared with previous ones. Second, we used a convenience sampling strategy in one province, which may compromise the representativeness of the sample. But we achieved a large sample size of 4862 health workers and a high response rate of 79.8%. Health workers across three tiers of the health system in various cities and counties were recruited for the study. Our study findings still provide new insights and reflect the situation to some extent. Third, we measured Yi Nao at the individual level but Yi Nao incidents are often characterised by gangs gathering and disturbing the hospital for a relatively longer period than a typical momentary violent case. A single Yi Nao incident usually affects a larger number of health workers in the facility. Further research is warranted to measure Yi Nao incidents and the impact at the organisational level. Furthermore, only a small number of the physical violence cases we investigated involved Yi Nao, thereby limiting our understanding of Yi Nao's role in the occurrence of physical violence. Our data suggest, however, a potential link between Yi Nao and higher rates of physical injuries resulting from WPV. To disentangle the role of Yi Nao in physical violence and its consequences, more victim-focused research will be needed. Finally, we excluded other forms of psychological and sexual violence against health workers in this study. Yet this allowed us to gain more insights about the three forms of WPV in China.
Conclusion {#s5}
==========
Our study found that health workers at higher level care facilities, especially secondary hospitals, had a higher risk of encountering physical violence and threats. More attention to WPV in secondary hospitals is warranted. Preventive measures, such as establishing organisational mechanisms and advocating to report, are urgently needed. Proper handling of violent incidents, including for both victims and perpetrators, are lacking, and tolerance to WPV is widespread. Formal guidelines for preventing and managing the aftermath of WPV are essential. Medical associations have the potential to assume a more active, leading role in tackling WPV in China.
Supplementary Material
======================
###### Reviewer comments
###### Author\'s manuscript
The authors thank the 46 participating health faculties in Zhejiang Province, China for their assistance with data acquisition.
**Contributors:** XDZ, LL and DW conceived and designed the study. SZY and NW collected data. SZY, DW and XDZ analysed the data and drafted the paper. LL, TH, KSS and NW provided constructive comments and revised the manuscript. All authors reviewed and approved the final version of the article.
**Funding:** This study was funded by the Zhejiang University Zijin Talent Program.
**Competing interests:** None declared.
**Patient consent for publication:** Not required.
**Ethics approval:** Ethics approval was obtained from the Ethics Committee of Zhejiang University (Approval No ZGL201706-3).
**Provenance and peer review:** Not commissioned; externally peer reviewed.
**Data availability statement:** No data are available.
[^1]: SZY and DW are joint first authors.
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Surgical resection is the standard treatment for localized non-metastatic pancreatic cancer. Data from the Surveillance Epidemiology and End Results (SEER) registry indicate that only about 10% of cases are able to undergo surgery with curative intent, and only a very small number of those are cured because of the high incidence of local relapse and early metastases \[[@B1]\]. Many clinical trials have been carried out using chemotherapy with or without radiation therapy following curative surgical resection, with the aim of preventing local and distant recurrence. With the exception of gemcitabine, neither chemotherapy nor radiation improved survival \[[@B2]\]. For those with locally advanced unresectable or metastatic disease, systemic chemotherapy remains the principal means of improving survival or alleviating cancer-related symptoms
The radiation-sensitive structures in the upper abdomen (small intestine, stomach, kidneys, liver, and spinal cord), prevent conventional radiation therapy to the pancreas or to the pancreatic bed from delivering adequate doses, and irradiation is usually accompanied by severe gastrointestinal intolerance \[[@B3]\]. This may explain in part the absence of survival benefit in patients with locally advanced pancreatic cancer who receive radiation therapy alone. However, 5-FU-based concurrent chemoradiation yields modest survival benefits in patients with locally advanced unresectable pancreatic cancer \[[@B4],[@B5]\]. Despite these findings, survival from pancreatic cancer is still poor, with approximately 23% of patients alive 12 months following diagnosis, and 5% alive at 5 years \[[@B1]\]. New radiation techniques including intensity modulated radiation therapy (IMRT), image guided radiation therapy (IGRT) and stereotactic radiosurgery make it possible to deliver optimally high doses to the target volume with minimal effect on adjacent radiosensitive tissues \[[@B6],[@B7]\]. Helical tomotherapy is a sophisticated image-guided IMRT based on the ring gantry concept, employing a combination of a megavoltage CT scanner and a linear accelerator \[[@B8],[@B9]\]. Capecitabine, a prodrug of 5-FU, is absorbed inert from the gastrointestinal tract and selectively metabolized to 5-FU in tumor cells. This selective conversion achieves higher levels of 5-FU in the tumor cells than can be obtained by intravenous administration of 5-FU. Additionally, radiation can magnify the tumor selectivity of capecitabine by upregulating thymidine phosphorylase in the tumor cells \[[@B10]\]. Capecitabine also acts as a radiation sensitizer by disturbing tumor cell DNA synthesis \[[@B11]\].
In this paper, we report our experience of concurrent administration of capecitabine with helical tomotherapy in patients with inoperable or recurrent pancreatic cancer. We achieved a highly conformal distribution of radiation doses and minimal treatment-related toxicities with excellent target volume responses.
Methods
=======
Patient population
------------------
Between October 2005 and February 2008, nineteen patients with pancreatic cancer were treated with concurrent chemoradiation using helical tomotherapy and capecitabine. They included patients with locally advanced and unresectable disease, and those with local relapse following curative resection or with metastatic disease. Patients who were older than 18 years, who understood the written informed consent document and who were willing to sign it, were eligible for inclusion. The medical records of these patients were reviewed retrospectively. This review was approved by the hospital institutional ethical committee, and written informed consent was obtained from each patient.
Radiotherapy
------------
Radiotherapy was provided by helical tomotherapy (Tomotherapy Incorporated, Madison, WI, USA). Two planning target volumes (PTV) were entered for each patient \[[@B3]\]. PTV1 consisted of the gross tumor volume (GTV) as determined by CT scan, or the tumor bed (in post-surgical cases). PTV2 consisted of the draining lymph nodes, comprising the nodes in the porta hepatis, celiac axis, superior mesenteric and retroperitoneal areas. PTV2 extended 2 cm below the target volume and did not have to include the inferior mesenteric nodes. Both targets were treated simultaneously in 25 daily fractions, 5 days a week. Helical tomotherapy delivered 55 Gy to PTV1 and 50 Gy to PTV2. In some patients with distant metastases (liver or lung), the metastatic lesions were also targeted as another PTV. The distribution of isodoses in the helical tomotherpy treatment planning is shown in Figure [1](#F1){ref-type="fig"}. The dose and volume constraints for the normal structures are listed in Table [1](#T1){ref-type="table"}. Figure [2](#F2){ref-type="fig"} is an average delivered dose-volume histogram for GTV and organ at risk. Capecitabine (Xeloda; Roche Pharmaceuticals, Nutley, NJ) was given at 1600 mg/m^2^/day in two doses on each day of radiation and continued for the duration of the radiation therapy \[[@B3]\].
{#F1}
{#F2}
######
Dose and volume constraints for organs at risk.
Structure Maximum dose constraint (Gy) Volume above limit (%) Maximum dose (Gy) Minimum dose (Gy)
-------------- ------------------------------ ------------------------ ------------------- -------------------
Liver 45.00 10.00 52.83 0.30
Right kidney 1.00 1.00 20.60 0.38
Left kidney 15.00 20.00 20.57 0.54
Small bowel 45.00 10.00 53.34 0.18
Stomach 50.00 10.00 52.95 0.44
Duodenum 10.00 1.00 14.07 0.60
Toxicity assessment
-------------------
Acute toxicity (occurring within 90 days of radiotherapy) was scored using the National Cancer Institute Common Toxicity Criteria (NCI CTC), version 2, morbidity scales \[[@B12]\]. Late toxicity was scored using the Radiation Therapy Oncology Group (RTOG) scale for late toxicity \[[@B13]\]. Patients were evaluated on a weekly basis.
Response assessment
-------------------
The response of each targeted lesion (defined as the in-field tumor response) was evaluated by comparing, by the RECIST criteria, tumor size in pre- and post-treatment CT images 8 weeks after completion of concurrent chemoradiation therapy (CCRT). Two different radiologist evaluated the response rate.
Statistical methods
-------------------
All statistics are descriptive. Survival was compared using the Kaplan-Meier method. Statistical analyses were performed using SPSS software, version 15.0, Chicago.
Results
=======
Patient and tumor characteristics
---------------------------------
The patient characteristics are shown in Table [2](#T2){ref-type="table"}. Twelve were male and seven were female. Median age was 64.0 (range, 46 - 83). Median duration from diagnosis to CCRT was 1.5 months (range, 0.2 - 63.3). The patients were classified with respect to disease status as follows: 1) eight had primarily unresectable disease without metastasis, and no history of previous treatment, 2) three had local relapse following complete resection, and 3) eight had metastatic disease in the liver, lung or peritoneum (three had metastases on first diagnosis and five had metastases that developed during the course of disease). Eight patients had previously received systemic chemotherapy.
######
Patient and tumor characteristics
Patient Sex Age Primary tumor site Previous operation Previous chemotherapy TNM (stage) Duration of follow-up after diagnosis (months) Site of metastasis Site of tomotherapy
--------- ----- ----- -------------------- -------------------- ----------------------------------------- ------------- ------------------------------------------------ -------------------- ---------------------
1 F 53 Head No No T4N1M0(IVA) 2.9 Pancreas
2 M 61 Body Yes Gemcitabine \#6, Cisplatin/Capecitabine T3N0M0 (II) 4.9 Pancreas
3 M 67 Tail No No T4N1M0(IVA) 1.5 Pancreas
4 F 76 Body No Gemcitabine \#5 T4N1M0(IVA) 7.6 Pancreas
5 M 57 Body No Gemcitabine/Capecitabine T4N1M1(IVB) 1.2 Liver Pancreas
6 F 64 Body, tail No No T4N1M0(IVA) 0.2 Pancreas
7 M 67 Body No No T4N1M0(IVA) 1 Pancreas
8 F 71 Body No Gemcitabine/Cisplatin \#3 T3N1M1(IVB) 8 Liver Pancreas
9 M 46 Body, tail No No T4N1M0(IVA) 0.7 Peritoneum Pancreas
10 F 80 Body, tail No No T3N1M0(III) 2.3 Pancreas
11 F 64 Tail No Gemcitabine/Cisplatin \#1 T4N1M1(IVB) 1.3 Liver Pancreas, Liver
12 M 59 Head No No T3N1M1(IVB) 1.4 Liver Pancreas
13 M 68 Body, tail No No T4N1M1(IVB) 0.6 Liver Pancreas, Liver
14 F 54 Neck, body No Gemcitabine/5 - FU \#2 T3N1M0(III) 5.5 Pancreas
15 M 57 Body Yes Gemcitabine, Cisplatin/5 FU \#6 T4N1M1(IVB) 5.2 Liver Pancreas
16 M 83 Head No No T4N1M0(IVA) 0.2 Pancreas
17 M 54 Head No No T4N1M0(IVA) 0.2 Pancreas
18 M 64 Yes Gemcitabine/xeloda \#9, Irinotecan \#2 M1(IVB) 63.3 Lung Lung
19 M 58 Head No No T4N1M0(IVA) 2.4 Pancreas
In-field tumor responses
------------------------
Twenty six lesions were targeted in nineteen patients (Table [3](#T3){ref-type="table"}). They included 15 pancreatic masses, 4 regional metastatic lymph nodes and 7 distant metastatic lesions. Of the 15 pancreatic masses, 8 showed partial responses (PR, 53.3%) and 7 stable disease (SD, 46.6%). Of the 4 regional metastatic lymph nodes, one showed PR (25.0%) and three, SD (75.0%). Of the seven distant metastatic lesions (six hepatic metastases and one pulmonary metastasis), 2 (a pulmonary lesion and a hepatic lesion) showed PR (28.6%) and 5, SD (71.4%). Although there were no complete responses (CR), the overall response rate was 42.3%. It is of interest that no target lesions showed in-field progression during the observation period. Figure [3](#F3){ref-type="fig"} illustrates a typical case of a pancreatic lesion treated with CCRT.
######
In-field tumor response rates of the target lesions after tomotherapy and concurrent capecitabine treatment
Target lesions CR PR SD PD
------------------------------ ------- ----------- ----------- -------
Pancreatic mass (n = 15) 0 (0) 8 (53.3) 7 (46.7) 0 (0)
Regional lymph nodes (n = 4) 0 (0) 1 (25.0) 3 (75) 0 (0)
Distant metastasis (n = 7) 0 (0) 2 (28.6) 5 (71.4) 0 (0)
Liver (n = 6) 0 (0) 1 (16.7) 5 (83.3) 0 (0)
Lung (n = 1) 0 (0) 1 (100) 0 (0) 0 (0)
Overall (n = 26) 0 (0) 11 (42.3) 15 (57.7) 0 (0)
CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease
Numbers in parentheses are percentages
{#F3}
Prognosis and survival
----------------------
The median duration of follow-up after CCRT was 6.5 months (range, 1.1-17.6, Table [4](#T4){ref-type="table"}). The one-year survival rate was 36.8%, and median survival time was 6.5 months (range 1.1-21.0). The median survival time in group I (patients with locally advanced disease without metastases) was 9.25 months (range, 2-18.4, Table [5](#T5){ref-type="table"}). In comparison with patients who had locally advanced and unresectable disease without metastases or a previous chemotherapy history, the others (those who had metastases at the time of CCRT, and a case with local relapse after previous curative surgery as well as those with a history of previous chemotherapy) showed poor survival (*p*= 0.063); 4.4 months (range, 1.1-21) versus 12.55 months (range, 6.5-18.4). Of the patients in group I, those who had no history of previous chemotherapy survived better than those with a history of previous chemotherapy (*p*= 0.0009); 12.55 months (range, 6.5-18.4) versus 3.9 months (range, 2-5.8).
######
Clinical outcomes in the nineteen patients treated with tomotherapy and concurrent capecitabine
Patient Overall In-field tumor response Duration of tumor response (months) Treatment-related toxicity Duration of follow-up after tomotherapy (months) Out-field progression state Cause of death other than cancer progression Duration of survival after tomotherapy (months)
--------- --------------------------------- ------------------------------------- ------------------------------ -------------------------------------------------- ----------------------------- ---------------------------------------------- -------------------------------------------------
1 Stable disease 5.9 fatigue (grade 2) 7.3 Progressed 7.3
2 Partial response hand foot syndrome (grade 1) 3.2 4.8
3 Stable disease 10.7 11.2
4 Stable disease 3.8 5.8 Progressed 5.8
5 1.1 Pneumonia 1.1
6 Partial response fatigue (grade 2) 6.8 13.9
7 Stable disease 6.9 15 Progressed 16.3
8 1.8 1.9
9 Partial response 2.4 nausea (grade 1) 4.4 Progressed 4.4
10 Partial response 13.6 13.9
11 Stable disease 4.1 fatigue (grade 1) 4.1 Stable disease Pulmonary thromboembolism 4.1
12 Stable disease 2.2 3.4 Progressed 3.9
13 Stable disease 2.5 6.5 Progressed 6.5
14 2 DUB, Pneumonia 2
15 Stable disease 10.5 10.5
16 Partial response 7.3 7.3 Stable disease Pneumonia 7.3
17 Partial response 14.9 18.4
18 Partial response 17.6 21
19 Partial response 3.2 6.2 Progressed 6.5
DUB, duodenal malignant ulcer bleeding
######
Survival of pancreatic cancer patients treated with tomotherapy and concurrent capecitabine
Group Characteristics Median duration of survival (months)
------- -------------------------------------------------------------------------- --------------------------------------
I Locally advanced without metastasis (n = 10) 9.25 (2.00-18.4)
No previous chemotherapy (n = 8) 12.55 (6.50-18.4)
Previous chemotherapy (n = 2) 3.90 (2.00, 5.8)
II Locally relapsed without metastasis following complete resection (n = 1) 4.80 (4.80)
III Metastatic disease (n = 8) 4.25 (1.10-21.00)
De novo (n = 3) 4.40 (3.90-6.50)
Relapsed (n = 5) 4.10 (1.10-21.00)
Data in parentheses are ranges of survival times
Progression of disease outside the targeted tumor volume (defined as the out-field progression) occurred in 7 patient. The median time to out-field progression was 3.8 months (range 2.2-7.3) with or without systemic chemotherapy following CCRT.
Toxicity
--------
Acute toxicity is summarized in Table [6](#T6){ref-type="table"}. As shown, only minor toxicities developed. The most common acute toxicity was grade 1 or 2 fatigue that occurred 2 to 3 weeks after the start of tomotherapy (three patients, 16.7%). Intriguingly, no treatment was interrupted due to gastrointestinal side effects. Only grade 1 nausea developed in one patient (5.6%). Grade 1 hand-foot syndrome related to oral capecitabine also developed in one patient (5.6%). None experienced hematologic toxicities during the treatment. All toxicities were manageable medically and regressed spontaneously, and they did not interfere with the scheduled radiotherapy. There were no treatment-related deaths and no grade 3 or 4 toxicity. Therefore, treatment was well tolerated by all patients.
######
Treatment-related toxicity
Grade 1 Grade 2
-------------------- --------- ----------
Fatigue 1 (5.6) 2 (11.1)
Nausea 1 (5.6) 0 (0)
Hand-foot syndrome 1 (5.6) 0 (0)
Data in parentheses are percentages
Discussion
==========
The majority of pancreatic cancer patients have advanced disease at the time of diagnosis due to a lack of symptoms and signs. Without treatment, mean survival time is 4-6 months and overall 5-year survival remains less than 5%. \[[@B14]\]. The only curative option is surgery, but only 10-20% of patients have tumors appropriate for radical resection \[[@B15]\]. Advanced pancreatic cancer is generally incurable and all therapies have significant limitations. The response to systemic chemotherapy is poor, with an approximately 20% response rate. The conventional radiation dose to the tumor volume is not large enough to cure patients because pancreatic tumors move markedly as patients breathe, and are surrounded by the duodenum, which is the dose-limiting organ \[[@B16]\]. Compared with chemotherapy alone or radiotherapy alone, chemoradiotherapy prolongs median survival somewhat, to approximately 9-12 months, in those with locally advanced unresectable disease \[[@B5]\].
Helical tomotherapy, a new radiotherapy system, is a helical IMRT with integrated CT imaging, offering highly conformal radiation with normal tissue sparing. The basis of image guidance is utilizing daily images gained in the treatment position in order to visualize daily organ variations and setup errors \[[@B17]-[@B19]\]. The radiation is discharged as a fan beam by a linear accelerator mounted on a turning gantry and is adjusted by a rapid pneumatically driven binary slit collimator \[[@B20]\]. The speed of gantry rotation and table movement is uniform for the entire fraction. Hence helical tomotherapy can provide significant conformal dose distributions at numerous locations \[[@B21]-[@B24]\].
Helical tomotherapy can treat multiple lesions more rapidly than conventional radiotherapy, for which multiple target points are necessary \[[@B20]\]. Moreover it is an ideal device for delivering multifocal, high-dose radiation without a significant increase in toxicity \[[@B9],[@B25]\]. Thus it allows us to treat patients with multiple targets including metastatic lesions.
The ideal concurrent chemotherapeutic agent in the therapy of pancreatic cancer should have both a systemic effect and radiosensitizing properties \[[@B16]\]. Capecitabine has a pronounced radiosensitizing effect on tumor cells such that DNA strand breakage induced by radiation is more difficult to repair \[[@B11]\]. The regimen described here takes advantage of the tumor-selective ability of capecitabine to enhance radiation effects within the tumor but not in the surrounding normal tissues. This can be ascribed to a higher 5-FU concentration in tumor cells and the induction of thymidine phosphorylase by the irradiation \[[@B10]\]. Also, the use of capecitabine is attractive because it is absorbed as an inert drug, causing little direct toxicity in the gastrointestinal tract.
Ben-Josef et al \[[@B3]\] treated 15 patients with unresectable or recurrent pancreatic cancer with IMRT and concurrent capecitabine. In that study, the regimen was well tolerated without significant toxicities, and efficacy was encouraging.
Another basis for offering radiotherapy to patients with pancreatic cancer is palliation of symptoms due to local invasion, such as biliary and gastrointestinal obstruction \[[@B26]\]. The drawbacks of radiotherapy include the acute and chronic toxicities of radiotherapy, particularly when the indication is palliation. Because of its ability to restrict the dose to normal organs and minimize radiation toxicities, helical tomotherapy may be an ideal palliative option for challenging cases of pancreatic cancer \[[@B27]\].
In our study, the overall in-field tumor response rate was 42.3%. Previous studies have reported 10-50% response rates for locally advanced pancreatic cancer with chemoradiotherapy \[[@B28]-[@B32]\]. The high response rate in our study is due to in-field assessment of responses. Considering the advanced stage of our patients, the in-field response rate is encouraging. It may be possible to increase this response rate by increasing the dose of capecitabine.
It may be noted that helical tomotherapy with concurrent capecitabine yielded excellent disease control within the radiation field, with an in-field disease control rate of essentially 100%. This could be thought to be a significant therapeutic benefit.
Median overall survival after tomotherapy was only 6.5 months. This was because of advanced stages of our study population (tumor stages III or IV). Our study included patients with locally advanced disease, local relapse following complete resection, and metastases.
Patients who had locally advanced disease without metastasis or a previous history of chemotherapy showed a tendency to survive longer than the others (12.55 versus 4.4 months) after tomotherpy. In our opinion, tomotherapy with concurrent capecitabine should be the first option for inoperable pancreatic cancer, especially in patients without metastases or a previous history of chemotherapy.
Although our patients were elderly, with a median age of 64, treatment was well tolerated. The majority of treatment-related toxicities were mild and transient. Only grade 1/2 fatigue, nausea and hand-foot syndrome developed, and they subsided with symptomatic care and without prematurely stopping radiotherapy. There was no direct treatment-related grade 3/4 toxicity or death. Therefore helical tomotherapy is a safe option in the treatment of advanced pancreatic cancer.
This study had several limitations. First, the number of cases was low. Second, the heterogeneity of the study population made direct comparison with other studies difficult. Third, long term treatment effects and late toxicities remain to be evaluated because the median follow-up time in our study was relatively short.
Although there was no in-field progression during the observation period, out-field progression occurred in seven patients. This observation provides a rationale for follow-up systemic chemotherapy after tomotherapy to prevent or delay out-field progression. Hence, subsequent chemotherapy such as gemcitabine alone or erlotinib combined with gemcitabine should be performed in eligible patients \[[@B33],[@B34]\].
There are only two examples of the clinical application of helical tomotherapy for locally advanced pancreatic cancer \[[@B35]\]. To the best of our knowledge, this is first comprehensive analysis of the clinical application of helical tomotherapy for a group of inoperable or recurrent pancreatic cancers.
Conclusions
===========
Our study demonstrates that helical tomotherapy with concurrent capecitabine is a feasible and safe option for locally advanced unresectable or metastatic pancreatic cancer. Our preliminary data yielded a high local control rate. Because of its ability to irradiate multiple targets simultaneously, helical tomotherapy could be an ideal palliative option for challenging cases of pancreatic cancer with metastases. Further large scale clinical trials are needed to verify the efficacy and safety of helical tomotherapy with concurrent capecitabine for treating advanced pancreatic cancer. Also, careful selection of those patients that stand to benefit from this regimen is needed.
Competing interests
===================
The authors declare that they have no competing interests.
Authors\' contributions
=======================
JJ participated in data collection, performed the statistical analysis and drafted the manuscript. CH conceived of the study, and participated in its design and coordination. JJ participated in data collection and helped to draft the manuscript. JK helped in data collection and analysis. YK helped in data collection and drafted the manuscript. BL, BK, HC, CK and IC helped to data analysis and drafted the manuscript. All authors read and approved the final manuscript.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Neural stem cells give rise to new neurons in the subgranular zone of the hippocampal dentate gyrus (DG) and the subventricular zone (SVZ) of the lateral ventricle throughout life. The generation of a mature neuron involves a stereotypic sequence of developmental steps including proliferation, cell cycle exit, neuronal fate determination, maturation and functional integration into the pre-existing neural circuit. These developmental stages can be distinguished on the basis of the expression of stage-specific marker proteins [@pone.0062693-Kempermann1].
Doublecortin (DCX) is a microtubule binding protein. The doublecortin (DCX) superfamily consists of 11 conserved members [@pone.0062693-Reiner1], all containing a DCX domain, which is necessary for microtubule binding [@pone.0062693-Kim1]. DCX is highly expressed in migrating neurons of the developing central nervous system [@pone.0062693-desPortes1], [@pone.0062693-Francis1], [@pone.0062693-Gleeson1]. In the adult mouse brain, DCX is almost exclusively expressed by immature newborn neurons in the DG and the SVZ/OB-system and is commonly used to distinguish immature neurons from non-neuronally committed precursors and mature neurons, and to estimate neurogenic activity [@pone.0062693-CouillardDespres1], [@pone.0062693-CouillardDespres2], [@pone.0062693-Ming1].
Mutations in the X-linked *Dcx* gene are associated with abnormal neuronal migration, and are causally linked to epilepsy, mental retardation, lissencephaly in male and subcortical laminar heterotopia in female human subjects [@pone.0062693-desPortes1], [@pone.0062693-desPortes2], [@pone.0062693-Dijkmans1]. Interestingly, there are species specific requirements for DCX function in the development of distinct forebrain regions. In humans, DCX is required for the lamination of the hippocampus and the neocortex [@pone.0062693-Corbo1]; in mice, only the lamination of the hippocampus is dependent on DCX function. RNAi-mediated knockdown of DCX causes heterotopia formation in the rat neocortex [@pone.0062693-Bai1] but not in the murine neocortex [@pone.0062693-Ramos1]. Short-hairpin (sh) RNA-mediated DCX knockdown in the early postnatal SVZ/OB system of mice causes abnormal neuronal migration and changes the fate of developing neurons [@pone.0062693-Belvindrah1]. Despite the widespread use of DCX as a marker for immature neurons in the adult neurogenic lineage, little is known about the specific function of DCX in adult neurogenesis. Analysis of DCX null mutant mice suggested that DCX is required for the migration of adult-born neurons in the SVZ/OB-system [@pone.0062693-Koizumi1]. DCX null mutant mice, however, lack DCX function already during embryonic development and thus do not allow to distinguish whether the observed migratory defects result from a direct function of DCX in adult-born neurons or result from defective CNS development. Here, we employ a MMLV-retrovirus based approach to overexpress or knockdown DCX specifically in the neurogenic lineage of the DG and the SVZ/OB-system during adulthood. Our results provide strong evidence that DCX is dispensable for the development of adult born neurons in wildtype mice.
Materials and Methods {#s2}
=====================
Animals {#s2a}
-------
All animal experiments were performed in accordance with the European Communities Council Directive (86/609/EEC). Stereotactic injections of retroviruses into the brain of adult mice were approved by the Government of Upper Bavaria. For all experiments, seven weeks old female C57BL/6-J mice were ordered from Charles River and retrovirally injected at an age of eight weeks. Mice were grouped housed in big rat cages under a 12 h light/dark cycle and had *ad libitum* access to food and water. Cages were containing a house and a running wheel.
Vector Construction {#s2b}
-------------------
For mouse moloney retrovirus (MMLV) -mediated expression of DCX, the cDNA of the murine DCX (oligos \# 796, \# 795; [Table 1](#pone-0062693-t001){ref-type="table"}) was tagged with FLAG (3x) and cloned into CAG-IRES-GFP [@pone.0062693-Jagasia1] or CAG-IRES-RFP to generate CAG-DCX-3xFLAG-IRES-GFP or CAG-DCX-3xFLAG-IRES-RFP. The CAG-RFP retrovirus has been described previously [@pone.0062693-Jagasia1], [@pone.0062693-Zhao1]. pcDNA^TM^6.2-GW/EmGFP-miR from Invitrogen was filled with a linker (\# 1015, \# 1016) to generate two BsmBI restriction sites and restriction sites (BglII, AvrII, XhoI, MfeI, HindIII) unique in the CAG-vectors. The generated vector was named pcDNA6.2-GW-EmGFP-filled. EmGFP-miR-flanking cassette was PCR amplified (\# 1025, \# 1027) from pcDNA6.2-GW-EmGFP-filled and cloned into pKSSP (pBluescript KS modified with a SfiI site near the KpnI site and a PmeI site next to the SacII site; from Fred Gage, Salk Institute, La Jolla, USA) to generate pKSSP-EmGFP-miR. Three different miRNAs against DCX were constructed using BLOCK-iT™ RNAi Express from Invitrogen miDCX(1) (\# 1072, \# 1073), miDCX(2) (\# 1074, \# 1075) and miDCX(3) (\# 1076, \# 1077) and cloned into BsmBI sites of pKSSP-EmGFP-miR to generate pKSSP-EmGFP-miDCX. EmGFP-miDCX was subcloned into CAG-GFP by replacing GFP to generate CAG-EmGFP-miDCX. For construction of the internal control vector CAG-RFP-miCtr-IRES-RFP, EmGFP was replaced in pcDNA6.2-GW-EmGFP-filled by PCR amplified (\# 1220, \# 1221) RFP from CAG-RFP. RFP-miR-flanking cassette was PCR amplified (\# 1222, \# 1027) from pcDNA6.2-GW-RFP-filled and cloned into pKSSP to generate pKSSP-RFP-miR. As a control miRNA, the sequence (\# 1170, \# 1171) of pcDNA™6.2-GW/miR-neg control plasmid from Invitrogen was used and cloned into pKSSP-RFP-miR to generate pKSSP-RFP-miCtr. RFP-miCtr was subcloned into CAG-IRES-RFP to generate CAG-RFP-miCtr-IRES-RFP.
10.1371/journal.pone.0062693.t001
###### Oligos used in this study.
{#pone-0062693-t001-1}
oligo name \# sequence 5′ --3′
------------------ ------ ------------------------------------------------------------------
DCX+1-for 796 GGGGCCGCCTGGGCCATGGAACTTGATTTTGGACA
DCX+1098-rev 795 GGGAATTCCATGGAATCGCCAAGTGA
miR_linker_for 1015 TGCTAGAGACGAGATCTCCTAGGCTCGAGCAATTGAAGCTTCGTCTCA
miR_linker_rev 1016 CCTGTGAGACGAAGCTTCAATTGCTCGAGCCTAGGAGATCTCGTCTCT
EmGFPmiRNA-for 1025 CAGGGCCGCCTCGGCCAATGGTGAGCAAGGGCGAGGAGCTG
miRNA-rev 1027 GTAGTTTAAACGGCCATTTGTTCCATGTGAGTGCT
miDCX(1)-for 1072 TGCTGATTACTTAATGCCTGCAAGGTGTTTTGGCCACTGACTGACACCTTGCACATTAAGTAAT
miDCX(1)-rev 1073 CCTGATTACTTAATGTGCAAGGTGTCAGTCAGTGGCCAAAACACCTTGCAGGCATTAAGTAATC
miDCX(2)-for 1074 TGCTGTGGAGTAGCACACTTTGAAGTGTTTTGGCCACTGACTGACACTTCAAAGTGCTACTCCA
miDCX(2)-rev 1075 CCTGTGGAGTAGCACTTTGAAGTGTCAGTCAGTGGCCAAAACACTTCAAAGTGTGCTACTCCAC
miDCX(3)-for 1076 TGCTGATTTCTAGATGCTTTGTCTCGGTTTTGGCCACTGACTGACCGAGACAACATCTAGAAAT
miDCX(3)-rev 1077 CCTGATTTCTAGATGTTGTCTCGGTCAGTCAGTGGCCAAAACCGAGACAAAGCATCTAGAAATC
miR-neg-for 1170 TGCTGAAATGTACTGCGCGTGGAGACGTTTTGGCCACTGACTGACGTCTCCACGCAGTACATTT
miR-neg-rev 1171 CCTGAAATGTACTGCGCGTGGAGACGTCAGTCAGTGGCCAAAACGTCTCCACGCAGTACATTTC
RFP+1-DraI-for 1220 GGCTTTAAAATGGCCTCCTCCGAGGACGTCATC
RFP+678-Dral-rev 1221 CCCTTTAAACGATCGACGGCCACGAAGTGCTTAGCTTAGGCGCCGGTGGAGTGGC
RFPmiRNA-for 1222 CAGGGCCGCCTCGGCCAATGGCCTCCTCCGAGGACGTCATC
Note: as the GFP signal in CAG-EmGFP-miDCX retrovirally transduced cells was not sufficient to study the dendrite morphology of transduced cells, CAG-EmGFP-miDCX-IRES-GFP was generated. Fluorescence signal was higher compared to IRES-less vector, but IRES-GFP containing vector was not resulting in an *in vivo* knockdown of DCX!
Retrovirus Preparation {#s2c}
----------------------
Retroviruses were produced from the plasmids above as described previously [@pone.0062693-Mu1]. Viruses were harvested four times; first harvest was three days after transfection and the following ones every second day. The obtained viral titers ranged between 10^5^--10^8^ colony forming units (cfu)/ml, dependent on the harvest number and the construct. For CAG-EmGFP-miDCX viruses only low titers (\<10^6^ cfu/ml) were obtained.
Stereotactic Injections {#s2d}
-----------------------
All experiments were conducted under the condition that animals have ad libitum access to a running wheel during the whole duration of the experiment. This measure was necessary to enhance proliferation and thus the number of transduced cells in the context of the loss-of-function experiments, because of the low titer of the CAG-EmGFP-miDCX vectors. In order to be able to compare gain- and loss-of-function results, animals in the gain-of-function experiments also had free access to a running wheel.
Mice were deeply anesthetized with a mixture of fentanyl (0.05 mg/kg bodyweight), midazolam (5 mg/kg bodyweight) and medetomidine (0.5 mg/kg bodyweight). Anaesthesia was reverted with a mixture of buprenorphine (0.1 mg/kg bodyweight), atipamezole (2.5 mg/kg bodyweight) and flumazenil (0.5 mg/kg bodyweight). Mice were stereotactically injected with 0.9 µl of CAG-DCX-3xFLAG-IRES-GFP and CAG-RFP (titer 1.11×10^7^ cfu/ml each) or CAG-EmGFP-miDCX (estimated titer \<10^6^ cfu/ml; precise titer could not been determined due to low GFP signal; undiluted usage) and CAG-RFP-miCtr-IRES-RFP (titer 1.11×10^7^ cfu/ml) retroviruses, into the left and right dentate gyrus (coordinates from bregma were −1.9 mm anterior/posterior, ±1.6 mm medial/lateral, −1.9 mm dorsal/ventral from dura) or into the left and right RMS (coordinates from bregma were +2.3 mm anterior/posterior, ±0.8 mm medial/lateral, −2.9 mm dorsal/ventral from dura). Group size was *n* = 4--5 mice per time point for dentate gyrus neurogenesis analysis, and *n* = 3 per time point for SVZ/OB neurogenesis.
Tissue Processing {#s2e}
-----------------
Mice were sacrificed with CO~2~ at the time points indicated. They were perfused for 5 min with PBS (pH 7.4), followed by 4% formaldehyde (FA) for 5 min. Brains were removed and postfixed in 4% FA over night at 4°C and were subsequently transferred to 30% sucrose in 0.1 M phosphate buffer. Forty, 80 or 120 µm-thick coronal brain sections were cut using a sliding microtome (Leica Microsystems). Brain sections were stored at −20°C in cryoprotect solution (0.05 M phosphate butter, 25% ethylenglycol v/v, 25% glycerol v/v).
Immunhistochemistry and Phenotyping of Transduced Cells {#s2f}
-------------------------------------------------------
Brain sections were washed three times in tris buffered saline (TBS) and blocked for 60 min in TBS supplemented with 3% normal donkey serum and 0.25% Triton X-100. Sections were incubated with primary antibodies diluted in blocking solution over one or two nights at 4°C. Primary antibodies against the following antigens were used: doublecortin (DCX; goat, 1∶250; Santa Cruz Biotechnology sc-8066), GFP (chicken, 1∶1.000; Aves Labs GFP-1020), RFP (rat, 1∶50; [@pone.0062693-Rottach1]), calretinin (rabbit, 1∶1.000; Swant 7699/4) and NeuN (mouse, 1∶50; hybridoma supernatant, Richard Mullen, University of Utah, Salt Lake City, UT). Sections were washed three times with TBS, blocked for 30 min and incubated with secondary antibodies, diluted 1∶250 in blocking solution for 2 h at room temperature. Secondary antibodies coupled to FITC, Cy3, Cy5 or CF633 (Biotium, 1∶500) from The Jackson Laboratory (Bar Harbor, USA) were previously resuspended in 400 µl water. After 10 min incubation with DAPI (10 mg/ml 4′,6-diamidino-2-phenylindole; 1∶10.000; Sigma-Aldrich) diluted in TBS, sections were washed twice in TBS and mounted in Aqua PolyMount (Polysciences). For phenotypical characterization of the retrovirally transduced cells, 40 µm sections containing the dentate gyrus or RMS were selected and immunstained. Selected sections from different animals were similar with regard to their position along the dorsoventral axis (for dentate gyrus analysis) and along the mediolateral axis (for RMS/OB analysis). Transduced cells were identified based on the expression of GFP and/or RFP. All transduced cells were analyzed by confocal microscopy for immunoreactivity for the respective marker (*n = *20--220 cells per animal and marker). Confocal single-plane images and *Z*-stacks were obtained on a Leica SP5 confocal microscope. Z-stacks were used for analysis of spines, dendritic morphology, phenotyping and migration analysis. Distance between single planes were 0.21 µm for spine analysis, 0.3 µm for dendritic morphology, and 1.5 µm for all other analyses.
Morphology {#s2g}
----------
For analysis of dendritic morphology, 120 µm thick sections were used to ensure that the section comprised the entire dendritic tree. For analysis of spines, 120 and 80 µm thick sections were used. RFP~only~-transduced control cells and yellow double transduced cells were morphologically analysed for the following parameters: complexity (Scholl analysis), number of branches, total dendritic length and number of spines per 10 µm. Only the RFP channel of confocal images of 2--6 RFP~only~ and yellow cells respectively per animal (*n* = 3) was analysed with Imaris7.2 (Bitplane).
Migration in RMS/OB System {#s2h}
--------------------------
Images were taken from three different positions along the RMS/OB system: rostral part of the RMS (RMS1), middle part of the RMS (RMS2), and core region of the OB where new neurons switch their migratory mode from tangential to radial migration (see [Figure 8E](#pone-0062693-g008){ref-type="fig"} and [Figure 4](#pone-0062693-g004){ref-type="fig"} in [@pone.0062693-Belvindrah1]). All virus-transduced cells in these positions were counted. Comparison of the ratio of yellow cells to all red cells (yellow+RFP~only~ cells) in these regions was performed to determine whether DCX manipulation altered positioning/migration along the RMS/OB system.
Statistical Analysis {#s2i}
--------------------
Unpaired Student's t-test was used for analysis. Differences were considered statistically significant at p\<0.05. All data are presented as mean ± s.d.
Transfection and Western Blot {#s2j}
-----------------------------
HEK 293T cells were co-transfected with the CAG-EmGFP-miDCX and the CAG-DCX-3xFLAG-IRES-RFP plasmids or alone with the CAG-DCX-3xFLAG-IRES-RFP plasmid as a negative control using the calcium phosphate method. 2.5 µg of DNA were transfected per plasmid. Proteins were extracted as described previously [@pone.0062693-Ehm1]. Nuclear and cytosolic extracts were pooled and 10 µg of whole cell extracts were analyzed by western blot.
Proteins were blotted on a PVDF membrane and blocked with 5% slim milk powder in PBS for 1h at room temperature (RT). Primary antibodies were diluted in a solution of 2% slim milk powder in PBST (0,05% Tween20 in PBS) and incubation was performed for 1 h at RT. Antibodies against the following antigens were used: FLAG (mouse, 1∶2.000; Sigma F1804) and β-actin (mouse, 1∶10.000; Abcam ab6276). Primary antibody incubation was performed under constant shaking for 1 h at RT. The membrane was washed three times with PBST and incubated with HRP-conjugated secondary antibodies diluted in 2% slim milk powder in PBST for 45 min at RT (mouse, 1∶10.000; The Jackson Laboratory 115-035-003). After three washes in PBST, blots were incubated with ECL solution (GE Healthcare). Chemiluminescence signals were detected with a CCD-camera (Fusion-SL 3500.WL; Peqlab) and proteins were quantified with the appropriate software (FusionCapt). FLAG signals were related to β-actin signals whereby DCX-3xFLAG single transfection signal was set to 1.
Both antibodies used in this study were raised in mice and the expected sizes were almost the same (∼50 kDa). As stripping of the blots for either FLAG or β-actin antibody was inefficient, it was impossible to hybridize one membrane with both antibodies. Therefore for each construct two blots were performed with the same amount of proteins and the same protein extracts.
Results {#s3}
=======
DCX is almost exclusively expressed in immature neurons of the two neurogenic niches, the hippocampal dentate gyrus (DG) and the subventricular zone (SVZ; [Figure 1](#pone-0062693-g001){ref-type="fig"}). Some postmitotic cells of the piriform cortex express DCX. A subpopulation of these cells retains immature characteristics and is associated with synaptic plasticity [@pone.0062693-Klempin1]. The main goal of this study was to investigate the role of DCX in adult neurogenesis, where it is commonly used as a marker for the immature state of adult born neurons. To address this question we performed gain- and loss-of-function studies.
{#pone-0062693-g001}
Gain of Function {#s3a}
----------------
Previous work demonstrated that voluntary wheel running accelerates maturation of newborn dentate granule neurons as evidenced by morphological and electrophysiological criteria and an earlier transition of newborn neurons from a DCX-positive to a DCX-negative stage [@pone.0062693-Piatti1]. Consistent with this observation, we found that approximately 42% of newborn neurons in mice housed under standard conditions expressed DCX [@pone.0062693-Jagasia1], whereas DCX expression was absent from newborn neurons at 28 dpi (days post injection), if animals have ad libitum access to a running wheel during the whole duration of the experiment.
To investigate if cells are kept in an immature state by overexpression/prolonged expression of DCX, eight weeks old female C57/BL6 mice were injected with CAG-RFP and CAG-DCX-3xFLAG-IRES-GFP retroviruses. Animals were sacrificed at 28 dpi ([Figure 2A](#pone-0062693-g002){ref-type="fig"}). Three populations of retrovirally transduced cells were expected: single transduced cells with either CAG-RFP (red RFP~only~ cells) or CAG-DCX-3xFLAG-IRES-GFP (green GFP~only~ cells) or double transduced cells (red and green = yellow cells). RFP~only~ cells were used as an internal control and compared with GFP~only~ and/or yellow cells.
{#pone-0062693-g002}
Immunstainings for DCX showed that, at 28 dpi, none of the RFP~only~ cells (white arrows) were positive for DCX, whereas all green cells (yellow arrows) were expressing DCX ([Figure 2B](#pone-0062693-g002){ref-type="fig"}). Brain sections were stained for calretinin, a calcium binding protein, which is expressed in immature dentate granule neurons [@pone.0062693-Brandt1]. 11.29% ±2.13% of the RFP~only~ control cells were positive for calretinin whereas 8.05% ±1.32% of DCX overexpressing green cells were calretinin positive (p = 0.04888412; [Figure 2C](#pone-0062693-g002){ref-type="fig"}), indicating that DCX overexpression does not maintain newly generated dentate granule neurons in a calretinin-positive immature state.
DCX knockout mice show branching defects in migrating interneurons [@pone.0062693-Kappeler1]. To investigate if morphological maturation is influenced by DCX, dendrite complexity (via Scholl analysis; [Figure 3A--C](#pone-0062693-g003){ref-type="fig"}), number of branches ([Figure 3D](#pone-0062693-g003){ref-type="fig"}), total dendritic length ([Figure 3E](#pone-0062693-g003){ref-type="fig"}) and the number of spines per 10 µm (8.40±1.87 spines/10 µm in RFP~only~ cells and 8.14±0.13 spines/10 µm in yellow cells) were analysed ([Figure 3F](#pone-0062693-g003){ref-type="fig"}). None of these parameters were changed in DCX overexpressing cells compared to RFP expressing control cells.
{#pone-0062693-g003}
Next, we investigated, whether migration of newborn cells in the DG is impaired by DCX overexpression. Under physiological circumstances, new dentate granule neurons migrate only 1--2 cell diameters from the SGZ towards the molecular cell layer. Around 80% of mature neurons are located in the inner part of the granular layer, 20% in the middle layer and only a small proportion of cells are located in the outer layer [@pone.0062693-Mathews1]. This short distance migration can be largely extended under pathological conditions such as knockdown of the schizophrenia susceptibility gene *Disc1* [@pone.0062693-Kim2] or seizures [@pone.0062693-OverstreetWadiche1], [@pone.0062693-Jessberger1], [@pone.0062693-Parent1], where newborn neurons can migrate even beyond the boundaries of the granule cell layer. DCX overexpressing green cells, however, were equally distributed like the RFP~only~ control cells ([Figure 2B](#pone-0062693-g002){ref-type="fig"}), strongly suggesting that DCX expression did not alter the migration of newborn neurons.
Taken together, our results indicate that prolonged expression/overexpression of the immature marker protein DCX does not influence maturation or migration of newly generated dentate granule neurons at 28 days post injection.
Loss-of-function {#s3b}
----------------
DCX is required for the lamination of the hippocampus during development [@pone.0062693-Corbo1] and was suggested to be required for the migration of adult-born neurons in the SVZ/OB-system [@pone.0062693-Koizumi1]. To address the question whether acute loss-of DCX function affects the maturation and/or migration of adult born cells in the DG or SVZ/OB-system of wildtype mice, a MMLV-retroviral expression vector was generated to knockdown DCX in neural stem cell progeny *in vivo*. Three different miRNAs against DCX were constructed using BLOCK-iT™ RNAi Express (Invitrogen).
Prior to virus production, knockdown efficiency was tested *in vitro* by co-transfection of HEK 293T cells with CAG-EmGFP-miDCX and CAG-DCX-3xFLAG-IRES-RFP and western blot against DCX ([Figure 4](#pone-0062693-g004){ref-type="fig"}). Constructs \#1 and \#3 were directed against a sequence located in the ORF while construct \#2 was directed against a sequence in the 3′UTR of DCX. Consequently for construct \#2, an *in vitro* knockdown of DCX was not expected as CAG-DCX-3xFLAG-IRES-RFP only contained the cDNA of DCX ([Figure 4B--C](#pone-0062693-g004){ref-type="fig"}, lane 2). Compared to CAG-DCX-3xFLAG-IRES-RFP only transfection, knockdown efficiency of construct \#1 was 10.5%. Construct \#3 showed much higher efficiency with a reduction of 66.9% in DCX expression ([Figure 4C](#pone-0062693-g004){ref-type="fig"}). To test if the miDCX vector was able to knockdown DCX *in vivo*, three female mice were injected with CAG-EmGFP-miDCX retrovirus for either construct \#2 or \#3 and analysed at 11 dpi ([Figure 5](#pone-0062693-g005){ref-type="fig"}). Both constructs were able to knock DCX down *in vivo*, whereby the efficiency of construct \#3 was higher. Hence, construct \#3 was used for subsequent *in vivo* loss-of-function studies.
{#pone-0062693-g004}
{#pone-0062693-g005}
For maturation and survival analyses, five animals were injected with CAG-GFP-miDCX(3) and CAG-RFP-miCtr-IRES-RFP (internal control) retroviruses and sacrificed for analysis at the following time points: 10, 14, 17 and 28 dpi ([Figure 6A](#pone-0062693-g006){ref-type="fig"}). Knockdown efficiency was determined by DCX immunostainings ([Figure 6C](#pone-0062693-g006){ref-type="fig"}). While 89.6±4.87% of the RFP~only~ control cells were positive for DCX, only 14.32±4.06% of the GFP cells were positive for DCX with comparable signal intensity as the red control cells. The remaining GFP-labelled cells were either DCX negative or had a barely detectable DCX signal.
{#pone-0062693-g006}
To test if maturation of the newborn cells is accelerated following DCX knockdown, immunostainings for the splicing factor NeuN, which is commonly used as a marker for mature neurons, were performed. At 10 dpi, 80% of the control cells were already NeuN positive ([Figure 6D](#pone-0062693-g006){ref-type="fig"}; in different experiments we detected in the meantime that roughly the same percentage of 5 dpi old cells were already NeuN positive; unpublished data). DCX knockdown did not significantly enhance the fraction of NeuN-expressing cells, demonstrating that loss-of-DCX does not promote the transition to a NeuN positive developmental stage.
Comparison of the position of DCX-knockdown cells and control cells, indicated that lack of DCX expression does not alter migration along the SGZ/molecular cell layer axis ([Figures 6B](#pone-0062693-g006){ref-type="fig"}, [2D](#pone-0062693-g002){ref-type="fig"}).
DCX knockout mice display branching defects in migrating interneurons [@pone.0062693-Kappeler1]. Analysis of 28 dpi cells, however, revealed no changes in complexity, the number of branches and the total dendritic length ([Figure 7](#pone-0062693-g007){ref-type="fig"}), indicating that acute loss-of-DCX does not alter the overall morphological development of new DG neurons.
{#pone-0062693-g007}
Finally, we determined whether DCX is required for the survival of new DG neurons. Coinjection of the CAG-GFP-miDCX(3) virus together with the internal control virus CAG-RFP-miCtr-IRES-RFP for at least two different time points, enables the investigation of the influence of DCX knockdown on the number of newborn cells by comparison of the ratios of yellow cells to all red cells (yellow+RFP~only~ cells) at the different time points, independently of variations in the viral titers and injection sites [@pone.0062693-Jagasia1], [@pone.0062693-Tashiro1]. If survival of newborn cells would be affected, the ratio would decline. Survival rate of the DCX lacking cells was determined by comparison of the ratios at 10 dpi and 28 dpi ([Figure 6E](#pone-0062693-g006){ref-type="fig"}). Ratios did not significantly change suggesting that the survival rate of new neurons was not affected by DCX ablation. To exclude the possibility that DCX affected survival prior to 10 dpi, the experiment was preformed for the 5 dpi and 10 dpi ([Figure 6E](#pone-0062693-g006){ref-type="fig"}'; ratios between experiments can strongly vary due to different amounts of virus particles in the different injection mixes, therefore always two time points per survival assay are needed at least). Again, ratios between these time points were comparable, indicating that DCX knockdown did not affect survival of newly generated neurons.
In the SVZ/OB-system newborn neurons migrate long distances from their place of birth in the SVZ tangentially along the RMS to the OB [@pone.0062693-Lois1] where they start to migrate radially until they reach their final destination where they differentiate into granule neurons and periglomerular interneurons and integrate into the olfactory network [@pone.0062693-Lois1], [@pone.0062693-Petreanu1]. DCX knockdown in new neurons of the SVZ/OB system causes abnormal neuronal migration and a fate change of developing neurons in the postnatal olfactory bulb 5--7 days post electroporation of the DCX knockdown vector [@pone.0062693-Belvindrah1]. In DCX null mutants, where DCX is already ablated during development, it has been shown that DCX is needed to maintain the bipolar shape and nuclear translocation during migration [@pone.0062693-Koizumi1]. To determine if acute knockdown of DCX in adult born neurons affects their migration, CAG-GFP-miDCX(3) and CAG-RFP-miCtr-IRES-RFP retroviruses were injected into the RMS of 8 week old female wildtype mice ([Figure 8A](#pone-0062693-g008){ref-type="fig"}). Like in the DG, DCX is efficiently knocked down in the SVZ ([Figure 8B](#pone-0062693-g008){ref-type="fig"}). miDCX(3) and control retrovirus transduced cells did not show apparent differences regarding their position in the RMS/OB system neither at 5 dpi ([Figure 8](#pone-0062693-g008){ref-type="fig"}) nor at 10 dpi ([Figures 9](#pone-0062693-g009){ref-type="fig"} and [10](#pone-0062693-g010){ref-type="fig"}). Moreover, RFP~only~ control cells and the green or yellow DCX ablated cells were migrating within the DCX-positive rostral migratory stream ([Figures 8](#pone-0062693-g008){ref-type="fig"}, [9](#pone-0062693-g009){ref-type="fig"}, [10](#pone-0062693-g010){ref-type="fig"}). Positioning of DCX-ablated cells relative to control cells was also investigated by determining the ratio of yellow cells to all red cells (yellow+RFP~only~ cells) in three different parts of the RMS/OB system: rostral part of the RMS (RMS1), middle part of the RMS (RMS2), and the core region of the OB where new neurons switch their migratory mode to radial migration ([Figure 10](#pone-0062693-g010){ref-type="fig"}). Ratios were comparable between different regions (RMS1∶0.47±0.09; RMS2∶0.55±0.02, OB: 0.48±0.04) confirming that DCX-ablated cells did not assume differential positioning along the dorsoventral axis of the RMS.
{#pone-0062693-g008}
{#pone-0062693-g009}
{#pone-0062693-g010}
Finally, the effects of DCX overexpression on migration were investigated during early phases of adult SVZ/OB neurogenesis. Eight weeks old female mice were injected into the RMS with CAG-RFP and CAG-DCX-3xFLAG-IRES-GFP retroviruses and analysed 10 days later. Similar to the knockdown experiments, DCX-overexpressing cells and control transduced cells were found in comparable positions along the RMS/OB system (Ratios RMS1∶0.59±0.12; RMS2∶0.68±0.02, OB: 0.59±0.09) ([Figure 11](#pone-0062693-g011){ref-type="fig"}).
{#pone-0062693-g011}
Thus, our data indicate that neither knockdown nor overexpression of DCX significantly alter migration in the RMS/OB system at the time points investigated.
Discussion {#s4}
==========
The microtubule binding protein DCX is expressed by migrating neuroblasts and immature neurons and is widely used as a stage-specific marker of adult neurogenesis but little is known about its specific function in adult neurogenesis. In this study, we have investigated the consequences of overexpression or knockdown of DCX in adult born neurons of wildtype mice. We did not observe any significant effect on morphological maturation of newly generated DG neurons or on migration of new neurons in either adult neurogenic system at the time points investigated.
For *in vivo* knockdown of DCX we have generated the retroviral CAG-EmGFP-miDCX vector, expressing a miRNA against DCX under the control of an RNA polymerase II promoter. Here we show that our vector has the potential to efficiently knockdown DCX *in vivo* ([Figures 5](#pone-0062693-g005){ref-type="fig"}, [6C](#pone-0062693-g006){ref-type="fig"}). By generation of this retroviral miRNA-vector we have created a useful tool for *in vivo* expression of miRNA and knockdown of other proteins of interest.
In DCX knockout mice it was shown that migrating interneurons show branching defects [@pone.0062693-Kappeler1]. Here we were investigating if overexpression or knockdown of DCX affected morphological maturation of adult born neurons. Morphological analysis of 10--17 dpi cells was precluded by unequal RFP distribution within the cells. Yet, with an average length of ∼1300 µm and an average number of ∼8 branch points per 10 µm in RFP~only~ control cells and DCX overexpressing or knockdown cells, no morphological defects in the dendritic compartment were apparent at 28 dpi. The relatively low expression levels of GFP precluded reliable axonal tracing of DCX overexpressing or knockdown cells. Thus, it remains to be determined whether DCX gain- and loss-of-function affects development of the axonal compartment.
We found no evidence that DCX gain- and loss-of-function altered migration of new DG neurons from the SGZ towards the granule cell layer. In principle newborn DG cells have the possibility to migrate in three dimensions from their place of birth in the subgranular zone into the granular layer ([Figure 2D](#pone-0062693-g002){ref-type="fig"}): medial/lateral (red arrows), septal/temporal (green arrows) and vertically to these two directions through the granular layer (blue arrow). As there is presently no live imaging method established so far to continuously follow the migration of a newborn cell in the dentate gyrus over several weeks, we cannot establish the medial/lateral, septal/temporal position where our retroviral labelled cells are born and cannot determine potential migration deficits along these axes. Hence, it is still formally possible, that DCX -- while not being required for migration from the SGZ towards the DG cell layer -- may be essential for medial/lateral and septal/temporal migration.
In the SVZ/OB-system newborn neurons migrate long distances from their place of birth in the SVZ tangentially along the RMS to the OB [@pone.0062693-Lois1] where they start to migrate radially until they reach their final destination and integrate into the olfactory network [@pone.0062693-Lois1], [@pone.0062693-Petreanu1]. DCX is needed for proper migration in the postnatal forebrain, which was shown in juvenile DCX mutant mice (age seven weeks) [@pone.0062693-Koizumi1] and by electroporation of shRNAs against DCX into the SVZ of P3/P4 wildtype mice [@pone.0062693-Belvindrah1], [@pone.0062693-Koizumi1]. In contrast to these reports, we did not observe migration defects of newborn neurons of the SVZ/OB-system after retrovirally miRNA-mediated DCX knockdown. In the present study control and manipulating retroviruses were co-injected into the same animal to precisely determine potential migration defects. This strategy allows to compare internal control cells with manipulated cells within one animal, thereby minimizing potential confounding factors, such as variability in the location of injections/electroporations between animals, variability in tissue preparation and generation of equivalent sections, and architectural differences in the SVZ/RMS. Thus, we consider it unlikely, that major migration phenotypes comparable to the phenotypes in the aforementioned studies remained undetected. We, however, cannot exclude that manipulation of DCX will affect later developmental steps of adult SVZ/OB neurogenesis.
The probably most significant difference between our study and the aforementioned studies is the timing of DCX ablation with regard to the age of the mice. In the present study, DCX is knocked down acutely in young adult mice and exclusively in adult born neurons. DCX null mutant mice lack DCX function already during embryonic development and thus do not allow to distinguish whether the observed migratory defects result from a direct function of DCX in adult-born neurons or result from defective CNS development. In the study of Belvindrah and colleagues [@pone.0062693-Belvindrah1], DCX is knocked down at P4. Early postnatal neurogenesis and adult neurogenesis have been found to generate morphologically and functionally distinct neuronal population for the OB neural circuit [@pone.0062693-Lemasson1]. It is therefore possible that the obvious difference between our findings and the findings of Belvindrah and colleagues reflect an age-dependent switch in the function of DCX in neuronal migration in the SVZ/OB system.
A second significant difference from the study of Belvindrah and colleagues is, that in our study, miDCX-retroviruses were injected into the RMS and not the SVZ. These targeting strategies may hit different population of newborn neurons, as dopaminergic periglomerular neurons are predominantly generated in the RMS [@pone.0062693-Hack1] while granule neurons are generated in the SVZ. Moreover, electroporation into the SVZ may have predominantly targeted early precursors of the SVZ prior to initiation of DCX expression, whereas our strategy may have targeted dividing neuroblasts in the RMS, which already initiated DCX expression and migration. Thus, it is possible that in the adult mouse brain, DCX may be required solely for initiation rather than sustenance of migration.
At present, it remains unclear, how adult-born neurons compensated for the extensive loss of DCX. We cannot exclude that ad libitum access to running wheels, which in the context of the low titer miRNA vector was necessary to increase the number of transduced cells, may enhance the expression of factors that render newly generated born neurons resistant to DCX loss-of-function. A second possibility would be that smallest amounts of remaining DCX expression following knockdown, which escape detection by immunstainings, may be sufficient for proper maturation of the cells. In our view, the most likely explanation would be that the doublecortin like kinase (DCLK), which is closely related to DCX, has redundant function to DCX during murine cortical [@pone.0062693-Deuel1], [@pone.0062693-Koizumi2], [@pone.0062693-Tuy1] and hippocampal development [@pone.0062693-Tanaka1] and is co-expressed with DCX in murine adult neurogenic niches [@pone.0062693-Saaltink1], might be sufficient to regulate proper development of newborn neurons following efficient DCX knockdown.
Generation of DCX conditional knockout mice will help to resolve these open questions and will greatly facilitate the future investigation of potentially hidden functions of DCX in adult neurogenesis.
We thank Katrin Wassmer for virus production, Fabian Gruhn and Marija Ram for excellent technical support and members of the Lie laboratory for helpful discussion and suggestions.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
[^2]: Conceived and designed the experiments: KM DCL. Performed the experiments: KM. Analyzed the data: KM. Contributed reagents/materials/analysis tools: DCL. Wrote the paper: KM DCL.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Opioid overdose deaths continue to increase in the United States, killing more than 42,000 people in 2016. The opioids detected in these cases, in increasing order, were methadone, natural and semi-synthetic opioids (e.g., oxycodone, hydrocodone), heroin and synthetic opioids (e.g., fentanyl, fentanyl-analogs). Synthetic opioids (excluding methadone) and heroin deaths specifically experienced a sharp increase from 2015 to 2016 (20 and 100%, respectively) (Seth et al., [@B102]). Fentanyl and its derivatives have been increasingly present as adulterants mainly in heroin, but also in other drugs such as cocaine and synthetic cannabinoids (Coopman and Cordonnier, [@B18]; Armenian et al., [@B5]), due to their ease of manufacturing and readily available precursors shipped from China (Armenian et al., [@B5]). In addition to being present in other drugs supply, fentanyl analogs have been also marketed as "research chemicals" and can easily be acquired over the internet. Due to their high potency and the increased use of heroin as an initiating opioid of abuse (8.7% in 2005 vs. 33.3% users in 2015) (Cicero et al., [@B17]; O\'Donnell et al., [@B78]), the number of opioid-related deaths have drastically increased in the recent years. Given that opioid novices have limited tolerance to opioids, a slight imprecision in dosing inherent in heroin use and/or the presence of potent fentanyl and analogs, can be fatal.
Fentanyl, its analogs (e.g., acetyl fentanyl, 3-methylfentanyl, alphamethylfentanyl, furanyl fentanyl) and the new generation synthetic opioids (e.g., AH-7921, U-47700, MT-45) have a chemical core structure totally different from morphine, a naturally occurring opioid from *Papaver somniferum* and reference compound of the opioids group; but all of them act on the opioid receptor (mu-receptor) reducing the intensity of pain and showing a high addiction potential. These opioid receptor agonists also induce dose-dependent respiratory depression (Pattinson, [@B87]), which is the main reason for their life-threatening risk (Ujváry et al., [@B112]). Fentanyl is approximately 200 times more potent than morphine, and the potencies of its analogs are variable, from 7 times more potent than morphine for butyrfentanyl and furanyl fentanyl, to more than 4,000 and 10,000 times for sufentanil and carfentanil, respectively (UNODC, [@B113]). The new generation opioids AH-7921 and MT-45 show similar potency to morphine (Brittain et al., [@B9]; EMCDDA, [@B30]), and U-47700 about 7.5 times more potent (Cheney et al., [@B15]).
Synthetic opioids are widely regulated by the United States Controlled Substances Act of 1970 (CSA) in order to control their use and distribution. As new compounds arise and threaten public safety, compounds can be emergency scheduled by the DEA to slow production and use of these harmful substances and aid in prosecution of drug diverters for a temporary period until the formal procedures have gone through (US Drug Enforcement Administration, [@B114]). Substances are classified into schedules in the CSA based on their safety, medicinal use and potential for abuse. A Schedule I substance is classified as having no currently accepted medical use and a high abuse potential. Examples of synthetic opioids in Schedule I include furanyl fentanyl, U-47700, acetyl fentanyl and 3-methyl fentanyl. Schedule II classified opioids have a high potential for abuse but have current medicinal uses like fentanyl which is used as an anesthetic and analgesic, as well as carfentanil, remifentanil and sufentanil (US Drug Enforcement Administration, [@B114]). Most recently, the DEA issued a temporary scheduling order for all fentanyl --related substances (to include all analog modifications) in February of 2018, which cover all substances that were not already classified into Schedule I of the CSA in an aggressive attempt to regulate the manufacture and subsequent trafficking of new synthetic opioids into the United States (Drug Enforcement Administration, [@B24]).
The expansion of these new synthetic opioids constitutes an important challenge in forensic toxicology. First of all, most of these substances are not detected in the routine screening and confirmation methods in the laboratory. Also, due to the low doses employed of these highly potent drugs, the concentrations expected in the biological samples are in the low ng to pg/mL or ng to pg/g range, requiring extremely sensitive methods of analysis. Recently, Marchei et al. ([@B62]) and Liu et al. ([@B59]) reviewed the currently available screening and confirmation methods of new synthetic opioids in biological and non-biological samples. As indicated by Marchei et al. ([@B62]), gas chromatography combined with mass spectrometry (GC-MS) and more frequently liquid chromatography tandem mass spectrometry (LC-MSMS) are the most common techniques due to their sensitivity and specificity. However, given the continued development of new derivatives, the major disadvantage of these target techniques, which employ quadrupole mass spectrometers, is that are limited by the reference standards available. High resolution mass spectrometry (time-of-flight, orbitrap) offers potential advantages to identify unknown compounds without the availability of a reference standard, but this technology is not readily available in most forensic laboratories (Marchei et al., [@B62]).
Regarding biological samples, most of these methods have been developed in blood or urine, and the target analytes are the parent compounds and rarely the metabolites (Marchei et al., [@B62]). In postmortem toxicology, other biological specimens such as vitreous humor, liver and brain are commonly analyzed. Unfortunately, fully validated methods for the determination of synthetic opioids in these specimens are lacking in the literature. This is in part due to the constant changes in illicit synthetic opioids being identified and laboratories being unable to justify the extensive time and cost associated with fully validating a method for a drug that may only be present in cases for a short time. Analytical methods in forensic toxicology are commonly validated in the corresponding biological sample following the guidelines published by the Scientific Working Group in Forensic Toxicology (SWGTOX) (Scientific Working Group for Forensic Toxicology, [@B100]) to guarantee the analytical quality of the measured concentrations. The analysis of metabolites in the different biological matrices may improve the interpretation of the results, extending the detection window and indicating if it was an acute or a delayed-death evaluating the metabolite-to-parent ratios. Recent publications about the identification of new metabolites of the synthetic opioids are available (Wohlfarth et al., [@B121]; Steuer et al., [@B108]; Watanabe et al., [@B119]; Krotulski et al., [@B56]); however, its application to authentic samples is still scarce (Poklis et al., [@B93]; Staeheli et al., [@B106]; Martucci et al., [@B65]; Allibe et al., [@B2]).
Besides the analytical challenges associated with synthetic opioids, due to the scarcity of available postmortem data, the interpretation of the results is extremely difficult. Conducting postmortem toxicology interpretation provides a number of very significant challenges to the forensic toxicologist. The range of postmortem specimens (blood, urine, vitreous humor, tissues, hair), the lack of reference databases, the presence of other substances (e.g., benzodiazepines, alcohol), opioid tolerance, and postmortem phenomena (postmortem redistribution and drug instability) complicates the interpretation of the analytical findings. Pichini et al. ([@B90]) and Zawilska ([@B123]) discussed non-fatal and lethal intoxications involving the new synthetic opioids, and Drummer ([@B25]) focused his review on fatalities due to these compounds.
The present review is focused on fentanyl derivatives and new generation opioids due to the limited knowledge concerning these substances and their high prevalence in opioid-overdose related cases. This work complements the previously published literature reviewing the current knowledge of postmortem toxicology of synthetic opioids and the chemical and pharmacological factors that may affect drug concentrations in the different matrices and therefore, their interpretation in postmortem samples. These factors include key chemical properties, essential pharmacokinetics parameters, postmortem redistribution and stability data in postmortem samples. All of these data are critically compared to postmortem data of natural opioids (morphine), semi-synthetic (oxycodone, hydrocodone, hydromorphone, and oxymorphone), and synthetic opioids (methadone and buprenorphine). The interpretation of drug intoxication in death investigation is based on the available published literature. This review serves to facilitate the evaluation of cases where synthetic opioids may be implicated in a fatality through the review of peer reviewed published case reports and research articles.
Methods {#s2}
=======
PubMed, Scopus and Google Scholar were searched for appropriate articles. Forensic case-reports and research articles of natural, semi-synthetic and synthetic opioids were reviewed up to May 2018. All articles were manually reviewed for content and references in each manuscript were further queried. Included articles were limited to peer-reviewed journals indexed by the Institute for Scientific Information (ISI) and published in English. Chemical properties were retrieved from the public databases PubChem (<https://pubchem.ncbi.nlm.nih.gov/>) and DrugBank (<https://www.drugbank.ca/drugs>).
Chemical and pharmacological properties {#s3}
=======================================
The chemical structure of the diverse synthetic opioids, including fentanyl and analogs, differs significally from the chemical structure of morphine and semi-synthetic opioids (e.g., oxycodone, hydrocodone, buprenorphine). Figure [1](#F1){ref-type="fig"} summarizes the chemical structure of selected classic opioids. Fentanyl is a piperidinyl derivative with moieties on the nitrogen and the 4-position (Figure [2](#F2){ref-type="fig"}). The different fentanyl derivatives show substitutions on the propionyl moiety (e.g., acetylfentanyl, acrylfentanyl, butyrfentanyl, furanyl fentanyl), phenethyl moiety (e.g., ohmefentanyl), N-phenyl ring (e.g., ocfentanil, 4-methoxy-butyrylfentanyl) and/or at the 4-piperidinyl-position (e.g., carfentanil). The chemical structures of the new generation synthetic opioids (AH-7921, U-47700, MT-45) are different from fentanyl. Figure [3](#F3){ref-type="fig"} shows 20 fentanyl derivatives and 3 new generation synthetic opioids not related to fentanyl. Due to the close chemical structure among fentanyl derivatives, some compounds, such as cyclopropyl fentanyl and crotonyl fentanyl, have exactly the same molecular formula, and therefore, the same molecular weight. As a consequence of this, special attention has to be paid in the development of the analytical methods for the determination of these compounds, and a complete chromatographic separation is required to guarantee their correct identification by gas or liquid chromatography coupled to mass spectrometry (GC-MS, LC-MSMS).
{#F1}
{#F2}
{#F3}
Chemically, opioids are predominantly basic drugs with pKa ranging from 7.5 to 10.9. The chemical parameter log P, the decimal logarithm of the partition coefficient Kp, is a useful indication of the lipophilicity of a compound. In the case of opioids, log P range is wide, from 0.8 (oxymorphone) to 5 (methadone). Morphine and related compounds show the lowest log *P* values (0.8--2). Fentanyl and analogs show a log P between 1.5 and 4.3. The high lipophilicity of fentanyl and its analogs enables rapid diffusion through membranes, including the blood-brain barrier. Also, this lipophilicity along with their basic characteristics make these group of drugs candidates to undergo postmortem redistribution. Table [1](#T1){ref-type="table"} summarizes the molecular weight, pKa and log P of selected opioids.
######
Monoisotopic molecular weight (g/mol), pKa and Log P of selected natural, semi-synthetic and synthetic opioids.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Group** **Analyte** **Monoisotipic molecular weight (g/mol)** **pKa** **Log P**
---------------------------------------- -------------------------------------------------------- ------------------------------------------- --------- -----------
Natural and semi-synthetic opioids Morphine 285.136 8.2 0.9
Codeine 299.152 9.2 1.3
Hydrocodone 299.152 8.6 2.0
Hydromorphone 285.133 8.6 1.6
Oxycodone 315.147 8.2 1.0
Oxymorphone 301.131 10.9 0.8
Buprenorphine 467.300 7.5 4.5
Synthetic opioids Fentanyl 336.220 8.8 3.8
Methadone 309.445 9.1 5.0
Tramadol 263.189 9.2 2.5
Synthetic opioids-Fentanyl derivatives alphamethylacetylfentanyl; acetyl-alpha-methylfentanyl 336.220 9.01 3.5
Alfentanil 416.253 7.5 2.8
Butyryl fentanyl;\ 350.235 8.77 4.3
butyr fentanyl
Carfentanil 394.225 8.05 3.7
3-methylcarfentanil;\ 408.241 8.36 4.2
lofentanil
4-fluorofentanyl; 4-FBF;\ 354.210 8.74 4.0
para-fluorofentanyl
beta-hydroxyfentanyl 352.215 8.28 2.9
alpha-methylfentanyl 350.235 9 4.2
cis-3-methylfentanyl;\ 350.235 9.08 4.3
3-MF; mefentanyl
beta-hydroxy-3-methylfentanyl; ohmefentanyl 366.230 8.59 3.4
Remifentanil 376.199 7.51 1.5
Sufentanil 386.202 8.86 3.6
3-methylthiofentanyl 356.192 9.07 4.2
-------------------------------------------------------------------------------------------------------------------------------------------------------------------
Volume of distribution (Vd) and protein binding also help to predict the drugs that may exhibit postmortem redistribution. Vd is defined as the volume into which the total amount of the drug would have to be uniformly distributed to reach the concentrations measured in plasma. It is expressed in L/kg of body weight (amount of drug in the body divided by the plasma drug concentration). Drugs highly bound to plasma proteins but not to tissue components would be expected to have a small Vd, while those drugs which distribute into muscle, adipose tissue and other intracellular components will have a high Vd. Drugs with a Vd greater than 3 L/kg are considered to have a greater potential to undergo postmortem redistribution. Table [2](#T2){ref-type="table"} summarizes the Vd and protein binding data currently available for selected opioids.
######
Critical pharmacological properties in postmortem toxicology, volume of distributon (Vd), protein bining and potency relative to morphine, of selected natural, semi-synthetic and synthetic opioids.
**Group** **Analyte** **Vd (L/kg)** **Protein binding (%)** **Potency relative to morphine** **References**
------------------------------------------- ---------------------------------------- --------------- ------------------------- ---------------------------------- ----------------------------------------------------------------
Natural and semi-synthetic opioids Morphine 1--6 30--40 1 Baselt, [@B7]
Codeine 2.5--3.5 7--25 0.3 Baselt, [@B7]
Hydrocodone 3.3--4.7 19--45 0.5--1 Patanwala et al., [@B86]; Baselt, [@B7]
Hydromorphone 2.9 20 5--10 Bruera et al., [@B11]; Patanwala et al., [@B86]; Baselt, [@B7]
Oxycodone 2.6 45 1 Patanwala et al., [@B86]; Al-Asmari et al., [@B1]
Oxymorphone 3 10--12 10 Patanwala et al., [@B86]; Smith, [@B104]
Buprenorphine 3--5 96 40 Dahan et al., [@B23]
Synthetic opioids Fentanyl 3--8 80--85 224 Jumbelic, [@B50]
Methadone 1--8 85--90 3--5 Patanwala et al., [@B86]; Baselt, [@B7]
Tramadol 3 20 0.1 Christoph et al., [@B16]; Oertel et al., [@B79]
Synthetic opioids-Fentanyl derivatives Acetylfentanyl NA NA 15 Higashikawa and Suzuki, [@B45]
Acrylfentanil NA NA 170 Ujváry et al., [@B112]
Alfentanil 0.4--1 92 72 Vardanyan and Hruby, [@B116]
Butyryl fentanyl; butyr fentanyl NA NA 7 Higashikawa and Suzuki, [@B45]
Isobutyrylfentanyl NA NA 1.3--6.9 Higashikawa and Suzuki, [@B45]
Carfentanil NA NA 10,000 Van Bever et al., [@B115]
Furanyl fentanyl NA NA 7 Higashikawa and Suzuki, [@B45]
alpha-methylfentanyl NA NA 56.9 Higashikawa and Suzuki, [@B45]
cis-3-methylfentanyl; 3-MF; mefentanyl NA NA 6000 Higashikawa and Suzuki, [@B45]
Remifentanil 0.35 70 220 Wax et al., [@B120]
Sufentanil NA NA 4,520 Niemegeers et al., [@B77]
Synthetic opioids-Not related to fentanyl AH-7921 NA NA 1 Hayes and Tyers, [@B44]
U-47700 NA NA 7.5 Cheney et al., [@B15]
MT-45 NA NA 1 EMCDDA, [@B30]
*NA, not available*.
One of the critical issues related to fentanyl, its derivatives and the new synthetic opioids, is the low concentrations expected in the biological samples (ng to pg/mL or ng to pg/g range) due to their high potency. However, the potency of these type of drugs varies considerably within this group, and therefore the concentrations reported show a wide range, depending on the drug. Table [2](#T2){ref-type="table"} summarizes the potencies relative to morphine for selected opioids.
Metabolism {#s4}
==========
The identification and quantification of metabolites in postmortem samples may improve the interpretation of the analytical results. The determination of metabolites may extend the window of detection, and also can be employed to calculate metabolite-to-parent ratios in urine and other biological samples to differentiate acute or delayed death. In certain cases, as it happens in morphine and buprenorphine, metabolites can be pharmacologically active. Although this type of information is limited in the case of the synthetic opioids, fentanyl, sufentanil, and alfentanil\'s metabolites are inactive in the opioid system (Schneider and Brune, [@B99]).
Although the utility of metabolite determination in biological samples is known, its application to authentic specimens is still scarce in the case of synthetic opioids due to the limited data available about their metabolism (Poklis et al., [@B93]; Staeheli et al., [@B106]; Martucci et al., [@B65]; Allibe et al., [@B2]). Recent publications about the identification of new metabolites of the synthetic opioids *in vivo* and *in vitro* are available (Wohlfarth et al., [@B121]; Steuer et al., [@B108]; Watanabe et al., [@B119]; Krotulski et al., [@B56]). While *in vitro* studies utilizing human liver hepatocytes or microsomes can identify multiple primary and secondary metabolites for a particular fentanyl derivative, actual human specimens typically show lower number and/or a different metabolite prevalence profile, so studies investigating the presence of the *in vitro* metabolites in authentic human samples are highly encouraged. Table [3](#T3){ref-type="table"} summarizes recent publications about the identification of new metabolites of synthetic opioids *in vitro* and *in vivo*.
######
*In vitro* and *in vivo* metabolism of synthetic opioids.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Compound** **Study type** **Matrix (species)** **Total \# phase I metabolites** **Major metabolites (decreasing order of relative intensity)** **Phase II metabolites** **Recommended target analytes in urine** **References**
-------------------------------------------- ---------------- --------------------------------------------------- ---------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ ----------------------------------------------------- --------------------------------------------------------------------------------------------------------------------------- ---------------------------
Acetyl Fentanyl\ *In vivo* Urine (humans) 6 -- Hydroxylated metabolite at phenylethyl ring-- Hydroxy-methoxy metabolite at phenylethyl ring Glucuronide of hydroxylated metabolites -- Hydroxylated metabolite at phenylethyl ring-- Hydroxy-methoxy metabolite at phenylethyl ring-- Acetyl fentanyl Melent\'ev et al., [@B70]
*In vitro* Pool human liver hepatocytes 7 -- N-dealkylated metabolite at the piperidine moiety-- Hydroxylated metabolites at the ethyl linker-- Dihydroxylation at phenylethyl ring Watanabe et al., [@B119]
*In vivo* Urine (human) 24 -- Hydroxy-methoxy metabolite at phenylethyl ring-- Hydroxy metabolite at the ethyl linker-- N-dealkylated metabolite at the piperidine moiety Glucuronides and sulfates of hydroxy-metabolites -- Hydroxy metabolite at the ethyl linker-- Hydroxy-methoxy metabolite at phenylethyl ring-- Acetyl fentanyl
*In vitro* Pluripotent stem cell-derived hepatocytes 6 -- N-dealkylated metabolite at the piperidine moiety-- Hydroxylated metabolite at phenylethyl ring-- Hydroxylated metabolites at the ethyl linker Kanamori et al., [@B51]
Acrylfentanyl *In vitro* Pool human liver hepatocytes 8 -- N-dealkylated metabolite at the piperidine moiety-- Hydroxylated metabolite at the piperidine moiety-- Hydroxylated metabolite at the ethyl linker Watanabe et al., [@B119]
*In vivo* Urine (human) 12 -- N-dealkylated metabolite the piperidine moiety-- Hydroxylated at the ethyl linker-- Dihydroxylated metabolite at the piperidine and at the ethyl linker-- Hydroxy-methoxy metabolite at phenylethyl ring Glucuronides of hydroxy-metabolites -- Hydroxylated at the ethyl linker-- Dihydroxylated metabolite at the piperidine and at the ethyl linker-- Acrylfentanyl
Butyrfentanyl\ *In vitro* Human liver microsomes 36 -- N-dealkylated metabolite-- Hydroxy-metabolite at butanamide chain-- Dihydroxy-metabolite at phenylethyl ring Steuer et al., [@B108]
*In vivo* Urine (human) -- Carboxy-metabolite at butanamide chain-- Hydroxy-metabolite at butanamide chain-- Carboxy at butanamide chain and hydroxy at phenylethyl ring metabolite Glucuronides of hydroxy-metabolites -- N-dealkylated metabolite-- Hydroxy-metabolite at butanamide chain-- Carboxy-metabolite at butanamide chain
Blood (human) -- Carboxy-metabolite at butanamide chain
Carfentanil\ *In vitro* Pool human liver hepatocytes 11 -- Monohydroxylated metabolite at of piperidine ring N-dealkylated metabolite Glucuronide of hydroxylated metabolite -- Monohydroxylated metabolite at of piperidine ring Feasel et al., [@B31]
Furanylfentanyl (Fu-F)\ *In vitro* Human hepatocytes Pooled human hepatocytes 13 -- Amide hydrolysis-- N-dealkylated metabolite-- Dihydrodiol metabolite at furan group Watanabe et al., [@B119]
*In vivo* Urine (human) 9 -- Amide hydrolysis-- Dihydrodiol metabolite at furan group-- Dihydrodiol at furan group and hydroxy at ethyl linker metabolite Glucuronide and sulfate of hydroxylated metabolites -- Dihydrodiol metabolite at furan group
*In vivo* Urine (human) -- Amide hydrolysis-- Dihydrodiol metabolite at furan group Sulfate metabolite of amide hydrolysis metabolite -- Sulfate metabolite of amide hydrolysis metabolite-- Dihydrodiol metabolite at furan group Goggin et al., [@B37]
*In vitro* Human liver microsomes 17 -- Despropnionyl fentanyl-- Monohydroxylated metabolite-- N-dealkylated metabolite Gaulier et al., [@B36]
*In vitro* HepaRG cell Line 17 -- Despropnionyl fentanyl-- N-dealkylated metabolite-- Dihydrodiol metabolite (at furan group) Glucuronide hydroxylated metabolite
4-Fluoro-isobutyrylfentanyl\ *In vitro* Pooled human hepatocytes 9 -- N-dealkylated metabolite of the piperidine moiety-- Monohydroxy metabolite at the piperidine ring or at the ethyl linker-- N-oxidation at the piperidine ring Watanabe et al., [@B119]
*In vivo* Urine (human) 13 -- N-dealkylated metabolite of the piperidine moiety-- Monohydroxy metabolite at the piperidine ring or at the ethyl linker-- Hydroxymethoxy metabolite at phenylethyl ring Glucuronide hydroxylated metabolites -- Monohydroxy metabolite at the piperidine ring or at the ethyl linker-- Hydroxymethoxy metabolite at phenylethyl ring
Isofentanyl\ *In vitro* Urine (rats) 11 -- N-dealkylation followed by hydroxylation of the alkyl and aryl moiety-- Hydroxylation of the propanamide side chain followed by oxidation to the carboxylic acid-- Hydroxylation of the benzyl moiety followed by methylation-- N-oxidation Glucuronides of hydroxy metabolites -- N-dealkylated metabolite Meyer et al., [@B72]
3-methylfentanyl\ *In vivo* Urine (rats) 9 /5 -- N-dealkylation followed by hydroxylation of the alkyl and aryl moiety-- Hydroxylation of the propanamide side chain followed by oxidation to the carboxylic acid-- Hydroxylation of the benzyl moiety followed by methylation Glucuronides of hydroxy metabolites -- N-dealkylated metabolite Meyer et al., [@B72]
Ocfentanil (OcF)\ *In vitro* Human liver microsomes 3 -- O-desmethyl metabolite-- Monohydroxylated metabolite at phenylethyl ring-- O-desmethyl metabolite hydroxylated at phenylethyl ring Glucuronide of O-desmethylated metabolite Allibe et al., [@B2]
*In vivo* -- Blood (human, *n* = 1)-- Bile (human, *n =* 1) 3 -- O-desmethyl metabolite-- Monohydroxylated metabolite at phenylethyl ring-- O-desmethyl metabolite hydroxylated at phenylethyl ring -- O-desmethylated-metabolite
AH-7921\ *In vitro* Human hepatocytes 11 -- N-demethyl metabolite-- N-dis-demethyl metabolite-- N-demethyl metabolite hydroxylated at cyclohexyl Glucuronide demethylated metabolite Wohlfarth et al., [@B121]
*In vivo* Urine (human) 10 -- N-demethylation-- N-dis-demethyl metabolite Glucuronide demethylated metabolite -- N-demethylation-- N-dis-demethyl metabolite
MT-45\ *In vitro* Rat hepatocytes 10 -- Hydroxy metabolite-- Dihydroxy metabolite-- 1-cyclohexyl-piperazine Glucuronides of hydroxy metabolites Montesano et al., [@B74]
*In vivo* Urine (rat) 10 -- Hydroxy metabolite-- Dihydroxy metabolite-- 1-cyclohexyl-piperazine-- OH-1-cyclohexyl-piperazine Glucuronides of hydroxy metabolites -- Hydroxy metabolite-- Dihydroxy metabolite
U-47700\ *In vitro* Human liver microsomes 4 -- N-desmethyl-U-47700-- N,N-didesmethyl-U-47700-- N-desmethyl-hydroxy-U-47700-- N,N-didesmethyl-hydroxy-U-47700 Krotulski et al., [@B56]
*In vivo* Urine (human, *n =* 5) 5 -- N-desmethyl-U-47700-- N,N-didesmethyl-U-47700-- N-desmethyl-hydroxy-U-47700-- N,N-didesmethyl-hydroxy-U-47700-- N,N-didesmethyl-N-desmethyl-U-47700 -- N-desmethyl-U-47700-- N,N-didesmethyl-U-47700
U-49900\ *In vitro* Human liver microsomes 5 -- N-desethyl-U-49900-- N,N-didesethyl-U-49900;-- N,N-didesethyl-N-desmethyl-U-49900-- N-desethyl-hydroxyo-U-49900-- N-desethyl-N-desmethyl-U-49900 Krotulski et al., [@B56]
*In vivo* Urine (human, *n =* 5) 5 -- N-desethyl-U-49900-- N,N-didesethyl-U-49900-- N,N-didesethyl-N-desmethyl-U-49900-- N-desethyl-hydroxyo-U-49900-- N-desethyl-N-desmethyl-U-49900 -- N,N-didesethyl-N-desmethyl-U-49900
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Fentanyl-derivatives metabolism studies showed similarities and differences from fentanyl metabolism pathways and rates. These different metabolic pathways observed for certain derivatives, demonstrate the need to perform individual metabolism studies for each new compound. In the case of fentanyl, only less than 8% of fentanyl is excreted unchanged. Approximately 85% of the dose is excreted within 72 h in feces and urine, the majority as metabolites mainly as norfentanyl generated by N-dealkylation at the piperidine nitrogen (McClain and Hug, [@B66]). Minor fentanyl metabolites are despropionylfentanyl, also known as 4-ANPP, which is formed by carboxamide hydrolysis, and hydroxyfentanyl and hydroxynorfentanyl metabolites, both hydroxylated at the propionyl moiety (Goromaru et al., [@B38]; Mahlke et al., [@B61]).
Several synthetic opioids follow a similar metabolic pathway to fentanyl. Alfentanil undergoes piperidine N-dealkylation to noralfentanil (Meuldermans et al., [@B71]). Major alpha-methylfentanyl metabolites in rats were norfentanyl and hydroxypropionyl norfentanyl metabolites, exactly as fentanyl (Sato et al., [@B98]). Meyer et al. ([@B72]) investigated the metabolism in rats of isofentanyl and 3-methyl fentanyl. After the administration of suspected recreational doses, the parent drugs could not be detected in urine and their common nor-metabolite was the predominant compound.
Patton et al. ([@B88]) detected high concentrations of acetylfentanyl and acetyl norfentanyl (\>16,500 ng/mL, 180 min post-dose) in urine samples from rats treated with a toxic dose of acetylfentanyl (3 mg/kg); however, Melent\'ev et al. ([@B70]), showed that the main pathway of the biotransformation of acetylfentanyl was hydroxylation by the phenylethyl moiety rather than N-dealkylation in authentic human samples. Melent\'ev et al. ([@B70]) and Watanabe et al. ([@B119]) recommended as target analytes in human urine hydroxy-methoxy at phenylethyl moiety and monohydroxylated metabolites, although the reported hydroxylation position in both publications was different. In both publications, the parent compound acetylfentanyl was highly abundant in urine samples, indicating that the parent drug is a suitable target.
Acrylfentanyl underwent N-dealkylation at the piperidine nitrogen producing the major nor-metabolite (Watanabe et al., [@B119]). The parent compound was also detected at high concentrations in urine samples. N-Dealkylation and monohydroxylation of the piperidine ring were the dominant metabolic pathways for carfentanil *in vitro* (Feasel et al., [@B31]). In that study, the authors observed a slow parent depletion in the hepatocytes. For 4-fluoroisobutyrylfentanyl the main metabolites identified in urine were the nor-metabolite, and monohydroxy metabolites at the piperidine ring or at the ethyl linker, as well as the parent compound. In terms of specificity, Watanabe et al., recommended as target compounds in urine the monohydroxy metabolites and the hydroxymethoxy metabolite (Watanabe et al., [@B119]).
In the case of butyrfentanyl, hydroxylation of the butanamide side chain followed by subsequent oxidation to the carboxylic acid represented the major metabolic step (Steuer et al., [@B108]). Although the norbutyrfentanyl was not among the most abundant metabolites in human samples in that study, the authors suggested its inclusion as a recommended target analyte because it showed a high intensity in the *in vitro* experiment. In authentic postmortem blood and urine samples, butyrfentanyl was still detected at 66 and 1,000 ng/mL, respectively.
Furanylfentanyl contains a furan group that affects its metabolic profile. This structure seemed to favor the amide hydrolysis, which is the main metabolite *in vitro* and *in vivo* (Watanabe et al., [@B119]). In terms of specificity of the target metabolites, Watanabe et al. ([@B119]) recommended the dihydrodiol-metabolite and Goggin et al. ([@B37]) recommended the same metabolite, as well as the sulfate of the metabolite that results from the amide hydrolysis. As it happened with butyrfentanyl (Steuer et al., [@B108]), the hepatocyte experiment also suggested high prevalence for the nor-metabolite, which was not significantly present in the authentic urine samples, illustrating the need to analyze human specimens. Furanylfentanyl parent compound was detected in authentic urine samples. For ocfentanyl, the predominant metabolite detected in blood, along with the parent drug, was the O-desmethylated metabolite (Allibe et al., [@B2]).
In the case of the new synthetic opioids not structurally related to fentanyl, different metabolic pathways has been reported. For AH-7921, the preferred metabolic sites were the amine function and the cyclohexyl ring. The two most dominant metabolites after hepatocyte incubation (also identified in a urine case specimen) were desmethyl and di-desmethyl AH-7921. Together with the glucuronidated metabolites, they were recommended as suitable analytical targets for documenting AH-7921 intake (Wohlfarth et al., [@B121]). In the case of MT-45, Montesano et al reported hydroxy-MT-45-glucuronide and di-hydroxy-MT-45-glucuronide as the most abundant metabolites in rat urine, while the parent drug was found at concentrations \<10 ng/mL after 300 min (Montesano et al., [@B74]). Although similar in chemical structure, U-47700 and U-49900 showed specific metabolites. N-Desmethyl-U-47700 was identified as the major metabolite in human urine specimens, and N,N-Didesethyl-N-desmethyl-U-49900 was identified as the most abundant metabolite present. Unlike U-47700 specimens, U-49900 was detected in low abundance in urine samples (Krotulski et al., [@B56]).
As indicated by Watanabe et al. ([@B119]), the target metabolites should generally be abundant, specific of the parent drug, and prevalent in most, if not all, case samples. Given the strong structural similarities among emerging designer fentanyls, many of them are coincidentally biotransformed to the exact same metabolite. This fact can make identification of the specific parent drug in a case difficult. The ability to identify minor metabolites that are unique and specific to the parent drug is therefore of considerable importance. 4-ANPP can be formed by fentanyl and other different fentanyl analogs metabolism, and it is also a precursor contaminant found in seized illicit fentanyl and analogs, so its presence is not particularly diagnostic. Other common metabolites are: acetylnorfentanyl from acetyl-alpha-methylfentanyl or acetylfentanyl (Watanabe et al., [@B119]); norfentanyl from fentanyl, beta-hydroxythiofentanyl and alpha-methyl-fentanyl (Sato et al., [@B98]); norcarfentanil from carfentanil, sufentanil and remifentanil (Feasel et al., [@B31]). 3,4-dichloro-N-(2-aminocyclohexyl)-N-methyl-benzamide is a common metabolite of U-47700 and U-49900, but it is not a major metabolite in urine for either compound (Krotulski et al., [@B56]).
Another important aspect of the metabolism is the identification of the enzymes involved. Pharmacokinetic interactions may be produced due to the presence of other substances metabolized by the same enzymes, ultimately affecting the drug blood concentrations. Fentanyl, sufentanyl and alfentanil are mainly metabolized by CYP 3A4 (Feierman and Lasker, [@B32]; Guitton et al., [@B41]). Steuer et al., identified CYP 3A4 and CYP 2D6 as the isoforms involved in the metabolism of butyrfentanyl (Steuer et al., [@B108]). Meyer et al., reported that CYP 3A4, CYP 3A5 and CYP 2C19 are involved in the metabolism of 3-methylfentanyl and isofentanyl and, in the case of isofentanyl, additionally CYP2D6 (Meyer et al., [@B72]). Remifentanil is the only family member of this class found to be \~95% metabolized in the blood and tissues by non-CYP enzymes, probably due to an easily accessible ester group allowing rapid hydrolysis by circulating blood esterases (Bürkle et al., [@B12]).
Concentrations in postmortem specimens and other findings {#s5}
=========================================================
The concentrations determined in postmortem specimens varied considerably depending on the type of synthetic opioid detected. Derivatives with potencies relative to morphine of more than 170, showed concentrations in femoral blood in the low ng/mL or pg/mL range, while those derivatives with potencies similar to morphine showed concentrations of hundreds, and even thousands, of ng/mL. An exception happens with furanyl fentanyl, which is seven times more potent than morphine (Higashikawa and Suzuki, [@B45]), but the reported femoral concentrations were less than 50 ng/mL. Typical morphine postmortem concentrations in blood in fatalities are from 200 to 2,300 ng/mL, for methadone 400 to 1,800 ng/mL, for buprenorphine 1.1--29 ng/mL and norbuprenorphine (active metabolite) 0.2--13 ng/mL (Baselt, [@B7]), and for oxymorphone 23--554 ng/mL (Crum et al., [@B21]). The potency of the different drugs affects their lethal levels, but other important issues, such as the presence of other CNS depressant drugs, and developed opioids tolerance, have to be taken into account in the interpretation of the concentrations. The derivative with the highest number of published cases was acrylfetanyl, and with the lowest MT-45. Table [4](#T4){ref-type="table"} summarizes the concentrations of the parent drugs found in case reports and articles where overdose due to a specific opioid was the cause of death.
######
Postmortem concentrations in different biological samples for synthetic opioids (median, range, number of cases).
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Analyte** **Blood (ng/mL)** **Vitreous humor (ng/mL)** **Brain (ng/g)** **Liver (ng/g)** **Urine (ng/mL)**
------------------------- ------------------- ---------------------------- ------------------ ------------------------ ------------------- ------------- --------------------- ----------------------
3-Methylfentanyl -- -- -- 0.4 (0.3--0.9) *n =* 3 -- -- -- --
4--fluorobutyr fentanyl -- -- -- 91--112\ -- 248\ 902\ 200\
*n =* 2 *n =* 1 *n =* 1 *n =* 1
Acetylfentanyl 223.5 (16--600)\ 270 (170--2,100)\ 220\ -- 140--240\ 620\ 1,000--1,100\ 2,660 (240--3,420)\
*n =* 12 *n =* 11 *n =* 1 *n =* 2 *n =* 1 *n =* 2 *n =* 4
Acrylfentanyl 0.2 (0.01--5)\ -- -- -- -- -- -- --
*n =* 42
Butyryl fentanyl 99 (66--145.2)\ 60.5 (39--220)\ -- -- 32\ 93--200\ 41--57\ 64\
*n =* 3 *n =* 3 *n =* 1 *n =* 2 *n =* 2 *n =* 1
Carfentanil 0.2 (0.01--0.5)\ 0.1--0.2\ 0.03\ -- -- -- -- --
*n =* 9 *n =* 2 *n =* 1
Fentanyl 11 (1--60)\ 13 (1.8--139)\ -- 13 (2--383)\ 14.8 (8--20)\ 49\ 78 (5.8--16,983)\ 97 (2.9--1,200)\
*n =* 207 *n =* 81 *n =* 66 *n =* 4 *n =* 1 *n =* 99 *n =* 31
Furanyl fentanyl 2.7 (0.4--42.9)\ 2.8\ -- -- -- -- -- --
*n =* 13 *n =* 1
Ocfentanyl 9.1 (3.7--15.3)\ 23.3 (3.9--27.9)\ -- -- 12.5\ 37.9\ 31.2\ 6--480\
*n =* 3 n3 *n =* 1 *n =* 1 *n =* 1 *n =* 2
AH--7921 350 (30--9,100)\ 480--3,900\ -- -- 190\ 7,700\ 530--26,000\ 760--6,000\
*n =* 13 *n =* 2 *n =* 1 *n =* 1 *n =* 2 *n =* 2
MT--45 520--660\ 1,300\ -- -- 260\ -- 24,000\ 370\
*n =* 2 *n =* 1 *n =* 1 *n =* 1 *n =* 1
U--47700 358 (189--1,460)\ 691.5 (260--1,347)\ -- -- 130\ (0.9--380)\ 142.1 (3.1--1,700)\ 1620.5 (360--4,600)\
*n =* 12 *n =* 4 (90--170)\ *n =* 3 *n =* 4 *n =* 4
*n =* 2
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
*3-Methylfentanyl references: (Ojanperä et al., [@B81])*.
*4-fluorobutyr fentanyl references: (Rojkiewicz et al., [@B96])*.
*Acetylfentanyl references: (Pearson et al., [@B89]; Poklis et al., [@B93]; Cunningham et al., [@B22]; Fort et al., [@B34]; McIntyre et al., [@B68]; Takase et al., [@B110]; Yonemitsu et al., [@B122]; Dwyer et al., [@B27])*.
*Acrylfentanyl references: (Butler et al., [@B13]; Guerrieri et al., [@B40])*.
*Butyryl fentanyl references: (Poklis et al., [@B92]; Staeheli et al., [@B106])*.
*Carfentanil references: (Shanks and Behonick, [@B103]; Swanson et al., [@B109]; Hikin et al., [@B46])*.
*Fentanyl references: (Anderson and Muto, [@B3]; Kuhlman et al., [@B58]; Martin et al., [@B64]; Coopman et al., [@B20]; Biedrzycki et al., [@B8]; Carson et al., [@B14]; Krinsky et al., [@B53], [@B54]; Palamalai et al., [@B83]; Marinetti and Ehlers, [@B63]; McIntyre et al., [@B67]; Bakovic et al., [@B6]; Moore et al., [@B75]; Pearson et al., [@B89]; Poklis et al., [@B93]; Rodda et al., [@B94]; Dwyer et al., [@B27])*.
*Furanyl fentanyl references: (Mohr et al., [@B73]; Guerrieri et al., [@B39]; Martucci et al., [@B65]; Papsun et al., [@B84])*.
*Ocfentanyl references: (Coopman et al., [@B19]; Dussy et al., [@B26]; Allibe et al., [@B2])*.
*AH-7921 references: (Karinen et al., [@B52]; Kronstrand et al., [@B55]; Vorce et al., [@B117]; Fels et al., [@B33])*.
*MT-45 references: (Papsun et al., [@B85]; Fels et al., [@B33])*.
*U-47700 references: (Elliott et al., [@B29]; Mohr et al., [@B73]; Dziadosz et al., [@B28]; Papsun et al., [@B84]; Rohrig et al., [@B95])*.
In several cases, multiple synthetic opioids were detected. Acetylfentanyl and fentanyl were frequently found together (Pearson et al., [@B89]; Poklis et al., [@B93]; Dwyer et al., [@B27]). Other combinations were butyryl fentanyl and acetyl fentanyl (McIntyre et al., [@B69]; Poklis et al., [@B92]), or U-47700 (Mohr et al., [@B73]); furanyl fentanyl and acetyl fentanyl (Papsun et al., [@B84]), acryl fentanyl (Butler et al., [@B13]), butyryl fentanyl (Mohr et al., [@B73]), fentanyl (Guerrieri et al., [@B39]), or carfentanil (Shanks and Behonick, [@B103]); carfentanil and fentanyl (Shanks and Behonick, [@B103]); and tetrahydrofuran fentanyl and U-49900 (Krotulski et al., [@B57]). The femoral concentrations reported in those combination cases were frequently below the range of the concentrations summarized in Table [4](#T4){ref-type="table"}. Acetylfentanyl median and concentration range in multiple synthetic opioids cases were 9.4, 0.4--240 ng/mL (*n* = 15); acrylfentanyl 0.3 ng/mL (*n* = 1); butyrfentanyl 14.9, 0.3--58 ng/mL (*n* = 4); carfentanil 0.08, 0.05--0.1 ng/mL (*n* = 2); fentanyl 8.2, 1.1--38 ng/mL (*n* = 14); furanyl fentanyl 1.7, 0.6--6.1 ng/mL (*n* = 4) and U-47700 17 ng/mL (*n* = 1).
In all of the reports mentioned in Table [4](#T4){ref-type="table"} and above, synthetic opioids were commonly detected with other drugs, especially other CNS depressants, such as benzodiazepines, ethanol and other opioids. This combination may produce a pharmacodynamic interactions and increase the risk of respiratory depression. This possible interaction between opioids, alcohol and benzodiazepines has been previously described for other opioids, such as buprenorphine (Häkkinen et al., [@B42]; Seldén et al., [@B101]), methadone (Jones et al., [@B49]; Pilgrim et al., [@B91]; Nielsen et al., [@B76]), oxycodone (Ogle et al., [@B80]), and heroin (Thaulow et al., [@B111]). Among the reviewed cases positive for synthetic opioids other than fentanyl, 44 reported as cause of death intoxication due to multiple drugs and 77 intoxication mainly due to one specific opioid. The manner of death was predominantly accidental (*n* = 99), and suicides were reported in 7 cases.
Postmortem redistribution and stability {#s6}
=======================================
Postmortem changes in drug concentrations can happen via postmortem redistribution (PMR) from tissues of a higher to a lower concentration. Physicochemical and pharmacological properties of the analytes, such as pKa, log P, volume of distribution (Vd) and protein binding, may indicate drugs that experience this postmortem phenomenon. Lipophilic basic drugs with a Vd \> 3 L/kg, such as fentanyl, may undergo PMR. Fentanyl has been reported to undergo extensive PMR (Luckenbill et al., [@B60]; Olson et al., [@B82]; Palamalai et al., [@B83]; Brockbals et al., [@B10]). In the case of the synthetic opioids, limited data is currently available about PMR, and as well as information about pKa, log P and Vd (Tables [2](#T2){ref-type="table"}, [3](#T3){ref-type="table"}). Staeheli et al. ([@B106]) reported postmortem concentration changes of butyrfentanyl and metabolites, suggesting these compounds were prone to PMR. PMR reports about other synthetic opioids are not currently available.
Based on currently published case reports and articles, the cardiac blood-to-femoral blood and liver-to-femoral blood ratios were calculated to predict candidates of PMR. Results are summarized in Table [5](#T5){ref-type="table"}. Due to the scarce amount of data available (1--4 cases per analyte), no conclusions could be drawn. Synthetic opioids showed median cardiac-to-femoral ratios around 1, and a tendency to accumulate in the liver. Regarding the distribution to vitreous humor, it may be slow showing higher concentrations in blood. Other factors, such as time of death and sample collection, or rapid vs. delayed deaths, has not been taken into account in this analysis due to the limited data available.
######
Postmortem concentration ratios in different biological samples for synthetic opioids (median, range, number of cases).
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Analyte** **Cardiac-to-femoral** **Liver-to-femoral** **Vitreous humor-to-femoral** **References**
------------------ ------------------------ ---------------------- ------------------------------- ------------------------------------------------------------------------------------------------------------------------------------
Acetylfentanyl 1.2 (0.8--1.6)\ 3.8--5.7\ 0.6--0.9\ Cunningham et al., [@B22]; Fort et al., [@B34]; McIntyre et al., [@B68]; Yonemitsu et al., [@B122]
*n =* 4 *n =* 2 *n =* 2
Butyryl fentanyl 0.6 (0.4--2.2)\ 0.4--0.9\ 0.3\ Poklis et al., [@B92]; Staeheli et al., [@B106]
*n =* 3 *n =* 2 *n =* 1
Fentanyl (0.7--4.6) *n =* 54 6.6 (1.4--539.4)\ 1.5 (1.1--1.8)\ Anderson and Muto, [@B3]; Krinsky et al., [@B53], [@B54]; Palamalai et al., [@B83]; McIntyre et al., [@B67]; Bakovic et al., [@B6]
*n =* 75 *n =* 3
Furanyl fentanyl 1.5\ -- -- Martucci et al., [@B65]
*n =* 1
Ocfentanyl 1.5 (1.1--3.1)\ 2\ 0.8\ Coopman et al., [@B19]; Dussy et al., [@B26]; Allibe et al., [@B2]
*n =* 3 *n =* 1 *n =* 1
AH--7921 0.4--1.1\ 1.2--2.9\ 0.4\ Vorce et al., [@B117]; Fels et al., [@B33]
*n =* 2 *n =* 2 *n =* 1
MT-45 2\ 36.4\ 0.4\ Fels et al., [@B33]
*n =* 1 *n =* 1 *n =* 1
U-47700 1.5 (0.7--2.6)\ 0.4 (0.003--8.9)\ 0.2--0.9\ Dziadosz et al., [@B28]; Rohrig et al., [@B95]
*n =* 4 *n =* 4 *n =* 2
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
PMR is still a controversial issue for classic opioids. Hargrove and Molina ([@B43]) showed insignificant redistribution of morphine from central sites within 24 h after death in bodies kept at 4°C, while Staeheli et al. ([@B107]) observed a significant increase of morphine concentration, although these changes were not relevant for forensic interpretation. Morphine-derivatives, such us hydrocodone (Saitman et al., [@B97]), codeine (Frost et al., [@B35]), and oxycodone (Brockbals et al., [@B10]), are unlikely to undergo substantial PMR changes. More lipophilic opioids with higher Vd, like methadone (Jantos and Skopp, [@B48]; Holm and Linnet, [@B47]; Brockbals et al., [@B10]), may undergo PMR.
Several studies have been conducted to evaluate stability of fentanyl and some of its derivatives in fortified biological samples, such as blood, plasma and urine. Eleven fentanils (fentanyl, norfentanyl, carfentanil, norcarfentanil, sufentanil, norsufentanil, lofentanil, 3-methylfentanyl, alfa-methylfentanyl, ohmefentanyl, and remifentanil acid metabolite), were stable in urine samples stored at −20°C or below for at least 2 months. However, remifentanil in urine samples decreased by approximately 90% within 1 week at room temperature and by more than 50% in samples stored for 1 week at 4°C. Because of the instability of that analyte, the authors recommended to analyze the primary metabolite, remifentanil acid (Wang and Bernert, [@B118]). Fentanyl and its metabolites norfentanyl, despropionylfentanyl and hydroxynorfentanyl were stable in urine after 3 freeze-thaw cycles, and after storage at −20°C for 2 months (Mahlke et al., [@B61]).
Fentanyl, norfentanyl, acetyl fentanyl and acetyl norfentanyl spiked into whole blood were stable after three freeze-thaw cycles and at room temperature for 72 h (Poklis et al., [@B93]). No loss of fentanyl concentration could be observed after 3 months of storage at 4--8°C and −20°C in blood samples at 5 and 10 ng/mL (Andresen et al., [@B4]). However, another study showed fentanyl and its metabolites norfentanyl, despropionylfentanyl and hydroxynorfentanyl lose up to 51.6% after 3 freeze-thaw cycles, and fentanyl and despropionylfentanyl up to 34.8% after storage at −20°C for 2 months (Mahlke et al., [@B61]). Furanylfentanyl showed no significant degradation in blood samples at 5 and 10 ng/mL 48 h room temp and at 4°C 7 days (Guerrieri et al., [@B39]) and up to 30 days (Mohr et al., [@B73]).
Regarding the new synthetic opioids not related to fentanyl, U-47700 was stable in blood refrigerated for up to 30 days (Mohr et al., [@B73]). AH-7921 was found to be stable for at least 21 days in blood and plasma at room temperature (Soh and Elliot, [@B105]). In the case of MT-45, a loss of 50% was observed after 12 months of storage (Papsun et al., [@B85]). Further studies are necessary to evaluate the stability of the different synthetic opioids and metabolites, and in additional biological samples of forensic interest, such as vitreous humor and tissues.
Conclusion {#s7}
==========
We performed a critical review of the currently available literature to assist in the toxicological interpretation of synthetic opioids postmortem cases. Synthetic opioids constitute a heterogenous group of compounds related or not to fentanyl, mostly basic and lipophilic, with a wide range of potencies related to morphine, from 1 to 10,000. Research has been conducted in the investigation of metabolic pathways and identification of target metabolites of fentanyl derivatives and non-structurally related synthetic opioids, showing similarities and differences from fentanyl depending on the compound. Postmortem concentrations seemed to correlate with their potency, although the presence of other CNS depressants, such as ethanol and benzodiazepines has to be taken into account. Further research is guaranteed to investigate postmortem redistribution phenomena of this class of compounds, and stability issues in postmortem samples.
Author contributions {#s8}
====================
MC and GC contributed conception and design of the review. MC, RC, and JP searched, organized, reviewed and analyzed the case reports and research articles. MC wrote the first draft of the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.
Conflict of interest statement
------------------------------
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
[^1]: Edited by: Francesco Paolo Busardò, Università degli Studi di Roma La Sapienza, Italy
[^2]: Reviewed by: Luke N. Rodda, University of California, San Francisco, United States; Salahadin Abdi, University of Texas MD Anderson Cancer Center, United States
[^3]: This article was submitted to Neuropharmacology, a section of the journal Frontiers in Pharmacology
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Localized surface plasmon resonance (LSPR) biosensing is a label-free technique that enables biomolecule detection with nanoscale spatial resolution. Since its introduction, the technique has found a broad range of applications including studies of DNA-protein interactions \[[@CR1]\], toxins \[[@CR2]\], proteins \[[@CR3], [@CR4]\], and vesicles \[[@CR5]\]. The small size of the individual sensors, typically 40 to 150 nm in diameter, has the potential to enable sensor miniaturization to a scale unapproachable by the closely related planar technique of surface plasmon resonance (SPR) biosensing. Also unique to LSPR is that the nanoplasmonic resonance condition is satisfied in a simple reflected or transmitted light geometry, common to both microscopy and spectroscopy applications, whereas SPR excitation requires incident light that is totally internally reflected. As a result, LSPR spectroscopy and imaging (LSPRi) are increasingly being integrated into live-cell microscopy experiments for the detection of secreted proteins while simultaneously imaging the cells with more traditional techniques such as fluorescence and bright field \[[@CR6]--[@CR9]\].
Continued progress in the field of nanoplasmonic biosensing is highly dependent upon meeting two design challenges related to surface functionalization. First, the reduced sensor area limits the number of conjugated ligands, making orientation critical for maximizing the probability of analyte capture. The most common ligands in biosensing are full antibodies, surface conjugated via their lysine residues or N-terminal amino groups. The typical antibody will have 60--80 of such residues resulting in a range of surface orientations, many of which are not optimal for analyte detection \[[@CR10]\]. Second, experimental evidence and theoretical models show the nanostructures can exhibit a wide range of electromagnetic field decay lengths into the solution with some as short as 5 to 15 nm \[[@CR11], [@CR12]\]. Such nanoscale distances are largely filled when using intact IgG antibodies that have typical dimensions of 14.5 nm × 8.5 nm × 4.0 nm \[[@CR13]\]. As a result, LSPR optimization requires at a minimum highly oriented full antibodies as ligands \[[@CR14]\] and preferably lower molecular weight molecules which can be modified to enhance preferential orientation.
There are a number of classes of biomolecules which have been engineered with the goal of meeting these criteria including single chain antibodies (ScFvs), Fab fragments, and aptamers. Orientated ScFvs, which are roughly one fifth the size of full antibodies, were recently shown to enhance LSPR sensitivity relative to whole antibody counterparts \[[@CR15]\]. In general, however, attempts to orient ScFvs and Fab fragments for improved biosensor sensitivity have produced decidedly mixed results \[[@CR16]--[@CR22], [@CR10]\]. In addition, ScFvs and Fab fragments with molecular weights of approximately 27 and 50 kDa, respectively, are still relatively large ligands for many LSPR biosensing applications, leaving room for improved sensitivity if the size can be further reduced.
Single domain antibodies (sdAb), also called nanobodies, are the recombinant variable domain derived from heavy chain-only antibodies found in camelids, i.e., camels, llamas, and alpacas. Consisting of only a single domain endows sdAb with a number of attractive properties for biosensor applications. First, their small size (∼ 15 kDa or ∼ 1/10 the size of a full IgG) allows for facile production in yeast or bacteria, alone or as fusions with effector proteins. Second, both their small size and being a single domain provides them with remarkable thermostability, being able to refold and recover their binding ability following heat or chemical denaturation. Finally, sdAb are derived from true antibodies; thus, they possess the high affinity and specificity for which antibodies are renowned. SdAb, which recognize a plethora of targets, have been developed, and orientated sdAb have recently been demonstrated to improve SPR biosensor sensitivity \[[@CR23], [@CR24]\]. Our primary focus has been on the development of robust reagents for the detection of bio-threat agents \[[@CR25], [@CR26]\]. Of those reagents, the sdAb which bind ricin have been the most studied, and thus make well defined tools for testing the utility of sdAb in novel sensor applications \[[@CR27], [@CR28]\].
Here, we compare the sensitivity of LSPRi for orientated versus randomly surface-conjugated sdAb. Using anti-ricin sdAb, we first demonstrate the ability to modify the carboxyl terminus with positively charged peptide tails and rhizavidin fusion proteins for preferential surface orientation. The positively charged tail was designed to aid in surface orientation by means of an electrostatic interaction with a negatively charged surface. The rhizavidin fusion construct was designed to assist in orientation by binding to a biotinylated surface, preferentially directing the sdAb into the solution. Using SPR imaging (SPRi), we show that these modified sdAb have enhanced surface conjugation efficiencies and improved sensitivity to ricin relative to the non-orientated sdAb. We then measured the sdAb sensitivity to ricin on an LSPRi platform and compared the results to the biotin-neutravidin binding pair, a commonly used standard for biosensing applications and an apt comparison since biotin is readily orientated and ricin and neutravidin have similar molecular weights \[[@CR3], [@CR29]\]. Despite the fact that the sdAb ligands were 62-fold greater in mass than biotin, the measured sensitivities were statistically equivalent. We conclude that the relatively small size and the ability to readily modify sdAb for orientated surface conjugation make them optimal ligand candidates for LSPR imaging and spectroscopy applications. The current sensitivity study lays the groundwork for future work in determining the LSPRi limit of detection of each construct and multiplexed toxin-sensing applications.
Experimental {#Sec2}
============
The sequence and binding characteristics of the ricin-binding sdAb C8 and D12 have been previously described \[[@CR27], [@CR28]\]; both have sub-nM affinity for the toxin ricin. The clone D12f is a version of D12 in which an unpaired cysteine was mutated to a serine causing no change in sdAb affinity. The C8 sdAb, lacking the upper hinge region, was cloned into pET22b for protein expression. Protein was purified using immobilized metal affinity chromatography followed by size exclusion chromatography on an FPLC as previously described \[[@CR26], [@CR30]\]. Protein sequences of all sdAb constructs are detailed in the [Supplementary Material](#Sec5){ref-type="sec"}.
The construct referred to herein as C8-zip is a genetic fusion of the C8 sdAb with a positively charged leucine zipper described by Oshea et al. \[[@CR31]\]. DNA including a glycine-serine linker followed by the zipper sequences was synthesized by GenScript (Piscataway, NJ) and included flanking Not I and Xho I restriction endonuclease sites. We cloned the linker-zipper sequence into pET22b and then mobilized the C8 sdAb into the modified pET22b through the Nco I and Not I sites; DNA sequencing was used to confirm the constructs. Protein was produced from the periplasmic space in a protocol identical to that used to purify C8. This C8-Zip construct had initially been prepared to facilitate sdAb heterodimer formation \[[@CR32]\].
The D12f-rhiz construct is a genetic fusion of the D12f sdAb with the biotin binding protein rhizavidin (rhiz). Unlike our previously described sdAb-rhiz fusion construct, the D12f-rhiz does not include the upper hinge sequence between the D12f and rhiz. Protein production was as described for our previous sdAb-rhiz fusions \[[@CR33]\].
Details of the LSPR chip fabrication and the LSPRi measurement setup have been published previously \[[@CR34], [@CR29]\]. In short, square arrays of gold nanostructures were patterned on No. 1.5 glass coverslips using electron-beam nanolithography (Raith GmbH) to expose a bilayer resist structure consisting of polymethyl methacrylate and ethyl lactate methyl methacrylate copolymer purchased from Microchem Corp. Each array measured 6 μm × 6 μm and consisted of 400 evenly spaced nanostructures separated by a pitch of 300 nm. The bases of the nanostructures were circular in cross section with diameters of 70 ± 5 nm and the heights were 75 ± 2 nm, giving a plasmonic resonance peak centered at ∼ 635 nm in 10 mM phosphate buffered saline (Thermo Scientific). Protein binding to the Au surface creates a perturbation in the local index of refraction which induces a redshift in the resonance peak as well as enhanced scattering. In the LSPRi measurement, this response is manifested as an increase in the nanoplasmonic array's brightness as imaged by the camera. We have shown, using analytes such as neutravidin and IgG proteins, that this response can be calibrated to the fractional occupancy of surface-bound receptors. Commercial SPRi has sensors that are orders of magnitude larger in surface area but operates on a similar physical principle. When using the same functionalization chemistries, results from the two techniques can be correlated \[[@CR8], [@CR34]\].
The chip was loaded onto a custom-built microfluidic assembly consisting of a 300-μL chamber to which the analyte solution was introduced using a peristaltic pump from Instech Laboratories (P720). All imagery was acquired using Zeiss Zen software, an inverted Zeiss Axio Observer microscope, a 40X/1.4 NA objective, and a thermoelectrically cooled 16-bit sCMOS camera from Hamamatsu (Flash 4.0) operated in 2 × 2 binning mode. Images were collected in a reflected light geometry using Koehler illumination, a 100 W halogen lamp, and crossed polarizers to reduce the collection of light scattered from the substrate. Error bars in the LSPRi response represent the standard deviation from nine averaged arrays. Unless otherwise shown, the error bars are within the size of the data markers.
SPRi measurements were conducted with Bio-Rad's XPR36 protein analysis system, the sensors of which consisted of a gold thin film deposited atop a glass prism for excitation of surface plasmon polaritons by total internal reflection. Once inserted into the instrument, each chip is microfluidically arrayed into 36 measurement spots. This enabled a range of sdAb ligands to be functionalized on a single chip in contrast to the LSPRi setup in which each experiment consisted of only one ligand type. All LSPRi and SPRi measurements were conducted at 25 °C. Error bars in the SPR response are the standard deviation from averaging five measurement spots on the same chip. Unless otherwise shown, the error bars are within the size of the data markers.
For SPRi studies of sdAb, the Bio-Rad chips were cleaned down to the bare gold surface by RF plasma ashing at 40 W in 300 mTorr of a 5 % hydrogen, 95 % argon mixture, and then functionalized with a two-component self-assembled monolayer (SAM) by immersion for 18 h in an ethanolic-based thiol solution (0.5 mM), consisting of a 3:1 ratio of SH-(CH~2~)~8~-EG~3~-OH (SPO) to SH-(CH~2~)~11~-EG~3~-COOH (SPC) (Prochimia Surfaces Sp.) for C8 and C8-zip studies. The chips were then rinsed with EtOH and dried under flowing nitrogen gas. Activation of the SPC component consisted of introducing a 33 mM: 133 mM ratio of N-hydroxysulfosuccinimide (sulfo-NHS) and 1-ethyl-3-\[3-dimethylaminopropyl\]carbodiimide hydrochloride (EDC) from Thermo Scientific in ultrapure water (DDW; EMD Millipore) for 5 min at a flow rate of 30 μL/min. This was followed by a 5-min rinse with DDW and the introduction of the sdAb under the range of concentration and pH conditions described in the "[Results and Discussion](#Sec3){ref-type="sec"}" section. For the D12f-rhiz studies, a 2-mg/mL solution of (+)-biotinyl-3,6-dioxaoctanediamine (amine-PEG~2~-biotin) from Thermo Scientific was introduced over the activated surface at 30 μL/min for 5 min followed by a range of D12f-rhiz concentrations as described in the "[Results and Discussion](#Sec3){ref-type="sec"}" and [Supplementary Material](#Sec5){ref-type="sec"} sections. Finally, in all cases, unreacted --COOH groups were blocked by flowing 0.1 M ethanolamine (Bio-Rad) in PBS for 5 min at 30 μL/min. For the LSPR chips, this process was repeated except all solutions were manually drop coated and washed while inside the custom-built microfluidic assembly.
The identical surface cleaning and SAM layer formation protocol was applied for LSPRi biotin-neutravidin studies except that the two-component SAM layer consisted of a 3:1 ratio of SH-(CH~2~)~8~-EG~3~-OH (SPO) to SH-(CH~2~)~11~-EG~3~-NH~2~ (SPN) (Prochimia Surfaces Sp.). LSPR chips were drop coated with 100 μL of 2 mg/mL sulfo-NHS-biotin (Thermo Scientific) in PBS, incubated for 30 min at room temperature, rinsed with DDW, and dried in flowing nitrogen.
Results and Discussion {#Sec3}
======================
Three ligands were used in this study: an unmodified sdAb (C8), a genetic fusion of C8 with a positively charged peptide (C8-zip), and a genetic fusion of the D12f sdAb with the biotin binding protein rhizavidin (D12f-rhiz). Both the C8 and D12f sdAb that compose the ligand provide the recognition function of the constructs and are specific for ricin. They bind the same epitope on the toxin and exhibit nearly identically sub-nM affinities for their cognate antigen. Structurally, the binding loops of sdAb are located on the opposite side from their C-terminus, enabling the construction of fusions to facilitate directional immobilization. We hypothesized that the positively charged lysines on the C-terminus of C8-zip protein would maximize the potential for directional immobilization on the negatively charged --COOH surface, resulting in an orientation with the tail coupled to the surface and the binding region oriented into solution for optimal antigen capture. In a similar way, the D12f-rhiz enables directional immobilization but via the biotinylated surface \[[@CR33]\].
To optimize the plasmonic sensitivity to ricin, a range of ligand concentrations were introduced over the EDC/Sulfo-NHS activated SPR surface followed by a saturating concentration of ricin (100 nM). Fig. [1a](#Fig1){ref-type="fig"} shows the SPR response to C8-zip in which the concentration was varied from 0.4 to 30 μg/mL in PBS, pH 7.0. At the higher ligand concentrations, there were diminishing returns with regards to ricin-binding capacity (Fig. [1b](#Fig1){ref-type="fig"}) with the 30 μg/mL exposed surface binding only 8 % more than the surface coated at 10 μg/mL. Exceeding this 30 μg/mL value resulted in a rapid increase in non-specific binding and minimal ricin sensitivity enhancement. C8 and D12f-rhiz exhibited similar surface saturation concentrations, lying between 1 and 30 μg/mL ([Supplementary Material](#Sec5){ref-type="sec"}). Ligand concentration values such as the 30 μg/mL value for C8-zip were defined as optimal based on these criteria of maximizing sensitivity and minimizing non-specific binding. A control study was also conducted in which a 100-nM ricin solution was first incubated with 5 μM of C8 for 45 min to block the ricin-binding sites. When the blocked ricin was introduced over a C8-functionalized surface, no response was detectable, demonstrating levels of non-specific binding below the SPRi limit of detection (Fig. [1b](#Fig1){ref-type="fig"}).Fig. 1**a** SPR response to the conjugation of C8-zip sdAbs for a range of ligand concentrations followed by the introduction of 100 nM ricin in (**b**). The control study in **b** consisted of a 100-nM ricin solution incubated with 5 μM of C8 for 45 min to block the binding sites. It was then introduced over a 4.0-μg/mL C8 functionalized surface. The vertical *dashed line* separates the association phase (*left*) in which the ligand or analyte solution is flowing over the surface from the dissociation phase (*right*) in which buffer flows over the surface
A study was then conducted of SPR sensitivity to ricin at each ligand's optimal concentration. The 36-array microfluidic architecture of the XPR36 allowed for these measurements to be taken simultaneously alongside control lanes, thus eliminating uncertainties from chip-to-chip variations. C8-zip exhibited the highest response to 100 nM ricin, 320 % higher than that of C8 and 42 % higher than that of D12f-rhiz (Fig. [2a](#Fig2){ref-type="fig"}). In order to help isolate sensitivity improvements due to ligand orientation, we next introduced ricin over C8 and C8-zip surfaces prepared with similar ligand responses (1300 ± 120 RU) which is indicative of equivalent ligand surface densities. Figure [2b](#Fig2){ref-type="fig"} shows the SPR response to 100 nM ricin for these surface conjugations. The C8-zip surface again exhibited the highest response to 100 nM ricin, 140 % higher than that of C8.Fig. 2**a** SPR response to 100 nM ricin of optimally prepared C8-zip, C8, and D12f-rhiz surfaces. **b** SPR analyte response to 100 nM ricin following C8-zip and C8 ligand conjugations with responses of 1300 ± 120 RU. Measurements were conducted in parallel by multiplexing a single sensor chip to eliminate chip-to-chip variations. The vertical *dashed line* separates the association phase (*left*) in which the analyte solution is flowing over the surface from the dissociation phase (*right*) in which buffer flows over the surface
To further investigate relative ligand orientation, we compared the analyte binding activity, *A*, of the conjugated C8 and C8-zip ligands, defined as the moles of captured analyte divided by the moles of surface-conjugated ligand. In SPR, the moles of a given molecule deposited on the surface is proportional to the instrument response, *X*, divided by the molecular weight, *m*, so that the activity is readily expressed as$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ A=\frac{X_A\cdot {m}_L}{X_L\cdot {m}_A} $$\end{document}$$
where *X*~*L*~ and *X*~*A*~ are the SPR responses in RU to the binding of ligand and analyte, respectively, *m*~*L*~ and *m*~*A*~ are the molecular weights of the ligand and analyte, respectively, and it is assumed that the ligands are monovalent.
To properly isolate the effect of orientation on *A*, care must be taken to work in the dilute ligand limit in order to minimize steric hindrance effects. If analyte molecules can repel one another at the surface, the result is a reduced activity value unrelated to ligand orientation. Figure [3](#Fig3){ref-type="fig"} compares the activity of C8-zip to C8 for a range of surface ligand densities by exposing the surface to ligand concentrations of 0.25 to 10 μg/mL for 300 s. The ligand surface density was calculated assuming a sensor calibration of 1 RU = 1 pg/mm^2^ \[[@CR35]\]. At higher surface densities, the activity values of the two ligands converge as expected for a surface dominated by steric hindrance. As the surface concentration is reduced, both exhibit marked increases with maximum C8-zip and C8 activities of 65 and 36 %, respectively. The results indicate that the positively charged tail of the C8-zip improved surface orientation and are consistent with the enhanced sensitivity measured in Fig. [2](#Fig2){ref-type="fig"}. We also observed improved sensitivity for the orientated sdAb when using SPR GLC sensor chips, designed by Bio-Rad for general amine coupling via a compact polymer layer with a binding capacity of approximately one protein monolayer ([Supplementary Material](#Sec5){ref-type="sec"}).Fig. 3Ricin saturation activity versus ligand surface density for C8-zip and C8. Ligand surface density was calculated assuming a sensor calibration of 1 RU = 1 pg/mm^2^. The ricin concentration was 100 nM for all experiments
For the corresponding nanoplasmonic response studies, a functionalization protocol identical to that used for the SPR study in Fig. [2a](#Fig2){ref-type="fig"} was applied to the gold nanostructures of the custom-made LSPR chips. Response data to 100 nM ricin were determined by averaging the mean intensity of nine arrays, each array consisting of 400 nanostructures (Fig. [4a, b](#Fig4){ref-type="fig"}). As with the SPR results, the C8-zip surface was the most sensitive of the three ligands, with a saturation response that was 104 % greater than that of C8 and 61 % greater than that of D12f-rhiz (Fig. [4c](#Fig4){ref-type="fig"}).Fig. 4**a** LSPRi of nine arrays, each array consisting of 400 nanostructures. The image is false colored *red* to indicate the resonance wavelength of 635 nm. **b** Scanning electron microscopy of an array of 400 nanostructures. **c** LSPRi response to 100 nM ricin for C8-zip, C8, and D12f-rhiz surfaces compared to that of a biotinylated surface for 100 nM neutravidin
To further characterize the sensitivity of the C8-zip functionalized surface, we conducted LSPRi studies in which the surface was functionalized with biotin and exposed to a 100-nM neutravidin solution, a commonly used standard in nanosensor characterization. Both of these receptor-ligand pairs have exceptionally high binding affinities and both ricin and neutravidin have approximately the same molecular weight (60 kDa). Biotin, however, is readily orientated and has a molecular weight of 244 Da, 62 times lower than that of C8-zip. Thus, the differences in LSPRi sensitivity to analyte should be reflective of ligand size and degree of orientation. Interestingly, Fig. [4c](#Fig4){ref-type="fig"} shows the responses of the C8-zip and biotinylated surfaces to their cognate analytes were statistically equivalent, demonstrating that the small size and highly oriented conjugation of the C8-zip molecules are well suited for LSPRi applications. Future work will build on these results to estimate the limit of detection for each construct.
Conclusions {#Sec4}
===========
When conjugated to plasmonic nanostructures, orientated sdAb exhibited enhanced LSPRi sensitivity which matched that of biotinylated surfaces to neutravidin, long considered a model system for small and orientated ligand studies. The additional facts that sdAb produce well in bacteria and yeast, are readily modified, and robust to denaturation make them well suited for a broad range of SPR and LSPR biosensing applications. The application of LSPRi to ricin detection is an important step forward in the design of small, readily deployable biosensors against bio-threat agents.
Electronic Supplementary Material
=================================
{#Sec5}
######
(DOCX 620 kb)
We are grateful for generous funding from the Naval Research Laboratory's Institute for Nanoscience and 6.1 base funds.
| {
"pile_set_name": "PubMed Central"
} |
The content published in Cureus is the result of clinical experience and/or research by independent individuals or organizations. Cureus is not responsible for the scientific accuracy or reliability of data or conclusions published herein. All content published within Cureus is intended only for educational, research and reference purposes. Additionally, articles published within Cureus should not be deemed a suitable substitute for the advice of a qualified health care professional. Do not disregard or avoid professional medical advice due to content published within Cureus.
Introduction
============
Mentorship is an essential component of professional development and its benefits are well described in the medical, business, and education literature \[[@REF1]-[@REF10]\]. Mentorship relationships are beneficial for both mentors and mentees, who experience increased career satisfaction, scholarship, and efficiency of academic promotion \[[@REF11]-[@REF12]\]. Models include one-on-one mentor-protege models \[[@REF13]\], institutional group programs \[[@REF14]-[@REF15]\], and "speed-mentoring" sessions with many mentors in serial succession \[[@REF16]\].
Recently, a collaborative, network-based model for mentorship has gained popularity in the business world: the Mastermind group \[[@REF17]-[@REF20]\]. Initially described by Napoleon Hill \[[@REF21]\], the Mastermind group is composed of multiple colleagues, including near-peers and those at different stages of their academic careers, who provide mentorship and career advice for each other through regularly scheduled meetings. The group benefits from the combined intelligence and accumulated experience of the participants.
Advances in technology now mean that the world can literally be right at our fingertips. Geographic bounding of the mentor-mentee relationship is no longer necessary \[[@REF22]\]. In fact, for effective mentorship through Mastermind Groups, it may be advantageous to connect academic physicians from different centers in mentorship opportunities, since these relationships may be less influenced by personal gains and local politics.
We describe the curriculum design, implementation, and program evaluation of a Mastermind group for the purpose of professional development of academic emergency physicians. Our outcome measures reflect participant experience, impact, and feasibility of the program.
Materials and methods
=====================
Participants
Academic Life in Emergency Medicine (ALiEM; [www.aliem.com](www.aliem.com)) conducted two Mastermind group sessions in 2017 for volunteer faculty within the organization. ALiEM is a digital health professions education organization with a globally distributed team of faculty who lead various online educational initiatives. The Mastermind group was intended as a career development exercise for the volunteer faculty in the organization. Participants were recruited from within ALiEM via the organizational Slack Channel (Slack Technologies Inc, San Francisco, CA) through a general message to all volunteer faculty at ALiEM, asking for voluntary participation in this faculty development session. Only voluntary faculty within the ALiEM organization were included and there were no exclusion criteria. Groups were limited to 6-8 participants to facilitate small group discussion. Based on the number of recruited participants, two separate groups were formed. One group held its meetings in January of 2017 and the second group met in October of 2017. Because of the innate geographic diversity of volunteer faculty within the ALiEM team, each group was formed of physicians from varying locations across the United States and Canada. While the volunteer faculty at ALiEM come from different specialties and health professions, all of the volunteers for this faculty development program were Emergency Medicine physicians. Demographics of the group are described in Table [1](#TAB1){ref-type="table"}.
###### Participant Demographics
-------------------------- ------------------------- ------------------------------- ------------------------------ -----------------------------------------------
Participant Demographics
Sex Female 57% (n=8) Male 43% (n=6)
Academic Rank Instructor 14% (n=2) Assistant Professor 65% (n=9) Associate Professor 7% (n=1) Full Professor 14% (n=2)
Geographic Location US West Coast 29% (n=4) US Midwest 36% (n=5) US East Coast 14% (n=2) Canada 21% (n=3)
-------------------------- ------------------------- ------------------------------- ------------------------------ -----------------------------------------------
Procedures of the Mastermind group
Each Mastermind group completed two homework assignments and two 90-minute videoconference meetings, using a structured, moderator-facilitated format. Meetings were conducted using participants' personal computers on Google Hangouts on Air© (Google Inc., Mountain View, CA). As pre-work, participants completed selected questions from a free self-assessment tool created as supplementary, online material for \"Stand Out\", a book authored by Dorie Clark, adjunct professor at Duke University's Fuqua School of Business and proponent of the Mastermind group model \[[@REF23]-[@REF24]\]. The self-assessment survey summarized participants professional strengths, weaknesses and current career trajectory. Example questions can be found in Figure [1](#FIG1){ref-type="fig"}. Given 6-8 participants per group and the selected questions for each participant to discuss, 90 minutes was thought to be the optimal length of time per session.
{#FIG1}
In the initial group meeting, the moderator encouraged participants to discuss their self-assessments, current projects, and career challenges. The moderator then facilitated comments from all of the other participants and kept track of time so that all participants could discuss their self-assessments during the 90-minute session. Prior to the second meeting, participants each contributed to a shared, digital document (Google Docs©, Google Inc.) with their suggestions for professional development resources personalized for each of the other participants in the group. The second meeting allowed discussion of these suggested resources, actionable 'next steps', and an accountability timeline for each participant. The free, cloud-based platforms and voluntary basis for the Mastermind groups resulted in a zero-cost innovation.
{#FIG2}
Data collection
In addition to collecting the Mastermind group session notes compiled by the facilitator and participants through the shared, digital document, we also surveyed the participants via anonymous online survey software (SurveyMonkey, San Mateo, CA) to understand participant reactions and obtain program evaluation information.
The post-intervention survey of this program was administered to participants. This survey consisted of three phases: demographic data, participant reactions to the experience, and perceived value. Respondents were asked to rate their Mastermind Group experiences using a Likert Scale ranging from 1 (not valuable, would not recommend to others) to 10 (very valuable, would highly recommend this to others). They were also asked to compare these sessions to prior mentorship experiences using a comparison scale (better than, same as, worse than other mentorship experiences). Finally, qualitative data was gathered about the participants' experiences via a required free-text box. See Appendix 1 for a listing of all the questions.
Analysis
We describe the demographic composition of the groups, themes discovered in the Mastermind shared digital document notes compiled by the facilitator and participants, and we perform descriptive statistics on our program evaluation survey data.
Results
=======
Participant demographics
The two groups included two full professors, one associate professor, nine assistant professors, and two instructors, representing 14 different academic medical centers across North America. 57% of participants were female (n=8), and 43% were male (n=6). Compiled summary data of the groups is provided in Table [1](#TAB1){ref-type="table"}.
Description of themes from the Mastermind session notes
A majority of participants received specific resource recommendations during the sessions, including readings (e.g., books, journal articles, blog posts), training courses, or conferences. Many also received introductory email referrals to specific individuals for additional mentorship. This was made possible given the breadth of networks among the participants. All participants had at least one identifiable 'next step' related to their reason for participating in the Mastermind group with the goal of being accountable to the group.
Survey results
A post-intervention survey was sent to a convenience sample comprised of the 14 participants, with a final response rate of 100%. The participants rated the Mastermind group experience as 9.4/10 on the Likert scale. When asked to compare Mastermind groups with prior mentorship experiences, 3/14 (21.4%) respondents noted that this was not applicable as this was their first formal mentorship experience, while the remainder of the participants, 11/14 (78.6%) rated these sessions as "much better than prior experiences", 10/10 on a Likert scale. Participants cited one of two reasons for participating in the Mastermind groups: need for career advice or assistance with a project. Overall, the participants described a synergy of energy, commitment to one another's longitudinal success, and benefit from the diverse range of talent and expertise in the group as reasons for preferring this model to other models of mentorship. Many of the members discussed plans to replicate this mentorship model in other settings.
Associated costs
Program organizers noted that they used commonly available, free services (Google Hangouts on Air), which allowed them to run this for free. Each individual spent a total of three hours in online discussion forums, as well as the time needed for pre-work completion. The participants and moderators were all uncompensated, volunteer faculty in the ALiEM organization.
Discussion
==========
Gottlieb and colleagues have previously described the opportunities and barriers to digital mentorship in a recent paper \[[@REF22]\]. This application of digital technologies displays that it is acceptable, feasible, and economical to establish digitally-based Mastermind groups between geographically disparate individuals.
Interestingly, the participants did not mention any specific barriers to the Mastermind group, though we acknowledge that at a minimum there is an opportunity cost and minor scheduling challenges inherent in our design. Traditional barriers in mentorship have been previously reported in the literature as: time required for mentorship, lack of academic recognition, lack of financial incentives, "authoritative boss-employee relationship", lack of availability, and a lack of a good selection of mentors \[[@REF25]\]. Our faculty development program overcomes several of these barriers by connecting like-minded individuals across institutions through real-time meetings and connecting a globally distributed team. It also makes it less likely for those at a higher professoriate rank to have any authority or boss-employee relationship with the junior members. By reaching outside of geographic restrictions, the catchment of available mentors was also widened, increasing the number of candidates for whom this program would be relevant.
Limitations
There may be an inherently biased sample of individuals who agreed to participate in our pilot Mastermind program. Many of the individuals that volunteer within the ALiEM organization have a high fluency and affinity for digital technologies. Their experiences, therefore, may not generalize to a population that finds the aforementioned online platforms difficult to navigate. Additionally, volunteers recruited from the ALiEM organization were all from North America and all fluent English speakers. This group of participants may not be a typical cross-section of junior, mid-career, or senior faculty members in terms of their values and preferences around remuneration or academic merit. Finally, volunteers participated in only two, ninety-minute sessions, from which it is difficult to discern any long-term outcome measures or follow-through on proposed 'next steps'.
Conclusions
===========
Our experiences suggest that the Mastermind conceptual framework is a feasible, zero-cost, and effective model for professional development for faculty who volunteer their time. This model was easily and cost-effectively replicated from the business literature to a cohort of physicians. Though the model was originally proposed as a method for in-person discussions, we report a more modern, online experience for professional development in our diverse, globally-distributed team.
The authors have declared that no competing interests exist.
Consent was obtained by all participants in this study. Minneapolis Medical Research Foundation issued approval 18-2235X. Exempt from IRB Review
**Animal subjects:** All authors have confirmed that this study did not involve animal subjects or tissue.
Survey Questions
1\. Which best describes your current professional stage? A. In residency or fellowship training B. 1-5 years out of training C. 5-10 years out of training D. 10+ years out of training
2\. What is your academic rank? A. Resident B. Instructor C. Assistant Professor D. Associate Professor E. Full Professor
3\. What is your gender? A. Female B. Male C. Prefer not to say
4\. What is your age? A. 20-29 B. 30-39 C. 40-49 D. 50-59 E. 60+
5. If you have participated in formal mentorship programs prior to this Mastermind Group, please describe (e.g., mandatory departmental mentorship pairing; \"speed-dating\" mentorship at a local conference; etc.) (Free Response)
6. Please rate your experience with the Mastermind Group. (1 = NOT Valuable; Would NOT Recommend This To Others; 10 = Very Valuable; Would HIGHLY Recommend This To Others)
7\. Please compare this with prior mentorship experiences. A. This program was MUCH WORSE than prior mentorship experiences B. This program was SLIGHTLY WORSE than prior mentorship experiences C. This program was SIMILAR to prior mentorship experiences D. This program was SLIGHTLY BETTER than prior mentorship experiences E. This program was MUCH BETTER than prior mentorship experiences F. Not Applicable
8. When comparing the Mastermind group with prior mentorship experiences (Question \#3), why did you feel this way? (Free Response)
9\. After this Mastermind Group experience, what is one thing you will do differently? (This may include career, goals, professional development, personal development, or any other aspects you deem important). (Free Response)
10. Did you obtain any resources recommended for you during the discussion (eg. books, articles, online resources)? If so, please describe what you obtained. (Free Response)
11\. Have you contacted any of the individuals who were suggested to you? If so, please describe your experience with them. (Free Response)
12\. Please describe any other actions that you took as a result of your MasterMind group experience. (Free Response)
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1}
============
In the United States, the prevalence of hip and knee osteoarthritis has increased substantially over the last 20 years and is the greatest cause of chronic disability in older adults [@bib1], [@bib2]. Although there are measures to slow the progression of the disease, elective total joint arthroplasty (TJA) is the recommended treatment after non-surgical measures have failed [@bib3], [@bib4]. TJA is a major surgical procedure, and recovery time can vary between patients with the most improvement in health-related quality of life quantified by the Quality of Wellbeing Index between 3 and 6 months postoperatively [@bib5]. Physical therapists play an important role in treating patients before and after TJA [@bib6]. The main goals of rehabilitation post-TJA are to maximize functional independence and to minimize complications [@bib7].
There are 2 common ways of assessing outcomes after TJA: the patient\'s assessment of his/her own function (patient-reported outcome measures or PROMs) and observed physical performance (performance-based outcome measures or PBOMs). Common PROMs include the Knee Injury and Osteoarthritis Outcome Score (KOOS), Hip Injury and Osteoarthritis Outcome Score (HOOS), and the Lower Extremity Function Scale (LEFS) [@bib8], [@bib9]. Examples of PBOMs include the Timed Up and Go (TUG), 6-Minute Walk Test, and the Stair Climbing Test [@bib10], [@bib11]. Currently, there is no absolute consensus in the literature on the appropriate PROMs and PBOMs following total hip or knee arthroplasty (THA or TKA) [@bib12], [@bib13], [@bib14]. However, the American Joint Replacement Registry (AJRR) 2016 guide recommends collection of PROMs including Veterans RAND 12 Item Health Survey or Patient Reported Outcome Measure Information System Global and HOOS or KOOS Jr [@bib15]. Also in a recent American Academy of Hip and Knee Surgeon symposium, the HOOS Jr and KOOS Jr were recommended for quality assessment in TJA [@bib16].
Both PROMs and PBOMs are useful and provide different clinical data. PROMs do not require a clinical visit, and therefore might be easier to collect than PBOMs especially when following a large number of patients [@bib17]. To utilize a more patient centered approach to medicine, the Center for Medicaid and Medicare Services has recently highly valued the use of PROMs based on goals stated by the National Quality Strategy and Institute of Medicine due to the Affordable Care Act [@bib18], [@bib19]. PROMs provide useful information about patients\' perceptions of physical function but are highly influenced by pain [@bib20]. However, patient perception may not correlate well with actual functional performance and may overstate functional improvement especially in the early postoperative period [@bib10], [@bib11]. PBOMs on the other hand can be harder to collect, but may provide important objective information about functional performance and progress through rehabilitation [@bib11], [@bib20]. Recent studies have recommended the use of both PROMs and PBOMs for evaluating patient progress after THA/TKA [@bib10]. McAuley et al [@bib21] found that physical therapists use a wide range of outcome measures when evaluating THA and TKA patients in Canada.
The aim of this study is to assess current and anticipated use of PROMs and PBOMs of physical therapists practicing in New England. There is very little known about outcome measures that therapists use pre-TJA and post-TJA. This information is important because orthopaedic surgeons and physical therapists work toward the same goal of optimizing patient recovery. The motivation for this study is to establish a foundation of current practice from which to develop standardized sets of outcome measures for orthopaedic surgeons and physical therapists to collect pre-TJA and post-TJA.
Material and methods {#sec2}
====================
The study was cross-sectional in design. It was executed as an online questionnaire requiring 10-15 minutes to complete distributed via email to licensed physical therapists practicing in New England (Maine, Vermont, New Hampshire, Massachusetts, Rhode Island, and Connecticut). A cover letter of instructions was developed, and reminder emails were sent 12, 21, and 36 days after the initial correspondence on July 15, 2015. Physical therapists who treated patients undergoing THA and/or TKA in the last 5 years were invited to complete the survey and those who had not were asked to decline. The online survey platform LimeSurvey was used and anonymity was ensured by assigning each response a random numeric code. The study was approved by the Committee on Human Subjects.
The survey had 4 sections consisting of a modified version of the survey developed by McAuley et al [@bib21] obtained with permission from the lead author. The first section documented location of practice, education background, and demographic characteristics of the therapist. The second and third sections evaluated the use of PROMs and PBOMs ([Table 1](#tbl1){ref-type="table"}). These measures were queried specifically in terms of clinical decision making (day-to-day thinking and reasoning that clinicians execute to plan, administer, modify, and evaluate a therapeutic intervention for a given patient after THA or TKA). These specific PROMs and PBOMs were chosen based on the work of McAuley et al and the Osteoarthritis Research Society International (OARSI) advisory group recommendations [@bib21], [@bib22]. Responders were asked to rate their current use of each measure on a 4-point scale (0 = not familiar, 1 = familiar no experience, 2 = some experience, 3 = considerable experience). The third section asked about anticipated future use of specific measures using a modified scale (0 = unable to rate, 1 = unlikely to use, 2 = likely to use, 3 = will use and recommend) ([Fig. 1](#fig1){ref-type="fig"}). The fourth section asked for their opinions about most valuable measures outright, other modalities used, and number of postoperative treatment sessions patient receive.Figure 1Example from Section 2 of questionnaire. ^∗^*P* \< .05.Table 1Outcome measures used in cross-sectional survey of New England physical therapists.PROMsPBOMsNumeric Pain Rating ScaleSit to Stand TestLEFSWalking SpeedOKS6-Minute walk testOHSTUGEQ-5DTimed Stair ClimbKOOSTinetti Mobility TestHOOSSingle Leg BalanceWOMACFunctional Reach Test[^1]
Initially, the survey was sent to 14 physical therapists in various practices throughout New England for feedback on language clarity and organization. Based on their comments the survey was modified.
Data were exported into an Excel spreadsheet and converted into SPSS. Analyses of responses were reported in frequencies and percentages and visualized with graphs for comparison ([Figs. 2](#fig2){ref-type="fig"} and [3](#fig3){ref-type="fig"}). Following the approach used by McAuley et al [@bib21] variables were dichotomized from the ordinal 4-point scales to used/familiar (3, 2) and not used/unfamiliar (1, 0). Paired sample t-tests were used to compare the use of each outcome measure for current and future use. Significance was set at *P* \< .05.Figure 2Comparison of responders\' current and future use of PROMs (a) and PBOMs (b) for clinical decision making (t-test, *P* \< .05). ^∗^*P* \< .05. EQ-5D, Euro-Quality of Life; OHS, Oxford Hip Score; OKS, Oxford Knee Score; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index.Figure 3Physical therapists\' overall rating of most valuable PROMs (a) and PBOMs (b). EQ-5D, Euro-Quality of Life; OHS, Oxford Hip Score; OKS, Oxford Knee Score; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index.
Results {#sec3}
=======
Seven hundred twenty-four emails were sent. Of the 724 emails, 95 failed to be transmitted. Therefore 629 surveys were successfully sent. Of those, 168 responses were received, and of those, 19 responses were not interested in completing the survey and 27 of them did not treat patients who had undergone TJA. Therefore, this produced 122 complete responses.
[Table 2](#tbl2){ref-type="table"} shows the demographic data of the physical therapists who completed the survey ([Table 2](#tbl2){ref-type="table"}). Physical therapists reported treating patients on average for 13.5 ± 0.5 sessions post-TKA and 11.2 ± 0.4 sessions post-THA.Table 2Demographics of responding physical therapists in percentages.Population setting Rural21.3 Mixed62.3 Urban16.4Years since graduation (y) \<513.4 5-912.3 10-1417.2 15-1913.9 20-249.8 \>2532.8Working status Full time88.5 Part time8.2 Per diem4.9Clinical setting Private practice clinic53.3 Private practice clinic associated with large organization10.7 Home/community care (eg, VNA)15.6 Outpatient clinic associated with academic hospital/medical center18.0 Inpatient acute care hospital5.7 Non-hospital inpatient rehabilitation facility2.5 Other8.2Continuum Pre-operative phase74.6 Immediate post-operative phase (first few wk)83.6 Sub-acute rehabilitation (wk to mo)89.3 Post-rehabilitation (mo to y)82.0Number of TJA patients treated/year \<2542.6 25-4938.5 50-7413.9 75-994.1Gender Female56.6 Male43.4Age (y) \<3011.5 30-3929.5 40-4929.5 50-5923.0 ≥606.6Professional degree Bachelor of Science23.8 Master of Science in Physical Therapy30.3 Doctor of Physical Therapy40.2[^2]
Regarding current and future use of PROMS for clinical decision making, responders most commonly use and recommend the Numeric Pain Rating Scale and the LEFS ([Fig. 2](#fig2){ref-type="fig"}a). More specifically, in relation to the Numeric Pain Rating Scale, 99.2% reported current considerable experience and 97.5% would use and recommend it in the future. In relation to the LEFS, 76.2% reported considerable experience and 77.0% would use and recommend it in the future. Therapists were more likely to use and recommend the Oxford Hip Score and Oxford Knee Score in the future than in the past for clinical decision making (t-test, *P* \< .05).
Among PBOMs used for clinical decision making, responders most commonly use and recommend the Single Leg Balance Test and the TUG ([Fig. 2](#fig2){ref-type="fig"}b). More specifically, in relation to the Single Leg Balance Test, 90.2% reported current considerable experience and 87.7% would use and recommend it in the future. In relation to the TUG, 93.4% reported considerable experience and 85.2% would use and recommend it in the future. Therapists were less likely to use and recommend the 6-Minute Walk Test, TUG, Tinetti Mobility Test, and the Functional Reach Test for future clinical decision making (t-test, *P* \< .05).
When respondents were asked to identify the most valuable PROMS, 70.5% recommended the LEFS and 68.0% the Numeric Pain Rating Scale ([Fig. 3](#fig3){ref-type="fig"}a). In regards to performance measures, 63.1% selected the TUG as the most valuable, along with 59.0% and 42.6% rating the Sit to Stand and the Single Leg Balance Test, respectively, as most valuable ([Fig. 3](#fig3){ref-type="fig"}b).
Discussion {#sec4}
==========
To our knowledge, this is the first description of practice patterns of a large number of physical therapists in the New England region. The results demonstrate that physical therapists use LEFS and Numeric Pain Rating Scale more than any other PROM for clinical decision making and program evaluation. Currently, physical therapists are often using PBOMs, and a large percentage are likely to use and recommend the Single Leg Balance Test, TUG, and Sit to Stand Test for use in the future. Physical therapists in the New England region use very similar tests independent of their clinical practice settings or level of training.
In contrast, the AJRR recommends general health-related quality of life measures like the Veterans RAND 12 or PROMIS 10 Global and joint-specific measures including the HOOS, KOOS, Oxford Knee, and Oxford Hip Scores, none of which were most valuable to therapists in this study [@bib15]. It is difficult to explain why they prefer the LEFS. LEFS is easy to implement and broadly applicable to all lower extremity sites in various stages of disability, but is not specific to hip and knee osteoarthritis [@bib23]. The HOOS and KOOS are joint specific, including the Western Ontario and McMaster Universities Osteoarthritis Index score, and have been shown to be more sensitive and responsive than the LEFS in total joint replacement [@bib24], [@bib25]. However, they take longer to administer. Pua et al [@bib26] did not find a significant difference in the absolute reliability of the LEFS when compared to the Western Ontario and McMaster Universities Osteoarthritis Index in patients with hip osteoarthritis and stated that the LEFS may be a good alternative.
All the PBOMs except for the Timed Stair Climb were used by a majority of responders, and the TUG and Single Leg Balance Test were recommended by a majority for future use. PROMs may overestimate patient mobility, especially in the immediate postoperative phase after both TKA and THA [@bib10]. Because therapists evaluate TJA patients multiple times in the early postoperative period, they may utilize PBOM more frequently to avoid overestimation of function during the early phase of rehabilitation [@bib11], [@bib14], [@bib20], [@bib27]. In fact, osteoarthritis Research Society International recommends that in addition to PROMs, the following PBOMs be used to assist in clinical decision making post-TJA: the Sit to Stand Test, Walking Speed Test, Timed Stair Climb, TUG, and 6-Minute Walk Test [@bib22]. It would be ideal if a single or small set of measures could assess function at all levels of rehabilitation, but perhaps no single test is able to assess all aspects and phases of recovery. PROMs and PBOMs also assess different time periods of recovery. PROMs generally assess a period of weeks of overall symptoms and function while PBOMs objectively measure function at a particular point in time. Both types of information are valuable in assessing patient recovery [@bib28].
Physical therapists and surgeons appear to utilize different tools to assess recovery after TJA with AJRR recommending that surgeons report only PROMs to the national registry. PROMs have the advantage of being patient centered and are consistent with performance-based initiatives [@bib4]. Although PROMs may not accurately assess early postoperative function, they are more responsive in the long term compared to PBOMs in measuring functional improvement after TKA [@bib10]. As physicians and therapists strive to provide patient-centered care, the patient perception of the outcome is important [@bib29], [@bib30]. This study showed that of the 16 outcome measures queried for clinical decision making, therapists indicated that they were less likely to use 4 of them in the future, all them PBOMs, and more likely to use 2 in the future, both PROMs ([Fig. 2](#fig2){ref-type="fig"}a and b). This may indicate dissatisfaction with therapist\'s current usage of PBOMs and interest in using PROMs more in the future. As surgeons and therapists work more closely, developing better understanding and consensus in the use of PROM and PBOM between surgeons and physical therapists will allow for improved assessment of TJA patient outcomes.
There were some potential limitations associated with this study. This study does have a low response rate of 20%, which may represent underlying biases. Another limitation is that the questionnaire did not ask about the timing of evaluation of PROMs and PBOMs. Outcome measures recorded at 2 weeks cannot be compared to one at 12 months. Timing is an important factor that may also be considered and evaluated in future studies. An adequate population of physical therapists in New England may not have been assessed to make generalizations about practice patterns. The HOOS and KOOS Jr are recommended by Medicaid and Medicare and might have been included; however, these shorter surveys did not have published information on scoring and were too new for many therapists to be familiar with at the time of this study [@bib16]. Certified Athletic Trainers, nurses, or occupational therapists who also treat this patient population were not included. "Some" or "considerable" experience was not defined, which is why the results were dichotomized. Another limitation is that 42.6% of the total cohort surveyed treat less than 25 joints/y, which might not be consistent with high volume centers.
Conclusions {#sec5}
===========
This article is the first to describe practice patterns of a large number of physical therapists treating knee and hip arthroplasty patients in the New England area. It demonstrates that therapists use and recommend the Numeric Pain Rating Scale and the LEFS for clinical decision making and program evaluation. Most are unfamiliar with the PROMs that surgeons more commonly use. While the use of PBOMs was more varied, therapists are likely to use and recommend the TUG and Single Leg Balance Test. As surgeons and therapists work more closely in a value-based system, coordination in the use of PROMs and PBOMs is needed for the assessment of TJA patient outcomes. This study highlights the need for more collaboration and consistency between the disciplines.
Appendix A. Supplementary data {#appsec1}
==============================
Conflict of Interest Statement for HalseyConflict of Interest Statement for ImadaConflict of Interest Statement for BlanksteinConflict of Interest Statement for Nelms
The authors would like to thank Bruce Beynnon, PhD and Catherine McAuley and colleagues. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
One or more of the authors of this paper have disclosed potential or pertinent conflicts of interest, which may include receipt of payment, either direct or indirect, institutional support, or association with an entity in the biomedical field which may be perceived to have potential conflict of interest with this work. For full disclosure statements refer to [https://doi.org/10.1016/j.artd.2017.08.003](10.1016/j.artd.2017.08.003){#intref0010}.
[^1]: EQ-5D, Euro-Quality of Life; OHS, Oxford Hip Score; OKS, Oxford Knee Score.
[^2]: VNA, Visiting Nurses Association.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#s0005}
===============
Gastrointestinal parasites account for huge losses experienced by the livestock industry worldwide ([@bb0165]). While presently the major way of control and treatment of these gastrointestinal parasites depends profoundly on the utilization of conventional anthelmintics, there are challenges associated with the use of these synthetic anthelmintics ([@bb0020]; [@bb0075]). The challenges include unavailability and inconsistent supply; high cost to resource-limited farmers, environmental pollution and the build-up of residues in animal products ([@bb0065]). The rapidly increasing resistance to anthelmintics may perhaps be as a result of the increased rate of dosing, under-dosing, prophylactic mass treatments and repeated chronic use of the same anthelmintic drugs ([@bb0135]). Regular use of these anthelmintics has also led to poor development of natural immunity against gastrointestinal parasites ([@bb0080]).
One practical way of developing cheaper and effective anthelmintics is to explore potential indigenous herbal remedies used as anthelmintics ([@bb0145]). Plant-derived anthelmintics is a promising area of research in trying to mitigate the challenges around the use of synthetic anthelmintic drugs in controlling gastrointestinal parasites. Alternative ethnoveterinary medicines, for example, plant extracts with anthelmintic properties are considered to be of immense potential in overcoming anthelmintic resistance ([@bb0090]). There have been many reports, mainly from Africa and Asia have indicated the effectiveness of medicinal plants against helminth infections in livestock ([@bb0005]; [@bb0025]). Hence, there is a need to investigate the potential active compounds from these plants.
Worldwide research has shown that some plants can be utilized to diminish the level of parasitism in livestock; thus are considered alternative options to the conventional chemical anthelmintics ([@bb0030]). *Elephantorrhiza elephantina* is a medicinal plant commonly utilized in ethnoveterinary medicine. Root decoctions infusions of *E*. *elephantina* are used to control gastrointestinal worms ([@bb0160]; [@bb0015]; [@bb0105]). The anthelmintic effects of this plant are due to the activity of secondary metabolites, such as tannins, alkaloids, saponins and glycosides ([@bb0125]; [@bb0025]; [@bb0040]). Phytochemicals are beneficial in the treatment of diseases and control of parasites that nourish on the walls of the gastrointestinal tract ([@bb0110]). Plants produce them as a defense mechanism. Research demonstrates the ability of these phytochemicals to protect humans and animals against infections ([@bb0085]).
However, reports on the phytochemical analysis of crude extracts from *E*. *elephantina* are limited. This study sought to quantify the phytochemical constituents of *E*. *elephantina* in ethanol, methanol and water extracts of the root. Furthermore, the study evaluated the *in vitro* anthelmintic potential of *E*. *elephantina* root extracts against adult *Paramphistomum cervi* in goats.
2. Materials and methodology {#s0010}
============================
2.1. Plant collection {#s0015}
---------------------
Roots of *E*. *elephantina* were collected from a characteristic populace in Mt. Frere, Alfred Nzo District Municipality, Eastern Cape, South Africa. Mt. Frere lies along latitude 30°55′0″ S and longitude 28°58′60″ E. Immediately after harvesting, the roots were rinsed in distilled water and air-dried in the shade for 10 weeks and after that were ground into powder using a grinder with 1 mm pore size sieve (IKA- Universal Mill M20, Laboratory and Scientific Equipment Co. Pty. Ltd). The root powder was kept in airtight containers at 4 °C until further analyses.
2.2. Extraction procedure {#s0020}
-------------------------
Three samples of 20 g of *E*. *elephantina* root powder were soaked in 200 mL each of ethanol, methanol, and distilled water. The mixtures were left on an orbital shaker for 24 h and then filtered under pressure using a Buchner funnel and Whatman filter paper (12.5 cm; 111^v^). Then the ethanol and methanol extracts were condensed in a rotary evaporator (Laborator 4000-efficient, Heidolph, Germany) while the water extract was freeze-dried (Vir Tis benchtop K, Vir Tis Co, Gardiner, NY). The dried extracts were used for the quantitative analysis of phytochemicals. The percentage yield was calculated for each extract using the formula:$$\text{Yield}\ \left( \% \right) = \frac{\text{Final weight}}{\text{Initial weight}} \times 100$$
2.3. Quantitative determination of phytochemicals {#s0025}
-------------------------------------------------
### 2.3.1. Alkaloids {#s0030}
The gravimetric analysis for total alkaloid content was used to determine alkaloids with a few modifications ([@bb0115]). A 0.5 mL of plant extract (1 mg/mL) was combined with 100 mL of 10% acetic acid in ethanol. The mixture was covered and left standing for 4 h. Afterwards, the mixture was filtered, and the filtrate was concentrated in a water bath at 60 °C to a quarter of its initial volume. Concentrated ammonium hydroxide (25%) was added to the mixture drop-wise until the formation of a precipitate was completed. The entire mixture was left to stand, then the accumulated precipitates were rinsed with 20 mL of 0.1 M dilute ammonium hydroxide and then filtered. The collected residue was oven-dried at 80 °C and weighed. The alkaloid content was calculated using the formula:$$\text{Alkaloid}\left( \% \right) = \frac{\text{Weight of residue}}{\text{Weight of sample taken}} \times 100$$
### 2.3.2. Condensed tannins {#s0035}
Determination of total condensed tannin was performed by applying the technique described by [@bb0150] with a few modifications. A 0.5 mL of plant extract (I m/mL) was added to 3 mL of 4% vanillin-methanol. A 1.5 mL of 37% hydrochloric acid was added to the mixture, thoroughly mixed and left standing for 15 min at room temperature. The absorbance of catechin standard (0.02 to I mg/mL) solution and plant extracts were recorded at 500 nm using the UV 3000 PC Spectrophotometer. Distilled water was used as the blank. Readings were taken in triplicates. The calibration curve was plotted using standard catechin. The readings were expressed as mg of catechin equivalent per g dry weight extract (mg CE/g).
### 2.3.3. Flavonoids {#s0040}
Flavonoid content was determined as described by [@bb0070] using the aluminium chloride colorimetric assay technique with some modifications. Briefly, 2 mL of distilled water and 0.15 mL of 5% sodium nitrite were put into a test-tube with 0.5 mL of the plant extract (1 mg/mL). The mixture was left standing for about 5--6 min at room temperature. After 6 min 0.15 mL of 10% aluminium chloride was added to the mixture and then was left standing for an additional 6 min. Afterwards, 1 mL of 4% sodium hydroxide was added to the mixture. A 1.2 mL of distilled water was poured to make up the volume of the mixture to 5 mL. The mixture was incubated for 15 min to develop colour. The absorbance was taken at 420 nm using a UV 3000 PC Spectrophotometer. Each assay was done in triplicates. The calibration curve was plotted using standard quercetin (0.2 to 1 mg/mL). The results were calculated as mg of quercetin equivalent for each g dry weight (mg QE/g).
### 2.3.4. Phenols {#s0045}
Phenol content was quantified through Folin-Ciocalteu\'s technique with a few modifications ([@bb0140]). A 2.5 mL of Folin-Ciocalteu reagent was added to 0.5 mL of each plant extract (I mg/mL) and vortexed for about 1 min. The mixture was held at room temperature for about 3--8 min. A 2.5 mL of 7.5% anhydrous sodium carbonate was then added and incubated in a water bath at 40 °C for 30 min to develop colour. Absorbance was then read at 765 nm using the UV 3000 PC Spectrophotometer. The blank was performed using distilled water. Each assay was done in triplicates. Standard gallic acid (0.02 to 1 mg/mL) was used to plot the calibration curve. The readings were expressed as mg of gallic acid equivalent for each g dry weight (mg GAE/g).
### 2.3.5. Saponins {#s0050}
The saponin content was determined following the technique of [@bb0115] with some modifications. A 20 mL of 20% ethanol was added to 0.5 mL (1 mg/mL) of the plant extract and put on a shaker for 30 min. The resulting mixture was then left in a water bath at 55 °C for 4 h. The mixture was filtered and extraction was repeated with 20 mL of 20% ethanol. The collected filtrates were concentrated to one-quarter of the original volume in a water bath at 90 °C. Afterwards, the concentrate was put into a 250 mL separating funnel and extraction was done twice with 20 mL diethyl ether. The ether layer was disposed of, while the aqueous layer was kept. A 20 mL *n*-butanol was poured to the aqueous layer, and then washing was done twice using 5 mL of 5% sodium chloride. The solution was placed in a water bath to evaporate. Then the samples were oven-dried at 40 °C. Calculation of saponin content was determined using:$$\text{Saponin}\mspace{2mu}\left( \% \right) = \frac{\text{Weight of final filtrate}}{\text{Weight of\ sample}} \times 100$$
2.4. Worm recovery {#s0055}
------------------
Worms were collected from freshly slaughtered goats at East London Abattoir in the Eastern Cape Province (latitude 32.9702° S and longitude 27.8872° E). Worm recovery was done following the method described by [@bb0035]. Immediately after slaughter, rumens were collected from the animals, ligated and quickly transported in a cooler box with ice to the Animal Science Laboratory at the University of Fort Hare. The contents of the rumens were washed, several times with phosphate-buffered saline (PBS) until the worms were free from debris. The species and description of the worms are as shown in [Table 3](#t0015){ref-type="table"}. The worms were then recovered from the rumens and kept in PBS at 37 °C for 2 h.
2.5. Preparation of solutions {#s0060}
-----------------------------
All test solutions and the standard drug solutions were freshly prepared before starting the experiments. Concentrations of 1.875, 3.75, 7.5 and 15 mg/mL were prepared for each extract by dissolving the water extract in PBS, while the ethanol and methanol extracts were dissolved in PBS mixed with 5% dimethyl sulfoxide (DMSO). A 5% DMSO/PBS mixture was used as the negative control after ascertaining that the addition of 5% DMSO did not affect the worms ([@bb0010]). Valbantel® (Albendazole 1.9% m/v and Closantel sodium 3% m/v Pfizer, South Africa) was used as the positive control after it was prepared in PBS to a final concentration of 1.875 mg/mL.
2.6. Experimental design and treatment procedures {#s0065}
-------------------------------------------------
Adult motility inhibition assay was done following the method of [@bb0050]. Ethanol, methanol and water extracts of *E*. *elephantina* were tested at different concentrations (1.875, 3.75, 7.5 and 15 mg/mL). Ten *P*. *cervi* adult worms were placed in each petri dish containing 3 mL of each of the above test solutions and incubated at 37 °C. Every concentration was tested in triplicate. Worm motility was observed at 2-hour intervals for 16 h, and each observation period lasted for 5--6 s. The number of motile worms was recorded per treatment during the observation period and expressed as a percentage. Time for motility inhibition was noted when no movement of any sort was observed except when the worms were shaken vigorously.
2.7. Statistical analysis {#s0070}
-------------------------
The results were presented as Mean ± SD. Fisher\'s Least Significant Difference (LSD) was used to compare the difference between the ethanol, methanol and water extracts at a 5% level (*P* \< *0*.*05*) using the Minitab 17 Statistical Software (2003). The mean percentages of worm motility inhibition were assessed using PROC GLM of SAS (SAS, 2003). Turkey post-hoc testing (SPSS version 22, Armonk, NY, 2013) was used to compare differences between treatment means. Two-way ANOVA test was performed to test for anthelmintic efficacy between treatment means at a 5% significance level (*P* \< *0*.*05*).
3. Results {#s0075}
==========
Methanol root extract exhibited higher yield (18.00%), followed by water (12.95%) and ethanol (9.91%) as shown in [Table 1](#t0005){ref-type="table"}. The results of the quantitative phytochemical screening showed that *E*. *elephantina* root contains high amounts of alkaloids, flavonoids, phenols, saponins and condensed tannins ([Fig. 1](#f0005){ref-type="fig"}, [Table 2](#t0010){ref-type="table"}). Higher alkaloid content was present in the water extract (93.24 ± 1.68%), followed by ethanol and methanol extracts (52.33 ± 1.38 and 34.77 ± 0.07%, respectively). There were significant differences in the saponin contents among the ethanol, methanol and water extracts at a 5% level (*P* \< *0*.*05*). Saponin content was highest in ethanol extract (83.28 ± 1.72%), followed by methanol extract (63.74 ± 0.75%) and water extract (50.83 ± 0.43%). Total condensed tannins in the roots were found to be (450.52 ± 0.15 mg CE/g) in ethanol extract, (354.53 ± 1.03 mg CE/g) in methanol extract and (41.31 ± 0.33 mg CE/g) in the water extract. A positive correlation was shown between standard catechin concentration and the absorbance values (R^2^ = 0.9986). Fisher\'s LSD analysis showed a significant difference in phytochemical quantities among the three solvents, which showed that ethanol had higher condensed tannin content than methanol and water extracts. The content of flavonoids was found to be high in ethanol (789.76 ± 0.21 mg QE/g) and methanol extracts (803.93 ± 0.13 mg QE/g) and there was no significant difference (*P* \> *0*.*05*) between them. However, the flavonoid content in ethanol and methanol extracts were significantly different (*P* \< *0*.*05*) from the water extract which contained lower content of flavonoids (403.44 ± 0.19 mg QE/g). A significant positive correlation was observed between standard quercetin concentration and the absorbance values (R^2^ = 0.998). The ethanol extract had the highest phenol content (476.11 ± 0.37 mg GAE/g), followed by methanol extract (462.86 ± 0.31 mg GAE/g) and lastly the water extract (359.91 ± 0.24 mg GAE/g). There was a significant difference (*P* \< *0*.*05*) between phenol contents in ethanol, methanol and water extracts. A positive correlation was observed between standard gallic acid concentration and the absorbance values (R^2^ = 0.9981).Table 1Percentage yield of extracts in ethanol, methanol and water solvents.Table 1SolventInitial weight of the powder (g)Final weight of the powder (g)Weight of the crude extract (g)Yield of crude extract (%)Ethanol2018.0191.9819.91Methanol2016.4013.59918.00Water2017.412.5912.95Fig. 1Phytochemical constituents in the ethanol, methanol and water extracts of *E*. *elephantina* roots. Results are expressed as means ± SD of three replicates. Bar graphs with dissimilar letter superscript in the same constituent are significantly different (*P* \< *0*.*05*).Fig. 1Table 2Content of alkaloids and saponins in extracts of *E*. *elephantina*.Table 2ExtractAlkaloids (%)Saponins (%)Ethanol52.33^b^ ± 1.3883.28^a^ ± 1.72Methanol34.77^c^ ± 0.0763.74^b^ ± 0.75Water93.24^a^ ± 1.6850.83^c^ ± 0.43[^1]Table 3Picture showing adult *P*. *cervi* recovered from the rumen of goats.Table 3Table 4*In vitro* effect of *E*. *elephantina* extracts on adult *P*. *cervi* in comparison with Valbantel.Table 4TreatmentConc\
(mg/mL)Motility inhibition (%) of adult *P*. *cervi* worms at different hours (Mean ± SD)024681012141618Ethanol1.875100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0090.0^be^ ± 0.0090.0^bg^ ± 0.0076.7^ae^ ± 15.2840.0^efg^ ± 10.0033.3^bcd^ ± 15.283.3^c^ ± 5.773.75100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0090.0^a^ ± 10.0086.7^ae^ ± 11.5583.3^agh^ ± 15.2866.7^be^ ± 5.7726.7^fgh^ ± 20.8226.7^bcd^ ± 15.280.0^c^ ± 0.007.5100.0^a^ ± 0.00100.0^a^ ± 0.0093.3^ac^ ± 5.7793.3^ad^ ± 5.7780.0^be^ ± 10.0070.0^ch^ ± 10.0013.3^c^ ± 5.776.7^h^ ± 5.773.3^d^ ± 5.770.0^c^ ± 0.0015100.0^a^ ± 0.00100.0^a^ ± 0.0086.7^ac^ ± 5.7770.0^be^ ± 10.0050.0^c^ ± 10.0030.0^e^ ± 10.003.3^cg^ ± 5.770.0^i^ ± 0.000.0^d^ ± 0.000.0^c^ ± 0.00Methanol1.875100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.7796.7^a^ ± 5.7796.7^a^ ± 5.7780.0^aj^ ± 10.0040.0^bc^ ± 10.0013.3^bc^ ± 5.773.75100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.7796.7^a^ ± 5.7793.3^ag^ ± 0.5.7786.7^ae^ ± 5.7770.0^bj^ ± 10.0033.3^bcd^ ± 15.286.7^bc^ ± 5.777.5100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0090.0^a^ ± 10.0086.7^ae^ ± 5.7776.7^agh^ ± 20.8270.0^bfh^ ± 10.0040.0^efg^ ± 10.0010.0^d^ ± 10.003.3^c^ ± 5.7715100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0083.3^ade^ ± 15.2880.0^be^ ± 10.0056.7^dh^ ± 5.7756.7^be^ ± 11.5533.3^g^ ± 5.773.3^d^ ± 5.770.0^c^ ± 0.00Water1.875100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.7796.7^a^ ± 5.7793.3^a^ ± 5.7756.7^b^ ± 15.2820.0^b^ ± 10.003.75100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.7796.7^a^ ± 5.7796.7^a^ ± 5.7786.7^a^ ± 5.7743.3^bc^ ± 5.7716.7^b^ ± 5.777.5100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.7793.3^ae^ ± 5.7783.3^cg^ ± 5.7770.0^bfh^ ± 10.0056.7^bef^ ± 15.2840.0^bc^ ± 0.006.7^bc^ ± 5.7715100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.7793.3^ad^ ± 5.7786.7^ae^ ± 5.7770.0^ch^ ± 10.0060.0^be^ ± 10.0036.7^cef^ ± 15.2833.3^c^ ± 5.770.0^c^ ± 0.00Positive control (Valbantel)1.875100.0^a^ ± 0.00100.0^a^ ± 0.0060.0^b^ ± 10.0020.0^c^ ± 10.0013.3^d^ ± 5.770.0^f^ ± 0.000.0^dg^ ± 0.000.0^i^ ± 0.000.0^d^ ± 0.000.0^c^ ± 0.00Negative control (PBS)100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.00100.0^a^ ± 0.0096.7^a^ ± 5.77[^2]
The extracts of *E*. *elephantina* showed time and dose-dependent anthelmintic activity on the adult *P*. *cervi*. Generally, all *E*. *elephantina* extracts showed anthelmintic effects against the adult worms ([Table 4](#t0020){ref-type="table"}; [Fig. 2](#f0010){ref-type="fig"}, [Fig. 3](#f0015){ref-type="fig"}, [Fig. 4](#f0020){ref-type="fig"}, [Fig. 5](#f0025){ref-type="fig"}, respectively). The extracts of *E*. *elephantina* roots significantly inhibited adult worm motility over the 16 h treatment period (*P* \< *0*.*05*). The positive control (Valbantel) showed 100% motility inhibition within 10 h while the negative control (PBS) showed 100% motility. Ethanol extracts showed a highly significant (*P* \< *0*.*05*) inhibition of motility (86.67 and 96.67%) at concentrations 7.5 and 15 mg/mL respectively, after 12 h of treatment. However, there was no significant difference in the inhibition of motility between ethanol (15 mg/mL) and the positive control; while ethanol (7.5 mg/mL) was significantly different from the positive control at 12 h post-exposure. Methanol and water extracts exhibited significantly different (*P* \< *0*.*05*) motility inhibition (96.67 and 66.67%, respectively) at concentration of 15 mg/mL 16 h post-exposure. There was no significant difference in the inhibition of motility by methanol (15 mg/mL) and positive control at 16 h post-exposure (*P* \> *0*.*05*).Fig. 2Time- and dose-dependent *in vitro* anthelmintic activity of ethanol, methanol and water extracts of *E*. *elephantina* (at a concentration of 1.875 mg/mL) in comparison with positive (Valbantel 1.875 mg/mL) and negative (PBS) controls on adult *P*. *cervi*.Fig. 2Fig. 3Time- and dose-dependent *in vitro* anthelmintic activity of ethanol, methanol and water extracts of *E*. *elephantina* (at a concentration of 3.75 mg/mL) in comparison with positive (Valbantel 1.875 mg/mL) and negative (PBS) controls on adult *P*. *cervi*.Fig. 3Fig. 4Time- and dose-dependent *in vitro* anthelmintic activity of ethanol, methanol and water extracts of *E*. *elephantina* (at a concentration of 7.5 mg/mL) in comparison with positive (Valbantel 1.875 mg/mL) and negative (PBS) controls on adult *P*. *cervi*.Fig. 4Fig. 5Time- and dose-dependent *in vitro* anthelmintic activity of ethanol, methanol and water extracts of *E*. *elephantina* (at a concentration of 15 mg/mL) in comparison with positive (Valbantel 1.875 mg/mL) and egative (PBS) controls on adult *P*. *cervi*.Fig. 5
4. Discussion {#s0080}
=============
The findings of this work have shown that the root extracts of *E*. *elephantina* exhibited anthelmintic activity against the adult *P*. *cervi* motility that was significantly different (P \< 0.05) from the positive control (Valbantel). These findings are similar to studies by [@bb0100] and [@bb0130] who reported anthelmintic activity of root fractions of *E*. *elephantina* against nematodes. The results of this study also exhibited time and dose-dependent anthelmintic activity which resulted in the inhibition of motility. Thus increasing the concentration of the extracts caused early motility inhibition. The time and dose-dependent anthelmintic effects could be due to more compounds with anthelmintic activity being present at higher concentrations of the extract ([@bb0095]). The ethanol extract was the most effective at the test concentration of 7.5 and 15 mg/mL compared to methanol and aqueous extracts. This could be attributed to other molecules present or the different mechanisms of action against the parasites. Methanolic and aqueous extracts exhibited significantly different (P \< 0.05) motility inhibition (96.67 and 66.67%, respectively) at a concentration of 15 mg/mL, 16 h post-exposure. The effect of most worm expellers such as albendazole is to cause worm paralysis leading to expulsion of the worm *via* faecal matter of the host ([@bb0060]). Therefore, the methanolic extract at the test concentration of 15 mg/mL showed a significant anthelmintic activity as compared to the positive control (Valbantel).
In this present study, the anthelmintic effects of *E*. *elephantina* could be attributed to one or more of the phytochemicals present in its roots such as alkaloids, condensed tannins, flavonoids, phenols and saponins. However, the precise mechanism of action of these phytochemicals against gastrointestinal nematodes is not well understood as their effects might be additive, synergistic, or antagonistic and act at single or multiple target sites ([@bb0180]; [@bb0040]). Tannins are postulated to cause anthelmintic activity by either binding to the proteins found in the gastrointestinal tract of the host or to glycoprotein on the cuticle of the parasite resulting in the death of the worms ([@bb0045]; [@bb0175]). Also, tannins are thought to interfere with energy generation by uncoupling oxidative phosphorylation ([@bb0120]; [@bb0170]) which can lead to worm death. On the other hand, saponins are reported to disrupt the cell membrane of the parasites thereby changing the morphology of the cells in the cuticle ([@bb0055]). Saponins may also affect feed intake and nourishment of the parasites, resulting in parasites dying ([@bb0040]). Alkaloids have been reported to act on the central nervous system and are thought to cause paralysis of worms ([@bb0120]; [@bb0170]).
5. Conclusions and recommendations {#s0085}
==================================
This study showed that aqueous, ethanol and methanol extracts of *E*. *elephantina* root exhibited anthelmintic activity: they inhibited motility in adult *P*. *cervi*. Therefore, further research is needed to determine their bioactivities *in vivo* as *E*. *elephantina* could contribute to the development of novel anthelmintic agents. The findings thereof may be useful in developing phytotherapic products that are more cost-effective, safer, and accessible with reduced risk of resistance than the conventional drugs presently in use.
Acknowledgement {#s0095}
===============
This work was supported by National Research Fund (10.13039/501100001321NRF) of South Africa (Innovation Master\'s Scholarship-Grant number 94851; Project T219 South Africa-Namibia Ethno-veterinary Project) and 10.13039/100014448Govan Mbeki Research and Development Centre (GMRDC) - University of Fort Hare.
Declaration of competing interest {#s0100}
=================================
The authors report that there are no known conflicts of interest associated with this publication. The authors alone are responsible for the content and writing of this article.
[^1]: Results are expressed as means ± SD of three replicates. Different superscripts within a column represent significant differences at (*P* \< *0*.*05*).
[^2]: ^abcdefghij^Means with the same lowercase letters within the same column are not significantly different (P \< 0.05).
| {
"pile_set_name": "PubMed Central"
} |
1.. Introduction {#sec1}
==================
Many recent advances in synchrotron X-ray imaging can be attributed to X-ray focusing optics (Ice *et al.*, 2011[@bb6]). These optics may operate *via* three possible principles: (i) diffraction, such as in Fresnel zone plates (Kirz, 1974[@bb10]) and multilayer Laue lenses (Kang *et al.*, 2006[@bb8]); (ii) total reflection, such as in Kirkpatrick--Baez (Kirkpatrick & Baez, 1948[@bb9]) and Wolter (Wolter, 1952[@bb27]) mirrors, and lobster-eye (Inneman *et al.*, 1999[@bb7]) and Kumakhov (Kumakhov & Komarov, 1990[@bb12]) lenses; and (iii) refraction, such as in prisms (Cederstrom *et al.*, 2000[@bb2]) and compound refractive lenses (CRLs) (Snigirev *et al.*, 1996[@bb24]). In the hard X-ray regime (*E* \> 15 keV), CRLs (linear arrays of refractive lenslets) are widely used due to their relatively low cost, ease-of-use and efficiency. Furthermore, their focal length can be actively varied by adjusting the number of lenslets (Vaughan *et al.*, 2011[@bb25]). However, the spatial resolution of CRL-based imaging systems is typically 50--100 nm (Schroer *et al.*, 2005[@bb20]), worse than that of other optics at comparable energies: 7 nm (Yamauchi *et al.*, 2011[@bb28]), 8 nm (Morgan *et al.*, 2015[@bb14]) and 20 nm (Vila-Comamala *et al.*, 2012[@bb26]) have been reported from microscopes based on mirrors, multilayer Laue lenses and Fresnel zone plates, respectively. Nonetheless, the advantages of CRLs make them uniquely suitable for *in situ* experiments where efficiency, large working distances and high X-ray energies are required. Under such circumstances, improving the spatial resolution of CRLs could facilitate new opportunities for multi-scale characterization.
One route to improving spatial resolution is by optimizing the CRL geometry (Chen *et al.*, 2014[@bb3]). Numerical optimization requires concise analytical expressions for parameters such as focal length, transmission and aberration. Furthermore, these expressions are essential for imaging techniques that involve sampling data in grids such as ptychography (Schroer *et al.*, 2008[@bb19]), scanning X-ray microscopy (Schroer *et al.*, 2005[@bb20]) or dark-field X-ray microscopy (Simons *et al.*, 2015[@bb22]). The optical theory of CRLs and CRL-based imaging systems has been addressed by various approaches such as ray-transfer matrices (RTMs) (Protopopov & Valiev, 1998[@bb17]; Pantell *et al.*, 2003[@bb15]) \[including Gaussian beam variants (Poulsen & Poulsen, 2014[@bb16])\], Monte Carlo ray tracing (Sanchez\~del Rio & Alianelli, 2012[@bb18]), wavefront propagation methods (Kohn, 2003[@bb11]) and others (Lengeler *et al.*, 1999[@bb13]). While these have greatly furthered the design and implementation of CRLs, no single formalism fulfills the core requirements for optimization: (i) simple, closed expressions, (ii) broad applicability to both condensing and full-field imaging systems, and (iii) consideration of both the thin-lens (where the focal length of the CRL far exceeds its length) and thick-lens conditions (where this approximation is no longer valid).
We present a formalism for CRL-based imaging systems utilizing an RTM approach to model archetypal X-ray imaging systems in a lens-by-lens manner, thus accounting for both thin- and thick-lens conditions. Attenuation by the lens material is calculated using RTMs to trace the ray position through the CRL. We provide exact analytical expressions for focal length, numerical aperture, spatial resolution, vignetting and chromatic aberration among other key optical parameters. These expressions form the basis of an efficient parametric optimization of the CRL and imaging geometry, which ultimately provides suggestions for future lens development routes.
2.. RTM formalism for CRLs {#sec2}
============================
2.1.. Assumed CRL and lenslet geometry {#sec2.1}
----------------------------------------
This formalism assumes a one-dimensional (1D) focusing geometry valid for both axisymmetric two-dimensional and planar 1D CRLs. The CRL is comprised of *N* identical parabolic and non-kinofirm lenslets (Fig. 1[▸](#fig1){ref-type="fig"}), each with radius of curvature *R*, aperture 2*Y* and center-to-center distance between successive lenslets *T* such that . For manufacturing reasons, lenslets have a small distance between the parabolic apices (*i.e.* a web thickness) that affects attenuation. There may also be a gap between adjacent lenslets, implying that the physical lenslet thickness is less than *T*. Such geometries limit *Y* and necessitate defining the physical aperture such that = .
2.2.. Background to the RTM approach and focusing behavior {#sec2.2}
------------------------------------------------------------
RTM analysis is a paraxial ray-tracing approach that assumes all rays propagate nearly parallel to the optical axis. It does not intrinsically consider diffraction and total reflection; however, these may be introduced *ad hoc*. The approach treats each photon as a ray with transverse position *y* and angle to the longitudinal optical axis α, within an optical system defined by a matrix (*i.e.* an RTM) that transforms an incident ray into an exit ray ,RTMs of compound systems may then be determined by multiplying the RTMs for the individual optical components. Previous RTM analyses of CRLs (Protopopov & Valiev, 1998[@bb17]; Pantell *et al.*, 2003[@bb15]; Poulsen & Poulsen, 2014[@bb16]) show that a single refractive X-ray lenslet can be described by three such components: a free-space propagation by *T*/2, a refracting thin lens with focal length *f* = (where δ is the refractive decrement) and a final free-space propagation by *T*/2. Because *f* is many times larger than *T*, each lenslet behaves like an ideal thin lens with the following transfer matrix,As the CRL is a linear array (*i.e.* a stack) of identical lenslets, its transfer matrix is = . This can be calculated through the matrix eigendecomposition theorem (Poulsen & Poulsen, 2014[@bb16]) (see §S1 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162) for derivation),Within this paraxial treatment, the parameter φ can be expressed as = = . Thus, is the refractive power of the CRL per unit length (Lengeler *et al.*, 1999[@bb13]) while is the critical angle for total external reflection (Schroer & Lengeler, 2005[@bb21]).
The trigonometric terms in equation (3)[](#fd3){ref-type="disp-formula"} imply periodicity with respect to . While attenuation by the lens means that CRLs practically operate within the first half-period (*i.e.* ), optical behavior differs markedly between the thin-lens limit (*i.e.* and correspondingly ) and the general thick lens case (*i.e.* all values of ). This formalism provides both cases in order to give straightforward access to the most common and important optical parameters for the vast majority of CRL geometries.
From equation (3)[](#fd3){ref-type="disp-formula"}, the focal length of the CRL as measured from its exit surface is given by the following two expressions (derived in §S2 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162)), which are identical to those given in the literature (Poulsen & Poulsen, 2014[@bb16]; Lengeler *et al.*, 1999[@bb13]),
2.3.. Ray transfer path {#sec2.3}
-------------------------
In order to predict the attenuation of the rays as they traverse the CRL, the RTM approach must be extended. Specifically, we require an expression for the position and angle of a given ray at the *center* of the *n*th lenslet as a function of its incident state . To this end, we compute the RTM of the CRL *after* the *n*th lenslet and back-propagate by *T*/2,Inserting from equation (3)[](#fd3){ref-type="disp-formula"} and simplifying (see §S3 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162)), are then Within the CRL, all rays have a sinusoidal trajectory that varies with distance *nT* with a period of . The physical aperture of the lenslets bounds this trajectory however, imposing the following criteria for participation in the focusing process,Rays may be excluded due to total reflection by the parabolic lenslet surfaces. In this case, the criteria for participation is \> . However, we note that, at typical X-ray energies, such reflection effects are only significant for lenslet geometries with impractically large values of . As such, we do not consider them further.
2.4.. Attenuation in CRLs {#sec2.4}
---------------------------
The attenuation *U* of a ray passing through a single lenslet at a distance *y* from the optical axis depends on the absorption coefficient μ of the lens material and the local material thickness = . Since the paraxial approximation implies that the variation of *y* and within the lenslet is negligible, the attenuation of the X-rays by the absorbing lenslet can be simply expressed using the Beer--Lambert law,Here, *H* is a box function of width that enforces the criteria in equation (7)[](#fd7){ref-type="disp-formula"}. The cumulative transmission of a ray as it travels through *N* lenslets is then the product of the individual attenuation contributions from each lenslet,The central expression is a geometric sum that can be solved analytically (see §S4 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162)). As all are a linear function of , has a Gaussian dependence on both parameters. Combined with the conditions imposed by *H*, this results in a bounded two-dimensional Gaussian transmission distribution, as shown in Fig. 2[▸](#fig2){ref-type="fig"}.
2.5.. Effective aperture and transmission efficiency of a CRL {#sec2.5}
---------------------------------------------------------------
The spatial acceptance function for a homogeneous and parallel incident beam can then be calculated from by inserting = 0 into equation (6)[](#fd6){ref-type="disp-formula"} (see §S5 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162)). This results in a 1D Gaussian transmission profile in with a root mean square (RMS) value ofFrom this, we can calculate the effective aperture : the diameter of a circular pinhole with the same total transmitted intensity as a (two-dimensional) CRL made from rotationally symmetric paraboloids. For all values of , we findThe transmission efficiency is then given byThese expressions for , and *t* provide a convenient means to compare the attenuation-limited properties of CRLs independent of the optical system they are operating in. Coupled with the expression for \[equation (4)[](#fd4){ref-type="disp-formula"}\], they constitute the simplest way to characterize CRL performance.
3.. CRL-based imaging systems {#sec3}
===============================
3.1.. The imaging condition {#sec3.1}
-----------------------------
The general imaging case describes both condensing (Schroer *et al.*, 2005[@bb20]) and full-field (Lengeler *et al.*, 1999[@bb13]) geometries comprising a source, a lens (either as an objective or condenser) and an image/detection plane. Note that these geometries are mathematically identical; condensing can be seen as full-field imaging of the source with a magnification ratio of less than one (shown schematically in Fig. 3[▸](#fig3){ref-type="fig"}).
In an imaging configuration, a ray originating from the source plane at travels a distance to the objective, where it is transformed by the CRL RTM before travelling a distance to a point on the detector plane. This transformation between and can be expressed aswhere is the matrixwhose components can be expressed in terms of asThe imaging condition implies that = 0, which leads to the general imaging formula (derivation in §S6 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162))The magnification of the imaging system (a positive number) is defined by = = . Combined with equation (16)[](#fd16){ref-type="disp-formula"}, this gives a set of two equations, which can be solved to give exact expressions for magnification and the imaging distances and , Note that and can never be negative in equations (18)[](#fd18){ref-type="disp-formula"} and (19)[](#fd19){ref-type="disp-formula"}, resulting in the following conditions on ,The number of lenses *N* necessary to achieve a given magnification and source-to-detector distance *L* = can be calculated by rewriting the imaging formula \[equation (16)[](#fd16){ref-type="disp-formula"}\] as follows,Despite a factor in the denominator, *N* can nonetheless be evaluated iteratively since (solution by fixed-point method). Furthermore, as *N* must be an integer number, and must be adjusted to fulfill the imaging condition. Consequently, a small deviation of the magnification from the target value must be accepted.
We observe that equation (13)[](#fd13){ref-type="disp-formula"} can also be used to describe the geometry of the back focal plane, which for these imaging systems is located at = . The intensity distribution at the back focal plane is closely related to the Fourier transform of the object in the sample plane, and as such can be used as a means for quantifying micro- and nano-scale periodicity (Ershov *et al.*, 2013[@bb4]). Inserting = into equations (13)[](#fd13){ref-type="disp-formula"}--(15)[](#fd15){ref-type="disp-formula"} gives = 0 andHence, X-rays emerging from the sample plane at the angle α~s~ will converge to position in the back focal plane.
3.2.. Attenuation in imaging systems {#sec3.2}
--------------------------------------
The spatial and angular acceptance of the CRL are defining characteristics of imaging systems, jointly defined by the attenuation properties of the lens and the specific geometry of the system. To find expressions for them, we first use equation (6)[](#fd6){ref-type="disp-formula"} to provide as a function of a ray's position and angle in the source plane, ,Inserting this into \[equation (9)[](#fd9){ref-type="disp-formula"}\] gives the complete acceptance function of the imaging system, which can be rewritten as follows (derivation in §S7 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162)),Notably, this is the product of a prefactor, a Gaussian with RMS and offset coefficient γ describing the system's angular acceptance, another Gaussian with RMS decribing the system's spatial acceptance (*i.e.* vignetting) and a box function representing the system's physical aperture. This physical aperture imposes a sharp cut-off to the acceptances \[see equation (7)[](#fd7){ref-type="disp-formula"}\]. In the same manner as equation (9)[](#fd9){ref-type="disp-formula"}, this is represented by which, from equation (23)[](#fd23){ref-type="disp-formula"}, is well approximated byThe angular acceptance, defined by its RMS and offset , describes the range of angles over which the lens collects radiation emitted from a point on the source plane and ultimately defines the theoretical image resolution achievable by the system. At any point in the field of view (*i.e.* at any ), the system will have a Gaussian angular acceptance with RMS given byThe offset coefficient γ describes the median angle accepted by the lens at a particular point on the sample plane . Notably, in the thin-lens case the characteristic distance is the weighted average of the imaging distance and the period of the sinusoidal divided by . The full derivation for γ is provided in §S7 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162).
The spatial acceptance has RMS and describes the reduction in brightness from the center of the optical axis towards its periphery, *i.e.* the maximum achievable field-of-view of the system. This is defined as the total acceptance of the system integrated across all incident angles . The general expression for the RMS of the vignetting function is given byThe leading term in the thin-lens limit is of the order , meaning that the vignetting does not originate from the material attenuation at the thin-lens limit. Instead, the thin-lens vignetting function, , is defined by the physical aperture of the small stack of lenses,The numerical aperture (NA) is a dimensionless number characterizing the range of angles accepted by the imaging system, and is therefore naturally related to . The parameter is regularly used in the context of visible-light systems where this range is sharply defined by the physical aperture of the system. However, the Gaussian nature of the X-ray acceptance function of typical thick CRLs makes such approaches inappropriate. Instead, the definition used for Gaussian laser systems is used here (Born & Wolf, 1999[@bb1]), in which the NA at a given position is given in terms of the ray angle where the transmission drops to ,Note that this reduces to NA = at the center of the field-of-view.
3.3.. Spatial resolution {#sec3.3}
--------------------------
Analytical expressions for the spatial resolution of the magnified image can now be derived from the magnification and the angular acceptance of the imaging system. Spatial resolution may be defined as the minimum distinguishable distance between two points at the source/sample plane (Born & Wolf, 1999[@bb1]). The degree of blurring of these points due to diffraction and the aberration inherent in lens-based optical systems is described by the point spread function (PSF), which can be calculated from the Fourier transform of the CRL pupil function (Born & Wolf, 1999[@bb1]). Neglecting aberration, the effective pupil function for the center of the image can be derived from equation (24)[](#fd24){ref-type="disp-formula"} by substituting = 0 and = , where is the ray position at the CRL entrance (*i.e.* = 0) (see §S8 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162)),where . For a wavenumber *k* = at X-ray wavelength λ, and ignoring constant prefactors, the point-spread intensity function at the source plane, , is then given byThe PSF has two components: a Gaussian with RMS σ~PSF~ = and a complex term comprising two error functions representing the effect of the physical aperture . The relative contributions of these components are shown in Fig. 4[▸](#fig4){ref-type="fig"}. Notably, in the absence of a physical aperture (*i.e.* if the CRL is limited by the attenuation of the lens material), only the Gaussian component of the PSF remains.
In classical optical systems, the resolution is often defined by the Rayleigh criterion, where two PSFs are regarded distinguishable when the maximum of one PSF coincides with the first minimum of the other (Born & Wolf, 1999[@bb1]). However, this is inappropriate in the case of a Gaussian or near-Gaussian PSF such as for CRLs, where such a minimum may not be present. Instead, we propose that the resolution be defined by the separation distance between two PSFs corresponding to a contrast ratio of *C* (where *C* is small when the contrast is poor). Using equation (32)[](#fd32){ref-type="disp-formula"}, this can be determined by numerically solvingIn the case of absorption-limited (*i.e.* Gaussian) CRLs, this gives a function in terms of λ, and *C*,The value of *C* necessary to distinguish details depends on the sampling statistics. In the case of low-intensity (*e.g.* dynamic) measurements, *C* should be greater than the equivalent for the Rayleigh criteria (approximately 0.26).
3.4.. Chromatic aberration {#sec3.4}
----------------------------
Imaging with a wide energy bandwidth can be advantageous due to the significant increase in photon flux. The bandwidth is ultimately defined by the type of X-ray source and conditioning optics: the raw spectrum from an undulator (*i.e.* the pink beam) is typically of the order of , while monochromators can provide bandwidths from to . As most conditioning schemes use a diffraction-based monochromator, the energy spectrum is typically Gaussian. As such, the spectrum can be defined around a nominal energy and intensity , in terms of the energy perturbation ∊, defined by = and the RMS bandwidth ,Small energy perturbations alter δ from its nominal value according to . Since the focal length of the lenslets depend on δ, CRLs are chromatic by nature. Under an ideal imaging condition at = 0, a ray departing from the center of the sample plane will strike the center of the detector plane, *i.e.* = = 0. At a slightly different photon energy, ; however, the same ray will be displaced in the detector plane, = 0. The position, , at which an incident ray from = 0 strikes the detector, can be approximated by inserting the chromatic expression for δ above into equation (13)[](#fd13){ref-type="disp-formula"}, and Taylor expanding to first order in ∊,where is a distance term for the imaging system given in terms of the nominal values for φ and *f* byThe ray is attenuated depending on its incident angle . Noting from equation (35)[](#fd35){ref-type="disp-formula"} that this angle may be written as = , the spatio-chromatic intensity distribution of the rays on the detector plane, , is then determined from equations (31)[](#fd31){ref-type="disp-formula"} and (34)[](#fd34){ref-type="disp-formula"} as (see §S9 of the [supporting information](http://scripts.iucr.org/cgi-bin/sendsup?ie5162))This distribution (Fig. 5[▸](#fig5){ref-type="fig"}) is consistent with experimental results in the literature (Falch *et al.*, 2016[@bb5]) and illustrates that the chromatic spread of intensity becomes broader as ∊ deviates further from zero.
The point-spread function as a result of this chromatic behavior, , is determined by integrating across ∊, and normalizing by the nominal magnification ,Notably, this is a Laplace distribution described by the characteristic width ,The combined PSF from a chromatic and diffraction-limited system will be a convolution of the two PSFs in equations (31)[](#fd31){ref-type="disp-formula"} and (38)[](#fd38){ref-type="disp-formula"}, shown in Fig. 5[▸](#fig5){ref-type="fig"}. While this does not have a convenient analytical expression, it can be readily solved by numerical methods.
4.. Optimization and the thick-lens limit {#sec4}
===========================================
The analytical expressions for key optical parameters such as spatial resolution, vignetting and aberration provide a foundation for numerical optimization. In the most general case, the aim is to optimize some figure-of-merit function with respect to X-ray wavelenth (λ), CRL geometry (*R, T, N*) and imaging geometry (, or *L*, ) for a given material (δ, μ).
In imaging systems, optimizing NA and can improve the diffraction- and chromatic-limited resolution \[equations (31)[](#fd31){ref-type="disp-formula"} and (38)[](#fd38){ref-type="disp-formula"}\] while increasing transmission efficiency. Because is directly coupled to \[equation (27)[](#fd27){ref-type="disp-formula"}\], it also determines the image vignetting profile. The multidimensional problem of optimizing for a given wavelength can, to a good approximation, be greatly simplified by assuming that *N* is large, *i.e.* . Then equation (26)[](#fd26){ref-type="disp-formula"} becomeswhere is the expression Equation (40)[](#fd40){ref-type="disp-formula"} confirms the well known belief that optical performance (*i.e.* ) is maximized when is large (Snigirev *et al.*, 1996[@bb24]). However, the expression also dictates that be maximized too. Plotting for the allowable range of and some typical values (Fig. 6[▸](#fig6){ref-type="fig"}) shows that *g* reaches a maximum value at large .
Correspondingly, there is a global optimum at = . For typical values of , this thick-lens limit is approximately at = , and is associated with some unusual imaging geometries,As discussed, the angular acceptance reaches its maximum at the thick-lens limit close to = . As the magnification approaches infinity, , and areThe first of these expressions show that the lenslet aperture *Y* is the critical geometrical term that fundamentally limits the optimal angular acceptance of CRL-based imaging systems. Similarly, *Y* is also a fundamental term for the vignetting profile and the chromatic spread. These expressions clearly demonstrate the need to miniaturize the lenslet geometry, as reducing *Y* is the only path to increasing the acceptance for a given set of parameters. Most importantly, the fact that *Y* = implies that there are a range of lenslet geometries that will satisfy the following conditions for optimization,
5.. Discussion {#sec5}
================
Following previous RTM formalisms (Protopopov & Valiev, 1998[@bb17]; Pantell *et al.*, 2003[@bb15]; Poulsen & Poulsen, 2014[@bb16]), this implementation is shown to be a versatile and effective tool for many aspects of the optimization of CRL-based X-ray imaging systems. It is particularly suitable for calculating optical parameters for systems operating in the thick-lens condition. Being a linear formalism, the approach does not account for spherical or higher-order aberrations; however, some effects of diffraction and refraction can be included. Similarly, CRL geometries where the lenslet profile varies along the CRL thickness direction \[*i.e.* adiabatic lenses (Schroer & Lengeler, 2005[@bb21]; Chen *et al.*, 2014[@bb3])\] are not intrinsically accounted for, but can be calculated numerically by substituting a function of, for example, or for *R* or *T*, respectively. In the case of astigmatic lenses such as two-dimensional CRLs produced by interdigitating 1D chips (Simons *et al.*, 2016[@bb23]), the RTM formalism can be readily extended to either two systems or a single system that can calculate the astigmatism analytically.
The optimization demonstrated in this work was largely unconstrained and thus represents a simplified imaging configuration. Applying the practical constraints of real beamline hutch geometries therefore requires additional constraints to the optimization. However, a significant advantage of the RTM formalism is its versatility and mathematical simplicity compared with, for example, Monte Carlo and wave propagation methods. As such, the number of free parameters could be increased without necessarily resulting in impractical computation times.
A major result of this work is that the spatial resolution is globally optimized at the thick-lens limit: = . This can be observed directly in Fig. 7[▸](#fig7){ref-type="fig"}, where is maximized and is minimized at this limit. Furthermore, the field-of-view and imaging geometries also remain practical at this thick-lens limit: is above 200 µm, *L* is 4 m and is 10 mm at the optimum in the example presented in Fig. 7[▸](#fig7){ref-type="fig"}.
The majority of CRL microscopes described in the literature (Simons *et al.*, 2015[@bb22]) operate quite far from this optimimum configuration, implying that there may be significant resolution gains from increasing the focal power of CRLs beyond their current state. This is particularly true in the case of pink-beam X-ray imaging, where the chromatic blurring is minimized at the thick-lens limit.
The formalism indicates that smaller *R* and *T* will offer superior performance, making a strong case for miniaturizing the lenslet geometry. However, the optimization also shows that the global maximum for can be reached by *any* lens geometry that satisfies equation (44)[](#fd44){ref-type="disp-formula"}. Thus, significant gains can still be had by reducing *R* or *T* alone. This is particularly important when manufacturing CRLs, as certain processes may prevent the realisation of specific lenslet geometries. For example, the performance of a CRL comprising indented two-dimensional metal lenslets may be improved by reducing the thickness of the lenslet rather than reducing the parabola radius, thus avoiding the shape-error and aberration that usually accompanies the production of small radii.
This formalism and optimization approach are key for the design of X-ray imaging experiments and instruments. The global optimum at small imaging distances implies that imaging systems with a small footprint will be optically superior. Such systems can offer improved mechanical stability and ultimately higher spatial resolution, and are therefore recommended on the basis of this work. The calculations performed here also shed light on the expected gains from upcoming fourth-generation synchrotron sources, which are characterized by unprecedented brilliance and a smaller energy bandwidth. As demonstrated in Fig. 7[▸](#fig7){ref-type="fig"}, the contribution of chromatic aberration to the PSF is expected to significantly decrease for lenses whose geometry approaches the thick-lens limit. We therefore predict that high-speed three-dimensional X-ray microscopy may be possible without the use of a monochromator.
6.. Conclusions {#sec6}
=================
This work describes a means of calculating and optimizing the optical properties of CRLs and some of the most common CRL-based imaging systems. Specifically:
\(i\) We developed a formalism with closed analytical expressions for key optical parameters, such as focal length, imaging distances, vignetting, spatial resolution and chromatic aberration. The expressions are relevant to the vast majority of X-ray microscopes, and pertain to both full-field imaging and condensing systems as well as thin- and thick-lens conditions.
\(ii\) We carried out an efficient global optimization on the archetypal X-ray imaging system. While this example was geometrically unconstrained, we note that practical limits to, for example, imaging distances can be easily incorporated.
\(iii\) The optimization identified that the optimum spatial resolution for any material and energy will occur at the thick-lens limit of = . Hence, manufacturing CRLs at this thick-lens limit is an opportunity for resolution enhancement.
\(iv\) This implies that the optimum resolution may be reached with many different CRL geometries. This creates a significant opportunity for tailoring CRL geometries for specific manufacturing processes, *e.g.* lithography or indentation, *etc*.
\(v\) Chromatic aberration is reduced near this thick-lens limit. This means that larger energy bandwidths can be used with thick lenses, potentially increasing imaging flux and providing a new opportunity for high-speed dynamic imaging experiments.
Ultimately, we hope that the expressions and optimization approach described here can be applied to improve the performance and design of full-field X-ray microscopes. For existing lens materials and manufacturing technology such as the example given in Fig. 7[▸](#fig7){ref-type="fig"}, one can expect such optimizations to yield improvements in numerical aperture and spatial resolution by a factor of two or more, while simultaneously allowing greater X-ray energy bandwidth and flux. With the advent of many new instruments, both for the laboratory and in synchrotrons, we believe this capability is of significant contemporary relevance.
Supplementary Material
======================
Supporting derivations for key equations in the main text.. DOI: [10.1107/S160057751602049X/ie5162sup1.pdf](https://doi.org/10.1107/S160057751602049X/ie5162sup1.pdf)
We thank Frederik Stöhr and Ray Barrett for useful discussions. In addition, we are grateful to the ESRF for providing beam time on ID06 and Danscatt for travel funding. HFP and SRA acknowledge support from the ERC grant 'd-TXM'. HS acknowledges support from a DFF-FTP individual postdoc grant.
{#fig1}
{ref-type="disp-formula"}, which can be approximated by two pairs of dashed lines corresponding to the entrance (horizontal pair) and exit (slanted pair) of the CRL.](s-24-00392-fig2){#fig2}
{#fig3}
{#fig4}
{#fig5}
{#fig6}
{#fig7}
| {
"pile_set_name": "PubMed Central"
} |
Roy Choudhury S, Johns SM, Pandey S. A convenient, soil‐free method for the production of root nodules in soybean to study the effects of exogenous additives. Plant Direct. 2019;3:1--11. 10.1002/pld3.135
This manuscript was previously deposited as a preprint at preprints.org; <https://www.preprints.org/manuscript/201809.0527/v1> (<https://doi.org/10.20944/preprints201809.0527.v1>).
1. INTRODUCTION {#pld3135-sec-0001}
===============
Nitrogen is an essential element for plant growth, development and productivity. Improving the nitrogen availability to plants results in significant increases in crop yields. Although present in huge quantities in the atmosphere (78% of earth\'s atmosphere), this nitrogen is not available to plants, unless fixed by biological nitrogen fixation (BNF). BNF happens by the activity of specialized groups of bacteria called rhizobia, which exists as symbionts with the roots of leguminous plants in specialized structures called root nodules. Root nodule formation is a sophisticated process that requires strict synchronization of bacterial infection and growth as well as plant organogenesis and nodule development. The successful interactions between the host plant and the soil bacteria of *Rhizobium* spp. begin with the secretion of flavonoids from plant roots. In response, the rhizobia produce lipochito‐oligosaccharides, known as nodulation factors or nod factors (NFs). The secreted NF from symbiotically compatible rhizobia directly bind with and activate the nod factor receptors (NFRs) of plants, which are LysM (Lysine motif)‐containing receptor like kinases (Limpens et al., [2003](#pld3135-bib-0036){ref-type="ref"}; Madsen et al., [2003](#pld3135-bib-0038){ref-type="ref"}; Radutoiu et al., [2003](#pld3135-bib-0056){ref-type="ref"}). NFR activation induces root hair deformation, curling, and consequently entrapment of bacteria in those root hairs. The entrapped bacteria form infection threads, which enters in the root hair cells and elongates from the root hair tips to the inner cells to initiate early infection. Additionally, active NFRs stimulate downstream signaling pathways through nuclear Ca^2+^ oscillations and Ca^2+^ spiking to begin nodule organogenesis from the cortical cells (Gleason et al., [2006](#pld3135-bib-0019){ref-type="ref"}; Tirichine et al., [2006](#pld3135-bib-0072){ref-type="ref"}). All these signaling and organogenesis events are considerably affected by the hormonal balance in plants (Ryu, Cho, Choi, & Hwang, [2012](#pld3135-bib-0059){ref-type="ref"}).
Phytohormones both positively and negatively regulate nodulation and nitrogen fixation in legumes. The positive effects of plant hormones auxins and cytokinins in nodule development have been established for a long time. Auxins are a prerequisite during the development and differentiation of nodule primordia and the formation of the vasculature within the nodules (Kohlen, Ng, Deinum, & Mathesius, [2018](#pld3135-bib-0028){ref-type="ref"}; Takanashi, Sugiyama, & Yazaki, [2011](#pld3135-bib-0068){ref-type="ref"}; Thimann, [1936](#pld3135-bib-0070){ref-type="ref"}). Similarly, cytokinins are responsible for the cortical cell division, differentiation, and nodule organogenesis (Frugier, Kosuta, Murray, Crespi, & Szczyglowski, [2008](#pld3135-bib-0016){ref-type="ref"}; Gonzalez‐Rizzo, Crespi, & Frugier, [2006](#pld3135-bib-0021){ref-type="ref"}; Reid et al., [2017](#pld3135-bib-0057){ref-type="ref"}). In addition to auxins and cytokinins, gibberellins (gibberellic acid, GA) are also involved during regulation of nodulation likely via their cross talk with cytokinin signaling pathways (Maekawa et al., [2009](#pld3135-bib-0039){ref-type="ref"}). Conversely, stress‐related hormones such as jasmonic acid (JA), salicylic acid (SA), and abscisic acid (ABA) typically reduce nodulation by disrupting NF‐induced Ca^2+^ spiking and downstream signaling pathways (Martinez‐Abarca et al., [1998](#pld3135-bib-0040){ref-type="ref"}; Nakagawa & Kawaguchi, [2006](#pld3135-bib-0046){ref-type="ref"}; Phillips, [1971](#pld3135-bib-0055){ref-type="ref"}).
Nodulation is an energy‐demanding process, therefore to control the number of nodules, legumes have evolved a systemic auto‐regulation of nodulation (AON) as well as local hormonal inhibition of nodulation, which are considered the negative feedback systems. The molecular mechanism of AON has been actively investigated using different supernodulation mutants, such as *hyper nodulation and aberrant root 1 (har1), super numeric nodules 1 (sunn)*, and *nodule autoregulation receptor kinase (nark)* in *Lotus japonicus, Medicago truncatula*, and *Glycine max*, respectively (Ferguson et al., [2010](#pld3135-bib-0014){ref-type="ref"}; Krusell et al., [2002](#pld3135-bib-0030){ref-type="ref"}; Nishimura et al., [2002](#pld3135-bib-0048){ref-type="ref"}; Oka‐Kira & Kawaguchi, [2006](#pld3135-bib-0050){ref-type="ref"}; Oka‐Kira et al., [2005](#pld3135-bib-0051){ref-type="ref"}; Searle et al., [2003](#pld3135-bib-0061){ref-type="ref"}). Numerous studies suggest that auxin, JA, and brassinosteroids (BRs) modulate AON signaling pathways (Kinkema & Gresshoff, [2008](#pld3135-bib-0027){ref-type="ref"}; Nakagawa & Kawaguchi, [2006](#pld3135-bib-0046){ref-type="ref"}; Oka‐Kira et al., [2005](#pld3135-bib-0051){ref-type="ref"}; Terakado, Yoneyama, & Fujihara, [2006](#pld3135-bib-0069){ref-type="ref"}) whereas ABA, JA, ethylene, and SA appear to act as during local inhibitory regulation of nodulation (Biswas, Chan, & Gresshoff, [2009](#pld3135-bib-0005){ref-type="ref"}; Ding et al., [2008](#pld3135-bib-0012){ref-type="ref"}; Oldroyd, Engstrom, & Long, [2001](#pld3135-bib-0052){ref-type="ref"}; Penmetsa & Cook, [1997](#pld3135-bib-0054){ref-type="ref"}; Sun et al., [2006](#pld3135-bib-0065){ref-type="ref"}). Coordinated action of the hormone levels and signaling controls nodule organogenesis and mature nodule development.
Previous studies on the hormonal control of nodulation are based on physiological approaches using a variety of leguminous species and exogenous application of phytohormones to study their effect on nodule formation. For example, exogenous application of cytokinins and auxins to pea root cortical explants induced cell proliferation required for root nodule formation (Libbenga, van Iren, Bogers, & Schraag‐Lamers, [1973](#pld3135-bib-0034){ref-type="ref"}). On the other hand, exogenous ABA reduced the number of root nodules by inhibiting the cortical cell divisions during nodule organogenesis (Phillips, [1971](#pld3135-bib-0055){ref-type="ref"}). GA, an important growth regulator, also modulates root nodule formation in legumes by exogenous application (Maekawa et al., [2009](#pld3135-bib-0039){ref-type="ref"}). SA, a key molecule in plant disease resistance, was shown to inhibit the indeterminate nodules of *Vicia sativa*, but not the determinate nodules of *Lotus japonicus* after exogenous application (van Spronsen et al., [2003](#pld3135-bib-0063){ref-type="ref"}). Although each hormone had a characteristic physiological effect, it was evident that different hormones may also follow additive, synergistic, or antagonistic interactions to regulate nodule formation.
The availability of excellent mutant populations in plants such as *L. japonicus* and *M. truncatula* provided genetic evidence for the effect of hormones on nodule formation. These plants serve as useful models due to their modest genome sizes, short seed‐to‐seed generation time, high plant transformation efficiency, and the formation of a restricted number of root nodules. These traits are also useful for performing highly controlled in vitro assays with multiple exogenous additives, which has led to several key discoveries (Bensmihen, [2015](#pld3135-bib-0004){ref-type="ref"}; Maekawa et al., [2009](#pld3135-bib-0039){ref-type="ref"}; Nakagawa & Kawaguchi, [2006](#pld3135-bib-0046){ref-type="ref"}; Stacey, McAlvin, Kim, Olivares, & Soto, [2006](#pld3135-bib-0064){ref-type="ref"}; Sun et al., [2006](#pld3135-bib-0065){ref-type="ref"}). Conversely, it is difficult to perform similar assays with exogenous additives in crop legumes such as soybean, due to their large stature, long life cycle, the formation of a huge number of root nodules, and the requirement of soil for nodule formation. To circumvent these problems, split‐root system was used in soybean to study the effects of various exogenous variable during rhizobia‐legume symbioses (Chaillou, Rideout, Raper, & Morot‐Gaudry, [1994](#pld3135-bib-0009){ref-type="ref"}; Gil‐Quintana et al., [2013](#pld3135-bib-0018){ref-type="ref"}; Lin, Gresshoff, & Ferguson, [2012](#pld3135-bib-0037){ref-type="ref"}; Singleton & Bohlool, [1984](#pld3135-bib-0062){ref-type="ref"}). Most recently, a new split‐root system was developed for continuous monitoring of soybean roots throughout the whole experiment after rhizobial infection (Hidalgo, Ruiz‐Sainz, & Vinardell, [2018](#pld3135-bib-0024){ref-type="ref"}). Although these protocols are useful, it is still technically challenging to do some of these experiments and is difficult to apply to a large plant population.
The goals of this study were to identify suitable media and growth conditions for efficient nodule development under in vitro conditions in soybean *G. max*, which is an important crop but not amenable to the standard plate‐based assays used to study nodule formation in *Medicago* or *Lotus* sp. In addition, the effect of each of the major phytohormones was analyzed on nodule development under a standard set of conditions. The results presented in the following sections describe a set of optimum growth and treatment conditions for soil‐free soybean nodulation and effects of phytohormones on it, which will be useful for the community at large.
2. MATERIALS {#pld3135-sec-0002}
============
Soybean seeds (*William 82*)Pots (2‐gallon) filled with soilrite (BM 7, 35% Bark mix, Berger, Saint‐Modeste, QC, Canada)Rock wool, cubic size (Hummert International, Earth City, MO, USA)Petri dishes (100 × 15 mm) (Corning^™^ Falcon^™^, Fisher Scientific, USA)Trays (2 square feet) (T.O. Plastics, Inc., Clearwater, MN, USA)Pot (85 × 85 mm) (T.O. Plastics, Inc.)Soilrite (Vermiculite:Perlite:Sand in 3:1:1 ratio) containing pots (Therm‐O‐Rock East Inc, New Eagle, PA, USA)Beakers (250 ml)Germination paper (Anchor Paper Company, St Paul, MN, USA)Light chamber under 16 h light/8 h dark conditions (100 μmol m^−2^ s^−1^) with 50% humidity and 25°C (day/night) temperatureGreenhouse with 50% humidity and 31/22°C (day/night) temperature.Nitrogen‐free nutrient solution (1000× stock solution, pH 6.5, sterile)
MacronutrientKSO~4~(87.135 gm/L)KH~2~PO~4~(68.045 gm/L)CaCl~2~, 2H~2~O(147.01 gm/L)MgSO~4~, 7H~2~O(123.24 gm/L)Fe‐EDTAEDTA(372.24 gm/L)FeSO~4~, 7H~2~O(278.02 gm/L)MicronutrientZnSO~4~, 7H~2~O(0.143 gm/L)CuSO~4~, 5H~2~O(0.030 gm/L)MnSO~4~, H~2~O(0.845 gm/L)H~3~BO~3~(1.855 gm/L)(NH~4~)~6~Mo~7~O~24~·4H~2~O(0.099 gm/L)Co(NO~3~)~2~(0.003 gm/L)NiSO~4~(0.026 gm/L)**13.** Vincent\'s rich medium (per liter)K~2~HPO~4~(0.5 gm)NaCl(0.1 gm)MgSO~4~, 7H~2~O(0.2 gm)Yeast Extract(0.4 gm)Mannitol(10.0 gm)pH 6.8
14Rhizobium strain: *Agrobacterium rhizogenes* strain K599 (USDA136)15Plant hormones:a(+/−) Abscisic acid (ABA), (Caisson Labs, Smithfield, UT, USA)bIndole‐3‐Acetic Acid (IAA), (Caisson Labs)c6‐Benzylaminopurine (BAP), (Caisson Labs)dEpibrassinolide ≥ 85% (BR), (Sigma‐Aldrich, Saint Louis, MO, USA)eGibberellic Acid (GA~3~), (Caisson Labs)fMethyl jasmonate, 95% (JA), (Sigma‐Aldrich)gMethyl salicylate *ReagentPlus* ^®^, ≥99% (GC) (SA), (Sigma‐Aldrich)
3. METHODS {#pld3135-sec-0003}
==========
3.1. Plant growth {#pld3135-sec-0004}
-----------------
Soybean seeds were grown in the greenhouse in two‐gallon pots filled with soilrite (BM 7 35%) for 12--14 days at (25/24°C (day/night) temperature, 40%--95% humidity, and ∼300--450 μmol/m^2^/s light intensity depending on the weather and time of the day. Twenty‐five to thirty seeds were grown in each pot, with equal spacing (Figure [1](#pld3135-fig-0001){ref-type="fig"}a). There should be no overlap of the seeds. The spacing is important for proper growth of the seedlings.
{#pld3135-fig-0001}
3.2. Hairy root formation {#pld3135-sec-0005}
-------------------------
Hairy roots were developed as per our previously established protocols (Roy Choudhury & Pandey, [2013](#pld3135-bib-0058){ref-type="ref"}). Briefly, three to four sterilized rock wool cubes were placed in one Petri dish and a hole was created in the middle of the rock wool by using 1 ml pipette tips. Six milliliter nitrogen‐free nutrient solution was applied in the hole of each rock wool cube. The trifoliate leaf from the upper part of a 2 weeks old soybean plant was cut using a razor blade and put in the hole of the rock wool cube (Figure [1](#pld3135-fig-0001){ref-type="fig"}b). Eight Petri dishes (\~30 trifoliate) could be placed in each two square feet tray.
The trays were covered with a plastic cover and kept in the light chamber with the light intensity 100 μmol m^−2^ s^−1^. Plants were grown for 2 weeks for generation of hairy roots. If needed, the nitrogen‐free nutrient solution was added in the rock wool cubes. After 2 weeks, the rock wool cubes were carefully removed and the plants with hairy roots were transferred to the 85 mm × 85 mm pots containing soilrite (vermiculite:perlite:sand at a 3:1:1 ratio). The plants were grown for one more week in the greenhouse (Figure [1](#pld3135-fig-0001){ref-type="fig"}c).
3.3. Rhizobium treatment {#pld3135-sec-0006}
------------------------
The appropriate *Rhizobium* strain was cultured in Vincent\'s rich medium containing chloramphenicol (20 μg/ml) for 3--4 days. The culture was spun down and resuspended in equal volume of nitrogen‐free nutrient solution, with the final concentration adjusted to 0.08 at OD~600~. Ten milliters of rhizobium culture was poured surrounding the roots in each pot and plants were grown for 2--4 days in the green house.
3.4. Hormone treatment {#pld3135-sec-0007}
----------------------
After rhizobium inoculation, plants were removed from the pot at specific time points (Figure [1](#pld3135-fig-0001){ref-type="fig"}d), wrapped in a roll between two sheets of germination paper pre‐wet with the nitrogen‐free nutrient solution. Six plants (in individual rolls) were kept in one 250 ml beaker containing 100 ml of nitrogen‐free nutrient solution (Figure [1](#pld3135-fig-0001){ref-type="fig"}e). Specific hormones were added to each beaker one time every week, immediately after transfer of the plants to the media. The 100 ml nitrogen‐free nutrient solution level was maintained by monitoring the solution level on alternate days and adding media as required. Control and treatment plants were treated similarly to avoid any non‐biological variation. Multiple replicates were performed for each treatment. Nodule numbers were counted after 4 weeks of growth (Figure [1](#pld3135-fig-0001){ref-type="fig"}f). The nodules were divided into small (\<0.5 mm in diameter), medium (0.5--2 mm in diameter), and large (\>0.2 mm in diameter) categories.
3.5. Statistical analysis {#pld3135-sec-0008}
-------------------------
All experiments were repeated two times independently and the data were averaged. Each replicate consisted of 24 plants per treatment per concentration. Statistical significance of results was calculated using Student\'s *t* test with a *p*‐value threshold of less than 0.05.
4. RESULTS AND DISCUSSION {#pld3135-sec-0009}
=========================
4.1. Optimization of soil‐free nodule development in soybean {#pld3135-sec-0010}
------------------------------------------------------------
Symbiotic nitrogen fixation in nodules plays a key role in the maintenance of soybean seed production. While in plants such as *Medicago*, their small stature allows for growth under in vitro conditions on sterile media plates, the large size of soybean plants precludes such a possibility. The study of soybean nodule development in soil by using different additives is relatively hard, inaccurate, and expensive. To overcome these issues, soil‐free nodule production under in vitro condition is a suitable choice. We optimized a method where the nodules were allowed to develop in germination papers, after initial infection in soil (Figure [1](#pld3135-fig-0001){ref-type="fig"}). Briefly, trifoliate leaves with a stalk were cut from 2‐week‐old plants grown on soil and inserted in the middle of sterilized rock wool cubes (Govindarajulu et al., [2009](#pld3135-bib-0022){ref-type="ref"}; Libault et al., [2009](#pld3135-bib-0033){ref-type="ref"}). The rock wool in cubes causes abrasion, which promotes the development of hairy roots, which are well developed after 2 weeks of growth. After this time plants with hairy roots were transferred to small pots (85 mm × 85 mm) containing soilrite (vermiculite:perlite:sand in a 3:1:1 ratio) and allowed to grow for 1 week. For nodule development, plants in pots were infected with compatible rhizobium strains. Infection of plants in soilrite was critical, because infecting plants in germination papers by direct application of rhizobia would result in uncontrolled bacterial infection on the paper itself. Plants were transferred 72 h post‐infection between the two sheets of pre‐wet germination paper and were arranged in rolls. One plant was placed per roll and each 250 ml beaker can accommodate six rolls. The plants were allowed to grow in nitrogen‐free nutrient solution and nodule number was counted after 4 weeks of further growth. For nodule development in soilrite, the plants can continue to grow in pots after infection, but will require one ½ gallon pot per plant. Achieving uniform root growth and nodule formation in soybean under in vitro conditions has been a challenge as it shows enormous inconsistencies in nodule numbers. In our experiments, the time of bacterial infection and the duration of subsequent growth of plants in soilrite were critical for efficient nodule development at later stages. To optimize the conditions for consistent, reproducible results, we transferred plants from soilrite to germination paper rolls at different time points after rhizobium infection and counted the nodules after 4 weeks of growth. Plant relocation time after rhizobium infection had a huge effect on nodule formation (Figure [2](#pld3135-fig-0002){ref-type="fig"}). No nodules were formed if we transferred the plants from 0 to 24 h after rhizobium infection, whereas \~2, 8, and 12 nodules on average were formed per plant if the plants were transferred 48, 72, and 96 h after rhizobium infection, respectively. Based on these results, all through our experiments, we have transferred the plants from soilrite to germination paper rolls 72 h post‐infection as it generates a reasonable number of nodules needed for any comparative analysis.
{#pld3135-fig-0002}
We are cognizant of the fact that the need to grow roots in soilrite for up to 72 h post‐infection excludes the possibility of assaying the effects of various additives during the early stages of infection such as on root hair curling, or infection thread formation. The method described in this research is therefore suitable for the study of additives at the later stages of nodule development only. However, as detailed in the next sections, it does result in reproducible and consistent effects of exogenous additives on nodule developments and is extremely resource‐ and cost‐effective.
4.2. Optimization of hormone treatment time {#pld3135-sec-0011}
-------------------------------------------
For this study, we focused on assaying the effects of various plant hormones as exogenous additives, as these have a significant influence not only on nodule initiation but also on organogenesis. Similar assays can be optimized for other additives such as salt, mannitol (for osmotic stress) etc. as needed. The hormone treatments were started when the plants were transferred from soilrite, and germination paper rolls containing plants were transferred to either control or treatment media. An obvious effect on nodule development was observed for each of the plant hormones.
4.3. Effect of ABA on root nodule development {#pld3135-sec-0012}
---------------------------------------------
Abscisic acid regulates multiple aspects of rhizobia‐legume symbiosis (Bano & Harper, [2002](#pld3135-bib-0001){ref-type="ref"}; Bano, Harper, Auge, & Neuman, [2002](#pld3135-bib-0002){ref-type="ref"}; Cho & Harper, [1993](#pld3135-bib-0010){ref-type="ref"}; Phillips, [1971](#pld3135-bib-0055){ref-type="ref"}). Because the addition of exogenous ABA causes a marked increase in the endogenous ABA levels in plants, the application of ABA to media in which plants are growing is a suitable system to study its effect on nodulation. In *L. japonicus*, the number of root nodules is reduced at higher ABA concentrations, and increased in the presence of abamine (which causes lower ABA concentrations) suggesting its inhibitory role during nodule development (Suzuki et al., [2004](#pld3135-bib-0067){ref-type="ref"}). In soybean, exogenous ABA decreased nodule number in both the wild‐type and a supernodulation mutant (Bano & Harper, [2002](#pld3135-bib-0001){ref-type="ref"}; Bano et al., [2002](#pld3135-bib-0002){ref-type="ref"}; Cho & Harper, [1993](#pld3135-bib-0010){ref-type="ref"}).
We evaluated the effect of ABA on soybean nodulation in a concentration‐dependent manner using ABA concentrations ranging from 2 to 25 μM; concentrations lower than 2 μM were ineffective under our growth conditions. An equimolar amount of ethanol was used as a control in all assays. Nodule formation showed extreme sensitivity to exogenously added ABA. On average, seven nodules were formed per plants after 4 weeks of growth under control conditions, which was reduced by 15%--30% in presence of 2--15 μM ABA (Figure [3](#pld3135-fig-0003){ref-type="fig"}). It was noticeable that plants treated with ABA had fewer large (\>2 mm in diameter) and medium (0.5--2 mm in diameter) compared to the control plants but a comparable number of small nodules (\<0.5 mm in diameter) were produced (Table S1). Concentrations higher than 15 μM severely affected nodule formation, with the nodule numbers reduced by more than 60% per plant in the presence of 20--25 μM ABA. At these higher concentrations, all types of nodules (large, medium, and small) were significantly affected. These data confirm the inhibitory effects of exogenous ABA on soybean nodule formation, and by extension of other abiotic stresses, which cause an increase in ABA concentrations in planta.
{#pld3135-fig-0003}
4.4. Effect of auxin (IAA) and cytokinin (BAP) on nodule development {#pld3135-sec-0013}
--------------------------------------------------------------------
Several studies have demonstrated that auxin and cytokinin intricately control root nodule formation. IAA (indole‐3‐acetic acid), a native auxin in plants, derived from the phenylpropanoid biosynthetic pathway is a key member of the auxin family. The root nodules have a higher IAA content than uninfected root tissues, suggesting a role for IAA in nodule development (Thimann, [1936](#pld3135-bib-0070){ref-type="ref"}). The higher IAA content of infected roots promotes root cells to undergo cell division, elongation, differentiation, and vascular bundle formation to develop nodules (Boivin, Fonouni‐Farde, & Frugier, [2016](#pld3135-bib-0006){ref-type="ref"}; Kuppusamy et al., [2009](#pld3135-bib-0031){ref-type="ref"}; Nagata & Suzuki, [2014](#pld3135-bib-0045){ref-type="ref"}; Suzaki et al., [2012](#pld3135-bib-0066){ref-type="ref"}). Moreover, rhizobia alter the root auxin balance, which is a prerequisite for nodule formation (Boot, van Brussel, Tak, Spaink, & Kijne, [1999](#pld3135-bib-0007){ref-type="ref"}; Hirsch, Bhuvaneswari, Torrey, & Bisseling, [1989](#pld3135-bib-0025){ref-type="ref"}; Mathesius et al., [1998](#pld3135-bib-0041){ref-type="ref"}; Pacios‐Bras et al., [2003](#pld3135-bib-0053){ref-type="ref"}; Wasson, Pellerone, & Mathesius, [2006](#pld3135-bib-0075){ref-type="ref"}). Besides auxin content, rhizobia‐legume symbiosis is also regulated by shoot‐to‐root auxin transport, which is an active transport process involving auxin efflux protein complexes (van Noorden, Ross, Reid, Rolfe, & Mathesius, [2006](#pld3135-bib-0049){ref-type="ref"}). Overall, it is well established that the changes in auxin accumulation and transport, both are essential for lateral root development, nodule primordium activation, and nodule organogenesis (Mathesius et al., [1998](#pld3135-bib-0041){ref-type="ref"}; Suzaki et al., [2012](#pld3135-bib-0066){ref-type="ref"}; Takanashi et al., [2011](#pld3135-bib-0068){ref-type="ref"}).
To test the effect of exogenous auxins on soybean nodule formation, plants were treated with different IAA concentrations ranging from 10 nM to 100 μM. IAA at 10 nM caused the most obvious phenotypic differences in the nodule number, although all treatment conditions resulted in higher nodule number per plant. Approximately four times more nodules were formed per plant at 10 nM IAA compared to control plants with a huge increase in the number of small‐ and medium‐size nodules (Figure [4](#pld3135-fig-0004){ref-type="fig"}a). Two times more nodules were formed compared to control conditions in the presence of 1 μM IAA, whereas a modest increase in nodule numbers was observed in the presence of 100 μM IAA in the media. These results indicate that even at a very high concentration, auxin still has a limited but positive effect on nodule formation. Furthermore, 10 nM exogenous auxin is an optimal concentration for increased nodule numbers in soybean.
{#pld3135-fig-0004}
Extensive research on *L. japonicus* and *M. truncatula* have demonstrated that cytokinins (CKs) are key players in the regulation of rhizobium infection and nodule development (Frugier et al., [2008](#pld3135-bib-0016){ref-type="ref"}; Gonzalez‐Rizzo et al., [2006](#pld3135-bib-0021){ref-type="ref"}; Tirichine et al., [2007](#pld3135-bib-0073){ref-type="ref"}). The activation of NF signaling pathway rapidly induces CK accumulation and response in hairy roots (Buhian & Bensmihen, [2018](#pld3135-bib-0008){ref-type="ref"}; Gamas, Brault, Jardinaud, & Frugier, [2017](#pld3135-bib-0017){ref-type="ref"}; Murray et al., [2007](#pld3135-bib-0044){ref-type="ref"}). The exogenous application of CKs also promotes cortical cell divisions and the expression of early nodulation markers in different legumes (Bauer, Ratet, Crespi, Schultze, & Kondorosi, [1996](#pld3135-bib-0003){ref-type="ref"}; Jimenez‐Zurdo, Frugier, Crespi, & Kondorosi, [2000](#pld3135-bib-0026){ref-type="ref"}; Mathesius, Weinman, Rolfe, & Djordjevic, [2000](#pld3135-bib-0042){ref-type="ref"}; Murray et al., [2007](#pld3135-bib-0044){ref-type="ref"}; Tirichine et al., [2007](#pld3135-bib-0073){ref-type="ref"}). To study the effect of exogenous applications of cytokinins on soybean nodulation, we included different concentrations of BAP (6‐Benzylaminopurine) (50 nM, 100 nM, 250 nM, 1 μM, and 10 μM) in the media when the plants were transferred from soilrite to germination paper rolls. We detected a clear, concentration‐dependent effect of BAP on nodulation. Approximately 2.5 times more nodules were formed in the presence of 50 nM BAP, whereas a modest increase was observed in the presence of 100 M BAP, compared to the control conditions (Figure [4](#pld3135-fig-0004){ref-type="fig"}b). Interestingly, BAP concentrations higher than 250 nM were inhibitory and lead to the development of fewer nodules compared to control media grown plants. For example, \~80% less nodule developed in the presence of 10 μM BAP. These data suggest that nodule formation in soybean strongly regulated by precisely controlled CKs level.
4.5. Effect of GA and BR on root nodule development {#pld3135-sec-0014}
---------------------------------------------------
Gibberellic acid is one of the vital growth regulators in higher plants. Several studies have highlighted the involvement of GA in the regulation of the rhizobium‐legume symbiotic interaction (Ferguson, Ross, & Reid, [2005](#pld3135-bib-0015){ref-type="ref"}; Lievens et al., [2005](#pld3135-bib-0035){ref-type="ref"}; Maekawa et al., [2009](#pld3135-bib-0039){ref-type="ref"}; McAdam, Reid, & Foo, [2018](#pld3135-bib-0043){ref-type="ref"}). Interestingly, both positive and negative effects of GA have been reported in previous studies. For examples, one study reported the inhibitory effects of an exogenous application of potassium gibberellate on nodule formation in *Phaseolus vulgaris* (Thurber, Douglas, & Galston, [1958](#pld3135-bib-0071){ref-type="ref"}). On the other hand, nodules aborted in pea mutant (*na‐1*) which was deficient in GA~3~, but were re‐established by the application of exogenous GA~3~ (Ferguson et al., [2005](#pld3135-bib-0015){ref-type="ref"}) implying requirement of GA for nodule formation. This discrepancy in the results could be either the concentration of GA used or the specific plant species used in the assays. To assess the effect of GA~3~ on soybean nodule formation, plants post‐infection were treated with different GA~3~ concentrations (10 nM to 1 μM) and compared with control media with no added GA~3.~ Plants treated with 10 nM and 100 nM GA~3~ showed considerably increased nodule formation with \~40% and \~20% more nodules, respectively. Interestingly, most of the nodules in these treatments were small (Table S1). However, similar to what was observed for cytokinins, at higher concentrations (1 μM) of GA~3~, the nodule formation was severely affected and \~70% fewer nodules were formed on treated roots compared with control media (Figure [5](#pld3135-fig-0005){ref-type="fig"}a). These data suggest that the reported discrepancies in the previous publications could be due to the different GA concentrations used in different assays. These observations suggest that the endogenous GA concentration in plants is tightly regulated to achieve effective nodulation.
{#pld3135-fig-0005}
Brassinosteroids play pivotal roles during many aspects of plant growth and development (Clouse & Sasse, [1998](#pld3135-bib-0011){ref-type="ref"}; Wei & Li, [2016](#pld3135-bib-0076){ref-type="ref"}). BRs affect both cell proliferation and cell elongation to control shoot and root lengths and hypocotyl growth in plants (Gonzalez‐Garcia et al., [2011](#pld3135-bib-0020){ref-type="ref"}; Tong et al., [2014](#pld3135-bib-0074){ref-type="ref"}). Although the effects of BRs are well documented on root and shoot growth, their effects on nodule development in different legumes are not as well described. In one example, BRs affected nodule development by reducing lateral root numbers in pea as is evident from the BR synthesis mutants *lk* and *lkb* and the BR response mutant *lka* (Ferguson et al., [2005](#pld3135-bib-0015){ref-type="ref"}). To test the effect of BRs on soybean nodule development, plants were treated with media containing different brassinolide (active BR) concentrations ranging from 10 nM to 1 μM. In comparison to plant grown on control media, a concentration‐dependent increase in nodule numbers was observed in plants grown on BR containing media. Highest nodule numbers were seen in response to 100 nM brassinolide, where ∼85% more nodules were present compared to the control media roots (Figure [5](#pld3135-fig-0005){ref-type="fig"}b). At lower (10 nM) and higher (1 μM) concentrations of exogenous BL, ∼50% more nodules were formed, suggesting the BRs are positive regulators of nodule formation.
4.6. Effect of SA and JA on root nodule development {#pld3135-sec-0015}
---------------------------------------------------
Salicylic acid strongly affects nodule formation at early stages of nodulation (Sato et al., [2002](#pld3135-bib-0060){ref-type="ref"}). The exogenous application of SA resulted in both reduced and delayed nodule formation on *Medicago* roots inoculated with wild‐type *S. meliloti*. Moreover, exogenous SA that inhibited nodulation also strongly reduced the growth of the bacterial symbiont (Martinez‐Abarca et al., [1998](#pld3135-bib-0040){ref-type="ref"}). The effect of SA was additionally documented in different legumes. For example, the inhibition of nodule formation after exogenous SA treatment was observed in plants like vetch (*Vicia sativa*), pea (*Pisum sativum*), and white clover (*Trifolium repens*) (van Spronsen et al., [2003](#pld3135-bib-0063){ref-type="ref"}). This study shows that the exogenous SA inhibits indeterminate but not determinate nodulation (van Spronsen et al., [2003](#pld3135-bib-0063){ref-type="ref"}), which is inconsistent with two other previous reports (Lian, Zhou, Miransari, & Smith, [2000](#pld3135-bib-0032){ref-type="ref"}; Sato et al., [2002](#pld3135-bib-0060){ref-type="ref"}), where they showed that exogenous application of SA to soybean, which forms determinate nodules, reduced nodulation. Another report demonstrated that SA enhances the efficiency of nitrogen fixation and assimilation in *Cicer arietinum* (Hayat, Hayat, Alyemeni, & Ahmad, [2014](#pld3135-bib-0023){ref-type="ref"}). To examine the role of exogenous SA in soybean nodulation in our system, we used different concentrations of SA ranging from 10 μM to 1 mM in the exogenous media. All SA‐treated soybean plants exhibited a significantly higher nodule number, ranging from 57% to 76% compared to the control media (Figure [6](#pld3135-fig-0006){ref-type="fig"}a). The differences were more prominent in case of higher concentration of SA treatment (100 μM to 1 mM). Our data thus suggest that exogenous SA levels positively regulate nodulation in soybean at these concentrations.
{#pld3135-fig-0006}
Jasmonic acid negatively regulates plants' response to the rhizobial bacterial signal, NF (Sun et al., [2006](#pld3135-bib-0065){ref-type="ref"}). Primarily, JA inhibits nodule formation by suppressing calcium spiking and the frequency of calcium oscillations to modulate the NF‐induced gene expression (Sun et al., [2006](#pld3135-bib-0065){ref-type="ref"}). In addition, even in the shoots, the AON pathway is modulated by JA (Kinkema & Gresshoff, [2008](#pld3135-bib-0027){ref-type="ref"}). However, a recent study by suppression of allene oxide cyclase in *M. truncatula*, an enzyme involved in committed step in JA biosynthesis, suggested that jasmonates are not involved in the development and function of root nodules (Zdyb et al., [2011](#pld3135-bib-0077){ref-type="ref"}). To examine if exogenous JA has any effect on soybean nodulation, we tested its effects at different concentrations (10 μM, 100 μM, and 1 mM). Surprisingly, at low concentration (10 μM), exogenous JA showed a positive effect and approximately twice as many nodules were formed compared to the control media. However, the nodule number was significantly inhibited (up to 95%) after increasing the JA concentrations from 10 to 100 μM. No nodules were formed in the presence of 1 mM JA (Figure [6](#pld3135-fig-0006){ref-type="fig"}b). These data suggest that regulation of nodule formation by JA is complex and is highly dependent on the exogenous concentrations. As hormones such as JA and SA not only modulate the signaling pathways in plants but are also a core part of plant--microbe interaction (Durner, Shah, & Klessig, [1997](#pld3135-bib-0013){ref-type="ref"}; Koornneef & Pieterse, [2008](#pld3135-bib-0029){ref-type="ref"}; Niki, Mitsuhara, Seo, Ohtsubo, & Ohashi, [1998](#pld3135-bib-0047){ref-type="ref"}), it is expected that different concentrations may have altered effects on growth and development versus survival.
5. CONCLUSION {#pld3135-sec-0016}
=============
Nodule development in legumes directly affects nitrogen fixation efficiency during plant growth. Here, we present a method for determining the effects of ABA, auxin, BAP, GA, BR, SA, and JA on soybean nodulation that is rapid, accurate, technically simple, and requires minimal resources. This method provides several advantages over other methods as these approaches do not require time‐consuming additional steps such as changing solvents and maintaining hormonal concentrations day‐by‐day nor use of large containers, which require large quantities of hormones, space, and tedious handling. Continued manipulations often increases potential technical errors. The method can also be applied to roots coming out of seeds, if generating the transgenic hairy roots is not a requirement. Moreover, the standardization of hormone concentrations and the description of resultant phenotypes will support further targeted studies, and in combination with additional genetic and genomic tools being developed in multiple labs, will greatly increase its use for the study of the effects of exogenous factors affecting nodulation in soybeans as well as in other larger legumes. Similarly, for the study of gene silencing, overexpression or gene editing on nodule development, Agrobacterium cells (*K599*) expressing the appropriate gene constructs can be used in this method by mixing it with a nitrogen‐free nutrient solution, which will result in the generation of transgenic soybean hairy roots.
AUTHOR CONTRIBUTION {#pld3135-sec-0019}
===================
Conceptualization, Sona Pandey; Data curation, Swarup Roy Choudhury and Sarah Johns; Formal analysis, Swarup Roy Choudhury; Funding acquisition, Sona Pandey; Investigation, Swarup Roy Choudhury; Methodology, Swarup Roy Choudhury and Sarah Johns; Project administration, Sona Pandey; Supervision, Sona Pandey; Writing -- original draft, Swarup Roy Choudhury; Writing -- review & editing, Sona Pandey.
Supporting information
======================
######
######
Click here for additional data file.
######
######
Click here for additional data file.
Research in the author\'s lab is supported by NIFA/AFRI (2015‐67013‐22964) and NSF (IOS‐1557942 and MCB‐1714693) grants. SMJ was supported by and NSF‐REU grant (DBI‐1659812).
The authors declare no conflict of interest associated with the workdescribed in this manuscript.
| {
"pile_set_name": "PubMed Central"
} |
**Citation:** Morita T, Rahman A, Hasegawa T, Ozaki A, Tanimoto T. The potential possibility of symptom checker. Int J Health Policy Manag. 2017;6(10):615--616. doi:10.15171/ijhpm.2017.41
Dear Editor, {#s1}
============
The access to medical care is unacceptably low worldwide in spite of the increasing demand for medical care due to population aging and increasing burden of non-infectious diseases. The probability of patients receiving at least one medicine for secondary prevention of cardiovascular disease was 19.8% in low-income countries, 30.7% in low- and middle-income countries, and 54.9% for upper-middle-income countries.^[@R1]^ Especially, low- to middle-income countries are seriously in short of professional medical staff. For example, in Bangladesh, the doctor-population and the doctors-nurse ratio is 1:12 690 and 2.5:1, respectively, which is among the lowest group in the world.^[@R2]^ In such resource-limited countries, patients cannot access to medical care in a timely manner. Thus, the tool to help patients with self-diagnosis and self-triage is urgently needed to mitigate the shortage of medical care resources.
Symptom checkers are algorithm-based tools for self-diagnosis and self-triage. The increasing access to the Internet enabled these kinds of web-based healthcare service including symptom checkers. Generally, symptom checkers make a diagnosis of a disease based on the data on the prevalence of disease and its sensitivity and specificity of symptoms. We would like to introduce the possible benefit of symptom checkers on public health.
One of the main functions of symptom checkers is to assist with triage, which is the same function of community health workers (CHWs) in resource-limited countries. In Bangladesh, community health workers trained by NGOs contribute to promote public health, which has been proved effective such as in reducing mortality among infants in rural areas.^[@R3]^ However, quality control or education cost for health workers are now becoming social issues.^[@R4]^ It is sometimes difficult or impossible for the doctors or health workers to maintain the quality of the treatment since they do not analyze the patient's symptoms with the medical database or due to lack of their up to date knowledge. They just do some specific tests to assess the health condition of the patients and using their sport observation they draw the conclusion and suggests medicine for the patient's remedy from diseases or illness. Symptom checkers could help CHWs to triage patients. In the future, symptom checkers would help keeping up the quality of primary care delivered by CHWs in resource-limited countries.
Though, prior researches suggest that symptom checkers may be less effective than physicians in diagnostic accuracy for now,^[@R5]^ to conclude the superiority of doctors to symptom checkers might be overhasty. This is because diagnostic accuracy of symptom checkers can be improved after appropriate feedback. One reason of low diagnostic accuracy is because there is no accurate database of these information for all diseases. However, with the feedback of the data of diagnoses along with patients' symptoms reported by doctors, the symptom checkers can update their incorporated database, making the diagnosis closer to doctors' diagnosis than before. After doctors' feedback, the updated database of symptom checkers can provide useful information for clinical education (eg, sensitivity and specificity of symptoms for diseases, prevalence of diseases). Therefore, symptom checkers could be more useful and helpful tools for medical staff.
Symptom checkers can not only be the beneficial tools for doctors but also provide the access to health care in low resource settings such as in rural areas or developing countries. We doctors should concern that our active cooperation with symptom checkers can contribute to improve public health.
Ethical issues {#s2}
==============
Not applicable.
Competing interests {#s3}
===================
Authors declare that they have no competing interests.
Authors' contributions {#s4}
======================
TM and TH developed the concept of the letter. TM and AR collected data. TM drafted the manuscript and all authors contributed substantially to its revision.
Authors' affiliations {#s5}
=====================
^1^Department of Internal Medicine, Soma Central Hospital, Soma, Japan. ^2^Department of Medicine, Shaheed Suhrawardy Medical College & Hospital, Dhaka, Bangladesh. ^3^Health Intelligence Center, The Institute of Medical Science, The University of Tokyo, Minato-ku, Japan. ^4^Department of Surgery, Minamisoma Municipal General Hospital, Fukushima, Japan. ^5^Department of Internal Medicine, Jyoban Hospital of Tokiwa Foundation, Fukushima, Japan.
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper and its Supporting Information files.
Introduction {#sec001}
============
The regulation of cell growth is fundamentally important for a wide range of biological processes \[reviewed in [@pgen.1006154.ref001], [@pgen.1006154.ref002], [@pgen.1006154.ref003]\]. A key signal transduction network regulating cell growth and proliferation in response to nutrients involves two related kinases, Target-of-Rapamycin (TOR) and Class I phosphatidylinositol 3-kinase (PI3K) \[[@pgen.1006154.ref004]--[@pgen.1006154.ref007]\]. A variety of nutritional and growth factor stimuli are known to activate PI3K, which converts the membrane phospholipid phosphatidylinositol-4,5-bisphosphate (PIP2) into phosphatidylinositol-3,4,5-trisphosphate (PIP3) \[[@pgen.1006154.ref008]\]. Levels of PIP3 are kept in check by Phosphatase and Tensin Homologue (PTEN), which hydrolyzes PIP3 back to PIP2. PIP3 is a signaling lipid that stimulates plasma membrane recruitment of proteins with PIP3-specific pleckstrin homology (PH) domains such as Akt (also known as protein kinase B) and phosphoinositide-dependent kinase 1 (PDK1). The colocalization of Akt and PDK1 at the membrane surface increases the rate at which PDK1 phosphorylates Akt at a regulatory site essential for its activation \[[@pgen.1006154.ref009]\]. Activated Akt is then able to phosphorylate numerous targets in the TOR/PI3K network, including Forkhead box subgroup O (FoxO) transcription factors and Tuberous Sclerosis Complex 2 (TSC2), an inhibitor of TOR \[reviewed in [@pgen.1006154.ref010]\]. Akt phosphorylation of both of these negative growth regulators attenuates their activities, thus promoting an increase in cell growth and biomass.
*Drosophila melanogaster* provide a useful genetic system for studying cell growth and PI3K signaling in the context of an intact organism. Tissue growth in *Drosophila*, as in mammals, depends upon the Insulin-like receptor (InR)/PI3K pathway and the interconnected amino-acid/TOR pathway \[reviewed in [@pgen.1006154.ref011], [@pgen.1006154.ref012]--[@pgen.1006154.ref015]\]. Class I PI3K is required for the growth of most if not all *Drosophila* tissues but the ways in which it is differentially regulated as a function of cell type and developmental stage are not yet fully clear. Nevertheless, some insights have been gained by experiments showing that there is selective tissue growth in larvae subjected to nutrient restriction (NR). At early larval stages, NR shuts down the growth of developing tissues, and prevents the stem cells of the central nervous system (neuroblasts) from re-entering the cell cycle after a period of quiescence \[[@pgen.1006154.ref016]--[@pgen.1006154.ref018]\]. At late larval stages, however, growth in neuroblast lineages is almost completely spared during NR, whereas it is approximately halved for the epithelial progenitors of adult structures (imaginal discs) and reduced to near zero in many other larval tissues \[[@pgen.1006154.ref019], [@pgen.1006154.ref020]\].
Among the larval tissues that are not spared during NR are two major organs of the *Drosophila* adipose axis: fat body cells (adipocytes) and oenocytes. The fat body provides the major storage depot in *Drosophila* for neutral lipids such as triglycerides, in the form of intracellular lipid droplets \[[@pgen.1006154.ref021]\]. This tissue is important for the maintenance of energy homeostasis during starvation and acts as a nutrient sensor. Depending upon amino acid levels, the fat body can either store or release lipid nutrients into the hemolymph, in the form of lipoproteins \[[@pgen.1006154.ref022]--[@pgen.1006154.ref024]\]. Oenocytes are endocrine cells specialized for lipid metabolism \[reviewed in [@pgen.1006154.ref025], [@pgen.1006154.ref026]\]. There are two morphologically distinct populations of *Drosophila* oenocytes, larval and adult (imaginal), each deriving from a separate pool of ectodermal progenitors \[[@pgen.1006154.ref027]--[@pgen.1006154.ref029]\]. Adult oenocytes synthesize species and sex-specific mixes of cuticular hydrocarbons that function in desiccation resistance and pheromonal communication \[[@pgen.1006154.ref030]--[@pgen.1006154.ref032]\]. These cuticular hydrocarbons are synthesized from very-long chain (VLC) fatty acids via a pathway requiring the cytochrome P450 enzyme Cyp4g1, a VLC fatty aldehyde decarbonylase \[[@pgen.1006154.ref033], [@pgen.1006154.ref034]\]. Larval oenocytes, on the other hand, are known to be essential for molting and synthesize VLC fatty acids required for waterproofing the tracheal system \[[@pgen.1006154.ref035], [@pgen.1006154.ref036]\]. Unlike most other cell types, larval oenocytes accumulate numerous lipid droplets during NR \[[@pgen.1006154.ref035]\]. This oenocyte NR response resembles, at least superficially, the fasting-induced build up of neutral lipids (steatosis) observed in mammalian hepatocytes. In mammals, this steatosis is thought to be a physiological response to elevated lipolysis in adipose tissue. Similarly, in *Drosophila*, fat-body specific overexpression of an ortholog of Adipose Triglyceride Lipase (ATGL) is sufficient to induce steatosis in the oenocytes of fed larvae \[[@pgen.1006154.ref035]\]. Hence, during NR, neutral lipid in the form of lipid droplets is lost from the fat body but gained by the oenocytes. The induction of lipid droplets during starvation requires the activity of the Lipophorin receptor (Lpr2) in oenocytes and so presumably involves the uptake of lipids released into the hemolymph \[[@pgen.1006154.ref036]\]. It nevertheless remains unclear which tissue-specific signaling mechanisms allow neutral lipid content in the fat body and in oenocytes to be simultaneously regulated in opposite directions during starvation.
Here, we characterize novel regulatory interactions between PI3K signaling, lipid metabolism and cell growth in the context of the *Drosophila* adipose axis. Tissue-specific and clonal genetic analyses reveal that oenocytes respond to nutrition and PI3K signaling very differently from the fat body. PI3K signaling stimulates neutral lipid storage in fat body cells but it inhibits this process in oenocytes. We identify two lipid oxidoreductases that regulate the balance between lipid storage and cell size in oenocytes and show that one of these is part of an oenocyte-specific regulatory circuit that modulates PI3K-dependent cell growth.
Results {#sec002}
=======
PI3K signaling inhibits lipid droplets in oenocytes but promotes them in fat body {#sec003}
---------------------------------------------------------------------------------
When raised on an optimal diet, *Drosophila* larvae grow (increase mass) by more than two orders of magnitude as they develop through three instars (L1-L3) over a four-day (\~96 hr) period. Fat body and oenocytes ([Fig 1A](#pgen.1006154.g001){ref-type="fig"}), in common with many other larval tissues, grow via an increase in cell size and ploidy rather than by cell division \[[@pgen.1006154.ref037], [@pgen.1006154.ref038]\]. We first compared how oenocytes and fat body respond to PI3K signaling. To assess the cell-autonomous effects of class I PI3K signaling upon fat body cell size and intracellular lipid droplets, genetic mosaic larvae were generated via Flp/FRT mediated activation of the GAL4/UAS system (Flp-out clones). A previous study found that overexpressing PI3K (UAS-Dp110) in Flp-out clones in the fat body is sufficient to increase cell size in well-fed larvae \[[@pgen.1006154.ref039]\]. During starvation it is known that neutral lipids, stored in intracellular lipid droplets, decrease in the fat body yet increase in oenocytes and neither tissue is able to grow significantly \[[@pgen.1006154.ref019], [@pgen.1006154.ref035]\]. To investigate the role of PI3K signaling in the regulation of lipid droplets during nutrient restriction (NR), larvae at the early-L3 stage (48hr after larval hatching) were switched from fed (yeast/cornmeal/agar) to NR (PBS/agarose) medium for 18 hr. Tissues were analyzed either at the early L3 stage (Fed~48~ control group) or 18 hr later (NR~66~ experimental and Fed~66~ control groups) ([Fig 1B](#pgen.1006154.g001){ref-type="fig"}). NR applied at early L3 induces developmental arrest such that, for NR~66~ larvae, their chronologically matched control is Fed~66~ but their developmentally matched control is Fed~48~. Figures therefore show comparisons between Fed~48~ and NR~66~ larvae, although in most cases the Fed~66~ time point was also analyzed. Following 18 hr of NR, the size of fat body cells is decreased but they are only partially depleted of stored neutral fat and so still contain numerous lipid droplets \[[@pgen.1006154.ref023], [@pgen.1006154.ref035] and [Fig 1C](#pgen.1006154.g001){ref-type="fig"}\]. Overexpression of PI3K resulted in increases in fat body cell size, lipid droplet diameter and neutral lipid content following NR (Figs [1C](#pgen.1006154.g001){ref-type="fig"}, [S1A](#pgen.1006154.s001){ref-type="supplementary-material"} and [S2A](#pgen.1006154.s002){ref-type="supplementary-material"}). The converse manipulation, expressing a dominant-negative PI3K (Dp110^DN^) resulted in the expected decrease in fat body cell size but it did not significantly alter lipid droplet content in fed conditions (Figs [1D](#pgen.1006154.g001){ref-type="fig"}, [S1B](#pgen.1006154.s001){ref-type="supplementary-material"} and [S2B](#pgen.1006154.s002){ref-type="supplementary-material"}). These results together suggest that a decrease in PI3K signaling is needed for neutral lipid loss from the fat body during NR but that this change alone is not enough to drive lipid loss in the fed state.
{ref-type="supplementary-material"} and [S2](#pgen.1006154.s002){ref-type="supplementary-material"} Figs for quantitations.](pgen.1006154.g001){#pgen.1006154.g001}
We next manipulated PI3K activity in oenocytes using a larval oenocyte-specific GAL4 driver (*PromE-GAL4*, \[[@pgen.1006154.ref031], [@pgen.1006154.ref036]\]). To monitor PI3K signaling levels during these manipulations, we used an antibody recognizing phospho-Akt (Ser505). Although technically challenging to detect, membrane localization of phospho-Akt provides a readout for most Akt activity \[[@pgen.1006154.ref019], [@pgen.1006154.ref040], [@pgen.1006154.ref041]\]. Control GFP-labeled oenocyte clusters showed higher and more consistent expression of membrane phospho-Akt in Fed~48~ than in NR~66~ larvae, consistent with the known nutrient-dependence of PI3K/TOR signaling ([Fig 1E and 1F](#pgen.1006154.g001){ref-type="fig"}). GFP-labeled oenocyte clusters with overactive PI3K signaling (*PromE\>Dp110*) are very large in size in both Fed~48~ and NR~66~ larvae and this \"giant oenocyte\" phenotype is associated with increased expression and membrane localization of phospho-Akt (Figs [1E, 1F](#pgen.1006154.g001){ref-type="fig"} and [S2C--S2F](#pgen.1006154.s002){ref-type="supplementary-material"}). Increased oenocyte size was also observed using a second method for boosting PI3K activity: expressing an activated myristoylated form of Akt (myr-Akt) in Flp-out clones ([S2G Fig](#pgen.1006154.s002){ref-type="supplementary-material"}). Conversely, we found that dominant-negative PI3K (*PromE\>Dp110*^*DN*^) decreases oenocyte size in both Fed~48~ and NR~66~ larvae and, in the latter context, p-Akt staining remains detectable, although weak and close to the limit of detection (Figs [1E, 1F](#pgen.1006154.g001){ref-type="fig"} and [S2C--S2F](#pgen.1006154.s002){ref-type="supplementary-material"}). A similar cell size decrease is also observed in Flp-out clones expressing *Dp110*^*DN*^ ([S2G Fig](#pgen.1006154.s002){ref-type="supplementary-material"}). During NR, we observed that oenocytes not only express weak p-Akt but they also retain higher expression of a GFP sensor for PIP3 levels \[tGPH, [@pgen.1006154.ref039]\] and weaker nuclear FoxO expression than do fat body cells ([S3 Fig](#pgen.1006154.s003){ref-type="supplementary-material"}). Oenocytes therefore sustain low-level PI3K signaling during NR. Nevertheless, decreasing this low level even further using *Dp110*^*DN*^ does not significantly block the normal 100-fold increase in lipid droplets during NR, suggesting that it is not an obligate positive input (Figs [1F](#pgen.1006154.g001){ref-type="fig"} and [S1C--S1E](#pgen.1006154.s001){ref-type="supplementary-material"}). Rather, PI3K signaling in oenocytes has primarily a negative input as *Dp110*^*DN*^ induces a low level of lipid droplets in the fed state and *Dp110* overexpression efficiently blocks them during NR (Figs [1E, 1F](#pgen.1006154.g001){ref-type="fig"}, [S1D and S1E](#pgen.1006154.s001){ref-type="supplementary-material"}). In the context of adult oenocytes, it was reported that lipid droplet induction during starvation requires Insulin-like receptor activation via Ilp6 secreted from the adult fat body \[[@pgen.1006154.ref042]\]. For larval oenocytes, however, we find that Ilp6 is not required for lipid droplet induction during NR ([S4A and S4B Fig](#pgen.1006154.s004){ref-type="supplementary-material"}). Together, the results thus far demonstrate that artificial overactivation of PI3K is sufficient to stimulate NR growth of both cell types in the larval adipose axis but that it has opposite effects on NR fat storage in lipid droplets: inhibiting it in oenocytes yet promoting it in the fat body.
Many lipid metabolic enzymes regulate lipid droplets but not oenocyte size {#sec004}
--------------------------------------------------------------------------
Before determining how PI3K signaling inhibits lipid droplets in a tissue-specific manner, we first investigated the oenocyte droplet induction mechanism itself. We focused on the metabolism of VLCFAs as these are known to be synthesized selectively in larval oenocytes \[[@pgen.1006154.ref035], [@pgen.1006154.ref036] and [Fig 2A](#pgen.1006154.g002){ref-type="fig"}\]. Consistent with a previous study \[[@pgen.1006154.ref036]\], in Fed~48~ larvae, *PromE-GAL4* driven RNAi knockdown of the malonyl-CoA synthesizing enzyme Acetyl-CoA Carboxylase (Acc) significantly increased oenocyte lipid droplets (Figs [2B](#pgen.1006154.g002){ref-type="fig"} and [S1F](#pgen.1006154.s001){ref-type="supplementary-material"}). This fed increase reflects an amount of lipid droplets per oenocyte that is considerably lower than that observed during NR in control larvae. It is known to be mediated via the uptake of fatty acids in fed larvae via Lpr2, a Lipophorin receptor \[[@pgen.1006154.ref036]\]. Unexpectedly, in NR larvae, we observed that Acc knockdown decreased lipid droplet induction, suggesting that fatty acid synthesis makes a contribution to starvation-induced oenocyte steatosis (Figs [2B](#pgen.1006154.g002){ref-type="fig"} and [S1G](#pgen.1006154.s001){ref-type="supplementary-material"}). In a previous study, *Acc* knockdown in oenocytes was also associated with tracheal flooding and systemic hypoxia \[[@pgen.1006154.ref036]\]. Importantly, however, using our genetic and dietary conditions, the majority of larvae expressing *PromE-GAL4* driven *UAS-RNAi* of *Acc* or other genes in this study do not display tracheal flooding even though knockdown efficiencies are over 90% ([S4C](#pgen.1006154.s004){ref-type="supplementary-material"} and [S5](#pgen.1006154.s005){ref-type="supplementary-material"} Figs). As the PromE knockdown phenotypes described in our study are observed in most if not all larvae, they are highly unlikely to be a secondary consequence of hypoxia due to flooding of the trachea. Similar to Acc, knockdown of Diacylglycerol acyltransferase 1 (DGAT1), a dedicated enzyme in the synthesis of TAGs \[[@pgen.1006154.ref043]\], gave a decrease in lipid droplet induction during NR in oenocytes (Figs [2B](#pgen.1006154.g002){ref-type="fig"}, [S1F and S1G](#pgen.1006154.s001){ref-type="supplementary-material"}). Conversely, *PromE* overexpression of Lsd-2 (perilipin-2), a protein that inhibits lipolysis to promote TAG storage in lipid droplets \[[@pgen.1006154.ref044]\], led to a statistically significant increase in oenocyte lipid droplets in Fed~48~ but not in NR larvae (Figs [2B](#pgen.1006154.g002){ref-type="fig"}, [S1F and S1G](#pgen.1006154.s001){ref-type="supplementary-material"}). This suggests that there is active lipolysis in the fed state and that this may help to prevent the accumulation of lipid droplets. Cytochrome P450 4g1 (Cyp4g1), with its redox partner Cytochrome P450 reductase (Cpr), together form a microsomal holoenzyme involved in converting VLCFAs into hydrocarbons in the oenocytes of adults \[[@pgen.1006154.ref033]\]. Strikingly, RNAi knockdowns for either enzyme or a *Cyp4g1* loss-of-function mutation \[[@pgen.1006154.ref035]\] attenuated lipid droplet induction during NR but had little effect in Fed~48~ larvae (Figs [2B, 2C](#pgen.1006154.g002){ref-type="fig"} and [S1F--S1H](#pgen.1006154.s001){ref-type="supplementary-material"}). These results strongly suggest that the Cyp4g1/Cpr holoenzyme is required for the induction of lipid droplets in larval oenocytes during NR. We next tested whether any of the above lipid biosynthetic enzymes regulating oenocyte lipid droplets might also control oenocyte size during NR. However, no significant differences in oenocyte volume were observed with *PromE* driven overexpression of Lsd2 or RNAi knockdown of Acc, Dgat1, Cyp4g1 or Cpr ([Fig 2D](#pgen.1006154.g002){ref-type="fig"}). Together with the previous results, this demonstrates regulation of oenocyte lipid droplets, not only by PI3K signaling but also by triglyceride and hydrocarbon biosynthetic enzymes. In contrast, regulation of oenocyte cell size is sensitive to PI3K signaling but not to the five tested triglyceride and hydrocarbon biosynthetic enzymes.
{ref-type="supplementary-material"} for quantifications.](pgen.1006154.g002){#pgen.1006154.g002}
Oenocyte Kar and FarO promote lipid droplets and suppress cell growth {#sec005}
---------------------------------------------------------------------
We next extended the oenocyte genetic analysis to two enzymes, Kar and FarO, which are not dedicated to triglyceride or hydrocarbon production but to more general roles in the synthesis of VLCFAs, VLCF aldehydes and VLCF alcohols ([Fig 2A](#pgen.1006154.g002){ref-type="fig"}). *CG1444* encodes a protein with YXXXK catalytic and putative NADH binding motifs found in many enzymes of the short-chain reductase/dehydrogenase (SDR) family \[[@pgen.1006154.ref045]\]. It has been reported to encode the only predicted 3-ketoacyl-CoA reductase (Kar) in the *Drosophila melanogaster* genome and is required in larval oenocytes for the synthesis of a putative VLCFA implicated in tracheal waterproofing and also in adult oenocytes for the synthesis of VLCFA-derived cuticular hydrocarbons \[[@pgen.1006154.ref034], [@pgen.1006154.ref036]\]. However, the activities and substrate specificities of the Kar enzyme (also known as Spidey) have yet to be directly determined. We raised an antibody against Kar and observed that the levels of Kar protein are substantially attenuated but nevertheless remain expressed in oenocytes during NR, whereas Acc levels do not noticeably change between fed and NR ([Fig 3A](#pgen.1006154.g003){ref-type="fig"}). Kar protein levels are also moderately decreased when PI3K signaling is inhibited in the oenocytes of Fed~48~ larvae ([Fig 3B](#pgen.1006154.g003){ref-type="fig"}). Hence, dietary nutrients and PI3K signaling not only regulate lipid droplet accumulation in oenocytes but also Kar expression. Another enzyme of VLCFA metabolism, fatty acyl-CoA reductase, converts fatty acyl-CoA esters into fatty alcohols. *CG18031* encodes one of several predicted fatty acyl-CoA reductases in *Drosophila*. It is selectively expressed in [o]{.ul}enocytes \[[@pgen.1006154.ref035]\] and will therefore be referred to as *FarO*. It encodes a protein with a Rossmann--fold NAD(P)H binding domain with a \[ST\]GXXGXXG motif and also a YXXXK active site motif as found in other fatty--acyl--CoA reductases ([Fig 3C](#pgen.1006154.g003){ref-type="fig"}). *Drosophila* FarO produced from a recombinant baculovirus in Sf9 cells is a microsomal enzyme that, in the presence of NADH or NADPH, can convert tetracosanoic-CoA ester (C24:0-CoA) to a product with an elution time identical to that of tetracosanol (C24:0) ([Fig 3D and 3E](#pgen.1006154.g003){ref-type="fig"}). Recombinant FarO also produced alcohols from C26:0-CoA, as well as from C26:0 and C28:0 fatty acids pre-incubated with CoA-SH. These biochemical experiments demonstrate that FarO is a fatty acyl-CoA reductase that reduces VLCFA-CoAs to VLCF alcohols.
{#pgen.1006154.g003}
We next tested the effects of knocking down Kar or FarO with RNAi. In the fat body, RNAi Flp-out clones for Kar or FarO did not detectably alter lipid droplets or cell size during NR ([S6A Fig](#pgen.1006154.s006){ref-type="supplementary-material"}). In oenocytes, however, RNAi knockdown of either gene did give a phenotype using *PromE-GAL4* or a second tissue-selective driver (see [Materials and Methods](#sec012){ref-type="sec"}). Knockdown of Kar but not FarO was associated with a significant increase of lipid droplets in the fed state, to a low level similar to that observed with Acc knockdown (Figs [4A](#pgen.1006154.g004){ref-type="fig"} and [S1I](#pgen.1006154.s001){ref-type="supplementary-material"}). Nevertheless, the knockdown of either Kar or FarO strongly decreased the accumulation of oenocyte lipid droplets during NR (Figs [4A](#pgen.1006154.g004){ref-type="fig"} and [S1J](#pgen.1006154.s001){ref-type="supplementary-material"}). Using a second genetic method, Flp-out clones, FarO RNAi was also observed to abrogate oenocyte droplet induction ([S6B and S6C Fig](#pgen.1006154.s006){ref-type="supplementary-material"}). Together, these findings demonstrate that the lipid reductases/dehydrogenases Kar and FarO are both required in oenocytes for the induction of lipid droplets during NR. Strikingly, we also observed that RNAi knockdown of either Kar or FarO with *PromE-GAL4* markedly increases the volume of Fed~48~ or NR~66~ oenocytes ([Fig 4A](#pgen.1006154.g004){ref-type="fig"} insets and [4B](#pgen.1006154.g004){ref-type="fig"}). Control and knockdown oenocytes are similar in size prior to larval hatching (0 hr) so excessive oenocyte size likely reflects a greater volume gain (i.e growth rate) during the first 48 hr of larval development. However, by 66 hr (Fed~66~), Kar and FarO knockdown oenocyte sizes are no longer significantly different from controls, even though oenocyte size remains PI3K-dependent at this later developmental stage ([S6D and S6E Fig](#pgen.1006154.s006){ref-type="supplementary-material"}). Given that controls catch up in size with knockdown oenocytes by 66hr, it may be that Kar or FarO RNAi do not alter final oenocyte size at the end of larval development (96 hr). We conclude more generally that, even though many oenocyte enzymes involved in triglyceride and hydrocarbon biosynthesis do not influence cell size, the two lipid oxido reductases Kar and FarO both play a key role in preventing excessive cell growth during early larval development.
{ref-type="supplementary-material"} for quantitations.](pgen.1006154.g004){#pgen.1006154.g004}
Kar regulates oenocyte nuclear size and endoreplication {#sec006}
-------------------------------------------------------
Kar and FarO could potentially regulate oenocyte size via several distinct mechanisms, such as osmotic swelling/shrinkage or the cell cycle. The developmental growth of many larval cell types, including oenocytes, is known to be accompanied by rounds of DNA replication without cytokinesis that lead to polyploidy and increasing nuclear size \[reviewed in [@pgen.1006154.ref038]\]. Following RNAi knockdown of Kar or FarO, we observed that NR cell volume increases are concomitant with nuclear enlargement ([Fig 5A](#pgen.1006154.g005){ref-type="fig"}). To visualize DNA replication directly, we then used the *in vivo* incorporation of the thymidine analogue 5-ethynyl-2\'-deoxyuridine (EdU). Consistent with previous studies, EdU is incorporated into the polyploid nuclei of many mid-L2 tissues including oenocytes in fed larvae. In contrast, when L2 larvae were subjected to NR for 12 hr before exposure to EdU in NR medium, nuclear incorporation in many tissues such as the salivary gland is near zero whereas in the majority, although not all, oenocytes it is detectable ([Fig 5B](#pgen.1006154.g005){ref-type="fig"}). This suggests that, unlike salivary gland cells, oenocytes retain a low-level of endoreplication for at least 12 hr during NR. Importantly, EdU incorporation in oenocyte nuclei during NR is markedly increased in the enlarged oenocytes generated by Kar knockdown (*PromE\>Kar Ri*) and can be even further increased in the giant oenocytes resulting from strong PI3K hyperactivation (*PromE\>Dp110*) ([Fig 5B](#pgen.1006154.g005){ref-type="fig"}). Kar therefore regulates a *bona fide* growth mechanism that increases both cell size and nuclear ploidy. Moreover, the opposing effects of Kar and PI3K upon oenocyte size correlate with their effects upon the oenocyte endocycle.
{#pgen.1006154.g005}
Kar inhibits p-AKT and PI3K signaling to limit oenocyte growth {#sec007}
--------------------------------------------------------------
To explore further the genetic interactions between lipid oxidoreductases and PI3K signaling, we made double combinations of Kar or FarO RNAi with dominant negative PI3K (*Dp110*^*DN*^). For oenocyte lipid droplets, double combinations were analyzed in the fed state as Dp110^DN^ does not have any significant effect in NR larvae. Kar knockdown and Dp110^DN^ each increase fed oenocyte lipid droplets to a moderate level and, not surprisingly, this was also observed when they were combined ([Fig 6A and 6B](#pgen.1006154.g006){ref-type="fig"}). A different result was obtained with FarO knockdown, which efficiently blocked the Dp110^DN^ increase of lipid droplets seen in Fed~48~ larvae ([S7A and S7B Fig](#pgen.1006154.s007){ref-type="supplementary-material"}). These results show that, in the context of oenocyte lipid metabolism, FarO and PI3K interact antagonistically. They also suggest that Kar and FarO interact differently with PI3K signaling. Genetic interactions of Kar and FarO with PI3K were also analyzed in the context of oenocyte size. For FarO knockdown, an intermediate oenocyte size phenotype was obtained with Dp110^DN^ precluding clear conclusions about regulatory connections ([S7A and S7C Fig](#pgen.1006154.s007){ref-type="supplementary-material"}). For Kar knockdown, however, the small cell volume obtained with Dp110^DN^ was epistatic in both fed and NR larvae ([Fig 6A and 6C](#pgen.1006154.g006){ref-type="fig"}). This shows that the increase in oenocyte size observed following Kar knockdown is strictly dependent upon PI3K activity.
{ref-type="supplementary-material"} for LipidTOX quantifications.](pgen.1006154.g006){#pgen.1006154.g006}
To investigate further the regulatory relationship between Kar, FarO and PI3K in larval oenocytes, we used immunostaining for phospho-AKT. No significant change in membrane phospho-Akt levels was detected following FarO knockdown in NR larvae, suggesting that it functions downstream or in parallel to PI3K during the negative regulation of oenocyte growth ([S8A Fig](#pgen.1006154.s008){ref-type="supplementary-material"}). Importantly, Kar knockdown using either the *PromE-GAL4* or the Flp-out clonal method gave a moderate and reproducible increase in membrane phospho-Akt expression during NR (Figs [7A, 7B](#pgen.1006154.g007){ref-type="fig"}, [S8B and S8C](#pgen.1006154.s008){ref-type="supplementary-material"}). Together with the genetic interaction analysis, these phospho-Akt results demonstrate that Kar plays a key role in inhibiting PI3K signaling and thus suppressing the inappropriate overgrowth of oenocytes.
{#pgen.1006154.g007}
Discussion {#sec008}
==========
PI3K regulation of fat body-oenocyte lipid metabolism {#sec009}
-----------------------------------------------------
During starvation, internal levels of nutrients decrease and growth is attenuated in many larval tissues including the fat body and oenocytes \[[@pgen.1006154.ref019], [@pgen.1006154.ref046]\]. Our study now suggests that the decrease in PI3K signaling during starvation plays a physiological role in permitting neutral lipids to be lost from the fat body and concomitantly gained in oenocytes. For the fat body, decreased PI3K signaling is a necessary but not a sufficient condition for neutral lipid loss, indicating that other regulators must also be important for the starvation response in this tissue. For larval oenocytes, reduced PI3K signaling is not only needed for lipid droplet induction during NR but it can also trigger inappropriate lipid droplets in the fed state. Hence, different tissue-specific responses to PI3K signaling may be an important component of the mechanism regulating lipid metabolism in the larval fat body-oenocyte axis.
It has been reported that starvation induces lipid droplets in adult oenocytes \[[@pgen.1006154.ref042]\], as is known to be the case for larval oenocytes \[[@pgen.1006154.ref035]\]. In contrast to our current findings in larvae, it was concluded in adult flies that starvation increases oenocyte PI3K signaling and that lipid droplet induction requires Insulin-like receptor signaling in oenocytes, triggered by Ilp6 secreted from the adult fat body \[[@pgen.1006154.ref042]\]. For larval oenocytes, however, the lipid droplet induction mechanism is very different as our study shows that it is compromised by high PI3K signaling and does not require Ilp6.
Oenocyte lipid droplet biogenesis requires VLCFA and hydrocarbon enzymes {#sec010}
------------------------------------------------------------------------
This study reveals that genes metabolizing fatty acids (Acc), triglycerides (Dgat1), VLCFAs (FarO and possibly Kar) and hydrocarbons (Cyp4g1 and Cpr) are all required for maximal lipid droplet induction in larval oenocytes during NR. For Acc and Kar, this starvation function in lipid droplet induction adds to their previously known roles in waterproofing the larval tracheal system \[[@pgen.1006154.ref036]\]. Our genetic analyses also suggest that derivatives of long and very long chain fatty acids, including hydrocarbons, could either regulate the induction of oenocyte lipid droplets or be part of their neutral lipid cargo. The fat composition of larval oenocyte lipid droplets is, however, not yet clear due to the inherent technical difficulty of obtaining enough purified material for meaningful lipidomic analysis. Given that the accumulation of as yet unidentified neutral lipids in larval oenocytes reflects increased fatty acid release from the fat body \[[@pgen.1006154.ref035]\], it is tempting to speculate that they may play an important metabolic role during starvation. Although the physiological functions of oenocyte lipid droplets have not yet been clearly demonstrated in larvae, oenocyte lipid uptake in adult flies is thought to play an important role in promoting lipid turnover and survival during starvation \[[@pgen.1006154.ref042]\].
Kar mediates a PI3K negative feedback loop during oenocyte growth {#sec011}
-----------------------------------------------------------------
An unexpected finding of this study is that two lipid dehydrogenase/reductase enzymes, Kar and FarO, regulate the balance between lipid droplets and the cell size of oenocytes. Knockdown of either enzyme blocks droplet induction and also increases cell size. Four other genes required for the induction of lipid droplets (Acc, Dgat1, Cyp4g1 and Cpr) did not detectably influence oenocyte size, arguing that blocking lipid droplet induction *per se* does not automatically lead to an increase in cell growth. Oenocyte size regulation therefore appears to be a selective property of a subset of lipid metabolic enzymes rather than all enzymes that synthesize/modify fatty acids. Kar is reported to be a 3-ketoacyl-CoA reductase for the elongation of long chain fatty acids into VLCFAs and it is known to be required in oenocytes to produce their derivatives, the cuticular hydrocarbons \[[@pgen.1006154.ref034], [@pgen.1006154.ref036]\]. A recent study, however, provides evidence that Kar, also known as Spidey, regulates ecdysteroid levels in larvae and so could function as a steroid dehydrogenase rather than a ketoacyl-CoA reductase \[[@pgen.1006154.ref047]\]. For FarO, we demonstrated biochemically that it is a *bona fide* VLC fatty acyl-CoA reductase that reduces VLCFA-CoA esters into their corresponding VLCF alcohols. It remains to be established, however, whether or not the functions of Kar or FarO in cell growth require their reductase/dehydrogenase catalytic activities. Intriguingly, we found that Kar is required for the induction of oenocyte lipid droplets during starvation, despite a decrease in its protein levels. Genetic analysis indicates that the lower levels of Kar during NR are compatible with its function in lipid droplet induction. Presumably less Kar is also needed during NR for its other function in limiting PI3K activity, which is itself much lower during NR than in the fed state. The Kar decrease during NR can be mimicked in fed larvae by blocking oenocyte PI3K signaling, indicating that PI3K is a positive regulator of Kar expression. Our study also provides two lines of evidence for the reciprocal regulation i.e. Kar-dependent repression of PI3K signaling. Hence, the increase in oenocyte growth following Kar knockdown requires PI3K activity and membrane phospho-Akt levels are elevated by Kar knockdown. The existence of such cross-regulation supports a model in which a PI3K negative feedback loop mediated by Kar prevents excessive oenocyte growth ([Fig 7C](#pgen.1006154.g007){ref-type="fig"}). So why does PI3K negative feedback mediated by Kar exist in some cell types (e.g oenocytes) but apparently not in others (e.g fat body cells)? One possibility is that this regulatory coupling is only important in those cell types that express high levels of Kar and dedicate a large fraction of their energy resources to metabolizing its substrates. Kar\'s role and indeed that of FarO could therefore be to ensure that critical physiological functions of oenocytes requiring VLCFA or ecdysteroid metabolism are not sacrificed at the expense of growth. Future biochemical and biophysical studies will be needed in order to elucidate the mechanisms by which Kar and FarO suppress oenocyte growth. Regulation of PI3K signaling at the level of membrane p-Akt was detected for Kar, although not for FarO. In principle, Kar could regulate PI3K signaling via effects upon ecdysone signaling and/or by changing the composition of the membrane where PI3K is active. With regard to the latter possibility, it is interesting that studies in various other biological contexts have shown that VLCFAs can modulate the activity of the PI3K signaling pathway. For example, docosahexenoic acid, a polyunsaturated VLCFA, inhibits PI3K signaling and the growth of prostate and colon cancer cells whereas it can stimulate PI3K signaling and the survival of neuronal cells \[[@pgen.1006154.ref048]--[@pgen.1006154.ref050]\]. Future studies will reveal whether oenocyte-like Kar regulation of PI3K exists in these contexts or in other tissues known to synthesize large quantities of VLCFAs and other specialized lipids such as those found in the mammalian liver and skin.
Materials and Methods {#sec012}
=====================
*Drosophila* strains and larval analysis {#sec013}
----------------------------------------
*Drosophila* were raised at 25°C using a standard cornmeal/agar diet unless otherwise stated. Larvae were staged at the L2/L3 molt according to morphology and early L3 larvae were either dissected immediately (the Fed~48~ group) or transferred for 18hr to PBS/1% low melting point agarose (the NR~66~ group) or for 18h to food (the Fed~66~ group) before dissection ([Fig 1](#pgen.1006154.g001){ref-type="fig"}). For all RNAi experiments, early L1 larvae were transferred to 29°C until dissection. For tracheal flooding assays, larvae were transferred 6-8hr after the L2/L3 molt from food to PBS in a petri dish and maintained submerged for 15 min using a wire mesh. The numbers of larvae with liquid in the main tracheal trunks were then scored. The oenocyte specific driver line *PromE-Gal4*, *UAS-CD8*::*GFP* \[[@pgen.1006154.ref031]\] is expressed from embryonic stage 17 onwards. Key results with *PromE-GAL4* were confirmed using a second oenocyte-specific GAL4 driver line *Cee-GAL4* ([*C*]{.ul}*yp4g1-Gal4*; [*e*]{.ul}*lav-Gal80*; [*e*]{.ul}*lav-Gal80*), which is strongly expressed from the L1 stage in oenocytes and salivary glands (and weakly in trachea) and was generated by combining a copy of *elav*-Gal80 on the 2nd and on the 3rd chromosomes (gifts of Matthias Llandgraf) with *Cyp4g1*-Gal4^*NP6073*^ (DGRC, Kyoto) to suppress unwanted neural Gal4 activity. To induce flp-out clones, *hs-flp*; *actin5C\>CD2\>Gal4* flies were crossed to the relevant *UAS* line and early L1 progeny were heat shocked at 37°C for 8 min. The larval progeny of GAL4 or UAS lines crossed to *w*^*1118*^ were used as controls. Oenocyte-less larvae were generated by crossing *PromE-Gal4* to a *UAS-reaper* line \[[@pgen.1006154.ref035]\]. Inducible RNA-interference (*UAS-RNAi*) lines for the genes *CG11198*^*KK108631*^ (*Acc*), *CG11198*^*GD8105*^ (*Acc*), CG1444^GD40949^ (*Kar*), CG1444^1444R-2^ (*Kar*), CG18031^GD30220^ (*FarO)*, CG31991^*GD6367*^ (*DGAT1*, *midway*), CG11567^GD46715^ (*Cpr*), CG11567^GD44232^ (*Cpr*), CG11567^KK107422^ (*Cpr*) and CG14049^GD45218^ (*Ilp6*) were obtained from VDRC and NIG-FLY. For the generation of inducible intron-spliced snapback *UAS-Cyp4g1* RNAi lines (O-194 and O-196), the cDNA clone GH05567 was used as a template and a 330 bp *Cyp4g1* fragment amplified with Pfu DNA polymerase using the following pairs of 5' tagged primers: GAGTACTAGTAAGAGGAGTCACGTGCGATTGTTG and GTTGACTAGTGCGAAGACTTTAGCCTGGATG. The PCR fragment of *Cyp4g1* was cloned as an inverted repeat into the *UAS-RNAi* vector pWIZ \[[@pgen.1006154.ref051]\] and introduced in *yw* hosts by P-element-mediated transformation \[[@pgen.1006154.ref052]\]. With the conditions used, *PromE-GAL4* produced pupal/adult lethality when crossed with *UAS-RNAi* lines for *Acc*, *Kar*, *Dgat1*, *Cyp4g1*, *Cpr* or *Dp110*^*DN*^ but not for *FarO*, *Dp110* or overexpression of *Lsd2*. *Cyp4g1*^*Δ4*^ is a protein-null allele \[[@pgen.1006154.ref035]\] and hemizygous mutant males were compared to *Cyp4g1*^*Δ4*^*/FM7c* female controls. Other stocks used were *Df(1)Ilp6* \[[@pgen.1006154.ref053]\], *Cg-GAL4;UAS-Dicer2*, *UAS-nlsGFP*, *UAS-Dp110DN*^*A2860C*^, *UAS-Dp110* \[[@pgen.1006154.ref054]\], *UAS-myrAkt1* \[[@pgen.1006154.ref055]\], *tGPH* \[[@pgen.1006154.ref039]\], *UAS-Lsd2* \[line RKF391, [@pgen.1006154.ref056]\]. Stocks obtained from the Bloomington *Drosophila* Stock Center (NIH P40OD018537) were used in this study.
Lipid droplet staining {#sec014}
----------------------
Oil Red O staining was performed as previously described \[[@pgen.1006154.ref035]\] and images were acquired using a Zeiss Axiophot 2 compound microscope. LipidTOX staining was performed after immunostaining: tissues were washed with PBS + 0.3% Triton. Then washed several times with PBS to remove the detergent and incubated O/N at 4°C with LipidTOX (HCS LipidTOX Red or Deep Red Neutral Lipid Stain, Life Technologies) at 1:1000 dilution.
Immunostaining {#sec015}
--------------
Larval pelts, cut along the dorsal side, were cleaned of internal tissues and immobilized in PBS cuticle face down with insect pins on silicone elastomer (Sylgard, Dow Corning) polymerized in a small petri dish. All subsequent reaction steps were done on pinned out pelts in the Sylgard dish, using a horizontal shaker. Pelts were then fixed in 4% formaldehyde (or 10% formaldehyde for P-Akt stainings) for 20 min. After permeabilization in PBST (PBS/0.1% TritonX-100) tissues were blocked for 1.5 hr in PBST containing 0.1% BSA, 10% NGS, then primary antibodies were added in block solution and incubated overnight at 4°C. After extensive washes, block solution containing secondary antibodies was added for 2 hr at room temperature, and then washed extensively. Washes (5x20 min) were in PBS/0.1% TritonX-100/0.1% BSA after incubation with primary and secondary antibodies. For P-Akt antibodies, samples were postfixed for 20 min in 4% formaldehyde. Pelts were then washed in PBST and mounted with the cuticle facing up in Vectashield (Vector Laboratories). Primary antibodies were used at the following dilutions: Acc (1:50 cat\#3662 Cell Signaling Technology), rabbit anti-Foxo (1:500, gift of P. Leopold), P-Akt Ser505 (1:50, cat\#4054 Cell Signaling Technology), mouse anti-Cyp4g1 (1:1000, 433-43-2e, gift of S. Kennel). Secondary goat antibodies used were Alexa Fluor F(ab\')2 fragments (1: 400, Invitrogen) and streptavidin-A555 (1ug/mL, Molecular Probes). Nuclear labeling was performed by incubating the tissues with Dapi (1:1000 Sigma). Images were collected on a Nikon Digital Eclipse C1 confocal microscope. Affinity purified rabbit anti-Kar antibodies (used at 1:1000) and rabbit anti-Cpr antibodies (used at 1:500) were generated by Cambridge Research Biochemicals via immunization with the peptides LGTRKRALRRLAKEQ (307--321, reference protein sequence AAF46291) and YLKNKQPQGSEEVKV (497--511, reference protein sequence AAF52367) respectively. RNAi knockdown efficiencies for Acc, Cpr, Cyp4g1 and Kar were quantified in ImageJ from maximum intensity z-stack projections by measuring mean pixel intensity across an oenocyte cluster, followed by subtraction of background as the mean pixel intensity measured in the area surrounding each cluster. Relative lipid content was quantified from maximum intensity z-stack projections of LipidTOX stainings of fat body cells and oenocytes. Mean pixel intensities across the entire fat body cell or oenocyte cluster were calculated using the \'integrated density\' value of the analysis measurement tool in Photoshop. Cell areas were manually selected according to CD8::GFP expression using the magnetic lasso tool. Similarly, mean pixel intensities for membrane p-Akt staining were quantified from single confocal sections by manually selecting the CD8::GFP positive membrane region of oenocytes.
EdU incorporation in vivo {#sec016}
-------------------------
Early L2 larvae were transferred to NR medium (PBS/0.5% low melting point agarose) for 12 hr and then removed, washed, and transferred to NR medium containing 0.1 mM EdU for 24 hr where larvae were seen "feeding" on the medium. Then larvae were dissected, fixed and EdU was detected using the Click-iT EdU Alexa Fluor 555 Imaging Kit (Life Technologies).
Oenocyte cell and nuclear volumes {#sec017}
---------------------------------
Oenocyte cell volumes were measured from 1--2 oenocyte clusters from 5 different larvae per genotype. Oenocyte clusters expressing a CD8::GFP reporter were scanned on a Nikon Digital Eclipse C1 confocal microscope. The Z-stack was adjusted to cover the entire cluster and sections spaced 2 μm apart were collected in three channels. The channels were separated using ImageJ software (<http://imagej.nih.gov/ij/>[)]{.ul} and the GFP channel was imported to Amira software (Visage Imaging, Inc.) for 3D volume analysis. The clusters were automatically segmented followed by manual verification of the segmentation process and the volume of each segmented cluster was calculated. The settings of the entire procedure were kept identical for all scanned clusters. Nuclear volumes were measured as above on individually segmented nuclei from DAPI stained images of 1 oenocyte cluster from 12 different larvae of each genotype. For fat body Flp-out clones, area measurements were calculated using the analysis measurement tool of Photoshop (Adobe Systems) with manual segmentation of GFP expressing cells using the magnetic lasso tool. For oenocyte Flp-out clones, volume measurements used Amira to segment the whole cluster automatically but with manual verification, followed by manual division into GFP-positive and GFP-negative cells. Volumes of GFP-positive cells were normalized to the mean volume of all GFP-negative oenocytes from the same cluster.
Production of recombinant FarO {#sec018}
------------------------------
The FarO ORF (clone LP02712, DGRC Bloomington) was subcloned into pENTR (Invitrogen), modified to remove the *Nco*I site \[[@pgen.1006154.ref057]\]. The ORF was amplified by PCR using InFusF1 and InFusR1 linker primers to attach 15 nt overlapping the pENTR insertion site. pENTR was amplified using pENTR F4 and pENTR R5 primers. All PCR reactions used proof-reading polymerase (Pfu Ultra II HS; Agilent). Products were purified from agarose gels, joined by Gibson assembly using the Infusion HD cloning kit (Clontech, Mountainview, CA), and completely sequenced at the Nevada Genomics Center to verify integrity. The ORF was then transferred into Baculo-Direct baculoviral DNA using LR recombinase (Invitrogen), and transfected into Sf9 cells by standard methods \[[@pgen.1006154.ref057]\]. A high titre P3 viral stock was produced by successive amplification of P1 and P2 stocks, and re-confirmed by sequencing. To produce recombinant FarO in Sf9 cells, liquid cultures were seeded at a concentration of 0.8x10^6^ cells/ml with 10% fetal bovine serum in SF900 unsupplemented media (Life technologies). Cultures were incubated at 27°C with shaking at 1,300 rpm on orbital shakers for 72 hr and centrifuged at 3000 rpm in the Beckman GS-6R centrifuge for 10 min at 4°C. The supernatant was discarded and the pellet was washed twice by successive resuspension in 5 ml of 100 mM Tris-HCl pH 7. The washed pellet was resuspended in cell lysis buffer \[(CLB); 10 ml 100 mM Tris-HCl pH 7 with the addition of 100 μM DTT, 0.5 mM PMSF and 10 μl of protease inhibitor cocktail (Sigma)\]. Pelleted cells were resuspended in 3ml CLB and sonicated using a Branson hand held sonifier (VWR Scientific) with 15 one second bursts repeated 3 times. 1ml aliquots were separated and microcentrifuged at 13,000 rpm for 20 min at 4°C. Supernatants were either used directly for functional assays or for microsomal preparations. Microsomes were prepared by centrifuging the lysate supernatant at 53,000 rpm in a Beckman Optima MaxE ultracentrifuge for 1 h at 4°C, removing the supernatant, and resuspending the microsomal pellet in cell lysis buffer. Genscript (Piscataway, NJ) was contracted to produce rabbit antisera to a FarO peptide (amino acids 412--425). FarO production in Sf9 cells was measured by Western blotting of microsomal and supernatant fractions. Recombinant housefly cytochrome P450 reductase (CPR) was similarly produced as described previously \[[@pgen.1006154.ref057]\].
FarO functional assays {#sec019}
----------------------
Recombinant FarO and CPR were assayed for reductase activity by incubating infected Sf9 cell lysate preparations with NAD(P)H and fatty acid (Sigma-Aldrich) or fatty acyl-CoA (Avanti Polar Lipids) substrates. Briefly, cell lysate supernatants in CLB were incubated in 100mM Tris HCl pH 7.0, 100 μM DTT, 0.5 mM PMSF and 10 μl of protease inhibitor cocktail (Sigma) supplemented with 120--150 μM 26:0--CoA or 24:0--CoA and 2.3 mM NADH (Fisher Scientific) or NADPH (Sigma Aldrich). All listed concentrations are final in 600--1000 μl reaction volumes. Samples were incubated at 30°C for 2 h and then extracted twice with hexane:ether (50/50, v/v) into glass vials. Samples were dried down to completion under N~2~ gas, resuspended in pure hexane and analyzed by gas chromatography using either a DB-5 column (Agilent) or a Shimadzu non-polar polysiloxane column, (catalog number: 220-94536-01, phase: SHR5XLB) with the following profile: injector temperature 150°C, FID temperature 300°C; program:160°C for 0.2 min, ramp to 265°C at 15°C/min, ramp to 295°C at 5°C/min, and hold for 5 min at 295°C.
Supporting Information {#sec020}
======================
###### Quantitations of neutral lipid content.
The relative lipid contents, calculated from neutral LipidTOX stainings, for experiments shown in the main Figures. (A,B) Flp-out clones in the fat body expressing Dp110 in NR (A) or Dp110^DN^ in Fed~48~ larvae (B). (C-E) *PromE-GAL4* controls in Fed~48~ and NR larvae (C) and *PromE-GAL4* driven expression of Dp110 or Dp110^DN^ in Fed~48~ (D) and NR (E) larvae. (F,G) *PromE-GAL4* driven expression of RNAi for Acc, Dgat1, Cyp4g1 and Cpr, or of Lsd2 overexpression in Fed~48~ (F) and NR (G) larvae. (H) *Cyp4g1*^*Δ4*^ mutant NR larvae. (I,J) *PromE-GAL4* driven expression of RNAi for Kar or FarO in Fed~48~ (I) and NR (J) larvae. Quantifications in (A-E) correspond to [Fig 1](#pgen.1006154.g001){ref-type="fig"}, (F-H) correspond to [Fig 2](#pgen.1006154.g002){ref-type="fig"} and (I,J) correspond to [Fig 4](#pgen.1006154.g004){ref-type="fig"}. In this and subsequent graphs, error bars represent 1 s.d. and asterisks show statistical significance in Student t tests (\*p\<0.05, and \*\*p\<0.001), compared to the one fold control condition unless otherwise indicated
(TIF)
######
Click here for additional data file.
###### Quantitations of cell size and membrane p-Akt expression.
(A, B) Fat body cell areas in Flp-out clones. Dp110 overexpression increases cell size relative to controls in NR larvae (A) and Dp110^DN^ expression decreases cell size relative to controls in Fed~48~ larvae (B). (C, D) Quantitation of Oenocyte membrane p-Akt expression. Membrane p-Akt expression increases significantly in Fed~48~ (C) or NR (D) larvae following *PromE-GAL4* driven Dp110 overexpression whereas the decreases following Dp110^DN^ expression are not statistically significant. (E-G) Relative oenocyte volumes in Fed~48~ (E, G) or NR (F) larvae significantly increase following *PromE-GAL4* driven Dp110 overexpression or Flp-out clonal expression of myrAkt. Oenocyte volumes significantly decrease following *PromE-GAL4* or Flp-out clonal expression of Dp110^DN^.
(TIF)
######
Click here for additional data file.
###### NR represses PI3K signaling more strongly in fat body than oenocytes.
Panels in top two rows show endogenous FoxO expression in fat body (left) and oenocytes (right, marked with *PromE\>mGFP*) from Fed~48~ and NR~66~ larvae. Redistribution of FoxO from the cytoplasm to the nucleus (associated with decreased PI3K signaling) during NR is more pronounced in the fat body than in oenocytes. The nuclei of NR oenocytes (white arrowhead) express less FoxO than the nuclei of neighboring NR epidermal cells (yellow arrowhead). Panels in the bottom row show expression of the *tGPH* reporter for PI3K activity in fat body (left) and oenocytes (right) from Fed~48~ and NR~66~ larvae. Membrane expression of EGFP fused to the pleckstrin homology domain of Grp1 (a readout for PI3K activity in some but not in all cell types) is clearly decreased during NR in fat body but in oenocytes the NR change is less noticeable. Scale bar is 10 μm.
(TIF)
######
Click here for additional data file.
###### Ilp6 is not required for NR induction of lipid droplets in larval oenocytes.
Clusters of oenocytes from NR larvae, marked with streptavidin-A555 and showing lipid droplets (LipidTOX) and nuclei (DAPI). (A) Larvae deficient for Ilp6 (*Df(1)Ilp6*) show lipid droplet induction similar to controls (*w*^*1118*^). (B) Larvae with fat body-specific RNAi knockdown of Ilp6 (*Cg\>Ilp6 Ri*) show lipid droplet induction similar to controls (*Cg-GAL4*). Scale bar is 20 μm. (C) Most *Acc*, *Kar*, *Cpr* and *Cyp4g1* RNAi larvae retain watertight trachea. Graph shows the percentage of larvae of various genotypes showing tracheal flooding at the early L3 stage. Asterisks indicates p\<0.01. Each larval genotype carried the oenocyte-specific driver Pro-*mE-GAL4* and the respective *UAS-RNAi* transgene indicated. The control genotype used was *PromE-GAL4 crossed to w*^*1118*^ and \"oenocyte-less\" refers to *PromE\>reaper*.
(TIF)
######
Click here for additional data file.
###### Quantitation of protein knockdowns with Acc, Kar, Cyp4g1 and Cpr RNAi.
\(A\) Antibody staining for Acc, Kar, Cyp4g1 and Cpr in oenocytes following *PromE-GAL4* driven RNAi in NR larvae. Scale bar is 20μm. (B) Staining for Kar in nlsGFP positive Flp-out clones expressing Kar Ri in oenocytes. Scale bar is 10μm. (C) Graph of relative immunostaining intensity for Kar protein in Kar Ri oenocyte Flp-out clones during NR. (D-G) Graphs of relative staining intensity in oenocytes for Acc (D), Kar (E), Cyp4g1 (F) and Cpr (G) proteins following *PromE-GAL4* driven RNAi knockdown of the corresponding genes in NR larvae.
(TIF)
######
Click here for additional data file.
###### FarO and Kar do not limit the size of oenocytes at the late larval stage.
\(A\) Neither Kar nor FarO RNAi detectably alter lipid droplets in the fat body. Flp-out clones for Kar RNAi (top row) or FarO RNAi (bottom row), marked with nlsGFP, show no detectable change in lipid droplets (LipidTOX), nuclear size or cell size in NR larvae. Nuclei are marked with DAPI and the scale bar is 10μm. (B, C) Oenocyte FarO Ri Flp-out clones in NR larvae, marked by nlsGFP, show decreased LipidTOX staining (B) and significantly decreased relative neutral lipid content (C) than control neighboring cells. Oenocyte Kar expression is not altered in FarO Ri Flp-out clones. (D, E) FarO and Kar do not regulate oenocyte volumes at the late larval stage. *PromE-GAL4* mediated expression of *Kar* or FarO RNAi alone or in combination with *Dp110*^*DN*^ in Fed~66~ larvae. In contrast to Fed~48~ and NR~66~ larvae, oenocyte volumes in Fed~66~ larvae are not significantly altered by *Kar* or *FarO* RNAi, although they remain PI3K dependent.
(TIF)
######
Click here for additional data file.
###### Genetic interactions between FarO and PI3K.
\(A\) In Fed~48~ larvae, *PromE\>Dp110*^*DN*^ induces a low level of oenocyte lipid droplets that are suppressed in *PromE\>FarO Ri Dp110*^*DN*^. In NR larvae, *PromE\>Dp110*^*DN*^ does not noticeably alter lipid droplet induction and thus inhibition of NR droplet induction is similar in *PromE\>FarO Ri* and in *PromE\>FarO Ri Dp110*^*DN*^. In both Fed~48~ and NR larvae, oenocyte volume in *PromE\>FarO Ri Dp110*^*DN*^ is intermediate between that of *PromE\>FarO Ri* alone and *PromE\>Dp110*^*DN*^ alone. Panels show single confocal sections and the scale bar is 10 μm. (B) Relative neutral lipid content of oenocytes from Fed~48~ larvae for the genotypes in panel A. Note that FarO is required for the increase in lipid droplets in fed larvae following PI3K inhibition. (C) Oenocyte volumes for the genotypes in panel A in Fed~48~ and NR~66~ larvae.
(TIF)
######
Click here for additional data file.
###### Kar but not FarO RNAi significantly increases oenocyte membrane p-Akt.
(A, B) Quantitation of membrane p-Akt intensity, relative to the control genotype, in NR oenocytes expressing FarO RNAi (A) or Kar RNAi (B). Membrane p-Akt intensity increases 1.7 fold with Kar RNAi but does not change significantly with FarO RNAi. (C) Panels show p-Akt staining in an oenocyte cluster (marked with mGFP) from control and Kar knockdown (*PromE\>Kar* RNAi) NR~66~ larvae. Kar RNAi knockdown is associated with increased membrane p-Akt expression. Scale bar is 10 μm.
(TIF)
######
Click here for additional data file.
We are grateful to Matthias Llandgraf, Jacques Montagne, Leslie Pick, and Joel Levine for fly stocks and to Pierre Leopold for antibodies. Stocks obtained from the Bloomington *Drosophila* Stock Center (NIH P40OD018537) and the Kyoto *Drosophila* Genetic Resource and Vienna *Drosophila* RNAi Centers were used in this study. We thank Eugenio Gutierrez for making the Cyp4g1 RNAi lines. We acknowledge technical assistance from Tsz Fong (Cindy) Cho, James MacRae and Liron Yanovsky and thank Andrew Bailey, Fumiaki Obata, Irina Stefana and Vanessa Tixier for advice and critical reading of the manuscript.
[^1]: The authors have declared that no competing interests exist.
[^2]: Conceived and designed the experiments: EC RM AS LPW CT GJB APG. Performed the experiments: EC RM AS LPW. Analyzed the data: EC RM AS LPW CT ZP APG. Contributed reagents/materials/analysis tools: CT ZP APG. Wrote the paper: EC CT APG.
[^3]: Current address: Department of Immunology and Cancer Research, Institute for Medical Research Israel Canada (IMRIC), Hebrew University-Hadassah Medical School, Jerusalem, Israel
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec0005}
===============
Ileorectal intussusception is a rare condition in adults in which the distal ileum, caecum, variable lengths of ascending and transverse colon, and associated mesentery invaginate into the rectum. Case reports on this rare condition in the adult population are limited [@bib0005; @bib0010; @bib0015; @bib0020].
2. Presentation of case {#sec0010}
=======================
We present the case of a 56-year-old man who presented to our hospital for investigation of rectal bleeding. The patient was an otherwise healthy male who had experienced two days of bright red rectal bleeding and colicky abdominal pain. On further questioning, he described an unintentional weight loss of 5 kg, anorexia, borborygmi, and altered bowel habit over a three-month period. Abdominal examination revealed a soft but slightly distended abdomen with a palpable mass in the right upper quadrant. A digital rectal exam demonstrated a mass in the rectum and rigid sigmoidoscopy revealed a large tumour in the rectum.
The patient had work up for a suspected rectal carcinoma with a staging CT scan. The CT revealed an extensive ileorectal intussusception ([Fig. 1](#fig0005){ref-type="fig"} red arrows). The intussusceptum consisted of distal ileum, caecum, ascending, and proximal transverse colon. The splenic flexure, descending colon, sigmoid colon, and rectum acted as the intussuscipiens. There was extensive involvement of the ileo-colic mesentery in the intussusceptum. A large soft tissue mass was demonstrated in the rectum and thought to represent the lead point ([Fig. 2](#fig0010){ref-type="fig"}b arrow). Also noted were extensive liver lesions, which were suspicious for colorectal metastases ([Fig. 1](#fig0005){ref-type="fig"} blue arrows). The CT revealed signs concerning for possible ischaemia in the distal portion of the intussusceptum ([Fig. 2](#fig0010){ref-type="fig"}a arrow) and a decision was made to immediately proceed to exploratory laparotomy.
A large caecal tumour, chronically intussuscepted into the rectum, was discovered during the laparotomy ([Fig. 3](#fig0015){ref-type="fig"}). The colon at the hepatic flexure was completely mobile due to a long mesentery and lack of posterior peritoneal attachment. There was no evidence of bowel ischaemia. A decision was made to carefully reduce the intussusceptum and avoid a total colectomy given the extent of the intussusception. The caecal mass was carefully milked back from the rectum to the transverse colon and an extended right hemicolectomy with a primary anastomosis was performed. Histology of the tumour revealed a mucinous caecal adenocarcinoma with clear surgical margins (Dukes Stage D AJPP stage pT3 N1a M1). The final histology was consistent with the mass originally biopsied with a rigid sigmoidoscope. The patient recovered well and was discharged home six days post operatively. The patient was considered for adjuvant chemotherapy and possible down staging of the liver metastasis before possible resection in a subsequent multidisciplinary meeting.
3. Discussion {#sec0015}
=============
Intussusception in adults is a rare diagnosis, accounting for only 5% of intussusception cases [@bib0025]. When cases occur in adults, they are generally secondary to a pathological solid lesion creating the lead point although cases of diverticulum, including Meckel's diverticulum, have been reported in the literature [@bib0030]. Adult intussusception is a difficult diagnosis to make since it may not present with the classical triad of crampy abdominal pain, vomiting and bloody stools. Adults can have a wide variety of symptoms such as abdominal pain, haematochezia, abdominal mass, altered bowel habits, and bowel obstruction [@bib0035]. Due to the varied clinical presentations, radiological imaging plays a key role in diagnosis. CT scanning of the abdomen is the most sensitive radiological investigation for adult intussusception and can be useful in establishing a diagnosis, locating a causative lesion and planning the operative strategy [@bib0025; @bib0040]. Due to the high rate of malignant cause of adult intussusception, laparotomy and bowel resection without preoperative reduction is the advocated management strategy and intraoperative reduction is often discouraged for fear of tumour rupture and potentially upstaging the cancer. However, no high level evidence is available to provide an answer to this question and the topic is debated [@bib0045; @bib0050; @bib0055].
What is unique in this case is the extent of the intussusception; there have only been a few case reports of adult ileo-rectal or ileo-anal intussusception [@bib0005; @bib0010; @bib0015; @bib0020]. The failure in the embryologic development of the mesenteries of the ascending and descending colon to blend with the posterior abdominal wall peritoneum by the process of zygosis contributes to this rare phenomenon. The colon is thus straighter and allows the intussusceptum to migrate large distances freely without obstruction.
4. Conclusion {#sec0020}
=============
Intussusceptions are a rare diagnosis in the adult population. Presentations are varied and radiological imaging plays a key role in diagnosis. In cases with extensive ileorectal or ileoanal intussusception, a balance needs to be reached between extensive resection that increases the morbidity of the procedure verses the risk of upstaging the malignancy. The decision to reduce the intussusception ultimately requires the application of surgical judgement specific to each patient. In our patient, the intussusception was reduced due to pre-existing distal metastasis.
Conflicts of interest {#sec0025}
=====================
None of the authors have any conflicts of interest to declare.
Funding {#sec0030}
=======
The article was funded entirely by the authors.
Ethical approval {#sec0035}
================
N/A.
Author contribution {#sec0040}
===================
Dr. Robertson -- article conception, image creation, literature review, writing the article and editing the article.
Dr. Due -- article conception, writing the article and editing the article.
Dr. Shimokawa -- image creation, writing the article and editing the article.
Dr. Yeow -- writing the article and editing the article.
Consent {#sec0045}
=======
Written informed consent was obtained from the patient prior to the writing of the case report.
Guarantor {#sec0050}
=========
Dr. Robertson will act as the guarantor for this article.
{#fig0005}
{#fig0010}
{#fig0015}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Due to advances in operation techniques and novel treatments (targeted therapy and PD-L1 immunotherapy) for patients with non-small cell lung carcinoma (NSCLC), the prognosis of NSCLC is improved. However, the 5-years survival rate of NSCLC patients is still \<20% and NSCLC remains a major health problem worldwide. Therefore, it is urgent to discover novel targets to regulates the carcinogenesis and metastasis of NSCLC and to develop new therapeutic agents for the treatment of NSCLC ([@B1], [@B2]). Recently, accumulating evidence indicates that abnormal regulation of acetylation processes plays a vital role in lung carcinogenesis. For example, histone deacetylases (HDACs) and histone acetyltransferases (HATs) can significantly change the nucleosome conformation of tumor cells through post-translational modifications of the N-terminal tails of core histones ([@B3], [@B4]). However, the expression level and clinical characteristics of HDAC10 in NSCLC tissue is not clear. Tumor-associated PD-L1 (B7 homolog 1, B7-H1) can block tumor-specific T cell-mediated immunity through inducing apoptosis of T cells, suppressing the secretion of cytokines and disturbing the function of activated T cells ([@B5], [@B6]). Moreover, recent findings suggested that activation of HDACs could induce PD-L1 expression in various types of cancer, especially in myeloma and B-cell lymphomas ([@B7]--[@B9]). Thus, we hypothesized that high-level expression of HDAC10 is closely associated with PD-L1 expression and poor prognosis of patients with NSCLC.
Herein, we evaluated HDAC10 and PD-L1 expression in patients with NSCLC. Furthermore, the correlation of HDAC10 and PD-L1 was analyzed. HDAC10 expression and poor prognosis in patients with NSCLC receiving pulmonary resection was also investigated.
Materials and Methods {#s2}
=====================
Patients
--------
From April 2004 and August 2009, a total of 180 pulmonary squamous carcinoma and adenocarcinoma patients from the 1st Affiliated Hospital of Jinzhou Medical University (Jinzhou, China) who were receiving complete pulmonary resection and systematic lymph node dissection were enrolled. During the enrollment period, all patients were pathologically diagnosed for the first time as having NSCLC. Based on the International Association for the Study of Lung Cancer TNM classification system, all of the patients with NSCLC were classified into TNM stages ([@B10]). The enrolled patient were required to have integrated clinicopathological information and follow-up data. Strictly based on the National Comprehensive Cancer Network Clinical Practice Guidelines on NSCLC, all of the patients received postoperative treatment ([@B11]). Exclusion criteria were as follows: patients receiving preoperative anticancer treatment (including neo-adjuvant chemotherapy, radiotherapy, or biotherapy), patients with previous or simultaneous cancers, patients who died within 1 month after surgery or died from non-cancer diseases.
Until August 30, 2014, all of the patients were followed up. In the retrospective study, patients who were still alive after the last follow-up were censored. Overall survival (OS) means the period between the time of surgery and the last follow-up or death.
Immunohistochemistry (IHC) and Scoring
--------------------------------------
Monoclonal mouse anti-HDAC10 antibody (Abcam, No. ab108934) and polyclonal rabbit anti-PD-L1 antibody (Abcam, No. ab213524) for IHC of NSCLC tissues were purchased from Abcam Inc. 180 samples of normal lung tissue were taken from tissue that was 10 cm from the cancer, and were used as para-cancer samples. In line with the viewpoints of two pathologists who were blinded to identity of the NSCLC tissues. Positive staining of HDAC10 and PD-L1 is taken to show the nuclear or cytoplasmic staining of tumor cells. Considering both the staining intensity and the proportion of cells stained, the IHC score was determined by a semi-quantitative method ([@B12]). The expression of analyzed makers were assessed using a semi-quantitative method, based on the intensity of color reaction (on a scale of + to +++), and the rate of immunopositive cells (within the ranges of 1--20%, 21--40%, 41--60%, 61--80%, 81--100%) ([@B13]). Based on these two variables, a numerical ratio of the markers expression was estimated and used for further analyses. The expression level was further classified into low expression (≤1), moderate expression (1.5-6) and strong positive expression (≥7.5) for both HDAC10 and PD-L1.
Database for HDAC10 and PD-L1 Expression in Patients With NSCLC
---------------------------------------------------------------
To provide evidence for our findings, we also searched HDAC10 expression in the Oncomine database ([@B14]). A total of 226 patients with NSCLC and 20 normal lung tissues were analyzed. Based on the Protein Atlas database ([@B15]), the expression of CD274 (PD-L1) in various types of cancer was also investigated. To verify the correlation of HDAC10 and CD274 in patients with NSCLC, an R2 database was used to analyze the relationship of HDAC10 and PD-L1 ([@B16]).
Statistical Analysis
--------------------
SPSS22.0.0 software was used to perform statistical analysis. The correlation of HDAC10 and PD-L1 expression was analyzed by Spearman\'s correlation method. OS (Overall Survival) was calculated by the Kaplan-Meier method and analyzed by log-rank test. Based on Cox\'s proportional hazard model, multivariable analysis of as an independent factor for survival was investigated. *P* \< 0.05 was considered statistically different.
Results {#s3}
=======
Expression Level of HDAC10 and Clinicopathological Features of all Patients With NSCLC
--------------------------------------------------------------------------------------
The clinicopathological features of all examined cases are shown in [Table 1](#T1){ref-type="table"}. Out of 180 NSCLC cases, 64 (35.6%) patients were classified in the high-level of the HDAC10 group. No significant differences were observed between HDAC10 expression and patient characteristics (gender, age, tumor size, and stages). Positive immunostaining of HDAC10 was mainly located in the nucleus and cytoplasm of carcinoma cells ([Figure 1A](#F1){ref-type="fig"}). Noticeably, we observed that the expression score of HDAC10 in NSCLC tissue of 180 patients was significantly higher than that in the corresponding para-cancer tissues (*P* \< 0.001). Meanwhile, we searched for HDAC10 expression in patients with NSCLC as compared to normal lung tissue in the Oncomine database. As shown in [Figure 1B](#F1){ref-type="fig"}, the expression level of HDAC10 in NSCLC tissues is 1.55-fold higher than that in normal lung tissues (*p* = 0.020, data were obtained from <https://www.oncomine.org/resource/login.html>). This result is consistent with our findings.
######
The clinicopathological features and analysis of all examined cases.
**Clinicopathological features** **OS**
---------------------------------- ------------ -------------------- --------- --------- ---------- -----------
**N (%)** **Median, months** **UVA** **MVA**
***P*** ***P*** ***HR*** **95%CI**
Gender 0.79 0.34 1.2 0.36--3.8
Male 100 (55.6) 54
Female 80 (44.4) 51.5
Age 0.13 0.12 1.7 0.85--3.6
\>60 98 (54.4) 52.5
≤60 82 (45.6) 56.5
Size 0.9 0.32 1.0 0.52--2.1
\>5 55 (30.6) 53
≤5 125 (69.4) 56
T 0.19 0.92 1.6 0.78--3.4
T1 36 (20) 66
T2 102 (56.7) 54
T3 32 (17.8) 49.5
T4 10 (6) 14
N 0.04 0.53 2.0 1--3.8
N0 74 (41.1) 63.5
N1 68 (37.8) 56
N2 32 (17.8) 43.5
N3 6 (3) 24
M 0.08 0.77 6.1 0.8--46
M0 178 (98.9) 54
M1 2 (1.1) 14
TNM 0.00 0.02 2.4 1.5--3.8
I 54 (30) 81
II 78 (43.3) 64
III 46 (25.6) 54
IV 2 (1.1) 16
PDL1 0.38 0.76 0.81 0.51--1.3
Low 78 (43.3) 62.5
Moderate 46 (25.6) 53.5
High 56 (31.1) 44
HDAC10 0.00 0.00 2.6 1.5--4.5
Low 56 (31.1) 72
Moderate 60(33.3) 54.5
High 64 (35.6) 43
{#F1}
High-Level HDAC10 Expression Is Positively Associated With PD-L1 Expression in Patients With NSCLC
--------------------------------------------------------------------------------------------------
Given that activation of HDACs could regulate PD-L1 expression in myeloma and B-cell lymphomas ([@B17], [@B18]), we further searched for PD-L1 (CD274) in the Protein Atlas website (<https://www.proteinatlas.org/ENSG00000120217-CD274/pathology>). As seen in [Figure 2A](#F2){ref-type="fig"}, PD-L1 expression in lung cancer is scored at a high-level for all types of cancer. Next, the PD-L1 expression level in NSCLC tissue and normal lung tissue were determined and quantified. The high level of positive staining of PD-L1 was observed in NSCLC tissues, as compared to normal lung tissue ([Figure 2B](#F2){ref-type="fig"}). Univariate and multivariate analysis showed that there is no significant correlation of PD-L1 expression and the prognostic survival after surgery (*P* = 0.76). What is more, the correlation of HDAC10 and PD-L1 expression in NSCLC tissues was analyzed. Patients with high-level expression of HDAC10 often show the overexpression of PD-L1 ([Figure 3A](#F3){ref-type="fig"}). Interestingly, the expression level of HDAC10 is positively correlated with PD-L1 expression in NSCLC tissue (*r* = 0.213, *P* \< 0.05). To provide evidence that the positive relationship of HDAC10 and PD-L1 (CD274) in patients with NSCLC, the correlation of these two targets (HDAC10 and CD274) was analyzed (<https://hgserver1.amc.nl/cgi-bin/r2/main.cgi>). As shown in [Figure 3B](#F3){ref-type="fig"}, the expression level of HDAC10 is positively associated with CD274 expression in patients with NSCLC (p=0.020, n=410). This result further confirmed our findings.
{#F2}
{#F3}
High-Level HDAC10 Expression Is Associated With a Poor Prognosis in Patients With NSCLC
---------------------------------------------------------------------------------------
According to the overall survival data from 180 patients with NSCLC, a Kaplan-Meier curve was drawn and is shown in [Figure 4](#F4){ref-type="fig"}. The follow-up time ranges from 22 to 118 months, and the average time is 72 months (median: 48 months). There were 31 deaths and 149 survivors at the last follow-up. We noticed that adenocarcinoma cases are usually at a lower stage than the squamous cell carcinoma cases, reflecting a better outcome of adenocarcinoma than that of squamous cell carcinoma. Furthermore, the result of univariate and multivariate analysis showed that HDAC10 expression is an independent prognostic factor for NSCLC OS. For stage-I and -II NSCLC, patients had shorter OS time than those with stage III (*P* = 0.02, [Figure 5](#F5){ref-type="fig"}). The survival time was significantly different between the groups with high-level and low-level expression of HDAC10 (*P* = 0.002, [Figure 5](#F5){ref-type="fig"}).
{#F4}
{#F5}
In addition, the OS data of 106 patients with NSCLC from the R2 database was analyzed (<https://hgserver1.amc.nl/cgi-bin/r2/main.cgi>) ([@B19]), and the Kaplan-Meier curve suggests that high-level expression of HDAC10 is positively associated with poor prognosis ([Figure S4](#SM4){ref-type="supplementary-material"}). This result provides proof that HDAC10 could be used as a poor prognosis indicator for patients with NSCLC.
Discussion {#s4}
==========
HDAC10, a member of the class II HDAC family, plays a key role in multiple biological processes, including cell proliferation, cell differentiation, cell cycle regulation, homologous recombination, DNA mismatch repair, autophagy, and so on ([@B20]--[@B24]). Moreover, HDAC10-mediated pathways are closely associated with tumorigenesis and metastasis in various types of cancer. For example, HDAC10 promotes cell proliferation through activation of AKT in lung cancer ([@B25]). HDAC10 also regulates the Let-7/HMGA2/Cyclin A2 signaling pathway and further impacts the G~2~/M phase transition in NSCLC ([@B26]). In human colorectal cancer, HDAC10 expression is associated with DNA mismatch repair genes. HDAC10 mRNA levels correlate with the platinum sensitivity of ovarian carcinoma cells ([@B27]). Suppression of HDAC10 could initiate HSP70-mediated autophagy and activate CMA (chaperone-mediated autophagy) and induce degradation of a CMA substrate in HeLa cells ([@B24]). Thus, HDAC10 may be a potential target for the diagnosis and treatment of many cancers.
Recently, researchers have found that class II HDACs (including HDAC10) are associated with poor prognosis independently of any clinicopathologic feature ([@B28]). Among the class II HDAC family, *HDAC10* is significantly correlated with poor prognosis in renal cancer, melanoma and gastric cancer patients ([@B6], [@B29], [@B30]). However, the expression of HDAC10 and its association with clinicopathological features in NSCLC is not clear. Herein, we analyzed 180 NSCLC patients receiving complete pulmonary resection and systematic lymph node dissection from April 2004 to August 2009. We observed that HDAC10 expression in lung cancer tissue is significantly higher than that in corresponding para-cancer tissue. Moreover, the Oncomine database demonstrated that HDAC10 expression in NSCLC tissues is 1.55-fold higher than in normal lung tissues ([Figure 1B](#F1){ref-type="fig"}) ([@B15]). This result confirmed that HDAC10 expression in NSCLC tissues is higher than that in normal lung tissues. Our findings suggest that HDAC10 could be a promising biomarker that might provide novel strategies for clinicians to improve efficacy for treatment of NSCLC. Furthermore, based on the overall survival data from 180 patients with NSCLC, a Kaplan-Meier curve showed that HDAC10 expression is an independent prognostic factor for NSCLC OS. Noticeably NSCLC patients in stage-I and -II had a shorter OS time than those with stage-III ([Figure 5](#F5){ref-type="fig"}). The survival time between groups with high-level and low-level expression of HDAC10 had significant differences. Meanwhile, the OS data of 106 patients with NSCLC from the R2 database suggests that high-level expression of HDAC10 is positively associated with poor prognosis ([Figure S4](#SM4){ref-type="supplementary-material"}). Our result provides proof that HDAC10 could be serve as a poor prognosis indicator for patients with NSCLC.
Tumor-associated PD-L1 (CD274) can block tumor-specific T cell-mediated immunity by inducing apoptosis of T cells, suppressing the secretion of cytokines and disturbing the function of activated T cells ([@B31], [@B32]). PD-L1 expression by tumors and its interaction with PD-1-expressing T cells in the tumor microenvironment can lead to immunotherapy tolerance. Targeting this co-inhibitory axis has proven clinically effective in the treatment of NSCLC ([@B33]). Meanwhile, recent findings suggested that activation of HDACs could induce PD-L1 expression in various types of cancer, especially in myeloma and B-cell lymphomas ([@B7], [@B8], [@B34]). Though the relationship of HDAC10 and PD-L1 expression with poor prognosis in patients with NSCLC is unclear, it is possible that HDAC10 is related to PD-L1 expression in NSCLC. Herein, we found a stronger positive staining of PD-L1 in NSCLC as compared to normal lung tissue. Interestingly, we noticed that HDAC10 is positively correlated with PD-L1 expression in patients with NSCLC. Based on the R2 database, the correlation of these two targets (HDAC10 and CD274) was evaluated. It provides evidence that HDAC10 is positively associated with CD274 expression in NSCLC (p=0.020). Targeting PD-L1 with monoclonal antibodies has significantly impacted the treatment landscape for NSCLC during the last 5 years. Due to the lack of definite biomarkers to select as optimal responders, only about 20% of patients with advanced NSCLC respond to PD-L1-based immunotherapy. Herein, we found that patients with high-level expression of HDAC10 often show PD-L1 overexpression. HDAC10 can be used as a potential biomarker for PD-L1 treatment. In addition, combination approaches to PD-L1-based immunotherapy are currently designed to re-energize the immune system with a synergetic mechanism and could achieve durable antitumor effects in NSCLC. Our findings also suggest the potential combination of PD-L1 inhibitors in NSCLC alongside targeted inhibition of HDAC10.
Given that the HDAC inhibitor (HDACi) pembrolizumab is currently in clinical trials for use in lung cancer patients ([@B35]), our findings will be helpful to establish clinical study protocols. What is more, our results suggested that HDAC is might act on the immune networks of lung cancer cells, including changes in the profiles of co-stimulatory molecules, co-stimulatory antigens and cytokines of cancer cells. Meanwhile, our results also indicated that there is an intimate connection between targeting epigenetics and PD-L1-based immunotherapy. Therefore, the expression level of HDAC10 in NSCLC might predict the therapeutic response to HDAC inhibitors. Similar to other targeted therapeutics, treatment response might be especially prominent in patients with overexpressed HDAC10 ([@B28], [@B36]). In addition, there were two limitations in this study. Firstly, we investigated the expression of HDAC10 and PD-L1 mostly in patients with TNM stage I to IIIa who had the opportunity to receive surgical therapy. For TNM stage-IV NSCLC patients, the data is unavailable. Secondly, our results suggest the connection of HDAC10 and PD-L1, while the precise mechanisms of PD-L1 expression in NSCLC will be addressed in the future.
Conclusions {#s5}
===========
Taken together, our findings suggest clinical relevance for HDAC10 and PD-L1 (CD274) expression and provide a rationale strategy for the combination of PD-L1 blockade and HDAC10 inhibition.
Data Availability Statement {#s6}
===========================
The datasets generated for this study are available on request to the corresponding author.
Ethics Statement {#s7}
================
This study was carried out in accordance with the recommendations of 1st Affiliated Hospital of Jinzhou Medical University Medical Ethics Committee. The protocol was approved by the 1st Affiliated Hospital of Jinzhou Medical University Medical Ethics Committee. All subjects gave written informed consent in accordance with the Declaration of Helsinki ([Figures S1](#SM1){ref-type="supplementary-material"}--[S3](#SM3){ref-type="supplementary-material"}).
Author Contributions {#s8}
====================
XL and YZ performed the experiments. YuW, RZ, TJ, LQ, QJ, JZ, JS, ZW, LW, TL, XM, YiW, and NW participated in the experiments and drafted the manuscript.
Conflict of Interest {#s9}
====================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We gratefully thank all the staff in the Department of Oncology, The First Affiliated Hospital of Jinzhou Medical University for collecting samples of patients with NSCLC. Tissue microarrays were co-evaluated by our hospital staff and Shanghai Outdo Biotech Company in a double-blind manner.
**Funding.** This project was supported by funds from the CSCO-HANSOH PHARMA Science Foundation of China (Grant No: Y-HS2019-29) and the Natural Science Foundation of Liaoning Province, China (Grant No: 201602304).
Supplementary Material {#s10}
======================
The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fonc.2020.00485/full#supplementary-material>
######
Proof of medical ethics.
######
Click here for additional data file.
######
Approve of ethics committee.
######
Click here for additional data file.
######
Biomedicine research informed consent form.
######
Click here for additional data file.
######
Kaplan-Meier survival curve of NSCLC patients with high- and low-HDAC10 expression. The data was obtained from the R2 dataset. (<https://hgserver1.amc.nl/cgi-bin/r2/main.cgi>).
######
Click here for additional data file.
[^1]: Edited by: Luisa Lanfrancone, European Institute of Oncology (IEO), Italy
[^2]: Reviewed by: Jan Theys, Maastricht University, Netherlands; Dimitrios Schizas, National and Kapodistrian University of Athens, Greece
[^3]: This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology
[^4]: †ORCID: Ning Wei [orcid.org/0000-0003-2723-4061](http://orcid.org/0000-0003-2723-4061)
| {
"pile_set_name": "PubMed Central"
} |
Many factors, including professional and personal relationships and activities, can influence the design, conduct, and reporting of the clinical science that informs health care decision. The potential for conflict of interest exists when these relationships and activities may bias judgement.[@B1] Many stakeholders---editors, peer reviewers, clinicians, educators, policymakers, patients, and the public---rely on the disclosure of authors\' relationships and activities to inform their assessments. Trust in the transparency, consistency, and completeness of these disclosures is essential.
Ten years ago, the International Committee of Medical Journal Editors (ICMJE) adopted the "ICMJE Form for the Disclosure of Potential Conflicts of Interest" as a uniform mechanism for collecting and reporting authors\' relationships and activities that readers might consider relevant to a published work.[@B2] The goal was to avoid the confusion (and often ensuing controversy) created when journals vary in how they collect and report this information. We believe a uniform disclosure form has been helpful, but problems remain. First, the software supporting the current form is increasingly problematic, making its use difficult or impossible for an increasing number of authors. More important, however, is that many authors and readers misunderstand, misapply, or misinterpret the disclosures.
Although some individuals violate the public trust by purposefully hiding relevant relationships and activities, we believe most authors are committed to transparent reporting and consider it as vital to the advancement of clinical science. Nonetheless, disagreement, confusion, and controversy regarding authors\' disclosures arise when opinions differ over which relationships and activities to report. An author might not report an item that others deem important because of a difference in opinion regarding what is "relevant," confusion over definitions, or a simple oversight. Some authors may be concerned that readers will interpret the listing of any item as a "potential conflict of interest" as indicative of problematic influence and wrongdoing, a concern often raised regarding the requirement to report publicly funded grants. For their part, some readers fail to recognize that their own relationships and activities influence how they assess the work of others and what they deem to be a "conflict" for others or themselves.
We propose several changes to the ICMJE disclosure form to help address these issues. First, words matter. Despite including the word "potential," a form entitled "... for the Disclosure of Potential Conflicts of Interest" may imply that any relationship or activity listed represents a problematic influence or wrongdoing. The proposed new title, "The ICMJE Disclosure Form," aims to dispel that interpretation and potential stigma. Second, we no longer ask authors to decide what might be interpreted as a potential conflict of interest. Authors disclose their relationships and activities so that readers can decide whether these relationships or activities should influence their assessments of the work. Further, to avoid omissions--inadvertent or purposeful--we now provide a checklist of relationships and activities for authors to complete.
We welcome feedback about the proposed new form, which is available with a link to provide comments, at [www.icmje.org](http://www.icmje.org). We will consider comments received by 30 April 2020, before finalizing and adopting a revised version. In the interim, the extant "ICMJE Form for the Disclosure of Potential Conflicts of Interest" will remain in use and available as a downloadable PDF at our website ([Supplementary Data](#S1){ref-type="supplementary-material"}).
In a further step to avoid inconsistencies and omissions, and to help ease the disclosure process for authors, some journals will change the mechanism by which disclosures are collected. Authors are required to provide disclosures to multiple entities (e.g., to academic institutions, continuing education providers, guideline and other committees as well as medical journals). Disclosing information repeatedly, with varying reporting requirements, formats, and definitions, is frustrating for authors and contributes to problematic and controversial discrepancies across disclosures. The ICMJE will therefore accept disclosures from web-based repositories. These enable authors to maintain an inventory of their relationships and activities and create electronic disclosures tailored to the requirements of entities such as ICMJE, without having to reenter information repeatedly.
ICMJE will accept disclosures from repositories that meet the following criteria: collection and reporting of relationships and activities consistent with ICMJE requirements; no fees for individuals to enter, store, or export their data; provision of disclosures to journals electronically as well as an option for journals without a digital interface; and compliant with the General Data Protection Regulation (GDPR).
One currently available repository that is consistent with these criteria is Convey ([www.convey.org](http://www.convey.org)), but we encourage the development of other repositories as necessary to meet regional, linguistic, and regulatory needs. A template that enables authors to create disclosures that emulate the extant "ICMJE Form for the Disclosure of Potential Conflicts of Interest" is already available at the Convey platform, and some of our journals have begun to collect author disclosures electronically in this way. This template will be updated to conform to the new ICMJE Disclosure Form when it is finalized, and all ICMJE journals can begin accepting disclosures in this manner. Ultimately, the currently employed PDF-based ICMJE form will be unavailable.
While no approach to disclosure will be perfect or foolproof, we hope the changes we propose will help promote transparency and trust. We look forward to your feedback.
This article is being published simultaneously in *Annals of Internal Medicine*, *BMJ (British Medical Journal)*, *Bulletin of the World Health Organization*, *Deutsches Ärzteblatt (German Medical Journal)*, *Ethiopian Journal of Health Sciences*, *JAMA (Journal of the American Medical Association)*, *Journal of Korean Medical Science*, *The Lancet*, *New England Journal of Medicine*, *New Zealand Medical Journal*, *Revista Medica de Chile (Medical Journal of Chile)*, *and Ugeskrift for Laeger (Danish Medical Journal)*.
**Disclaimer:** Dr. Sahni\'s affiliation as representative and past president of the World Association of Medical Editors (WAME) does not imply endorsement by WAME member journals that are not part of the ICMJE.
SUPPLEMENTARY MATERIAL
======================
###### Supplementary Data
ICMJE Form for Disclosure of Potential Conflicts of Interest.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .44}
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
As the second leading cause of death after heart disease,[@b1-ott-11-5397] cancer includes a group of diseases characterized by irregular cell division and proliferation. Lung cancer is the most deadly type.[@b2-ott-11-5397] The main reason behind the dismal survival statistics is that most lung cancer patients are terminally ill and cannot be cured with existing therapies.[@b3-ott-11-5397] As far as we know, there is little information about quercetin (QCT) regulating the ultrastructural changes of lung cancer.
QCT, or 3, 3′, 4′, 5, 7-pentahydroxyflavone, 1 of the 6 subclasses of flavonoids, has a wide range of biological activities.[@b4-ott-11-5397] The antitumor, antiallergic, and anti-inflammatory effects of QCT have been extensively reviewed.[@b5-ott-11-5397],[@b6-ott-11-5397] Their biological activities mainly include electron transfer of free radicals, the activation of antioxidant enzymes, and the ability to inhibit oxidative stress.[@b7-ott-11-5397] There is evidence that QCT is capable of targeting different types of cancer cells including those of leukemia, breast cancer, esophageal cancer, colon cancer, prostate cancer, nasopharyngeal carcinoma, endometrial cancer, and lung cancer.[@b8-ott-11-5397],[@b9-ott-11-5397] The proliferation of these malignant cells may be inhibited by QCT; however, the exact molecular mechanisms underlying the effects of QCT are unknown. For the treatment of these pulmonary diseases, the required therapeutic agents must be taken for extended periods of time. Moreover, the long-term use of QCT may generate many undesirable side effects such as tubule adenoma, renal failure, and liver cancer.[@b10-ott-11-5397],[@b11-ott-11-5397] In addition, QCT has low water solubility, poor absorption, and rapid metabolism (bioavailability of about 1%--5%),[@b12-ott-11-5397] all of which can produce in vivo results that differ from the powerful in vitro efficacy of QCT. Therefore, the loading of QCT in nanoscale droplets can improve its pharmacokinetic profile, and, with the use of ultrasound-triggered rupturing, enable effective drug delivery to provide anticarcinogenic effects.[@b13-ott-11-5397],[@b14-ott-11-5397]
RGD is a short peptide containing arginine--glycine--aspartic acid and widely exists in the human body. The extracellular matrix and adhesion proteins in blood, including fibrinogen, fibronectin, and collagen, usually contain RGD sequences.[@b15-ott-11-5397] RGD peptide acts as a recognition site for integrins and their ligands, and allows for adhesion between mediated cells, extracellular matrix, and cells. In recent years, targeting angiogenesis has become an important topic in cancer research. It is generally believed that tumor growth, invasion, and metastasis are caused by angiogenesis. If the blood vessels feeding the tumor were insufficient, the tumor would be necrotic or apoptotic.[@b16-ott-11-5397] At the same time, molecular biological studies have shown that integrins exist on the cell surface and play an important role in tumor angiogenesis.[@b17-ott-11-5397] Integrin receptors, especially ανβ3, are highly expressed in some types of tumor cells and vascular endothelial tumor cells, but not in the normal vessels.[@b18-ott-11-5397] It is also indicated that exogenous RGD peptides can competitively inhibit the ligand binding of integrins, thus inhibiting angiogenesis and migration of tumor cells. At the same time, tumors can be target-marked and anticancer drugs can be target-delivered.[@b19-ott-11-5397]--[@b21-ott-11-5397] In the present work, we encapsulated QCT in nanoscale liposome using the lipid carriers. The surface morphology and particle size distribution were characterized. The drug loading (DL%), encapsulation ratio (ER%), and in vitro release of the drug were also studied. In addition, pharmacokinetics and antitumor studies of RGD-QCT liposomes were evaluated.
Materials and methods
=====================
Materials
---------
QCT was purchased by Natural-Standard Biopharma Co., Ltd (Shanghai, People's Republic of China). Distearoyl-L-a-phosphatidylethanolamine (DSPE)-PEG2000-RGD was provided by the HuiJia Biopharma Co., Ltd (Xiamen, People's Republic of China). Soybean phosphatidylcholine, DSPE-PEG2000, and cholesterol were obtained from Sinopharm Chemical Reagent (Shanghai, People's Republic of China). A549 cell line was purchased from Kobai Biomedical Ltd., Co. (Nanjing, People's Republic of China). All other reagents were obtained from Sinopharm Chemical Reagent. Methanol and acetonitrile (chromatographic grade) were obtained from Sigma (Aldrich, St Louis, MO, USA). Deionized water used throughout the research was produced using a Milli-Q System (Millipore Corporation, Burlington, MA, USA).
Animals
-------
Experiments were carried out on rats weighing 220±20 g and C57BL/6 mice weighing 20±2 g. The animals were kept in cages in a room at a temperature of 25°C±2°C, with a 12:12 light--dark cycle. There was a free supply of food and water. All experiments were carried out in strict accordance with the rules of experimental animal care used by Chinese national health institutions.
Preparation of liposomes
------------------------
RGD-modified liposomes containing QCT were prepared by the thin-film hydration method as described previously.[@b22-ott-11-5397] Briefly, QCT, soybean phosphatidylcholine, (DSPE)-PEG2000-RGD, and cholesterol (molar ratio: 2, 55, 5, and 38, respectively) were dissolved in 5 mL of chloroform to form a mixed solution; then the organic solvent was removed under reduced pressure at 40°C by rotary evaporation to form a thin solid film on the inner walls of the round-bottomed flask. This film was then flushed with nitrogen for 30 minutes and stored overnight in a desiccator to remove any traces of chloroform. Glucose and mannitol (1:1, w/w) were dissolved in PBS (pH =7.4). The lipid film was then hydrated with 5 mL of PBS (pH =7.4) at 55°C by rotation (180 rpm ×0.5 hour) to form (DSPE)-PEG2000-RGD-LPs/QCT ([Figure 1A](#f1-ott-11-5397){ref-type="fig"}, appearance morphology). For the preparation of different QCT LPs distearoyl-L-a-phosphati-dylethanolamine-polyethylene glycol 2000-RGD-liposomes (\[DSPE\]-PEG2000-LPs/QCT; \[DSPE\]-PEG2000-RGD-LPs/QCT), a similar procedure was carried out.
Characteristics
---------------
Nanoliposome morphology was observed with an optical microscope. In addition, high-resolution images were obtained of the lyophilized particles using a transmission electron microscope (Philips CM120, Philips, Amsterdam, the Netherlands). In practice, the LPs solution containing 0.1% (w/v) phosphotungstic acid was applied to the copper mesh carbon film and observed by electron microscopy at 80 kV. Particle size distribution and average diameter of (DSPE)-PEG2000-RGD-LPs/QCT were determined by dynamic light scattering using a NICOMP 380 Submicron Particle Sizer (Particle Sizing Systems, Santa Barbara, CA, USA) equipped with a 5 mW heliumneon laser at 632.8 nm. Zeta potential was measured on the same samples prepared for size analysis.
The QCT loading was determined as previously described.[@b23-ott-11-5397] Briefly, QCT was extracted from the LPs with methanol; then, the QCT concentration was determined at 254 nm using high-performance liquid chromatography method with a QCT calibration curve (1.0--10 µg/mL concentration range). The calibration curve was A =0.0654C +0.0031, *r*=0.997. Then, DL% and ER% were calculated according to [Eq (1)](#fd1-ott-11-5397){ref-type="disp-formula"} and [Eq (2)](#fd2-ott-11-5397){ref-type="disp-formula"}: $$\text{DL}\% = \frac{W_{M}}{W_{P} + W_{M}\ } \times 100$$ $$\text{ER}\% = \frac{W_{M}}{W_{F}} \times 100$$where W~P~ is the weight of the initial feeding polymer, W~M~ is the weight of drug incorporated in microspheres, and W~F~ is the weight of the initial feeding drug.
Stability
---------
The liposomes were stored for 6 months at 4°C and 20°C, respectively. The morphology and drug content of liposomes were detected regularly.
In vitro release
----------------
The in vitro drug release from different formulations was determined in PBS (pH=7.4) containing 10% alcohol at the temperature of 37°C±0.5°C. Hundred milligram QCT-related formulations were filled into a dialysis bag (molecular weight cut-off =12,000) immersed fully in 18 mL of the release medium under magnetic stirring at 75 rpm. At the scheduled time points, 1 mL of the release medium was taken out and replaced with the same amount of fresh release medium. The QCT contents were determined by the high-performance liquid chromatography method. All release experiments were carried out as triplicates. All measurements were used to calculate cumulative drug release. The release behavior of QCT in LPs was analyzed with the different release model.
In vivo evaluation
------------------
All the in vivo experimental protocols were approved by the animal care committee of the Second Hospital of Anhui Medical University, and all experiments were conducted in strict accordance with the laboratory animal care and usage guidelines adopted by the National Institutes of Health (Hefei, People's Republic of China). The pharmacokinetics of QCT preparations was observed in 18 rats after intravenous administration. Rats were randomly divided into 3 groups, given a 5 mg/kg intravenous dose of QCT injection, (DSPE)-PEG2000-LPs/QCT, and (DSPE)-PEG2000-RGD-LPs/QCT via the tail vein. Blood samples (0.5 mL) were collected from the orbital cavity into heparinized tubes at 10 and 30 minutes and 1, 2, 4, 8, 12, 24, and 48 hours after intravenous administration. The blood was immediately processed for plasma by centrifugation at 2,000× *g* for 10 minutes. Plasma samples were frozen at −70°C until analysis.
In vivo imaging in tumor-bearing mice
-------------------------------------
The noninvasive optical imaging systems were used to observe the real-time distribution of the functional targeting LPs in A549 cells of xenograft mice. DiR, a lipophilic near-infrared probe, was encapsulated into the LPs as a fluorescence probe at a dose of 0.1 mg/kg. For the preparation of different QCT LPs containing DiR, a similar process was carried out. Tumor-bearing mice models were built according to the previous report. In brief, 1×10^9^ A549 cells were suspended in 200 µL Dulbecco's Modified Eagle's Medium and injected into the right forelimb flank of mice. When tumor volume reached \~100--150 mm^3^, mice were injected with free QCT + DiR, DiR + (DSPE)-PEG2000-LPs/QCT, and DiR + (DSPE)-PEG2000-RGD-LPs/QCT (0.2 mg/kg) via the tail vein (3 each group). After anesthesia, the mice were scanned with the body image system (GelLogic 212 PRO, Carestream Health, Rochester, NY, USA) at 2, 4, 10, and 24 hours, and the systems were equipped with an excitation band-pass filter at 750 nm and an emission at 780 nm. The exposure time of the fluorescence images was 60 seconds.
Antitumor efficacy in tumor-bearing mice
----------------------------------------
C57BL/6 mice were inoculated with A549 cells as described. The treatments began on the day when the tumor volume reached 100--150 mm^3^, which was the eighth day of the whole experiment. On day 8, the mice were randomly divided into four groups (blank LPs, QCT injection, \[DSPE\]-PEG2000-LPs/QCT, and \[DSPE\]-PEG2000-RGD-LPs/QCT) (n=10). A total of 10 mg/kg samples were injected intravenously via the tail vein every 3 days. Four administrations were carried out in total. During the study, the tumor size and body weight were measured 3 times a week. On the 30th day, the mice were sacrificed and the resected tumor weighed. The test drug administration procedure was as follows: weighing and first injection on day 0, 3, 9, and 12 and euthanasia on day 30. The diameter of tumor was measured using a digital caliper. The tumor volume (mm^3^) was calculated with the following formula: tumor volume = length × width[@b2-ott-11-5397] ×0.5. In the course of the study, rats were regularly weighed to monitor potential toxicity.
Results and discussion
======================
Characteristics
---------------
As shown in [Figure 1Ba and Bc](#f1-ott-11-5397){ref-type="fig"}, the surface morphology of (DSPE)-PEG2000-RGD-LPs/QCT was observed by transmission electron microscopy. The LPs were spherical, smooth and uniform in size, and suitable for intravenous administration. Various strategies, such as polymeric nanoparticles, microspheres, liposomes, and solid lipid nanoparticles, were assessed for the sustained delivery of activated agents to the lungs. Peptide-modified target preparation has become a popular drug carrier system in recent years. In this study, water-insoluble reagents, such as QCT, were found to be easily dispersed in lipid solution and entrapped by the thin-film hydration method. The method is simple and reproducible with high entrapment efficiency and ideal DL. The ER% and DL% of prepared LPs were 89.2%±7.4% and 9.2%±1.3% (n=6), respectively, and the mean diameter was 93.4±7.2 nm for 3 batches. The polydispersity index was 0.14±0.02, and the zeta potential was −20.4±0.6 mV ([Table 1](#t1-ott-11-5397){ref-type="table"}). The results of in vitro experiments showed that the particle size of liposomes was suitable for the fenestrated vasculatures of cancer tissues through the enhanced permeability retention effect. In addition, compared with normal liposomes, the size of (DSPE)-PEG2000-RGD-LPs/QCT was not significantly affected by RGD modification. The zeta potentials of (DSPE)-PEG2000-RGD-LPs/QCT were negative due to the addition of DSPE-PEG2000. High ER% and DL% of drugs in LPs would help prevent the rapid leakage during the process of drug delivery and the accumulation of drug in tumor tissues.
Stability
---------
The stability data of (DSPE)-PEG2000-RGD-LPs/QCT showed that after storage at 4°C or room temperature 20°C for 6 months, the surface morphology and other physicochemical properties of QCT had no notable changes ([Table 1](#t1-ott-11-5397){ref-type="table"} and [Figure 1Bb](#f1-ott-11-5397){ref-type="fig"}).
In vitro release
----------------
The release behavior of QCT in different formulations in vitro was studied by the dialysis bag method. The release profiles of free QCT, (DSPE)-PEG2000-LPs/QCT, and (DSPE)-PEG2000-RGD-LPs/QCT were shown in [Figure 2](#f2-ott-11-5397){ref-type="fig"}. We observed that the release of free QCT was rapid, while the cumulative release rates of the other 2 QCT liposomes were much slower, followed by a sustained release. In the free QCT group, 90% of QCT was released in the first 2 hours. In contrast, only 30% or 65% of QCT were released from DSPE-PEG2000-RGD-LPs/QCT within the initial 2 or 48 hours (*P*\<0.01).
The kinetics of in vitro release was analyzed according to the zero-order, first-order, and diffusion controlled release mechanisms. As shown in [Table 2](#t2-ott-11-5397){ref-type="table"}, by analyzing the amount of drug released vs the square root of time, a relatively high correlation coefficient was obtained, indicating that the release followed the Higuchi kinetic model. The results showed that (DSPE)-PEG2000-RGD-LPs/QCT had good sustained release effect.
In vivo evaluation
------------------
Pharmacokinetic studies were carried out in rats using different QCT formulations. The time course of the plasma concentrations of QCT injection, (DSPE)-PEG2000-LPs/QCT, and (DSPE)-PEG2000-RGD-LPs/QCT are summarized in [Figure 3](#f3-ott-11-5397){ref-type="fig"}. [Table 3](#t3-ott-11-5397){ref-type="table"} lists the pharmacokinetic parameters calculated from the plasma drug concentration vs time profiles.
There were significant differences in half-life (*t*~1/2~), area under the curve (AUC~0--~*~t~*), AUC~0--∞~, mean residence time, and plasma clearance between the QCT injection and (DSPE)-PEG2000-RGD-LPs/QCT (*P*\<0.05). As shown, after the single QCT injection, the plasma drug concentration quickly reached the maximum level (4,563.2±398.5 ng/mL) in 10 minutes, before it decreased rapidly, leaving around 10% of the maximum concentration value 2 hours later. This implied that a rapid in vivo elimination of QCT existed in rats. In the case of intravenous administration, the in vivo profile of (DSPE)-PEG2000-LPs/QCT was smoother than the QCT injection group. The *t*~1/2~ and AUC~0--∞~ of (DSPE)-PEG2000-LPs were 2.72-fold and 3.62-fold higher, respectively, compared with free drug. Therefore, it is reasonable to conclude that LPs could significantly extend the role of QCT in vivo (provide higher bioavailability). Meanwhile, (DSPE)-PEG2000-RGD-LPs/QCT provided higher AUC~0--∞~, mean residence time, and *t*~1/2~ when compared with (DSPE)-PEG2000-LPs/QCT. However, (DSPE)-PEG2000-RGD-LPs/QCT showed decreased clearance when compared with the (DSPE)-PEG2000-LPs/QCT. There was no significant difference in pharmacokinetic parameters between the 2 liposomes.
In vivo imaging in tumor-bearing mice
-------------------------------------
The biodistribution of (DSPE)-PEG2000-RGD-LPs/QCT was evaluated by in vivo fluorescence imaging, and [Figure 4](#f4-ott-11-5397){ref-type="fig"} shows the real-time imaging observation after intravenous administration of varying DiR formulations. Results showed that strong fluorescent signals were observed in blood circulation at the tumor site after administration of (DSPE)-PEG2000-RGD-LPs/QCT + DiR, and fluorescent signals remained at the tumor site for 24 hours. In contrast, the fluorescent signal of the (DSPE)-PEG2000-LPs/QCT + DiR group could not be detected at the tumor site at 24 hours. Free DiR was mainly distributed in the liver without any signal in the tumor site, suggesting that the free DiR was not specific to the tumor tissue. According to the experimental results, the high tumor targeting ability of (DSPE)-PEG2000-RGD-LPs/QCT + DiR might be related to the combination of enhanced permeability retention effect and receptor-mediated cell uptake. Although in vitro data provided an important preliminary background to the promotion of the potential use of RGD-modified QCT LPs, these results should be further confirmed and verified in vivo.
Antitumor efficacy in tumor-bearing mice
----------------------------------------
[Figure 5](#f5-ott-11-5397){ref-type="fig"} summarizes body weight changes and tumor volume changes after treatments with varying QCT formulations. [Figure 5A](#f5-ott-11-5397){ref-type="fig"} shows tumor volume changes. (DSPE)-PEG2000-RGD-LPs/QCT-treated mice showed a significantly suppressed tumor growth and there was no recurrence after 30 days of treatment, while free QCT- and (DSPE)-PEG2000-LPs/QCT-treated groups, similar to the control group, showed ever-increasing tumor growth. As can be seen in [Figure 5B](#f5-ott-11-5397){ref-type="fig"}, the body weight of mice in treatment groups did not decline significantly compared with blank LPs groups. [Figure 6](#f6-ott-11-5397){ref-type="fig"} exhibits the hematoxylin and eosin staining assay of hearts, livers, spleens, lungs, and kidneys in tumor-bearing mice. Compared with the control group, there was no obvious abnormality or organ damage after treatments with RGD-modified preparations. The sustained release of QCT from DSPE-PEG2000-LPs revealed its applicability as a drug delivery system that minimizes the exposure of healthy tissues and increases the accumulation of therapeutic drug in tumor sites. In lung cancers, the overexpression of cell surface receptors is often exploited for targeted delivery of therapeutics with ligand-/antibody-modified nano-drug delivery vehicles. The integrin (ανβ3) receptor is of particular interest, since its expression is high in tumor endothelium and tumor cells. Using RGD peptide to target integrin in tumor vascular endothelium is a well-known strategy to suppress angiogenesis and metastasis. The expression of integrins is relatively weak in normal cells. Overall, the results indicated that the RGD-modified QCT LPs could more significantly increase tumor inhibition ability compared with ordinary liposomes.
**Disclosure**
The authors report no conflicts of interest in this work.
{#f1-ott-11-5397}
{#f2-ott-11-5397}
{#f3-ott-11-5397}
{#f4-ott-11-5397}
{#f5-ott-11-5397}
{#f6-ott-11-5397}
######
The characteristics and stabilities data of (DSPE)-PEG2000-RGD-LPs/QCT before and after storage at different temperatures (n=3)
Temperature Time (months) Particle size (nm) Zeta potential (mV) ER (%) DL (%) Polydispersity index
------------- --------------- -------------------- --------------------- ---------- --------- ----------------------
20°C 0 93.4±7.2 −20.4±0.6 89.2±7.4 9.2±1.3 \<0.14
1 93.7±6.8 −21.3±0.8 88.4±8.1 9.0±1.5 \<0.16
3 94.3±7.5 −21.7±0.5 87.7±6.9 9.1±1.2 \<0.18
6 95.2±8.1 −20.8±0.3 87.1±7.8 8.9±1.1 \<0.19
4°C 1 93.9±6.2 −21.5±0.4 88.5±6.5 9.1±1.2 \<0.16
3 94.4±8.1 −21.6±0.1 88.1±6.3 9.2±1.4 \<0.15
6 95.1±6.3 −20.9±0.5 87.5±7.1 8.9±1.2 \<0.18
**Abbreviations:** DL, drug loading; (DSPE)-PEG2000-RGD-LPs/QCT, an RGD-modified nanoliposomes containing quercetin; ER, encapsulation ratio; QCT, quercetin.
######
Dissolution kinetic parameters of QCT from (DSPE)-PEG2000-RGD-LPs/QCT (n=3)
Model Formulations
---------------------- ---------------------------------- --------
Zero-order equation Q =4.344 t--0.422 0.9318
First-order equation ln(1--Q) =6.143t +2.115 0.9224
Higuchi Q =5.941*t*~1/2~--1.953 0.9924
Weibull's equation ln(1/\[1--Q\]) =−3.261lnt +1.491 0.9698
**Abbreviations:** (DSPE)-PEG2000-RGD-LPs/QCT, an RGD-modified nanoliposomes containing quercetin; QCT, quercetin.
######
Pharmacokinetic parameters of QCT after intravenous administration of QCT injection, (DSPE)-PEG2000-LPs/QCT, and (DSPE)-PEG2000-RGD-LPs/QCT to rats (n=6)
Parameter Intravenous administration
--------------------------- ---------------------------- ------------------ -------------------------------------------------------------
*t*~1/2~ (min) 31.4±7.9 82.5±8.6 92.6±7.3[a](#tfn3-ott-11-5397){ref-type="table-fn"}
AUC~0--~*~t~* (ng⋅min/mL) 3,251.4±332.4 11,330.6±1,126.5 12,789.6±1,226.5[a](#tfn3-ott-11-5397){ref-type="table-fn"}
AUC~0--∞~ (ng⋅min/mL) 3,418.6±326.9 12,192.5±1,227.9 13,592.5±1,327.9[a](#tfn3-ott-11-5397){ref-type="table-fn"}
MRT (min) 29.5±3.5 64.7±7.4 72.9±4.6[a](#tfn3-ott-11-5397){ref-type="table-fn"}
CL (L/kg/min) 0.31±0.14 0.12±0.05 0.09±0.03[a](#tfn3-ott-11-5397){ref-type="table-fn"}
**Notes:**
*P*\<0.05: vs QCT injection; AUC~0--~*~t~*, area under the concentration--time curve from time 0 to the final measurable concentration.
**Abbreviations:** AUC~0--∞~, area under the concentration--time curve pushed to infinity; CL, plasma clearance; (DSPE)-PEG2000-RGD-LPs/QCT, an RGD-modified nanoliposomes containing quercetin; MRT, mean residence time; QCT, quercetin; *t*~1/2~, half-life.
[^1]: These authors contributed equally to this work
| {
"pile_set_name": "PubMed Central"
} |
Gravity is a relentless fact of life on Earth. With each change in posture, the vector of all hydrostatic pressure gradients is altered, and regional pressures affected. Due to the elongated shape of the human body, our cardiovascular system is particularly sensitive and when upright gravity reduces cardiac output by remarkable 2 L/min. In response to the gravitational stress, an intricate system of baroreceptors and compensating blood pressure reflexes have evolved. The most important baroreceptors are located just below the brain and thus at an advantageous position to monitor and safeguard blood pressure and flow to the brain.
The cerebral perfusion pressure is defined as the pressure gradient across the brain and calculated by the difference between the arterial blood pressure at brain level and the intracranial pressure (ICP) (Petersen et al. [2016](#phy214039-bib-0009){ref-type="ref"}). To maintain perfusion pressure to the brain, elevations in ICP are counterbalanced by equal elevation in arterial blood pressure (Guild et al. [2018](#phy214039-bib-0003){ref-type="ref"}). The role of ICP is thus well‐recognized in pathology and any physician will cringe at the thought of what elevated ICP will do to cerebral perfusion, especially in the case of head‐trauma where cerebral autoregulation may be temporarily impaired or knocked out. Despite this, the role of ICP in normal everyday regulation of cerebral perfusion is largely unrecognized -- this is in part because the pressure range of ICP has been believed to be very small compared to arterial blood pressure and therefore ignored, and perhaps moreover because of the invasive nature of ICP‐recordings and the limited available data on ICP variability in healthy humans. With this well‐written article, Stok et al. ([2019](#phy214039-bib-0010){ref-type="ref"}) open up this important question of role of intracranial pressures and cerebrospinal fluid movement for cerebral autoregulation.
Cerebral autoregulation is a broad term that indicates the ability of the human brain to maintain appropriate blood flow despite changes in arterial blood pressure. A broad array of intrinsic mechanisms and systemic neural reflexes contribute to cerebral autoregulation. The term "static" cerebral autoregulation refers to the brains ability to maintain relatively constant flow within mean arterial blood pressure range of some 60--150 mmHg (Lassen [1974](#phy214039-bib-0005){ref-type="ref"}). The more recently coined term "dynamic" cerebral autoregulation describes the brain\'s ability to compensate for rapid changes in perfusion pressure. The buffering capacity of the cerebral vascular bed depends on the frequency of the fluctuation in perfusion pressure; high‐frequency fluctuations in arterial blood pressure are translated more directly to cerebral blood flow velocity, while slower changes are better dampened indicating more efficient autoregulation. Furthermore, a rise or fall in arterial partial pressure of CO~2~, which is a potent cerebral vasodialator, can increase or decrease cerebral blood flow independently of the autoregulation (Lennox and Gibbs [1932](#phy214039-bib-0007){ref-type="ref"}). Conversely, low arterial blood pressure alters the CO~2~ reactivity of the cerebral vasculature (Harper and Glass [1965](#phy214039-bib-0004){ref-type="ref"}) and a decrease or increase in CO~2~ partial pressures can improve or attenuate dynamic cerebral autoregulation (Aaslid et al. [1989](#phy214039-bib-0001){ref-type="ref"}). Thus, both direct and indirect effects of CO~2~ on the cerebral vasculature interfere with cerebral autoregulation adding to the complexity of the integrated physiology.
Daily fluctuates in ICP are determined by the sum of the volumes of intracranial arterial and venous pressure and the cerebrospinal fluid (CSF) pressure. The overall brain pressure is, so to say, the resultant balance of these three fluid columns. Within the rigid confinements of the skull, the second to second arterial inflow is, of course, perfectly matched by venous outflow and each pulse wave is furthermore buffered by CSF movements to and from the spinal cavity. The three fluid systems are thus synchronized and interact in a compensatory fashion.
Because of the eccentric placement of our brain at the very top of these fluid columns, the daily postural fluctuations in ICP are quite significant (Petersen et al. [2016](#phy214039-bib-0009){ref-type="ref"}). Traditionally, cerebral autoregulation is assessed by transfer function from systemic blood pressure to changes in cerebral perfusion. If the gravitational vector is kept constant, that is, body position is maintained, changes in blood pressure and ICP/cerebral perfusion pressure are congruent and this assessment holds true. However, during a change in posture, the arterial blood pressure at heart level is affected less than ICP and cerebral perfusion pressure because the hydrostatic indifference point is located close to heart level (Petersen et al. [2014](#phy214039-bib-0008){ref-type="ref"}). As dynamic changes in the gravitational vector affect regional arterial pressure and ICP/cerebral perfusion pressure differently, it is possible that dynamic cerebral autoregulation is over‐ or underestimated. Furthermore, compliance of the brain‐tissue and thus pressure‐wave propagation within the brain is affected by posture, which may in itself also affect cerebral autoregulation. In other words, ICP may affect cerebral autoregulation differently in upright versus supine postures and could be considered an independent modifying factor in dynamic regulation of cerebral blood flow.
Dr. Stok and colleagues (Stok et al. [2019](#phy214039-bib-0010){ref-type="ref"}) illustrate the complex interaction of gravitational fluid‐shifts, regional pressures, and respiration for cerebral autoregulation. Using both static and dynamic manipulation of the gravitational vector by whole‐body head‐up tilt and sinusoidal oscillations, responses in cerebral blood flow velocity and arterial blood pressure are reported. Further attempts to interpret data are done by mathematic modeling of CSF movements to and from the spinal canal. Despite the fact that no final conclusions can be drawn from the results, Stok et al. deserves much credit for the truly integrative approach to unravel the complex physiology behind cerebral autoregulation. The article raises the important question of the role of normal gravitational ICP fluctuations for cerebral autoregulation.
Changes in posture, whether static or dynamic, may be used to manipulate the magnitude and direction of the gravitational vector and the resultant hydrostatic pressure gradients. However, the only way to eliminate hydrostatic gradients altogether is by weightlessness which thus constitutes an important tool for investigating effects of gravitational stress. The indications of impaired intracranial pressure regulation in some astronauts during and following long‐term spaceflight (Lee et al. [2017](#phy214039-bib-0006){ref-type="ref"}) along with indications of possible impaired cerebral autoregulation (Blaber et al. [2011](#phy214039-bib-0002){ref-type="ref"}) begs further investigation of effects of gravity and weightlessness on cerebral perfusion and function.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Public health ethics is seen from the view of communities and populations. The ethical considerations related to forced quarantine *i*mpinges on the rights of individuals to protect the wellbeing of the greater public, in an effort to curb an epidemic, a pandemic or to ensure that neither starts. In these situations, ethical conflicts are bound to occur because public health and the safety of the community often trumps individual rights. This chapter will examine the ethical questions pertaining to public health ethics that arose from the Ebola outbreak of 2014/2015. During the outbreak, it is said that the United States of America went through extraordinary lengths to give two tow of it citizens, experimental drugs, outfit a plane with an incubation pod and monitors; airlift them out of the "red zone" to a special, ward at Emory University hospital which houses the most sophisticated infectious disease unit in the United States. Many people were outraged that the individual rights of these two citizens appeared to be more important than that of the public. The fear of Ebola Virus Disease spreading in the country turned to hate mail hurled at doctors at Emory. (Usborne [@CR23]). The practice of forced quarantine will be a focus of the discussions as the risks of emerging and re-emerging infectious diseases increase with modern travel that transports people, to different continents within a few hours via airplanes, ships and trains that cross borders at exceedingly increasing speeds. With this "speed travel" comes the speed of rapid transport of communicable, infectious diseases, bio-hazards and toxic, biological agents; across borders. In the United States of America, rules and regulations that govern the use of quarantine are in place to safeguard the public against the transmission of infectious diseases to the general public.
This chapter attempts to explain differences between medical and public ethics, using quarantine, as an example. It also explains the differences between quarantine and isolation. There are existing laws, regulations, as well as, ethical considerations that affect health practice and practitioners when a forced quarantine occurs. Examples from the countries around the world are used for discussion purposes in an effort to express a global experience and to outline historical perspective.
International Law {#Sec2}
=================
Historically, international law played a key role in global communicable disease by providing guidelines for the control of diseases as well as playing a major role in the surveillance of global communicable diseases. During the nineteenth century, in lieu of a world global health body which could harmonize the different laws within European nations, international law dominated the field. There were inconsistent regulations on quarantine until the exchange of information and the establishment of international health organizations (Aginam [@CR2]). Communicable diseases shape global health since pandemics and epidemics know no borders. International law has been of importance since it reduces cross-border vulnerability to these diseases.
Quarantine and Isolation {#Sec3}
------------------------
In 1377, the first known record of modern day quarantine was introduced in Dubrovnik, on Croatia. In 1424, the first *lazeretto* was opened in Venice because of the Plague. It was on the island of Santa Maria di Nazareth. In 1467, the system in Venice was adopted by Genoa (Tognotti [@CR22]). The origin of the word "quarantine" is said to be from Venetian Italian and a variant of the term "*quaranta giorni*" which was used as a designated period of 40 days that ships were isolated before crew and passengers could come ashore during the Plague and Black Death (CDC [@CR7]). Quarantine is government enforced and can apply to humans, animals and can be used at borders or within countries. Clearly, this is the historical perspective of quarantine from a global north perspective. Isolation is when a sick person is kept separate from others. This differs from quarantine where people who have been exposed but are not ill, are separated from the general public.
### Patrick Sawyer: Ebola Virus Disease in Lagos[1](#Fn1){ref-type="fn"},[2](#Fn2){ref-type="fn"} {#Sec4}
On July 20th 2017, a Liberian national, by the name of Patrick Sawyer, flew to Lagos, the former administrative and current commercial capital of Nigeria from Monrovia with a brief stopover in Lomé in Togo. Although there are no accurate statistics on the population of Lagos, in 2016, the National Population Commission of Nigeria, stated that the population of Lagos was over 21 million. Lagos became the largest city on the continent of Africa, surpassing Cairo back in 2012 (World Population Review [@CR27]). Patrick Sawyer arrived in Lagos with the intention to proceed to Calabar, Nigeria, for a government conference. Some reports state that he was seeking the healing ministry of one of the Nigerian pastors, in Calabar, because he knew he was sick and possibly had Ebola Virus Disease (EVD). Mr. Sawyer's actions were reported to have been deliberate; as he reportedly evaded health warnings and the health protocols for the spread of infectious disease and lied to the staff at the hospital about not having contact with anyone who had EVD. Patrick Sawyer had nursed his sick sister who was confirmed to have EVD. She later succumbed to the disease. According to Patrick Sawyer's wife, he evaded contact with people at the Monrovia airport. In an interview with Mrs. Sawyer, she stated that her husband travelled to Nigeria because of the better health system. He knew he was ill and wanted to be treated (Mai-Duc [@CR16]). Patrick Sawyer was the index case, (the first known case) of EVD in Nigeria. Professionally, Patrick Sawyer was a lawyer and the National Public Health Officer for Arcelor-Mittal, headquartered in Luxembourg. Patrick Sawyer was a naturalized American citizen with his family residence in the state of Minnesota, although he was consulting for the government of Liberia, at the time (Mai-Duc [@CR16]). Information regarding Mr. Sawyer's education and social status is worthy of note, as it rules out the case of a person who may not know or understand the lethal and highly infectious nature of EVD. Against this backdrop, one can only assume that Patrick Sawyer, as the public health officer knew he was not supposed to travel and therefore should have quarantined himself to avoid exposing others to risk of contracting EVD.
Patrick Sawyer underscored the tension that might arise in balancing self-determination and autonomy of individuals and protecting the greater good of the society at large. Considering the social, health and psychological impact the Sawyer case had in Lagos and Nigeria at large, there is no doubt that the potential benefits for self-quarantine would certainly have outweighed by far, the individual rights of Mr. Sawyer. However, the debate on whether or not one should self-quarantine themselves is more complex than what many can imagine. For example, in the event that a person does not self-quarantine, important questions may arise about the ethical considerations for having that person under mandatory quarantine in an effort to protect the public from the spread of an outbreak? In the event that mandatory quarantine is considered, questions may equally arise regarding infringing and violating their individual rights? At face value, these questions and considerations may sound simple, yet too complex to resolve!
### West Point: Government of Liberia {#Sec5}
On August 20, 2014, the Liberian government, led by President Ellen Johnson-Sirleaf, imposed a 21 day quarantine in an effort to contain the EVD outbreak, on a sprawling area known as West Point. The Liberian government were at a loss as to how to contain the outbreak and the decision to cordon off this community, a week after the government had declared the outbreak a public health emergency, did not go well with the international community. West Point, a lower socio-economic slum of about 800,000 people, came under strict military enforced quarantine after a holding center for suspected Ebola disease victims was ransacked. Seventeen suspected EVD patients escaped; mattresses and infected materials were stolen. The government of Liberia felt the need for barbed wire and wooden checkpoints to go up around West Point. In a Time magazine interview, President Johnson-Sir leaf cited the attack on the holding center as the reason for the quarantine. She further went on to say that the attack "put the entire community at risk, hence the government had to protect them from themselves." (MacDougal [@CR15]). President Johnson-Sirleaf was making reference to the residents of West Point.
West Point residents struggle to eke out a living by trade and barter such as, selling their catch of fish for the day, to people in wealthier neighborhoods. Contrary to international advice, President Johnson Sirleaf, imposed mandatory quarantine enforced by the military. The Liberian government got caught up in the fear and the turmoil of the Ebola epidemic which led to violent clashes. There were many scuffles with the police and West Point residents threw bottles and stones at the authorities, attempting to escape the makeshift checkpoints. Liberian security forces opened fire on the rioting crowd and killed a 15-year-old, Shakie Kamara,[3](#Fn3){ref-type="fn"} wounding two other teenagers in the resulting melee. There was no autopsy performed on Shakie Kamara and the Ministry of Defence forces took possession of his body (MacDougall [@CR15]). The quarantine imposed area was cordoned off for a total of 10 days (Butty [@CR6]). On Friday, August 30th, the quarantine was lifted amongst much jubilation.
### Thomas Eric Duncan: Texas Health Presbyterian Hospital[4](#Fn4){ref-type="fn"},[5](#Fn5){ref-type="fn"} {#Sec6}
Thomas Eric Duncan, a 42 year-old man from Liberia, although feeling ill, flew to Texas in the United States. He went to Texas Presbyterian Hospital. After arriving at the hospital and complaining about symptoms, he was given antibiotics and some other medications and released from the hospital. He had no medical insurance. Thomas Duncan vomited on the sidewalk outside his apartment and was again taken to the same hospital. Upon arrival this time, he was admitted. Tests were run and they came back positive for Ebola Virus Disease.
Thomas Duncan's family and the staff of Texas Presbyterian Hospital faced a 21 day quarantine because they had been exposed to Thomas Eric Duncan who came from Liberia with EVD. There was a series of events that led to the fear that not only was Thomas Duncan discriminated against, but also the African community, many of whom had to be quarantined.
Reports suggest that Thomas Duncan was treated unfairly and unfortunately died on October 8, 2014 however, there is no empirical evidence to prove it. According to the Washington Post newspaper, some Ebola patients were rushed to advanced medical facilities contrary to what happened with Thomas Duncan. Duncan's family stated that he did not receive the expected standard of care owing to his lack of medical insurance, his race and poor background. The family stated that they requested for everything that they thought would save his life and a family member asked if he could donate blood, but this request was not heeded by the hospital staff. The hospital made an out of court settlement with the family on the strength that Duncan should have been admitted the first time he was taken to the hospital. (Moyer [@CR17]). Had Duncan been hospitalized and quarantined appropriately, by staff who were wearing PPE, others would not have been exposed to EVD or infected with EVD and the questions would not have arisen.
The Director of the National Institute of Allergy and Infectious Disease which is part of the National Institutes of Health (NIH), stated that the hospital made a mistake by not admitting Duncan during his first visit. However, he stressed that he did not think that Duncan was treated differently compared to others (referring to allegations of Duncan being black, poor and uninsured (Moyer [@CR17]). In this case, the ethics pertaining to economic support and the delivery of care would need to be questioned (Kominski [@CR12]). According to the director, some of the staff who had cared for Thomas Duncan became infected because some of their skin on their necks and face was exposed. They also did not wear shoe covers which made them vulnerable to the deadly infection of Ebola virus disease (Jaffe [@CR8]). Two nurses, Nina Pham and Amber Vinson, both exposed to Thomas Eric Duncan, tested positive for EVD. The two nurses were just two out of one hundred workers who had contact with Thomas Eric Duncan at Texas Presbyterian Hospital.
### Carnival Cruise Ship[6](#Fn6){ref-type="fn"} {#Sec7}
On October 17, 2014, a Carnival cruise ship with 4000 passengers headed back to the United States after not allowed to dock in Cozumel, Mexico. This happened because a laboratory technician who handled Thomas Eric Duncan's biological samples was on board the ship. Although the laboratory technician showed no symptoms, she was to self-monitor. After 3 days of the cruise, the lab technician and her husband self-quarantined. On arrival at Belize, the United States requested permission for her to board a flight back to the United States, however the Belizean government denied that permission. Other passengers were allowed to disembark. The laboratory technician and her spouse self-quarantined from Thursday until the ship docked in Texas the following Sunday. Nevertheless, the nation of Mexico isolated this ship by not allowing it to dock on their shores. Belize, refused passage for this passenger to board a flight back to the United States. In addition, some students who had been on the ship were kept out of school. This is an example of how people can be stigmatized as a result of suspected contact with EVD (Bever [@CR3]).
Ethical conflicts were inevitable especially in the United States of America. The reason for this is the history and environment of the United States which allows for the expression of individual rights. Quarantine puts healthy people who are not yet exhibiting symptoms of a communicable disease into a very confined and restricted environment. Most people who enter quarantine usually never become sick (Wynia, [@CR26]).
### Nurse Kaci Hickox[7](#Fn7){ref-type="fn"}: Civil Rights Violations, Liberty, Due Process and Systematic Change. {#Sec8}
In the Autumn of 2014, states to include Maine, New York and New Jersey took onstricter standards for returning health care workers than was required by the United States federal government. On October 22, 2014, Governor Christie announced enhanced Ebola preparedness plans for the state of New Jersey. Two days later, Governors Christie (New Jersey) and Cuomo (New York) passed mandatory quarantine for health care workers returning from working with patients with EVD in West Africa. The former U.N Secretary General, Ban-Ki Moon cautioned against these restrictions because healthcare workers were critical to the efforts in curbing EVD. (Sherwood and Jenkins [@CR20]).
Kaci Hickoxarrived at Newark Liberty International Airport, on October 24, 2014, from Sierra Leone where sheworked with Médecins sans frontiers (MSF) during the EVD outbreak. She passed through the new screening for all travellers from West Africa, as she transited to a connecting flight enroute to Maine. Her temperature was taken with a temporal scanner which showed elevated temperature readings however when taken orally her temperature was normal. She had been questioned aggressively for several hours by several people upon arrival at the airport; taken with several police cars in tow, sirens and lights on, and held illegally,in a tent on hospital grounds, against her will (Sherwood and Jenkins [@CR20]). Kaci Hickox was forced into mandatory quarantine in New Jersey and held against her wishes for three dayswithout legal counsel. Two tests for EVD were found to be conclusively negative.
When Kacy was released, she went back to Maine where she was toself-quarantine, as mandated by the state, for a total of 21 days. Kaci Hickoxwent for a bike ride with her boyfriend defying state voluntary quarantine guidelines. The American Civil Liberties Union and lawyers for Kaci Hickox, filed afederal civil rights law suit stating her civil rights had been violated. Shesoughtboth compensatory and punitive damages. In October, 2014, a judge in the state of Maine overturned the state's forced quarantine and reversed a court order however ordered that Kaci Hickox self-monitor, take her temperature daily and get travel approved by state officials. Governor LaPage (Maine) stated that the judge put Kaci Hickox rights above public safety. (Nicks [@CR18]).
Systematic changes concerning quarantine in the state of New Jersey are intended to be a new guideline to be used for other states. These include the concern that forehead scanners can be inaccurate and an oral reading is preferable, right to communication, right to legal counsel prior to hearings, the right to privacy, as long these do not interfere with the needs of the public (Santora [@CR19]).
Quarantine Strategy {#Sec9}
-------------------
A good quarantine strategy would take into consideration all members of the public to include those under quarantine. It would ensure that when a quarantine is imposed,people have access to liberty and due process (Santora [@CR19]), in addition to the basic necessities such as water and food. It would also be critical to ensure the people have access to health care that would curtail the spread of the communicable diseases to others within a population. The strengths of quarantine lie in the power of the state to impose it. However, the flip side of this is that it can be a weakness as it is important to have checks and balances such as a committee made up of multiple heads of institutions saddled with the responsibility of engaging those affected at various levels, rather than one person or entity making unilateral decisions. This would keep the balance of fairness. The United States has a federal law and institutions such as the Center for Disease Control (CDC) which have been authorized to take action in the event of quarantine. Africa now has an organization as the African Union and the CDC partnered to form the new Africa CDC (Africa CDC [@CR1]). The strategy for Africa and quarantine is that it has to embody the African culture and must include stakeholder engagement. Failure to do so will result in a failed quarantine that could cost the public dearly.
### Community Engagement {#Sec10}
It is important to consider community engagement during an outbreak and most especially when quarantine is considered necessary, such as the case of West Point. Unfortunately, most national governments easily engage the military to reinforce quarantine, without considering sensitizing the affected communities on the value and importance of the exercise. Lack of community and stakeholder sensitization can create tensions between the government and the affected community, human rights activities and civil societies who may resist a potentially beneficial intervention. Community engagement should be considered to avoid unnecessary tension. The World Health Organization Country representative for Liberia had previously advised that community consent was necessary prior to the quarantine of West Point. However, this was not sought by the Liberian government, thereby contributing to the civil unrest.
### Other Ethical Considerations {#Sec11}
Professionals working in field of healthcare have an immense responsibility to act professionally whether they work with individual patients or with populations or communities. Professional ethics is part and parcel of the duty of professionals in the domain who are responsible for those who have entrusted the professionals with their health. This is a complex field to navigate most especially because there is no current framework for ethics in public health (Wilson and Mabhala [@CR25]; Kass [@CR9]). There is a risk of the transfer of deadly infectious disease, as was seen with the Severe Acute Respiratory Syndrome (SARS) in 2013 and the 2014 EVD outbreak in West Africa. This risk, also carries with it the possibility of threats of bioterrorism, by nature (Tognotti [@CR22]).
#### Stigma and Shame {#FPar1}
In Africa, another ethical consideration of quarantine is related to the associated stigma of those who are quarantined. For example, survivors of EVD in West Africa were initially ostracized and shunned by their own community members, and in some cases by their own families. Survivors often experienced psychosocial trauma such as feelings of guilt, isolation and shame, as well as stigmatization. Survivors were also frequently attacked, evicted and threatened by community members. Many were excluded from their communities and their properties burned and destroyed because the communities believed they still harboured the deadly virus and could infect others.
#### Right to Life {#FPar2}
Should citizens of more affluent countries be given priority by their countries and saved from the burden of disease by being airlifted out of a grave situation, whilst others are unable to get basic treatment and left to die? (Kass [@CR10]).This individualistic view examines considerations that focus on autonomy while public health ethics supports the rights of the community. Essentially, everyone has a right to life. While the experience of life differs dramatically between different contexts, there is a general appreciation that optimal quality of life should comprize of a healthy life, free from disease, including any form of physical and psychological distress. Against this backdrop, every human being, regardless of their socioeconomic background, ought to have an optimal life experience. Health care provides a means towards restoring the divergent quality of life to its optimal standard. Unfortunately, this was never the case during the EVD outbreak. The outbreak saw numerous cases where individuals from specific nationals received preferential treatment over others. Thus, individuals from well-resourced countries, those who were affluent or who came from wealthy families were more likely to receive the best quality of health care compared to the poor black persons from resource-poor countries where the health systems were already dilapidated and severely constrained. In fact, it is fair to say that the outbreak exposed the pathetic disparities that exist between the well resourced and poor resourced countries and individuals more profound.
#### Culture and Traditions {#FPar3}
Survivors of EVD were stigmatized, in part because cultural beliefs such as the concept of having been bewitched, brought by those who were afflicted, affected survivors and their families(Van Bortel et al. [@CR24]).It is clear that the African perception of a personhood or self, lies within how or where that persona can be found. Since there are no logs or manuscripts chronicling the journey of self, persona and humanness; disease and death, this can only be found in African philosophy and evidenced by the way the African people live and breathe, the essence of African "ness" and how they perceive themselves. Although traditional African societies have been eroded by the Western values, health had a high value and this value was what made other avenues in life possible (Tangwa [@CR21]). Quarantine within the West African context also falls right in the middle of treading within the delicate balance of tradition and cultural, vis-a-vis public health. For example, the way Africans perceive the concept of health and illness, death and the dying and personhood, as a whole differ, significantly from the public health construct, which borrows heavily from the western epistemology. Despite its good intentions, quarantine is likely to create tensions with individual rights as well as general African culture and traditions.
Government Blunders {#Sec12}
===================
The Liberian government made several blunders, the first recorded one concerning quarantine was the case of Patrick Sawyer, outlined above. Sawyer was able to travel to Nigeria when he was supposed to have been on the no-fly list, under observation and in self-quarantine. This mistake which could have been corrected at several pointsprior to Sawyer's leaving Liberia and his arrival in Nigeria. This error, put the people of Lagos metropolis at high risk of EVD, causing the infection of 20 people out of whomeightdied.
In Monrovia, 17 patients escaped from a holding center for suspected EVD patients. After the escape, the Liberian government through the Ministry of Information, claimed the patients were rounded up and taken to the John Fitzgerald Kennedy Hospital; a claim which a doctor at the hospital denied. The quarantine of West Point was also a disaster with the use of force; batons, sticks and teargas as well as bullets, to control the rioting by the unarmed residents. This led to some human rights activist expressing dismay at the unjustified shooting of a teenager in an epidemic situation. This show of force was totally unwarranted (MacDougall [@CR15]).
Conclusion {#Sec13}
==========
Public health ethics is geared towards communities and populations, unlike medical ethics which supports the rights of individuals (Kass, [@CR9]). Ethical considerations related to public health can be seen through two main views. These views are from that of the healthcare worker and the associated professionality of that worker. Ethical considerations in regards to public health and the practice of quarantine primarily centers on the loss of freedom for the individual being quarantined. West Africa is a region that is susceptible to emerging infectious diseases. It is crucial that the region examines its surveillance system, restrictions concerning biological specimens, and has a true model to ensure that fairness and equity prevails. The quarantine of West Point under military guard is an example of how leaders make decisions against international advice, out of fear and disregard for people of lower economic classes who are helpless and do not have a voice. The decision to impose a mandatory, military enforced quarantine at a time of such fear; interrupted livelihoods of people who were already at a serious economic disadvantage some of whom are yet to recover. A lot has happened in the field of Public health since the Ebola outbreak of 1976, in the Central African nation of Zaire (now known as the Democratic Republic of the Congo), in a remote rainforest and a village known as Yambuku, when Peter Piot discovered the Ebola virus and provided epidemiological evidence of the path of the virus. (Boseley [@CR4]). The examples shared in this chapter, show how an outbreak can spill over to neighbouring towns, villages and countries, like the situation with both Patrick Sawyer who flew to Lagos, Nigeria and Thomas Duncan who flew to Texas, United States.
Professionals, working in field of healthcare and public health have an immense responsibility to act professionally whether they work with and treat patients individually or if they are public health professionals dealing with populations of people in communities, regions or at a national level. Professional ethics is part and parcel of the duty of professionals in the domain who are responsible for those who have entrusted the professionals with their health. This is a complex field to navigate most especially because there is no current framework for ethics in public health. (Wilson and Mabhala [@CR25]; Kass [@CR9]). There is a very real risk of the transfer of deadly infectious disease, as was seen with the Severe Acute Respiratory Syndrome (SARS) in 2013 and the 2014 EVD outbreak in West Africa. This risk, also carries with it the possibility of threats that are bioterrorist by nature (Tognotti [@CR22]).
<https://www.premiumtimesng.com/investigationspecial-reports/166560-exclusive-how-liberian-govt-cleared-patrick-sawyer-to-travel-to-nigeria-while-under-observation-for-ebola.html>
<https://www.premiumtimesng.com/news/166660-ebola-why-patrick-sawyer-travelled-to-nigeria-wife.html>
<https://www.nytimes.com/2014/08/30/world/africa/quarantine-for-ebola-lifted-in-liberia-slum.html>
<https://www.thedenverchannel.com/news/u-s-world/dallas-texas-ebola-patient-thomas-duncan-dies-at-texas-health-presbyterian-hospital>
<https://www.washingtonpost.com/news/post-nation/wp/2014/10/08/texas-ebola-patient-has-died-from-ebola/?utm_term=.609115f8a0ce>
<https://www.independent.co.uk/news/world/americas/ebola-cruise-ship-in-utter-panic-as-mexico-and-belize-refuse-to-let-it-dock-9804428.html>
<https://www.nbcnews.com/storyline/ebola-virus-outbreak/kaci-hickox-maine-nurse-quarantined-ebola-scare-sues-new-jersey-n449491>
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Approximately 20% of human breast cancers possess an amplification of human epidermal growth factor receptor 2 (HER2), and this overexpression was previously correlated with aggressive tumor behavior and poor patient outcomes \[[@CR1]\]. HER2 (ErbB2, or HER2/neu) is a member of the HER tyrosine kinase receptor (TKR) family, which includes three other members: the epidermal growth factor receptor (EGFR or HER1), HER3, and HER4. Homo- and hetero-dimerization of ligand-bound HER receptors results in activation of multiple pathways, including the p44/42 mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K)/AKT pathways, which regulate cell proliferation and apoptosis \[[@CR2]--[@CR4]\]. HER2 is the preferred heterodimerization partner of the other HER receptors, although it does not have a known natural ligand. Its activation is mediated by homodimerization, or by ligand-mediated stimulation of another HER receptor through heterodimerization. Targeted therapies such as monoclonal antibodies or tyrosine kinase inhibitors (TKI) have provided a significant improvement in patient outcome by targeting HER2 or its dimerization partners. The monoclonal antibody trastuzumab and the TKI lapatinib have both been approved by the FDA and the EMA in combination with chemotherapy \[[@CR5], [@CR6]\], and both have proven efficacy in the clinical setting \[[@CR7], [@CR8]\]. However, resistance still occurs in patients because of the complexity, robustness and redundancy of the HER biological network \[[@CR9], [@CR10]\].
Obesity has been described as a risk factor not only for cancer development but also for a decreased sensitivity to cancer treatment \[[@CR11]--[@CR14]\]. Epidemiological studies have demonstrated that trastuzumab and lapatinib were less efficient in overweight and obese patients than in other patients in terms of free relapse survival and overall survival \[[@CR10], [@CR15]\]. Indeed, with a 10-year follow-up the overall survival of lean women treated with trastuzumab was 38% higher than for obese women. Moreover, several in vitro studies have demonstrated a decrease of efficacy of targeted therapies on breast tumor cells when cells are cultured in the presence of adipocytes or adipocyte-conditioned medium (CM) \[[@CR16]--[@CR19]\]. The main mechanisms identified for obesity-mediated resistance of cancer cells to anticancer agents have been described in vitro or in mice and include alterations of the intracellular signaling pathway PI3K/AKT/mTOR \[[@CR16]\], inhibition of cell cycle blockade and inhibition of apoptosis \[[@CR20]--[@CR22]\]. In a previous study, our team showed that adipocyte-secreted factors decreased the efficacy of trastuzumab on BT-474 and SKBR-3 breast cancer cell lines \[[@CR16]\]. More recently we have shown that MVP could be involved in adipocyte-induced resistance of breast cancer cells to doxorubicin \[[@CR23]\].
Several mechanisms of resistance have been described specifically for anti-HER2 targeted therapies. These include i) the expression of a truncated form of HER2 called p95HER2 unable to bind trastuzumab \[[@CR24]\] ii) an alteration in ADCC mechanisms \[[@CR25]\] iii) a defect in cell cycle arrest and/or apoptosis \[[@CR26]\] iv) an alteration in phosphorylation of intracellular signaling pathways \[[@CR27]\] v) the action of alternative tyrosine kinase protein activation in case of HER2 blockade \[[@CR17], [@CR28]\] vi) drug efflux through efflux pumps such as P-gp \[[@CR29]\] and vii) an upregulation of estrogen receptors (ER) \[[@CR30]\].
In this study, we analyzed the role of proximal adipose tissue in the resistance of breast cancer cells to small molecule targeted therapies such as the TKI lapatinib. Lapatinib is a small molecule that binds to the intracellular domain of the TKR and inhibits the activation of downstream signalization pathways. Lapatinib shares some of the mechanisms of resistance described for anti-HER2 targeted therapies. However, since it binds to the intracellular domain of HER2, the truncation of HER2 into p95HER2 and alterations in ADCC mechanism should not modify its activity, we did not study these mechanisms as a potential mechanism of resistance and focused our analyses on cell cycle arrest. Indeed, the cytotoxic effect of lapatinib has been described to modify the activation of the PTEN/AKT/mTor pathway and to block the cell cycle in G1 phase \[[@CR31]\]. Several proteins are required for cell cycle progression, such as P27, AKT, cyclin D1 and E2F3. Phosphorylated AKT is involved in cell cycle progression by phosphorylating P27, thereby preventing cell cycle blockade. The non-phosphorylated form of P27 inhibits the action of cyclins, while E2F3 is implicated in cyclin D1 gene transcription.
After demonstrating the reduction of cell cycle blockade induced by lapatinib in the presence of adipocyte-conditioned medium, we reproduced the protective effect of adipocyte-secreted factors on tumor cells against lapatinib, according to studies carried on other compounds \[[@CR16], [@CR20], [@CR32]--[@CR34]\] and investigated the effect of adipocyte-conditioned medium on cell proliferation. In order to investigate whether the protective effect of adipocyte-secreted factors is dependent on HER2 expression, we explored the lapatinib-induced cytotoxic effect on different breast cancer cell lines in the presence or absence of adipocyte-conditioned medium. We reproduced these results in vivo in SCID mice using patient derived normal human adipose tissue. To understand the mechanisms of resistance to lapatinib induced by the proximity of adipocytes and to identify the different agents involved in these mechanisms, we performed different physical and chemical treatments on the adipocyte-conditioned media. In parallel, we investigated alterations occurring on breast tumor cells following the contact with adipocyte-conditioned media and the exposure to lapatinib, both at transcriptional and protein levels. Finally, we used pharmacological agents to modify the metabolism of adipocytes in order to determine the possibility to overcome adipocyte-induced resistance of tumor cells to lapatinib.
Methods {#Sec2}
=======
Cell culture {#Sec3}
------------
The preadipocyte cell line 3T3, the fibroblast cell line NIH3T3 and the tumor cell lines MDA-MB-453, MDA-MB-361, MDA-MB-231, and MCF-7 were cultured in complete DMEM medium (Life technologies), supplemented with 10% fetal calf serum (FCS), 2 mM L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin). The tumor cell linesBT-474 and SKBR3 were cultured in complete RPMI medium with the same supplementation as DMEM. All cells were cultivated at 37 °C in presence of 5% CO~2~.
To induce the differentiation of 3T3 cells, confluent cells were incubated in differentiation medium (DMEM supplemented with 10% FCS plus 50 nM insulin (Sigma, 259,278)) up to 14 days. During the two-week incubation, pre-adipocytes differentiated into adipocytes (80--90% of the cells are differentiated after 14 days) and accumulate lipid droplets in their cytoplasm. The CM from pre-adipocytes (3T3-CM) and adipocytes (\#3T3-CM) were harvested, centrifuged at 300 g for 5 min and either directly used or stored at − 20 °C before use. The control medium for these CM is complete DMEM (medium) for 3T3-CM and differentiation medium (\#medium) for \#3T3-CM.
\#3T3 cells were incubated with 5 μM salbutamol, 18 μM terbutaline, 0,1 μM isoprenaline, 29,6 μM dobutamine, 45 μM propranolol, 45 μM atenolol, 15 μM insulin, 100 μM acipimox and 20 μM etomoxir.
Protein denaturation, exosome isolation and lipid sequestration {#Sec4}
---------------------------------------------------------------
The \#3T3-CM was heated at 96 °C during 1 h in order to denature the proteins.
Exosomes were isolated from the CM by differential centrifugation. In brief, the CM was centrifuged at 3000 g for 30 min to remove cell debris then at 10,000 g for 60 min at 4 °C to separate vesicles from exosomes. The exosomes and the soluble factors were separated by ultracentrifugation at 100,000 g for 90 min. We evaluated the purification using the Nanosight® device that detects and quantifies the exosomes with a laser at 405 nm.
Cytotoxicity MTT assay {#Sec5}
----------------------
BT-474, SKBR3 and MCF-7 cells were seeded in 96 wells plates at 20,000; 8000; 3000, cells per well respectively, in 50 μL media. The 3T3-CM, \#3T3-CM, human adipocyte line hMAD-CM (hMADs-CM) and human fibroblast NIH3T3-CM (NIH3T3-CM) were added to the wells. The next day, drugs were added to the wells at a range of concentrations and cells were incubated for 72 h at 37 °C, 5% CO~2~. MTT was then added in each well and incubated for an additional 4 h. The supernatant was discarded, and a solution composed of isopropanol/H~2~O/HCl (90/9/1, v/v/v) was added. The optical density was determined at 540 nm using Multiskan device.
Cell cycle analysis {#Sec6}
-------------------
BT-474 cells seeded in 6 well plates at 420,000 cells per well in 1.5 mL were incubated with 1.5 mL of either \#3T3-CM or \#medium or either 3T3-CM or medium, reproducing the MTT assays conditions for the indicated times. Each condition was performed in triplicate in at least three separate experiments, with or without lapatinib at 1 μM. Cells were harvested by trypsination and dead cells were numbered in the supernatant. After washing with PBS, the cells were incubated with propidium iodide solution (0.05 μg/mL) in the dark at 4 °C for 30 min. Cell cycle was measured by flow cytometry using a BD LSRII flow cytometer using BD FACSDiva software (BD Biosciences).
Reverse transcription and quantitative PCR {#Sec7}
------------------------------------------
BT-474 cells were exposed to \#3T3-CM or 3T3-CM for 24 h prior to exposure to lapatinib. Cells were harvested, and RNA was extracted using a RNeasy Mini Kit (Qiagene®). To determine the impact of pharmacological agents on \#3T3 cells, these were exposed for 6 h prior to RNA extraction using successively Qiazol® and formaldehyde.
Real-time qPCR was performed on a LightCycler Nano (Roche), detecting SYBR Green I signal transmission. The expression of genes was normalized to the housekeeping gene ribosomal 28S. Relative expression was determined using the ΔCt method.
Patient samples {#Sec8}
---------------
The body mass index (BMI) was calculated as follows: weight (kg)/height^2^ (m^2^). Subcutaneous adipose tissue obtained by lipoaspirate was immediately transported to the registered Cell Therapy Unit (Hospices Civils de Lyon, ETI/16/M/001). Briefly, after centrifugation (1962 g for 3 min), oil (upper phase) and tumescent phase (lower phase) were removed. Then, adipose tissue was digested with collagenase (0.1 U/ml, NB6 collagenase (GMP-grade, Serva Electrophoresis Roche, Indianapolis, USA) at 37 °C for 45 min and under constant shaking. Digestion was stopped by adding Dulbecco's Modified Eagle's Medium (DMEM with glutamax, Gibco (Invitrogen, Carlsbad, USA) containing 10% fetal calf serum (FCS, HyClone, Logan, USA). After centrifugation at 300 g for 5 min, floating adipocytes were discarded and cells from the stromal-vascular fraction (SVF) were pelleted, rinsed with medium, and centrifuged (300 g for 5 min at 20 °C). Cells were counted using 0.4% trypan blue (hospital pharmacy preparation) and viability was evaluated. Adipocytes were resuspended in DMEM for 24 h and supernatants were extracted and frozen at − 80 °C before use. Patient provided written informed consent.
In vivo studies {#Sec9}
---------------
All animal procedures were performed in accordance with the European Union directive 86/609/EEC. Experiments were performed under individual permit and in animal care facilities accredited by the French Ministry of Agriculture. The study was approved by the local animal ethics committee (University Claude Bernard Lyon I, protocol number DR-2014-64). The study was conducted using severe combined immunodeficiency (SCID) mice, with five mice per group, purchased from Charles River. Mice bearing xenografts of normal human adipose tissue were obtained by the subcutaneous injection of 1 mL of abdominal adipose tissue obtained from patients undergoing plastic surgery who had provided informed consent. After 1 week, BT-474 tumors were grafted subcutaneously in contact with the adipose xenograft. Treatments were initiated when the tumor volume was 150 mm^3^, with daily orally administration of lapatinib 135 mg/kg for 2 weeks. Tumor growth was directly measured using a caliper. The volume was calculated using the following formula V = 4/3 x π x R^3^ where R is the radius measured. Each mouse was euthanized after experimentation with carbon dioxide.
Statistical analysis {#Sec10}
--------------------
Means ± SD or representative experiments are shown when experiments were repeated several times (*n* ≥ 3). Statistical significance was evaluated using paired Student's *t*-test on the means of at least three independent in vitro experiments. Unpaired Student's *t*-test was used for in vivo experiment. IC50 were calculated using Compusyn software. *p*-values below 0.05 (\*) or 0.01 (\*\*) were considered significant while "ns" stands for not significant data.
Materials {#Sec11}
---------
Human breast cancer cell lines BT-474, SKBR3, MDA-MB-453, MDA-MB-361, MDA-MB-231, MCF-7 and the fibroblast cell line NIH3T3 were all obtained from ATCC. The 3 T3-F442A (named 3 T3 cells here) murine preadipocyte cell line was kindly provided by Dr. Catherine Muller (Toulouse, France). The human multipotent adipose-derived stem cell (hMADS) were provided by Dr. Christian Dani, *UMR 6543 CNRS,* Nice, France and cultured as described previously \[[@CR16]\].
Lapatinib was purchased from Sigma Aldrich while phenylephrine, clonidine, epinephrine, dobutamine, yohimbine, propranolol and atenolol and ibrutinib were purchased from BioScience. Acipimox and etomoxir were obtained from Adooq Bioscience and terbutaline, prazosin, salbutamol, afatinib and AZD4547 were purchased from Selleckchem.
The primers used for PCR were: AKT forward primer 5′-tctggcttcatcggcagt-3′, AKT reverse primer 5′-gatcgcactccctgtctagg-3′, cycline D1 forward primer 5′-tacaaccaggcagcggata-3′, cycline D1 reverse primer 5′--agccacccagaattagacacc-3′, P27 forward primer 5′-ccctagagggcaagtacgagt-3′, P27 reverse primer 5′-agtagaactcgggcaagctg-3′, E2F3 forward primer 5′-acgaagtccagatagtccaaaaa-3′, E2F3 reverse primer 5′-ataccccatcgggtgactg-3′, FABP4 forward primer 5′-ggatggaaagtcgaccacaa-3′, FABP4 reverse primer 5′-tggaagtcacgcctttcata-3′, LPL forward primer 5′-tttgtgaaatgccatgacaag-3′, LPL reverse primer 5′-cagatgctttcttctcttgtttgt-3′, HIF1α forward primer 5′-catgatggctccctttttca-3′ and HIF1α reverse primer 5′-gtcacctggttgctgcaata-3′.
Results {#Sec12}
=======
Adipocyte-conditioned medium reduces lapatinib-induced cell cycle blockade in tumor cells {#Sec13}
-----------------------------------------------------------------------------------------
To assess lapatinib-induced cell cycle blockade, we stained the SKBR3 cells with propidium iodide and performed flow cytometry analyses of cells cultured in control medium or in \#3T3-CM in the presence or absence of lapatinib. Figure [1](#Fig1){ref-type="fig"}a shows that the percentage of cells in G0/G1 phase was increased by 23.4% after exposure to lapatinib when SKBR3 were in control medium. The increase was lower for cells incubated in \#3T3-CM (13.2%). The percentage of cells in S phase decreased from 14.3 to 5.8% when cells were incubated in control medium after lapatinib exposure whereas it decreased from 17.4 to 14.7% when incubated in \#3T3-CM. The proportions of cells in G2/M phase followed the trend with a lower lapatinib-induced decrease for the tumor cells exposed to \#3T3-CM than in control medium. Fig. 1Conditioned medium from adipocytes reduces the lapatinib-induced cell cycle blockade in tumor cells. **A)** Lapatinib-induced cell cycle blockade was investigated on SKBR-3 cells incubated in control medium (a) or in adipocyte-conditioned medium (\#3T3-CM) (b). Cells were exposed for 24 h to lapatinib before staining by propidium iodide and FACS analyses were performed to evaluate the percentage of cells in the different cell cycle phases. *n* ≥ 3. **B)** The expression of genes involved in the cell cycle progression was measured on tumor cells after exposure to lapatinib in presence (+) or not (−) of \#3T3-CM. Cells were exposed for 12 h to lapatinib before RNA extraction and mRNA measurement by SYBR Green device. Fold changes were normalized for the cells exposed to lapatinib on the cells in the same condition not exposed to lapatinib. n ≥ 3
The results obtained when cells were incubated in control medium are consistent with the classical effect of lapatinib. However, this effect was weaker when SKBR3 cells were incubated in \#3T3-CM, suggesting that lapatinib is less effective under these conditions. Interestingly, \#3T3-CM had no effect on the proportion of cells in S phase as described in other studies \[[@CR20]\].
After extraction and purification of cellular RNA, we performed quantitative RT-PCR assays to evaluate the transcription of genes coding for P27, cyclin D1, AKT and E2F3 (Fig. [1](#Fig1){ref-type="fig"}b). We normalized the fold changes obtained from three experiments for each condition with lapatinib on the cells in the same condition without lapatinib in order to highlight the transcription changes induced by the drug. As expected, in control medium, the exposure to lapatinib increased the expression levels of P27 and decreased those of AKT, cyclin D1 and E2F3 in tumor cells while the CM had no effect on its own (data not shown). For the cells exposed to \#3T3-CM, P27 gene expression levels did not vary after exposure to lapatinib and those of cyclin D1 and E2F3 were higher than in cells cultured in control medium. The levels of AKT mRNA remained similar within all conditions.
Adipocyte-conditioned medium induces a resistance in HER2+ breast cancer cells to several tyrosine kinase inhibitors {#Sec14}
--------------------------------------------------------------------------------------------------------------------
To evaluate and quantify the role of adipocytes in the sensitivity of tumor cells to lapatinib, we performed MTT assays and evaluated the drug-mediated cytotoxicity on several breast cancer cell lines with different levels of expression of HER2 in the presence of \#3T3-CM or control medium.
The sensitivity of each cell line to lapatinib was quantified by calculating the mitochondrial metabolic activity that we correlated with the percentage of viable cells after exposure to lapatinib at 1 μM on each cell line (Fig. [2](#Fig2){ref-type="fig"}a). Lapatinib had a cytotoxic effect on all the HER2+ (BT-474, SKBR3, MDA-MB-453 and MDA-MB-361) cell lines but not on HER2- cells (MDA-MB-231 and MCF-7). However, this effect was significantly reduced when the HER2+ cells were cultured in \#3T3-CM. These results show that \#3T3-CM can induce resistance to lapatinib-induced cytotoxicity in HER2+ breast cancer cells. Moreover, the protective effect was found in both BT-474 cells which are HER2+/ER+/PR+ and SKBR3 cells, which are HER2+/ER−/PR-, suggesting that the effect is independent from estrogen and progesterone levels. Fig. 2Adipocyte-conditioned medium induces a tumor cell resistance to various tyrosine kinase inhibitors in Her2 positive cell lines. **A)** BT-474, SKBR3, MDA-MB-453, MDA-MB-361, MDA-MB-231 and MCF-7 tumor cell lines were incubated 24 h in adipocyte-conditioned medium (\#3T3-CM) or in control medium then we added 0,1 μM of lapatinib and we measured the mitochondrial metabolic activity by MTT that we correlated with the percentage of viable cells. *n* ≥ 3. *P* values were calculated by comparing for each cell line the percentage of viable cells in presence of \#3T3-CM to the percentage of viable cells in control medium after exposure to lapatinib. **B)** Under the same conditions of incubation as in **A**) BT-474 cells were exposed to tyrosine kinase inhibitors lapatinib, ibrutinib, afatinib and AZD4547. n ≥ 3. The IC50 values for each therapeutic agent were measured and we calculated the ratio and evaluated the *p* values of the value in presence of \#3T3-CM to the control media condition. \**p* \< 0,05 \*\**p* \< 0,01
Lapatinib is not completely HER2-specific since it also binds to EGFR. In order to investigate whether \#3T3-CM-induced resistance of tumor cells to lapatinib was specific of one or another TKR blockade, we performed MTT experiments to determine the half-maximal concentration (IC50) of other TKIs on BT474 cells. Ibrutinib is a TKI directed against Bruton tyrosine kinase but also against HER2, EGFR and HER3 \[[@CR35]\]. Afatinib blocks HER2, EGFR and HER4 \[[@CR36]\] and AZD4547 is an anti-fibroblast growth factor receptor (FGFR) \[[@CR37]\]. Ratios were calculated using IC50 values obtained were cells were exposed to \#3T3-CM over the IC50 values of cells in control medium for each drug and for each experiment (Fig. [2](#Fig2){ref-type="fig"}b). Quality controls for cytotoxicity assays included OD values greater than 0.2. All assays were performed with technical triplicates in at least three separate experiments.
The results revealed that the IC50 values were increased approximately two- to seven-fold for lapatinib when tumor cells were in contact of \#3T3-CM in comparison to control medium. IC50s of ibrutinib were three and four-fold higher in \#3T3-CM than in control medium. The exposure to \#3T3-CM increased the IC50 of afatinib four-fold compared to the IC50 obtained in control medium. Finally, the IC50 for AZD4547 increased approximately five-fold for the BT-474 cells incubated in the \#3T3-CM. Thus, the protection induced by exposure to adipocyte-CM was observed for every TKI tested.
Exposure to adipocyte-derived factors is required during exposure to lapatinib to induce resistance {#Sec15}
---------------------------------------------------------------------------------------------------
In order to investigate whether the \#3T3-CM-induced resistance of tumor cells to lapatinib required the presence of one (or several) soluble agent(s) during the exposure to lapatinib, we changed the conditions of \#3T3-CM exposure (Fig. [3](#Fig3){ref-type="fig"}a). Cells were cultured (\#3T3-CM or control medium) for 24 h and the medium was removed and replaced by new medium (\#3T3-CM or control medium) shortly before the exposure to lapatinib. On the one hand, it appears that the BT-474 cells exposed for 24 h to \#3T3-CM then placed in control DMEM medium prior to lapatinib exposure were as sensitive to lapatinib as the cells that had not been exposed to \#3T3-CM. On the other hand, we observed a reduced sensitivity of tumor cells to lapatinib-induced cytotoxicity when these were first placed in control medium then secondarily exposed to the \#3T3-CM. These results suggest that exposure to adipocyte-derived factors is required concomitantly to exposure to lapatinib to induce resistance to this agent. Fig. 3Adipocyte-conditioned medium-induced resistance of tumor cells to lapatinib is specific to adipocyte cell lines and the constant presence of adipocyte-conditioned medium is necessary to observe the resistance. **A)** BT-474 cells were seeded on the day 1 in control medium or adipocyte-conditioned medium (\#3T3-CM) and the medium were changed on day 2 prior exposure to lapatinib and evaluation of IC50. n ≥ 3. **B, C** and **D** BT-474 cells were cultured in \#3T3CM, in human adipocyte-conditioned medium (\#hmads), in fibroblast-conditioned medium (NIH-3T3-CM), in preadipocyte-conditioned medium (3T3F442A-CM) or in conditioned medium from fresh human adipocytes from three donors (ASC18032, ASC15011 and ASC18019). We evaluated the IC50 of lapatinib in each condition. n ≥ 3. The ratio of IC50 for cells in control medium to the IC50 for cells in the various conditions. \* *p* \< 0,05 \*\* *p* \< 0,01
The protective effect is specific of adipocyte-conditioned medium {#Sec16}
-----------------------------------------------------------------
To confirm that the protective effect of adipocyte-conditioned medium could be observed with another adipocyte cell line we used the hMAD human adipocyte cell line. Using these cells, we also observed resistance of tumor cells to lapatinib in presence of hMAD-CM (Fig. [3](#Fig3){ref-type="fig"}b). Conversely in the presence of a non-adipocyte fibroblast-CM no resistance was observed, showing that the conditioned-medium from adipocytes specifically inhibits the activity of lapatinib on breast tumor cells (Fig. [3](#Fig3){ref-type="fig"}c). Additionally, \#3T3-CM did not stimulate cell proliferation (supplementary Figure [1](#MOESM1){ref-type="media"}). We also incubated the tumor cells in fresh CM of human adipocytes obtained from three lean healthy patients (Fig. [3](#Fig3){ref-type="fig"}d). There was a trend towards higher IC50 values of lapatinib on cells exposed to these fresh adipocyte conditioned supernatants although the difference was only significant in patient ASC18011.
Proximity with adipose tissue protects HER2+ cancer cells from lapatinib in vivo {#Sec17}
--------------------------------------------------------------------------------
To verify whether adipose tissue could exert a protective effect in vivo, we used a new and original adipose tissue xenograft model in SCID mice, which we recently reported \[[@CR16]\]. Subcutaneous adipose tissue obtained from patients undergoing plastic surgery was injected subcutaneously into SCID mice to form a stable xenograft. BT-474 tumor fragments were secondarily grafted in contact with the adipose xenograft, to reproduce the contact between tumor and adipose tissue (Fig. [4](#Fig4){ref-type="fig"}). Fig. 4Proximity of adipose tissue protects tumors from the lapatinib-induced cytotoxicity in vivo. Tumor growth of BT-474 breast tumors in mice treated with 135 mg/kg of lapatinib daily were measured in four groups of five SCID mice. The animals of two groups were implanted with human adipose tissue (adip) 1 week prior to tumor xenografts. Animals were either treated with lapatinib (lapa) or not treated (nt)
Our results showed that lapatinib treatment significantly reduced BT-474 tumor growth in mice (lapa group) in comparison to the untreated group (nt group) (*p* = 0.005). Indeed, the tumor volume was 1700 ± 112 mm^3^ for the mice in the nt group and 600 ± 95 mm^3^ for those in the lapa group after 19 days, demonstrating the efficacy of lapatinib on the reduction of breast tumor growth. Conversely, lapatinib had no significant effect on tumors implanted in contact with adipose tissue xenografts (adip lapa group vs adip group). However, the tumor growth seemed to be impaired when the tumor was in contact with adipose tissue, a phenomenon that we have also observed with other models (data not shown). These results suggest that tumors in direct contact with adipose tissue are not sensitive to lapatinib.
Soluble products issued from adipocyte lipolysis are likely to be responsible for the reduced antitumor activity of lapatinib on breast tumor cells exposed to adipocyte-conditioned medium {#Sec18}
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Due to their metabolism and endocrine functions, adipocytes produce a large variety of molecules including proteins and lipolytic products and release them as soluble molecules or contained in microparticles such as exosomes. In an attempt to identify the type of molecules responsible for the resistance effect induced to tumor cells, we performed different physical and chemical treatments on \#3T3-CM. We heated the \#3T3-CM at 95 °C in order to degrade the proteins, including adipokines and we added lipid-free BSA to sequester the lipids. We also isolated the exosomes by differential centrifugation and ultracentrifugation. Results shown in Fig. [5](#Fig5){ref-type="fig"}a suggest that the resistance of breast cancer cells to the effect of lapatinib depends on a thermolabile factor but not on free fatty acids. Indeed after \#3T3-CM heating, we no longer observed the increase of lapatinib's IC50 in breast tumor cells exposed to \#3T3-CM compared to the same cells exposed to control medium, suggesting that thermolabile factor(s) had been degraded. Conversely, when we attempted to sequester the lipids by adding BSA and filtered the \#3T3-CM, the protective effect of CM was maintained. After ultracentrifugation, we found that the \#3T3-CM containing only the exosomes did not induced resistance to lapatinib-mediated while the \#3T3-CM containing the soluble factors induced resistance to lapatinib (Fig. [5](#Fig5){ref-type="fig"}b). Fig. 5Thermolabile soluble factor(s) are responsible for adipocyte-conditioned medium-induced resistance of breast cancer cells to lapatinib. **A)** adipocyte-conditioned medium (\#3T3-CM) were heated at 95 °C in order to denature the protein and delipidated BSA was added to the CM prior to exposure of tumor cells and to lapatinib. n ≥ 3. **B)** \#3T3-CM were centrifuged to separate the microvesicles then ultracentrifuged to precipitate the exosomes and to isolate soluble factors. n ≥ 3. *P* values were calculated by comparing the conditions to the control medium. \* *p* \< 0,05
As the secretome of adipocytes is very complex, we also attempted to pharmacologically modulate the metabolism of adipocytes in order to modify the adipocyte secretome of factors released from metabolic reactions. At first, as the metabolism of adipocytes is highly dependent on adenosine AMP, ADP and ATP \[[@CR38], [@CR39]\] we incubated the adipocytes with one or the other of these metabolites or with other molecule affecting lipolysis such as etomoxir that is a lipolysis inhibitor (Fig. [6](#Fig6){ref-type="fig"}a). We observed that the \#3T3-CM from adipocytes exposed to agents such as ADP, AMP and adenosine maintained their ability to reduce the sensitivity of cancer cells to lapatinib. However, in the case of the CM from adipocytes exposed to etomoxir, we found a significantly higher reduction of viable tumor cells after exposure to lapatinib suggesting that the exposure to etomoxir could modify the production or release of soluble agent(s) responsible for the resistance phenotype provided by the CM. Fig. 6Inhibition of lipolysis in adipocytes reverses adipocyte-conditioned medium-induced resistance of tumor cells to lapatinib. **A)** Adipocytes were exposed to different metabolic modulators. After 6 h, the supernatants were harvested and added to BT-474 cells prior to exposure to lapatinib. BT-474 were also exposed to control medium or adipocyte-conditioned medium (\#3T3-CM). n ≥ 3. **B)** adipocytes were exposed to lipolysis stimulators or inhibitors. The supernatants were harvested and used as in **A**) and IC50 values were calculated for each condition. n ≥ 3. **C)** mRNA was extracted from adipocytes after exposure to terbutaline, salbutamol, dobutamine, isoprenaline, atenolol and propranolol and the expressions of the genes involved in lipolysis stimulation were evaluated and normalized on the expression of the genes in adipocytes cultured without treatment. n ≥ 3. \*\* *p* \< 0,01
As etomoxir modulates lipolysis by inhibiting the transport of fatty acyl chain from the cytosol into the mitochondria, we tried to verify our hypothesis that a soluble agent from lipolysis was responsible for the observed resistance by modulating the lipolysis via other pathways. We stimulated or blocked the lipolysis via β-adrenergic receptors. Indeed, the stimulation of a β-adrenergic receptor in adipocytes results in the increase of HIF1α transcription factor \[[@CR40]\] and leads to the activation of lipolysis. We also investigated whether the effect of another lipolysis inhibitor, acipimox, an anti-lipolytic nicotinic acid and insulin derivative that is known to block lipolysis and to stimulate lipogenesis after food intake. As shown in Fig. [6](#Fig6){ref-type="fig"}b, we found that the \#3T3-CM from adipocytes exposed to the antagonists of beta 1 and 2 adrenergic receptors called propranolol and atenolol did not induce resistance of BT-474 cells to lapatinib. This could be explained by the fact that these beta blockers are only effective when there this a beta-agonist present. Similar observations were made with other lipolysis inhibitors such as insulin and acipimox. Conversely, the exposure of adipocytes to pure or mixed beta 1 and beta 2 adrenergic receptor agonists such as salbutamol, terbutaline, isoprenaline and dobutamine produced a \#3T3-CM that induced a resistance of tumor cells to lapatinib-mediated cytotoxicity at least as strong as \#3T3-CM from adipocytes alone. To confirm that these agonists and antagonists were effectively modulating the receptors and altering lipolytic activities in adipocytes, we performed RT-PCR analysis of HIF1α as well as LPL and FABP4, the latter being involved in lipolysis (Fig. [6](#Fig6){ref-type="fig"}c). The results revealed a two to eight-fold increase in HIF1α expression in adipocytes exposed to the agonists compared to the unexposed adipocytes. LPL and FABP4 expression were also higher, particularly in adipocytes exposed to salbutamol. Conversely, the exposure to beta-blockers leads to a drop of the levels of mRNA for these three genes. These data are consistent with the hypothesis that adipose tissue induces a resistance to lapatinib of tumor cells via the release of lipolytic products in the microenvironment.
Discussion {#Sec19}
==========
Adipose tissue and breast tumors are in close contact and adipocytes actively participate in tumor progression through the secretion of various adipokines \[[@CR41]\]. Previous studies have shown that adipocyte-secreted factors could promote tumor cell resistance to various chemotherapeutic agents as well as antibodies. A similar effect has been demonstrated for multiple myeloma cells in response to melphalan, bortezomib, dexamethasone and doxorubicin \[[@CR20], [@CR32]\] and for breast cancer cells in response to trastuzumab, T-DM1 and gemcitabine \[[@CR16], [@CR33]\]. It has been found by Brady et al. that blocking mTOR could reverse lapatinib resistance of cancer cells \[[@CR42]\]. Moreover, another study demonstrated that the activation of mTOR could elude HER2 blockade \[[@CR43]\]. To our knowledge, no study has investigated the role of adipocyte-CM in breast cancer cell resistance to lapatinib or other TKIs.
Our results show that adipocyte-CM reduces the sensitivity of HER2+ breast tumor cells to the cytotoxic activity of lapatinib as well as other TKIs. These results have been confirmed in two different HER2+ cell lines and using conditioned medium from two different types of adipocytic cell lines. They are also in accordance with other studies using a CM from adipocytes demonstrating an adipocyte-induced resistance of tumor cells towards therapeutic agents. In our results, the cells exposed to \#3T3-CM did not proliferate differently from cells in control medium. This suggests that the protective effect is not attributable to a reduced growth rate of cells due to adipocyte-derived factors. For the first time we also show that proximity to adipose tissue in vivo reduces sensitivity to lapatinib of established tumors. This confirms the relevance of adipocyte-derived factors in the in vivo setting.
Our results show a reduced lapatinib-induced cell cycle blockade when tumor cells are exposed to adipocyte-derived factors. This was observed both by direct analysis of the cell cycle and the expression of genes coding for the proteins involved in cell cycle progression. This type of resistance has already been described in the literature for chemotherapeutic agents \[[@CR20]--[@CR22]\]. Additional experiments could be performed to determine lapatinib-mediated reduction of expression of phosphorylated proteins in tumor cells such as HER2, EGFR, AKT, ERK1/2 in the presence or absence of adipocyte conditioned medium. Such results would highlight the role of downstream signaling pathways in the adipocyte secreted factor-mediated resistance of breast cancer cells to lapatinib.
Our results obtained using \#3T3-CM fractionation and pharmacological modulation suggest that lipolysis pathway products are involved in the protective effect towards lapatinib. Modulation of lipolysis in differentiated adipocytes by etomoxir and acipimox, which act at two different levels of the metabolic reaction, were found to reduce \#3T3-CM mediated resistance of breast tumor cells to lapatinib \[[@CR44], [@CR45]\]. An opposite effect was observed with beta-adrenergic receptor agonists. Additional metabolomic approaches are required to determine the soluble factor(s) involved in the resistance of tumor cells to lapatinib by measuring the metabolites in the \#3T3-CM from adipocytes after exposure to the modulators of lipolysis. Such analyses have already been performed in supernatants from adipocytes \[[@CR46]\].
Several mechanisms regarding resistance of tumor cells to lapatinib have been described in the literature \[[@CR47]\]. An alternative hypothesis is that lapatinib could be sequestered or inactivated by adipocyte-derived soluble factor(s). Another mechanism of resistance could be the activation of compensatory pathway such as the increase activity of hepatocyte growth factor receptor (HGFR) or c-Met. The ligand of HGFR is the hepatocyte growth factor which is secreted by adipocytes. Other adipocyte-secreted factors have been described to promote tumor cell resistance to TKI, including the tissue inhibitor metalloproteinase-1 (TIMP-1), interleukin-6 (IL-6), the fibroblast growth factor (FGF) and neuregulin-1 (NRG-1) \[[@CR48]--[@CR52]\]. All these factors could play a role in the adipocyte-secreted factor mediated resistance of cancer cells to lapatinib observed in this study and it would be useful to quantify them or to evaluate their activity.
A possible limitation of our study includes the fact that we have used a murine adipocytic cell line and human breast cancer lines. Additional studies using a human pre-adipocytic line would be of interest. Another caveat is that we have explored the effect of proximal, rather than distal, adipose tissue on sensitivity to therapy. The use of obese mice models would constitute a means to address this issue. Finally, we have analyzed a limited number of kinase inhibitors in our models. A more extensive screening of kinase inhibitors with different affinities for various enzymes would be informative.
Conclusions {#Sec20}
===========
The investigation of lipid metabolic reprogramming in the tumor microenvironment suggests a major role of cancer associated adipocytes \[[@CR53], [@CR54]\]. By using in vitro and in vivo approaches, we demonstrated that peritumoral adipocytes are likely to be involved in the resistance of breast tumor cells to lapatinib-mediated cytotoxicity. The effects of adipose cells are mediated via secreted factors that appear to affect HER2 associated downstream signaling pathways. This phenomenon is due to thermolabile soluble factors and is reversed when adipocytes are exposed to inhibitors of lipolysis. Further studies are necessary to identify the precise mechanism of this resistance and to allow the development of sensitizing strategies to circumvent adipose-mediated resistance to lapatinib in breast cancer.
Supplementary information
=========================
{#Sec21}
**Additional file 1.**
ADCC
: Antibody-dependant cell-mediated cytotoxicity
AKT
: Threonine kinase
CM
: Conditionned medium
EGFR
: Epidermal growth factor receptor
ER
: Estrogen receptor
ERK1
: Extracellular signal-regulated kinase-1
FABP4
: Fatty acid binding protein 4
FGFR
: Fibroblast growth factor receptor
HER
: Human epidermal growth factor receptor
HER2
: Human epidermal growth factor receptor-2
HER3
: Human epidermal growth factor recpetor-3
HER4
: Human epidermal growth factor receptor-4
HIF1α
: Hypoxia Inducible Factor 1 Subunit Alpha
hMADS
: human multipotent adipose derived stem cell
IL-6
: Interleukin-6
LPL
: Lipoprotein lipase
MAPK
: Mitogen-activated protein kinase
mTOR
: Mechanistic target of rampamycin kinase
MTT
: 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide
NRG-1
: Neuregulin-1
P-gp
: Permeability-glycoprotein
PI
: Inorganic phosphate
PI3K
: Phosphatidylinositol 3-kinase
PTEN
: Phosphatase and tensin homolog
TIMP-1
: Tissue inhibitor metalloproteinase-1
TKI
: Tyrosine kinase inhibitor
TKR
: Tyrosine kinase receptor
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
=========================
**Supplementary information** accompanies this paper at 10.1186/s40360-020-00436-z.
Not applicable.
AG and CD designed and performed experiments, interpreted results and wrote the manuscript. MND, LM, SB, AC, ML and ELM designed and performed experiments. LC and LPJ interpreted results and wrote the manuscript. KC designed experiments and interpreted results. All authors have read and approved the manuscript.
This work was funded in part by the French National Cancer Institute (INCA, public funding organization) grants INCa_INSERM_DGOS_12563 and PLBIO 2013--106. Funding bodies provided financial support for reagents.
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
All animal procedures were performed in accordance with the European Union directive 86/609/EEC. Experiments were performed under individual permit and in animal care facilities accredited by the French Ministry of Agriculture. The study was approved by the local animal ethics committee (Université Claude Bernard Lyon I, protocol number DR-2014-64).
Patients provided written consent.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
Colorectal cancer is one of the most common malignant tumors in the world. In Korea, the incidence of colorectal cancer has been increasing since 2000, as the lifestyle has become westernized. Accordingly, the number of colorectal cancer surgeries has also increased precipitously \[[@B1]\].
Postoperative ileus is one of the most common complications after abdominal surgery \[[@B2]\]. Postoperative ileus increases abdominal discomfort, pain, the number of nosocomial infections and the length of hospital stay and, as a result, can lead to an overall increase in the medical costs. The incidence of postoperative ileus in laparoscopic surgery is lower than that in conventional open surgery. However, the surgical extent and dissection area are wide in laparoscopic colorectal cancer surgery, which could be a risk for postoperative ileus \[[@B3]\]. Thus far, various treatments have been attempted to prevent and shorten the duration of ileus. Choi and O\'Connell \[[@B4]\] reported that it was safe to start early feeding as soon as possible after conventional open colorectal surgery and that the early-feeding group had a shorter hospital stay than the late-feeding group. Stewart et al. \[[@B5]\] also reported that early feeding after surgery could decrease postoperative ileus and shorten hospital stay. However, approximately 20% of the patients showed difficulties with or did not tolerate early feeding \[[@B5]\]. Thus, as an alternative to early feeding, \"sham feeding\" by gum chewing was introduced and was found to lead to an early recovery of intestinal motility and shortened hospital stay \[[@B6]\].
Gum chewing promotes the cephalic-vagal reflex, which is activated when eating food, and it is thought to promote the secretion of hormones related to intestinal motility \[[@B7]\]. However, few studies have reported on the relationship between gum chewing and recovery from postoperative ileus after laparoscopic surgery. In this study, we aimed to examine the effect of gum chewing after laparoscopic colorectal cancer surgery.
METHODS
=======
We reviewed the medical records of patients who underwent laparoscopic colorectal cancer surgery from January 2012 to November 2012 at Incheon St. Mary\'s Hospital, The Catholic University of Korea School of Medicine.
In this study, patients who needed to have a stoma (temporary or permanent) made or had postoperative complications such as leakage were excluded. We divided the patients into 2 groups: group A consisted of 67 patients who did not chew gum; group B consisted of 65 patients who chewed gum. Patients in the gum-chewing group were asked to chew gum, starting at the first postoperative day, 3 times a day, approximately 10-20 minutes at a time, until normal feeding was resumed. Postoperative care of both groups was the same and was in accordance with the institution\'s clinical pathway program with the exception of gum chewing. We analyzed the short-term clinical outcomes between the two groups to evaluate the effect of gum chewing. In this study, P \< 0.05 was considered statistically significant.
Patients were considered appropriate for discharge if they passed flatus, passed urine freely, ambulated independently, and tolerated a diet and if pain was managed with oral medication, the patient was willing to leave the hospital and had adequate home support, and there was no evidence of postoperative complication. These factors all contributed to the length of the postoperative hospital stay, and as mentioned above, patients who had postoperative complications were excluded in this study.
RESULTS
=======
No significant differences were observed between the 2 groups in terms of patient\'s demographics, American Society of Anesthesiologists physical status score, body mass index (BMI) or prior history of abdominal surgery ([Table 1](#T1){ref-type="table"}). The types of surgical procedures for colorectal cancer were not significantly different between the 2 groups. The first passage of gas was slightly earlier in group B, but did not show a significant difference. However, the length of hospital stay was 6.7 days in group B, which was significantly shorter than that in group A, 7.3 days (P = 0.018) ([Table 2](#T2){ref-type="table"}).
DISCUSSION
==========
Normal gastrointestinal motility is maintained through complex mechanisms involving the central nervous system, gastrointestinal system, hormones, neurotransmitters and inflammatory reactions \[[@B8]\]. Postoperative gastrointestinal motility is normally recovered within several hours for the small intestine, 24-48 hours for the stomach, and 3-5 days for the large intestine \[[@B9]\]. Postoperative ileus is one of the most common complications after abdominal surgery \[[@B2]\]. Although its pathogenic mechanism is still controversial, factors such as overstimulation of the sympathetic nervous system, increased blood catecholamine levels, electrolyte imbalance, irritation of the peritoneum or retroperitoneum, use of narcotic analgesics and type of surgical procedure are thought to contribute to postoperative ileus \[[@B10], [@B11]\].
Recently, the importance of neurohormonal factors has been highlighted, and the splanchnic nerve is known to play an important role in the pathogenic mechanism of ileus \[[@B12]\]. After surgery, vasoactive intestinal polypeptide levels increase, which directly suppresses the contraction of the smooth muscles of the small intestine \[[@B13]\], but the secretion of hormones that promote intestinal motility, such as gastrin, neurotensin and pancreatic polypeptide, decreases \[[@B11]\]. Pain increases the secretion of substance P, which suppresses intestinal motility \[[@B14], [@B15]\], and the use of analgesics affect the enteric nervous system, increasing enteric input and decreasing gastrointestinal motility and peristalsis.
In addition a large surgical site has been shown to lead to severe tissue damage, which, in turn, activates local inflammatory factors, thereby causing ileus \[[@B16]\]. For this reason, the incidence of postoperative ileus in laparoscopic surgery is lower than that in conventional open surgery \[[@B3]\]. Rapid recovery of intestinal motility after abdominal surgery could decrease the length of postoperative hospital stay and the medical costs.
Gum chewing is a type of \"sham feeding\" that activates the cephalic-vagal reflex in a manner similar to that when eating food and stimulates the motility of the duodenum, stomach and rectum. Gum chewing increases the serum concentrations of gastrin, neurotensin and pancreatic polypeptide, stimulates the motility of the duodenum, stomach and rectum, and promotes intestinal motility \[[@B17]-[@B19]\]. Several studies reported that the first passage of gas was faster and the length of postoperative hospital stay were significantly less in the gum-chewing group; also, rapid recovery of intestinal motility was shown in the gum-chewing group \[[@B6], [@B20]\].
Although we used feeding protocols that commenced with sips of water on the second postoperative day, with the diet being escalated only after passage of flatus, there has been a move toward early postoperative feeding regimes as part of enhanced recovery after surgery (ERAS) program. The ERAS program features a push toward earlier gastrointestinal recovery and discharge, and some patients feel discomfort with ERAS feeding protocol. In this case, we think that gum chewing would lead to improved gut and functional recovery and make patients feel more comfortable.
This study had some limitations. First, although the data were collected prospectively, this study was a retrospective study. Second, the study volume was too small to evaluate the exact effect of sham feeding. A larger scale prospective randomized study is needed to evaluate the exact effect of sham feeding on the recovery from laparoscopic colorectal cancer surgery. Third, the use of analgesics could affect the recovery of intestinal motility, so the effect of analgesics must be taken into consideration \[[@B21], [@B22]\]. However, all patients in this study used patient-controlled analgesia; unfortunately we were not able to determine the correlation between the use of analgesics and recovery of intestinal motility. Fourth, we recommend patients for discharge according to the criteria discussed above, but some refused to be discharged from the hospital for the reasons other than medical reasons; thus, the length of postoperative hospital stay might have been influenced by that decision.
Gum chewing is extremely easy and very cost-effective. We also think that gum chewing can help prevent dehydration of the mouth. In addition, the process of chewing, the sweet taste, and the smell of gum satisfied the patient\'s appetite, leading to increased patient comfort.
In conclusion, this study showed that the length of the postoperative hospital stay was shorter in the gum-chewing group, and we think that gum chewing is an easy and cost-effective method to reduce the length of the postoperative hospital stay for laparoscopic colorectal cancer surgery. In future studies, we expect to elucidate the effect of gum chewing on the postoperative recovery more clearly.
No potential conflict of interest relevant to this article was reported.
######
Demographics of patients
Values are presented as mean ± standard deviation or number (%).
ASA, American Society of Anesthesiologists.
######
Short-term clinical outcomes between the two groups
Rt, right; Lt, left.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-materials-10-00304}
===============
Controlled drug-delivery systems were designed to deliver drugs at desirable times and/or to specific sites for achieving a therapeutic purpose \[[@B1-materials-10-00304]\]. For developing drug-delivery carriers, the major challenges were preparation of a suitable biomaterial that could ensure an excellent drug-releasing rate at the required dose to a target location, while also being nontoxic with good biocompatibility \[[@B2-materials-10-00304]\]. In recent years, considerable attention has been paid to the synthesis of intelligent hydrogels for the purpose of drug delivery because of their responsiveness to external stimuli such as pH, solvent composition, ionic strength, temperature and light, and other factors \[[@B3-materials-10-00304],[@B4-materials-10-00304]\].
Temperature- and pH-sensitive hydrogels represent the most investigated class of intelligent hydrogels, and have been used extensively \[[@B5-materials-10-00304],[@B6-materials-10-00304],[@B7-materials-10-00304],[@B8-materials-10-00304],[@B9-materials-10-00304]\]. Variations in pH occur at several body sites, such as the gastrointestinal tract, blood vessels, and vagina, and therefore suitable pH- and temperature-responsive hydrogels are needed for drug release. Temperature- and pH-sensitive hydrogels can protect a drug from enzymatic hydrolysis or being destroyed by stomach acid, and therefore they can be considered as the ideal candidate for a drug-delivery system \[[@B10-materials-10-00304]\]. Nevertheless, it should be noted that most traditional temperature- and pH-sensitive hydrogels generally have several limitations. Firstly, the traditional hydrogels usually suffer from poor biodegradability and biocompatibility. Secondly, the drug-loading and encapsulation efficiency are limited. Finally, another serious limitation of the normal hydrogel as a potential drug carrier is its poor sustained release capability \[[@B11-materials-10-00304]\]. Therefore, for the preparation of hydrogels possessing nontoxicity, biocompatibility, high drug-loading and encapsulation efficiency, and controlled drug delivery, it is very necessary to change their structure.
In recent years, natural polymer-based hydrogels have aroused broad interests because of their unique adjustable structure, and properties such as the low cost and good biodegradability and biocompatibility \[[@B3-materials-10-00304],[@B12-materials-10-00304],[@B13-materials-10-00304],[@B14-materials-10-00304],[@B15-materials-10-00304]\]. Hemicellulose, a renewable plant polysaccharide, ranks second to cellulose in terms of lignocellulosic biomass content, and accounts for 1/4 to 1/3 of agriculture residues \[[@B16-materials-10-00304]\]. Due to the low cost and good biodegradability and biocompatibility, the high value-added utilization of hemicellulose has attracted much attention \[[@B17-materials-10-00304]\]. Xylan, as the most common hemicellulose, has unique physiological properties, such as inhibition of cell mutation; promotion of cell adhesion, proliferation, and innate immunological defense; and anticancer effects. These properties make xylan suitable for preparation of hydrogels used in drug release and biomedical engineering \[[@B8-materials-10-00304],[@B18-materials-10-00304]\]. Moreover, hydroxyl and carboxylic groups on the xylan chains provide more opportunities for chemical or enzymatic modifications, while maintaining their native structure. Especially, the introduction of carboxyl groups and unsaturated double bonds can improve the pH response performance and the reaction efficiency of hemicellulose-based hydrogels, thereby extending their application in drug release. Maleic anhydride (MAH), as a small molecule of multifunctional groups, can react chemically with carbohydrate hydroxyls to form polymeric intermediates--MAH derivative, which contains both carboxyl groups and unsaturated double bonds. The carboxyl groups and double bonds in these derivatives can improve the pH response and adjust the cross-linking density of hydrogels and thus can be used as functional groups in drug delivery \[[@B19-materials-10-00304],[@B20-materials-10-00304]\]. Therefore, MAH-modified xylan (MAHX) can also be imparted with potential applications in the preparation of hydrogels for drug delivery.
In this work, new pH- and thermoresponsive hydrogels were prepared by the cross-linking polymerization of MAHX with *N*-isopropylacrylamide (NIPAm) and acrylic acid (AA) using *N*, *N*'-methylene-bis-acrylamide (MBA) as the cross-linker under UV irradiation to form MAHX-*g*-P(NIPAm-co-AA) hydrogels, and their application in the drug-release system was studied. The aim of this study was to improve the encapsulation efficiency and regulate the drug controlled release behaviors of hydrogels by changing the degree of substitution (DS) of MAHX and the hydrogels' structure. The physical and chemical properties of hydrogels were characterized by the swelling ability, the lower critical solution temperature (LCST), scanning electron microscopy (SEM) and Fourier-transform infrared spectroscopy (FTIR). Acetylsalicylic acid is widely used as a model compound because of its prevention of and therapy for thrombotic diseases, and it represents structures which form parts of many drug substances (benzene ring and carboxyl group) \[[@B21-materials-10-00304]\]. Theophylline, belonging to alkaloids, is a bronchodilator widely used in asthma therapy, and can exist either as the anhydrate or the monohydrate depending on the conditions under which it is stored \[[@B22-materials-10-00304]\]. The two compounds were applied as drug models for the study of drug-delivery performance of prepared hydrogels. Moreover, the cytocompatibility of MAHX-based hydrogels was studied by MTT methods.
2. Experimental Section {#sec2-materials-10-00304}
=======================
2.1. Materials {#sec2dot1-materials-10-00304}
--------------
Beech wood xylan (*M*~w~ of 130,000 g·mol^−1^) was obtained from Sigma Aldrich, Germany. 1-Butyl-3-methylimidazoliumchloride (\[BMIM\]Cl) ionic liquids (ILS) were bought from Lanzhou Greenchem ILS, LICP. CAS (Lanzhou, China). MBA (98%) and acetylsalicylic acid (99%) were supplied by Aladdin Reagent Company Limited (Shanghai, China). LiOH, maleic anhydride (A. P.), and theophylline were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). NIPAm (98%), AA (98%), 2,2-dimethoxy-2-phenylacetophenone (DMPA, 99%) *N*-methyl pyrrolidone (NMP, 99%) were obtained from Guangzhou Chemical Reagent Factory. The NIH3T3 cells were supplied by School of Chemical Engineering, Jinan University (Guangzhou, China), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide (MTT) was supplied by Sigma-Aldrich (St. Louis, MO, USA), and fetal bovine serum (FBS) was supplied by SiJiqing Bio-engineering Material Company (Hangzhou, China). All chemical reagents were analytical reagent grade.
2.2. Preparation of MAHX or MAHX-Based Hydrogels {#sec2dot2-materials-10-00304}
------------------------------------------------
### 2.2.1. Preparation of MAHX {#sec2dot2dot1-materials-10-00304}
MAHX was synthesized according to our previous work \[[@B23-materials-10-00304]\]. Dry xylan (0.33 g) was dissolved into \[BMIM\]Cl ionic liquids (2.5%, *w*/*w*) to form the mixed solution. The mixture was stirred at 90 °C for 4 h under the protection of nitrogen to guarantee complete dissolution of xylan. Then, 0.0025 g of LiOH and the required amount of MAH ([Table 1](#materials-10-00304-t001){ref-type="table"}) were added and stirred for 80 min at 80 °C. After the required time, the mixture was cooled to room temperature, precipitated with 95% (*w*/*w*) ethanol, and centrifuged at 4000 rpm for 30 min. The precipitate was washed by 95% (*w*/*w*) ethanol. In the end, obtained MAHX was dried at 45 °C in a vacuum oven for 16 h. The DS of MAHX was determined according to the method in our previous work \[[@B23-materials-10-00304]\].
### 2.2.2. Preparation of MAHX-Based Hydrogels {#sec2dot2dot2-materials-10-00304}
MAHX-based hydrogels were synthesized by the cross-linking polymerization under UV irradiation. The detailed synthesis conditions are summarized in [Table 1](#materials-10-00304-t001){ref-type="table"}. Briefly, MAHX with different DS (in [Table 1](#materials-10-00304-t001){ref-type="table"}) was dissolved in distilled water at a concentration of 5% (*w*/*w*) and stirred until MAHX was completely dissolved. Under nitrogen atmosphere, a certain amount of NIPAm, AA, and MBA were added, respectively. Photoinitiator (DMPA 5% *w*/*w*, based on MAHX weight) dissolved in the NMP solution (2.5% *w*/*w*) was added to the mixed solution. Then, the solution was transferred into a glass beaker with a diameter of 5 cm and irradiated under ultraviolet light (365 nm and 40 W) at the intensity of \~199.9 mW/cm^2^ \[[@B24-materials-10-00304],[@B25-materials-10-00304]\] for 6 h at room temperature. The obtained hydrogels were washed thoroughly in 250 mL deionized water for 5 days to remove unreacted chemicals, and the used deionized water was changed twice a day. Finally, the obtained hydrogels were freeze-dried (105 Torr, 48 h, and −70 °C).
2.3. The Swelling Behavior Studies of Hydrogels {#sec2dot3-materials-10-00304}
-----------------------------------------------
Swelling studies of the MAHX-based hydrogels were carried out by the gravimetric method. The dried hydrogels were swollen in buffer solutions with different pH values (pH 1.5 and pH 7.4) at different temperatures (from 25 °C to 37 °C), respectively, to reach the equilibrium swelling state. The ionic strength of the buffer solution was adjusted by NaCl to keep it constant at 0.5 M. All samples were measured three times and the average taken. The design formula of the equilibrium swelling ratios (*S~eq~*) is shown as follows \[[@B26-materials-10-00304]\]: where *W~eq~* and *W~d~* are the weight of swelling hydrogels and dried hydrogels, respectively.
2.4. Compression Test of Hydrogels {#sec2dot4-materials-10-00304}
----------------------------------
Compressive stress measurement was taken using an Instron Universal Testing Machine 5565. Hydrogels were cut into columns with a diameter of 20 mm and height of 20 mm using scalpels. The testing was carried out with a speed of 2 mm/min until the compression ratio reached to 50% in a room with constant temperature and humidity (25 °C and 50% humidity).
2.5. Characterization {#sec2dot5-materials-10-00304}
---------------------
SEM (Hitachi S3700, Tokyo, Japan) with an acceleration voltage of 10 kV was carried out to study and compare the surface morphology of xylan, MAHX, and MAHX-based hydrogels. FTIR spectra of samples were recorded in KBr pellets on Nicolet 750 (Sarasota, FL, USA) and the wavelength ranged from 4000 to 400 cm^−1^. Differential scanning calorimetry (DSCQ200, TA Instrument, Wilminton, NJ, USA) analysis was utilized for studying the LCST of hydrogels. The thermal analyses were performed from 15 °C to 55 °C on the swollen hydrogels under a dry nitrogen atmosphere with a flow rate of 25.0 mL/min and a heating rate 5 °C/min.
2.6. Loading of Acetylsalicylic Acid and Theophylline into Hydrogels {#sec2dot6-materials-10-00304}
--------------------------------------------------------------------
The dried hydrogels were submerged in ethanol solution containing 10% (*w*/*v*) acetylsalicylic acid and theophylline, respectively, and stirred for 24 h; loading was performed under the dark environment \[[@B27-materials-10-00304]\]. Then, hydrogels were dried in a vacuum oven. The weight was recorded when the drug-loaded hydrogels reached a constant weight. The content of acetylsalicylic acid and theophylline in hydrogels and the encapsulation efficiency were estimated by the following equations \[[@B28-materials-10-00304]\]: where *W~d~* and *W~h~* represent the total weight of drug in hydrogels and the total weight of dried hydrogels, respectively. *W~a~* and *W~m~* refer to the actual drug-loading weight and the theoretical maximum drug-loading weight, respectively.
2.7. In Vitro Drug Release {#sec2dot7-materials-10-00304}
--------------------------
The dried hydrogels were ground into powder and then measurements were carried out with a physical property measurement system (PPMS-9, Quantum Design Inc., San Diego, CA, USA) at room temperature. The release behavior of drug-loaded hydrogels was characterized using a horizontal oscillator with a shaking speed of 50 rpm at 37 °C. The drug-loaded hydrogels were immersed into the artificial gastric liquid (pH = 1.2, HCl--NaCl buffer solution) and the artificial intestinal liquid (pH = 7.4, Tris--NaCl buffer solution), respectively. Five milliliters of each buffer solution was collected at 1 h intervals. The amount of acetylsalicylic acid and theophylline was determined by a UV/V spectrophotometer (SHIMADZU UV1800, Kyoto, Japan) at 296 nm and 270 nm, respectively. Meanwhile, a volume of medium equal to the solution removed for testing was added back to maintain a constant volume. All samples were measured three times for in vitro release tests and the average taken. The percentage of drug released was estimated as follows: where *W~dt~* is the weight of drug at time t and *W~∞~* is the total weight of loaded drug in hydrogels, respectively.
2.8. Cell Viability Study {#sec2dot8-materials-10-00304}
-------------------------
The cell viability assay was conducted by MTT method using NIH3T3 cells to measure the cell proliferation and the cell activity in the solution including prepared hydrogel powder (gel-3). NIH3T3 cells (2 × 10^3^ cells per well) were seeded in 96-well plates (100 µL) and cultivated for 24 h. Then, different concentrations of hydrogel samples (1.6, 0.8, 0.4, 0.2, 0.1, and 0.05 mg/mL) were put into the cells and cultivated for 24 h and 72 h. Then, the medium was removed and 30 μL MTT (5 mg/mL) was added, followed by 4 h of incubation at 37 °C. Then, the supernatant was removed, and 150 µL of DMSO was added to dissolve the formazan precipitate. The absorbance of the solution was measured by microplate reader (Bio-Rad 550, Hercules, CA, USA) at 570 nm to determine the optical density (OD) value. Each group was measured three times and the average taken. The cell viability was calculated by following formula \[[@B29-materials-10-00304]\]: where *OD~gel~* is the optical density of cells cultured on hydrogels and *OD~ctrl~* is the optical density of the control group.
3. Results and Discussion {#sec3-materials-10-00304}
=========================
3.1. Temperature and pH-Dependent Swelling Behaviors of Hydrogels {#sec3dot1-materials-10-00304}
-----------------------------------------------------------------
The swelling ability determined the physical properties and applications of hydrogels. To study the influence of the hydrogel composition (DS of MAHX) and environment conditions (temperature and pH) on the swelling ratio of hydrogels, a systematic study on the swelling characteristic was carried out and the results are shown in [Figure 1](#materials-10-00304-f001){ref-type="fig"}. The S~eq~ values of MAHX-based hydrogels (MAHX-gels) in pH 7.4 at different temperatures were illustrated in [Figure 1](#materials-10-00304-f001){ref-type="fig"}a. Obviously, the S~eq~ was affected by the temperature and the DS of MAHX. The S~eq~ of all the samples were decreased with the increase of temperature from 25 °C to 37 °C. Besides, the fastest decline in the S~eq~ of hydrogels occurred from 33 °C to 35 °C, revealing that the LSCT of hydrogels was distributed between 33 °C and 35 °C. This could be explained by the fact that when the temperature was above the LCST, NIPAm moieties became dehydrated, leading to the shrinkage of the network of hydrogels and thus decreasing the swelling ratio \[[@B30-materials-10-00304]\]. The S~eq~ of hydrogels at 37 °C in pH 7.4 firstly increased from 23.03 g/g to 31.03 g/g with the increase of the DS of MAHX from 0.12 (gel-1) to 0.48 (gel-3). However, further increasing the DS of MAHX from 0.48 (gel-3) to 0.65 (gel-4) caused the swelling ratio to decrease. The hydrophilicity of hydrogels could be improved by the modification of xylan and be enhanced by increasing the DS of MAHX and, consequently, increasing the swelling ratio. Meanwhile, when xylan was modified by MAH, due to the introduction of carboxyl groups, many highly active carbon--carbon double bonds were introduced into the xylan backbone. During the hydrogel formation, these carbon--carbon double bonds could cross-link with AA and NIPAm, resulting in a higher cross-link density, reducing the pore diameter of hydrogels, and thus decreasing the swelling ratio \[[@B31-materials-10-00304]\].
As shown in [Figure 1](#materials-10-00304-f001){ref-type="fig"}b, the S~eq~ of MAHX-gels decreased faster in pH 1.5 than in pH 7.4 ([Figure 1](#materials-10-00304-f001){ref-type="fig"}a). This was because the existence of many hydroxyl and carboxyl groups in the hydrogel network promoted the formation of hydrogen bonds, thereby inhibiting the expansion of the network of hydrogels in the acidic system. The difference was that the S~eq~ of MAHX-gels decreased with the increase of the DS of MAHX. This could be explained that the content of the carboxyl groups was increased with the increase of the DS of MAHX, resulting in stronger hydrogen-bonding interaction and higher cross-link density, thus decreasing the swelling ratio.
3.2. The Swelling and Deswelling Behaviors of Hydrogels in Buffers {#sec3dot2-materials-10-00304}
------------------------------------------------------------------
The swelling and deswelling behavior of hydrogels are important parameters used to evaluate its reversibility and durability. [Figure 2](#materials-10-00304-f002){ref-type="fig"} illustrates reversible swelling--deswelling behaviors of gel-3 in the buffer solutions with pH 1.5 and 7.4 at 37 °C. When the swollen gel-3 in the buffer solutions (pH 7.4) was transferred into the solution of pH 1.5, the S~eq~ of gel-3 was sharply decreased from 30.86 g/g to 4.25 g/g, 29.95 g/g to 4.38 g/g, and 30.52 g/g to 4.45 g/g, respectively. At a low pH region, the carboxylates (--COO--) in gel-3 were generally protonated to become COOH and thus led to the deswelling of the hydrogel. However, when the hydrogel was transferred into the solution of pH 7.4 again, the COOH became ionized, and the resulting electrostatic repulsion in the network led to the reswelling of hydrogels \[[@B32-materials-10-00304]\]. After three swelling--deswelling cycles, there was no obvious decrease in the swelling ratio of gel-3 in either pH 1.5 or pH 7.4. This indicated that hydrogels had good durability and reversibility, implying that these hydrogels have potential application in drug-delivery systems. A similar result was also reported for other hemicellulose-based poly (acrylic acid) hydrogels \[[@B33-materials-10-00304],[@B34-materials-10-00304]\].
The reversible swelling--deswelling behaviors of gel-1, gel-2, and gel-4 are displayed in [Figure S1 (Supporting Information file)](#app1-materials-10-00304){ref-type="app"} in buffer solutions at pH 1.5 and 7.4 at 37 °C. After three swelling and deswelling cycles, there was no obvious decrease in the swelling ratio of gel samples in either pH 1.5 or pH 7.4, which indicated that hydrogels had good durability and reversibility.
3.3. Morphological Analysis {#sec3dot3-materials-10-00304}
---------------------------
[Figure 3](#materials-10-00304-f003){ref-type="fig"} displays the SEM images of xylan (a); MAHX (b); and freeze-dried gel-1 (c); gel-2 (d); gel-3 (e); and gel-4 (f). Compared with the native xylan ([Figure 3](#materials-10-00304-f003){ref-type="fig"}a), the surface of MAHX ([Figure 3](#materials-10-00304-f003){ref-type="fig"}b) became coarser due to the attachment of MAH to the xylan backbone, which led to the change in the structural morphology of xylan. All the hydrogel samples displayed a honeycomb-like architecture. Moreover, the pore volume of hydrogels was increased with the increase of the DS of MAHX. The hydrophilicity of hydrogels was increased with the increase of the DS of MAHX, resulting in a larger swelling ratio, and thus increasing the pore volume. In addition, compared to gel-1 and gel-2, the network structure of gel-3 was more uniform. This was because the cross-link density of the hydrogel was increased with the increase of the DS of MAHX, resulting in a cross-link density perfect for forming the cross-linking network. However, further increasing the DS of MAHX caused the continued increase in the cross-link density of hydrogels, thereby resulting in an uneven crosslinking density, which induced a nonuniform network of gel-4. Interestingly, compared to the other gels (gel-1, gel-2, and gel-4), gel-3---with even larger pores and thinner walls---also had the highest drug-loading and encapsulation efficiency, as will be discussed. The same conclusions were obtained by Cao et al. and Zhao et al. \[[@B33-materials-10-00304],[@B35-materials-10-00304]\]. Thus, the gel-3 was the desirable candidate for drug delivery.
[Table 2](#materials-10-00304-t002){ref-type="table"} shows the mechanical properties of the observed hydrogels. The compress stress firstly increased from 53.34 kPa to 68.25 kPa with the increase of the DS of MAHX from 0.12 (gel-1) to 0.48 (gel-2). Further increasing the DS of MAHX from 0.48 (gel-1) to 0.65 (gel-4) resulted in the compress stress decreasing from 68.25 kPa to 67.56 kPa. The change trend of elasticity modulus for hydrogels was the same with the compress stress, and gel-3 had the highest compression strength and elasticity modulus, as shown in [Table 2](#materials-10-00304-t002){ref-type="table"}. Obviously, a denser network was observed with the increase of the DS of MAHX, thus, resulting in stronger compression strength. However, further increasing the DS of MAHX caused the continued increase in the cross-link density of hydrogels, consequently resulting in an uneven cross-linking density, which induced a vulnerable network of hydrogels \[[@B26-materials-10-00304]\]. This well agreed with the results obtained from the swelling ratio and SEM.
3.4. FTIR Spectra of Xylan, MAHX and Prepared Hydrogels {#sec3dot4-materials-10-00304}
-------------------------------------------------------
[Figure 4](#materials-10-00304-f004){ref-type="fig"} illustrates the FTIR spectra of xylan, MAHX, and gel-3. In the spectrum of xylan, the absorption at 3436 cm^−1^, 2920 cm^−1^, 1630 cm^−1^, 1460 cm^−1^, 1162 cm^−1^, 1040 cm^−1^, 980 cm^−1^, and 895 cm^−1^ are characteristic absorption bands of xylan \[[@B34-materials-10-00304],[@B36-materials-10-00304]\]. The bands at 2923 cm^−1^ and 1040 cm^−1^ are assigned to the C--H and C--O--C stretching vibration band of xylan, respectively. A sharp band at 895 cm^−1^ is associated with β-glycosidic linkages between the sugar units, indicating that the xylose residues forming the backbone of the macromolecule are linked by β-form bonds \[[@B37-materials-10-00304]\]. The wide absorbance at 3436 cm^−1^ is assigned to the hydroxyl stretching vibration of xylan. Compared with xylan, the absorption at 3436 cm^−1^ of MAHX and gel-3 decreased significantly, which indicated that esterification had occurred between the hydroxyl groups in xylan and the anhydride groups of the MAH. In the spectrum of MAHX, there are a few new bands compared with the spectrum of xylan in [Figure 4](#materials-10-00304-f004){ref-type="fig"}. The bands at 1736 and 1170 cm^−1^ are assigned to carbonyl stretching vibrations of C=O in anhydride groups and the C--O vibration in MAH, respectively \[[@B19-materials-10-00304]\], indicating the successful attachment of MAH onto xylan \[[@B23-materials-10-00304],[@B26-materials-10-00304]\]. In the spectrum of gel-3, the bonds at 1635 cm^−1^ and 1560 cm^−1^ are assigned to stretching vibration of --N--H in NIPAm \[[@B24-materials-10-00304],[@B38-materials-10-00304],[@B39-materials-10-00304],[@B40-materials-10-00304]\]. The bonds at 1452 cm^−1^ and 1382 cm^−1^ are assigned to the asymmetric absorption peak of COO^−^ and bending vibration of C--H originating from AA, respectively \[[@B41-materials-10-00304]\]. These signal changes observed in the FTIR spectra confirmed the successful synthesis of the target hydrogels.
3.5. LCST of Hydrogels {#sec3dot5-materials-10-00304}
----------------------
[Figure 5](#materials-10-00304-f005){ref-type="fig"} exhibits DSC heating scans of gel-1, gel-2, gel-3, and gel-4. The LCST of gel-1, gel-2, gel-3, and gel-4 emerged at 33.4 °C,33.7 °C,34.5 °C, and 34.2 °C, respectively. The LCST of hydrogels firstly increased with the increase of the DS of MAHX and then decreased. Gel-3 showed the highest LCST of 34.5 °C nearest to body temperature. It could be deduced that the hydrophilicity of the hydrogel networks increased with the increase of the DS of MAHX, increasing the combination of intermolecular forces, especially the hydrogen-bonding interaction between the hydrogel matrixes and water molecules in the system. This implied that breaking of hydrogen bonds required more energy, which made the transition temperature of LCST become higher and broader \[[@B42-materials-10-00304]\]. Nevertheless, further increasing the DS of MAHX led to the nonuniform cross-linking of the hydrogel network, which led to the decrease in the LCST. Therefore, the DS of MAHX used to prepare hydrogels had an important effect on the LCST of hydrogels. Thus, gel-3 with LCST of 34.5 °C nearest to the body temperature was the most suitable as the carrier for drug controlled release in the biomedical field.
3.6. In Vitro Drug Release of Hydrogels {#sec3dot6-materials-10-00304}
---------------------------------------
The drug-loading and encapsulation efficiency of gel-1, gel-2, gel-3, and gel-4 for acetylsalicylic acid are shown in [Table 2](#materials-10-00304-t002){ref-type="table"}. Both the drug-loading and encapsulation efficiency of hydrogels were firstly increased and then decreased with the increase of the DS of MAHX. Gel-3 had the highest drug-loading and encapsulation efficiency. The interpretation is that gel-3 had a more suitable pore diameter and ordered networks than gel-1 and gel-2, which provided more electrostatic interaction for acetylsalicylic acid, while the irregular internal structure of gel-4 had a lower affinity and electrostatic interaction than gel-3 \[[@B28-materials-10-00304]\]. Therefore, gel-3 was the most suitable candidate for the application as the carrier for drug controlled release.
To investigate whether gel-3 had special drug release behavior and whether the type of drug had an effect on the drug-release behavior of gel-3, acetylsalicylic acid and theophylline were used as model drugs for studying the in vitro release behavior of gel-3 at 37 °C. The cumulative acetylsalicylic acid and theophylline release at different pH conditions (pH 7.4 and pH 1.5) are shown in [Figure 6](#materials-10-00304-f006){ref-type="fig"}. It was found that the release rate of acetylsalicylic acid from gel-3 in simulated intestinal fluids (pH 7.4) increased quickly during the initial 4 h and then slowed down and approached the release equilibrium ([Figure 6](#materials-10-00304-f006){ref-type="fig"}a). The release equilibrium was achieved at 8 h and the maximum release percentage was 94.56%. While in the simulated gastric fluids (pH 1.2), gel-3 showed a slower and steadier acetylsalicylic acid release rate. The maximum release percentage (24.25%) was achieved within 4 h. The cumulative release value of acetylsalicylic acid in the simulated intestinal fluids (pH 7.4) was higher than that in the simulated gastric fluids (pH 1.5) during the releasing process. This could be attributed to the difference in the swelling ratios of hydrogels in different mediums.
Gel-3 showed the same trend for the release of theophylline ([Figure 6](#materials-10-00304-f006){ref-type="fig"}b). The release equilibriums were achieved at 7 h in pH 7.4 (87.25%) and 5 h in pH 1.5 (35.6%), respectively. In comparison, gel-3 presented the better release behavior for the acetylsalicylic acid release than that for theophylline release. This could be attributed to the existence of --COOH in acetylsalicylic acid, which could adjust the release rate fast and flexibly in different pH solutions. When the gel was put into the solution of pH 7.4, these --COOH groups in acetylsalicylic acid would be ionized into --COO--, the electrostatic repulsion forces would induce the expansion of the hydrogel network, and, finally, result in a faster release rate. When the hydrogel was put into a lower pH value (pH=1.5), the --COO-- groups would be generally protonated into --COOH with the pH decrease and further result in a slower release rate \[[@B33-materials-10-00304]\]. Besides, it might be speculated that the drug release behavior was controlled by both the performance of hydrogels and the nature of drugs.
To study the sequential release behaviors, the acetylsalicylic acid-loaded hydrogel was firstly put into simulated gastric fluids (pH 1.5) for 3 h and then transferred into simulated intestinal fluids (pH 7.4) for 7 h, and the results are shown in [Figure 7](#materials-10-00304-f007){ref-type="fig"}. The initial release of acetylsalicylic acid in simulated gastric fluids (pH 1.5) was fast during the first 2 h and then slowed down and achieved the release equilibrium within 3 h. The maximum release percentage was 24.26% at 3 h. In the simulated intestinal fluid (pH 7.4), the release rate increased rapidly and reached to 52.36% at 4 h, nearly double that of the first 3 h in pH 1.5. The release equilibrium was achieved at 8 h and the cumulative release percentage was 90.5%. Moreover, the chromogenic reaction between acetylsalicylic acid and ferric chloride (FeCl~3~) was carried out to test whether acetylsalicylic acid was hydrolyzed or not, which is displayed in [Figure S2 (Supporting Information file)](#app1-materials-10-00304){ref-type="app"}. The results showed that no chromogenic reaction happened after dropping FeCl~3~ into the acetylsalicylic acid solution released by gel-3. This implied that acetylsalicylic acid was not hydrolyzed during the releasing process. Therefore, these hydrogels could efficiently control the release of drug in gastric fluids and intestinal fluids, and had the desired protective effect in the stomach for oral drug delivery. When hydrogels were transferred to the simulated intestinal fluid (pH 7.4), most of acetylsalicylic acids were released from the hydrogel network. Thus, prepared hydrogels had the intestinal-targeted drug-delivery function.
3.7. Cytocompatibility of Hydrogels {#sec3dot7-materials-10-00304}
-----------------------------------
NIH3T3 cells were employed as model cells for evaluating the cytocompatibility of gel-3 by the MTT method \[[@B43-materials-10-00304]\]. [Figure 8](#materials-10-00304-f008){ref-type="fig"} illustrates the NIH3T3 cell viability under the different concentrations of gel-3 (0.05--1.6 mg/mL) after cultivating for 24 h and 72 h. After cultivating for 24 h, the cell viability of NIH3T3 cells was firstly increased from 73.6% to 88.1% with the decrease of the gel-3 concentration from 1.6 mg/mL to 0.4 mg/mL, and then decreased from 88.1% to 64.1% with further decreasing of gel-3 concentration from 0.4 mg/mL to 0.05 mg/mL. This could be interpreted as a higher concentration of hydrogel samples having an inhibitory effect on NIH3T3 cells at the beginning of the cultivating (24 h), while the hydrogel showed a significant decrease in cell survival at a lower concentration of hydrogel. This might be because the decrease of xylan content resulted in the decrease of cell compatibility. As expected, the cell viability of NIH3T3 cells was enhanced to some degree at different concentrations of gel-3 after cultivating for 72 h. These results confirmed that the xylan-based hydrogel had favorable biocompatibility with NIH3T3 cells. The same conclusions were obtained by Voepel et al. and Zhao et al. \[[@B35-materials-10-00304],[@B44-materials-10-00304]\].
[Figure 9](#materials-10-00304-f009){ref-type="fig"} illustrates the micrographs of NIH3T3 cell distribution under the different concentrations of gel-3 samples after cultivating for 24 h and 72 h. The NIH3T3 cells represented an aggregate distribution and they were relatively lower in number after cultivating for 24 h, while the NIH3T3 cells appeared to have a significant proliferation when they were distributed densely and uniformly after cultivating 72 h. The results were consistent with the cell viability shown in [Figure 8](#materials-10-00304-f008){ref-type="fig"}. The observation and results above were indicative of the promising potential of xylan-based hydrogels in human drug-delivery fields because of their excellent biocompatibility.
4. Conclusions {#sec4-materials-10-00304}
==============
In summary, new maleic anhydride modified xylan (MAHX)-based hydrogels with temperature/pH dual sensitivity and controllable drug delivery behavior were obtained by the cross-linking polymerization under UV light. These hydrogels presented good swelling--deswelling properties, honeycomb-like architecture, and temperature/pH dual sensitivity. Moreover, the pore volume, the mechanical properties, and the drug release rate of hydrogels could be controlled by the DS of MAHX. In vitro, the cumulative release rate of acetylsalicylic acid for MAHX-based hydrogels was higher than that of theophylline. Besides, in the gastrointestinal sustained drug release, the acetylsalicylic acid release rate was extremely slow at initial 3 h in the gastric fluid (24.26%), and then the cumulative release rate reached 90.5% after sustained release for 5 h in the stimulated intestinal fluid. Importantly, MAHX-based hydrogels had satisfactory biocompatibility with NIH3T3 cells. Therefore, MAHX-based hydrogels as drug carriers have potential application in human drug-delivery fields.
This work was supported by grants from National Natural Science Foundation of China (No. 21406080), the Foundation (No. KF201608) of Key Laboratory of Pulp and Paper Science and Technology of Ministry of Education/Shandong Province of China, and the Fundamental Research Funds for the Central Universities of SCUT.
######
Click here for additional data file.
Supplementary Materials: The following are available online at [www.mdpi.com/1996-1944/10/3/304/s1](www.mdpi.com/1996-1944/10/3/304/s1). Figure S1. Reversible swelling-deswelling behaviors of gel-1, gel-2 and gel-4 in the buffer solutions of pH 1.5 and 7.4 at 37 °C. Figure S2. The color change in the acetylsalicylic acid solution released by gel-3 before (a) and after (b) the addition of FeCl.
Wei-Qing Kong and Cun-Dian Gao performed the experiments; Wei-Qing Kong and Shu-Feng Hu analyzed the data; the paper was written under the direction and supervision of Jun-Li Ren, Li-Hong Zhao, and Run-Cang Sun; Wei-Qing Kong is responsible for the writing of this work.
The authors declare no conflict of interest.
{#materials-10-00304-f001}
{#materials-10-00304-f002}
{#materials-10-00304-f003}
{#materials-10-00304-f004}
{#materials-10-00304-f005}
{#materials-10-00304-f006}
{#materials-10-00304-f007}
{#materials-10-00304-f008}
{#materials-10-00304-f009}
materials-10-00304-t001_Table 1
######
Degree of substitution (DS) of maleic anhydride-modified xylan (MAHX) and the preparation conditions of MAHX-based hydrogels.
Samples MAH/xylan DS of MAHX NIPAm/MAHX (g/g) AA/MAHX (g/g) MBA/MAHX (g/g)
--------- ----------- ------------ ------------------ --------------- ----------------
gel-1 1:1 0.12 0.1 10:1 0.1
gel-2 2:1 0.28 0.1 10:1 0.1
gel-3 4:1 0.48 0.1 10:1 0.1
gel-4 8:1 0.65 0.1 10:1 0.1
AA: acrylic acid; MBA: *N*,*N*'-methylene-bis-acrylamide; NIPAm: *N*-isopropylacrylamide.
materials-10-00304-t002_Table 2
######
The compressive properties and drug-loading and encapsulation efficiency of hydrogels.
Samples Compression Stress (kPa) Elasticity Modulus (kPa) Drug ^a^ Loading (%) Encapsulation Efficiency (%) LCST (°C)
--------- -------------------------- -------------------------- ---------------------- ------------------------------ -----------
gel-1 56.34 ± 3.68 78.54 ± 9.48 55.26 86.56 33.4
gel-2 60.42 ± 4.23 128.25 ± 6.56 58.2 89.62 33.7
gel-3 68.25 ± 3.34 168.67 ± 8.54 60.68 92.25 34.3
gel-4 67.56 ± 2.68 164.34 ± 8.78 58.45 90.56 34.2
LCST: lower critical solution temperature; ^a^ acetylsalicylic acid drug loading.
| {
"pile_set_name": "PubMed Central"
} |
We seek to define the gene regulatory networks that govern heart development and disease. Recently, we discovered that the hearts of neonatal mice can fully regenerate after partial surgical resection or myocardial infarction, but this capacity is lost early in life. We are currently exploring the molecular underpinnings of the neonatal regenerative response of the heart, with the long-term goal of discovering combinations of genes and drugs that promote cardiac repair and regeneration. Promotion of cardiomyocyte proliferation through activation of the Yap pathway and modulation of epicardial signaling systems have shown efficacy in enhancing these processes. We are also optimizing strategies for reprogramming of cardiac fibroblasts toward a cardiac cell fate as a means of replacing cardiomyocytes in injured hearts. We have shown that four transcription factors can cooperatively reprogram fibroblasts into cardiac-like myocytes in vitro, albeit relatively inefficiently. Forced expression of these factors in dividing non-cardiomyocytes in mice also allows reprogramming into functional cardiac-like myocytes, improves cardiac function and reduces adverse ventricular remodeling following myocardial infarction. Screens for small molecules and microRNAs that enhance cardiac reprogramming have revealed new insights into the mechanistic basis of this process and have allowed further optimization in human cells. Opportunities and obstacles in the path toward mammalian cardiac regeneration will be discussed.
| {
"pile_set_name": "PubMed Central"
} |
**Purpose:** Currently, the outcomes of aesthetic facial treatments are limited to patient-reported satisfaction and patient, clinician, or observer rating scales. These assessments are highly subjective and incapable of evaluating skin function and structure below the superficial epidermis. The purpose of this study was to determine the practicality of non-invasive 3D imaging and 0.33mm microbiopsy methods for the objective assessment of facial skin following laser treatment. In addition, we evaluatedskin repair following microbiopsy extraction.
**Methods:** Twelve patients received a single facial treatment with a 1470nm/2940nm laser. Assessments were performed before treatment, seven days, three weeks, and three months post-treatment. VISIA imaging was used to assess skin texture. High-resolution, non-invasive 3D skin imaging- ultrasonography and optical coherence tomography (OCT)- were used to measure epidermal/dermal thickness, blood flow, surface roughness, wrinkle depth and attenuation coefficient. 0.33mm diameter microbiopsies were collected for histological and gene expression evaluation. In order to assess scar formation following microbiopsy, fifteen subjects were randomized to receive no biopsy, biopsy on the left nasolabial fold, or biopsy on the right nasolabial fold. Six blinded raters assessed subject facial photos at baseline, one month, and three months post-biopsy to evaluate for a visualized scar.
**Results:** The non-invasive measurements were able to detect significant changes after treatment in a variety of parameters, including improvements in skin surface, a decrease in attenuation coefficient at 3 weeks (p\<0.05), and an increase in blood flow 0.5mm to 0.7mm deep during the 5 day- to 3 week-period following treatment. (p\<0.05). Microbiopsy analysis revealed an increase in epidermal hyaluronic acid expression assessed by immunostaining and increased expression of inflammatory genes IL-1 beta at seven days post-treatment compared to untreated or three months post-treatment. No subjects exhibited scar formation post-biopsy. There was no visible evidence at the biopsy site seven days after the procedure, and OCT imaging showed a complete closing of the column created by the biopsy punch approximately 48 hours post-biopsy. At both one-month and three-months post-biopsy, six blinded photography reviewers were unable to determine whether a scar was present based on statistical analysis.
**Conclusions:** Non-invasive imaging successfully provides objective measurements of skin structure, texture, and blood flow to assess the effectiveness of facial skin rejuvenation treatments. Furthermore, microbiopsies can objectively evaluate the effects of these treatments at molecular level without scarring or pain. A single treatment with the 1470nm/2940nm laser significantly improved skin appearance at three weeks following treatments without significant changes in the tissue at the molecular level, as assessed by microbiopsy.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-nutrients-12-01872}
===============
Cerebrovascular diseases, such as stroke, and heart diseases, including angina pectoris and myocardial infarction, are the leading causes of death worldwide. Even for patients who recover, a major economic burden is placed on both the patient and society, as patients must undergo long-term treatments and extended bed rest. Hence, there is a need to improve the vascular function of patients in an effort to prevent vascular diseases. Accordingly, our laboratory has been studying the effects of food materials on vascular function and attempting to elucidate the mechanisms through which functional foods may prevent vascular disease \[[@B1-nutrients-12-01872],[@B2-nutrients-12-01872],[@B3-nutrients-12-01872]\].
Blood vessels consist of three layers: the outermost tunica adventitia, the tunica media, and the innermost tunica intima, which contains vascular endothelial cells (VECs). Nitric oxide (NO) is released from VECs to protect blood vessels by regulating their contraction and relaxation and by preventing thrombus formation caused by attachment of white blood cells, and other blood components, to the vascular endothelium \[[@B4-nutrients-12-01872],[@B5-nutrients-12-01872]\]. Meanwhile, if VECs become damaged by oxidative stress induced by reactive oxygen species, or oxidised low-density lipoprotein (LDL), the production of NO is suppressed, thereby increasing the risk of cardiovascular diseases \[[@B6-nutrients-12-01872]\]. Accordingly, improving NO production by VECs is critical for protecting blood vessels. In our previous research, *Raphanus sativus* cv. Sakurajima Daikon (Sakurajima radish; [Figure 1](#nutrients-12-01872-f001){ref-type="fig"}A) extracts were found to induce NO production, and its active ingredient was identified as trigonelline ([Figure 1](#nutrients-12-01872-f001){ref-type="fig"}B) \[[@B1-nutrients-12-01872]\]. Sakurajima radish, produced in Kagoshima, Japan, is famously known as the world's largest radish cultivar \[[@B7-nutrients-12-01872],[@B8-nutrients-12-01872]\]. However, no studies have yet evaluated the concentration of trigonelline that becomes absorbed and subsequently transferred into the blood following human consumption of Sakurajima radish. Moreover, although the bioregulatory functions of trigonelline, including improved hypertension, diabetes, and central nervous system disease, have been reported \[[@B9-nutrients-12-01872],[@B10-nutrients-12-01872]\], the effects of Sakurajima radish on human endothelial function have not been characterised.
Therefore, in this study, we evaluated trigonelline contents in various plants, including several varieties of Sakurajima radish, including 'Sakurajima Ogojo', which is a smaller F1 variety with a high germination rate cultivated in Kagoshima Prefecture; the 'Native farm species', which is a larger variety with a germination rate of approximately 75% \[[@B11-nutrients-12-01872],[@B12-nutrients-12-01872]\]; and 'Others', which represents other varieties obtained from the markets. Furthermore, we assessed the effects of cooking and processing Sakurajima radish on trigonelline and conducted the first trial of Sakurajima radish in humans.
2. Materials and Methods {#sec2-nutrients-12-01872}
========================
2.1. Materials {#sec2dot1-nutrients-12-01872}
--------------
Sakurajima radish, Aokubi radish (*Raphanus sativus var. Longipinnatus*), coffee cherry (*Coffea arabica*), and squash (*Cucurbita maxima*), cultivated in Kagoshima, Japan, were used in this study. Different varieties of Sakurajima radish were obtained, including 'Sakurajima Ogojo', an F1 variety with lower occurrence of hollow cavity and pores; 'Native farm species', are the largest varieties and have been inherited for many years; and 'Others' which represented a mixture that do not classified other varieties. Seeds of Sakurajima radish were obtained from the Japan Agricultural Cooperative (Kagoshima, Japan). Porcine VECs were purchased from Cosmo Bio Co., Ltd. (Tokyo, Japan). Trigonelline and methanol for high-performance liquid chromatography (HPLC) were purchased from FUJIFILM Wako Chemical Corporation (Osaka, Japan).
2.2. Quantification of Trigonelline {#sec2dot2-nutrients-12-01872}
-----------------------------------
The roots and leaves were separated and cut into small pieces. The roots were processed using a homogeniser and lyophilised to generate powdered raw material. One mL of methanol/H~2~O/acetic acid solvent (95.0%/9.5%/0.5%, *v*/*v*/*v*) was added to 25 mg raw material and mixed in a vortexer, followed by 5 min of ultrasonic treatment. The sample was centrifuged twice at 1600× *g* for 10 min at 4 °C (cooling centrifuge 3500; KUBOTA Corporation Co., Ltd., Tokyo, Japan); the supernatant was then collected and concentrated by drying. The dry sample material was weighed and dissolved in an appropriate solvent prior to use in experiments.
After filtration through a 0.45-μm filter (Toyo Roshi Kaisha, Tokyo, Japan), the samples were analysed by HPLC with an ultraviolet-visible adsorption detector and a photodiode array detector (Extrema, Jasco, Tokyo, Japan). The conditions for HPLC analysis were as follows: the C18 reversed-phase column (COSMOSIL 5C~18~-AR-300, 5 μm, 4.6 mm I.D., 250 mm; NACALAI TESQUE, INC., Kyoto, Japan) was maintained at 40 °C, and detection was conducted at 265 nm. The mobile phase consisted of 10 mM phosphoric acid solution (A) and methanol (B). We used a gradient of 0 min with 10% solution B, and 0--10 min with a direct increase in solution B of up to 40%. The flow rate was 0.7 mL/min, and the injection volume was 10 μL.
2.3. Cooking and Processing {#sec2dot3-nutrients-12-01872}
---------------------------
After harvesting Sakurajima radish, 1 g of the washed product was weighed to obtain a raw sample. The samples used for analysis of the effects of boiling were sliced thinly and boiled at 100 °C for 1, 10, 20, and 30 min; whereas the samples used for analysis of the effects of high-temperature cooking were baked in an oven or fried in oil for 1, 5, 10, 15, and 20 min at 180 °C. The samples used for analysis of the effects of freezing were stored in a deep freezer at −80 °C after initial freezing in liquid nitrogen. In addition, salted samples were immersed in 5% salt for 24 h, and pickled samples were boiled with vinegar (1 mL), water (1 mL), sugar (0.2 g), and salt (0.06 g; pH 3.8) for 24 h.
2.4. First Trial of Sakurajima Radish Examining the Vasodilator Property in Humans {#sec2dot4-nutrients-12-01872}
----------------------------------------------------------------------------------
Fourteen healthy volunteers (seven men and seven women, age 33.9 ± 6.7 years) participated in this study. Sakurajima radish (native species) was consumed at 170 g/day (trigonelline content: 61.2 mg), which was assumed to be sufficient for promoting NO production from VECs \[[@B1-nutrients-12-01872]\]. Other lifestyle factors were not altered. The intake period was ten consecutive days, and participants were permitted to either eat the entire 170-g portion, which was packaged using a vacuum packer (MINI JUMBO, RO18765-1; Nichiwa Electric Corporation, Tokyo, Japan), at once or divide it into multiple servings. Additionally, there were no limitations to the method used to cook the radish. Blood pressure (BP), pulse, and weight were measured before consumption and then again after ten days of consumption of Sakurajima radish. Blood samples were collected before consumption and then again after ten days of consumption of Sakurajima radish, and general biochemical tests (white blood cell \[WBC\]; haemoglobin \[Hb\]; platelet \[Plt\]; LDL cholesterol \[LDL-C\]; high-density lipoprotein cholesterol \[HDL-C\]; triglyceride \[TG\]; fasting plasma glucose \[FPG\]; uric acid \[UA\]; blood urea nitrogen \[BUN\]; creatinine \[Cr\]; sodium \[Na\]; potassium \[K\]; chloride \[Cl\]; aspartate transaminase \[AST\]; and alanine transaminase \[ALT\]) were performed. For analysis of vascular endothelial function, flow-mediated dilation (FMD) values were determined using an ultrasonic diagnostic imaging device (UNEXEF18VG; UNEX Co., Nagoya, Japan), which is an automated edge detection system for measurement of brachial artery diameter. FMD represents endothelium-dependent, largely NO-mediated, dilatation of conduit arteries in response to an imposed increase in blood flow and shear stress and is a tool for examining the pathophysiology of cardiovascular diseases to potentially identify individuals who are at increased risk of future cardiovascular events \[[@B13-nutrients-12-01872]\]. FMD was automatically calculated as the percentage change in peak vessel diameter from the baseline value. Percentage of FMD \[(peak diameter − baseline diameter)/baseline diameter\] was used for analysis \[[@B14-nutrients-12-01872]\]. Impaired endothelial function plays an important role in the initiation of atherosclerosis, and brachial FMD is a predictor of coronary artery disease severity \[[@B15-nutrients-12-01872]\]. To evaluate arteriosclerosis, brachial-ankle pulse wave velocity (baPWV) values were examined using a BP pulse wave tester (BP-203RPE III; Omron Healthcare, Kyoto, Japan). The blood concentration of trigonelline was determined by centrifugation (1000× *g*, 5 min) of plasma, and values were then quantified by HPLC as described above. This test was conducted with the approval of the institutional Ethics Review Committee by an outside expert (approval no. 31-1), and written informed consent was obtained from all participants.
2.5. Statistical Analyses {#sec2dot5-nutrients-12-01872}
-------------------------
Significant differences among groups were assessed using Student's *t*-tests. Data are represented as means ± standard deviations (SDs). Results with *p* values \< 0.05 were considered statistically significant. In the human trial, discrete data were presented as frequencies and percentages, and continuous variables were presented as means ± SDs. Continuous variables were compared using either *t*-tests or Mann Whitney U tests, and results with *p* values \< 0.05 were considered statistically significant.
3. Results {#sec3-nutrients-12-01872}
==========
3.1. Comparison of Trigonelline Content in Sakurajima Radish {#sec3dot1-nutrients-12-01872}
------------------------------------------------------------
Of the approximately 300 types of plants investigated, those in which trigonelline was detected were radish, coffee, and squash. When the amount of trigonelline in the root extracts of Sakurajima radish was measured and converted per weight of raw Sakurajima radish, raw Sakurajima radish was found to contain 360 ng/mg trigonelline. Setting this value as 100%, [Table 1](#nutrients-12-01872-t001){ref-type="table"} shows the relative values of trigonelline contents in other plants. Aokubi radish contained only 1.75% the trigonelline content as Sakurajima radish, and the same amount as squash. In contrast, fresh coffee cherry contained approximately 81.7% of the trigonelline as Sakurajima radish. However, the source of coffee that is often consumed does not originate from coffee cherry, but rather the seeds obtained by removing fermentable skin and pulp from coffee cherry. In addition, coffee cherry seeds cannot be eaten in their natural form, but instead by first be roasted at high temperatures. The trigonelline value decreased to 17.15% after roasting at 185 °C for 15 min. After extraction with hot water at 100 °C (French press method), coffee cherry lost up to 0.01% of its trigonelline content. Hence, the amount of trigonelline consumed from Sakurajima radish is overwhelmingly high.
3.2. Differences in Trigonelline Content between Sakurajima Radish Varieties {#sec3dot2-nutrients-12-01872}
----------------------------------------------------------------------------
The average trigonelline content in Sakurajima radishes was 260--360 ng/mg, which was significantly higher than that of other plants. However, although there was a tendency for the amount of trigonelline in "Sakurajima Ogojo" to be higher than that in the other varieties, no significant differences were observed as each variety showed large individual differences ([Figure 2](#nutrients-12-01872-f002){ref-type="fig"}). These findings suggest that all varieties of Sakurajima radish contained high amounts of trigonelline.
3.3. Effects of Cooking and Processing Sakurajima Radishes on Trigonelline Contents {#sec3dot3-nutrients-12-01872}
-----------------------------------------------------------------------------------
[Figure 3](#nutrients-12-01872-f003){ref-type="fig"}A shows the effects of cooking and processing radishes on trigonelline content, with the amount of trigonelline contained in raw Sakurajima radish set at 1. Although trigonelline is a water-soluble substance, all of it was detected in the radish when boiled for approximately 1 min. After boiling for more than 10 min, the amount of trigonelline was reduced to approximately half, and trigonelline was detected in the infusion. Since no change was observed in the total amount, it is considered that trigonelline was not decomposed by heating at 100 °C but rather was discharged from the radish to the surrounding infusion ([Figure 3](#nutrients-12-01872-f003){ref-type="fig"}B). After high temperature cooking, the centre temperature of the Sakurajima radish reached 180 °C. Heating at 180 °C for 10 min did not affect trigonelline, however, after 15 min it was found to decompose. Meanwhile, at 185 °C, a small amount of trigonelline was decomposed even after heating for 1 min; and after 10 min only very low levels were retained in the radish. At 195 °C, it decomposed rapidly after heating for only 1 min, and after 5 min it was barely detectable ([Figure 3](#nutrients-12-01872-f003){ref-type="fig"}C). Furthermore, the level of trigonelline was reduced by 5% in Sakurajima radish after one month of freezing, and was suppressed by only 8% after 11 months ([Figure 3](#nutrients-12-01872-f003){ref-type="fig"}D). Almost all trigonelline was detected in the radish regardless of changes in osmotic pressure due to salt and in pH due to vinegar. These findings also demonstrate that trigonelline was stable for cooking and processing Sakurajima radish.
3.4. First Trial of Sakurajima Radish Examining the Vasodilator Property in Humans {#sec3dot4-nutrients-12-01872}
----------------------------------------------------------------------------------
Following the first trial of Sakurajima radish in humans, no changes were observed in BP, or body weight ([Table 2](#nutrients-12-01872-t002){ref-type="table"}).
However, a trend towards reduced pulse was found after ten days (73.6 ± 11.0/min versus 70.9 ± 11.7/min, *p* = 0.07), which was not clinically significant. Moreover, biochemical examination of blood samples following ingestion of Sakurajima radish for ten days showed a significant decrease only in TG (110.1 ± 123.4 mg/dL versus 60.8 ± 33.0 mg/dL, *p* = 0.04; [Table 3](#nutrients-12-01872-t003){ref-type="table"}). Meanwhile, platelet counts were only slightly decreased (27.2 ± 5.9 × 10^4^/μL versus 26.1 ± 5.8 × 10^4^/μL, *p* = 0.10), which was not clinically significant. Furthermore, although trigonelline was not detected in any of the blood samples before consumption of Sakurajima radish, 1.6 ± 0.2 mg/mL trigonelline was detected after ten days of consumption ([Figure 4](#nutrients-12-01872-f004){ref-type="fig"}). This indicates that trigonelline was absorbed following intake of Sakurajima radishes. In addition, the %FMD, which is normal above 6.0, was significantly improved (6.7 ± 1.6% versus 9.4 ± 1.9%, respectively; *p* = 0.0016) following consumption for ten days. Lastly, baseline brachial artery diameter was 3.5 ± 0.5 mm before and 3.4 ± 0.6 mm after ten days consumption (*p* = 1.0). Hence, the baPWV did not show any significant differences (1184.4 ± 201.4 versus 1179.8 ± 197.8, respectively; *p* = 0.4).
4. Discussion {#sec4-nutrients-12-01872}
=============
Following comparison of the amount of trigonelline within the tested food ingredients, we found that Sakurajima radish served as the most significant source. In fact, the Aokubi radish, which belongs to the same Japanese radish family, contains only approximately 1/57 of the trigonelline content of Sakurajima radish. These results support those previously reported, which indicated a high NO-producing capacity of Sakurajima radishes \[[@B1-nutrients-12-01872]\]. Moreover, coffee products that are commonly consumed may contain low levels of trigonelline, unless prepared via specific manufacturing methods that do not require roasting \[[@B16-nutrients-12-01872]\]. Although the biosynthetic pathway of trigonelline has not been fully elucidated, the production of trigonelline from water-soluble vitamin nicotinic acid is likely catalysed by *N*-methyltransferase \[[@B17-nutrients-12-01872],[@B18-nutrients-12-01872],[@B19-nutrients-12-01872]\]. Considering that many plants use nicotinamide and nicotinic acid as final products, it is postulated that Sakurajima radish and coffee cherry synthesise the final product trigonelline via nicotinamide and nicotinic acid as intermediate products by activation of trigonelline synthase.
The effects on trigonelline following cooking and processing were also examined, and it was shown that trigonelline is highly stable even as a water-soluble substance. In fact, after boiling Sakurajima radish at 100 °C, trigonelline was detected in the radish for approximately 1 min, meanwhile, the amount of trigonelline was reduced to approximately half, and detected in the infusion after boiling for 10 min. In Japanese cuisine, Sakurajima radish is also commonly baked in the oven or fried in oil, hence, the effects of high temperature cooking were also investigated. Interestingly, following 10 min of baking or frying the trigonelline at 180 °C remained within Sakurajima radish rather than becoming exuded, as was observed following boiling. Furthermore, trigonelline was found to decompose when exposed to temperatures of 180 °C for more than 15 min, or temperatures ≥ 185 °C for any amount of time. This is reasonable considering the trigonelline reduction results obtained following roasting coffee shown in [Table 1](#nutrients-12-01872-t001){ref-type="table"}.
In addition, unlike Aokubi, which can be cultivated in approximately 3 months, Sakurajima radish requires a cultivation period of 5 to 6 months, with a harvest period of only 3 months per year. Hence, effective preservation techniques must be employed to allow for the distribution of Sakurajima radish throughout the entire year. Freezing, may be one such method as retention of trigonelline in frozen Sakurajima radish was very high (95%; [Figure 3](#nutrients-12-01872-f003){ref-type="fig"}A), and was found to be stable over an 11 months period. Taken together, these results suggest that Sakurajima radish can be subjected to general cooking and processing methods while retain high trigonelline contents, consistent with the many types of Sakurajima radish (pickles, sprinkles, rice crackers, Japanese sweets, frozen sweets, jelly drinks, and chips) found in the market.
The first trial of Sakurajima radish examining the vasodilator property in human found no significant differences in the blood biochemical parameters between the two groups i.e., those that consumed radish and those that did not, save for TG values. TG-rich lipoproteins represent causal risk factors for atherosclerotic cardiovascular diseases, and all-cause mortality \[[@B20-nutrients-12-01872]\]. A previous study suggested that trigonelline might reduce triglyceride levels, which was supported by our data \[[@B21-nutrients-12-01872]\]. Trigonelline was not detected in any blood samples before consumption of Sakurajima radish; meanwhile significant levels of trigonelline were detected after consumption of Sakurajima radish for ten days. This indicates that trigonelline was absorbed following consumption of Sakurajima radish. Furthermore, FMD is a surrogate marker of endothelial function that can complement clinical symptoms of structural arterial disease, while facilitating early diagnosis and prediction of cardiovascular diseases outcomes. The %FMD, which is considered normal above 6.0, was significantly improved in individuals who consumed Sakurajima radish for ten days compared to those that did not (9.4% ± 1.9% versus 6.7% ± 1.6%, respectively; *p* = 0.0016). Meanwhile, the baPWV did not exhibit significant differences between the two groups. Although endothelial dysfunction contributes to progression of structural arterial stiffness, consumption of Sakurajima radish for ten days does not appear to be sufficient to affect baPWV. Hence, longer periods of ingestion may be required to significantly affect baPWV and contribute to the prevention of atherosclerosis.
Recently, trigonelline contained in coffee was reported to have various bioregulatory functions in cells and animal models, including hepatic lipid accumulation, lipotoxicity \[[@B22-nutrients-12-01872]\], suppression of diabetes mellitus \[[@B23-nutrients-12-01872]\], and suppression of diabetic nephropathy \[[@B24-nutrients-12-01872]\]. In addition, human studies have reported the effects of consuming coffee containing trigonelline, including acute effects on glucose tolerance \[[@B25-nutrients-12-01872]\], management of patients with Parkinson's disease \[[@B26-nutrients-12-01872]\], and reduced levels of spontaneous DNA strand breaks in WBCs \[[@B27-nutrients-12-01872]\]. However, coffee also contains abundant amounts of caffeine, which has a stimulating effect, and during roasting, coffee metabolites undergo complex Maillard reactions, producing melanoidins and other degradation products; thus, the results are often difficult to interpret. Nevertheless, herein we found that the trigonelline contained in Sakurajima radish likely has the capacity to directly contribute to improved human vascular endothelial function. Kuroda et al. found that the underlying mechanism for stimulating NO production by trigonelline, an active constituent of Sakurajima radish involves endothelial NO synthase activation by the phosphorylation of Ser1177 and the dephosphorylation of Thr495, which is triggered by elevated concentrations of cytoplasmic Ca^2+^, resulting from the activation of Ca^2+^ channels in VECs \[[@B1-nutrients-12-01872]\]. The molecular mechanisms of vasodilation in human by Sakurajima radish in this study appear consistent with those reported by Kuroda et al.
5. Conclusions {#sec5-nutrients-12-01872}
==============
In this study, we compared the trigonelline contents of various plants, including Sakurajima radish varieties. Importantly, we found that Sakurajima radish contained approximately 60 times more trigonelline than Aokubi radish or squash. Furthermore, fresh coffee cherries, which contained the second highest amount of trigonelline, showed reduced trigonelline contents after roasting, suggesting that the coffee generally consumed by humans may be low in trigonelline. Therefore, Sakurajima radish was identified as the most suitable source of trigonelline. Moreover, although there were large individual variations among varieties of Sakurajima radish, the differences were not significant, suggesting that all types of Sakurajima radish likely contained high amounts of trigonelline. Evaluation of the effects of cooking and processing revealed that trigonelline was relatively resistant to heating at 100 °C for approximately 1 min; however, trigonelline contents in radish decreased rapidly as the elapsed time increased and detected in the infusion. Additionally, trigonelline was decomposed after heating for 10 min at 180 °C or 1 min at 185 °C and 190 °C. In contrast, the compound was quite stable after freezing for up to 11 months. Based on these results, we conducted the first trial of Sakurajima radish consumption in humans. Ingestion of 170 g/day Sakurajima radish for ten days effectively increased blood trigonelline concentrations and significantly improved %FMD, which represents vascular endothelial function. Overall, our findings revealed that trigonelline contained in Sakurajima radish is likely to directly contribute to the improvement of vascular endothelial function in humans. Thus, Sakurajima radish may have applications as a functional food for enhanced vascular endothelial function. Despite the strengths and positive results obtained in this study, we note that the efficacy of Sakurajima radish was only investigated in healthy Japanese volunteers. Thus, further large-scale clinical trials in patients with vascular disease are needed to clearly establish the clinical benefits of Sakurajima radish.
The authors would like to thank Fumio Yagi, Youngil Kim, Yoshihiko Atsuchi, Shoichi Take, and Hitoshi Imuta for advice with the experiments.
Conceptualisation, K.K.; Methodology, T.K., and K.K.; Formal Analysis, M.S., and Y.N.; Investigation, M.S., Y.N., T.K., M.K., and K.K.; Resources, M.K.; Data Curation, T.K. and K.K.; Writing---Original Draft Preparation, M.S.; Writing---Review & Editing, Y.M. and K.K.; Supervision, N.A., D.-C.C., L.R.J., and Y.M.; Project Administration, K.K.; Funding Acquisition, K.K. All authors have read and agreed to the published version of the manuscript.
This work was supported in part by JSPS KAKENHI \[grant number 20H02936\]; the Japan Society for Bioscience, Biotechnology, and Agrochemistry; ROHTO Pharmaceutical Co., Ltd.; and Public Foundation Yonemori-seishinikuseikai. The funding agencies had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. All authors have provided consent for publication.
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
{#nutrients-12-01872-f001}
{#nutrients-12-01872-f002}
######
(**A**) Effects of cooking and processing on trigonelline content. (**B**) Time-dependent changes in trigonelline content in radish (black circles), infusion (white circles), and toral (broken line) after heating in hot water at 100 °C. (**C**) Changes in trigonelline contents after heating in an oven at 180 °C (black circles), 185 °C (white circles), or 190 °C (black triangles). (**D**) Time course of changes in trigonelline contents when stored in a freezer at −30 °C. All values are represented as means ± SD. For each time point, means with a different letter are significantly different, *p* \< 0.05.
{#nutrients-12-01872-f004}
nutrients-12-01872-t001_Table 1
######
Comparison of trigonelline amounts in different products.
Plant Type Relative Value (%)
----------- ------------ --------------------
Radish Sakurajima 100
Aokubi 1.75
Coffee Fruit 81.74
Roast 17.15
Extraction 0.01
Buttercup Seeds 1.50
Squash Fruit 1.00
nutrients-12-01872-t002_Table 2
######
Changes in blood pressure (BP), pulse, and body weight.
Before After *p* Value
-------------- ------ ------------- -------------- -----------
Systolic BP mmHg 120 ± 13.6 115.8 ± 10.8 0.72
Diastolic BP mmHg 70.8 ± 11.0 68.2 ± 10.7 0.66
Pulse /min 73.6 ± 13.7 70.9 ± 11.7 0.07
Body weight kg 58.1 ± 9.4 58.0 ± 9.4 1.0
nutrients-12-01872-t003_Table 3
######
General biochemical examination of blood collected from participants.
Before After *p* Value
------- ----------- ----------------- ----------------- -----------
WBC /μL 5785.7 ± 1287.8 6228.6 ± 1658.1 0.12
Hb g/dL 14.6 ± 1.6 14.5 ± 1.5 0.58
Plt ×10^4^/μL 27.2 ± 5.9 26.1 ± 5.8 0.10
LDL-C mg/dL 108.5 ± 21.1 113.6 ± 29.0 0.33
HDL-C mg/dL 79.5 ± 18.3 77.7 ± 14.9 0.47
TG mg/dL 110.1 ± 123.4 60.8 ± 33.0 0.04
FPG mg/dL 87.5 ± 4.4 90.4 ± 7.1 0.08
UA mg/dL 4.4 ± 1.5 4.9 ± 1.6 0.23
BUN mg/dL 13.4 ± 3.0 12.6 ± 2.8 0.24
Cr mg/dL 0.7 ± 0.2 0.7 ± 0.2 0.71
Na mEq/L 142.0 ± 1.4 140.6 ± 1.3 1.00
K mEq/L 4.2 ± 0.3 4.2 ± 0.2 1.00
Cl mEq/L 101.6 ± 1.4 103.1 ± 1.0 1.00
AST IU/L 20.5 ± 4.7 20.3 ± 4.2 0.89
ALT IU/L 16.1 ± 8.4 16.5 ± 8.1 0.65
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Nonalcoholic fatty liver disease (NAFLD) is characterized by the excessive accumulation of hepatic lipid and hepatic steatosis in the absence of alcoholism \[[@B1]\]. It is emerging as the most common cause of liver diseases, with a global prevalence of 25% approximately. More importantly, NAFLD is considered to be a dangerous complication of nonalcoholic steatohepatitis (NASH), advanced fibrosis, cirrhosis, and hepatocellular carcinoma, with evidence of NAFLD at markedly increased risk of adverse outcomes, including liver-specific morbidity and mortality \[[@B2]\]. Recently, increases in high-fat diet (HFD) intake and thus induced obesity have demonstrated parallelism with a global increase in NAFLD. Furthermore, studies evidence that NAFLD contributed to the characteristics of a series of diseases, such as metabolic syndromes, like type II diabetes mellitus, hypertension, and cardiovascular events, making NAFLD a major public health concern worldwide \[[@B3], [@B4]\].
Although the presence of steatosis is a requisite and typical for NAFLD, the development of NAFLD is a complex process that is affected by numerous mechanisms including genetic, metabolic, lifestyle, and gut microbiome \[[@B5]\]. They lead to increased metabolic substrate (mainly lipids and carbohydrates) delivery to the liver and increased visceral adipose tissue, featured with the excessive accumulation of free fatty acids (FFA), triglycerides (TG), and proinflammatory mediators. These changes alter lipid and glucose metabolism, produce insulin resistance, and create a proinflammatory milieu that induce oxidative stress and modify cell-cell crosstalk, triggering cell injury, apoptosis, or cell death \[[@B6]\]. Variable activation of these processes and the consequent metabolic response determines the development of disease phenotype and its progression to fibrosis and cirrhosis. Lifestyle intervention is recommended for the majority of NAFLD patients, and its benefit is also witnessed by a wide range of clinical studies \[[@B7], [@B8]\]. However, confined to the difficulties of long-term exercise, pharmaceutical treatment aimed at alleviating hepatic steatosis or protecting the liver from additional injury is necessary.
To expand the range of pharmaceutical options for NAFLD treatment, recent studies also focus on identifying active agents or herbal extracts that can ameliorate NAFLD. With a multi-ingredient and multipath pharmacological action, traditional Chinese medicine (TCM) is compatible with the complex pathogenesis of NAFLD, which can be used to mitigate NAFLD and associated diseases \[[@B9]\]. Kangtaizhi granule (KTZG) is a Chinese medicine compound prescription that is composed of eight kinds of TCM, Radix Puerariae, Rhizoma Dioscoreae, Sophora japonica oyster, mulberry leaves, Polygonatum odoratum, mulberry, and papaya. Some of these eight TCM are partly consumed as a part of the human diet with abundant bioactivities. Our previous study demonstrated that KTZG has the efficacy to protect life from NAFLD, with the ability to ameliorate abnormal liver function and lipid metabolism \[[@B10]\]. Although the effect of KTZG in the clinic has been confirmed, the clear mechanisms under this efficacy are rather elusive. In this study, with network pharmacology and experiment validation in high-fat diet rats and FFA-treated HepG2 cells, we aimed to provide basic data for a better understanding of the pharmacological effects and potential mechanisms of KTZG in preventing of NAFLD.
2. Materials and Methods {#sec2}
========================
2.1. KTZG Bioactive Compounds and NAFLD-Related Target Screening {#sec2.1}
----------------------------------------------------------------
For the eight herbs that make up KTZG, the Traditional Chinese Medicine Systems Pharmacology (TCMSP) database, TCM-Mesh database, and Traditional Chinese Medicine Integrative Database (TCMID) database were used to obtain the related chemical information of these eight herbs, including the bioactive compound name; molecular weight; structural formula; absorption, distribution, metabolism, and elimination (ADME) parameters; and CAS ID. Based on the screened compounds in KTZG, compound-related targets were predicted using the SwissTargetPrediction and STITCH online databases. Canonical SMILES of each compound was searched from PubChem and submitted to these two databases, then predicted targets were obtained. The keyword "NAFLD" was inputted in the following databases including the Therapeutic Target Database (TTD), DrugBank, Online Mendelian Inheritance in Man (OMIM), and DisGeNET to screen the disease-related targets. Then, the coexistent targets between compounds and disease were identified as KTZG-related targets for NAFLD treatment.
2.2. Network Construction and Analysis {#sec2.2}
--------------------------------------
Based on the identified compounds and predicted targets of KTZG, the interaction network between compounds and targets was visualized with Cytoscape 3.2.1 software. The protein-protein interactions of the targets were also analyzed with the STRING online tool and visualized with Cytoscape 3.2.1 software. The protein\'s topological attribute was analyzed with the software plugin tool "Network analyzer".
2.3. KEGG Pathway Enrichment Analysis {#sec2.3}
-------------------------------------
For the predicted targets, the underlying biological information was also explored. These target-enriched KEGG pathway networks were analyzed with the Cytoscape 3.2.1 software plugin tool ClueGO; *P* value of the KEGG pathway term less than 0.05 was set up as the criterion.
2.4. High-Performance Liquid Chromatography (HPLC) Analysis of KTZG {#sec2.4}
-------------------------------------------------------------------
HPLC analysis of KTZG was performed on a Waters e2695 HPLC system equipped with a 2998 PDA detector. An Agilent ZORBAX SB-C18 column (4.6 mm × 150 mm, 5 *μ*m) was used for the chromatographic separation, and solvent A (acetonitrile) and solvent B (0.01% glacial acetic acid in distilled water) composed the solvent system. The chromatographic separation was conducted according to the following solvent gradient: 0-30 min, 10%-15% A; 30-60 min, 15%-35% A; and 60-70 min, 35%-90% A. the injection volume was 10 *μ*L, flow rate was 0.8 mL/min, detection wavelength was 254 nm, and column temperature was 30°C \[[@B11]\].
2.5. Animal Diets and Tissue Sample Preparation {#sec2.5}
-----------------------------------------------
4-week male SD rats were purchased from the Shanghai SLAC Laboratory Animal Co. Ltd., China, and maintained under specific pathogen-free conditions at 22\~25°C temperature, 60% humidity, 12 h dark-light cycle, and free access to water and normal diet. After a week of acclimatization, the rats were randomly divided into five groups (*n* = 12 per group): control group: rats were fed with normal diet; NAFLD group: rats were fed with HFD; KTZG 0.75 group: rats were fed HFD and KTZG 0.75 mg/kg/d, i.g.; KTZG 1.5 group: rats were fed with HFD and KTZG 1.5 mg/kg/d, i.g.; and KTZG 3 group: rats were fed with HFD and KTZG 3.0 mg/kg/d, i.g. Rats were fed with normal or HFD for 10 weeks; at the same time, KTZG was administrated from the week of 5 to 10 for a total of 6 weeks. The same volume of saline was given to the control and NAFLD groups. After last administration and overnight fasting, body weight was recorded, and all rats were euthanized using CO~2~. Blood was collected from a cardiac puncture and centrifugated at 3000 rpm for 15 min to obtain the serum samples. Liver tissues were also removed and weighted and then stored immediately at -80°C for further analysis. All experiments were performed with the approval from the Zhejiang Chinese Medical University using guidelines for the care and use of laboratory animals.
2.6. Biochemical Analysis {#sec2.6}
-------------------------
Serum content of total TG, total cholesterol (TC), total bilirubin, high-density lipoprotein (HDL), low-density lipoprotein (LDL), alanine aminotransferase (ALT), and aspartate aminotransferase (AST) was detected using an automatic analyzer (Roche, Basel, Switzerland) according to the manufacturer\'s introduction.
2.7. Liver Histopathological Examination {#sec2.7}
----------------------------------------
The liver tissues were fixed in 10% paraformaldehyde for 48 h and embedded in paraffin, then the tissues were cut into 3\~5 *μ*m sections and stained with hematoxylin and eosin (HE). All the specimens were examined under a light microscope (Olympus, Tokyo, Japan). The tissue sections were also stained with oil red O solution to assess the lipid droplet accumulation in liver tissues. The specimens were examined under a light microscope, and the oil red O-positive areas were quantified using Image-Pro Plus 6.0 software (IPP, Media Cybernetics, Inc., USA).
2.8. Immunohistochemical Assay {#sec2.8}
------------------------------
Liver tissue sections were incubated with the primary antibody at 4°C overnight, then washed with PBS for three times and further incubated with the horseradish peroxidase anti-rabbit secondary antibody for 15 min at room temperature. Then, the sections were washed with PBS for 3 times and stained with DBA solution. After terminating the reaction with water, the sections were further counterstained with hematoxylin; the tissue sections in each group were observed under a microscope.
2.9. Cell Culture and Induction of Hepatic Steatosis {#sec2.9}
----------------------------------------------------
The HepG2 hepatocyte cell line (iCell Bioscience Inc., China) was cultured in Dulbecco\'s Modified Eagle\'s Medium (DMEM, Hyclone) supplemented with 10% FBS and penicillin/streptomycin mixture at 37°C in a 5% CO~2~ atmosphere. For induction of hepatic steatosis, HepG2 cells were incubated with 1 mM of FFA (oleic acid and palmitic acid, 2 : 1) in serum-free medium containing 1% fat-free bovine serum albumin (BSA) to stimulate lipid accumulation for 24 h. Serum samples containing KTZG were obtained from the KTZG 0.75, 1.5, 3 mg/k group normal diet-fed SD rats and named the L-KTZG, M-KTZG, and H-KTZG groups, respectively. Then, the cells were further incubated in DMEM-contained serum (20 *μ*L) containing with KTAG for another 24 h. The control group and FFA group HepG2 cells were incubated in DMEM-contained normal serum (20 *μ*L) without KTZG.
2.10. MTT Assay {#sec2.10}
---------------
HepG2 cells (1 × 10^4^ cells/well) were seeded in a 96-well plate and allowed to adhere to the wells. Then, cells were incubated in serum-free DEME containing 1 mM of FFA, with or without KTZG treatment at different doses for 24 h. Then, MTT reagent (BBI Life Sciences) was added for another coincubation of 1 h. After that, the optic density (OD) value was measured at 590 nm using a CMax Plus ELISA reader (MD, USA).
2.11. Intracellular TG Level Assay {#sec2.11}
----------------------------------
HepG2 cells (1 × 10^4^ cells/well) were seeded in 6-well plates and allowed to adhere to the wells. Then, cells were incubated in DMEM containing 1 mM of FFA, with KTZG treatment at different doses for 24 h. Then, intracellular TG levels were detected with the TG assay kit (Cayman Chemical, USA) according to the manufacturer\'s instructions.
2.12. Oil Red O Staining {#sec2.12}
------------------------
Oil red O staining was also performed to evaluate the intracellular lipid accumulation. Briefly, the prepared HepG2 cells were fixed with 10% formalin in PBS for 1 h and then stained with oil red O solution for 30 min at room temperature. Then, cells were washed with distilled water, and then the stained lipid droplets within cells were observed under a light microscope. The lipid accumulation was quantified through the dissolving of stained lipid droplets with isopropanol, and the absorbance was measured at 490 nm. Control cells were treated with 1% BSA only.
2.13. Glucose Uptake Assay {#sec2.13}
--------------------------
Glucose uptake was detected using a fluorescence assay. HepG2 cells (1 × 10^4^ cells/well) were seeded in a 6-well plate and treated as described above. Then, the medium was removed, and 100 *μ*mol/Lof 2-NBDG was added and coincubated in serum-free DMEM medium at 37°C for 30 min. Then, the relative fluorescence images were observed under a Ts2-FC fluorescence microscope (Nikon, Japan).
2.14. Immunofluorescence Staining {#sec2.14}
---------------------------------
HepG2 cells were prepared as described above and fixed with 10% formalin in PBS for 1 h. Then, the fixed cells were washed with PBS for three times, permeabilized in 0.5% Triton X-100, and then blocked with 5% normal goat serum. After incubation with primary p-AKT (ab23875, 1 : 250, Abcam, USA), p-mTOR (ab131538, 1 : 250, Abcam, USA), and secondary antibody IgG H&L (Alexa Fluor® 594) (ab150076, 1 : 250, Abcam, USA), cell nuclei were stained with DAPI (Beyotime Institute of Biotechnology, China). Then, the relative fluorescence images were obtained using a Ts2-FC fluorescence microscope (Nikon, Japan).
2.15. Western Blot Assay of Liver Tissues and HepG2 Cells {#sec2.15}
---------------------------------------------------------
Liver tissues or HepG2 cells were lysed with RIPA buffer (Beyotime Institute of Biotechnology, China), and the protein concentrations were measured by the BCA method using a BCA protein assay kit (Beyotime Institute of Biotechnology, China). Then, equal amounts of protein samples (30 *μ*g) were loaded onto 10% SDS-polyacrylamide gel and electrophoretically transferred to polyvinylidene difluoride (PVDF) membranes. The membranes were blocked with the blocking solution and incubated with the primary antibodies against PPAR-*γ*, ab59256, 1 : 1000; SREBP-1, ab28481, 1 : 5000; p-AKT, ab8805, 1 : 500; FAS, ab82419, 1 : 1000; SIRT1, ab12193, 1 : 2000; p-AMPK, ab23875, 1 : 1000; AMPK, ab80039,1 : 1000; p-mTOR, ab131538, 1 : 1000; mTOR, ab32028, 1 : 5000; and GLUT2, ab192599, 1 : 5000 (Abcam, USA) overnight at 4°C. After that, the membranes were washed with TBST for three times and further incubated with horseradish peroxidase- (HRP-) conjugated secondary antibodies (ab20272, 1 : 5000). The blots were visualized using Millipore\'s enhanced chemiluminescence and quantified using Image-Pro Plus 6.0 software (IPP, Media Cybernetics, Inc., USA).
2.16. Statistical Analysis {#sec2.16}
--------------------------
Data were expressed as mean ± SD and analyzed with SPSS software; statistical significance was determined by Student\'s *t*-test for comparisons between two groups. Differences were considered to be significant when *P* \< 0.05.
3. Results {#sec3}
==========
3.1. TKZG-Compound-Target Network {#sec3.1}
---------------------------------
TKZG was prepared by eight TCM. Based on TCM-related databases, the chemical constituents in these eight TCM were screened ([Figure 1(a)](#fig1){ref-type="fig"}). In addition, 1362 proteins that target these compounds were predicted, and 152 protein targets related to NAFLD were screened. As a result, 79 coexisted targets were identified as TKZG\'s targets for NAFLD treatment ([Figure 1(b)](#fig1){ref-type="fig"}). The compound-target network is also shown in [Figure 2(a)](#fig2){ref-type="fig"}, including 42 compounds and 79 proteins, forming 296 edges.
3.2. Hub Target Identification {#sec3.2}
------------------------------
The PPI network was constructed to analyze the interactions of these identified targets ([Figure 2(b)](#fig2){ref-type="fig"}). 77 targets were mapped into the network with 1015 interaction edges, forming a complex interaction network. The node size was positively correlated to the node degree; some proteins were specified with a high degree, including PPARG (degree = 54), ADIPOQ (degree = 54), AKT1 (degree = 54), TNF (degree = 49), SREBF1 (degree = 48), and SIRT1 (degree = 46).
3.3. KEGG Pathway Analysis {#sec3.3}
--------------------------
79 target-enriched KEGG pathways were also analyzed; the KEGG pathway network is also presented in [Figure 2(c)](#fig2){ref-type="fig"}. The network was composed of 84 terms of pathways, forming a tight pathway interaction network. Some pathways that the identified targets enriched were directly related to the disease NAFLD, including the NAFLD pathway, type II diabetes mellitus, insulin resistance, insulin signaling pathway, glucagon signaling pathway, regulation of lipolysis in adipocytes, and other signaling pathways like the AMPK and mTOR signaling pathways. These signaling pathways directly interacted with NAFLD-associated pathways and marked with similar color cluster.
3.4. HPLC Chemical Fingerprint of KTZG {#sec3.4}
--------------------------------------
Ten batches of KTZG were prepared, and the typical chromatograms from these 10 batches with good quality control are shown in [Figure 3(a)](#fig3){ref-type="fig"} by HPLC analysis. Thirteen peaks were identified on the HPLC chemical fingerprint of KTZG ([Figure 3(b)](#fig3){ref-type="fig"}). Among these peaks, six peaks, including 3′-hydroxy puerarin (no. 1), puerarin (no. 2), daidzin (no. 6), rutin (no. 7), daidzein (no. 12), and quercetin (no. 13), were identified by referring to the corresponding standards.
3.5. KTZG Improved Liver Organ Index and Serum Lipid Profile {#sec3.5}
------------------------------------------------------------
In the NAFLD model rat with HFD, the liver organ index was significantly increased compared to that in the normally fed control group rat (*P* \< 0.05) ([Figure 4(a)](#fig4){ref-type="fig"}). But for model rats with KTZG doses reaching 1.5 g/kg and 3 g/kg, the organ index was significantly decreased (*P* \< 0.05 and *P* \< 0.01, respectively). Compared to the control group, serum contents including TG, TC, and LDL-c in the model group were significantly increased (*P* \< 0.05) ([Figure 4(b)](#fig4){ref-type="fig"}). But KTZG treatment inhibited these increased lipid profiles in the serum, especially in the 1.5 g/kg and 3 g/kg dosage groups with statistical significance. For HDL-c, the levels in these groups were correspondingly opposite. In addition, serum AST and serum ALT in the NAFLD group were significantly higher than those in the control group (*P* \< 0.01), and the serum AST and ALT levels in the KTZG group were significantly decreased than those in the NAFLD group (*P* \< 0.05).
3.6. KTZG Alleviated Hepatic Steatosis and Liver Lipid Accumulation in NAFLD Rats {#sec3.6}
---------------------------------------------------------------------------------
HE staining showed that compared to the control group with normal liver morphology, NAFLD group liver tissues showed significantly hepatic steatosis, vacuolation, hepatocellular hypertrophy, and lipid droplet accumulation ([Figure 4(c)](#fig4){ref-type="fig"}), whereas the degree of hepatic steatosis and the size of the lipid droplets were alleviated in the liver of the rats with KTZG treatment. Oil red O results also evidenced that there were significant lipid droplets and lipid accumulation in liver tissues of NAFLD rats, and KTZG treatment reduced the HFD-induced lipid accumulation in liver tissues ([Figure 4(d)](#fig4){ref-type="fig"}).
3.7. Effect of KTZG on Expression of Lipid-Related Genes and AMPK/mTOR Signaling in Liver Tissues {#sec3.7}
-------------------------------------------------------------------------------------------------
Expressions of the lipid-related genes identified from network pharmacology and the proteins in AMPK/mTOR signaling were detected. Western blot results in [Figure 5(a)](#fig5){ref-type="fig"} showed that the protein expressions of PPAR-*γ* were significantly decreased, and SREBP-1 and p-AKT were significantly increased in the NAFLD group compared to the control group (*P* \< 0.01). KTZG treatment significantly increased the expression of PPAR-*γ* in the 1.5 and 3 g/kg groups (*P* \< 0.05) and decreased the expression of SREBP-1 in the 3.0 g/kg group (*P* \< 0.05) and p-AKT in the 0.75, 1.5, and 3 g/kg groups (*P* \< 0.05, *P* \< 0.05, and *P* \< 0.01, respectively). Compared to the control group, the p-AMPK expression was significantly decreased in the NAFLD group, and KTZG treatment increased the expression significantly in the 3 g/kg group (*P* \< 0.05); p-mTOR expression was significantly increased in the NAFLD group, and KTZG treatment increased the expression significantly in the 1.5 g/kg and 3 g/kg groups (*P* \< 0.05 and *P* \< 0.01, respectively) ([Figure 5(b)](#fig5){ref-type="fig"}). Immunohistochemical results also showed that compared to the control group, the expression of FAS was increased in the NFALD group but increased with KTZG treatment and the Sirt1 expression was decreased in the NFALD group but increased in the KTZG groups ([Figure 5(c)](#fig5){ref-type="fig"}).
3.8. Cell Viability of HepG2 Cells Treated with KTZG {#sec3.8}
----------------------------------------------------
The MTT assay was used to evaluate the effects of KTZG in different concentrations on HepG2 cell viability. As the result indicated, KTZG treatment at the tested concentrations did not exhibit significant toxicity in cell viability ([Figure 6(a)](#fig6){ref-type="fig"}).
3.9. KTZG Decreased Intracellular Lipid Accumulation in FFA-Induced HepG2 Cells {#sec3.9}
-------------------------------------------------------------------------------
Significantly high intracellular TG level was induced by an exposure to FFA 1 mM for 24 h of HepG2 cells ([Figure 6(b)](#fig6){ref-type="fig"}), whereas in KTZG-treated cells, TG accumulation showed a significant decrease compared to that in FFA-only treated cells (*P* \< 0.01). Oil red O staining showed that there was significant lipid droplet accumulation in FFA-treated HepG2 cells (*P* \< 0.01), but for the middle and high concentrations of KTZG along with FFA treatment, the lipid accumulation was significantly decreased compared to those for the FFA-only treated cells (*P* \< 0.05 and *P* \< 0.01, respectively) ([Figure 6(c)](#fig6){ref-type="fig"}).
3.10. KTZG Attenuated Glucose Uptake in FFA-Induced HepG2 Cells {#sec3.10}
---------------------------------------------------------------
To determine the effect of KTZG on glucose uptake in HepG2 cells, the 2-NBDG glucose uptake assay was performed in HepG2 cells ([Figure 6(d)](#fig6){ref-type="fig"}). The result showed that the intensity of the glucose-labeled fluorescence was significantly decreased in response to the FFA treatment (*P* \< 0.01). However, HepG2 cells treated with KTZG improved the glucose uptake, with higher fluorescence intensity, compared to cells treated with FFA only. Protein expression of GLUT2 also showed that the expression of GLUT2 was decreased in FFA-treated cells but substantially increased in response to KTZG.
3.11. Effect of KTZG on Expression of Lipid-Related Genes in FFA-Induced HepG2 Cells {#sec3.11}
------------------------------------------------------------------------------------
Protein expressions of the lipid-related genes PPAR-*γ*, SREBP-1, p-AKT, FAS, and SIRT1 were detected in FFA-induced HepG2 cells. Western blot results in [Figure 7(a)](#fig7){ref-type="fig"} showed that the protein expressions of PPAR-*γ* were significantly decreased and SREBP-1, p-AKT, FAS, and SIRT1 were significantly increased in FFA-induced HepG2 cells compared to untreated cells (*P* \< 0.01). In KTZG-treated HepG2 cells, significantly increased expressions of PPAR-*γ* were observed in KTZG low- to high-treated groups (*P* \< 0.05); decreased expressions of SREBP-1, p-AKT, and FAS were observed in the KTZG high-treated group (*P* \< 0.05).
3.12. KTZG Ameliorates Lipid Accumulation and Hepatic Steatosis in FFA-Induced HepG2 Cells via the Regulation of AMPK/mTOR-Dependent Signaling {#sec3.12}
----------------------------------------------------------------------------------------------------------------------------------------------
Compared to the control group, the p-AMPK expression was significantly decreased in FFA-induced HepG2 cells (*P* \< 0.01), and KTZG treatment increased the expression significantly in the H-KTZG group (*P* \< 0.01); the p-mTOR expression was significantly increased in FFA-induced HepG2 cells (*P* \< 0.01), and KTZG treatment increased the expression significantly in the M-KTZG and H-KTZG groups (*P* \< 0.01) ([Figure 7(b)](#fig7){ref-type="fig"}). Immunofluorescence staining also showed that compared to the control group HepG2 cells, p-AMPK expression was decreased in FFA-induced HepG2 cells significantly (*P* \< 0.01) but increased in the KTZG-treated groups significantly (*P* \< 0.05 or *P* \< 0.01) ([Figure 8(a)](#fig8){ref-type="fig"}). p-mTOR was increased in FFA-induced HepG2 cells (*P* \< 0.01) but suppressed with KTZG treatment (*P* \< 0.05 or *P* \< 0.01) ([Figure 8(b)](#fig8){ref-type="fig"}). FFA-induced HepG2 cells were further treated with AMPK inhibitor compound C (10 *μ*L) or mTOR inhibitor rapamycin (10 *μ*L) for 24 h. Western blot results showed that compared to FFA+compound C group cells, the protein expressions of p-AMPK in FFA+KTZG+compound C group cells were significantly increased (*P* \< 0.05), and the protein expressions of p-mTOR in FFA+KTZG+compound C group cells were significantly suppressed (*P* \< 0.05) ([Figure 9](#fig9){ref-type="fig"}).
4. Discussion {#sec4}
=============
NAFLD is defined as excessive hepatic lipid accumulation in the form of lipid droplets, also known as hepatic steatosis, in the liver with minor or no alcohol intake. It encompasses a broad spectrum of hepatic damage stages such as isolated hepatic steatosis, NASH, liver fibrosis, and cirrhosis or may even progress to hepatocellular carcinoma \[[@B12]\]. In the current study, we explored the effect and underlying mechanism of KTZG on NAFLD with *in vivo* and *in vitro* experiments. As the results indicated, in HFD-fed rats, KTZG treatment significantly improved liver organ index and serum lipid profiles and alleviated hepatic steatosis and liver lipid accumulation; in FFA-treated HepG2 cells, KTZG treatment significantly decreased intracellular TG levels, lipid accumulation, and attenuated glucose uptake. In addition, network pharmacology and experiment study showed that this protective effect of KTZG from hepatic steatosis might be associated with its ability to regulate the AMPK/mTOR signaling pathway.
Excess FFA are converted to TG and stored in hepatocytes, and excessive accumulation of TG in hepatocytes is the pathologic hallmark of NAFLD. The treatment strategies are also generally aimed at reducing TG accumulation. Other lipid profiles, including TC and LDL-C, are also accumulated in hepatocytes with NAFLD. TG/HDL-C and TC/HDL-C ratios are associated with the severity of NAFLD, and the patients with higher lipid ratios had a significantly greater risk for advanced NAFLD \[[@B13], [@B14]\]. In this study, aberrant lipid profiles were emerged in SD rats fed with HFD; the serum TG, TC, and LDL-C levels were significantly increased; and TG/HDL-C and TC/HDL-C ratios were higher than normal rats; TG levels in HepG2 cells treated with FFA were also increased significantly. In addition, lipid droplet accumulation and liver pathologic damage-like hepatic steatosis were observed in the liver tissues and HepG2 cells with oil red O staining. KTZG treatment restored these altered lipid profiles in HFD rats and HepG2 cells; with the TG, TC, and LDL-C levels decreased, lipid accumulation and hepatic steatosis were also ameliorated significantly.
Although studies focused on this Chinese medicine compound prescription is limited, there are still some studies that could be found for the eight TCM of KTZG against NAFLD. In HepG2 cells, the water extract of *Puerariae radix* treatment could attenuate the hepatic lipoprotein production and secretion; the intracellular total and free TC concentrations also decreased \[[@B15]\]. Metabolomic analysis also revealed that by improving metabolism disorders and inhibiting oxidative damage, the total isoflavones from *Radix Puerariae* could exhibit the therapeutic potential in diabetic rats \[[@B16]\]. In high-fat diet fed with Rhizoma Dioscoreae Tokoronis extracts, mice were found to have lower increases in body and epididymal adipose tissue weights, a lessened occurrence of hepatic steatosis, and a significant decrease in TG, TC, and LDL-C than mice that were fed with high-fat diet only \[[@B17]\]. The mulberry and mulberry leaves are used in TCM as the remedy for obesity, hyperlipidemia, and metabolic disorders \[[@B18]\]. Mulberry leaf extracts could attenuate dyslipidemia and lipid accumulation in obesity-related NAFLD mice, via downregulating the lipogenesis enzymes, inflammation, and oxidative stress while upregulating the lipolysis markers \[[@B19]\]. Mulberry fruit extract could ameliorate lipid accumulation and inhibit the increased levels of TC, TG, and LDL-C but restore the level of HDL-C in HFD-fed rats and protect liver tissue against NAFLD damage through the inhibition of mitochondrial oxidative stress \[[@B20]\].
Network pharmacology and HPLC analysis also clarified the major chemical components of KTZG. Some chemicals, including puerarin, rutin, daidzein, and quercetin, were identified from KTZG using both network pharmacology and HPLC analysis. Wang et al. \[[@B21]\] found that puerarin mediated activation of the PARP-1/PI3K/AKT pathway, and further improvement in fatty acid metabolism could ameliorate NAFLD in C57BL/6J mice fed with a high-fat high-sucrose diet. In addition, the expressions of genes involved in hepatic fatty acid synthesis including PPAR-*α*, SREBP-1, and FAS were also attenuated with puerarin coadministration. In both fat-challenged murine liver tissues and HepG2 cells, rutin treatment was shown to significantly lower TC content and the abundance of lipid droplets and also was able to restore the expression of lipid-related genes, PPAR-*α*, and downstream targets, CPT-1 and CPT-2, while suppressing those of SREBP-1c, DGAT-1, and DGAT-2, as well as ACC \[[@B22]\]. Daidzein was reported to act like a PPAR-*γ* activator, stimulating the adipogenic differentiation in 3T3-L1 adipocytes, and in obese mice, daidzein could inhibit hypertrophy in fat cell size and improved insulin sensitivity, concomitant with upregulation of PPAR-*γ* in fat tissue \[[@B23]\]. Results of *in vivo* and *in vitro* T2DM-induced NAFLD and quercetin treatment models revealed that quercetin could alleviate the serum transaminase and IL-1*β*, IL-6, and TNF-*α* levels; recover oxidative stress; and markedly reduce T2DM-induced histological alterations and lipid accumulation of the livers, accompanied by the restoration of the increased serum total bile acid and the decreased liver total bile acid \[[@B24]\]. These studies indicated that the potential therapeutic effect of NAFLD against HFD-induced lipid accumulation and liver steatosis of NAFLD may owe to these identified chemicals from KTZG.
Based on the identified chemicals in KTZG, some lipid-related genes targeted to these chemicals, including PPAR-*γ*, SREBP-1, FAS, and SIRT1, were also identified with network pharmacology. PPAR-*γ* is critically implicated in the metabolic regulation of lipid and lipoprotein levels, such as TG, and blood glucose. Some drugs like fibrates and glitazones, with the ability to deduce TG levels in NAFLD treatment, are PPAR agonists; they act through the activation of nuclear receptors of the PPAR family, thereby regulating genes involved in TG metabolism \[[@B25], [@B26]\]. FAS is an important enzyme for fatty acids in controlling lipid synthesis, and SREBP-1 could regulate the expression of FAS to increase lipid synthesis in the liver or adipose tissues \[[@B27]\]. Moderate Sirt1 overexpression protects mice from developing NAFLD, and sirt1-deficient mice witness increased body weight and triggered hepatic steatosis \[[@B28]\]. In HFD-fed rats and FFA-induced HepG2 cells, except for the alleviation effect of KTZG on lipid accumulation and hepatic steatosis, our results showed that KTZG could regulate the expression of these lipid-related genes, upregulating PPAR-*γ* and Sirt and downregulating FAS and SREBP-1. These indicated that KTZG exhibit the therapeutic potential against NAFLD which might be via the active chemicals targeting these lipid-related genes, inhibiting the activity of transcription factors for lipid synthesis, thus inhibiting lipid accumulation. The puerarin identified from KTZG was reported to decrease the expression of lipogenic enzymes, FAS and SREBPs, and activate the AMPK signaling pathway to exert a regulatory effect on lipid accumulation \[[@B29]\]. A study focused on rutin found that in both HFD-challenged mouse liver tissues and HepG2 cells, rutin treatment significantly lowered TG content and the abundance of lipid droplets; furthermore, rutin treatment was able to restore the expression of PPAR-*α* and AMPK, while suppressing the expression of lipid-related genes, SREBP-1 and FAS \[[@B22], [@B30]\]. Daidzein and quercetin are also reported to have the potential to alleviate hepatic steatosis and lower lipid accumulation; daidzein promotes the expression of fatty acid oxidation- and oxidative phosphorylation-related genes via an ERR*α* pathway to decrease lipid accumulation \[[@B24], [@B31]\].
The AMPK/mTOR signaling pathway is a master regulator of metabolism and critically involved in metabolic diseases. Activation of AMPK could inhibit the synthesis of fats and protect against diet-induced NAFLD; mTOR is an important signal molecule downstream of AMPK, which also plays a central role in autophagy \[[@B32]\]. In a genetically engineered mouse model, liver-specific AMPK activation reprograms lipid metabolism, reduces liver steatosis, decreases expression of inflammation and fibrosis genes, and leads to significant therapeutic benefits in the context of diet-induced obesity \[[@B33]\]. Other activators of AMPK also have the capability of inhibiting lipid and TC synthesis pathways, lowering hepatic and systemic lipid and TC levels, and these effects require AMPK activity in the hepatocytes \[[@B34]\]. The protective effect of the AMPK/mTOR signaling pathway against NAFLD is also associated with the regulation of autophagy in liver cells. In this study, the expression of p-AMPK was significantly decreased in HFD-fed rats and FFA-induced HepG2 cells; p-mTOR was significantly increased in HFD-fed rats and FFA-induced HepG2 cells. KTZG treatment activated the expression of p-AMPK and suppressed the expression of p-mTOR in liver tissues and hepatocytes, indicating that the AMPK/mTOR signaling pathway may participate in the protective mechanisms of KTZG against NAFLD. Furthermore, compound C, an AMPK inhibitor, and rapamycin, a mTOR inhibitor, were further used to confirm the role of the AMPK/mTOR signaling pathway in KTZG for NAFLD treatment.
5. Conclusion {#sec5}
=============
In conclusion, the present study demonstrated that KYZG had a protective effect against NAFLD; its treatment significantly improved the liver organ index and serum lipid profiles and alleviated hepatic steatosis and liver lipid accumulation in HFD diet-fed rats; in FFA-treated HepG2 cells, KTZG treatment also decreased intracellular TG levels and lipid accumulation and attenuated glucose uptake significantly. In addition, these protective mechanisms of KTZG might be associated with its ability to regulate the AMPK/mTOR signaling pathway.
Data Availability
=================
The data used to support the findings of this study are included within the article.
Conflicts of Interest
=====================
The authors declare that there is no conflict of interest regarding the publication of this article.
Authors\' Contributions
=======================
Zhang JX and Ye XM designed the study, completed the data collection, and contributed to writing of the manuscript; Zhang JX, Du HX, and Shen ML performed the experiment and revised the manuscript; and Zhang JX, Zhao ZQ, and Ye XM analyzed the data and drafted all figures. All authors read and approved the final manuscript.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
{#fig5}
{#fig6}
{#fig7}
{#fig8}
{#fig9}
[^1]: Guest Editor: Kai Wang
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
There is a growing appreciation for the neurodevelopmental underpinnings of many psychiatric disorders. While the importance of the growth and structuring of the brain has long been recognized for disorders that begin in childhood such as autism, language disorders or mental retardation, there is consensus building that adult-onset mental disorders also have origins early in neurodevelopment. In order to expand our understanding of these origins, the National Institute of Mental Health (NIMH) is investing in studies of developmental neurobiology, choosing one of its strategies to be the support of "research to improve our basic understanding of the development, structure, and function of neural circuits, with a focus on those most relevant to mental disorders" (NAMHC, [@B110]). Forwarding this agenda requires a better understanding of the neurobiology of neural stem cells (NSCs) and the factors that regulate them in the developing brain.
Neural stem cells can be defined as self-renewing, multipotent cells that are present in both the embryonic and adult brain. Several recent research findings demonstrate that psychiatric illness may begin with abnormal specification, growth, expansion and differentiation of embryonic NSCs. For example, candidate susceptibility genes for schizophrenia, autism and major depression include those specifying the signaling molecule Disrupted In Schizophrenia-1 (DISC-1), the homeodomain gene engrailed-2 (EN-2), and several receptor tyrosine kinases (RTKs), including brain-derived neurotrophic factor (BDNF) and fibroblast growth factors (FGF), all of which have been shown to play important roles in NSCs or neuronal precursors. Hypotheses about how some of these factors relate to psychiatric disorders are now the focus of much neurodevelopmental work (see several reviews: Arnold et al., [@B2]; Buckley et al., [@B14]; Eisch et al., [@B42]; Thomas and Peterson, [@B152]; Jaaro-Peled et al., [@B72]; Vaccarino et al., [@B159]).
This review will lay out some aspects of the new frontier represented by stem cells for understanding the origins of mental illness. In section I, we will review embryonic central nervous system (CNS) development, focusing on forebrain NSCs and the role of the RTKs, such as FGF receptors (FGFRs), in the earliest developmental processes. RTKs are transmembrane proteins that, upon binding a ligand presented or secreted by nearby cells, phosphorylate intracellular signaling molecules and transduce cell-to-cell signaling in the brain. FGF receptors (FGFRs) are among the earliest RTKs expressed in the developing brain, but also continue to be expressed in the mature brain.
Following this review of NSCs and FGFs in embryonic development, in section II we will describe neurobiological findings in clinical populations that support disrupted embryonic stem cell activity. In section III, adult NSC functioning and connections with mental illness will be reviewed. In section IV, we will discuss how new research with inducible pluripotent stem cells derived from patients with neuropsychiatric disorders may further our understanding of developmental aspect of psychopathology and reveal potential targets for psychiatric treatment.
Embryonic CNS Development
=========================
An introduction to neurogenic stem cells (NSCs)
-----------------------------------------------
Perturbations in early development of the CNS may increase an individual\'s susceptibility to neuropsychiatric disorders. These perturbations may occur at the earliest stages of development, when the primordium of the brain is a sheet of neuroepithelial stem cells. The functioning of NSCs is tightly regulated by both intrinsic and extrinsic factors (Johnson et al., [@B73]). Intrinsic factors that shape the development of NSCs include transcription factors, non-coding RNAs and covalent modifications of chromatin in the nucleus (epigenetic modifications). All of these factors, a description of which is outside the scope of this review, can influence stem cells and their proliferation, or the migration and terminal differentiation of their progeny into defined neural cell types with specific connectivity (Figure [1](#F1){ref-type="fig"}).
{#F1}
The specific "mix" of intrinsic factors that characterizes distinct neural stem cell function and identity acts in conjunction with a local "mix" of extracellular signaling molecules bathing NSCs. There are four major classes of secreted extracellular signaling molecules that are expressed in the developing brain during embryogenesis and that participate in the patterning of the nervous system -- FGFs, WNTs, Sonic Hedgehog (SHH) and Bone Morphogenetic Proteins (BMP). FGFs diffuse from the anterior neural ridge, a region corresponding later in development to the commissural plate, which is the foremost rostral boundary of the telencephalon; WNTs and BMPs emanate from the cortical hem, comprising the medial margin of each hemisphere; BMPs are secreted from the roof plate, the dorsal region in between the cerebral hemispheres; and SHH diffuses from the ventral portion of the neural tube or floor plate (Figure [2](#F2){ref-type="fig"}). In addition to these long-range signals, the balance between Notch ligands and Notch receptors, which are membrane bound, strongly influences neural stem cell fate (Johnson et al., [@B73]). Extracellular signals shape CNS morphogenesis and regulate cell fate by influencing the specific "mix" of intrinsic factors present at specific locations and times in the developing CNS (Figure [1](#F1){ref-type="fig"}).
{#F2}
One mechanism by which changes in embryonic NSCs could lead to behavioral symptoms include an imbalance between production of specific types of excitatory and inhibitory neurons, resulting in abnormal levels of activation in cortical circuits (Rubenstein and Merzenich, [@B135]). A second mechanism may involve impairments in the relative size of cortical areas receiving specific thalamic inputs or sending projections to subcortical stations that play an important role in emotional/behavioral regulation (Figure [1](#F1){ref-type="fig"}).
WNT, SHH, and BMP decrease in expression with age, indicating that their primary function is in establishing early identities of NSCs. Continued expression of WNT and SHH in the adult NSC niches regulates stem cell proliferation (Lai et al., [@B84]; Lie et al., [@B90]; Palma et al., [@B118]). In contrast, FGFs continue to be widely expressed and play a role not only in adult NSC niches (Zheng et al., [@B175]) but also in the maturation of the postnatal cerebral cortex.
FGF ligands, stem cell amplification and cortical neurogenesis
--------------------------------------------------------------
Fibroblast growth factor ligands are peptides that act both intracellularly and through secretion into the extracellular space. There are 22 known FGFs which act upon the four membrane bound FGFRs. Amongst the FGF ligands, 13 are known to be expressed in the CNS during embryonic development (Fgf1,2, 3,7,8, 9,10,13,15,16,17,18,22) in specific regions of the neuroepithelium (Figure [2](#F2){ref-type="fig"}). Three of the receptors, FGFR1, FGFR2 and FGFR3 are present in the embryonic brain. Indeed, FGFRs are among the earliest RTKs expressed in brain development.
Two FGF ligand molecules must bind a receptor dimer in order to cause receptor activation. FGF receptors, akin to other members of the RTK family of proteins, cross-phosphorylate their partner upon ligand binding, triggering the activation of three main intracellular pathways, the Ras/MAP Kinase, PI3 kinase, and PLCγ/Protein Kinase C (Schlessinger, [@B139]). The cascades eventually impinge upon the transcriptional machinery in the cell nucleus. Although RAS/MAPK and PI3K pathways are known to be important mediators of FGF signaling in the developing CNS, the relative role of each of these signaling pathways and of the other putative nuclear functions of FGF signaling for transcriptional regulation in stem/progenitor cells and biological functions are still unclear.
Concurrently with patterning in the developing dorsal telencephalon, NSCs expand in number. Through a developmental switch not yet fully understood, after the majority of this expansion has occurred, stem cells then begin to generate neuronal precursors in a neurogenic phase that lasts for approximately 6 days in rodents and 10--12 weeks in primates (Caviness et al., [@B22]; Rakic, [@B132]) (Figure [1](#F1){ref-type="fig"}). Cortical excitatory neurons are derived from NSC that line the dorsal telencephalic ventricle. The primary stem cells in this ventricular zone (VZ) are called radial glia because of their expression of glial markers such as GFAP and GLAST, and their cellular morphology. Radial glial cells have an apical end foot attachment at the ventricle, a cell body that is near the ventricle, and a long radial process that is attached at the pial surface (Levitt et al., [@B89]). Radial glia can undergo self-renewing cell divisions, or asymmetric cell divisions that directly give rise to neurons (Noctor et al., [@B112]). Another product of radial glial division are committed neurogenic progenitors that migrate to the subventricular zone (SVZ), above the VZ, where they in turn proliferate to give rise to neurons. The committed neuronal progenitors of the SVZ, referred to as intermediate progenitor cells (IPCs) express the transcription factor TBR2 and lack the self-renewal properties of true stem cells (Pontious et al., [@B124]). However, their proliferation is important for the expansion of cortical layers, as demonstrated by the decrease in cortical surface area and thickness in mice lacking *tbr2* (Arnold et al., [@B3]; Sessa et al., [@B141]).
Fibroblast growth factor signaling is important for the regulation of neurogenesis in the developing cortex. Studies *in vitro* originally suggested that the ability of a cortical NSC to stop self-renewing and begin the differentiation process was somehow delayed by increased FGF signaling, resulting in an expanded stem cell pool (Kilpatrick and Bartlett, [@B78]). The first *in vivo* demonstration was provided by injection of an FGF ligand, FGF2, in rat embryonic brain ventricles, which resulted in an expanded cortex with increased excitatory neuron production (Vaccarino et al., [@B160]). Conversely, the deletion of the *fgf2* gene resulted in a cortex with reduced numbers of glutamatergic excitatory neurons, particularly in the anterior neocortex (Raballo et al., [@B125]; Korada et al., [@B81]). This was not due to a change in the cell cycle or by alterations in cell survival, suggesting that FGF signaling might affect the early amplification of stem cells or their immediate descendants (Raballo et al., [@B125]). This was confirmed by later work performed on FGF receptor knockout mice (see FGFs and the Developing Dorsal Telencephalon).
Similarly, mice with reduced *fgf8* gene expression have decreased proliferation and increased numbers of apoptotic cells in the developing telencephalon (Fukuchi-Shimogori and Grove, [@B48]; Garel et al., [@B50]; Storm et al., [@B149]). However, reducing the gene dosages of *fgf15* has opposite effects (Borello et al., [@B12]) with *fgf15* expression in the telencephalon promoting cell differentiation, inhibiting proliferation, and promoting the expression of the *coup-tf1* transcription factor, which plays a role in the development of layer four neurons and posterior cortex (Gimeno et al., [@B53]; Borello et al., [@B12]). Therefore, the combination of different FGFs and other cell extrinsic signaling proteins expressed in the neurogenic period may regulate the behavior of stem cells and the production of neuroblasts in a precise sequence, resulting in the establishment of a cortex with the correct number of neurons.
FGFs and the developing dorsal telencephalon
--------------------------------------------
An essential feature of secreted morphogens during development is their ability to regulate the neurogenic process in a spatially restricted manner. Two distinct areas of the dorsal telencephalon are of relevance to the pathophysiology of neuropsychiatric disorders, the cerebral cortex and the hippocampus, the first developing primarily prenatally, and the second both prenatally and postnatally. In this section we will focus more on the prenatal role of FGF receptors on cortical development. Later sections will discuss the role of FGF on hippocampal development and relate these phenomena to neuropsychiatric disorders.
FGF8, FGF17 and FGF18 are expressed in the anterior telencephalon (Figure [2](#F2){ref-type="fig"}), and evidence suggests that this high rostral and low caudal gradient of FGF signaling contributes to prefrontal cortex (PFC) development (Fukuchi-Shimogori and Grove, [@B48]; Cholfin and Rubenstein, [@B28]). A given amount of FGFs in the developing cortical field can determine whether stem cells will form specific sub-regions such as somatosensory cortex or PFC (Fukuchi-Shimogori and Grove, [@B48]). Alterations in the level of FGFs have been shown to lead to abnormal behaviors in animals (Scearce-Levie et al., [@B138]). FGF8 is negatively regulated by BMP (Crossley et al., [@B36]; Ohkubo et al., [@B114]) and in turn, FGF8 restricts WNT3a expression, confining it into the hem region (Shimogori et al., [@B143]). WNT3a expression in posterior hem regions is important for the formation of the hippocampus (Lee et al., [@B87]).
If we consider the secreted signaling molecules expressed by NSCs in the cortical primordium, only disruptions in the FGF pathways have thus far resulted in major defects in cortical development (Hebert et al., [@B64]; Cheng et al., [@B26]). Although overexpression of a stabilized β-catenin, which is downstream of the WNT pathway, increases cortical size (Chenn and Walsh, [@B27]), the disruption of the WNT pathway affects the hippocampus rather than the cortex (Galceran et al., [@B49]). Thus, normal embryonic cortical development primarily depends on FGF signaling, and in order to better understand the mechanisms mediating these roles, several FGF receptor mutant models have been generated.
Mice that transiently overexpress a dominant negative *fgfr1* gene that interferes with normal functioning of all receptor types during early embryogenesis have a smaller cortex, particularly in the frontal and temporal regions, and reduced numbers of excitatory neurons (Shin et al., [@B144]). Furthermore, these mice exhibited hyperactivity. Specific receptor knockouts have subsequently shown the relative contribution of each FGF receptor to cortical development. In order to avoid embryonic lethality of *fgfr1* and *fgfr2* systemic gene knockouts, *fgfr* alleles containing loxP site have been recombined *in vivo* with a variety of Cre lines, including the *foxg1* knock-in Cre, *nestin*-Cre and *hgfap*-Cre lines. Mice lacking *fgfr1* in radial glial progenitors driven by *hgfap*-Cre (which targets the dorsal telencephalon and hippocampal anlage), exhibited severe reduction of hippocampal volume, almost complete absence of midline telencephalic commissures due to abnormal development of midline glia, and decreased inhibitory interneuron number in the cortex and hippocampus (Ohkubo et al., [@B115]; Smith et al., [@B146]; Muller Smith et al., [@B108]). In contrast, the number of cortical excitatory neurons was not decreased in mice lacking *fgfr1* alone (Muller Smith et al., [@B108]). However, mice lacking *fgfr2* alone or in combination with *fgfr1*, via *hgfap*-Cre mediated recombination, showed a decrease in cortical excitatory neurons and volume, both of which were more pronounced in the medial prefrontal area of the cortex (Stevens et al., [@B148]). The mechanism of these abnormalities leading to a loss of cortical excitatory neurons resides in the ability of FGFR2 signaling to induce radial glial stem cells to re-enter the cell cycle, particularly in anterior regions. Therefore, FGFR2 support prefrontal cortical development by promoting the self-renewal or maintenance of cortical stem cells. These data converge with previous work demonstrating that PFC size is reduced by knockout of FGF17, a ligand for FGFR2 (Cholfin and Rubenstein, [@B29]).
Consistent with the decreased number of excitatory projection neurons in PFC in conditional knockouts for *fgfr2* driven by the *hgfap*-Cre transgene, glutamatergic synapses in the bed nuclei of the stria terminalis (BST), an area that receives projections from the PFC and in turn projects to the hypothalamus, were also reduced in these mice. A decreased number of GABAergic neurons in the BST was also noted in these animals, which is likely to be secondary to decreased glutamate input, as FGFR2 was not targeted in this region (Stevens et al., [@B148]). These changes may have important consequences for the correct functioning of limbic circuitry.
Mice constitutively lacking FGFR3 through the germline have skeletal defects but have not been reported to have abnormal cortical morphogenesis. (Colvin et al., [@B30]; Oh et al., [@B113]). On the other hand, mice with activating mutations of *fgfr3* have an increase in the rate of the cell cycle of cortical stem cells in early neurogenesis and an increase in the generation of TBR2+ IPCs at later stages, leading to increased cortical size and cortical cell number (Inglis-Broadgate et al., [@B71]; Thomson et al., [@B154], [@B153]). The caudal and lateral areas of the cortex were most affected, reflecting the natural gradient of *fgfr3* expression. Thus, FGFR2 and FGFR3 both appear to influence the appropriate proliferation of stem cells in different regions of cortex (anterior and posterior, respectively), suggesting that the regulation of signaling by each of these receptors individually and in combination may be critical for the appropriate expansion of different cortical areas. Compound mutation for *fgfr1*, *fgfr2* and *fgfr3* in the early anterior neural tube (driven by *foxg1*-Cre) results in an almost complete agenesis of both dorsal and ventral telencephalic regions (Gutin et al., [@B57]; Hebert and Fishell, [@B63]).
The downstream targets that mediate FGF regulated cellular events are still under investigation. Relevant to the early patterning processes in the cortex, FGF8-mediated activation of FGFRs is known to repress genes that specify dorsal and posterior cell fate including *coup-tf1*, *emx2*, and *wnt8b* (Crossley et al., [@B36]; Garel et al., [@B50]; Storm et al., [@B149]). FGF and other signaling factors also induce RAS/MAPK pathways, supporting NSC proliferation and self-renewal, and PI3K/AKT pathways, supporting cell survival. Several transcription factor genes are also activated by FGF signaling in NSCs as well as neuronal progeny, including *pea3*, *erm* (*etv5*), and *er81* (etv1) some of which are anteriorly expressed (Hasegawa et al., [@B59]; Cholfin and Rubenstein, [@B29]) and may be involved in cell differentiation or fate. The multiplicity of targets demonstrates how master regulatory factors such as FGF may exert multifaceted roles in stem cells and their progenies.
Cortical interneuron development
--------------------------------
The ganglionic eminences of the ventral telencephalon give rise to the striatum which emerges from the lateral ganglionic eminence (LGE), and to cortical and hippocampal GABA interneurons which are born in the medial and caudal ganglionic eminences (MGE and CGE). Parvalbumin (PV)+ and somatostatin (Sst)+ interneurons arise from the MGE, while calretinin+ interneurons arise largely from the CGE (Wonders et al., [@B165]). The interneurons from the MGE and CGE travel to the cortex via tangential migration during embryogenesis and progressively integrate into the cortical circuitry. Postnatally they undergo a protracted maturation and develop into their mature interneuron subtypes through processes that are largely unknown.
FGFR1 and FGFR2 appear to regulate early patterning of the ganglionic eminences before neurogenesis (Gutin et al., [@B57]). We shown that loss of *fgfr1* or *fgfr2* in the cerebral cortex via *hgfap*-Cre mediated recombination, which does not target the ganglionic eminences, results in a decrease in PV+ and Sst+ interneurons within the cortex (Smith et al., 2008). Work in our lab suggests that this deficit develops postnatally, and does not arise from defects in the patterning or proliferation of stem cells and progenitors in the ventral telencephalon. FGFR1 may act in the developing postnatal neocortex to support the survival and maturation of GABA interneurons.
Clinical Studies
================
Biological mechanisms of neuropsychiatric disorders can be studied in clinical populations using multiple lines of inquiry. Structural brain imaging can reveal changes in brain region volume while functional imaging can demonstrate altered functioning of brain regions, both of which may originate in early developmental processes, particularly when found in patients presenting early in the course of illness or young subjects at high risk for psychiatric disorders. More specific mechanisms relevant to stem cell biology may be examined in the post mortem brain as well as genetic analysis of patients. Converging evidence from populations of patients with autism, schizophrenia and affective disorders suggests that stem cell biology is implicated in neuropsychiatric etiology and pathophysiology.
Autism
------
The brain structure of patients with autism spectrum disorders (ASD) including autism, Asperger\'s Disorder (AD) and pervasive development disorder not otherwise specified (PDD-NOS) has been found to be abnormal using several different approaches. It is now widely believed that at least some of the deficits are present very early in life and that abnormal embryonic brain development may be a contributor to later structural deficits.
The head circumference of some patients with autism has long been known to be larger (Davidovitch et al., [@B38]; Woodhouse et al., [@B166]; Fombonne et al., [@B47]; Miles et al., [@B103]) because of abnormal acceleration in growth in early infancy (Courchesne et al., [@B33]). While no studies have been published on the neuroanatomy of high-risk individuals before a diagnosis of autism is made, retrospective data have shown that children with macrocephaly and autism do not have increased head circumference at birth (Lainhart et al., [@B85]; Courchesne et al., [@B33]; Dementieva et al., [@B39]; Redcay and Courchesne, [@B133]) but begin to show larger head measures at about 4 months of age (Gillberg and de Souza, [@B52]; Courchesne et al., [@B34]; Redcay and Courchesne, [@B133]). These findings suggest that autism may be underlain by either problems in early postnatal life and/or processes of embryonic development on which these postnatal events depend. Structural imaging of individuals with autism has shown that differences in brain volume, including both white and gray matter, diminish after the age of 5 years (Hazlett et al., [@B61]), although some studies have reported increases in gray matter volume in adolescents and adults with ASD (Lotspeich et al., [@B93]; Palmen et al., [@B119]; Hazlett et al., [@B62]), particularly in PFC (Mitchell et al., [@B106]).
Although there is no clear mechanism accounting for the dysregulation in the trajectory of brain growth in ASD, one hypothesis stipulates that it is the consequence of altered regulation of neural stem cell proliferation or differentiation arising before birth (Vaccarino et al., [@B159]). Further evidence that early embryonic developmental events are implicated in the pathophysiology of autism comes from post mortem studies demonstrating a fundamental change in cortical structure. Patients with autism were shown to have an increased packing density of mini-columns, which are vertical (radial) assemblies of neurons thought to be anatomically and functionally interconnected (Casanova et al., [@B19], [@B20]).
Several underlying mechanisms could explain macrocephaly and minicolumn pathology, all based on altered embryonic cortical development (Figure [3](#F3){ref-type="fig"}). The first is an increase in the number of radial units in the embryonic cerebral cortex, which in turn is thought to depend upon an increase in the number of "founder" NSCs in the cortical primordium (Rakic, [@B132]) (Figure [3](#F3){ref-type="fig"}). This mechanism is supported by the occurrence of mutations in *pten*, a gene that regulates embryonic stem cell proliferation (Eng, [@B43]) in a small number of autistic patients with macrocephaly (Butler et al., [@B15]). Abnormal expression of this gene in NSCs would likely result in an intrinsic alteration of stem cells. Interestingly, an animal model of *pten* mutations shows increased brain size and social deficits (Kwon et al., [@B83]), although this mutation was in differentiated neurons, not intrinsically affecting NSCs. In Fragile X syndrome, which frequently presents with symptoms of autism, fetal NSCs have been shown to differentiate into neurons at greater rates (Castren et al., [@B21]) and to misexpress multiple genes involved in proliferation and differentiation (Bhattacharyya et al., [@B8]). Mutant embryonic NSCs isolated from mice lacking the fragile X mental retardation protein (FMRP) due to a deletion in the *fmr1* gene differentiate in greater numbers into immature neurons (Castren et al., [@B21]). These findings are similar to those obtained in *Drosophila* germline stem cells lacking an ortholog of the *fmr1* gene (Yang et al., [@B169]). Thus, the *pten* and *fmr1* mouse models of ASD support the hypothesis that an intrinsic abnormality in NSC is responsible for features of these disorders. Two members of the *TF-II* family of transcription factors involved in Williams syndrome, another disorder with abnormal social behavior, have been shown in mice to regulate specific gene targets that may be involved in embryonic stem cell differentiation (Makeyev and Bayarsaihan, [@B97]).
{#F3}
GABA interneurons surround each minicolumn, and some genetic evidence and post mortem data suggest a GABAergic abnormality in ASD (Fatemi et al., [@B45]). Multiple genes involved in the development and function of the GABAergic system, including *dlx5*, have been involved in Rett syndrome, a developmental abnormality with autistic features (Horike et al., [@B67]); furthermore, alterations in the development of GABAergic neuron circuitry have been found in mice lacking the *Methyl-CpG binding protein 2 (MeCP2)* gene, whose mutations are responsible for Rett syndrome (Medrihan et al., [@B100]; Zhang et al., [@B173]). Lastly, a large number of synaptic-related genes have been implicated in small subsets of patients with ASDs (in total, accounting for probably less than 3--4% of the cases) (Buxbaum, [@B16]; Radyushkin et al., [@B126]). Thus, it appears that abnormalities in both the early-determined size and scaffolding of the cerebral cortex and later developing synaptic connections may play a role in individual cases of autism.
Schizophrenia
-------------
In schizophrenia, retrospective studies have suggested that head circumference is decreased at birth and developmental delays are present in early childhood, both of which implicate prenatal and early postnatal alterations in forebrain development (Cannon et al., [@B18]). The occurrence of prodromal symptoms in the majority of cases (Hafner et al., [@B58]), as well as the presence both prior to and after onset of illness of neuropsychological deficits (Crespo-Facorro et al., [@B35]) also implicates disruption of forebrain development. Neuropsychological dysfunction implicates the PFC in patients with schizophrenia, the cortical region that has expanded most extensively in mammalian evolution and that has been suggested to rely on key evolutionary developments in stem cell function (Martinez-Cerdeno et al., [@B99]).
Altered hippocampal development has also been implicated in the pathogenesis of schizophrenia (Kobayashi, [@B80]). Total brain gray matter and hippocampal volumes are reduced at the time of first episode in schizophrenia (Ohnuma et al., [@B116]; Gur et al., [@B56]) and at times later in the disease course, volume reductions have been observed in PFC, hippocampus and temporal lobe (Shenton et al., [@B142]). Longitudinal studies with structural imaging show that both the PFC and the medial temporal lobe become progressively smaller during the period when psychosis develops and that dorsal prefrontal regions experience even more loss as illness progresses, consistent with a neurodegenerative process (Pantelis et al., [@B120]). Post mortem studies of the cerebral cortex of patients with schizophrenia have shown reductions in neuropil (Selemon et al., [@B140]), as well as in GABAergic cells expressing PV and reelin (Guidotti et al., [@B55]; Fatemi et al., [@B46]; Hashimoto et al., [@B60]). Post mortem analysis of the hippocampus also shows evidence for an abnormal GABAergic system and suggests that important signaling pathways are altered, including WNT and TGFbeta (Benes et al., [@B6], [@B7]; Todtenkopf and Benes, [@B156]; Arnold et al., [@B2]). While these post mortem findings may reflect changes in the adult brain due to disease progression and medication effects, alterations in cortical neurodevelopmental processes could also create vulnerabilities that develop later in life.
Genes that are implicated in stem cell regulation, including *neuregulin*, *disc-1*, *wnt* related genes, *bdnf* and *fgfr1*, have been associated with schizophrenia in genetic association and post mortem studies, suggesting NSCs dysregulation in at least some cases. Mouse models lacking *fgfr1* embryonically have smaller hippocampi and cortical interneuron deficits similar to those in patients with psychosis (Ohkubo et al., [@B115]). Similarly, deficient *bdnf*, *neuregulin* and *disc-1* genes in mice mimic various aspects of the disorder (Ayhan et al., [@B4]; Brandon et al., [@B13]; Meyer and Morris, [@B102]). Conversely, mutations found in some patients with schizophrenia have significant effects on cortical stem cell development. Mice lacking the genes within the 22q11 mutation (velocardiofacial syndrome, a known chromosomal abnormality predisposing to schizophrenia) have abnormal neurogenesis, specifically affecting upper cortical layers (Meechan et al., [@B101]). As in autism, patients with schizophrenia are likely heterogenous, with pathology in some determined by a component of neuronal functioning that arises postnatally and in others, determined by earlier disruptions of patterning and neurogenesis.
Affective disorders
-------------------
Patients with major depression have been found to have smaller hippocampi (Videbech and Ravnkilde, [@B163]) and smaller anterior cingulate cortical regions (Caetano et al., [@B17]). Connectivity and functioning of prefrontal and cingulate cortices have also been shown to be abnormal in individuals with major depression (Drevets, [@B40]; Grimm et al., [@B54]; Vasic et al., [@B162]). A recent longitudinal study of people at high familial risk for developing depression show thinning of the cerebral cortex, particularly in right-sided frontal and parietal areas (Peterson et al., [@B123]). This thinning was present even in children and adolescents prior to the onset of any mood episode, suggesting that structural changes that result from abnormal early cortical pruning during infancy or adolescence may be a predisposing factor in depression. In patients with bipolar disorder with psychotic symptoms, neuroanatomical and neuropsychological findings are similar to those found in patients with schizophrenia (Murray et al., [@B109]). In non-psychotic bipolar disorder, patients presenting at the first episode of illness have shown decreased volume in prefrontal and temporal cortex (Hirayasu et al., [@B65]; Strakowski et al., [@B150]; Kasai et al., [@B77]). Bipolar disease progression is associated with ventricular enlargement and stronger reductions of the PFC which correlate with impaired functioning during manic episodes (Bearden et al., [@B5]; Blumberg et al., [@B9]).
Post mortem studies have demonstrated that prefrontal cortical regions have abnormal densities of pyramidal, GABAergic and glial cells in patients with depression. Medium to large pyramidal neurons appear to be lower in density and smaller in soma size (Rajkowska et al., [@B131], [@B130]; Law and Harrison, [@B86]) while smaller neurons and those in layers III, V may be increased in density (Chana et al., [@B24]; Rajkowska, [@B128]). Glial cell density in patients with depression appears significantly reduced (Ongur et al., [@B117]; Rajkowska et al., [@B131]; Cotter et al., [@B32], [@B31]; Uranova et al., [@B158]). These findings suggest that abnormal prenatal development of stem/progenitor cells that generate the neurons populating different cortical layers and abnormal development of glial elements may account for structural brain abnormalities in depression.
Post mortem work has also examined the levels of signaling factors in patients with depression. In frontal cortical regions, FGF ligand and receptor levels as well as BDNF receptor levels were shown to be altered in patients with depression. Findings generally show reduced expression (Evans et al., [@B44]; Sibille et al., [@B145]) but some point to upregulated receptor levels (Tochigi et al., [@B155]). Altered gene expression in the mature cortex may suggest a role for FGF and other important mediators of neuronal growth in mood disorders in the adult brain. However, they may also reflect a genetic defect that may have contributed to an earlier altered developmental trajectory. Parallel findings in model animal systems (Chen et al., [@B25]) and patients with mood disorders (Rajkowska et al., [@B129]; Valentine and Sanacora, [@B161]) suggest a role for FGF in both embryonic and mature brain functions (Figure [4](#F4){ref-type="fig"}).
{#F4}
Adult Neural Stem Cells
=======================
Neural Stem Cells in the adult brain are present in regions around the ventricles, specifically in the SVZ, which gives rise to neurons that migrate to the olfactory bulb via the rostral migratory stream. A region in the hippocampal dentate gyrus (DG), the subgranular zone (SGZ), also contains NSCs which generate DG granule cells. Interestingly, over 80% of the interneurons in the olfactory bulb and 10% of granule cells in the DG are replaced by this mechanism in the period of about 1 year, suggesting that a portion of these neurons are under constant turnover (Lemasson et al., [@B88]; Ninkovic et al., [@B111]; Imayoshi et al., [@B69]).
The contribution of adult stem cells to regeneration and neuroplasticity in adult patients with mental illness has received relatively more attention than the role of embryonic NSCs in these disorders. Interestingly, as more is understood about the intrinsic and extrinsic regulation of both embryonic and adult NSCs, common mechanisms are being identified which may underlie disruption in both types of NSCs in disease states (Bordey, [@B11]).
While the identity of NSCs in the adult brain is not clearly documented, it appears likely that they have many features characteristic of astroglia (Imura et al., [@B70]; Alvarez-Buylla and Lim, [@B1]; Liu et al., [@B91]). Like embryonic radial glia, adult SVZ NSCs have GABA~A~ receptors which limit their proliferation in response to GABA released by neuroblasts, suggesting negative feed-back regulation *via* non-synaptic GABAergic signaling between neuroblasts and GFAP-expressing NSCs (Liu et al., [@B92]).
Neuronal stem cells in the adult brain pass through several stages from quiescent multipotent cell (astroglial in nature, slowly dividing) to amplifying neural progenitors which rapidly proliferate and generate immature neurons, which finally mature into granule cells or olfactory interneurons (Ma et al., [@B94]). A large proportion of postnatally generated neurons die and only some are incorporated into the existing circuitry. It has been proposed that this pace may be adaptive in order to enhance the sensitivity of the system while the new neurons fully integrate. Environmental and activity-dependant factors may be critical for regulating how these cells pass appropriately through these stages and hence how newly born neurons will participate in normal brain functioning and potentially compensate for pathological states.
Like embryonic NSCs and radial glial cells, adult astroglial NSCs and progenitor cells in the SVZ and SGZ express several RTKs including *fgfr1*, *fgfr2* and *fgfr3, bmpr-1a/alk3* as well as *egfr*, and have active SHH and Notch signaling systems. Adult NSC maintenance and self-renewal depends on secreted FGF, BMP, SHH, and EGF, acting in conjunction with the cell-attached Notch signaling system (Machold et al., [@B96]; Zheng et al., [@B175]; Lie et al., [@B90]; Mira et al., [@B105]). For example, the number of proliferating cells in the adult SVZ is reduced by 40% in *fgf2* KO mice (Zheng et al., [@B175]). FGF2 increases cell cycle re-entry and BMP promotes quiescence in adult NSCs, and thus, as in the embryo, the balance of these 2 extrinsic factors is likely to be important for proper regulation of adult NSC proliferation and differentiation (Ming and Song, [@B104]; Mira et al., [@B105]).
Much evidence has been found linking major neuropsychiatric disorders to altered regulation of adult NSCs and to the molecular systems involved in their maintenance and differentiation. The SGZ of the DG has been the region of the most research focus. DG functioning underlies mood, memory, and reward, which are impaired in schizophrenia, major depression and bipolar disorder. In animal models of depression, schizophrenia, and bipolar disorder, the activity of SGZ stem cells is reduced. For example, chronic stress leads to a depression-like phenotype in rodents, suppresses neurogenesis in the SGZ and reduces FGF2 expression in hippocampus (Warner-Schmidt and Duman, [@B164]). Findings from post mortem studies have demonstrated that numbers of proliferating cells in the DG are decreased in psychotic disorders (Reif et al., [@B134]); furthermore, abnormal FGF expression has been documented in the post mortem hippocampus of depressed and schizophrenic patients (Gaughran et al., [@B51]). In this study, neither antidepressant treatment nor the presence of affective disorders were found to be correlated with altered numbers of hippocampal dividing cells. However, other post mortem work has demonstrated that patients who received antidepressant treatment did show a greater number of neural progenitor and dividing cells in the DG than untreated patients with depression (Boldrini et al., [@B10]). Furthermore, antidepressant treatment increases hippocampal neurogenesis in both rodents (Malberg et al., [@B98]) and non-human primates (Perera et al., [@B121]). Chronic antidepressant administration to rodents has been shown to increase FGF2 expression in astrocytes and neurons within and outside of the hippocampus, suggesting that FGF signaling may be involved in the neurogenic action of antidepressants. In support of this idea, the infusion of FGF2 into the ventricles of adult rats increases cell proliferation and reduces depression-like behavior in rodents (Turner et al., [@B157]). Also, conditional knockout of *fgfr1* in neuronal stem cells by *nestin*-Cre results in reduced proliferation and neurogenesis in the adult DG (Zhao et al., [@B174]). New data suggest FGF2 may be specifically involved in the modulation of anxiety/fear behavior in rodents (Perez et al., [@B122]). However, it is currently controversial whether ongoing hippocampal neurogenesis is required for the therapeutic action of antidepressants. Some have proposed that hippocampal neurogenesis may be limited to mediating the beneficial action of antidepressants on cognition. One problem is the limitations of animal models of depression in which behavior may rely on mechanisms that are peripheral to human disease (Santarelli et al., [@B137]; Sahay and Hen, [@B136]; Holick et al., [@B66]; David et al., [@B37]). Hence, the specific role of FGF2 on the development and function of systems that subserve affective and cognitive functions in humans is still somewhat unclear.
Gene products implicated with neuropsychiatric illnesses *via* association studies have been shown to affect hippocampal neurogenesis, most likely by working intrinsically within stem and progenitors cells. The gene with the strongest association to schizophrenia, *disc-1*, has been shown to regulate the maturation of newborn neurons in the mature brain. When expression of *disc-1* is suppressed, neurons mature too quickly (Kim et al., [@B79]). This may involve a disruption of the balance between those signaling pathways that maintain neuronal stem cells, such as FGF, and those that advance their differentiation into neurons, such as BMP2/4. A role in progenitor cell maintenance and neuronal differentiation within hippocampal NSCs has been also identified for systems that regulate gene expression by methylating or acetylating specific histone proteins in stem and progenitor cells (Wu and Sun, [@B167]; Yu et al., [@B172]). A gene that has been associated with some forms of autism, *gadd45b*, may also play a role in these pathways as it increases progenitors in the DG after electroconvulsive treatment, perhaps by controlling *fgf* gene expression by DNA methylation (Ma et al., [@B95]). In summary, neuropsychiatric disorders may have links to stem cell biology in the mature as well as the developing brain.
The Future of Stem Cells and Interventions: Integrating Human and Animal Knowledge
==================================================================================
Work in animal models reveals fundamental processes of brain development that are important to generate and test hypotheses concerning the pathogenesis of human mental disorders. This opens the possibility of manipulating stem cells in clinical populations. Avenues of research are utilizing general knowledge about stem cells to (1) use manipulated stem cells in a rehabilitative capacity in patients and (2) use stem cells from patients to understand pathophysiology and test candidate drugs for reversing abnormal cellular functions. We will review briefly here how knowledge about stem cell biology from both humans and animal models is being applied to develop these new research directions.
Research on the use of stem cells for rehabilitation was originally motivated by the potential for direct development of new neurons from stem cells that might replace lost neurons that are at the center of pathophysiology. For example, in degenerative illnesses such as Parkinson\'s disease and Amyotrophic Lateral Sclerosis, specific populations of neurons, dopaminergic cells and spinal motor neurons respectively, are progressively lost. Stem cell research has sought to use either embryonic or other types of stem cells to generate replacement neurons that could be transplanted and incorporated into the CNS of patients (for a Review, see Hynes and Rosenthal, [@B68]). This work has required knowledge acquired from animal models of the transcription and signaling factors that determine dopaminergic and spinal motor neuron fate and maturation. As researchers have pursued this goal, it has been shown that stem cells in animal model transplant recipients may have an indirect trophic role, i.e., reducing cell death, increasing metabolic functions and protecting remaining neurons from insults (Jung et al., [@B76]; Rafuse et al., [@B127]; Yasuhara et al., [@B170]). These findings may hold promise for future application of stem cell-derived therapies to neuropsychiatric disorders. Disorders such as schizophrenia and depression may have a neurodegenerative component, similar to neurological disorders for which the supportive role of stem cells has been demonstrated. The regions of the brain involved may be different for major affective and psychotic disorders than degenerative motor diseases -- e.g., cortex and hippocampus -- but these regions may be even better candidates for the application of stem cell support therapy given their nearby neurogenic zones. Extrinsic factors that influence the stem cell niches, like FGF and BMPs, will likely have an important role to play in these potential therapies.
Induced pluripotent stem cells (iPSCs) also hold great promise in their application to neuropsychiatric disorders. These pluripotent stem cells can be derived from skin fibroblasts or blood lymphocytes by introducing a combination of genes that confer stem cell like properties, essentially reprogramming the differentiated cells to behave like embryonically derived stem cells (Takahashi et al., [@B151]; Yu et al., [@B171]; Muller et al., [@B107]). Takahashi and colleagues reported the first generation of human iPSCs in 2007. They achieved this by introducing the *oct3/4*, *sox2*, *klf4*, and *c-myc* genes into fibroblasts with retroviral vectors. As iPSCs reproduce the essential features of human embryonic stem cells (hESC) they can be coaxed to give rise to neurons. It has already been shown that hESC lines can give rise to mature, physiologically active neurons using different protocols, and many of these protocols have already been successfully applied to iPSC (Johnson et al., [@B75]; Chambers et al., [@B23]; Kriks and Studer, [@B82]; Soldner et al., [@B147]). Furthermore, by using different combinations of secreted patterning gene products such as FGFs, SHH and WNTs, different subtypes of neurons can be generated such as forebrain specific neurons, midbrain dopaminergic or GABAergic neurons (Yan et al., [@B168]; Eiraku et al., [@B41]; Chambers et al., [@B23]). Here again, decades of animal research in neurodevelopment is paying off, in that the accumulated knowledge in patterning and cell fate specification events in normal development will aid in the development of *in vitro* differentiation paradigms for generating specific neuron subtypes. More needs to be done and human and animal research must proceed in parallel.
Our inability to examine the direct consequences of gene sequence variations in the living CNS at the transcript and biological levels is a formidable challenge to the investigation of the genetic pathophysiology of psychiatric disorders. The derivation of iPSCs may allow us to investigate basic neurobiological aspects of neuropsychiatric disorders using patient-derived cell lines. For the first time, scientists will be able to use patient-derived cells to investigate developmental properties of the nervous system, such as neural stem cell proliferation, neural differentiation, and synapse formation either *in vitro* by using cell culture techniques, or *in vivo* by transplantation into a host animal, such as the developing rodent brain. One of the most promising aspects of iPSCs is the potential to examine the effects of specific human gene variants upon neural differentiation and function, by creating iPSCs from patients with particular genetic mutations. Furthermore, the gene expression profiles of neurons derived from such patients-derived neural stem cell lines can be examined. One potential complication of such studies is the large number of genetic variants present in the human genome. While on the one hand it is useful to begin to catalog such variants, associating them with particular phenotypes will require sophisticated statistical analyses and a large number of samples. Prior to the development of iPSCs, the standard approach for investigating the role of individual genes in brain development was to create a transgenic mouse model with either gene inactivation, gene overexpression, or mutations that mimic the mutations found in human disease. This remains a powerful approach, and indeed, animal work has provided useful experimental frameworks in which the functional impact of specific gene mutations can be tested. However, there are important differences in the development of the rodent brain and the human brain, which are reflected in the diversity of gene splice variants expressed during animal and human brain development (Johnson et al., [@B74]). While work in animals continues to represent an essential reference point against which differences in gene expression and function found in humans can be compared, the use of iPSCs will allow us to test hypotheses about the effects of specific gene variants and mutations in developing human cells. Using iPSC models, we should be able to ask questions such as, how mutations in the *disc-1* gene, together with a specific constellation of gene mutations/variants present an individual patient, lead to abnormal cell proliferation or neuronal differentiation in human NSCs? We may also be able to identify abnormalities in stem cell development in patients for which no specific genetic mutation has yet been identified. Such research may also provide means by which to test candidate pharmacological agents for their ability to normalize the function of neural cells derived from patients that may have abnormal functions.
The study of NSCs in multiple model systems will continue to inform our understanding of normal brain structure and function. The application of this information to patient populations is slowly shedding light on the pathophysiology of major neuropsychiatric diseases. The advent of new techniques for manipulating human cells may accelerate this process of discovery and may hold real promise for improving the lives of patients with mental illness.
Conflict of Interest Statement
==============================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
[^1]: Edited by: Ronald S. Duman, Yale University School of Medicine, USA
[^2]: Reviewed by: Amelia J. Eisch, University of Texas Southwestern Medical Center, USA
[^3]: This article was submitted to Frontiers in Neurogenesis, a specialty of Frontiers in Neuroscience.
| {
"pile_set_name": "PubMed Central"
} |
I[NTRODUCTION]{.smallcaps} {#sec1-1}
==========================
Diabetes, a global escalating public health problem, primarily because of the increasing prevalence, is estimated to affect 285 million individuals worldwide\[[@ref1]\] (Approximately 90% have type 2 diabetes mellitus) and causes hundreds of billions of dollars of economic damage each year. Global estimates for the year 2030 predict a further growth of almost 50%.\[[@ref2]\] In 2000, it is estimated that 2.8% of world\'s population had diabetes mellitus and that by 2030 this number will be 4.4% of the world\'s population. According to WHO\[[@ref2]\] the 'top' three countries in terms of the number of type 2 diabetes mellitus (T2DM) individuals with diabetes are India (31.7 million in 2000; 79.4 million in 2030), China (20.8 million in 2000; 42.3 million in 2030) and the US (17.7 million in 2000; 30.3 million in 2030). This increase is a warning sign for Indian health care system to be vigilant for adequate diabetes mellitus management.
Type 2 diabetes mellitus is a complex and pleomorphic metabolic disorder, characterized by defects in insulin secretion and insulin action which lead to hyperglycemia.\[[@ref3]\] With the onset of this chronic condition and the associated co-morbidities (hypertension, abdominal obesity, dyslipidemia and insulin resistance; together termed as metabolic syndrome), a life-long reduction in quality of life and premature mortality due to micro and macro-vascular complications can be expected.\[[@ref4]\] Genetic background and environment factors are likely to be important in determining susceptibility to associated micro and macro-vascular complications, but exposure of tissues to chronic hyperglycemia is the main initiating factor. Thus, the primary therapeutic goal is to reduce plasma hyperglycemia. For many years, it is generally accepted that exercise is a cornerstone of diabetes management, along with dietary and pharmacological interventions. Based on a number of large randomized controlled trials, current guidelines from the American Diabetes Association (ADA)\[[@ref5]\] acknowledge the therapeutic strength of exercise intervention. Although, there are guidelines on treatment patterns for T2DM patients, it appears that most patients with T2DM are treated pharmacologically.\[[@ref6]\] It is not known whether this is a reflection of poor adherence to lifestyle modifications in T2DM patients or because clinician\'s perceptions of and experiences with lifestyle recommendations and interventions are less effective (for whatsoever reasons) for managing T2DM patients.
Presently available literature regarding the effects of lifestyle modification, in terms of increasing the quantity and quality of physical activity, on overall glycemic control independently or in combination with hypoglycemic drugs are scarce and less conclusive. The majority of studies which supplement weight reduction and exercise programs with appropriate drug therapy have been done on western population. There is a paucity of such studies in Indian population more so specifically in Gujarat population. In India, considering the economic/time constraint and other factors (such as availability of suitable training center, awareness etc), structured exercise training may be available only to a subset of patients with type 2 diabetes, increased physical activity (unstructured activity) is more feasible and can be easily followed.
Therefore, with the hypothesis that short-term six month regular physical activity (structured exercise/unstructured activity) undertaken by T2DM patients, with minimal diet modification may boost glycemic control, an attempt has been made in this study to assess the effect of structured exercise training and unstructured activity interventions on glycemic control along with other biochemical and anthropometric parameters in T2DM patients.
Primary outcome for the result was change in hemoglobin A1c value from baseline to the end of the intervention and secondary outcomes were measures of anthropometry, plasma lipid levels and blood pressure. In addition, we assessed the proportion of T2DM patients in whom cardiovascular risk factors (CVRFs)/profile were not under control as established by the NCEP-ATP III (National Cholesterol Education Program-Adult Treatment Panel III) guidelines.\[[@ref7]\]
M[ATERIALS AND]{.smallcaps} M[ETHODS]{.smallcaps} {#sec1-2}
=================================================
This was a randomized six month exercise intervention study conducted from October 2011 to July 2012 (Box 1 shows the flow of participants from enrollment to follow--up) \[[Figure 1](#F1){ref-type="fig"}\]. Before randomization, all enrolled T2DM participants (*n*: 300) entered a one-month run-in phase to reduce dropout and maintain adherence. Participants performed 15 min of aerobic exercise\[[@ref8]\] and one set of nine resistance exercises\[[@ref8]\] {four upper body exercises (bench press, seated row, shoulder press, and pull down), 3 leg exercises (leg press, extension and flexion), abdominal crunches and back extensions}, at moderate intensity\[[@ref8]\] under professional trainer and physiotherapist. Only persons who attended \>75% of the scheduled 24 run-in sessions were eligible for randomization. The study protocol was approved by the human research ethical committee and informed consent was obtained (after providing a detailed study overview) from all the patients before enrolling into the study.
{#F1}
Inclusion criteria for enrollment includes sedentary, 30 to 60 year old adults of either sex with type 2 diabetes mellitus (for more than a year) of HbA1c levels of 6.5% or higher, residing in and around Ahmedabad (Gujarat), and attending the diabetic clinics at B. J. Medical College and Civil Hospital, Ahmedabad (Gujarat). Sedentary was defined as not exercising more than 20 min on three or more days a week. The history of diabetes mellitus was based on patient self report of a prior physician diagnosis. Patients with overt albuminuria, congestive cardiac failure, preexisting macro-vascular condition, any severe illness (such as malignancy, severe infection, respiratory disease, kidney disease, liver disease), impairment of speech, hearing, vision or cognition, suffering from a serious diabetes complication, using insulin, changes during the previous three months in oral hypoglycemic agents, continuous or periodic use of corticosteroids, pregnant females or who had given birth within the preceding six weeks, orthopedic constraints or any musculoskeletal injury or joint or peripheral vascular disease sufficient to impede exercise or who had participated in regular physical exercise (more than two 30 min sessions/week of moderate/vigorous aerobic exercise or one 30 min session/week of resistance training) during the preceding six months, serious exertion hypertension or any medical condition that prevented participants from adhering to the protocol or exercising safely, lack of approval by physician and patients showing disinterest were excluded from the study.
After one-month run-in phase, participants were randomly allocated (randomization sequence was computer generated) in equal numbers to the structured exercise training, unstructured activity and control groups, stratified by sex and age. Structured exercise training was defined as an intervention in which patients were engaged in planned, individualized and supervised exercise programs (combination of aerobic and resistance exercise). Unstructured activity was defined as an intervention in which patients were not engaged in supervised exercise training, but received advice to increase the "physical activity" which refers to any bodily movement produced by skeletal muscles that results in an expenditure of energy and includes a walk for 45 min/day, broad range of occupational, leisure and daily activities and control group in which patients reverted to pre-study physical activity levels. Later two groups were asked to maintain the said activity during the six-month study period.
Participants, physiotherapist and trainers could not feasibly be blinded to group assignment after randomization, but the main study outcomes were measured by blinded objective methods. Dietician recommended (based on standard guidelines)\[[@ref6][@ref9]\] a diet (which was not greatly different from individual\'s routine diet) to all participants that would not cause weight loss to minimize dietary variability among groups. Physicians were requested not to alter the medications (antihypertensive, lipid-altering, or hypoglycemic) during the six month intervention unless it was medically necessary and if any change occurs throughout the study, it was well documented.
Structured exercise training group participants followed an exercise plan consists of a combination of aerobic and resistance exercise, conducted under supervision of professional trainer and physiotherapist. Participants exercised six-times weekly,\[[@ref6][@ref10]\] Each session lasted for 45 min and training progressed gradually in intensity. Each exercise session had a five minute warm-ups and cool-down period, consisting of very light exercises and stretching, to allow a gradual warming/cooling of the muscles. During 12^th^ week the exercise dose was reduced by one-third to provide a recuperation week.
Nine resistance exercises were performed on three non-consecutive days/week using weight-stacked machines (Multi Station Gym Khare Enterprises Pvt. Ltd, Sangli.). {The weight loading was set at 70-85% one repetition maximum, determined for each exercise at week 0 to assess the muscle strength.\[[@ref11]\] One repetition maximum is defined and determined as: Following a low-intensity warm-up, participants performed four trials (separated by a one-minute resting interval) using varying moderate-heavy weights to determine the highest weight that could be lifted with only one repetition through the full range of motion with correct technique}. Each session consisting of two sets of four upper body exercises, three sets of three leg exercises and two sets each of abdominal crunches and back extensions (with one-min rest between sets). Each set consisted of 12 repetitions. Weight or resistance was increased by 5 to 10 pounds when the participant was able to complete 12 repetitions for each set of exercises on 2 consecutive exercise sessions while maintaining proper form. On other three days of week, aerobic activities were performed on a bicycle ergometer or treadmill. At week 0, participants self-selected a walking/cycling pace and the grade increased by 2% every 2 min until exhaustion. The same speed/grade was used for baseline and then the intensity was increased on a weekly basis.
All patients were studied as outpatient. Patients were interviewed for medical and nutritional history. Present and past history of each case was recorded in detail regarding their general information i.e., name, age, sex, address, religion, occupation, economic status, nutritional and personal habits, education, medication and history suggestive of any systemic illness. Each patient was then examined for various anthropometric parameters: Weight (Kg) and height (meters) were measured (using Omron digital body weight scale HN-286 and SECA 206 wall mounted metal tapes respectively). Body mass index (BMI) was calculated by weight (Kg)/height squared (m^2^).\[[@ref12]\] Waist circumference was assessed in the standing position, midway between the highest point of the iliac crest and the lowest point of the costal margin in the midaxillary line. Hip circumference was measured at the level of the femoral greater trochanter. Total body fat (%) was measured using bioelectrical impedance analyzer (Omron HBF-362). All anthropometric measures reflect the average of 3 measurements (measured by same person on same instrument to avoid inter-instrument and inter personal variation). Blood pressure was measured three times in the seated position after 10 min of rest with a standard manual mercury sphygmomanometer (Diamond Deluxe Industrial Electronics and Products). The recorded pressure of the three measurements was averaged. Patients were assigned to a category of hypertensive status according to the Seventh Report of the Joint National Committee, JNC 7.\[[@ref13]\] Age was defined as the age at the time of interview (though no documentary proof had been entertained) and the date of diagnosis of diabetes mellitus was obtained from the patient.
After an overnight fast of 12 h, venous sampling was done at baseline and at 6 m (post-intervention). Serum and plasma was separated by centrifugation of blood sample and were subjected for analytical procedures. Glucose (Glucose oxidase method, CV %: 3.4),\[[@ref14]\] cholesterol (Cholesterol oxidase method, CV %: 3.9),\[[@ref15]\] triglycerides (Enzymatic method, CV %: 3.6),\[[@ref16]\] high-density lipoprotein cholesterol (HDL-C) (Phosphpotungstic method, CV %: 4.7)\[[@ref17]\] low-density lipoprotein cholesterol (LDL-C) (CV %: 3.6)\[[@ref18]\] and HbA1c (Immunoturbidimetric method, CV %: 3.9),\[[@ref19]\] were measured in fully automated analyzer (Bayer express plus). Quality was controlled using standard solutions. To ensure safety, participants had monthly visits with a physician who reviewed glucose levels to identify hypoglycemic risk (HbA1c, fasting and 2 h post prandial glucose levels was assessed monthly for the same).
The prevalence of the metabolic syndrome was assessed according to the NCEP-ATP III criteria;\[[@ref7]\] when three or more of the following five conditions were present: Abdominal obesity (waist circumference \>102 cm in men, \>88 cm in women); serum triglycerides equal to or greater than 150 mg/dl; HDL cholesterol less than 39 mg/dl in men and 45 mg/dl in women; systolic blood pressure equal to or greater than 130 mm Hg and/or diastolic blood pressure equal to or greater than 85 mm Hg; and fasting plasma glucose \>110 mg/dl or use of hypoglycemic medication. In addition, we assessed the proportion of participants in whom CVRFs were not under control as established by the NCEP-ATP III guidelines.\[[@ref7]\]
All analyses were performed using SPSS statistical software (SPSS, version 15.0). Prior to hypothesis testing, data were examined for normality. Non--normally distributed variables were logarithmically transformed before analysis. Two-group comparisons were made using X^2^ or Fisher\'s exact tests (when any expected cell frequency was \<5) for categorical variables and Student\'s t-tests or one-way ANOVA for continuous variables. For all analyses, two-sided probability values \< 0.05 were considered statistically. Power and sample size calculation yielded 61 participants/group. It was based on a predicted HbA1c difference of 0.65 HbA1c units with an SD of effect of 1.2 HbA1c units, α=0.05, 1− β=0.85 and an expected dropout rate of 15%. This study exceeded this sample size.
R[ESULTS]{.smallcaps} {#sec1-3}
=====================
We enrolled 300 patients (56% males) with T2DM, 21 were excluded during run-in phase due to inadequate adherence. Of the 279 randomized participants, 21 withdrew {due to lack of time or loss of interest (8 from unstructured activity group; 6 from structured exercise training group and seven from control group)} and 13 participants were excluded for exercise noncompliance (Compliance with structured exercise training was defined as 75% of prescribed sessions and was determined as the number of exercise sessions completed at the prescribed training loads per total number of prescribed sessions). There was no difference between exercises versus non-exercise groups in the number of participants who withdrew (*P* \> 0.66). From baseline to end, the median exercise training attendance was 84%. [Table 1](#T1){ref-type="table"} shows the clinical characteristics of the randomized patients. One of every 6 patients reported to actively smoke. The mean reported duration of diabetes mellitus in the randomized participants was 6.1 + 4.2 years with a HbA1c level of 7.5 + 0.6% and there were no significant differences in age, duration of the disease, weight, BMI, and sex distribution between groups (*P* \> 0.46) at baseline. Elevated body weight was a concomitant health disorder for most patients. Ninety two percent showed a body mass index equal to or greater than 25.0 kg/m^2^, and 56% of the patients (95% confidence interval \[CI\], 51-61%) were obese (BMI \> 30.0 kg/m^2^). Baseline systolic blood pressure was well controlled, most likely because of the high percentage of patients taking medications. Only 51 (18.27%) of the patients were not receiving antihypertensive medications. Two or more antihypertensive drugs were being taken by 57% of the patients and 12% were treated with four different blood-pressure lowering drugs. As for glycemic control, biguanides and sulfonylurea group of drugs were taken by most patients.
######
Clinical characteristics of randomized patients at baseline
Tables [2](#T2){ref-type="table"} and [3](#T3){ref-type="table"} shows overall results for baseline, post- intervention and change in various primary and secondary outcome parameters across the groups respectively, providing details on within group changes and intergroup analyses. Unstructured activity alone was not found to have a statistically significant effect on HbA1c reduction (0.14%, 95% CI, 0.09-0.22%; *P* = 0.12). Structured exercise training reduced HbA1c by 0.59% (95% CI, 0.52-0.68; *P* = 0.03), which is considered both statistically and clinically significant. The absolute change in HbA1c in the structured exercise training group Vs the control group was 0.69% (95% CI, 0.62-0.75%; *P* = 0.02) whereas in unstructured activity group changes in HbA1c was not significant compared with those in the control group 0.24% (95% CI, 0.19 - 0.31%; *P* = 0.09). In a subgroup analysis limited to participants with a baseline HbA1c value \>7%, both the unstructured (0. 48% (95% CI, 0.42-0.54%; *P* = 0.04) and structured exercise training (0.77% (95% CI, 0.69 -0.83%; *P* \< 0.01)) groups experienced significant decline in HbA1c Vs the control, whereas among participants with baseline hemoglobin A1c values less than 7%, significant reduction occurred only in the structured exercise training group. Even after excluding 17 participants due to any changes in oral hypoglycemic medications, results were similar to those of the overall study sample.
######
Baseline and post-intervention values of primary and secondary outcomes variables
######
Baseline to post-intervention changes in primary and secondary outcomes variables
Changes in blood pressure; total cholesterol, HDL cholesterol, LDL-cholesterol and the atherogenic index factors (total cholesterol/HDL-cholesterol and LDL-cholesterol/HDL-cholesterol ratio) did not statistically significantly differ within (baseline to follow-up) and among groups. However, structured exercise was related to a significant decrease of 21.85 mg/dl post-intervention (95% CI, 13.82-29.75; *P* \< 0.05) in triglycerides.
The average baseline to post-intervention changes in body weight were −0.5kg (*P* = 0.79) in the control group, −3.7 kg (*P* = 0.21) in the structured exercise training group and 1.4 kg (*P* = 0.48) in the unstructured activity group. In the body mass index comparisons for the exercise groups Vs control group, no significant post-intervention differences were found. However, total body fat percentage did show significant improvement in structured exercise training group (*P* = 0.04), as did waist circumference (*P* = 0.030).
At baseline, the metabolic syndrome was diagnosed in 71.9% (95% CI, 67.1-76.8%) of participants, as per NCEP-ATP III criteria.\[[@ref7]\] Most patients showed at least one uncontrolled CVRF \[[Figure 2](#F2){ref-type="fig"} and [Table 4](#T4){ref-type="table"}\]. In fact, only 17 patients (6.09%, 95% CI, 5.7-8.8%) had all CVRFs under control. Approximately two thirds (192 patients, 68.81%; 95% CI, 64.6-72.9%) showed two or more uncontrolled CVRFs. Of the CVRFs, lipids were not optimally controlled in 60% of the patients (95% CI, 55-65%); diabetes was not optimally controlled (HbA1c \< 7.5%) in 61% (95% CI, 53-67%); arterial hypertension was not optimally controlled in 36% (95% CI, 27-46%); weight was not optimally controlled in 56% (95% CI, 45-55%); and 17% (95% CI, 13-20%) were active smokers. Intergroup differences in prevalence of metabolic syndrome and CVRFs were not statistically significant. Baseline to post-intervention changes in the prevalence of CVRFs were significant only in structured exercise training group (\>4CVRFs: 2.15% Vs 1.35%, *P* \< 0.05; 3CVRFs: 39.78% Vs 29.79%, *P* \< 0.05; 2CVRF: 27.95% Vs 33.78%, *P* \< 0.05) \[[Table 4](#T4){ref-type="table"}\], even here changes in proportion was mainly because of significant improved glycemic control.
{#F2}
######
Number of uncontrolled cardiovascular risk factors
D[ISCUSSION]{.smallcaps} {#sec1-4}
========================
Based on a number of large randomized controlled trials, current clinical guidelines\[[@ref5][@ref20][@ref21]\] acknowledge the therapeutic strength of exercise intervention. Despite this recommendation, it is estimated that only 30-40% of people with T2DM\[[@ref22]\] engage in regular exercise program indicating that their rate of participation is significantly below.
The primary finding from this randomized, exercise trial involving individuals with type 2 diabetes is that supervised structured (combination of aerobic and resistance exercises) exercise training were more efficacious than unstructured activity (increased physical activity) in achieving declines in HbA1c. Although both structured and unstructured training provide benefits, only the former was associated with significant reductions in HbA1c levels. Though in subgroup analysis (in either the structured or the unstructured group), we found that exercise-induced improvements in glycemic control were greater among persons with higher baseline hemoglobin A1c values (\>7%). Among persons with lower baseline hemoglobin A1c values (\<7%), only combined aerobic and resistance training improved values; unstructured training alone did not. The potential importance of good glycemic control for the reduction of cardiovascular disease risk was supported in a meta-regression study,\[[@ref23]\] which demonstrated an exponential relationship between fasting glucose concentrations and the incidence of cardiovascular events. Furthermore, cumulative benefit across other outcomes was also greater in the structured training group compared with other participating groups, thus may results in substantial improvement in CVD risk factors.
To our knowledge, this is the first randomized trial in Gujarat population involving individuals with type 2 diabetes to assess the effect of structured exercise training and unstructured activity interventions on glycemic control and to determine whether unstructured activity is associated with similar declines in HbA1c as compared with those associated with structured exercise. Our results demonstrate that in patients with type 2 diabetes, a recommendation to increase physical activity was beneficial (0.14% HbA1c reduction), but was not bringing significantly declines in HbA1c, whereas, structured training is associated with a HbA1c decline of 0.59%. (*P* = 0.030). A reduction in HbA1c of this magnitude is clinically significant and close to the difference between conventional and intensive glucose-lowering therapy in the United Kingdom Prospective Diabetes Study (UKPDS).\[[@ref24]\] In the UKPDS, subjects receiving intensive treatment with insulin or sulfonylurea\'s had HbA1c averaging 0.9% below the conventional treatment (7.0% vs 7.9%; *P* \< 001) and had significant reduction in diabetes-related clinical end points (40.9 Vs 46 events per 1000 patient-years; *P* =0.03).\[[@ref25][@ref26]\]
For each 1% increase in the level of HbA1c, the relative risk of CVD increases by 1.18%,\[[@ref27]\] whereas each 1% decrease in HbA1c levels is associated with a 37% reduction in micro-vascular complications and a 14% reduction in myocardial infarctions.\[[@ref28]\] Thus, our observed 0.59% reduction in HbA1c levels might be expected to produce 21% decrease in risk of micro-vascular complications and an 8% reduction in cardiovascular disease risk.
Because HbA1c reduction in type 2 diabetes is associated with improved insulin resistance\[[@ref20]\] and increased physical workout (structured exercise/unstructured activity) have distinct mechanisms to elicit these effects,\[[@ref20]\] it is expected that these interventions would result in greater metabolic effects, virtually bringing beneficial effects on all aspects of metabolic syndrome, one way or other. However, we did not find statistical support for the existence of a relationship between these exercise approaches (short term structured/unstructured) and an improved weight/BMI, or blood pressure, total cholesterol, HDL-C and LDL-C levels. There are two possible reasons for this discrepancy. Firstly, because many patients were on antihypertensive and lipid lowering drugs, and over the last decade, both lipid lowering therapy\[[@ref29][@ref30]\] and blood pressure lowering therapies\[[@ref31][@ref32]\] have been proven effective to improve cardiovascular outcome in T2DM patients. The effectiveness of these drugs may explain why the additive benefits of intensive glycemic control were not observed in this study. Similar results were obtained in even large and long-term clinical trials such as ACCORD, ADVANCE and VADT.\[[@ref33]\] Other reason may be that exercise approaches employed in this study were of short duration. To achieve greater changes, perhaps exercise interventions prescribing higher levels of exercise quantity might be necessary\[[@ref34]\] to positively affect body composition (weight/BMI) and lipid levels in individuals with T2DM on antihypertensive and/or lipid lowering drugs. Despite this, one of the important outcomes of this study is that exercise does not need to reduce body weight to have a beneficial impact on glycemic control. Exercise training decreases hepatic and muscle insulin resistance and increases glucose disposal through a number of mechanisms that would not necessarily be associated with body weight changes. The mechanisms were extensively reviewed by Ivy *et al*.,\[[@ref35]\] and include increased post-receptor insulin signaling,\[[@ref36]\] increased glucose transporter protein and messenger RNA,\[[@ref37]\] increased activity of glycogen synthase\[[@ref38]\] and hexokinase,\[[@ref39]\] decreased release and increased clearance of free fatty acids,\[[@ref35]\] increased muscle glucose delivery due to increased muscle capillary density\[[@ref39][@ref40][@ref41]\] and changes in muscle composition favoring increased glucose disposal.\[[@ref42][@ref43]\]
Nonetheless, this study has few limitations. Firstly, sampling may not be representative to all T2DM patients as the participants were probably more adherent to exercise and healthier on average than the general population with type 2 diabetes, as we excluded patients who were receiving insulin or who had advanced diabetes complications thus findings cannot be generalized to patients who cannot or do not wish to undertake exercise programs. Secondly, effect of different medications (antihypertensive, hypoglycemic and lipid lowering) were not taken into account, and fact is that a significant number of patients were receiving antihypertensive and lipid lowering drugs along with hypoglycemic medication and it well known that different drugs individually or in combination affect glucose metabolism (e.g., diuretics and beta blockers negatively affect glucose metabolism. In contrast, angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, calcium antagonists, and alpha-receptor blockers are thought to be metabolically neutral.). Still, the results of our study are particularly relevant, as our trial was not designed to study effects of different medications, rather our study have contributed to disclose metabolic changes in T2DM patients on routinely prescribed drugs. Thirdly, the proxy definition of diabetes mellitus was used in the study and auto antibodies screening such as Anti-GAD analyses was not assessed for patients. Fourthly, unstructured activity group did not receive a similar amount of supervision. Still, taken together, these results provide important information for clinical practice. Future research should include longer interventions with better quantification of body composition changes.
C[ONCLUSION]{.smallcaps} {#sec1-5}
========================
This RCT demonstrates important findings regarding the prescription of exercise training. First, unstructured activity and structured training, each are associated with HbA1c decrease, but the magnitude of this reduction is different across the two training modalities. Second, the extent of HbA1c reduction is positively related to the baseline value of HbA1c, combined aerobic and strength training, may be the preferred first-line treatment option for individuals with lower baseline HbA1c values, but if glycemic control is poor, then even increased physical activity would also improve the hemoglobin A1c value, but the structured training having combination of aerobic and resistance exercise would be better. Based on this, it is concluded that T2DM patients should be stimulated to participate in specifically designed exercise intervention programs. More clinical research is warranted to establish the efficacy of exercise intervention in a more differentiated approach for type 2 diabetes subpopulations within different stages of the disease and various levels of co-morbidity. In the interim, our analysis adds support to the idea that exercise is a cornerstone of diabetes therapy and findings of this study strongly support the ADA guidelines\[[@ref20]\] recommendation that in the treatment regimen of T2DM, optimal physical activity programs consist of regular physical activity combined with structured training must be added to supplement the effect of pharmacological intervention.
**Source of Support:** Nil
**Conflict of Interest:** None declared
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Visual categorization allows us to effortlessly interpret a wide range of sensory information into a limited number of meaningful categories. This process enables the efficient response to novel stimuli and is the foundation for visual perception and memory. Though initial category learning and well-practiced categorization have been well-studied at both the behavioral (Fabre-Thorpe, [@B11]; Ashby and Maddox, [@B2]) and neural levels (Reber et al., [@B51]; Kanwisher, [@B31]; Haxby et al., [@B23]; Seger, [@B55]), few studies have directly compared these time points in learning nor the large scale network changes in the transition from initially learned to well-practiced categorization.
Initial visual category learning relies on representations in early visual cortex and inferior temporal cortex (ITC), though other mechanisms employed and brain regions involved depend on the strategy that subjects use. According to COVIS (COmpetition between Verbal and Implicit Systems), a prominent theory of the neural basis of category learning, when the rule that separates categories is verbalizable, brain regions supporting working memory such as prefrontal cortex (PFC) and parts of the basal ganglia (BG) that include the head of the caudate nucleus are implicated (Ashby and Maddox, [@B2]). In contrast, when the category rule is not verbalizable, a procedural system that depends on the tail of the caudate nucleus is involved. Studies demonstrating initial category learning deficits in patients with damage to the PFC (Barcelo and Knight, [@B3]) and BG (Maddox et al., [@B41]) support this theory. The involvement of the medial temporal lobe (MTL) in initial category learning is unclear: some studies have demonstrated that MTL damage significantly impairs visual category learning (Hopkins et al., [@B27]) while others have shown that forms of category learning can occur without an intact MTL (Knowlton and Squire, [@B35]; Knowlton et al., [@B36]). Studies have also implicated the premotor cortex (PMC) in learning new visual categories, especially for categories that require a stereotyped motor response (Halsband and Freund, [@B20]; Boettiger and D\'Esposito, [@B5]).
Though PFC, PMC, BG, and MTL have shown to be important to initially learning visual categories, damage to these regions does not consistently produce deficits in the automatic recognition of well-established categories (Freedman et al., [@B14]; Zgaljardic et al., [@B66]; Squire et al., [@B56]). There are some reports, however, of patients with MTL damage that present with object discrimination difficulties, suggesting that the MTL may facilitate higher-level object processing (Lee et al., [@B37]). Additionally, lateral frontal regions have feedback connections to object recognition regions in ITC and studies have shown that lateral frontal regions modulate ITC to help facilitate successful recognition (Fuster et al., [@B16]; Barcelo et al., [@B4]; Gazzaley et al., [@B19]). A goal of the current study is to further characterize the supporting roles that the PFC, PMC, BG, and MTL play in well-established categorization and how the interactions between these regions and ITC change with category learning.
In contrast to patients with PFC, PMC, BG, and MTL lesions, patients with ITC lesions present with severe deficits in recognizing well-established categories (Warrington, [@B63]). These patients can often describe an object in their visual field in great detail, including color, texture, and shape but are unable to integrate this information to identify the object. Thus, ITC has a crucial role in organizing perceptual features into an integrated percept, essentially linking perception with recognition. In addition to this role in representing and recognizing well-established categories, ITC has also been shown to be modified with learning. Studies have generally demonstrated increases in ITC activity and enhanced neural tuning after extensive training with novel categories (Gauthier et al., [@B17]; Op de Beeck et al., [@B47]; Jiang et al., [@B29]) and when comparing category experts with category novices (Gauthier et al., [@B18]). These training-related changes have been demonstrated in the regions that preferentially respond to stimuli before training, such as the lateral occipital cortex for shapes, as well as in changes in the overall pattern of activity across ITC (Op de Beeck et al., [@B47]). Though these studies clearly demonstrate ITC changes with learning, the interpretation of these changes and overall organization of ITC are intensely debated (Tarr and Gauthier, [@B59]; Haxby et al., [@B23]; Kanwisher and Yovel, [@B33]). Current models suggest that extensive category learning is accompanied by changes in activity in regions representing that object or process, such as in the fusiform face area for faces, which may reflect the recruitment of new processes or modifications in object representations (Tarr and Gauthier, [@B59]; Kanwisher and Yovel, [@B33]). An alternative model suggests that changes distributed across all of ITC may be more important to category learning and representation than any specific region(s) (Haxby et al., [@B23]).
A goal of the current study is to compare the regional and network changes that occur during extensive category learning to better characterize the mechanisms of category learning in ITC.
In addition to characterizing changes within PFC, PMC, BG, MTL, and ITC we also seek to explore the overall changes in involvement of these regions in initial categorization as compared to later in learning. Though there are many variations of how this could occur, we propose two general possibilities: (1) there could be a shift from the network including PFC, PMC, BG, MTL, and ITC regions to a new network more focused on visual regions or (2) the same network could be utilized for visual categorization both early and late in learning and with practice there could be a redistribution of activity within the same network (for a review of these mechanisms see Kelly and Garavan, [@B34]). Several studies demonstrate a shift from the initial network of regions engaged to a new network when training involves recruiting different strategies at the beginning and end of practice. For example, Poldrack et al. ([@B49]) showed that during classification learning, initially MTL structures were recruited and later, when learning was more associative, activation shifted to more basal ganglia involvement. Additionally, Fletcher et al. ([@B12]) demonstrated a decrease in right fronto-parietal activity and increase in left fronto-parietal connectivity during the acquisition of artificial grammar rules. As far as category learning is accompanied by salient shifts in cognitive and neural strategies, we might expect to see a shift in the network from PFC, PMC, BG, MTL, and ITC early in learning to a network more focused around ITC regions since damage to these regions are the only lesions that consistently impair the retrieval of well-established categories.
An alternative to shifting to a new network is that the same network of PFC, PMC, BG, and MTL may serve as a permanent scaffold throughout all stages of learning: these regions initially facilitate category learning, retrieval, and decision-making and could continue to be involved in further updating, retrieving, and making decisions as categories become more well-established. This model suggests that a similar network including PFC, PMC, BG, MTL, and ITC is recruited throughout category learning, though parts of this network such as the PFC and BG may be less active late in learning when updating visual representations and retrieval demands are less pronounced (Wagner et al., [@B61]). Thus, another goal of the current study is evaluate the overall network changes that accompany visual category learning and decide between these two models.
Standard univariate analyses of regional activation changes could be useful to determine whether the same network or a different network is recruited with category learning. However, univariate analyses cannot assess if regions that change with learning are functionally connected. One approach to this issue is to measure the activity covariance between a region known to be involved in the task and the rest of the brain. By comparing this covariance map before and after training, it could help characterize the changes in task-related functional networks that accompany categorization training. A functional MRI method that is well-suited for this approach is coherence analysis. In coherence analysis, a reference or seed region is identified and the time series in this region is correlated, in the frequency domain, with the time series of every other voxel in the brain (Sun et al., [@B57], [@B58]). In this way, coherence analysis provides a task-related network with the seed region and we can measure how this network differs between initially learned and well-practiced categorization. The main advantage of coherence over simply correlating activity with a seed region is that coherence does not depend on the estimate of the hemodynamic response function or a model of neural activity. Thus, coherence is not affected by regional differences in hemodynamic responses whereas correlating activity is biased to produce high correlations between regions with similar hemodynamic responses (Muller et al., [@B45]). Also, by using partial coherence, we can measure the task-induced relationship between two regions while factoring out the stimulus-locked response (see Sun et al., [@B57] for further details).
In the current study, by using two categorization tasks, a novel task (100 training trials) and a well-practiced task (4250 training trials) with similar stimuli, we assess initially learned and well-practiced categorization in a single fMRI session. We chose to use faces as stimuli because faces have been shown to be obligatorily processed in a focal ITC region (right fusiform face area-FFA, Kanwisher et al., [@B32]; McKone et al., [@B42]), which can be functionally localized and assessed for learning-related changes. Additionally, the right FFA can be used as a seed region in the coherence analysis to identify categorization-related networks in order to assess how networks change with visual expertise.
Materials and Methods
=====================
Subjects
--------
Ten right-handed subjects ranging in age from 20--27 (*M* = 22.4) were recruited from the University of California, Berkeley. All participants were screened against medical, neurological, and psychiatric illnesses, and for use of prescription medications. All subjects gave written informed consent prior to participation in the study according to the procedures approved by the University of California, Berkeley Committee for Protection of Human Subjects.
Training session
----------------
Lifelike faces were created from the Faces composite face making software (Faces version 3.0). Using a template face, two categorization tasks were designed: the eyebrow--mouth task and the forehead--nose task (Figure [1](#F1){ref-type="fig"}). In the eyebrow--mouth task, the eyebrow height and mouth height varied in 2 mm increments to make 10 faces, while the other features remained constant (Figure [1](#F1){ref-type="fig"}A). In the forehead--nose task, the forehead height and nose length varied in 2 mm increments to make 10 faces while the other features remained constant (Figure [1](#F1){ref-type="fig"}B). In each task, subjects had to integrate information from both facial features to achieve optimal accuracy. To ensure that the results were not due to specific feature properties, half the subjects (5) trained with the eyebrow--mouth task and the other subjects trained with the forehead--nose task. At the beginning of training, subjects were shown a matrix of faces (for example, Figure [1](#F1){ref-type="fig"}A) and told a verbal description of the categorization task. For example, "faces with higher eyebrows and lower mouths are generally in category 1 and faces with lower eyebrows and higher mouths are generally in category 2." Next, subjects received 250 trials of self-paced computer training where they were presented with a face and had to respond by pressing one of two buttons with their right hand designated to each category. Feedback (blue "correct"/red "incorrect") was provided immediately after each trial to further facilitate category learning. After 250 trials, subjects took a break and received a feedback matrix that showed their accuracy and reaction time for each face and their monetary bonuses (+0.02 per correct trial, −0.01 per incorrect trial). Subjects used this information to try to boost their performance. Two 250 trial sessions were performed on the first training day and three were performed on each of the five training days thereafter. To ensure subjects were learning a general strategy and not memorizing individual faces, parts of the template face changed each day (see Figure [2](#F2){ref-type="fig"}). However, parts of the template faces that could potentially affect performance (such as the hair for the forehead--nose rule) were not changed.
{#F1}
{#F2}
Pre-fMRI session
----------------
The fMRI scan was performed on the seventh day after training initiation. Before scanning, subjects received 100 trials of feedback training on the task they learned for the last 6 days. After this review of the well-practiced task, subjects received the new categorization task. If subjects received 6 days of training on the eyebrow--mouth task, they were given the forehead--nose task (and vice versa). Identical to the procedure on day 1 with the well-learned task, subjects received explicit instructions to categorize faces and performed 100 trials of feedback training in order to attain a steady level of performance. During the pre-fmri training and in the fMRI scanner, subjects performed the well-practiced and initially learned tasks with the same template face (see Figure [1](#F1){ref-type="fig"}). At the beginning of each scan, subjects were told which task they were performing. Additionally, initially learned and well-practiced tasks were blocked (three scans in a row) to eliminate confusion on which task they were performing.
fMRI session
------------
While in the scanner, subjects performed the categorization tasks and a one-back task using blocks of face and scene stimuli (courtesy of Nancy Kanwisher, MIT). The categorization tasks in the scanner had several differences from training. First, two new face configurations for each task were introduced during the scanning session (see Figure [1](#F1){ref-type="fig"}, dashed boxes). These configurations had extreme feature values and were easy to classify. They were introduced to assess if subjects were learning general strategies (subjects would effortlessly apply the strategy to the novel configurations) or were memorizing specific feature configurations (subjects would notice the new configurations). In contrast to the training in which stimulus presentation was self-paced, in the scanner faces were displayed for 2 s to control for exposure duration across trials and participants. Also, no feedback was provided during scanning to promote consistent strategy use throughout the scanning session. Lastly, instead of a 200-ms inter-trial interval (ITI), a 12-s ITI was used to allow hemodynamic responses to return to baseline. Subjects performed three runs in a row of categorization with the initially learned task and three runs in a row of categorization with the well-practiced task. The order of initially learned task runs and well-practiced task runs were counterbalanced across subjects. Each run was 5 min 36 s and contained 24 trials.
After the categorization tasks, subjects performed a one-back task with faces and scenes. In the one-back task, subjects were shown 16-s blocks of either novel faces, scenes, or fixation. During the face and scene blocks, 20 images were shown for 500 ms with a 300-ms fixation cross between each image. To keep the subject\'s attention focused on the images throughout the task, subjects were instructed to press both thumbs on the response pad when the current image was the same as the image immediately preceding it (on average, one response was required for each block of images). There were seven blocks of each type and the scan lasted 5 min and 20 s.
Image acquisition
-----------------
Functional images were acquired using a gradient echoplanar sequence (TR = 2000 ms, TE = 28 ms, matrix size = 64 × 64, FOV = 22.4 cm) sensitive to BOLD contrast. Each functional volume consisted of 18 × 5 mm thick axial slices with 0.5 mm gap between each slice, providing whole brain coverage except for portions of the inferior cerebellum and the most superior extent of the parietal lobe. For each scan, 30 s of gradient and RF pulses preceded data acquisition to allow steady-state tissue magnetization and allow the subject to habituate to the scanner noise before performing the task. Stimuli were presented using Eprime software (Psychological Software Tools, Pittsburgh, PA) and all stimuli subtended a visual angle of approximately 5°. Participants viewed images in the scanner via back-projection onto a custom screen mounted at the participant\'s chest level and viewed via an angled mirror mounted inside the head coil. Responses were made using a hand-held fiber optic button box.
fMRI data analysis -- univariate
--------------------------------
Functional images acquired from the scanner were reconstructed from *k*-space using a linear time-interpolation algorithm to double the effective sampling rate. Image volumes were corrected for slice-timing skew using temporal sinc-interpolation. Data were preprocessed with SPM2 (Wellcome Department of Cognitive Neurology, London). Images were realigned using a six-parameter, rigid-body, least-squares alignment and spatially smoothed with an 8 mm FWHM Gaussian kernel. Univariate Statistical analyses were performed on individual subjects' data with a modified general linear model (GLM) as implemented in SPM2. The fMRI time series data was modeled as a series of events with a 2-s duration convolved with a canonical hemodynamic response function (HRF). The resulting functions were used as covariates in a general linear model. For analysis of the categorization task, two covariates were used to model the fMRI data: one for the initially learned task and one for the learned task. For the analysis of the face-scene one-back, two covariates were used: one for faces and one for scenes. These covariates, along with a basis set of cosine functions that high-pass filtered the data, were included in a general linear model. The least squares parameter estimates of height and best fitting canonical HRF for each condition were used in pairwise contrasts. For each subject, images of parameter estimates for each contrast of interest were spatially normalized to an EPI template based on the MNI305 stereotactic space (Collins et al., [@B8]). This was accomplished using a 12-parameter affine transformation together with a non-linear transformation involving cosine basis functions. Volumes were then resampled to 2-mm cubic voxels. These normalized contrasts were submitted to a second-level one-sample *t*-test, in which the mean estimate across participants at each voxel was tested against zero. For exploratory analyses (see Table [1](#T1){ref-type="table"}), regions of activation were identified using an uncorrected two-tailed threshold of *p* \< 0.001 and a minimum cluster size of at least five contiguous voxels. For visualization purposes, unthresholded (Figure [4](#F4){ref-type="fig"}) and thresholded (Figure [5](#F5){ref-type="fig"}) statistical parametric maps were overlaid onto a normalized T1-weighted image using MRIcro software (www.mricro.com). To determine the significance of regions in this exploratory analysis, we applied a whole-brain family-wise correction (*p* \< 0.05). Additionally, to determine the significance *a priori* regions, we performed a small volume correction (using family-wise error rate of *p* \< 0.05) for each region using MNI anatomical regions of interest. This included left and right hippocampi, parahippocampal gyri, middle frontal gyri, inferior frontal gyri, precentral gyri, fusiform gyri, inferior temporal gyri, caudate, and putamen.
Definition of face selective regions
------------------------------------
For each subject, the fusiform face area (FFA) was functionally defined using the contrast of novel faces minus scenes in the face-scene one-back task (Kanwisher et al., [@B32]). The FFA was defined by taking the peak voxel within the middle fusiform gyrus that responded more to the faces than to scenes and selecting the nine most significant contiguous voxels to the peak voxel. If the threshold had to be dropped below a *t*-value of 1.5 to find the peak voxel, the region was deemed unreliable and was excluded from further analyses. Activity within the FFA region of interest (ROI) was averaged across all voxels. This procedure was repeated for both hemispheres. Using this procedure, we successfully localized the right FFA in 10/10 subjects and left FFA 9/10 subjects. We performed ROI analyses on the right and left FFA for the univariate categorization model and for the right FFA for the coherence model (see below). For each subject, parameter estimates yielded by the GLM were extracted for each covariate and averaged within each ROI. These parameter estimates served as the dependent measures for across-subject "random-effects" analyses.
fMRI data analysis -- partial coherence
---------------------------------------
To identify networks of functional connectivity for the right FFA, we generated coherence and partial coherence maps using the task-specific coherence between the right FFA seed and all other voxels. One potential limitation of coherence is that it can be driven by a stimulus-locked response. For example, two regions could have high coherence because they are independently responsive to a stimulus rather than because they are part of a common functional network. Partial coherence takes into account the stimulus-locked response and estimates any remaining coherence between two time series after (see Sun et al., [@B57]) for further discussion of partial coherence). Visual categorization is a very rapid process and simply using coherence analyses may only allow the examination of learning-related changes in stimulus- and response-locked regions such as early visual/inferior temporal and motor regions. Using partial coherence would allow the examination of other regions shown to be involved in categorization that are not stimulus- or response-locked such as hippocampus, basal ganglia, prefrontal cortex, and premotor cortex. This led us to perform our analyses on the partial coherence maps rather than coherence maps. To identify practice-related changes in functional interactions, we then contrasted these task-specific partial coherence maps. This procedure is described briefly below, see Sun et al. ([@B57]) for further detail.
### Selection of reference voxels
For each subject, we used the average of the voxels within the right fusiform face area (rFFA), defined in the univariate analysis as described above, as a seed for the coherence analyses.
### Generation of condition-specific time-series
To generate condition-specific time-series, the data were separated into initially learned task and well-practiced task blocks. Each time-series each had a total of 1008 data points.
### Estimation of condition-specific coherence maps
Coherence is the normalized cross-covariance of two time-series and is defined by the magnitude-squared of the cross-spectrum divided by the power spectra of both time-series. Here, we calculate the coherence by estimating the cross-spectrum and power spectra using Welch\'s periodogram-averaging method.
Specifically, the power spectra were estimated by averaging the magnitude-squared discrete Fourier transform of short overlapping segments of the condition specific time-series from each voxel. Each segment was 64 data points in length, mean-centered and windowed with a 64-point Hanning window; the segments were overlapping by 32 data points. As compared to calculating the power spectrum with a single discrete Fourier transform of the entire time-series, averaging the spectra over several shorter segments decreases the variance of the power spectral estimate (Welch\'s method). Similarly, the cross-spectrum was estimated by averaging the cross-spectra of shorter segments, where the cross-spectra was calculated multiplying the discrete Fourier transform of short segments of the condition specific time-series of the reference region (rFFA) with the complex conjugate of the discrete Fourier transform of the condition specific time-series of all other voxels in the brain.
We then generated coherence maps for the seed ROI for each condition using the estimate of the band-averaged coherence within the bandwidth of the hemodynamic response function (0--0.15 Hz).
### Contrasts of condition-specific coherence maps
To identify changes in functional connectivity across conditions, we contrasted the initially learned task and well-practiced task coherence maps. We applied an arc-hyperbolic tangent transform to the coherency, as described in (Rosenberg et al., [@B52]), so that the difference of the coherency magnitudes approaches a zero-centered normal distribution. This transformation allows us to apply a parametric random-effects group analysis (a two-tailed, one-sample *t*-test) on the difference maps to determine regions with significantly different connectivity with the seed ROI across conditions. These difference maps were then normalized and submitted to a second-level one-sample *t*-test, in which the mean estimate across participants at each voxel was tested against zero. Identical to the univariate analysis, we first performed an exploratory analysis of the initially learned vs. well-practiced contrast using an uncorrected two-tailed threshold of *p* \< 0.001 and a minimum cluster size of at least five contiguous voxels. To determine the significance of regions in this exploratory analysis, we applied a whole-brain family-wise correction (*p* \< 0.05). Additionally, to determine the significance *a priori* regions, we performed a small volume correction (using family-wise error rate of *p* \< 0.05) for each region using MNI anatomical regions of interest. This included left and right hippocampi, parahippocampal gyri, middle frontal gyri, inferior frontal gyri, precentral gyri, fusiform gyri, inferior temporal gyri, caudate, and putamen.
Results
=======
Behavioral results
------------------
### Training
Subjects showed consistent improvement throughout categorization training in both reaction time and accuracy (Figure [3](#F3){ref-type="fig"}). At the end of training, subjects were highly accurate at the task (*M* = 96%, SD = 2.3) and obtained mean reaction times of less than a second (*M* = 950 ms, SD = 120). To determine if introduction of a new template face at the beginning of each day affected performance, we compared the first blocks from days 2--6 with the last blocks from days 1--5 and found no significant difference \[*t*(9) = 0.92, *p* \> 0.41; first blocks *M* = 89%, *M* = 1156 ms; last blocks *M* = 87%, *M* = 1232 ms\]. This suggests that subjects learned a general strategy and did not simply memorize the specific faces each day.
{#F3}
On day 7, immediately preceding the scanning session, subjects were able to successfully learn the new task within 100 trials (*M* = 84%, SD = 4.2; *M* = 1380 ms, SD = 171). This performance was significantly faster and more accurate than the initial training session of the learned task \[ACC *t*(9) = 2.92, *p* \< 0.05, RT *t*(9) = 3.35, *p* \< 0.05; learned task *M* = 77%, SD = 4.8; *M* = 1602 ms, SD = 193\], demonstrating that some skill or general task strategy was transferred from the practiced task to the new task.
### fMRI session
Comparing the well-practiced and initially learned tasks across all blocks demonstrates that subjects were adept at executing the well-practiced (*M* = 94%, SD = 2.8; *M* = 1202 ms, SD = 193) than the initially learned tasks (*M* = 86%, SD = 3.8; *M* = 1472 ms, SD = 211). As expected, performance was significantly faster \[*t*(9) = 2.92, *p* \< 0.05\] and more accurate \[*t*(9) = 3.35, *p* \< 0.05\] when performing the learned task. The behavioral results also suggest that subjects did not learn or drastically change their strategy during the fMRI session, as there was no significant difference between scans 1, 2, and 3 (see Figure [3](#F3){ref-type="fig"}). Reaction times for the well-practiced task were slower during scanning than the final training block \[*t*(9) = 4.95, *p* \< 0.05\], most likely because during training the task is self-paced and in the scanner the pace is slower and fixed. After the scanning session, subjects were asked if they were aware of the four novel face configurations introduced in the scanning session. None of the subjects reported noticing the novel configurations during scanning, though overall subjects performed nearly perfect at classifying these new faces (*M* = 99%, SD = 0.5). This suggests that subjects were learning a general rule rather than memorizing exemplars.
Neuroimaging results
--------------------
### Categorization-related regions
Figure [4](#F4){ref-type="fig"} shows the group-averaged univariate parameter estimate maps (positive only) and maps of partial coherence with the right FFA for the categorization task. The univariate and partial coherence analyses show very similar results. Both analyses implicate early visual regions, inferotemporal cortex, medial temporal lobe, thalamus, inferior parietal lobe, lateral prefrontal cortex, anterior cingulate cortex, supplementary motor area, premotor, and motor regions. However, there was positive partial coherence in several brain areas such as the anterior prefrontal cortex, posterior cingulate/precuneus, and temporal/parietal regions that were not activated in the univariate maps.
{#F4}
### Initially learned vs. well-practiced categorization performance -- univariate
Although similar networks were involved in both initially learned and well-practiced face categorization, there were several regions that differed in activity during execution of the initially learned versus well-practiced tasks (see Table [1](#T1){ref-type="table"} and Figure [5](#F5){ref-type="fig"}). However, after application of more strict corrections including the whole-brain family-wise error rate (*p* \< 0.05) and small volume corrections, no regions were significantly different between the two conditions. Though there was a slight trend for the left FFA to be more active during the learned task \[*t*(8) = 1.92, *p* = 0.22\], neither the right nor left FFA showed significant differences in activation between the two tasks (Figure [6](#F6){ref-type="fig"}).
######
**Regions that show activity and partial coherence differences between initially learned and well-practiced tasks (corrected at *p* \< 0.001). Hypothesized regions are italicized**.
Region Hemisphere MNI coordinates (*x,y,z*) BA *t*-Value
------------------------------------- ----------------------- ------------ --------------------------- ------- -----------
**UNIVARIATE ANALYSIS**
Well-practiced \> initially learned *Inf. Temporal Gyrus* Right 70, −32, −18 20 6.09
Inf. Parietal Right 56, −54, 42 40 6.06
Supra marginal Right 64, −38, 32 39/40 5.89
Mid. Temporal Gyms Right 66, −52, 10 21 5.83
Initially learned \> well-practiced Mid. Occipital Gyrus Right 30, −86, 16 19 5.31
Insula Right 34, 20, −8 48 5.24
*Mid. Frontal Gyrus* Left −38, 16, 32 44 4.97
*Inf. Frontal Gyrus* Left −34, 8, 36 44 5.68
*Premotor* Left −30, −26, 48 6 5.84
**PARTIAL COHERENCE ANALYSIS**
Well-practiced \> initially learned Mid. Occipital gyrus Right 36, −88, 28 19 5.96
*Fusiform* Right 50, −46, −20 37 5.96
*Fusiform* Left −26, −50, −24 37 5.94
*Inf. Temporal Gyrus* Left −44, −52, −4 37 5.42
Sup. Temporal Gyrus Right 50, −10, −10 21 8.28
Insula Left −36, −4, −2 48 7.55
Inf. Parietal Left −38, −38, 44 10 6.67
*Hippocampus* Left −16, −4, −22 49 5.54
*Premotor* Left −36, −6, 46 6 6.08
*Premotor* Right 66, 6, 28 6 7.30
Initially learned \> well-practiced Sup. Parietal Right 38, −54, 72 7 5.08
{#F5}
{#F6}
### Initially learned vs. well-practiced categorization performance -- partial coherence
The partial coherence analysis revealed a general increase in connectivity with practice: nearly all regions that showed significant coherence changes were more coherent with the right FFA during categorization with the well-learned (see Table [1](#T1){ref-type="table"} and Figure [5](#F5){ref-type="fig"}). Additionally, the functionally defined left FFA was significantly more coherent with the right FFA during the well-practiced task compared to the initially learned task \[*t*(8) = 3.22, *p* \< 0.05\]. A small but significant focus in the left anterior hippocampus (abutting the amygdala and parahippocampal gyrus) demonstrated more partial coherence with the right FFA during the learned task. Lastly, a right premotor region, corresponding to area 6 in the precentral gyrus, was significantly more coherent with rFFA during the well-learned task. Univariate and partial coherence values from these regions are displayed in Figure [6](#F6){ref-type="fig"}.
Increased signal-to-noise ratio (SNR) could have produced both increased univariate activity as well as increased partial coherence during the practiced. However, there was positive coherence in several brain areas such as the anterior prefrontal cortex, posterior cingulate/precuneus, and temporal/parietal regions that were not activated in the univariate maps, suggesting that partial coherence is a slightly different measure. Additionally, several regions show univariate changes in activity (see Table [1](#T1){ref-type="table"}) but no partial coherence changes and vice versa. An example that suggests that partial coherence shows categorization-specific results is the difference between the univariate and partial coherence analyses in the initially learned vs. well-practiced contrast. At a lower threshold (*p* \< 0.001 uncorrected), a left PMC region approximately 10 mm ventral to Rolando\'s genu area (the location of hand primary motor cortex, Herve et al., [@B26]) was more active for the initially learned task than the well-practiced task. This is likely due to the lengthened reaction time/duty cycle for the initially learned task and not specific to visual categorization. In contrast, a premotor region approximately 20 mm more anterior (outside the hand area) was implicated in the coherence analysis, showing greater partial coherence with the right FFA during the well-practiced task. This is consistent with studies showing greater premotor involvement during the execution of a stereotyped response, and suggests this region is specific to increased proficiency with the categorization task. Together, this suggests that univariate and partial coherence measure slightly different aspects of brain function and that partial coherence measures aspects that may be more specific to categorization.
Discussion
==========
The current results demonstrate that well-practiced visual categorization utilizes a similar network of brain regions as newly learned categorization rather than recruiting an alternative network. The regions implicated in the task-related univariate analysis were highly overlapping with the regions implicated in the partial coherence analysis, suggesting that regions active during categorization are also functionally connected with the right fusiform face area (FFA). The results also demonstrate an overall training-related increase in functional connectivity between task-related regions with learning. This suggests that visual category expertise occurs through strengthening connections between task-related regions. Though the network activity and connectivity remained largely the same for the initially learned and well-practiced task, there were also notable regional differences. Partial coherence analyses revealed increases in functional connectivity between task-related regions and the right FFA with practice. Specifically, the left FFA was more coherent with the right FFA during the learned task, suggesting a greater coordination between visual regions as a mechanism of visual category learning. Additionally, left MTL and right PMC showed greater coordination with the right FFA during the learned task, suggesting their involvement in the retrieval and representation of well-learned categories.
Overall networks involved early and late in categorization training
-------------------------------------------------------------------
The results demonstrate that a similar distributed network of regions supports well-practiced visual categorization, rather than distinct networks. This suggests that with visual category learning there are relatively small changes within the same network rather than a shift to a new network (Kelly and Garavan, [@B34]). This also suggests that similar cognitive and neural strategies were used for both tasks and throughout learning subjects became more efficient at utilizing the same strategy rather than developing a new strategy (Jonides, [@B30]). These results are consistent with recent fMRI studies of category prototype learning (Little and Thulborn, [@B39], [@B40]; Little et al., [@B38]). In these tasks, subjects explicitly learned to classify dot-patterns into one of four categories based on their similarity to category prototype patterns. Little and colleagues found that there was no significant change in either the distribution or magnitude of the BOLD signal between initial categorization (without any training) and well-practiced categorization (after either 750 or 2150 trials of training) (Little and Thulborn, [@B39], [@B40]; Little et al., [@B38]).
Though the current category learning task used faces instead of dot-patterns, had fewer categories and fewer exemplars, and more training, the current results are consistent with Little et al.'s findings. Our results further demonstrate that, in addition to eliciting a similar distributed activation pattern, the functional connectivity between these distributed regions is similar throughout learning. Together, this suggests that visual categorization is accomplished by a similar functionally connected network rather than the recruitment of a new network. Studies with additional training will be useful in further delineating the time course of the involvement of this network. Additionally, exploring if the current results generalize to other learning strategies/tasks and fMRI designs would be useful. Because the current task assessed initially learned and well-practiced performance in one fMRI session, there may have been some transfer in the learning strategy between the tasks, biasing the results to show the same network at different time points in learning. Taking separate scans early and late in learning would useful to compare to the current results. Also, the current study provided feedback and small monetary rewards to motivate rapid learning. However, categories can be learned incidentally in the absence of any reward and may rely on mechanisms distinct from reward-based category learning (Reber et al., [@B50]). Future studies would be helpful to determine if the recruitment of a consistent network throughout learning applies to these other forms of category learning.
In addition to the recruitment of a consistent network, it is also notable that there was an overall increase in partial coherence and relatively few decreases with learning (see Figure [5](#F5){ref-type="fig"} and Table [1](#T1){ref-type="table"}). This suggests that stronger coordination between task-related brain regions may be a general mechanism underlying improvements in performance. Recent studies of impaired populations support this idea (Bokde et al., [@B6]; DeGutis et al., [@B10]; He et al., [@B24],[@B25]). Bokde et al. ([@B6]) compared healthy controls to mild cognitive impairment (MCI) patients during a face matching task and found overall greater correlations between the right fusiform gyrus and task-related regions in healthy controls compared to MCI patients. DeGutis et al. ([@B10]) demonstrated that improvements in face processing in a prosopagnosic following rehabilitation training correspond with widespread increases in functional connectivity with the right FFA and very few decreases. These studies suggest that better performance is related to greater functional connectivity. However, this effect may be specific to complex tasks that require coordination among a broad network of regions. Schwartz et al. ([@B54]) extensively trained subjects on visual texture discrimination and found that learning is specific to the trained eye and that there is a decrease in functional connectivity between visual and frontal regions after training. Future studies varying task complexity and amount of training are necessary to better characterize the timing and task constraints that produce connectivity changes. Additionally, the use of repetitive transcranial magnetic stimulation (rTMS) to create virtual lesions in regions that significantly increase their coherence would be useful to assess the behavioral relevance of functional connectivity changes.
Though the results demonstrate an overall similar functionally connected network being employed early and late in category learning, there were a few notable increases in regional connectivity with learning in ITC, MTL, and PMC, which are described below.
Increased connectivity between face-selective regions
-----------------------------------------------------
Our results suggest that increased coordination between the left and right FFA support improvements in visual processing with category learning. Univariate ROI analysis of right and left FFA showed no significant difference in the magnitude of activity between initially learned and well-practiced task. This is in contrast to studies demonstrating that perceptual training generally increases activity in regions that represent or process the trained stimuli (Op de Beeck et al., [@B47]; Jiang et al., [@B29]). However, the results demonstrate significantly greater partial coherence between the right and left FFA during the well-learned task, suggesting increased coordination between the processes performed by or representations in these regions. In the right FFA, these processes or representations are likely related to computing specific spatial relations between facial features and those involved in integrating feature identities and spacings into a holistic percept (Yovel and Kanwisher, [@B65]). In the left FFA, these processes or representations are likely related to more parts-based analysis of faces, as this region has shown more activity when matching face parts compared to whole faces (Rossion et al., [@B53]). Thus, our findings suggest that training increased the coordination between holistic processing/representations in the right FFA and parts-based processing/representations in the left FFA.
This learning-related increased in coordination between ITC regions may reflect that well-established categories are represented by functional connections between ITC regions. Several studies have demonstrated reliable distributed patterns of both supra- and sub-threshold voxels throughout ITC when viewing a variety of well-established categories such as chairs, shoes, and scissors (Ishai et al., [@B28]; Haxby et al., [@B23]; Hanson et al., [@B22]; O\'Toole et al., [@B48]). Additionally, Op de Beeck et al. ([@B47]) showed that visual training produces distinct distributed activation patterns to trained stimuli not predicted by pre-training activations. This suggests that distributed activations are integral to category representations as well as learning new categories. A recent study suggests that these distributed activations are also functionally connected (Moeller et al., [@B43]). Using simultaneous microstimulation and fMRI in macaques, Moeller and colleagues stimulated face- and object-selective patches while measuring fMRI activity in other face and object patches. They found that microstimulation activated, to varying degrees, distinct networks of ipsi- and contralateral patches, demonstrating that these regions are functionally connected. The current results add to this finding by suggesting that as categories become more established, their functional connections in ITC strengthen and further suggest that functional connectivity changes may precede activity changes during visual learning. Future studies with additional training and other object categories would be useful to further characterize these learning-related changes in ITC.
High-level feature binding in left medial temporal lobe
-------------------------------------------------------
Similar to the left FFA, left medial temporal lobe (MTL) partial coherence increased with learning, likely due to increased proficiency in individuating the face stimuli. This finding is consistent with recent studies demonstrating that MTL regions are more active when making judgments about well-learned compared to poorly learned information (Yanike et al., [@B64]; DeGutis and D\'Esposito, [@B9]). This is also consistent with studies showing that MTL regions are important for visual perception, in particular making fine-grained object discriminations that rely on conjunctions of features (Bussey et al., [@B7]). These fine-grained discrimination mechanisms are used more when making more specific judgments about objects and faces (such as "zebra" instead of "living thing"). Correspondingly, left MTL has shown to be more active when making more specific compared to less specific object categorizations (Tyler et al., [@B60]) and is recruited during face individuation (Furl et al., [@B15]). Additionally, left MTL is more active in car and bird experts while viewing their stimuli of expertise compared to novices (Gauthier et al., [@B18]). Together, this suggests that the left medial temporal lobe may be involved in binding object features when making subordinate level category judgments and is increasingly involved with visual category learning. Future studies with additional types of visual categories will be important to determine if the involvement of the medial temporal lobe is specific to subordinate level expertise. Also, studies with higher resolution fMRI could determine the specific contributions of subregions of the medial temporal lobe (e.g. hippocampus, parahippocampus, or perirhinal cortex) to these effects.
This role of the left MTL in individuation and feature binding may be particular to categories that are initially learned in an explicit manner. Nomura et al. ([@B46]) compared explicit and implicit category learning and found increased left hippocampus (HC) during explicit category learning. Also, Reber et al. ([@B50]) showed that left HC is more involved in explicit rather than implicit category retrieval. Furthermore, DeGutis and D\'Esposito ([@B9]) found that left HC responded more when retrieving explicitly learned exemplars farther rather than closer to the category boundary. This HC category boundary effect was present even while performing a perceptual task with the same stimuli where subjects were not instructed to explicitly categorize the stimuli, suggesting the involvement of the left HC is relatively automatic. The current study extends these findings by suggesting that the left MTL is automatically recruited when successfully retrieving stimuli throughout extended learning and its involvement is likely particular to explicitly learned categories.
It is notable that, in contrast to the MTL, the basal ganglia (BG) did not show significant activity or connectivity changes with learning. Previous reports show that basal ganglia is integral to the initial stages of both implicit and explicit category learning (Seger, [@B55]) and is also recruited for categorization judgments after extended training (DeGutis et al., [@B10]). We previously demonstrated, after extensive category training on a similar task as the current study, that the BG was more responsive to faces close-to as opposed to far-from the category boundary (DeGutis et al., [@B10]). Unfortunately, this contrast was underpowered in the current design. It is possible that the BG distance-to-boundary effect is present both early and late in learning, resulting in no significant change in its involvement throughout learning in the current results. Alternatively, the BG may be less involved in the initially learned task because a fair amount of learning has already taken place in this task due to transfer from the well-practiced task, making initial learning less dependent on the BG.
Premotor cortex, retrieval, and response selection
--------------------------------------------------
In addition to ITC and left MTL, our results demonstrate that right PMC increased coherence with the right FFA during the well-practiced task, suggesting its involvement in the retrieval or representation of category responses. The right lateralization of this region suggests that it may not necessarily be related to subjects' right-handed category response. However, recently learned motor skills have been shown to be supported by regions specific to the learned movement whereas long-term learning (∼4 h of training over 3 weeks) has shown to involve more of a bi-hemispheric network (Floyer-Lea and Matthews, [@B13]). Thus, this region could be part of the premotor network involved in executing the category response. The increased coordination of the right PMC with the right FFA through learning is consistent with the finding that PMC damage impairs the retrieval of previously learned responses to visual stimuli (Halsband and Passingham, [@B21]). Additionally, these results are in line with Wallis and Miller ([@B62]) demonstration that responses of PMC neurons are selective to well-learned rules and that rule-selective activity in PMC precedes PFC and BG (Muhammad et al., [@B44]). These results also fit well with the late stage learning predictions of a recent model of categorization automaticity by Ashby et al. ([@B1]) (though not the early stage prediction of basal ganglia involvement) in which extended procedural category learning leads to strengthened cortical-cortical connections from sensory association areas directly to premotor cortex, in this case from right FFA to right PMC.
Summary
-------
Explicit categorization of recently learned visual categories is accomplished by a dynamic interaction of inferotemporal cortex, medial temporal lobe, prefrontal, premotor, and motor cortices. Both initially learned and well-practiced categorization recruits this network of regions, rather than the recruitment of a distinct network with practice. With practice, the connectivity between the right FFA and this network is strengthened. In particular, visual analysis is more efficiently accomplished perhaps due to increased connectivity between ITC regions. Additionally, subjects improve at individuating stimuli and retrieving categories likely through increased ITC and MTL connectivity. Finally, right premotor cortex shows increased connectivity with ITC, likely related to increasing efficiency in selecting the appropriate category response.
Conflict of Interest Statement
==============================
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We would like to thank Dr. Felice Sun for developing the coherence analysis and for help with applying this analysis to the current study. We would like to thank Dr. Shawn Ell and Dr. Charlotte Boettiger for feedback and useful comments. This work was supported by a grant from the National Institute of Health.
[^1]: Edited by: Russell A. Poldrack, University of California, USA
[^2]: Reviewed by: Stephen A. Engel, University of Minnesota, USA; Shawn Ell, University of Maine, USA; Russell A. Poldrack, University of California, USA
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper.
Introduction {#sec001}
============
Obligate intracellular bacteria exclusively replicate inside the cells of their host and they are predominantly transmitted through mother-offspring relationships \[[@pone.0128660.ref001]\]. As a consequence, some endosymbionts have adopted a strategy consisting of manipulating their host reproduction to maximise their transmission \[[@pone.0128660.ref002]\]\[[@pone.0128660.ref003]\]\[[@pone.0128660.ref004]\]. Indeed, reproductive parasite endosymbionts either favour the fitness of the infected females through cytoplasmic incompatibility (CI) \[[@pone.0128660.ref005]\]\[[@pone.0128660.ref006]\], or induce sex-ratio biases towards females in host progenies \[[@pone.0128660.ref007]\] through male-killing (death of male progeny) \[[@pone.0128660.ref008]\]\[[@pone.0128660.ref009]\], thelytokous parthenogenesis \[[@pone.0128660.ref010]\]\[[@pone.0128660.ref011]\], or feminization of genetic males \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref013]\]. The alphaproteobacterium *Wolbachia* that infects many arthropod species and filarial nematodes is the only endosymbiont known to induce all four of these effects \[[@pone.0128660.ref014]\]\[[@pone.0128660.ref015]\]\[[@pone.0128660.ref016]\]. Considered the most widespread endosymbiont on the planet, *Wolbachia* infects at least 40% of insect species \[[@pone.0128660.ref016]\], some chelicerate species (mites, spiders and scorpions \[[@pone.0128660.ref017]\]\[[@pone.0128660.ref018]\]\[[@pone.0128660.ref019]\]), and crustaceans, in which at least 61% of terrestrial isopod species are infected \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref020]\].
In feminization induced by *Wolbachia*, host genetic sex determination is manipulated by the endosymbiont which converts genetic males into functional phenotypic females. *Wolbachia-*induced feminization is the most common phenotype observed in terrestrial isopod crustaceans \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref021]\]\[[@pone.0128660.ref022]\]\[[@pone.0128660.ref023]\], even if CI strains have been described in three species \[[@pone.0128660.ref024]\]\[[@pone.0128660.ref025]\]\[[@pone.0128660.ref026]\]. Despite extensive studies carried out since the discovery of *Wolbachia* as the feminizing agent of the isopod *Armadillidium vulgare* in the 1970s \[[@pone.0128660.ref022]\]\[[@pone.0128660.ref027]\]\[[@pone.0128660.ref028]\], no underlying molecular mechanism has been described to date \[[@pone.0128660.ref002]\].
In *A*. *vulgare*, embryos that inherited *Wolbachia* developed into functional females, in which bacteria prevented the development of the androgenic gland. This gland secretes the androgenic hormone responsible for the differentiation of primary and secondary male sexual characters \[[@pone.0128660.ref002]\]\[[@pone.0128660.ref029]\]. *A*. *vulgare* gonad differentiation has been described in detail by Suzuki and Yamasaki \[[@pone.0128660.ref030]\] and it occurs within a period of ten to fifteen weeks after the release of juveniles from the female ventral pouch. Eight post-embryonic stages were defined, each corresponding to an intermolt stage. Gonads differentiate during stages 4 to 8 and androgenic glands progressively develop at the top of each testis from stage 6 to 8 \[[@pone.0128660.ref030]\]. *Wolbachia* is thought to act before or during sexual differentiation in order to inhibit male gonad differentiation, and convert genetic males into phenotypic females. When incomplete feminization occurs, this leads to intersexes ranging from sterile intersex males (iM) exhibiting female genital apertures and hypertrophied androgenic glands, to functional intersex females harboring typical male brushes on forelegs \[[@pone.0128660.ref031]\]\[[@pone.0128660.ref032]\]\[[@pone.0128660.ref033]\]. It has been hypothesized that this incomplete feminization is linked with low density of *Wolbachia* \[[@pone.0128660.ref034]\].
All these intersex phenotypes regularly observed in natural isopod populations can be experimentally produced by injection of *Wolbachia* in adult males. *Wolbachia*-induced iM may be obtained when the donor of the feminizing *Wolbachia* strain and the recipient belong to the same species \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref035]\]\[[@pone.0128660.ref036]\]. In the case of interspecific transfers of *Wolbachia* in terrestrial isopod hosts, the efficiency of the feminizing strains decreases with phylogenetic distance of the recipient, leading to *Wolbachia* elimination, an absence of effect, death of the recipient or conservation of the effect \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref034]\]\[[@pone.0128660.ref035]\]\[[@pone.0128660.ref036]\]. It has been shown that the feminizing strain of *A*. *vulgare*, *w*VulC, is able to induce the development of female secondary sexual characters in *Cylisticus convexus* adult males \[[@pone.0128660.ref035]\], whereas the CI-inducing strain of *C*. *convexus*, *w*Con, is able to induce CI when transinfected to *A*. *vulgare* \[[@pone.0128660.ref025]\]. These observations showed that the *Wolbachia*-induced phenotypes are due to the injected strains and do not depend on the genetic background of the hosts. *C*. *convexus* may therefore be a suitable model to uncover the mechanisms of feminization. Hence, it would be relevant to compare feminization induced by the same strain of *Wolbachia*, *w*VulC, in two distinct host species. However, induction of female secondary sexual characters by *w*VulC in *C*. *convexus* adult males does not necessarily mean that there is feminization of *C*. *convexus*, as feminization is the conversion of the male genetic sex into a fully functional female. Moreover, since feminization is supposed to take place before or during sexual differentiation, it is essential to distinguish between the feminizing action of the bacteria linked to sexual differentiation and an alternative action of the bacteria unrelated to sexual differentiation (*i*.*e*., without any link to feminization) but fortuitously happening during sexual differentiation. Such a confounding effect may be uncovered by analyzing a feminizing *Wolbachia* strain in two distinct genetic backgrounds having different timings of sexual differentiation.
In this study, we first described the developmental stages of *C*. *convexus* by comparison with those of *A*. *vulgare* and identified a one-stage shift of sex differentiation timing between the two species. *w*VulC from *A*. *vulgare* was also injected into uninfected *C*. *convexus* adult females and males to test whether: (i) *w*VulC can be vertically transmitted from mother to offspring, and (ii) *w*VulC induces feminization in its new host (feminization being assessed by the appearance of female secondary sexual characters in adult males, as well as the presence of intersexes and female bias in transinfected female progenies). Altogether, our results provide formal demonstration of a feminizing effect of *w*VulC in *C*. *convexus*.
Materials & Methods {#sec002}
===================
Transinfection of the *w*VulC Feminizing *Wolbachia* Strain in *C*. *convexus* {#sec003}
------------------------------------------------------------------------------
Isopods were reared at 20°C with food *ad libitum* (dead lime tree leaves and carrots) under natural photoperiod, except those in cross-breeding and juveniles which were reared under a 18L:6D photoperiod. Uninfected 10 month-old *C*. *convexus* males and females (from our laboratory line AW, derived from individuals caught in Villedaigne, France, in 1997) were infected by *w*VulC. The solution containing *Wolbachia* was prepared using ovaries from five *A*. *vulgare* females naturally harbouring the *w*VulC strain (from our laboratory line ZN, derived from individuals caught in Celles sur Belle, France, in 1991). The presence of *w*VulC was specifically checked by PCR and sequencing of the *wsp* sequence as described in Cordaux *et al*. \[[@pone.0128660.ref021]\]. Ovaries were crushed in 500 μL of Ringer solution (NaCl 394 mM; KCl 2 mM; CaCl~2~, 2H~2~O 2 mM; NaHCO~3~ 2 mM), and the resulting suspension was filtered through a 5 μm pore membrane (Sartorius Stedim Biotech). One microliter of the solution diluted twice with Ringer solution was directly injected in the general body cavity of *C*. *convexus* animals through a small hole pierced in each individual's cuticle using a thin glass needle. A total of 16 males and 37 females were transinfected with *w*VulC. As controls, 6 males and 20 females were injected with Ringer solution only. After 6 months, the 49 surviving females (33 treated animals and 16 controls) were cross-bred with uninfected males. At birth, juveniles were separated from the parents and at three months old, males and females were separated. Fourteen months after injection, the 11 surviving males were dissected in Ringer solution, and sexual characters were directly examined under binocular magnifier (50X); including gonad morphology, male copulating pleopods, size of uropods (larger in males than in females), male brushes on pereiopods, and female genital apertures. In order to improve the contrast of the image (especially for gonads which are white), dark-field was used.
Finally, 14 months after injection, total DNA of gonads of all surviving animals (treated and control males and females) and gonads, nervous cords, and head tissues of all progenies were extracted using the Qiagen DNeasy Blood and Tissue kit according to the manufacturer\'s instructions. Before DNA extraction, gonads of all progenies were examined under binocular magnifier (50X) as described above. Success of DNA extractions was checked by PCR amplification of the mitochondrial marker COI according to Folmer *et al*. \[[@pone.0128660.ref037]\]. The presence of *Wolbachia* was tested by PCR amplification of at least two of three bacterial genes (*wsp*, *ftsZ* or *recR*) using *Wolbachia*-specific primers, according to Badawi *et al*. \[[@pone.0128660.ref038]\], Braig *et al*. \[[@pone.0128660.ref039]\], and Werren *et al*. \[[@pone.0128660.ref040]\].
Developmental Study and Microscopy Observations {#sec004}
-----------------------------------------------
*C*. *convexus* females and males from our uninfected AW laboratory line were cross-bred. At birth, juveniles were separated from their parents. Sexual development of *C*. *convexus* was investigated in offspring of uninfected females and *w*VulC-transinfected females (obtained as described above). Molts were checked every 2 or 3 days from birth until stage 7. Ventral calcium white plates and evacuation of gut content occur a few days before each molt, which allows accurate developmental stage identification. Gonads of juveniles (3--14 individuals for each stage) were harvested right after molts for microscopic observations. Each gonad was fixed for 2h in a fixative solution (9% glutaraldehyde, 0.3 M sodium cacodylate, 3% NaCl; 1/1/1). Gonads were then washed for 1h in PBS solution and for 10 minutes in bi-distilled sterile water. Tissues were mounted between slide and slip cover in a citifluor drop before bright-field light microscope observation (200X for stage 1--3 gonads and 100X for stage 4--7 gonads). Photographs were taken as mosaics with ZEN software (ZEISS) when the object was larger than the microscopic field. Images were reconstructed with ZEN.
Statistical Tests {#sec005}
-----------------
The proportion of females and intersexes (F+I) was calculated for each brood. To estimate a bias towards the proportion of F+I considering animals of all progenies, we compared with a χ^2^ test the observed frequencies of males and F+I of infected and uninfected individuals with the mean frequency observed in the controls. We also compared for each type of clutch (clutch with infected individuals and clutch with no infected individuals) the total frequencies of males and F+I with the mean frequency observed in the controls.
Results {#sec006}
=======
Sexual differentiation of *C*. *convexus* {#sec007}
-----------------------------------------
We investigated gonad differentiation from birth to gonad maturity in 43 *C*. *convexus* individuals produced by uninfected females. The development was followed during the first 7 molts (17 weeks post-birth) and used to define post-embryonic developmental stages. Among the 29 individuals for whom the sex could be assessed (15 females and 14 males), there was a balanced sex-ratio (51.7% females; χ^2^ = 0.0345 df = 1; p = 0.85).
A few hours after birth, juveniles experienced a first ecdysis molt (defining start of stage 1). Duration of the following intermolts ranged from 2 to 3 weeks. Until the end of stage 2, juveniles were still mancas since they harbored only 6 pairs of functional pereiopods. At stage 2, the 7th pereiopod pair developed but remained non-functional and folded down the ventral face. Finally, stage 3 marked the first stage of larvae, as the 7th pereiopod pair was functional. Female genital apertures and male copulating pleopods were visible at stage 5. Other secondary sexual characters such as foreleg brushes or longer uropods in males developed after stage 7.
Individuals from each stage were also dissected to follow gonad development. During stages 1 and 2, no distinct morphological difference between male and female gonads was observed. Juvenile gonads constituted undifferentiated tubes slightly swollen in the middle and harboring suspensory filaments from which testis and *vas deferens* in males, or oviduct in females will differentiate (Fig [1a](#pone.0128660.g001){ref-type="fig"} and [1b](#pone.0128660.g001){ref-type="fig"}). Stage 3 began \~30 days after birth and marked the start of gonad differentiation. In the male gonad, the first suspensory filament at the top of the gonad was elongating into testis, whereas the second one was just starting to differentiate. At the opposite end, the gonad was elongating into *vas deferens*. The central swollen region also tended to elongate ([Fig 1c](#pone.0128660.g001){ref-type="fig"}). At stage 4, two weeks later, the first two testis were fully formed whereas the third one was still differentiating ([Fig 1d](#pone.0128660.g001){ref-type="fig"}). At stage 5, two weeks later, testis 1 and 2 now harbored an androgenic gland. At the bottom of the gonad, the seminal vesicle was still differentiating. The onset of spermatogenesis was also observable in the first testis ([Fig 1e](#pone.0128660.g001){ref-type="fig"}). After three more weeks, at stage 6, spermatogenesis started in the three testis, which all presented an androgenic gland at their top. The seminal vesicle was now fully differentiated, as the lumen (still empty) was visible along its full length ([Fig 1f](#pone.0128660.g001){ref-type="fig"}). Finally, at stage 7, sperm filled the seminal vesicle and *vas deferens* ([Fig 1g](#pone.0128660.g001){ref-type="fig"}).
{#pone.0128660.g001}
In females, at stage 3, the oviduct was differentiating and the gonad was elongating (Fig [1h](#pone.0128660.g001){ref-type="fig"} and [1i](#pone.0128660.g001){ref-type="fig"}). From stage 4, other suspensory filaments regressed ([Fig 1i](#pone.0128660.g001){ref-type="fig"}). Then, at stage 5, the very beginning of oogenesis was observed in the upper part of the gonad ([Fig 1j](#pone.0128660.g001){ref-type="fig"}). At stage 6, the second and third suspensory filaments almost disappeared and the posterior suspensory filament was regressing ([Fig 1k](#pone.0128660.g001){ref-type="fig"}). Finally, at stage 7, gonads continued to mature and newly developed oocytes were arranged in one line along the longitudinal axis of the ovary ([Fig 1l](#pone.0128660.g001){ref-type="fig"}).
*w*VulC-Induced Feminization of *C*. *convexus* Adult Males {#sec008}
-----------------------------------------------------------
Bouchon *et al*. \[[@pone.0128660.ref035]\] showed that injection of *w*VulC into *C*. *convexus* adult males induced the formation of female genital apertures and hypertrophied androgenic glands after 14 months (which correspond to iM observed in *A*. *vulgare* males transinfected with *w*VulC). While this observation led the authors to suspect a feminizing effect of *w*VulC in *C*. *convexus*, it does not constitute a formal demonstration of a feminizing effect which requires: (i) transmission of the bacteria to the next host generation and, (ii) full conversion of genetic male individuals into functional females during development.
First, we injected *w*VulC into *C*. *convexus* adult males, as did Bouchon *et al*. \[[@pone.0128660.ref035]\]. The difference was that while the recipient individuals used in Bouchon *et al*. \[[@pone.0128660.ref035]\] originated from a *C*. *convexus* line infected with the *w*Con *Wolbachia* strain, here we used a *Wolbachia*-free *C*. *convexus* line, to exclude any putative interaction between the transinfected *w*VulC strain and the native *w*Con strain that may have contributed to the observations made by Bouchon *et al*. \[[@pone.0128660.ref035]\]. Consistent with the observations by Bouchon *et al*. \[[@pone.0128660.ref035]\], 10 of 11 *w*VulC-transinfected *C*. *convexus* males exhibited hypertrophied androgenic glands (Fig [2a](#pone.0128660.g002){ref-type="fig"} and [2b](#pone.0128660.g002){ref-type="fig"}), and 6 of them harbored one (2 individuals) or two (4 individuals) female genital apertures ([Fig 2c](#pone.0128660.g002){ref-type="fig"}; [Table 1](#pone.0128660.t001){ref-type="table"}). Other male sexual characters, such as the size of uropods, copulating pleopods, and pereiopod brushes were not altered. *Wolbachia* was detected by PCR in 9 of the 10 transinfected males exhibiting an intersexual phenotype ([Table 1](#pone.0128660.t001){ref-type="table"}). All males harboring female genital apertures were infected by *Wolbachia*. *Wolbachia* was not detected in any of the controls, which exhibited all expected male sexual characters (Fig [2d](#pone.0128660.g002){ref-type="fig"},[2e](#pone.0128660.g002){ref-type="fig"} and [2f](#pone.0128660.g002){ref-type="fig"}; [Table 1](#pone.0128660.t001){ref-type="table"}).
{#pone.0128660.g002}
10.1371/journal.pone.0128660.t001
###### Characteristics of transinfected *C*. *convexus* adult males.
{#pone.0128660.t001g}
Treatment AG[^a^](#t001fn001){ref-type="table-fn"} fga[^b^](#t001fn002){ref-type="table-fn"} *Wolbachia* [^c^](#t001fn003){ref-type="table-fn"} 2nd Pr[^d^](#t001fn004){ref-type="table-fn"} Ur1 (μm)[^e^](#t001fn005){ref-type="table-fn"} Ur2 (μm)[^e^](#t001fn005){ref-type="table-fn"} Pl (μm)[^e^](#t001fn005){ref-type="table-fn"} Telson (μm)[^e^](#t001fn005){ref-type="table-fn"}
----------- ------------------------------------------ ------------------------------------------- ---------------------------------------------------- ---------------------------------------------- ------------------------------------------------ ------------------------------------------------ ----------------------------------------------- ---------------------------------------------------
*w*VulC H 2 \+ brush 1092 2206 1332 3799
H 2 \+ brush 1852 1893 1231 3724
H 2 \+ brush 2344 2293 1461 3812
H 2 \+ brush 856 779 1428 3668
H 1 \+ brush 2441 2392 1302 3655
H 1 \+ brush 2540 NA 1279 3738
H 0 \+ brush 1611 2135 1180 3741
H 0 \+ brush 2453 2067 1043 3491
H 0 \+ brush 2079 1933 1131 3928
H 0 \- brush 2727 2509 1322 3781
N 0 \- brush 2229 2125 973 3257
Ringer N 0 \- brush 2465 2299 963 3316
N 0 \- brush 2373 1367 942 3272
N 0 \- brush 2071 2176 972 3274
N 0 \- brush NA NA NA NA
^a^ AG = Androgenic gland; N = Normal androgenic gland; H = Hypertrophied androgenic gland
^b^ Number of female genital apertures (fga)
^c^ Presence of *Wolbachia* detected by PCR using two molecular markers: *wsp* and recR.
^d^ Presence of brush on the second pair of pereiopods (Pr)
^e^ Size of uropods (Ur1 and Ur2), pleopod 2 (Pl) and telson
Vertical Transmission of *w*VulC, Sex-Ratio Bias and Intersexes in *C*. *convexus* {#sec009}
----------------------------------------------------------------------------------
To investigate whether *w*VulC can be vertically transmitted in *C*. *convexus*, *w*VulC was injected in adult females originating from the same *Wolbachia*-free *C*. *convexus* line used in the previous experiments. Vertical transmission of *w*VulC to offspring of 10 transinfected mothers, whose infection was confirmed by PCR, was investigated by PCR testing of all 159 offspring (64 males, 93 females and 2 intersexes based on external sexual characters) once they were adult (*i*.*e*., more than 6 months old). *w*VulC was detected in 25 individuals; 21 of which presented a female phenotype, 3 a male phenotype and 1 an intersex phenotype ([Table 2](#pone.0128660.t002){ref-type="table"}).
10.1371/journal.pone.0128660.t002
###### Vertical transmission of *w*VulC in *C*. *convexus* and proportion of females and intersexes (F+I).
{#pone.0128660.t002g}
Treatment *Wolbachia* [^a^](#t002fn001){ref-type="table-fn"} M F I Total Proportion F+I
------------------------------------- ---------------------------------------------------- --------- --------- -------- --------- ----------------
mother *w*VulC +; 3F 0 3 0 3 1.00
offspring with at least 1 *w*VulC + 1F 0 8 1 9 1.00
7F; 1I 1 12 4 17 0.94
5F 5 15 0 20 0.75
3M; 4F 5 9 1 15 0.67
1F 6 11 0 17 0.65
**Total** **3M; 21F; 1I** **17** **58** **6** **81** **0.79**
mother *w*VulC +; 6 11 7 24 0.75
offspring *w*VulC- 11 18 1 30 0.63
6 5 0 11 0.45
10 1 2 13 0.23
**Total** **33** **35** **10** **78** **0.58**
mother *w*VulC-; 11 22 33 0.67
offspring *w*VulC- 15 26 41 0.63
17 20 37 0.54
11 13 24 0.54
11 12 23 0.52
18 18 36 0.50
11 8 19 0.42
9 3 12 0.25
**Total** **103** **122** **225** **0.54**
^a^ Presence of *Wolbachia* detected by PCR using two molecular markers: *wsp* and *ftsZ*.
F = Female, M = Male, I = Intersex.
Strikingly, after dissection of the 159 individuals, we found that 14 individuals initially scored as males based on external sexual characters actually exhibited an intersex phenotype at the gonad level, harboring a whole gradient of male and female gonad morphology such as: male-like gonads with hypertrophied androgenic glands ([Fig 3a](#pone.0128660.g003){ref-type="fig"}), male-like gonads with oocytes in the seminal vesicle ([Fig 3b](#pone.0128660.g003){ref-type="fig"}), and co-occurence of both male-like and female like gonads within the same individual (with both male copulating pleopods and female genital apertures in two individuals; [Fig 3c](#pone.0128660.g003){ref-type="fig"}).
{#pone.0128660.g003}
With respect to the 159 offspring from the 10 *w*VulC transinfected *C*. *convexus* mothers, the overall proportion of F+I was 68.6%, which was significantly higher (Χ^2^ = 13.16; df = 1; p\<10^-3^) than in the 225 controls (54.2%) ([Table 2](#pone.0128660.t002){ref-type="table"}). When considering the 25 individuals in which *w*VulC was detected by PCR, the proportion of F+I was 88.0%, which was significantly higher than in the 134 individuals in which *w*VulC was not detected (64.9%; Χ^2^ = 5.85; df = 1; p = 0.016) or the 225 controls (54.2%; Χ^2^ = 11.49; df = 1; p\<10^-3^). Moreover, the proportion of F+I in the 6 broods in which *w*VulC was detected in at least one individual (79.0%) was significantly higher than in the 4 broods in which *w*VulC was not detected (57.7%; Χ^2^ = 15.08; df = 1; p\<10^-3^) or in the 8 control broods (54.2%; Χ^2^ = 20.05; df = 1; p\<10^-3^) ([Table 2](#pone.0128660.t002){ref-type="table"}). The offspring from the controls presented a balanced sex-ratio (54.2% females, no intersex; Χ^2^ = 1.60; df = 1; p = 0.21).
To further investigate the feminizing action of *w*VulC in *C*. *convexus* during early development, gonads of 36 juveniles from a set of 5 *w*VulC-transinfected mothers were dissected during developmental stages 3 to 7. The number of post-embryonic molts was similar for juveniles produced by transinfected mothers compared to juveniles produced by uninfected mothers. Among the 36 dissected juveniles, 6 males and 16 females exhibited normal gonads, while the remaining 14 individuals were intersexes harboring altered gonads, ranging from unusual morphology (hypertrophy of androgenic glands; [Fig 3d](#pone.0128660.g003){ref-type="fig"}) to intersexual morphology (co-development of spermiduct and oviduct; Fig [3e](#pone.0128660.g003){ref-type="fig"},[3f](#pone.0128660.g003){ref-type="fig"} and [3g](#pone.0128660.g003){ref-type="fig"}). Such alterations in gonad morphology were never observed in juveniles produced by uninfected mothers. As for adults produced by transinfected *C*. *convexus* females, we observed a significant bias towards F+I in juveniles (Χ^2^ = 13.3; df = 1; p\<10^-3^) produced by transinfected mothers (83.3%) compared with the juveniles produced by uninfected mothers for which we also investigated gonad morphology during development (51.7%).
Discussion {#sec010}
==========
Comparison of Sexual Differentiation Timing between *C*. *convexus* and *A*. *vulgare* {#sec011}
--------------------------------------------------------------------------------------
*C*. *convexus* gonad morphology is the same as that observed in several Oniscidea, including *A*. *vulgare* \[[@pone.0128660.ref030]\]\[[@pone.0128660.ref041]\]. Indeed, male gonads present three testes, each surmounted by an androgenic gland. Each testis develops from suspensory filaments inserted onto the seminal vesicle, which extends by the *vas deferens* \[[@pone.0128660.ref041]\]. The duration of *C*. *convexus* embryonic development (\~17 weeks), from marsupium release to stage 7, is apparently longer than that of *A*. *vulgare* (\~10 weeks, according Suzuki and Yamasaki \[[@pone.0128660.ref030]\]). However, absolute duration is not directly comparable between the two studies. This is because post-embryonic development depends on rearing conditions such as temperature, which was different between the *A*. *vulgare* experiments (25°C) \[[@pone.0128660.ref030]\] and our study (21°C). Using our rearing conditions, *A*. *vulgare* post-embryonic development lasts for \~15 weeks; 5 weeks longer than in Suzuki and Yamasaki \[[@pone.0128660.ref030]\]. Nonetheless, molt stages are identical in the two species, demonstrating that they are a stable and robust temporal measurement of post-embryonic development.
The most significant difference observed between *C*. *convexus* and *A*. *vulgare* is that *C*. *convexus* post-embryonic development lasts for 7 stages whereas that of *A*. *vulgare* lasts for 8 stages ([Fig 4](#pone.0128660.g004){ref-type="fig"}) \[[@pone.0128660.ref030]\]. In *C*. *convexus*, undifferentiated gonads start to differentiate at stage 3, one stage earlier than in *A*. *vulgare* ([Fig 4](#pone.0128660.g004){ref-type="fig"}) \[[@pone.0128660.ref030]\]. Gonad differentiation continues until stage 5 when testes are filled with mature sperm, which occurs at stage 6 in *A*. *vulgare* \[[@pone.0128660.ref030]\]. Finally, differentiated gonads mature during stages 6 and 7 in *C*. *convexus*, while this occurs during stages 7 and 8 in *A*. *vulgare* ([Fig 4](#pone.0128660.g004){ref-type="fig"}) \[[@pone.0128660.ref030]\]. Hence, regarding gonad development, *C*. *convexus* sexually differentiates earlier (one stage shift) than *A*. *vulgare*, during the same number of stages ([Fig 4](#pone.0128660.g004){ref-type="fig"}). However, differentiation of female genital apertures and male copulating pleopods occurs at stage 5 in both species \[[@pone.0128660.ref042]\].
{#pone.0128660.g004}
Another study previously described the beginning of gonad morphological differentiation in two other Oniscidea species \[[@pone.0128660.ref041]\]. In *Porcellio dilatatus*, gonad morphology is similar to *C*. *convexus* and *A*. *vulgare*, while *Helleria brevicornis* has a different gonad morphology. Indeed, *H*. *brevicornis* male gonads also presents three testes implanted on the seminal vesicle, but only harbor two androgenic glands, one attached in the middle of the seminal vesicle and the other one on the *vas deferens* \[[@pone.0128660.ref041]\]. Sexual differentiation of *P*. *dilatatus* begins at the same stage as *C*. *convexus* (stage 3), whereas *H*. *brevicornis* begins one stage earlier, at stage 2 \[[@pone.0128660.ref041]\]. Altogether, these results indicate that there is a fair amount of diversity in sexual differentiation timing in terrestrial isopods.
Vertical Transmission of *w*VulC in *C*. *convexus* {#sec012}
---------------------------------------------------
We detected *Wolbachia* in \~16% of descendants produced by transinfected *C*. *convexus* females, showing that *w*VulC had been vertically transmitted from the maternal lineage. However, *w*VulC transmission rate in the heterologous host *C*. *convexus* (\~16%) is currently lower than in its natural host *A*. *vulgare* (\~82%) \[[@pone.0128660.ref021]\]. Transinfection success with the native *w*Con strain would likely be higher, but the low *w*VulC transmission rate in *C*. *convexus* is not surprising as it concerns the first generation of infection in the host, produced by horizontally infected mothers. Indeed, the host and the bacteria are not adapted to each other as it is the first generation of vertical transmission. After artificial horizontal transfer, it is common that the symbiont is lost through generations \[[@pone.0128660.ref043]\]\[[@pone.0128660.ref044]\]\[[@pone.0128660.ref045]\]\[[@pone.0128660.ref046]\], although sometimes the symbiont can durably settle in the host \[[@pone.0128660.ref047]\]\[[@pone.0128660.ref048]\]\[[@pone.0128660.ref049]\]\[[@pone.0128660.ref050]\]. Even with a low transmission rate, these results indicate that *w*VulC can pass two out of three filters that allow the bacteria to settle durably in the new host after horizontal transfer \[[@pone.0128660.ref051]\]. First, *w*VulC maintains itself in *C*. *convexus*, including in germinal cells \[[@pone.0128660.ref001]\]. Second, *w*VulC is transmitted to the next generation. It is noteworthy that *w*VulC does not invade *Armadillo officinalis*, as transinfected *Wolbachia* seem to totally disappear \[[@pone.0128660.ref035]\]. Moreover, *w*VulC also failed to invade a new species when transferred to *P*. *dilatatus* due to the death of the recipient host after an excessive autophagic reaction \[[@pone.0128660.ref052]\]\[[@pone.0128660.ref053]\]. However, previous experimental transfers showed that *w*VulC can also efficiently invade *A*. *nasatum* and be transmitted to progenies whose males are feminized by the bacteria \[[@pone.0128660.ref034]\]. Therefore, the third filter for *w*VulC to pass in *C*. *convexus* would be the ability to induce feminization in its new host, as this would increase bacterial transmission.
Feminization Effect of *w*VulC in *C*. *convexus* {#sec013}
-------------------------------------------------
In terrestrial isopods, *w*VulC injection in males strongly disturbs male secondary sexual characters, leading to intersexes in several recipient hosts, including *C*. *convexus* \[[@pone.0128660.ref035]\]\[[@pone.0128660.ref036]\]. In this study, all *C*. *convexus* transinfected adult males in which *Wolbachia* was detected presented hypertrophied androgenic glands. In contrast, only two thirds of them harbored one or two female genital apertures, suggesting a gradual feminizing effect of the bacteria. As hypothesized by Rigaud *et al*. \[[@pone.0128660.ref027]\], feminizing efficiency might rely on *Wolbachia* density. Indeed, transinfection of serial dilutions of feminizing *Wolbachia* extracts in *A*. *vulgare* adult males showed that the more the extract was diluted, the less efficient feminization of males became \[[@pone.0128660.ref027]\].
We also investigated feminization effect of inherited *w*VulC in progenies of transinfected *C*. *convexus* mothers. We showed that from the very first generation of *C*. *convexus* which has vertically inherited *Wolbachia*, *w*VulC seems to be able to induce a sex-ratio bias towards females ([Table 2](#pone.0128660.t002){ref-type="table"}). However, the correlation of *Wolbachia* presence only in females and intersexes is not perfect. Infected males harboring no morphological abnormalities have never been observed in *A*. *vulgare* natural populations hosting *w*VulC \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref021]\]. Nonetheless, terrestrial isopod males can be infected with feminizing strains of *Wolbachia* in *Oniscus asellus* and *Porcellionides pruinosus* \[[@pone.0128660.ref054]\]\[[@pone.0128660.ref055]\]\[[@pone.0128660.ref056]\]. This change of feminization efficiency induced by *w*VulC is not surprising as the extended phenotype induced by *Wolbachia* in their native host is rarely conserved in a transinfected host, especially for feminization \[[@pone.0128660.ref034]\]\[[@pone.0128660.ref057]\]. Thus, perfect feminization may require a specialized relationship between the host and the symbiont as it requires fine manipulation of host physiology.
Sometimes, in *A*. *vulgare*, incomplete feminization can occur as demonstrated by the presence of intersexes \[[@pone.0128660.ref012]\]\[[@pone.0128660.ref033]\]. In *C*. *convexus*, about 10% of the progenies produced by transinfected mothers were intersexes. It is noteworthy that intersexes have never been observed; neither in controls, nor in uninfected *C*. *convexus* lines raised in our laboratory for almost 20 years. Intersexes are expected to be derived from incomplete feminization caused by *Wolbachia* infection, although we detected *w*VulC by PCR in only one intersex. If so, it is possible that the transmission rate we calculated (\~16%) is underestimated.
The feminizing effect of *w*VulC can be directly observed in *C*. *convexus* juveniles produced by transinfected mothers, as attested by the altered gonads with intersexual phenotype during post-embryonic development. This suggests that *w*VulC alters gonad differentiation of the new host, as in *A*. *vulgare* \[[@pone.0128660.ref058]\]. As fewer normal males (16.6%) than expected (48.3%) were observed in these *C*. *convexus* juveniles, we assumed that the 14 individuals with intersexual gonads are genetic males that are being feminized (as expected in the natural host *A*. *vulgare*) \[[@pone.0128660.ref002]\].
Sex-ratio bias towards females in infected progenies, intersexual phenotypes in progenies throughout development, and appearance of female secondary sexual characters in transinfected adult males, together clearly indicate that *w*VulC has a feminizing effect in *C*. *convexus*. Feminization occurs after vertical inheritance, despite an earlier sexual differentiation when compared to *A*. *vulgare*. This suggests that feminization mechanisms induced by *w*VulC are not highly specialized with respect to host sexual differentiation \[[@pone.0128660.ref026]\]. The absence of coevolution between *w*VulC and its heterologous host *C*. *convexus* may explain a lower transmission rate and a less efficient feminizing effect than in the native host of *w*VulC. We are currently obtaining additional generations of *w*VulC-infected *C*. *convexus* isopods using crosses involving infected female individuals, which should select for *Wolbachia* genotypes best adapted to the new host. The establishment of a stable line of *C*. *convexus* infected with the feminizing *w*VulC strain constitutes an asset to study the molecular mechanisms of feminization. Indeed, it will allow further comparative studies between the natural and well-studied model *w*VulC/*A*. *vulgare* and the new model *w*VulC/*C*. *convexus*, which has different sexual differentiation timing. Such an experimental system will enable discrimination between the confounding effects of the feminizing action of *w*VulC linked to sexual differentiation from an alternative action unrelated to feminization, but fortuitously occurring during sexual differentiation.
We thank Joanne Bertaux, Maryline Raimond and Catherine Debenest for technical assistance and discussion about the results. We are also grateful to Ben Makepeace (University of Liverpool) for comments on the manuscript and English corrections.
[^1]: **Competing Interests:**RC is an Academic Editor at PLOS ONE. This does not alter the authors\' adherence to PLOS ONE policies on sharing data and materials.
[^2]: Conceived and designed the experiments: MB PG RC. Performed the experiments: MB. Analyzed the data: MB PG RC. Contributed reagents/materials/analysis tools: PG RC. Wrote the paper: MB PG RC.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Multiple sclerosis (MS) treatment has evolved tremendously over the last two decades. Whereas a patient diagnosed with MS 30 years ago had very little hope of a future without severe neurological disability, most patients who receive the diagnosis now have access to a variety of pharmacological and non-pharmacological treatments that can modify their disease evolution \[[@CR1]\]. In the present scenario, MS management aims to reach a state of no (or minimal) evidence of disease activity in patients \[[@CR2]\]. Therefore, while individuals with MS might not have considered parenthood in the past due to the potential disability, patients nowadays seem to have a different attitude towards having children. Family planning for patients with MS includes advice from neurologists regarding disease control at conception and during pregnancy and the puerperium. Since reactivation of well-controlled MS may lead to relapses and accumulated neurological disability, simple withdrawal of drugs for patients who intend to conceive is not an option. On the other hand, not all treatments presently recommended for MS are considered safe during conception, pregnancy and/or breastfeeding \[[@CR3]\]. The objective of the present study was to summarize the practical and evidence-based recommendations for neurologists, obstetricians and urologists who discuss family planning with women and men with MS.
A panel of neurologists with experience in MS reviewed several aspects of the disease with regard to human reproduction. Their recommendations were established based on published evidence and the specialists' expert opinion. The review of the literature was not limited by date and included specific words for each individual subject reviewed by the specialists. The authors used PubMed, Medline, SciELO, LILACS, the Cochrane Library and Google Scholar to identify relevant papers. Only papers with a title and abstract in English were reviewed; from the articles selected, the lists of references were further searched for potentially relevant citations. After 2 months of reviews, the panel met for a full day of discussions and elaboration of the material below. All recommendations presented here are evidence-based and were included with the approval of at least 80% of the authors.
This paper is a comprehensive review of medical literature on previously conducted studies and does not contain any studies with human participants or animals performed by any of the authors. All statements regarding the literature are cited in the references.
Level of Evidence {#Sec2}
=================
For obvious ethical reasons, there are no double-blind or placebo-controlled randomized studies on men or women with MS who intend to conceive, or on pregnant and/or breastfeeding women. The papers selected for this article typically come from anecdotal reports, case series, observational studies and national or international databases, using retrospective, prospective or cross-sectional cohorts of patients. Therefore, there is no evidence coming from Class I or II studies in this population of patients with MS, and all recommendations in the literature are, at best, Level C \[[@CR4]\].
General Recommendations for Patients with MS Who Want to Have Children {#Sec3}
======================================================================
Although both men and women with MS should have the disease under control before conceiving a child, this recommendation is particularly important for women. The risk of relapses in the post-natal period showed an independent correlation with the 12-month \[[@CR5]\] and 24-month \[[@CR6]\] annualized relapse rate preceding pregnancy. In addition, a higher number of postnatal relapses increased the risk of disability progression in women with MS \[[@CR5]\]. Therefore, it is of essence that women with MS who intend to become pregnant have their disease under control for at least 1 year before conceiving. Should the patient have a new diagnosis of MS, the recommendation is to wait for at least 1 year in order to attempt to control disease activity before conceiving.
Antenatal Care for Women with MS {#Sec4}
================================
Pregnancy in a woman with MS does not constitute a case of "high-risk" pregnancy. Antenatal care follows the general recommendations, with the same scheme of dietary supplementation of folic acid, vitamins, minerals and iron that pregnant women need to have. Likewise, cessation of alcohol and tobacco use is important \[[@CR7]\]. One aspect of antenatal care that may give rise to greater controversy is supplementation of vitamin D among pregnant women with MS. The plasma levels of vitamin D have not been found to be associated with the postnatal relapse rate in MS \[[@CR8], [@CR9]\], and general obstetric and neonatal outcomes seem to have little relationship to vitamin D supplementation during pregnancy \[[@CR10]\]. Therefore, there is no particular reason to aim for high plasma levels of vitamin D in pregnant women with MS.
In summary, antenatal care for women with MS should follow the general recommendations of the gynecology and obstetrics societies of each country.
Vaccines {#Sec5}
========
Obstetricians and gynecologists have very good opportunities to incorporate vaccination into the standard clinical care for women \[[@CR11]\]. They specifically care for pregnant women who, along with their fetuses, can be particularly vulnerable to vaccine-preventable diseases and their related complications \[[@CR12]\]. In patients with MS, vaccination can have peculiarities related to the disease itself and its treatment \[[@CR13], [@CR14]\].
Vaccination should be discussed with the patient before conception and should follow the recommendations and immunization calendar of each country. In general, if the woman has not been previously exposed or vaccinated against a certain disease, the risk/benefit of immunization during pregnancy should be assessed individually \[[@CR15]\]. Inactivated vaccines, such as seasonal influenza and tetanus, can be used safely even during pregnancy. Vaccines with bacteria and those with attenuated viruses can be used before the woman conceives and may be used in the final trimester of pregnancy. These include Bacillus Calmette--Guérin (BCG) against tuberculosis, oral typhoid, measles-mumps-rubella (MMR), varicella/zoster and rotavirus \[[@CR16]\]. Other vaccines that may be used during pregnancy on an exception basis include pneumococcal polysaccharide vaccine and hepatitis B. The healthcare staff need to be aware that patients with MS should not receive live-virus vaccine (for example, yellow fever). These patients should be immunized only in some very specific situations, through case-by-case indication.
The Effect of Pregnancy on MS Relapses (Short-Term Outcomes) {#Sec6}
============================================================
Acute demyelinating relapses of MS can be worrisome during pregnancy and the puerperium. While pregnancy status typically reduces the number of clinical relapses, this number is not zero, and some patients may require treatment while pregnant. Reactivation of MS inflammatory activity is a characteristic feature of the postnatal period, and treatment of relapses may be necessary in the puerperium.
The mechanism underlying the changes in relapse rate during gestation and after delivery are ultimately a function of cytokine and hormone levels in the woman. Pregnancy is associated with downregulation of cell-mediated immunity, ultimately resulting in a shift towards a T-helper 2 (Th2) cytokine profile \[[@CR17]\]. This shift leads to reduced levels of Th1 cytokines (interferon gamma and tumor necrosis factor alpha) and increased levels of Th2 cytokines (interleukins IL-4 and IL-10), which are essential for tolerance of the fetus during pregnancy \[[@CR18]\]. In addition, the beneficial immunomodulatory effects of high levels of estrogen during pregnancy \[[@CR19]\] may explain the overall reduction in MS activity during gestation.
Management of Relapses During Pregnancy {#Sec7}
=======================================
Fortunately, most women with MS will not develop neurological symptoms during or after pregnancy \[[@CR20]\]. Relapses occurring during pregnancy, particularly in the first trimester, should be treated with corticosteroids only when they significantly affect the mother's activities of daily living \[[@CR21]\]. Although conflicting data have been published, corticosteroids used during pregnancy have been associated with orofacial cleft \[[@CR22], [@CR23]\]. There are no reports of other therapies (including immunoglobulin) for treatment of relapses during the gestational period in women diagnosed with MS.
Summarizing the recommendations regarding therapy for MS relapses during pregnancy, intravenous 3-day pulse of methylprednisolone (1 g/day) can be used if necessary. Prolonged oral administration of prednisone should be avoided, as well as the use of corticosteroids with placental effects, such as dexamethasone or betamethasone \[[@CR24]\].
Management of Relapses After Delivery {#Sec8}
=====================================
Studies from various countries show that relapses occur in 12--39% of women with MS during the puerperium \[[@CR20], [@CR25]--[@CR28]\]. A low relapse rate preceding pregnancy \[[@CR6], [@CR21]\] and exposure to immunomodulatory drugs at the time of conception \[[@CR6], [@CR29]\] appear to be the only factors associated with lower numbers of postnatal relapses.
Following studies with conflicting results regarding the role of breastfeeding in preventing postnatal relapse \[[@CR30], [@CR31]\], a meta-analysis concluded that this protective effect, if present, is modest \[[@CR32]\]. The recommendation is to encourage breastfeeding for its beneficial effect as a whole \[[@CR33]\], with the important exception of (re)starting MS therapy with drugs that are detectable in breast milk. These drugs will be discussed later in this paper.
The use of intravenous immunoglobulin (IVIG) has generated conflicting results in published papers. Although IVIG has a good safety profile, a recent meta-analysis demonstrated an ineffective cost--benefit profile from prescribing IVIG for prevention of relapses during the puerperium \[[@CR34]\]. Different therapeutic schemes have been used; there are some uncontrolled studies and some studies controlled with 30-year-old historical patient data \[[@CR34]\]. Briefly, using the present information on IVIG, approximately six women must be treated to avoid one relapse, which translates into a cost of over US\$80,000 to prevent of a single postnatal MS relapse. Therefore, the evidence-based recommendation does not include IVIG in the management of postnatal reactivation of MS.
Monthly pulses of 1 g methylprednisolone have been proposed for prevention of relapse during the puerperium, but only two case series have been published \[[@CR35], [@CR36]\]. This approach has shown good results and is potentially safe, since the corticosteroid virtually disappears from breast milk within 4 h \[[@CR36], [@CR37]\]. Milk from this period could be discharged and the baby could be breastfed 4 h after the mother's infusion. It is an inexpensive option that could be studied in detail and thus could become part of the recommendations for protection against puerperal relapses.
The Effect of Pregnancy on Disability (Long-Term MS Outcomes) {#Sec9}
=============================================================
Pregnancy does not, in itself, negatively affect the course of MS \[[@CR38]\]. While there are studies showing that parity was associated with better outcomes regarding disability progression in MS \[[@CR39]\], others have not obtained the same results \[[@CR40]\]. Irrespective of this divergence, the recommendation regarding pregnancy for women with MS remains a matter of reaching disease control before conception.
The Effect of MS on Obstetric, Neonatal and Delivery Outcomes {#Sec10}
=============================================================
There are studies reporting higher rates of premature birth and lower birth weight among babies born to women with MS \[[@CR38], [@CR41], [@CR42]\], while other studies did not show such results \[[@CR43], [@CR44]\]. All authors seem to agree that there is a higher rate of cesarean section among women with MS. This finding might be a reflection of fatigue, spasticity of lower limbs, slower progression of labor and/or pelvic organ dysfunction, which can all be features of MS \[[@CR45]\]. The recommendation is that it is the obstetrician's prerogative to indicate induced labor or a cesarean section.
There are no reported negative outcomes from the use of epidural, peridural, caudal, spinal, subarachnoid or intrathecal analgesia in MS \[[@CR46]\]. Epidural analgesia can be safely used in women with MS who are giving birth \[[@CR47]\]. Pregnancies in women with MS do not need to be classified as "high-risk" pregnancy. However, it is our recommendation that hospital-assisted delivery is a better option than "natural home birth", "labor and delivery alone", "water birth" and similar alternatives.
Symptomatic Drugs Often Used for Patients with MS {#Sec11}
=================================================
Patients with MS often present other conditions such as depression, fatigue, spasticity or gait abnormalities, which can all worsen during and after pregnancy. The drugs used to treat these conditions typically consist of small molecules taken orally. Except for antidepressants, data on the safety of these treatments are sparse.
Large-cohort studies \[[@CR48]--[@CR51]\] on the use of antidepressants by pregnant women and nursing mothers have indicated that these drugs are not particularly teratogenic. Conflicting data in the literature have suggested that antidepressants may have a small effect on fetal growth, while associations between paroxetine and cardiac defects, between citalopram and craniofacial malformations, and between venlafaxine and pulmonary hypertension in newborns have been described \[[@CR51]--[@CR54]\]. Women who use antidepressants late in pregnancy do not have lower production of breast milk, as was reported in the past \[[@CR55]\].
The pharmacological profile of sertraline suggests that this drug might be an appropriate choice when antidepressant treatment is required during pregnancy. Sertraline has little interaction with other systems and presents linear pharmacokinetic characteristics, with a half-life of 24--26 h. Sertraline is also considered to be a good option for treating depression in nursing mothers \[[@CR56]\].
Amantadine is often used to manage fatigue in MS, although its effect is disputable, and the recommendation of amantadine for this purpose is off-label \[[@CR57], [@CR58]\]. There are only anecdotal reports of exposure to amantadine during pregnancy, but due to teratogenicity in animal studies and occasional reports of severe malformation in humans, this drug should not be prescribed for pregnant women \[[@CR59]--[@CR61]\].
Modafinil is another drug used off-label for treatment of fatigue and cognitive dysfunction in MS \[[@CR62], [@CR63]\]. The effect of modafinil in MS, if any, is small and is poorly studied. There are no data in humans justifying its use during pregnancy, given that the effects on the mother and child are virtually unknown.
Fampridine is recommended for improving gait velocity for patients with MS. There is only one literature report regarding fampridine exposure during pregnancy, which showed good outcomes \[[@CR64]\]. The recommendation is not to prescribe fampridine to women who intend to have children, or to pregnant or breastfeeding patients with MS.
Baclofen is a gamma-aminobutyric acid agonist used primarily as a muscle relaxant to improve spasticity. Although intrathecal baclofen seems to be a safe alternative during pregnancy in very specific cases \[[@CR65]\], oral baclofen should not be prescribed at all \[[@CR66]\]. There are no formal contraindications for breastfeeding while using baclofen, as the drug is detected only in small amounts in breast milk \[[@CR65]\].
Men with MS Who Want to Have Children {#Sec12}
=====================================
While research on pregnancy among women with MS has increased considerably over recent decades, investigations in men with MS who father children still have a long way to go. Men with MS may present fertility impairment and sexual dysfunction, while semen itself may be affected by drugs.
Relative to control subjects, men with MS have been reported to have lower baseline levels of luteinizing hormone (LH), follicle-stimulating hormone (FSH) and testosterone \[[@CR67]\]. Total sperm count, sperm motility and percentage of normal sperm morphology have been found to be lower in patients with MS than in controls \[[@CR67]\]. These findings indicate that men with MS may exhibit a state of hypogonadotropic hypogonadism that can affect their fertility. It is interesting to observe that men with hypogonadism showed a higher risk of developing MS in large-population studies \[[@CR68], [@CR69]\], while low baseline levels of testosterone in men with MS were associated with worse disability outcomes \[[@CR70]\]. Another large-population study showed that infertile men were at higher risk of developing not only MS, but also rheumatoid arthritis, psoriasis and thyroiditis \[[@CR71]\].
In addition to hormonal changes associated with MS, it is well known that 50--90% of men affected by MS can experience erectile dysfunction, ejaculatory dysfunction, orgasmic dysfunction and/or reduced libido \[[@CR72], [@CR73]\]. Higher rates of depression and fatigue have been described in patients with MS and sexual dysfunction \[[@CR74], [@CR75]\].
There are few studies on the potential effect of MS therapy on men who have fathered children. The use of interferon beta and glatiramer acetate by the father has not been associated with worse neonatal outcomes \[[@CR76]--[@CR78]\]. There is still a lack of data on other therapeutic options for MS.
Assisted Reproductive Technology {#Sec13}
================================
Assisted reproductive technology (ART) is used to treat infertility, in relation to both the woman's ovule and the man's sperm. While ART does not seem to influence the risk of MS relapses in men \[[@CR79]\], women can have higher MS activity when undergoing the same procedure \[[@CR80]\]. At least in theory, the use of gonadotropin-releasing hormone (GnRH) antagonists should be a better option for women with MS undergoing ART \[[@CR80]\]. GnRH antagonists have the same clinical outcomes as GnRH agonists, without leading to a decrease in estrogen \[[@CR81]\]. The potential protective effect of estrogen and estradiol in MS has been known for a few years \[[@CR82]\].
The Effect of MS Therapy on Childbearing {#Sec14}
========================================
The majority of studies on pregnancy and MS have reported on relapse rates among the mothers and on the effect of disease-modifying drugs (DMDs) on newborns. A recent comprehensive review by Vaughn et al. \[[@CR83]\] summarized the data regarding the potential risks of the use of DMDs during pregnancy and their prescribing recommendations. All the published data come from observational reports and may not be considered very robust \[[@CR84]\]. On the other hand, the overly simplistic solution of withdrawing all treatments from women who intend to conceive (or are already pregnant) does not take into consideration the potential severity of MS \[[@CR85]\]. Severe rebounding of MS activity may occur when fingolimod or natalizumab is withdrawn during pregnancy (or at the time of pregnancy planning) \[[@CR86]--[@CR91]\]. In the modern era, when decision-making relating to MS does not allow for anything less than optimal disease control \[[@CR92]\], it is unacceptable that young women are simply told to stop therapy without being informed of the consequences of this action. Furthermore, as previously discussed, low relapse rates in the postpartum period are associated with good disease control at conception \[[@CR6]\]. The present knowledge on DMDs can serve as guidance for specific recommendations for men and women with MS who want to have children. For easier understanding of this complex subject, DMDs are classified and presented here as "self-injectable drugs", "oral drugs" or "monoclonal antibodies". A summary of the data on each drug is shown in Table [1](#Tab1){ref-type="table"}. More detailed information is presented below.Table 1Characteristics of disease-modifying drugs used for treating multiple sclerosis that are relevant to reproductive issuesDrug (commercial brand name)Main mechanism of actionExposure during pregnancy (published cases)Main concerns according to literatureRecommendation on the label (FDA and/or EMA)Pre-conception washout periodBreastfeedingPanel recommendationSelf-injectable drugs Interferon β (Avonex^®^, Rebif^®^, Betaseron^®^)↓ Antigen presentation\
↓ T-cell proliferation\
Suppresses inflammation\> 3000↑ Pre-term delivery (37 weeks)\
↑ Rates of caesarean deliveryShould be avoided during pregnancy, unless the benefits outweigh the risksNot requiredNo specific concernsMaintain use irrespective of the patient's wish to conceive\
Keep drug during pregnancy and breastfeeding if deemed appropriate Glatiramer acetate (Copaxone^®^)Shifts inflammatory profile to an anti-inflammatory one (Th1 →Th2)\> 5000No specific concernsCan be used during pregnancy if benefits outweigh the risksNot requiredNo specific concernsMaintain use irrespective of the patient's wish to conceive\
Keep drug during pregnancy and breastfeeding if deemed appropriateOral drugs Fingolimod (Gilenya^®^)↓ Number of circulating lymphocytes by preventing their egress from secondary lymph organs\< 500Reports of fetal malformations\
Discontinuing the drug may lead to rebound of activityShould not be used2--3 monthsNot recommended (identified in breast milk)Discontinue fingolimod for at least 2 months prior to conception\
Do not breastfeed Dimethyl fumarate (Tecfidera^®^)Targets antioxidant mechanisms\
Anti-inflammatory effects via Treg\< 200No increased risk of fetal abnormalities or adverse pregnancy outcomes have been observedShould not be used, as there are not enough safety dataNot establishedNo data availableThe drug should be discontinued in patients intending to conceive\
However, the short-life and non-cumulative profile of this drug may be a favorable aspect of it\
Do not breastfeed Teriflunomide (Aubagio^®^)↓ Pyrimidine synthesis, leading to reduced number of lymphocytes\
↓ Antigen presentation\
Shifts inflammatory profile to an anti-inflammatory one (Th1 → Th2)\< 100 (leflunomide and teriflunomide)Severe malformations in an animal exposureIf pregnancy is diagnosed, the drug must be eliminated with activated charcoal, cholestyramine or colestipol hydrochlorideProceed with accelerated elimination in both men and women prior to conceptionContraindicatedDespite great worries about teratogenicity, initial findings have been reassuring to mothers and fathers exposed to teriflunomide\
Should be avoided by men and women willing to conceive\
Do not breastfeed Cladribine (Mavenclad^®^)↓ Lymphocyte subpopulations\
↓ Lymphocyte transit through the BBB\< 50No specific concerns, but very small number of casesShould not be used, as there are not enough safety data6 months for both men and womenNo data availableShould be avoided by men and women willing to conceive\
Do not breastfeedMonoclonal antibodies Natalizumab (Tysabri^®^)Blocks leukocyte attachment to cerebral endothelial cells\
↓ Inflammation at the BBB and inside the central nervous system\< 400No specific concerns up to the 30th week\
After that, hematological abnormalities in the neonateShould be avoided during pregnancy, unless the benefits outweigh the risksDiscontinuing the drug may lead to important rebound of activityNot recommended (identified in breast milk)Discontinue in patients intending to conceive only if disease is strictly under control\
Can be used with great care up to 30 weeks of pregnancy\
Do not breastfeed Alemtuzumab (Lemtrada^®^)↓↓↓ Lymphocytes by targeting CD52\< 50No specific concerns, but very small number of casesPatients should avoid pregnancy throughout therapy with alemtuzumabShort half-life and does not cross placental barrier in the first weeksNot recommendedAt least in theory, conception 1 month after alemtuzumab infusion should not pose a problem\
Breastfeeding can be carried out if planned prior to resuming infusions Ocrelizumab (Ocrevus^®^)↓↓↓ B lymphocytes by targeting CD20\<300\
(including data on rituximab)B-cell depletion in neonatesPatients should avoid pregnancy when using B-cell-depleting therapyShort half-life and does not cross placental barrier in the first weeksNot recommendedBreastfeeding can be carried out if planned prior to resuming infusionsNote that data were obtained only for the original product and not for copies or biosimilar drugsSigns and abbreviations: *FDA* Food and Drug Administration, *EMA* European Medicines Agency, *BBB* blood--brain barrier, *Th* helper T lymphocyte, *Treg* regulatory T lymphocyte, ↓ decreases, ↑ increases, ↓↓↓ greatly decreases
It is extremely important to understand that the data presented in this paper are not necessarily true for biosimilar drugs (interferon beta and monoclonal antibodies), follow-on glatiramer acetate and generic oral drugs. The information reported in the present review comes from the original patented drugs.
Self-Injectable DMDs {#Sec15}
====================
The "older" category of immunomodulatory drugs used for treating MS comprises interferon beta and glatiramer acetate. Long-term data and the large number of cases of maternal exposure to these drugs have provided reassuring results regarding their safety during pregnancy. There are no specific recommendations for discontinuation of these treatments within the context of family planning.
Interferon Beta {#Sec16}
---------------
Different formulations of interferon beta have been used for treating MS, and outcomes relating to exposure during pregnancy have been reported from all of them. Two of the formulations consist of the naturally occurring amino acid sequence of interferon beta and are known as interferon beta-1a, with the commercial names Avonex^®^ (intramuscular administration) and Rebif^®^ (subcutaneous administration). The third formulation is known as interferon beta-1b (Betaseron^®^) and consists of a modified amino acid sequence, presenting a cysteine-to-serine mutation at amino acid 17, along with a deletion of the amino terminal methionine \[[@CR93]\].
Interferon beta targets immune cells, which gives rise to reduced antigen presentation and T-cell proliferation and induces altered cytokine and matrix metalloproteinase expression that culminates in suppression of inflammation \[[@CR94]\]. The interferon beta molecule is over 20 kDa in size and is not believed to cross the placenta \[[@CR95]\]. Pegylation of interferon beta results in a prolonged half-life of the active substance, with an extended dosage interval, thus enabling fewer injections and improving adherence to treatment \[[@CR96]\]. Pegylation stabilizes the molecule by protecting it from degradation and proteolysis \[[@CR97]\], thereby leading to better bioavailability, with a potential beneficial impact on efficacy \[[@CR98]\]. This new formulation is also available for treating MS under the commercial name Plegridy^®^.
In total, over 3000 reports of the use of interferon beta during any stage of pregnancy can be found in the literature, and the outcomes are not considered deleterious. Earlier reports on exposure to interferon beta during pregnancy pointed towards a higher risk of spontaneous abortion and low birth weight \[[@CR99], [@CR100]\], but later data seem reassuring. More recent papers with higher numbers of patients have suggested that the percentages of live births, spontaneous abortions and malformations resulting from exposure to interferon beta are similar to those observed in the general population \[[@CR101]--[@CR105]\]. Although some later studies do correlate interferon beta exposure with lower birth weight, the newborns in those studies were still over 3200 g on average \[[@CR106]\]. The World Health Organization defines "low birth weight" as "weight at birth of less than 2500 g" \[[@CR107]\]. Therefore, the widespread notion that interferon beta can lead to low birth weight in newborns is not correct. Higher incidence of cesarean deliveries and pre-term (37 weeks) birth have also been correlated with exposure to interferon beta in some reports \[[@CR45]\].
The European Medicines Agency and the US Food and Drug Administration label interferon beta as a drug that should be avoided during pregnancy unless the benefits of its use outweigh the risks. The recommendation is to maintain use of the drug irrespective of the patient\`s wish to conceive. Once the diagnosis of pregnancy has been established, the use of interferon beta throughout gestation will be a decision made jointly by the physician in charge and the patient, taking into consideration the risk/benefit profile of this therapeutic option.
The proportion of interferon beta that is transferred to breast milk is very low, and the estimate breastfed infant dose is 0.006% of the maternal dose \[[@CR108]\]. In addition, when given orally, interferon beta has shown no systemic biological effect \[[@CR109]\]. Therefore, women who intend to breastfeed may use interferon beta without concerns that this might affect the child.
Few cases have been reported among men undergoing treatment with interferon beta who father children, and there is no compelling evidence of obstetric or neonatal risks \[[@CR77], [@CR78]\].
Glatiramer Acetate {#Sec17}
------------------
Glatiramer acetate consists of acetate salts of synthetic polypeptides, containing four naturally occurring amino acids: [l]{.smallcaps}-glutamic acid, [l]{.smallcaps}-alanine, [l]{.smallcaps}-tyrosine and [l]{.smallcaps}-lysine. This complex molecule has an average length of 45--100 amino acids, with a molecular weight of 5--9 kDa. A substantial percentage of the therapeutic dose of glatiramer acetate is hydrolyzed at the site of the injection and interacts locally with peripheral blood lymphocytes \[[@CR110]\]. This drug has been used worldwide for two decades and has shown no teratogenic or mutagenic effects \[[@CR111]\]. In late 2016, the European Medicines Agency updated the label for Copaxone^®^, such that it is no longer considered contraindicated during pregnancy. According to the Food and Drug Administration, Copaxone^®^ is a "Category B" drug, meaning that no risks have been shown in animal fetuses, but there are no well-controlled studies related to human pregnancy. Reports on over 5000 cases of branded Copaxone^®^ exposure during pregnancy reinforce the lack of teratogenic effects \[[@CR112], [@CR113]\]. These findings provide important knowledge for better counseling for women with MS who intend to become pregnant and should not forgo the use of DMDs while attempting to conceive. Continuous use of glatiramer acetate throughout pregnancy can also be discussed with patients who may require this particular approach \[[@CR114]\].
Breastfeeding while using glatiramer acetate is potentially safe. This drug is rapidly degraded after subcutaneous injection and cannot be detected in the plasma, urine or feces \[[@CR115]\].
There are only two specific reports on men undergoing treatment with glatiramer acetate who fathered children \[[@CR77], [@CR78]\]. No association between paternal exposure to glatiramer acetate at the time of conception and a risk of adverse outcomes was shown.
Oral Drugs {#Sec18}
==========
While the "older" self-injectable DMDs discussed above are large molecules that generally do not cross the placenta or have any presence in breast milk, the "newer" oral drugs are small molecules that behave differently. Fingolimod freely crosses the blood--brain barrier \[[@CR116]\] and the placental barrier \[[@CR117]\]. Teriflunomide can cross the placental barrier \[[@CR117]\] and should not be used by men or women with MS who intend to conceive. It is likely that dimethyl fumarate can cross the placenta as well \[[@CR117]\], while it is still uncertain whether cladribine can reach the fetus \[[@CR118]\]. It is generally better to avoid the use of small molecules that can potentially reach the fetus among women who want to have children. The amount of data in pregnancy exposure registries on the use of these oral drugs for treating MS remains insufficient. As safety data accumulates through pharmacovigilance programs, the recommendation for avoiding oral drugs during pregnancy may change.
The washout period for these oral drugs varies; for fingolimod it has been established as 2--3 months. In order to decrease the period between DMD withdrawal and conception, discontinuation of oral MS therapy should be planned such that it is only implemented 2 or 3 months after withdrawal of oral contraceptives. Pregnancy after stopping the use of birth control pills may be achieved after only one cycle in approximately 20% of women, and may take up to 1 year for 80% \[[@CR119]\].
Fingolimod {#Sec19}
----------
Fingolimod was the first oral drug to be approved for MS therapy \[[@CR120], [@CR121]\]. It is a sphingosine-1-phosphate receptor modulator that is administered orally once daily, and its use leads to a reduction in the number of circulating lymphocytes by preventing their egress from secondary lymph organs \[[@CR120]\]. This oral therapy has been shown to be effective at a dose of 0.5 mg/day in double-blind, placebo-controlled studies and in trials comparing it with interferon beta-1a. Studies continue to assess the long-term efficacy and safety of newer DMDs, including fingolimod, as Faissner and Gold reinforce in a comprehensive 2018 review \[[@CR121]\].
Animal data on exposure to fingolimod during gestation suggests that birth malformation can occur, possibly due to the influence of sphingosine-1-phosphate on embryogenetic vascular formation \[[@CR122]\]. Data in relation to human pregnancy are scarce, but fetal abnormalities have been reported at higher rates than what would be expected for the general population \[[@CR123]\]. The present recommendation is to discontinue fingolimod for at least 2 months prior to conception \[[@CR123], [@CR124]\]. As fingolimod can be identified in human breast milk, this treatment should not be resumed if the mother intends to breastfeed \[[@CR125]\].
Regarding fingolimod withdrawal in the context of pregnancy planning, the neurologist needs to be aware of the risks of disease rebound following fingolimod discontinuation \[[@CR86], [@CR91], [@CR126]\]. There are no predictive factors defining the rebound risk group, and patients should be closely monitored after stopping fingolimod \[[@CR126]\].
Dimethyl Fumarate {#Sec20}
-----------------
Dimethyl fumarate derives from free fumaric acid, a substance that has poor gastrointestinal absorption \[[@CR127]\]. The ester derivatives of fumaric acid have better absorption but may still lead to gastrointestinal symptoms that affect tolerance to the drug \[[@CR127]\]. Although relatively new for treating MS, fumarates have been used to treat psoriasis since the mid-1990s \[[@CR128]\].
Dimethyl fumarate targets antioxidant mechanisms and exerts anti-inflammatory effects through stimulation of regulatory T lymphocytes \[[@CR129]\]. The drug can modify a variety of proteins involved in T-cell activation through its electrophilic activity \[[@CR130]\]. Dimethyl fumarate can reduce the production of nitric oxide synthase and decrease the levels of proinflammatory cytokines, including those that depend on mediation by nuclear factor kappa B (NF-κB) \[[@CR131]\]. It is likely that the drug can induce a wider spectrum of immunological changes in patients with MS, but the studies performed so far suggest that the main mechanism of action involves elimination of proinflammatory activated T cells \[[@CR129]\].
The data on fetal exposure to dimethyl fumarate are very limited, but no increased risk of fetal abnormalities or adverse pregnancy outcomes have been observed thus far \[[@CR83], [@CR132]\]. It is not known whether dimethyl fumarate or its metabolites are present in human milk.
There are no reports in the literature suggesting acute reactivation of MS following discontinuation of dimethyl fumarate. Likewise, the drug is known to have a short half-life, and there is no evidence of its accumulation \[[@CR133]\]. Therefore, pregnancy planning following withdrawal from dimethyl fumarate should not pose the same concerns observed with other drugs, as discussed above.
Teriflunomide {#Sec21}
-------------
Teriflunomide is the active metabolite of leflunomide, a drug that has been successfully used for treating rheumatoid arthritis and psoriatic arthritis for many years \[[@CR134]\], and has been approved for treating MS. It induces reversible inhibition of dihydroorotate dehydrogenase, a critical enzyme in the synthesis of pyrimidine. Lymphocytes require pyrimidine for proliferation, and thus teriflunomide limits the number of circulating peripheral lymphocytes. In addition, this drug affects the antigen presentation of dendritic cells and stimulates the Th1--Th2 shift in the lymphocyte profile \[[@CR134]\].
Teriflunomide was originally labeled as a substance with high teratogenic effects. The drug label states that it should be avoided at all costs among men and women with MS who intend to have children \[[@CR135]\]. Animal studies on maternal exposure to pyrimidine inhibitors have shown that the drug can induce neural tube defects, cleft palate, tail deformities, limb malformations, abnormalities in vertebrae (cervical to sacral), membranous ventricular septum defect and persistent truncus arteriosus \[[@CR136]\]. The abnormalities observed in animal fetuses have been dose-dependent, and very high doses rendered the fetus unviable. Malformations were a function of dihydroorotate dehydrogenase inhibition, and early exposure gave rise to the possibility of multiple congenital abnormalities \[[@CR137]\]. Since teriflunomide can be detected at low levels in human semen, it is recommended that men undergoing therapy with this drug avoid reproduction \[[@CR138]\].
Despite all the concerns regarding teratogenicity, no adverse outcomes relating to human pregnancy have been seen with leflunomide \[[@CR139], [@CR140]\] or teriflunomide \[[@CR138]\]. The number of pregnancies studied remains small, but the initial findings have been reassuring to mothers and fathers who might be overwhelmed with worries after reading the drug label.
Elimination of teriflunomide is slow, and the drug may still be detected in human blood 2 years after discontinuation \[[@CR141]\]. Accelerated elimination of teriflunomide with activated charcoal, cholestyramine or colestipol hydrochloride is recommended when the drug needs to be eliminated rapidly \[[@CR141], [@CR142]\]. Particularly in relation to pregnancy planning, the recommendation for accelerated elimination of the drug is important, but this also needs to be implemented as soon as a pregnancy is diagnosed in women with MS who are using teriflunomide.
Accelerated elimination can be achieved through the following schemes: (1) cholestyramine 8 g three times daily for 11 days; or (2) activated charcoal 50 g twice daily for 11 days; or (3) colestipol (colesevelam hydrochloride), four 625-mg tablets in the morning plus three 625-mg tablets in the evening \[[@CR141]\]. It would be of great help to physicians if the company commercializing teriflunomide for treatment of MS could guarantee provision of the full dose of an accelerated elimination drug whenever necessary.
Teriflunomide is contraindicated during breastfeeding \[[@CR138]\].
Cladribine {#Sec22}
----------
Cladribine is a chlorinated deoxyadenosine prodrug that is activated by intracellular phosphorylation to become an active purine nucleoside analogue \[[@CR143]\]. Oral cladribine may have beneficial immunological effects in the treatment of MS through the targeting of specific lymphocyte populations \[[@CR144], [@CR145]\]. In addition, cladribine appears to reduce the availability of vascular cell adhesion molecule 1 (VCAM-1), intercellular adhesion molecule 1 (ICAM-1) and E-selectin, all of which are important for lymphocyte transit through the blood--brain barrier \[[@CR146]\]. Despite initial concerns regarding adverse events during the use of cladribine (lymphocytopenia, herpes zoster infections and malignancies), this drug has been approved for treating MS in several countries now.
The recommended dose of cladribine in MS is 3.5 mg/kg body weight over 2 years, or 1.75 mg/kg per year. Each course of treatment consists of two treatment weeks: one at the beginning of the first month and one at the beginning of the second month of the respective treatment years (4 or 5 days of 10 mg or 20 mg as a single daily dose, depending on body weight). Although this therapeutic scheme is very convenient, the long-term systemic effects of the drug may raise concerns regarding pregnancy over this 2-year period. Data on pregnancy and cladribine are scarce, and despite the lack of reported adverse outcomes, this drug should be avoided among patients who intend to conceive \[[@CR147]\]. Men and women using cladribine should not conceive for at least 6 months after the last dose. Recent research strongly suggests that cladribine can exert its effects mainly via B-cell depletion \[[@CR148]\]. Therefore, with caution and attention to all potential differences there may exist among drugs, data on pregnancy in patients with MS using other drugs with similar mechanisms of action might ultimately help building a "B-cell depletion" safety database.
Cladribine is contraindicated for women who breastfeed \[[@CR147]\].
Monoclonal Antibodies {#Sec23}
=====================
Monoclonal antibodies are increasingly being used for disease management. However, experience with and data on the potential reproductive and developmental toxicities relating to these agents remain sparse \[[@CR149]\]. Monoclonal antibodies cannot be transported across the placenta by means of simple diffusion, since they are hydrophilic molecules with a molecular mass exceeding 100 kDa. They require active transportation across the placental barrier via a specific receptor-mediated mechanism \[[@CR150]\], which may not take place until after several weeks of placental development \[[@CR151]\]. Therefore, it would be expected that early maternal exposure to monoclonal antibodies does not negatively affect fetal organogenesis \[[@CR152]\]. This consideration is important, since women with MS undergoing therapy with monoclonal antibodies tend to be patients with more aggressive neurological disease. Discontinuation of their therapy might lead to severe disease reactivation, and therefore, planning for conception and pregnancy among these patients is an extra challenge for the physician in charge.
Natalizumab {#Sec24}
-----------
Natalizumab was the first monoclonal antibody developed for treating MS and is still one of the most potent therapies for disease control. Natalizumab is a humanized monoclonal antibody that binds to integrins of both α4β1 (very late antigen 4, VLA-4) and α4β7 (lymphocyte Peyer's patch adhesion molecule 1, LPAM-1) \[[@CR153]\]. This unique mechanism of action blocks leukocyte attachment to cerebral endothelial cells, thus reducing inflammation at the blood--brain barrier and inside the central nervous system \[[@CR154]\]. The main impediments to more widespread use of this drug relate to potential development of progressive multifocal leukoencephalopathy (PML) in patients with MS undergoing therapy with natalizumab \[[@CR155]\]. The recommended dose of natalizumab is 300 mg administered by means of infusion every 4--8 weeks \[[@CR156]\].
Over 350 cases of maternal exposure to natalizumab have been reported with complete outcomes \[[@CR157]\]. This database shows that although the overall rate of birth defects was higher than that expected for a healthy population (5.05%), there were no specific patterns of malformations that would suggest a drug effect \[[@CR157]\]. The babies' weights were within expected values, and no differences in the spontaneous abortion rate or gestational age at delivery were observed relative to the general population \[[@CR157]\].
In Germany and Italy, case series of early exposure to natalizumab during pregnancy have been reported \[[@CR158]--[@CR161]\]. The obstetric and neonatal outcomes have been unremarkable in these cases of early exposure (up to 12 weeks of pregnancy). On the other hand, the use of natalizumab in the third trimester of pregnancy has been systematically correlated with hematological abnormalities in babies \[[@CR162], [@CR163]\]. In two newborns, long-term in utero exposure to natalizumab resulted in a reduced T-lymphocyte chemotaxis rate, which may have compromised their early-life host defense \[[@CR164]\]. Reactivation of MS (observed clinically and/or via MRI) was reported in 95.5% of patients with MS who discontinued natalizumab due to pregnancy \[[@CR165]\]. The same study showed worsening of disease disability in 27.3% of these patients. Disease reactivation is one of the most difficult aspects of discontinuing natalizumab in any patient, and during pregnancy it poses an additional challenge \[[@CR166]\]. There are reports of severe MS reactivation after withdrawal of natalizumab in relation to pregnancy planning \[[@CR89]\] and after pregnancy was diagnosed \[[@CR87], [@CR88], [@CR90]\]. At least in theory, natalizumab might be administered until the 30th week of pregnancy in cases of very aggressive disease, but this recommendation is based upon very few cases in the literature \[[@CR157], [@CR159]\].
In order to avoid serious complications over the course of the treatment with natalizumab, family planning needs to be discussed with the patient before treatment initiation.
Breastfeeding is not recommended during natalizumab use, since the drug can be identified in breast milk \[[@CR153], [@CR167], [@CR168]\]. Although the levels of natalizumab in breast milk are minute, breastfeeding safety cannot be determined at this time \[[@CR168]\].
Alemtuzumab {#Sec25}
-----------
In 1983, a group of researchers in Cambridge, UK, developed an IgM rat antibody that depleted lymphocytes, thus improving the chances of success in organ transplantation \[[@CR169]\]. This antibody evolved to an IgG2b CAMPATH-1 Mab compound, called Cambridge Pathology 1 (CAMPATH-1), and it proved to be a very potent immunosuppressive drug targeting only white blood cells \[[@CR169]\]. CAMPATH-1 was later studied for a possible role in purging of lymphocytes prior to autologous transplantation to treat acute lymphocytic leukemia \[[@CR170]\]. Subsequently, it became clear that the antibody of this potent drug targeted CD52 \[[@CR171]\]. Nomenclature standards for biological therapies led to changing the name of the drug to alemtuzumab. Alemtuzumab not only results in dramatic depletion of the T-cell population, but also seems to affect the complex reconstitution of lymphocytes in the immune repertoire \[[@CR172]\]. Although alemtuzumab has only recently been incorporated into the therapeutic options for treating MS, a group in Cambridge has used it to treat MS since the early 1990s \[[@CR173]\]. The drug was approved by the European Medicines Agency and the Food and Drug Administration in 2014 as therapy for MS with cytolytic properties against CD52 of lymphocytes. It is used in infusions of 12 mg/day for five consecutive days in the first year, and for three consecutive days 12 months later.
The number of pregnancies with alemtuzumab exposure is relatively small and has not yet reached 200 cases \[[@CR174]\]. The data on these pregnancies have typically been presented at congresses and conferences, and no definitive recommendations can be made at this time. No adverse outcomes have been associated with the use of alemtuzumab.
Alemtuzumab has a short half-life, ranging from 2 to 32 h after the first administration and 1--14 days after the last dose in patients with leukemia, who use higher doses than those used for MS cases \[[@CR175]\]. As with other monoclonal antibodies, alemtuzumab will not cross the placental barrier in the first weeks of embryogenesis. Therefore, at least in theory, conception occurring while alemtuzumab is being used should raise no particular concerns regarding disease in the fetus, and no particular recommendations for washout should be necessary. However, the autoimmune disorders that the mother may develop during therapy with alemtuzumab may pose a problem \[[@CR176]\]. For example, autoimmune thyroid disease, which affects 30--40% of patients undergoing therapy with alemtuzumab \[[@CR177]\], may translate into hypothyroidism during pregnancy.
Breastfeeding is contraindicated during the use of alemtuzumab, but infusions can be organized in such a way as to overcome this problem.
Ocrelizumab {#Sec26}
-----------
Until recently, T cells were considered to be the main players in inflammation leading to demyelination and degeneration in MS. Trials with rituximab, which provides B-cell depletion, yielded conflicting results, and DMD development continued to target T lymphocytes \[[@CR178]\]. Although trials with rituximab did not yield very favorable results in MS cases \[[@CR178]\], some groups have had good experience with this drug in treating their own patients \[[@CR179]\].
Ocrelizumab was developed for the purpose of targeting B lymphocytes in MS and is the most recently approved drug for MS therapy. It is a humanized anti-CD20 monoclonal antibody that depletes B cells through antibody-dependent cellular cytotoxicity \[[@CR180]\]. It is used in infusions at doses of 300 mg (first infusion) and 300 mg (second infusion) 2 weeks later. Thereafter, the dose is 600 mg every 6 months.
The data on exposure to ocrelizumab during pregnancy amounts to only 13 cases with clear exposure to the drug during the embryogenetic period \[[@CR181]\]. For rituximab, there have been approximately 200 cases of exposure to autoimmune diseases in general \[[@CR182]\]. Clear exposure within 6 months of the infusion has been reported in 102 cases, seven of which were patients with MS \[[@CR183]\]. Therefore, there are no data to define the safety of maternal exposure. The half-life of rituximab is 21 days \[[@CR184]\], and it is 26 days for ocrelizumab \[[@CR185]\]. Again, at least in theory, conception occurring during the use of B-cell-depleting therapy should not raise any particular concerns regarding disease in the fetus and thus should avoid concerns regarding washout periods. When used throughout pregnancy, rituximab has been associated with B-cell depletion in the baby \[[@CR183], [@CR186]\] and lymphoid tissue abnormalities in exposed animals \[[@CR187]\]. Babies with this hematological finding showed spontaneous recovery of B-cell levels after 6 months and had no complications.
Breastfeeding is contraindicated during the use of ocrelizumab, but scheduling postnatal infusions can overcome this problem.
Conclusion {#Sec27}
==========
The present study summarizes recommendations for physicians who discuss family planning with women and men with MS. The authors believe that reproductive issues in MS will remain an important issue for researchers and clinicians. Results from real-world databases and pharmacovigilance will add to the knowledge on this subject and will continue to change family planning for people with MS. A disease that once was a reason not to have children is now a potentially controllable disease. Medications with unknown effects on fetuses are now extensively studied regarding maternal and paternal exposure at conception. Fetal exposure to drugs used for MS treatment do not constitute a reason for abortion due to fear of the unknown.
**Enhanced digital content**
To view enhanced digital content for this article go to 10.6084/m9.figshare.6984785.
Funding {#FPar1}
=======
This study was made possible by an unconditional grant from TEVA Pharmaceutical Brazil. The grant funds provided standard accommodation and meals for participants to meet in São Paulo, SP, Brazil, in March 2018. Expenses for open-access publication were also covered by the grant. Two non-medical scientific personnel (G&C Scientific Writing, Brazil) attended the meeting to record the discussion among the experts and to revise the final material prior to submission for publication. They did not write the paper and only revised the final format for accuracy in relation to recorded discussions at the meeting. The scientific writers received standard rates of payment for their work. There was no payment for any of the MS experts participating in any stage of the review or article preparation. All authors had full access to all of the data in this study and take complete responsibility for the integrity of the data and accuracy of the data analysis.
Medical writing and other editorial assistance {#FPar2}
==============================================
Carla Isobel Elliff and Gerson Fernandino Andrade Neto are non-medical scientific writers working at G&C Scientific Writing. They do not work for any pharmaceutical company and hold no royalties or interest in any particular pharmaceutical compound. They did not write the paper but provided support in recording the meeting and revising/editing the text, table and references in the final article submitted to this journal. They guaranteed, through recordings and notes, that all subjects discussed in the experts' meeting were included in the final version of the article. For this work, they received a flat fee from TEVA Pharmaceuticals.
Authorship {#FPar3}
==========
All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article, take responsibility for the integrity of the work as a whole, and have given their approval for this version to be published.
Disclosures {#FPar4}
===========
Yara Dadalti Fragoso received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Tarso Adoni received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Joseph B. Bidin Brooks received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Alessandro Finkelsztejn received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Paulo Diniz da Gama received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Anderson K. Grzesiuk received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Vanessa Daccach Marques received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Monica Fiuza K. Parolin received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Henry K. Sato received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Daniel Lima Varela received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound. Claudia Cristina F. Vasconcelos received unconditional support from Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals to attend medical events and has been called as speaker and/or advisor and/or participant in phases III and IV of clinical trials for Bayer-Schering, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi-Genzyme and TEVA Pharmaceuticals. The author and relatives do not work for any pharmaceutical companies and hold no royalties or interest in any particular pharmaceutical compound.
Compliance with ethics guidelines {#FPar5}
=================================
This paper is a comprehensive review of medical literature on previously conducted studies and does not contain any studies with human participants or animals performed by any of the authors. All statements regarding the literature are cited in the references.
Open Access {#d29e1740}
===========
This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (<http://creativecommons.org/licenses/by-nc/4.0/>), which permits any noncommercial use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Overuse injuries of the lower limb associated with intensive weight bearing exercise are a significant problem for athletes and military recruits, with estimated incidence of running-related injuries reported to range from 20% to 79% \[[@CR1]\]. Lower limb overuse injuries are generally recognised as having multifactorial aetiologies \[[@CR2]\]. Some of the most common injuries, such as Achilles tendinopathy, medial tibial stress syndrome, patellofemoral pain and lower limb stress fractures, are reported to be more prevalent in those with altered foot function \[[@CR3],[@CR4]\].
The potential mechanisms linking variations in dynamic foot function with lower limb overuse injury may be related to altered lower limb biomechanics and subsequent changes in tissue stress \[[@CR5]\]. This is supported by laboratory-based research using uninjured participants, which suggests that variations in foot posture (flat- and normal-arched feet) are associated with systematic differences in lower limb kinematics \[[@CR6]-[@CR8]\], kinetics \[[@CR4],[@CR9],[@CR10]\], muscle function \[[@CR11]-[@CR16]\] and tendon morphometry \[[@CR17]\].
While laboratory-based research is important for understanding potential mechanisms linking foot function and lower limb overuse injury, field-based prospective studies are required to determine whether foot function is a risk factor for lower limb overuse injury. Our accompanying systematic review \[[@CR18]\] found that static measures indicating greater foot pronation were associated with an increased risk of patellofemoral pain and medial tibial stress syndrome. However, the small effects suggest that static measures may not adequately represent dynamic foot function. A substantial number of prospective studies have utilised a variety of measurement techniques in order to quantify dynamic foot function and its relationship with lower limb overuse injury \[[@CR19]-[@CR46]\]. However, it is unclear if there are consistent findings across different measures, or whether particular foot function characteristics are risk factors for specific overuse injuries. Enhanced knowledge regarding this may lead to the development of targeted preventative strategies.
Therefore, the aim of this systematic review was to: (i) identify and appraise the current evidence for the prospective link between dynamic foot posture and lower limb overuse injury; and (ii) provide guidance for future research in this area. This review represents the second component of a two-part systematic review on foot posture-related risk factors for lower limb overuse injury.
Methods {#Sec2}
=======
The systematic review protocol was developed in consultation with guidelines provided by the Preferred Reporting of Systematic Reviews and Meta-Analysis (PRISMA) Statement \[[@CR47]\].
Search strategy {#Sec3}
---------------
MEDLINE, CINAHL, Embase and SPORTDiscus were searched from inception until April 2014. Medical Subject Headings (MeSH) were exploded to include relevant subheadings, in addition to keywords specific to the research question (Additional file [1](#MOESM1){ref-type="media"}). The search was limited to adult human participants and English language publications. To ensure identification of all relevant studies, reference lists of appropriate narrative and systematic reviews were hand searched, and discussion with field experts (e.g. physiotherapists, podiatrists) was conducted regarding known important publications. A cited reference search for each included paper was also completed in Google Scholar.
Eligibility criteria {#Sec4}
--------------------
All studies identified by the search strategy were exported to Endnote version X5 (Thomson Reuters, Philadelphia), by a single investigator (GJD). Abstracts and then full text versions were reviewed by two authors (GJD, MMFS) to determine eligibility. Discrepancies were resolved in consultation with a third reviewer (GSM). Initial eligibility criteria were: (i) prospective cohort study design; (ii) quantitative measurement of foot posture or function at baseline (static or dynamic); and (iii) prospective collection of specific or non-specific lower limb overuse injury surveillance data over a specified time period. Specific lower limb overuse injuries were defined as injuries with a single diagnosis, while non-specific lower limb overuse injuries included injuries without a specific diagnosis or where multiple overuse types of injuries were pooled by the study reviewed. After retrieval of studies that fulfilled the initial eligibility criteria, suitable studies were separated into those that investigated dynamic measures of foot function (i.e. measured during walking or running), and those that investigated static measures of foot posture. This review focused on dynamic measures as risk factors, while static measures are addressed in the accompanying review \[[@CR18]\].
Quality assessment {#Sec5}
------------------
Assessment of the methodological quality of the included studies was performed using the Epidemiological Appraisal Instrument (EAI) \[[@CR48]\]. This instrument is designed to assess the quality of cohort (prospective and retrospective) studies. The EAI consists of 43 items separated into five domains --- (i) reporting, (ii) subject/record selection, (iii) measurement quality, (iv) data analysis and (v) generalisability of results \[[@CR48]\]. Items on the EAI were scored as "Yes" (score of 2), "Partial" (score of 1), "No" (score of 0), "Unable to determine" (score of 0) or "Not Applicable" (item excluded). The EAI has demonstrated good/excellent validity, and good to excellent intra-rater (Kappa coefficient range 52 to 60), and inter-rater reliability (Kappa coefficient = 90% \[95% CI; 87 to 92%\]) \[[@CR48]\]. For the purpose of this review, the wording of all 43 items was modified slightly to improve clarity and rater interpretation. No items were removed or modified, in order to maintain validity (Additional file [2](#MOESM2){ref-type="media"}).
Two raters (GJD, NJC) independently evaluated each study while blind to author and publication details. For any discrepancies in assessment of items between the two raters, a meeting occurred and consensus was achieved. To evaluate the overall quality of the studies, average scores across the 43 items were calculated, with a maximum possible score of two (i.e. as individual items are scored '0', '1' or '2', the maximum 'average' score across 43 items is two). A ranking system was used to evaluate the quality of evidence, whereby studies were classified as being high (EAI ≥ 1.4), moderate (EAI 1.1 to \<1.4), or low quality (EAI \< 1.1) \[[@CR47]\].
Data management {#Sec6}
---------------
Two investigators (GJD, GSM) extracted data regarding study characteristics, including publication details (year, author, country), participant characteristics (number of injured and uninjured, age, sex, inclusion and exclusion criteria, population \[i.e. military\]) and study methods (dynamic foot function measurement, examiner details, injury outcome, duration of study and covariates investigated). To facilitate calculation of effects, means and standard deviations (SD) were extracted for injured and uninjured participants for continuous foot function variables, while raw counts were extracted for nominal variables.
Where appropriate data was not provided in the publication, authors were contacted with a request to provide additional data. Where studies described specific variables but did not publish data, it was recorded as 'not reported' (NR) and, for the purpose of the analysis, assumed that the variable investigated was not significantly different between the injured and the uninjured population.
Statistical methods {#Sec7}
-------------------
Inter-rater reliability of the raters' EAI scores was evaluated using a descriptive analysis. Differences between rater scores for "Yes", "Partial", "No", and "Unable to determine" were calculated, with a difference of zero indicating perfect agreement and a difference of 1 indicating near perfect. The rating "not applicable" was excluded from analysis because no interpretation was required for this rating.
For continuous foot function variables, standardised mean differences (SMD) were calculated as the difference between injured and uninjured group means, divided by the pooled standard deviation \[[@CR49]\]. SMDs and 95% confidence intervals (CI) were calculated using the 'Effect Size Calculator' from the Centre for Evaluation and Monitoring \[[@CR50]\]. Interpretation of the SMD was based on previous recommendations, where \> 1.2 was considered large, 0.6 to 1.2 moderate, and \< 0.6 small \[[@CR51]\]. For nominal scaled foot function variables, risk ratios (RR) and 95% CI were calculated using the 'Confidence Interval Calculator' from the Physiotherapy Evidence Database (PEDro) \[[@CR52]\]. This was represented as the number of participants with lower limb overuse injury in the group with the associated factor (e.g. delayed time to peak force), divided by participants with lower limb overuse injury in the group without the associated factor. A RR \> 1.0 indicated that the lower limb overuse injury was more likely to be found in participants with the risk factor present. A small effect was indicated by a RR ≥ 2.0, and a large effect ≥ 4.0 \[[@CR53]\]. Effects were considered statistically significant if the associated 95% CI did not contain zero for the SMD, or one for RR.
Evidence-based recommendations {#Sec8}
------------------------------
In order to provide recommendations based on statistical findings, while incorporating the methodological quality of included papers, a scale regarding levels of evidence was utilised, based on previous work by van Tulder et al. \[[@CR54]\].
***Strong evidence:*** pooled results derived from three or more studies, including a minimum of two high quality studies that are statistically homogenous; may be associated with a statistically significant or non-significant pooled result.
***Moderate evidence:*** statistically significant pooled results derived from multiple studies that are statistically heterogeneous, including at least one high quality study; or from multiple moderate quality or low quality studies which are statistically homogenous.
***Limited evidence:*** results from one high quality study or multiple moderate or low quality studies that are statistically heterogeneous.
***Very limited evidence:*** results from one moderate quality study or one low quality study.
***No evidence:*** pooled results insignificant and derived from multiple studies regardless of quality that are statistically heterogeneous.
Results {#Sec9}
=======
Search results {#Sec10}
--------------
Across the two parts of this systematic review (static foot posture and dynamic foot function), a total of 33,518 citations were retrieved from the electronic database search. Following the sequential review of titles, abstracts and full texts, as well as removing studies that were not prospective cohort studies, 80 studies were eligible (Figure [1](#Fig1){ref-type="fig"}). Of these, 12 studies investigated dynamic foot function variables, and were included in this part of the review \[[@CR27],[@CR29],[@CR35],[@CR38]-[@CR46]\]. Due to inconsistencies in outcomes measured, pooling of data was not possible.Figure 1**Search results through the review process.**
Quality assessment {#Sec11}
------------------
Quality scores ranged from 0.44 to 1.20 (out of a possible total score of 2.00) (Additional file [3](#MOESM3){ref-type="media"}). With the exception of one moderate quality study \[[@CR43]\], all studies were rated as low quality \[[@CR27],[@CR29],[@CR35],[@CR37]-[@CR42],[@CR44]-[@CR46]\]. In terms of inter-rater reliability across 35 items included in the quality assessment, 24 items had perfect or near perfect agreement between raters. That is, these items were awarded the same score or there was a maximum of one point difference in scoring. For a further 10 items, the raters had near perfect agreement for 80% of the articles reviewed. Item 10 ('reporting of adverse effects') displayed the lowest agreement, with perfect or near perfect agreement for only 5/12 studies. Percentage agreement across the 35 items ranged from 17 to 100%.
All studies clearly reported the aim and objective (item 1) and that foot posture was measured prospectively before longer-term follow up of injury (item 28) \[[@CR27],[@CR29],[@CR35],[@CR38]-[@CR46]\]. Eleven studies clearly defined the assessment of foot function (item 2) \[[@CR27],[@CR29],[@CR35],[@CR38]-[@CR45]\] and eight studies clearly defined the lower limb overuse injury of interest (item 3) \[[@CR29],[@CR35],[@CR39],[@CR41]-[@CR45]\]. None of the included studies provided an adequate description of all intrinsic or extrinsic covariates or how these were adjusted for in the analysis (items 11, 12, 13, 36 and 37) (e.g. footwear worn, skill level or playing surface). Furthermore, no study provided an adequate report of the reliability and validity of foot function or injury outcome measurement of interest (items 25, 26, 31 and 32). Three studies provided an adequate standardisation procedure for assessing foot function (item 27) \[[@CR39],[@CR42],[@CR45]\] and five studies reported standardisation of injury outcome (item 33).
Clear reporting of all data was present in four studies (items 14 and 15) \[[@CR29],[@CR39],[@CR40],[@CR46]\]. However, the remaining seven studies primarily reported data only for significant relationships \[[@CR27],[@CR35],[@CR38],[@CR42]-[@CR45]\], while one study did not report any data \[[@CR41]\]. Only one study reported effects for all results (odds or risk ratios) (item 16) \[[@CR29]\]. With respect to generalisability of results, nine studies received a score of "Partial" (item 43) as results were deemed to be applicable to similar population groups to those investigated \[[@CR29],[@CR35],[@CR38]-[@CR45]\].
Study characteristics {#Sec12}
---------------------
The 12 included studies incorporated a total of 3,773 participants. Table [1](#Tab1){ref-type="table"} presents a summary of study characteristics. The participant population varied, with five studies investigating military personal \[[@CR27],[@CR29],[@CR39],[@CR41],[@CR43]\], five studies investigating runners \[[@CR38],[@CR40],[@CR42],[@CR44],[@CR46]\], and two studies investigating cohorts of physical therapy students \[[@CR35],[@CR45]\]. The types and incidence of lower limb overuse injuries reported were: tibial and femoral stress fractures, 8.7 to 10.0% \[[@CR29],[@CR39]\]; iliotibial band syndrome, 9.4% \[[@CR29],[@CR40]\]; patellofemoral pain, 4.0 to 17.0% \[[@CR27],[@CR29],[@CR42]-[@CR44]\]; medial tibial stress syndrome, 7.9% \[[@CR41]\]; Achilles tendinopathy, 5.1 to 15.8% \[[@CR29],[@CR44]\]; and non-specific lower limb overuse injuries, 14.0 to 20.6% \[[@CR35],[@CR38],[@CR45]\].Table 1**Summary of study characteristicsPopulationObservation period (activity, duration)Injury outcomeInjured groupUninjured groupGait assessmentFoot function measureN total (n females)Age (mean ± SD)N total (n females)Age (mean ± SD)**Hesar et al., \[[@CR38]\]Athletics club members3 running sessions/week; 10 weeksLL overuse injury27 (22)41 ± 8104 (89)39 ± 11Barefoot; 15 m runway; self-selected running speedPlantar loading (Footscan)Hetsroni et al., \[[@CR27]\]Military personal4 month basic training coursePatellofemoral painNRNRNRNRBarefoot; treadmill running at 5 km/hrRearfoot kinematics (Ariel Dynamics Inc.)Hetsroni et al., \[[@CR39]\]Military personal4 month basic training courseTibial and femoral stress fracturesDependent on outcome variable investigatedNRDependent on outcome variable investigatedNRBarefoot; treadmill running at 5 km/hrRearfoot kinematics (Ariel Dynamics Inc.)Kaufman et al., \[[@CR29]\]Military personal25 week training courseLL overuse injuryDependent on outcome variable investigatedNRDependent on outcome variable investigatedNRBoots and barefoot; self-selected walking speed (no mean or range presented)Plantar pressure ratios -- dynamic arch index (\<4.14 cavus, \>8.10 planus)Noehren et al., \[[@CR40]\]Female runnersIndividual non-specified running programs over a 2 year periodIliotibial band syndrome18 (18)26Dependent on outcome variable investigated28'Standard running shoe'; running along a 25 runway at a speed of 3.7 m/sRearfoot kinematics (Vicon)Noehren et al., \[[@CR46]\]Female runnersIndividual non-specified running programs over a 2 year periodPatellofemoral pain15 (15)27 ± 1015 (15)27 ± 10'Standard running shoe' (Nike, Pegasus); running along a 25 run way at a speed of 3.7 m/sRearfoot kinematics (Vicon)Sharma et al., \[[@CR41]\]Male infantry recruits26 week military trainingMedial tibial stress syndrome37 (0)NR239 (0)NRBarefoot; self selected walking speed (no mean or range presented)Plantar loading (Footscan)Thijs et al., \[[@CR42]\]Novice recreational runners10 week start to run programmePatellofemoral pain17 (16)39 ± 1085 (NR)37 ± 9Barefoot; walking at a self-chosen, moderate velocity (no mean or range presented)Plantar loading (Footscan)Thijs et al., \[[@CR43]\]Military personal6 week basic military trainingPatellofemoral pain36 (19)19 ± 248 (NR)19 ± 1Barefoot; walking at a self-chosen, moderate velocity (no mean or range presented)Plantar loading (Footscan)Van Ginckel et al., \[[@CR44]\]Novice runners10 week start to run programmeAchilles tendinopathy10 (2)38 ± 1153 (45)40 ± 9Barefoot; self-selected jogging pace (no mean or range presented)Plantar loading (Footscan)Willems et al., \[[@CR35]\]Physical education studentsUniversity physical education courseLL overuse injury46 (29)NR167 (NR)NRBarefoot; 3.3 m/s within a boundary of 0.17 m/sPlantar loading (Footscan)/ankle, knee and hip kinematics and kinetics (Proreflex)Willems et al., \[[@CR45]\]Physical education studentsUniversity physical education courseLL overuse injury46 (29)NR167 (NR)NR'Neutral running shoe'; 3.3 m/s within a boundary of 0.17 m/sPlantar loading (Footscan)/ankle, knee and hip kinematics and kinetics (Proreflex)LL = lower limb; NR = not reported.
Prior to prospective investigation, eight of the 12 studies investigated dynamic plantar loading (i.e. plantar pressure) \[[@CR29],[@CR35],[@CR38],[@CR41]-[@CR45]\], six investigated kinematic variables \[[@CR27],[@CR35],[@CR39],[@CR40],[@CR45],[@CR46]\] and one investigated rearfoot joint moments \[[@CR45]\] (Additional files [4](#MOESM4){ref-type="media"} and [5](#MOESM5){ref-type="media"}). A large number of plantar pressure variables were evaluated. Baseline measures of foot function were commonly performed during unshod gait \[[@CR27],[@CR29],[@CR35],[@CR38],[@CR39],[@CR41]-[@CR44]\], although four studies obtained measures during shod gait \[[@CR29],[@CR40],[@CR45],[@CR46]\]. Gait was assessed during treadmill walking at 5 kilometers per hour \[[@CR27],[@CR39]\], or during overground walking or running at a self-selected speed \[[@CR29],[@CR35],[@CR38],[@CR40]-[@CR46]\]. Only four studies that investigated overground running reported mean values of the speed at which participants were observed, ranging between 3.3 to 3.7 metres per second \[[@CR35],[@CR40],[@CR45],[@CR46]\].
Dynamic foot function variables as risk factors for lower limb overuse injuries {#Sec13}
-------------------------------------------------------------------------------
We found evidence supporting foot function as a risk factor for lower limb overuse injuries. There was *limited to very limited evidence* supporting (i) plantar loading and kinematic variables as risk factors for patellofemoral pain; (ii) plantar loading variables for Achilles tendinopathy; and (iii) plantar loading and kinematic variables for various non-specific lower limb overuse injuries. This is illustrated in Figure [2](#Fig2){ref-type="fig"}. For a complete reference of significant and non-significant findings for all injuries investigated, refer to Additional files [4](#MOESM4){ref-type="media"} and [5](#MOESM5){ref-type="media"}.Figure 2**Plantar pressure risk factors for: (A) patellofemoral pain (during walking); (B) patellofemoral pain (during running); (C) Achilles tendinopathy; and (D) non-specific injuries.** Force/pressure includes force time integral, impulse.
Patellofemoral pain {#Sec14}
-------------------
### Plantar loading variables {#Sec15}
There was *limited evidence* for plantar loading variables as a risk factor for patellofemoral pain, see Figures [2](#Fig2){ref-type="fig"}A and B. Participants who developed patellofemoral pain had earlier relative time to peak force in the lateral heel (SMD −0.56, 95% CI −1.09 to −0.37) and greater peak force in the second (0.65, 0.12 to 1.17) and third (0.60, 0.07 to 1.12) metatarsal regions during running \[[@CR42]\]. Those who developed patellofemoral pain also demonstrated greater lateral centre of pressure (COP) displacement (−0.47, −0.90 to −0.03) and lower maximal displacement velocity of the mediolateral COP (−0.85, −1.29 to −0.39) during the 'forefoot contact phase' of walking \[[@CR43]\].
### Kinematic variables {#Sec16}
There was *very limited evidence* for kinematic variables as a risk factor for patellofemoral pain, see Figure [3](#Fig3){ref-type="fig"}A. A single study \[[@CR27]\] investigated rearfoot kinematics, reporting opposite findings for the left and right sides. Greater pronation velocity on the left was a significant risk factor for patellofemoral pain development (quartile 4 versus quartile 3: RR 3.43 95% CI 1.32 to 8.96). Conversely, reduced pronation velocity of the right foot was a significant predictor of patellofemoral pain development (quartile 4 versus quartile 3: 0.38, 0.15 to 0.92). The authors did not specify whether the outcome (i.e. greater or reduced pronation velocity) was related to the side affected by patellofemoral pain.Figure 3**Kinematic risk factors for: (A) patellofemoral pain; and (B) non-specific injuries.**
Achilles tendinopathy {#Sec17}
---------------------
### Plantar loading variables {#Sec18}
There was *very limited evidence* for plantar loading variables as a risk factor for mid-portion Achilles tendinopathy, evaluated in one study \[[@CR44]\], see Figure [2](#Fig2){ref-type="fig"}C. Participants who developed Achilles tendinopathy exhibited significantly earlier time to peak force in the medial heel (SMD −0.716, 95% CI −1.39 to −0.02) and lateral heel (−1.08, −1.77 to −0.37), and delayed time to initial contact in the second metatarsal region (−1.00, −1.69 to −0.29). They also demonstrated greater peak force (0.84, 0.14 to 1.52) and a higher absolute force time integral (0.81, 0.11 to 1.49) in the fifth metatarsal region. In addition, those that developed Achilles tendinopathy displayed less anterior-posterior center of force (COF) displacement for the whole foot (−0.95, −1.64 to −0.25), greater laterally directed force in the forefoot at 'forefoot flat' (−0.88, −1.57 to −0.18) and a more posterior COF position at 'last foot contact' (−0.95, −1.63 to −0.24). During forefoot push-off, those that developed Achilles tendinopathy displayed more posterior COF displacement (−0.75, −1.43 to −0.05).
Non-specific lower limb overuse injuries {#Sec19}
----------------------------------------
There was *limited evidence* for plantar loading variables as a risk factor for non-specific lower limb overuse injuries, see Figure [2](#Fig2){ref-type="fig"}D.
### Plantar loading variables - discrete plantar regions {#Sec20}
Participants who developed a non-specific lower limb overuse injury exhibited delayed initial lateral heel contact (SMD 0.60, 95% CI 0.35 to 0.86) and terminal heel contact in the second and third metatarsal region (0.43, 0.18 to 0.68; 0.37, 0.12 to 0.62, respectively) \[[@CR35]\]. In the fifth metatarsal region, an increase in peak force (0.52, 0.09 to 0.95 \[[@CR38]\]) and absolute force-time integral (0.57, 0.14 to 1.00 \[[@CR38]\]), as well as delayed time until initial contact (0.32, 0.07 to 0.57 \[[@CR35]\]) were risk factors for non-specific overuse injury. However, contrary to these findings, Willems and colleagues reported lower fifth metatarsal region peak pressure (−0.44, −0.70 to −0.19) \[[@CR35]\] and absolute impulse (−0.31, −0.56 to −0.05 \[[@CR45]\]; −0.42, −0.67 to −0.17 \[[@CR35]\]) in those who developed non-specific lower limb overuse injuries.
### Plantar loading variables - time-specific gait events {#Sec21}
At first foot contact, participants who developed a non-specific lower limb overuse injury had a more laterally directed COP (SMD −0.47, 95% CI −0.73 to −0.22) \[[@CR45]\] and a more anterior COP position (0.31, 0.06 to 0.56) \[[@CR35]\]. At first metatarsal contact, participants who developed a non-specific lower limb overuse injury had greater lateral force as indicated by three mediolateral regional force ratios (−0.55, −0.97 to −0.12; −0.57, −0.99 to −0.13; −0.59, −1.02 to −0.16) \[[@CR38]\]. At forefoot flat, there was a lower velocity of the medio-lateral (−0.64, −1.07 to −0.21) and anterior-posterior displacement of the COF (−0.46, −0.88 to −0.03); and a more anterior COF position (0.61, 0.18 to 1.04) in those that developed non-specific lower limb overuse injuries \[[@CR38]\]. Willems et al. \[[@CR35],[@CR45]\] reported greater medial pressure as indicated by two pressure ratios (0.47, 0.22 to 0.72 \[[@CR35]\]; 0.40, 0.09 to 0.59 \[[@CR45]\]) and a more medially directed COP (0.38, 0.13 to 0.63) \[[@CR35]\]. At heel-off, participants who developed a non-specific lower limb overuse injury had a more laterally directed COF (−0.70, −1.13 to −0.27) \[[@CR38]\]. Contrary to this finding, Willems et al. \[[@CR35],[@CR45]\] reported greater medial pressure, as indicated by two pressure ratios (0.33, 0.07 to 0.58 \[[@CR35]\]; 0.33, 0.08 to 0.58 \[[@CR45]\]) in those who developed overuse injuries. At last foot contact, participants who developed a non-specific overuse injury had a more laterally directed COP (−0.81, −1.07 to −0.55) \[[@CR35]\], and more posterior COP position (−0.53, −0.79 to −0.28) \[[@CR35]\].
### Plantar loading variables (phase-specific gait events) {#Sec22}
During the initial contact phase, participants who developed a non-specific lower limb overuse injury had a more laterally directed plantar force (SMD −0.43, 95% CI −0.85 to −0.001) \[[@CR38]\]. Contrary to this finding, Willems et al. \[[@CR45]\] reported a more medially directed pressure, as indicated by one pressure ratio (0.57, 0.31 to 0.82) and a more medially directed COP displacement (0.61, 0.36 to 0.86).
Hesar et al. \[[@CR38]\] found that, during the forefoot contact phase, participants who developed a non-specific lower limb overuse injury had greater lateral COF displacement (−0.84, −1.27 to −0.40). Contrary to this finding, Willems et al. \[[@CR35],[@CR45]\] reported a greater medial pressure (0.54, 0.29 to 0.79) \[[@CR35]\] and a more medially directed COP displacement (0.58, 0.33 to 0.83 \[[@CR35]\]; 0.31, 0.05 to 0.56 \[[@CR45]\]).
Participants who developed a non-specific lower limb overuse injury had a more laterally directed COF displacement (−0.61, −1.03 to −0.17 \[[@CR38]\]) during the foot flat phase, and a more medially directed COF during the forefoot push off phase (0.52, 0.09 to 0.94 \[[@CR38]\]). Contrary to this latter finding, Willems et al. \[[@CR35],[@CR45]\] reported a more laterally directed pressure during forefoot push off, as indicated by one pressure ratio (−0.35, −0.60 to −0.09) \[[@CR45]\], and a more laterally directed COP displacement (−0.84, −1.09 to −0.58 \[[@CR35]\]; −0.37, −0.62 to −0.12 \[[@CR45]\]).
### Kinematic variables {#Sec23}
There was *limited evidence* for kinematic variables as a risk factor for non-specific lower limb overuse injuries, see Figure [3](#Fig3){ref-type="fig"}B. For the rearfoot segment, participants who developed a non-specific lower limb overuse injury exhibited a greater maximal eversion position (SMD 0.37, 95% CI 0.12 to 0.62) \[[@CR35]\], eversion excursion (0.36, 0.10 to 0.61 \[[@CR35]\]; 0.31, 0.06 to 0.56 \[[@CR45]\]), mean eversion velocity (0.37, 0.12 to 0.62) \[[@CR35]\], time to maximal eversion (0.39, 0.14 to 0.64) \[[@CR45]\], maximal eversion velocity (0.39, 0.14 to 0.64 \[[@CR35]\]; 0.29, 0.03 to 0.54 \[[@CR45]\]), mean inversion velocity (0.44, 0.18 to 0.69) \[[@CR35]\], maximal re-inversion velocity (0.41, 0.16 to 0.66) \[[@CR45]\], and mean re-inversion velocity (0.31, 0.06 to 0.56) \[[@CR45]\].
In the forefoot segment, participants who developed a non-specific lower limb overuse injury exhibited greater maximal abduction velocity (0.62, 0.37 to 0.88) \[[@CR35]\] and abduction excursion (0.36, 0.10 to 0.61 \[[@CR35]\]; 0.31, 0.06 to 0.56 \[[@CR45]\]). One study derived a three-dimensional pronation angle from eversion, abduction and dorsiflexion excursions, and reported that participants who developed a non-specific lower limb overuse injury exhibited greater three-dimensional pronation excursion (0.49, 0.23 to 0.74) \[[@CR45]\].
Other lower limb overuse injuries {#Sec24}
---------------------------------
There was no evidence supporting dynamic foot function as a risk factor for any other lower limb overuse injury. Non-significant effects were found for iliotibial band syndrome \[[@CR29],[@CR40]\] and stress fractures \[[@CR29]\].
Discussion {#Sec25}
==========
This systematic review evaluated current evidence for dynamic foot function as a risk factor for the development of lower limb overuse injuries. From six of the twelve studies included, we found very limited evidence that plantar pressure and kinematic variables representing dynamic foot function are associated with an increased risk of patellofemoral pain, Achilles tendinopathy and non-specific lower limb overuse injury \[[@CR35],[@CR38],[@CR42]-[@CR45]\]. Notably, significant findings reported across the studies had small to moderate effect sizes, and many 95% confidence intervals included zero, indicating non-significant findings.
Plantar pressure patterns associated with patellofemoral pain differed for walking and running gait. Risk factors in walking gait included greater lateral COP displacement and lower maximal displacement of the medio-lateral COP \[[@CR42]\], whereas for running gait risk factors included earlier time to peak force in the lateral heel and greater peak force in the second and third metatarsal region \[[@CR43]\]. While it is difficult to suggest a mechanism linking these plantar pressure differences with the development of patellofemoral pain, Thijs et al. \[[@CR42],[@CR43]\] speculated that these findings may indicate a resultant reduction in foot pronation during the loading phase of gait, and subsequent reduction in shock attenuation at the foot. This could increase transfer of ground reaction forces to more proximal structures, such as the patellofemoral joint.
Plantar pressure patterns associated with Achilles tendinopathy were evident from one study investigating jogging gait, and included earlier time to peak force in the lateral heel, less posterior COF displacement/more posterior COF position, greater laterally directed force and delayed time to initial contact in the second metatarsal region \[[@CR44]\]. Van Ginkel and colleagues \[[@CR44]\] speculated that these findings may indicate a more lateral foot roll-over following heel strike and diminished forward force transfer from the rearfoot to the forefoot. It is plausible that differences in force transfer across the foot may lead to altered loading of the Achilles tendon and contribute to injury, but this requires further evaluation.
Another consideration is that increased lateral loading at the foot is an adaptive response to proximal mechanics that increase medial lower limb loading. Prospective studies have shown that increased hip adduction during overground running \[[@CR46]\] and increased hip internal rotation when landing from a drop jump \[[@CR22]\] are risk factors for the development of patellofemoral pain. Furthermore, cross-sectional studies have reported deficits in neuromuscular control of the hip in those with patellofemoral pain \[[@CR55]-[@CR61]\] and Achilles tendinopathy \[[@CR62],[@CR63]\]. Further research is required to better understand the relationship between proximal and distal mechanics during gait, and risk of overuse injury development.
In contrast to evidence we found regarding plantar pressure, we found very few kinematic risk factors for lower limb overuse injuries. Our search strategy identified only one study that investigated kinematic risk factors for patellofemoral pain, which presented contradictory findings, no prospective studies that investigated kinematic risk factors for Achilles tendinopathy and two studies that reported differences in rearfoot eversion and forefoot abduction as risk factors for non-specific injuries \[[@CR35],[@CR45]\]. Whilst cross-sectional findings indicate differences in foot kinematics in people with patellofemoral pain \[[@CR64]\] and Achilles tendinopathy \[[@CR49]\], we found a lack of prospective kinematic data to indicate the temporal relationship between foot kinematics and overuse injury. Thus, at this time it is difficult to draw conclusions as to whether altered foot kinematics is a clear risk factor for lower limb overuse injuries.
In addition to necessitating more kinematic studies, consideration needs to be given to the method of measuring foot kinematics. Considering that overuse injuries generally involve cumulative exposure to load, it is plausible that those who develop overuse injuries demonstrate subtle kinematic differences that are not detectable by current kinematic measures. This is supported by previous findings regarding a lack of biomechanical coupling of plantar pressure indices and angular movements recorded between the calcaneus and the tibia \[[@CR65]\]. Further studies are required to increase understanding of this relationship, which could be achieved using more sophisticated three-dimensional and multisegment foot modeling techniques, and more clinically applicable measures of foot function.
Not surprisingly, it was difficult to identify a systematic pattern of plantar loading and kinematic risk factors for the category of 'non-specific injuries'. For example, significant risk factors were evident for greater lateral and medial directed COP, as well as increases and decreases in pressure-related outcomes in the fifth metatarsal region. While these findings indeed add evidence of a relationship between dynamic foot function and lower limb injury, the nature of the relationship is unpredictable, and likely relates to the variability of injuries evaluated under the term 'non-specific injuries'. Therefore, with the advancement and availability of diagnostic algorithms and imaging for lower limb injury, future research should avoid pooling all injuries, and instead focus efforts on exploring conditions that are discrete and well-defined. This is likely to enhance identification of injury-specific risk factors.
Interestingly, we found no evidence that dynamic foot function is a risk factor for iliotibial band syndrome or lower limb stress fractures including the foot. Findings from Noehren et al. \[[@CR46]\] indicated that aberrant hip mechanics may be a stronger risk factor for iliotibial band syndrome than dynamic foot function. They reported that increased hip adduction during running, but not rearfoot eversion, was a predictor of patellofemoral pain development in a cohort of 400 female runners \[[@CR46]\]. This is logical given the proposed mechanism of iliotibial band syndrome, where increased tension on the iliotibial band compresses the lateral femoral epicondyle \[[@CR46]\].
The lack of foot specific injuries (e.g. plantar fasciitis, metatarsal stress fracture) associated with dynamic foot function is another unexpected finding. Although Kaufman et al. \[[@CR29]\] reported that dynamic pes planus in shoes, measured as the ratio of midfoot contact area to total contact area, was a significant predictor of lower limb stress fracture (one third of which involved the foot), our effect size calculations were not significant. This is because the authors set significance at 0.10, whereas we used the more conventional alpha of 0.05. Because of the large number of variables evaluated, this is the more conservative approach to reduce the risk of type II error. An earlier study also reported that pronated foot type (i.e. static foot posture) was a significant risk factor for metatarsal stress fractures, while a supinated foot type was a risk factor for tibial and femoral stress fractures \[[@CR66]\]. However, the static x-ray measure of foot type used in this study may not correlate with dynamic foot function. It is plausible that lower limb stress fractures are more a function of bony overload due to the application of external loads, rather than the biomechanical characteristics of the foot. This is in part supported by the use of military cohorts in both studies \[[@CR29],[@CR66]\]. The influence of dynamic foot function on the development of lower limb stress fractures should be investigated in civilian populations to ascertain this.
Plantar loading variables were the most abundant risk factor identified for lower limb injury, albeit a relatively low risk with small to moderate effect sizes. In terms of the clinical application of these findings, it is difficult to map the plantar pressure risk factors to specific static foot types. De Cock et al. \[[@CR67]\] reported that participants with low arched feet had a more laterally directed COP across the gait cycle. This is consistent with our plantar pressure findings relating to patellofemoral pain and Achilles tendinopathy. Conversely, Wong et al. \[[@CR68]\] investigated the effect of foot morphology on center-of-pressure excursion during barefoot walking. Their findings indicated that more supinated foot types displayed a larger area of lateral COP excursion, and, conversely, more pronated foot types displayed a smaller area of lateral COP excursion. However, these findings were taken over the entire gait cycle, rather than the discrete phases evaluated in the prospective studies included in this review. In light of the volume of studies that use plantar pressure measures to evaluate dynamic foot function, there is a clear need for further studies to investigate methods of transferring plantar pressure information to clinically relevant measures.
Nevertheless, having some limited knowledge of the pattern of plantar loading risk factors may serve to inform the design of new and existing interventions that may redistribute or counter-balance plantar loading patterns observed in people at risk of injury. For example, arch-contoured foot orthoses alter plantar pressure systematically by reducing pressure in the forefoot and heel regions, and redistributing pressure to the midfoot \[[@CR69]\]. With this in mind, there is evidence from pooled data from randomised clinical trials (RCTs) that foot orthoses are effective in preventing lower limb overuse injuries \[[@CR70]\], as well as evidence from high-quality RCTs that foot orthoses reduce symptoms associated with patellofemoral pain \[[@CR71]\]. In the absence of evidence regarding kinematic effects, our findings suggest that foot orthoses may exert their clinical effects by redistributing plantar pressure (i.e. alter the magnitude, location and temporal patterns of reaction forces at the foot-orthosis interface). However, this requires further investigation.
Whilst this review has highlighted specific measures of dynamic foot function that are risk factors for lower limb overuse injuries, there are several limitations to the identification of these risk factors in a clinical practice setting. Firstly, while findings indicate the direction of altered plantar loading that may increase the risk of development of Achilles tendinopathy or patellofemoral pain, there are no reported thresholds of when an individual is deemed at risk (e.g. peak force in forefoot region exceeding 150 N). Future investigations are required to establish clinical guidelines and screening criteria for these risk factors. Secondly, the assessment of plantar pressures and three-dimensional kinematics requires expensive and sophisticated equipment that is not readily available in clinical practice settings, as well as specialised training in performing and processing these measurements. Future studies should investigate the translation of these laboratory-based measures to clinically applicable measures.
There are also limitations associated with the included studies. The majority of studies evaluated foot function while walking or running barefoot, which may limit the generalisability of findings to shod gait. While it is acknowledged that there are limitations associated with measuring plantar pressures and kinematics while wearing shoes, this is the condition that most closely resembles gait during daily and sporting activities. There were also differences between studies in the evaluation of overground versus treadmill gait analysis. As different gait patterns have been observed for treadmill and overground gait \[[@CR72],[@CR73]\], it may be inappropriate to measure dynamic foot function during treadmill gait in habitual overground runners, and vice versa. This may lead to a discrepancy between dynamic foot function measured during testing, and foot function during cumulative usual activity. A further limitation of this systematic review is that the methodological quality of the majority of included studies was generally poor. This was largely related to inadequate reporting of foot function measures, covariates, and non-significant results. Thus, the findings should be considered with this in mind. In order to enhance the overall quality of research in this field, future prospective studies should comply with published guidelines for minimum standards of reporting \[[@CR74]\].
Conclusion {#Sec26}
==========
This systematic review identified very limited evidence, with small to moderate effect sizes, that dynamic foot function during walking and running is a risk factor for patellofemoral pain, Achilles tendinopathy, and non-specific lower limb overuse injuries. More lateral plantar loading patterns were found to be risk factors for patellofemoral pain and Achilles tendinopathy. Findings from three studies indicate that there is no evidence that dynamic foot function is a risk factor for iliotibial band syndrome or lower limb stress fractures. At present, it is unclear whether these risk factors can be identified clinically (without sophisticated equipment), or modified to prevent or manage overuse injuries. Future prospective studies should address methodological limitations, avoid grouping different lower limb injuries in analyses, and explore clinically meaningful representations of dynamic foot function.
Additional files {#Sec27}
================
Additional file 1:**Search strategy.**Additional file 2:**Epidemiological Appraisal Instrument used to rate the quality of the 12 included studies.**Additional file 3:**Results from quality assessment using the Epidemiological Appraisal.** Instrument (12 included studies).Additional file 4:**Presentation of plantar loading variables across the 12 studies.**Additional file 5:**Presentation of kinematic and kinetic variables across the 12 studies.**
Please see related article: [www.jfootankleres.com/content/7/1/55](http://www.jfootankleres.com/content/7/1/55)
**Competing interests**
The authors declare that they have no competing interests.
**Authors' contributions**
GSM, MMFS, BSN, IBG, CB, SEM and NJC conceived the idea for this review. GSM, MMFS, BSN, IBG designed and piloted the search strategy. GJD undertook the search. Title and abstracts were reviewed by GJD and GSM. Quality appraisal was undertaken by GJD and NJC. Study information and data was extracted by GJD and GSM. The manuscript was drafted by GJD, GSM, MMFS, BSN, IBG, CB, SEM, and NJC. All authors have read and approved the final manuscript.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
At present, lung transplantation is still the best treatment for end-stage lung disease. Chronic rejection, the pathological manifestation of which appears as bronchiolitis obliterans (BO), also known as bronchiolitis obliterans syndrome (BOS), is the main factor affecting the long-term survival of patients who have undergone lung transplantation \[[@CR1], [@CR2]\]. Based on "The Registry of the International Society for Heart and Lung Transplantation: Thirty-second Official Adult Lung and Heart-Lung Transplantation Report," it was estimated that 5 years after lung transplantation 50% of the recipients would have developed BOS and 76% would have developed BOS 10 years after transplantation \[[@CR3]\]. In recent years, the study of lung transplantation and BOS has been facilitated by the establishment of several relatively mature experimental models of orthotopic left-lung transplantation in rats and mice, based on the cuff technique or other anastomotic methods \[[@CR4]--[@CR6]\].
Hypoxia-inducible factor-1 (HIF-1) is a transcription factor that becomes activated under hypoxia. It is involved in the immune/inflammatory reaction in various pathophysiological statuses, encoding or regulating more than 50 kinds of proteins including erythropoietin, glycolytic enzymes, induced-NOS, and vascular endothelial growth factor (VEGF) \[[@CR7]--[@CR10]\]. HIF-1, as a heterologous dimer, is composed of a functional alpha subunit and a structural beta subunit, and its biological effect is achieved by HIF-1 alpha (HIF-1α), which mediates the recruitment of bone marrow-derived angiogenic cells to increase arterial remodeling and angiogenesis during transplantation. Among the numerous target proteins associated with HIF-1, the protein of most concern is the VEGF, which is well-known for its pro-angiogenic and pro-fibrogenic effects. It has six isoforms, from VEGF-A to VEGF-F, where VEGF-A specifically promotes the division and proliferation of vascular endothelial cells and the migration of inflammatory cells. It is confirmed that HIF-1α signaling inhibition can downregulate VEGF induced fibrinogenesis and angiogenesis in vitro \[[@CR11]\]. Recent studies have found that VEGF-A can bind to VEGF receptor-2 (VEGFR-2) to enhance vascular permeability, promote the expression of cytokines, and induce the inflammatory cells' chemotaxis, which may play a role in the inflammatory response of transplanted organs \[[@CR8], [@CR12], [@CR13]\].
The oxygen concentration in the graft is different from normal after transplantation due to the change in anatomic structures and the involvement of immune, inflammation, and other internal environmental factors \[[@CR7], [@CR14]--[@CR16]\]. At present, it is not clear whether this change would affect the expression of HIF-1 and its downstream proteins, and relate to the rejection of the allograft. Thus, we studied the expression of HIF-1α and VEGF-A, and the relationship between these proteins and BO, in rats after orthotopic left-lung transplantation.
Material and methods {#Sec2}
====================
Animals {#Sec3}
-------
Fischer 344 (F344) and Lewis (LEW) male rats were purchased from the experimental animal center at Zhejiang Academy of Medical Sciences. They were reared in the SPF laboratory of the Key Laboratory of Multiorgan Transplantation at the Chinese Ministry of Health (The First Affiliated Hospital, Zhejiang University, Hangzhou). Animals weighing 250--300 g were used as both donors and recipients. All experimental animal protocols were reviewed and approved by the ethics committees at The First Affiliated Hospital and the Experimental Animal Center, Zhejiang University, China.
Orthotopic lung transplantation {#Sec4}
-------------------------------
The orthotopic left-lung transplantation was performed in the F344-to-Lewis rat strain combination as the allogeneic group (*n* = 24). As described in previous studies \[[@CR4]\], the F344 rats (RT1^1v1^) were used as donors and the Lewis rats (RT1^1^) as recipients. In the control group (*n* = 12), both the donors and the recipients were the Lewis rats. We had two trained laboratory technicians who dealt with donors and recipients, respectively. To reduce the time of cold ischemia and anesthesia of the recipient as much as possible, when one technician separates the donor lung hilar, the other one can start recipient surgery, so that pulmonary hilar anastomosis can be performed as soon as possible. Recipient operation time is from cutting the recipient rat skin, to the end of the skin suture. Warm ischemia time is composed of two parts, the first part is from donor rats superior vena cava disconnection to lung lavage completion, the second part starts with anastomosis, ends of the recipient pulmonary artery opening. Cold ischemia time is the transition time between the first and the second part of the warm ischemia time. Recipients of all groups received the same drug treatment, and specifically, with ciclosporin (Sandimmune, 50 mg/ml; Novartis, Nürnberg, Germany) on the first 10 days and lipopolysaccharide (LPS; L2654, from *Escherichia coli*, 026:B6, Sigma-Aldrich, Steinheim, Germany) once on the 28th day after transplantation.
Histology {#Sec5}
---------
The recipients were sacrificed 90 days after transplantation. The lungs were immediately fixed in situ by an intratracheal instillation of 4% formaldehyde for 24 h with subsequent paraffin embedding. The tissue sections were prepared and stained with hematoxylin-eosin (HE) and Masson's trichrome staining. The staging of the lung allograft rejection was completed according to the International Society for Heart and Lung Transplantation guidelines \[[@CR2]\]. The severity of fibrosis (SF) was measured on an arbitrary scale of 0 to 4 for reference \[[@CR17], [@CR18]\].We conducted the histopathologic analysis in a blinded fashion.
Immunohistochemistry {#Sec6}
--------------------
Immunohistochemistry staining was performed by the streptavidin peroxidase complex method to detect HIF-1α, VEGF-A, and VEGFR-2. The immunohistochemistry score was the product of the staining intensity and percentage of positive cells. The staining intensity was graded from 0 to 3+ (0 for colorless, 1+ for pale yellow, 2+ for brown-yellow, and 3+ for saddle-brown). The score for the percentage of positive cells was from 0 to 4+ (0 for negative, 1+ for fewer than 10% positive cells, 2+ for 10--50% positive cells, 3+ for 51--75% positive cells, and 4+ for over 75% positive cells). The immunohistochemistry score ≥ 3 was considered positive immune response.
Statistical analysis {#Sec7}
--------------------
All numerical data of the experimental results were presented as mean ± standard deviation and categorical data as count and percentage. The statistical analyses were accomplished using the GraphPad Prism 5.0 (GraphPad Software, San Diego, CA). Tests were performed with the Pearson's chi-square test for count data and the student T test for measurement data. All *p*-values were two-tailed and differences were considered statistically significant at *p* \< 0.05.
Results {#Sec8}
=======
Animal model of lung transplantation {#Sec9}
------------------------------------
We have successfully completed a total of 35 rat lung transplantations, with an average operation time for the recipient of about 27 min and a 97.2% success rate (35/36). Cold ischemia time is about 10 min, and warm ischemia time is about 12.5 min. There were two deaths in the allogeneic group: one was caused by a pleural hemorrhage on the third day after surgery and the other by a digestive obstruction on the fiftieth day after the operation. In the isogeneic group, one rat's pulmonary vein was torn during the operation, eventually leading to death by hemorrhagic shock and another died 3 days after the surgery due to a pulmonary embolism. So finally, we have 22 allogeneic recipients in the allogeneic group and 10 isogenic recipients in the control group.
Each recipients received the chest computed tomography scan at specific points \[[@CR19]\]. We selected a typical example from each group (Fig. [1](#Fig1){ref-type="fig"}). On day 27, the computed tomography images of the two groups were similar for clear lung fields and unobstructed bronchus (Fig. [1](#Fig1){ref-type="fig"}a and e). On day 30 after surgery, the diffuse infiltration of bilateral lung parenchyma was seen in every model and the surrounding of the bronchus was more serious as the LPS was applied intratracheally (Fig. [1](#Fig1){ref-type="fig"}b and f). Sixty days after transplantation, lung fields appeared clearer than the previous scan (Fig. [1](#Fig1){ref-type="fig"}c) while the allogenic left lungs still had an increased lung field density and emphysema appeared in the right native lungs (Fig. [1](#Fig1){ref-type="fig"}g). Ninety days postoperatively, the lung fields of the isografts and the right native lungs of the allogenic recipients were clear (Fig. [1](#Fig1){ref-type="fig"}d and h). However, the allogenic left lungs showed disseminated high density shadows with local atelectasis and mediastinum shifted to the left (Fig. [1](#Fig1){ref-type="fig"}h).Fig. 1High-resolution computed tomography imaging of lung isografts (**a**-**d**) and allografts (**e**-**h**). The chest computed tomography imaging was performed in the anesthetized animal on 27 (**a** and **e**), 30 (**b** and **f**), 60 (**c** and **g**), and 90 (**d** and **h**) days after transplantation. The white arrows in left pulmonary point to the algate aerated bronchus
By observing the paraffin sections stained with HE under a light microscope, we found airway remodeling in sections of the allogenic left lungs, and the bronchial lumina displayed severe stenosis or occlusion (Fig. [2](#Fig2){ref-type="fig"}a), obviously different from the native right lungs and the isografts (Fig. [2](#Fig2){ref-type="fig"}b). The inflammatory cells predominantly consisted of mononuclear leukocytes that infiltrated the bronchial mucosa and the area surrounding the bronchus. Around the bronchioles, there are proliferative bronchial mucosal epithelial cells and smooth muscle cells. In addition to bronchiolitis, It may be accompanied by abnormal thickening of arterial intima (Fig. [2](#Fig2){ref-type="fig"}a). Fibrous connective tissue, such as collagen, was stained in blue when Masson's trichrome staining was applied. We found dense bands of collagen surrounding bronchioles, replacing large areas of lung parenchyma, in sections of the allogenic left lungs (Fig. [2](#Fig2){ref-type="fig"}c), which differ from the isografts (Fig. [2](#Fig2){ref-type="fig"}d).Fig. 2Histopathology of lung allografts (**a** and **c**) and isografts (**b** and **d**) in HE and Masson's trichrome staining. The magnification of pictures was 200×. HE, hematoxylin-eosin staining; Allo left, allogenic left lung; Iso left, isogenic left lung
Expression of HIF-1α increased in the transplanted lung {#Sec10}
-------------------------------------------------------
We used immunohistochemical staining with antigen-specific antibodies to better quantify and locate the proteins. As described above, we divided the pathological sections into positive or negative immune response groups, also called high- or low-expression groups, according to the immunohistochemistry score. The percentages of high-expression of HIF-1α, VEGF-A, and VEGFR-2 in the allogeneic group were 56.82, 45.45, and 54.55%, compared with 35, 20, and 40% in the isogeneic group (Fig. [3](#Fig3){ref-type="fig"}a), respectively. When we counted the data on the left and right lungs individually, we found that the positive rates of high-expression of HIF-1α, VEGF-A, and VEGFR-2 in the isograft were similar to the right lung in situ of the two groups, but significantly lower than in the allograft (77.27, 63.64, and 68.18%, respectively) (Fig. [3](#Fig3){ref-type="fig"}b). Simultaneously, we found that the locations of the different proteins inside the cells were not exactly the same. HIF-1α was mainly expressed in the nucleus because it was a transcription factor, which mainly functions in the nucleus, whereas VEGF-A and VEGFR-2 were mainly observed in the cytoplasm (Fig. [4](#Fig4){ref-type="fig"}).Fig. 3Expression of HIF-1α, VEGF-A and VEGFR-2. The positive rate of related protein's expression is different between **a** the isografts (*n* = 10) and the allografts (*n* = 22), and **b** in different position of lung. Iso, isograft; Allo, allograft; HIF-1α, Hypoxia inducible factor-1α; VEGF-A, vascular endothelial growth factor-A; VEGFR-2, vascular endothelial growth factor receptor-2; R, right lung; L, left lungFig. 4Immunohistochemistry of HIF-1α, VEGF-A and VEGFR-2 in lung grafts or in situ. Protein expression was assessed as positive (**a**, **d** and **g**) or negative (**b**, **c**, **e**, **f**, **h** and **i**) depending on the immunohistochemistry score in different position. Positive immune response group had more brown-staining cells in particular areas than the negative group. The magnification of pictures was 200×. HIF-1α, Hypoxia inducible factor-1α; VEGF-A, vascular endothelial growth factor-A; VEGFR-2, vascular endothelial growth factor receptor-2
Expression of HIF-1α related to the severity of pulmonary fibrosis {#Sec11}
------------------------------------------------------------------
The severity of pulmonary fibrosis, evaluated in the Masson's trichrome staining, was mainly dependent on the thickness and number of fibers in the connective tissue. The SF of the allograft was 3.05 ± 0.84, apparently higher than the right lung in situ (1.09 ± 0.61, *P* = 3.60 × 10^− 9^) and the left lung of the isograft (1.16 ± 0.62, *P* = 5.39 × 10^− 7^), respectively. And, there were no statistical difference between the left and right lung of the isograft (1.16 ± 0.62, 1.10 ± 0.57) (Fig. [5](#Fig5){ref-type="fig"}a).Fig. 5The severity of pulmonary fibrosis of allografts and isografts. Expression of HIF-1α, VEGF-A and VEGFR-2 is related to the severity of pulmonary fibrosis. **a** n = 10 (Iso) and n = 22 (Allo); **b** *n* = 32 (HIF-1α(−)) and *n* = 32 (HIF-1α(+)); **c** *n* = 40 (VEGF-A(−)) and *n* = 24 (VEGF-A(+)); **d** *n* = 32 (VEGFR-2(−)) and *n* = 32 (VEGFR-2(+)); All data shown are the mean ± standard deviation. ^\*\*\*^*p* \< 0.01 and ^\#\#\#^*p* \< 0.01 with T test. F score, The severity of fibrosis score; Iso, isograft; Allo, allograft; HIF-1α, Hypoxia inducible factor-1α; VEGF-A, vascular endothelial growth factor-A; VEGFR-2, vascular endothelial growth factor receptor-2; R, right lung; L, left lung
Then, all samples were divided into two groups to compare whether there were statistical differences in SF, based on whether the immunohistochemistry score ≥ 3 of the corresponding protein. The results showed that the SF was higher in samples with high protein expression (HIF-1α: 2.50 ± 1.09 vs. 1.06 ± 0.67, *P* = 3.88 × 10^− 10^; VEGF-A: 2.79 ± 1.03 vs. 1.18 ± 0.71, *P* = 7.53 × 10^− 13^; VEGFR-2: 2.31 ± 1.19 vs. 1.25 ± 0.81, *P* = 6.13 × 10^− 10^) (Fig. [5](#Fig5){ref-type="fig"}b-d). These results indicated that the expression of HIF-1α, VEGF-A, and VEGFR-2, and the severity of pulmonary fibrosis, was positively correlated.
Correlation between HIF-1α, VEGF-A, and VEGFR-2 {#Sec12}
-----------------------------------------------
By calculating the correlations among HIF-1α, VEGF-A, and VEGFR-2, we found that HIF-1α and VEGF-A, VEGF-A and VEGFR-2, and HIF-1α and VEGFR-2 were all positively correlated (Table [1](#Tab1){ref-type="table"}).Table 1Correlation between HIF-1α, VEGF-A and VEGFR-2ParametersnVEGF-A (−) (*n* = 40)VEGF-A (+) (*n* = 24)χ^2^*p*-value HIF-1α (−)3230 (20)2 (12)26.672.41 × 10^−7^ HIF-1α (+)3210 (20)22 (12)ParametersnVEGFR-2 (−) (*n* = 32)VEGFR-2 (+) (*n* = 32)χ^2^*p*-value HIF-1α (−)3222 (16)10 (16)9.002.70 × 10^−3^ HIF-1α (+)3210 (16)22 (16)ParametersnVEGFR-2 (−) (*n* = 32)VEGFR-2 (+) (*n* = 32)χ^2^*p*-value VEGF-A (−)4026 (20)14 (20)9.601.95 × 10^−3^ VEGF-A (+)246 (12)18 (12)All samples are from both side of all recipient lung (the number of allogeneic group is 44 and the number of control group is 20) and divided into two groups based on whether the immunohistochemistry score ≥ 3 of the corresponding protein. (−) represents immunohistochemistry score \< 3; (+) represents immunohistochemistry score ≥ 3; *HIF-1α* Hypoxia inducible factor-1α, *VEGF-A* Vascular endothelial growth factor-A, *VEGFR-2* Vascular endothelial growth factor receptor-2
Discussion {#Sec13}
==========
The aim of our study was to explore the relationship between BO after lung transplantation in rats and the expression of HIF-1, etc. In the pretest, we searched for the appropriate animal model. Based on the literature \[[@CR6], [@CR20]\], we tried several models and found that there were different outcomes in the allografts for different strains. For instance, when Sprague-Dawley rats were used as recipients, there were very few rejections, yet when Brown Norway rats acted as recipients, the grafts were prone to acute rejection. As for trachea transplantation, although it is similar to BOS in immunology and morphology, we thought that the orthotopic lung transplantation model can better simulate chronic rejection if technical difficulties can be overcome. Finally, we selected the F344-to-Lewis rat strain combination as reported by Atanasova et al. \[[@CR4]\] Based on our results, we believe that the chronic rejection outcomes demonstrated by this model resemble the 2007 revised criteria for the diagnosis of lung rejection \[[@CR2]\]. On the ninetieth day after transplantation, the allograft sections showed bronchial mucosal epithelial hyperplasia with a large number of inflammatory cells infiltrating the bronchial mucosa and the area surrounding the bronchus. This was accompanied by the presence of coarse collagen and a large amount of proliferative smooth muscle. In addition to the proliferation of arterial intima, this resulted in concentric or eccentric stenosis or occlusion of the bronchial lumina. This morphology is similar to the BOS of human lung transplantation.
Chronic rejection of lung transplantation is the main limitation for long term survival of lung transplantation patients, which is associated with airway fibrosis. Studies demonstrated that the airways might be relatively hypoxic after lung transplantation, which had an influence upon the metabolism of graft \[[@CR7], [@CR16]\]. In our study, we found that the expression of HIF-1α, VEGF-A, and VEGFR-2 in the allografts was more than isografts. As is well-known, the HIF-1α was an important nuclear transcription factor that can be activated under hypoxic conditions to mediate increased transcription of genes involved in angiogenesis, growth factor signaling and oxygen transport to either increase O~2~ delivery or decrease O~2~ consumption \[[@CR7], [@CR10], [@CR21], [@CR22]\]. HIF-1--mediated transcriptional responses activate the expression of angiogenic cytokines and growth factors that stimulate angiogenesis by recruiting bone marrow-derived angiogenic cells as well as endothelial cells. The recent studies have shown that HIF-1α has a crucial role in critical aspects of cell metabolism and mammalian embryogenesis, and contributes to pathogenesis for a variety of diseases, such as hereditary erythrocytosis, pulmonary arterial hypertension, cancer and airway constrictive diseases \[[@CR23]\].
In the acute phase of hypoxia, the tissue's blood supply could mainly be affected by the angiotensin system and, when entering the chronic hypoxia process, this could compensate for a lack of oxygen through vascular remodeling. VEGF-A and VEGFR2 are a pair of ligands that induce angiogenesis in vivo. Abraham, Paulus, and others have found that early in transplantation, even in the donor organs preservation period, the level of HIF-1α in grafts was raised, and the transcription of the VEGF-A gene was up-regulated. This would directly affect the vascular permeability and the degree of edema after transplantation and, ultimately, the outcomes of the grafts \[[@CR9], [@CR12], [@CR13], [@CR24]\]. Our team expects that although HIF-1α, VEGF-A, and other proteins were up in the early stage of transplantation, their influence was perhaps not on the allograft rejection, but on increasing the vasopermeability and the recruitment of inflammatory cells. Due to bronchial arteries are not re-established, which are the main arterial blood supply for pulmonary, the grafts would be in hypoxemia after transplant surgery \[[@CR13], [@CR16], [@CR22]\]. As a result, we suppose that if the state of hypoxia had not been changed, HIF-1α and downstream proteins might have had a continuous high expression. However, as there was no access to any relevant literature, our team completed the experiment and the results confirmed our hypothesis.
In other solid organ transplantation studies, researchers have found that HIF-1α, VEGF, and NOS were released in the anoxic condition. Macrophages, monocytes, and other inflammatory cells were then recruited and induced to secrete cytokines, such as transforming growth factor-1β, thrombospondin-1 and so on, which might promote the occurrence of graft fibrosis \[[@CR7], [@CR11], [@CR13], [@CR14], [@CR24]--[@CR27]\]. Researchers have recently used normal human bronchial epithelial cells in vitro to confirm that HIF-1a is associated with pulmonary fibrosis after transplantation \[[@CR11]\]. Yet, Jiang et al. have shown a pro-angiogenic role of HIF-1α to mediates airway microvascular repair and thereby attenuate airway fibrotic remodeling by using a model of orthotopic tracheal transplantation. Studies demonstrated that chronic anoxia impaired HIF-1-dependent VEGF expression in graft \[[@CR10]\]. When exogenous intervention upregulated HIF-1α, graft airway perfusion accelerated recovery by increasing expression of angiogenic factors and recruitment of host bone marrow-derived angiogenic cells to the airway \[[@CR28]\]. The contradiction between the complex mechanisms and the phenomenons may be due to the different periods of research objects, as well as the characteristics of self-protection and the limitations of self-repair. In this way, it would be of great significance to explore the role of HIF-1α in different models and stages, for its apparently opposing roles.
According to the revision of the 1996 working formulation for the standardization of nomenclature in the diagnosis of lung rejection \[[@CR2]\], BO is defined as a progressive fibrous obliteration of the small airways, an all-or-none diagnosis, with no index for judging the degree of severity. Thus, to more comprehensively assess the extent of graft rejection, we introduced the arbitrary scale of severity of fibrosis to score the sections \[[@CR17], [@CR18]\]. We found that the expression of HIF-1α, VEGF-A, and VEGFR2 was related to the degree of graft fibrosis. We confirmed a link between the HIF-1α-VEGF-A pathway and BO, in a long-term survival model of orthotopic lung transplantation in rats for the first time.
Conclusions {#Sec14}
===========
We focused on the status of HIF-1α, VEGF-A, and VEGFR2 in BO after lung transplantation and found a link between these three proteins. According to our data, we think that VEGF-A, as a target protein of HIF-1, has specific binding to the VEGFR2 transmembrane protein, activating downstream secondary messengers to regulate the transcription of the relative genes involved in angiogenesis, pro-fibrogenesis, and other pathophysiological manifestations. In conclusion, in our model of orthotopic left-lung transplantation, BO was observed and the expression of HIF-1α, VEGF-A and VEGFR2 in the allograft tissue was up-regulated. It is thus suggested that the activation of the HIF-1α-VEGF pathway may be involved in the development and prognosis of BO after lung transplantation.
BO
: Bronchiolitis obliterans
BOS
: Bronchiolitis obliterans syndrome
F344
: Fischer 344
HE
: Hematoxylin-eosin
HIF-1α
: Hypoxia inducible factor-1 alpha
LEW
: Lewis
LPS
: Lipopolysaccharide
Ltx
: Lung transplantation
SF
: Severity of fibrosis
VEGF-A
: Vascular endothelial growth factor-A
VEGFR-2
: Vascular endothelial growth factor receptor-2
This manuscript has been edited for English language by the professional academic editing company Armstrong-Hilton Limited.
Funding {#FPar1}
=======
This study was supported by National Natural Science Foundation of China (Project 81373161) and three awards from the Health Department of Zhejiang province, China (2012ZDA014, 2012ZDA017, 2012KYA081) and a program of Chinese Medicine Research Program of Zhejiang Province (2012ZQ016).
Availability of data and materials {#FPar2}
==================================
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
HX, YY, CZ and JH contributed to the conception of the study. AA, WL and YY gave valuable suggestions to the experiments. HX, ZZ and CZ performed all the experiments and statistical analyses. HX, CZ and JH drafted the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate {#FPar3}
==========================================
All animal experiments were approved by the ethical review committees from The First Affiliated Hospital and the Experimental Animal Center, Zhejiang University, China.
Consent for publication {#FPar4}
=======================
Not applicable.
Competing interests {#FPar5}
===================
The authors declare that they have no competing interests.
Publisher's Note {#FPar6}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-ijms-21-03165}
===============
Diabetes is a serious, long-term condition with a major impact on the lives and well-being of individuals, families, and societies worldwide. The global diabetes prevalence in 2019 is estimated to be 9.3% (463 million people), rising to 10.2% (578 million) by 2030 and 10.9% (700 million) by 2045 \[[@B1-ijms-21-03165]\]. Population aging is also increasing dramatically throughout the world, especially in developing countries, creating pressures on the health system as well as social security services and policies. In Vietnam, diabetes is projected to be one of the top seven diseases leading to death and disability by 2030 \[[@B2-ijms-21-03165],[@B3-ijms-21-03165]\]. With the increasing prevalence of diabetes, there are approximately 5.76 million people with diabetes currently living in Vietnam. The age-adjusted comparative prevalence of diabetes in the population of Vietnam was approximately 6% in 2017 \[[@B2-ijms-21-03165]\]. Nowadays, many people are familiar with type 1 or type 2 diabetes mellitus, however, there is another form of diabetes that has just recently been identified, known as type 3 diabetes (T3DM). This lesser-known type manifests as insulin resistance within the brain and has major potential to impact neurocognition and contributes to the etiology of Alzheimer's disease \[AD\]. AD has already been identified as the sixth leading cause of death in the United States, and the fifth leading cause of mortality in people 65 and older \[[@B4-ijms-21-03165]\]. It has no current cure, but treatments for symptoms are available and research continues. Neurotransmitter deficits, degenerated neurons, synaptic dysfunction, extracellular buildup of β-amyloid (Aβ) and intracellular neurofibrillary tangles (NFT) are the major crude disfigurements present in AD \[[@B5-ijms-21-03165]\]. To produce Aβ peptides of different lengths such as Aβ38, Aβ40, and Aβ42 due to the active enzymatic component of the γ-secretase complex, presenilin 1 (PSEN1), and PSEN2, the amyloid precursor protein (APP) cleaves at several sites within the membrane. Unfortunately, diabetes is following right behind AD as the seventh leading cause of mortality and is projected to affect almost half a billion people by the year 2045 \[[@B1-ijms-21-03165]\]. Both diseases have been recognized to have multifactorial interactions involving both the environment and to a lesser degree, genetics. Yet, insulin insensitivity has been linked to memory deficits, cognitive decline, and many of the characteristic symptoms that have been displayed in AD. At the same time, type 2 diabetes has remained one of the most adjustable risk factors for the development of AD. DM may be classified into four clinical categories: type 1, type 2, type 3, and type 4. Type 1 diabetes (T1D) is mainly due to β-cell destruction, mostly leading to absolute insulin deficiency. The type 2 diabetes (T2D) is due to a progression of insulin secretary defect concomitantly with insulin resistance. Insulin resistance is a common phenomenon, closely associated with obesity, and defined as the inability of target tissues to respond normally to insulin. Insulin resistance typically precedes the onset of type 2 diabetes by several years. T2D is a risk factor for dementia and for AD, the most common type of dementia. T1D is mainly observed in children and young adults, while T2DM is more common among adults and is responsible for 90% of the incidences globally \[[@B6-ijms-21-03165]\]. Some epidemiological studies suggest that insulin resistance increases the risk for dementia and AD, even in nondiabetic populations.
A recently discovered form has been suggested to be termed type 3 diabetes mellitus (T3DM) by scientists. These scientists have tried to define it as a metabolic syndrome that may lead to abnormalities linked to progressive brain insulin resistance with consequent impairment of central insulin signaling processes, accumulation of neurotoxins, neuronal stress, and resulting in a course of neurodegeneration \[[@B7-ijms-21-03165],[@B8-ijms-21-03165]\]. In vitro and animal studies indicated that insulin resistance can contribute to the pathogenesis of AD through multiple different pathways \[[@B7-ijms-21-03165]\]. Endocrine abnormalities---especially diabetes---are common in AD, which is also regarded as a type of diabetes. Diabetes having an influence on memory processing (recognition and retrieval), morphology of brain (brain atrophy) and synaptic communication is a well demonstrated hazardous aspect that influences the pathology of AD \[[@B9-ijms-21-03165]\]. In addition, the hyperinsulinemia impairment of insulin signaling and insulin resistance are the vital factors that make the sense of keeping insulin at the center stage of both pathologies irrespective of genotype \[[@B10-ijms-21-03165]\]. Many recent studies have indicated that impaired hippocampus insulin signaling impairs the memory and other executive functions, attributing to the decline of insulin signaling and concurrent development of insulin resistance \[[@B11-ijms-21-03165],[@B12-ijms-21-03165],[@B13-ijms-21-03165]\]. This deliberation advocates a strong link between hyperinsulinemia and insulin resistance and the resultant pathologies like T3D and AD \[[@B14-ijms-21-03165]\]. Peripheral insulin resistance leads to decrease insulin signaling in CNS, followed by alteration in brain metabolism. Increased Aβ toxicity, Tau hyperphosphorylation, oxidative stress and neuroinflammation are attributed to central insulin resistance, which leads to neurodegeneration. The work provides the relationship between T3DM and AD based on the fact that both the processing of amyloid-β (Aβ) precursor protein toxicity and the clearance of Aβ are attributed to impaired insulin signaling in the brain. Furthermore, insulin-related therapeutic strategies are suggested to succeed in the development of therapies in AD by slowing down their progressive nature or even halting their future complications. [Figure 1](#ijms-21-03165-f001){ref-type="fig"} reveals the concept of T3D regarding AD and its approaches for treatment and prevention.
2. Insulin and Glucagon Signaling in the Central Nervous System (CNS) {#sec2-ijms-21-03165}
=====================================================================
Insulin is a hormone that regulates glucose levels in the blood, is produced by the beta cells of the islets of Langerhans in the pancreas, and consists of two polypeptide chains connected by disulfide linkages. Insulin initiates its action by binding to transmembrane glycoprotein receptors formed by two α and two β-subunits \[[@B14-ijms-21-03165]\]. Insulin binding to α-subunits of the receptors fabricate confirmative alterations that lead to its activation and autophosphorylation of several Tyr residues at β-subunit cytosolic region \[[@B15-ijms-21-03165],[@B16-ijms-21-03165]\]. Autophosphorylated remnants are then acknowledged by the insulin receptor substrates (IRS), out of which IRS-1 and IRS-2 are the two major players and the common intermediaries in insulin signal propagation. IRS is ideal and suitable for the configuration of molecular complexes which mediates intracellular signaling pathways. Insulin and insulin-like growth factors (IGF-1) connect to tyrosine kinase receptors, the insulin receptor (IR) and IGF-1. Insulin binding is highest in the olfactory bulb, cerebral cortex and hippocampus. Furthermore, insulin receptors are also expressive on endothelial cells of the blood--brain barrier and are responsible for transport of insulin and IGF-1 through the blood--brain barrier (BBB) into CNS \[[@B17-ijms-21-03165]\]. While the exact mechanism of how insulin gets into the brain still remains controversial, insulin circulating in the blood can cross the BBB through a receptor-mediated active transport system \[[@B17-ijms-21-03165]\]. This pathway is consistent with studies showing that insulin levels in the cerebrospinal fluid (CSF) increase proportionally with blood insulin after peripheral insulin infusion \[[@B15-ijms-21-03165],[@B16-ijms-21-03165],[@B17-ijms-21-03165]\]. However, the amount of insulin produced in the brain and whether this pool of insulin is physiologically relevant still remains elusive. It is possible that both the centrally and peripherally derived pools of insulin are important for signaling in the brain.
Insulin and IGF-1 are conferred with functions which are important for neuronal survival and the maintenance of CNS integrity. Insulin receptors and insulin signaling affect glucose homeostasis, neuronal integrity and cognition through influencing several receptor-mediated mechanisms including calcium influx, neurotransmitter build-up and synaptic connections, apoptosis, and neurogenesis \[[@B17-ijms-21-03165]\]. Insulin also regulates expression and levels of GABA, NMDA and AMPA-mediated mechanisms which have a strong influence over long-term potentiation (LTP) and long-term depression (LTD). Furthermore, insulin is crucially involved in expansion and preservation of excitatory synapses \[[@B18-ijms-21-03165]\] and dendritic spine formation through the activation of AKT--mTOR and Ras-related pathways \[[@B19-ijms-21-03165],[@B20-ijms-21-03165]\] which are integral to insulin signaling \[[@B21-ijms-21-03165]\]. Insulin also influences cell survival by modulating apoptotic pathways and the intermediates involved in the apoptotic cascade \[[@B22-ijms-21-03165],[@B23-ijms-21-03165]\].
The presence of insulin in the brain was first reported by Havrankova et al. \[[@B24-ijms-21-03165]\] who used radioimmunoassay to determine high levels of insulin in brain extracts. Also, high insulin concentrations had then been reported not only in the human brain but also in several experimental animals \[[@B25-ijms-21-03165]\]. Recently, the production of insulin in the CNS has also been widely studied, and suggestions of possible insulin biosynthesis in the brain have been based on different experimental evidence. Evidence of the presence of insulin mRNA was found in the periventricular nucleus of the rat hypothalamus by in situ hybridization \[[@B26-ijms-21-03165]\]. The molecular mechanisms involved in the production and secretion of insulin in the CNS reveal similarities between beta cells and neurons, particularly in relation to ATP-sensitive K+ channel depolarization that have been demonstrated \[[@B27-ijms-21-03165]\]. This depolarization-induced release of insulin was able to be inhibited by cycloheximide, and was specific for neurons, but not for astrocytes \[[@B28-ijms-21-03165]\]. Interestingly, dysfunctions in the insulin receptors (IRs)-mediated processes might be due to abnormalities in IR activation, lowered insulin availability, and compromised IR-triggered downstream mechanisms resulting in a broad range of brain disorders \[[@B29-ijms-21-03165]\]. Also, IR associated with RNA polymerase II in the nucleus, with striking enrichment at promoters genome-wide have recently been demonstrated \[[@B30-ijms-21-03165]\]. These results reveled that IR interacts with transcriptional machinery at promoters, and identify a pathway regulating genes linked to insulin's effects in physiology and related diseases \[[@B30-ijms-21-03165]\]. Thus, through influencing any of these pathways, insulin alters the neuronal performance and integrity which may result in defects in learning, memory and other features of AD. Previous studies indicated that brain insulin was equally reduced in AD patients and age-matched controls, indicating that reductions in brain insulin are likely a result of age, not AD \[[@B31-ijms-21-03165]\]. Ultimately, a greater understanding of insulin in the brain relative to the severity of AD and age-matched controls needs to be obtained in order to fully comprehend insulin's function in healthy and diseased brains. Thus, reduced insulin levels in the CNS can lead to reduced levels of antiamylogenic proteins, and both the overproduction and an impaired clearance of Aβ.
3. The Role of Type 3 Diabetes in Glucose Homeostasis {#sec3-ijms-21-03165}
=====================================================
The key to understanding the relationship between diabetes and these other areas begins with the role of energy homeostasis in diabetes. Energy homeostasis is a well-regulated process that depends on the coordination between feeding behavior and energy expenditure. The control of energy homeostasis in humans has received much attention in recent years due to alterations caused by onset of conditions such as obesity and diabetes. There are two distinct features of adult neurons that make them vulnerable to either neuronal cell death or a diseased state such as neurodegeneration or neuronal loss. The first feature is that fully differentiated (adult) neurons are permanently postmitotic cells, which lack regenerative ability \[[@B32-ijms-21-03165]\]. Therefore, when adult neurons are exposed to any cellular stresses such as lack of adenosine triphosphate (ATP) moieties or energy crisis or oxidative stress, they either die or experience apoptosis, or degenerate or cause neuronal degeneration and loss, and thus predispose neurodegenerative diseases \[[@B32-ijms-21-03165]\]. The second important feature is that brain neurons or tissues are highly demanding excitable cells, in which more than 40% of the present ATP is used to keep neurons viable or alive \[[@B33-ijms-21-03165]\]. There are two sources of brain glucose that involve cortical glucose metabolism stimulation through basal insulin levels \[[@B34-ijms-21-03165]\] and astrocytic glycogen conversion to glucose that is stimulated by the activation of glial β-adrenoceptors. The increase in glucose uptake is transported by insulin-sensitive glial glucose transporter type 1 (GLUT1) to the plasma membrane for neuronal use. Therefore, the balanced cellular glucose transportation depends on astrocytes and glucose transporters that are expressed in the brain \[[@B35-ijms-21-03165]\].
Moreover, a glucose homeostasis defect might be important in the pathogenesis of T3DM due to impaired glucose uptake as a result of impaired glucose metabolism in the brain. The mechanisms that are involved in glucose transportation abnormalities include brain insulin resistance and intracellular glucose metabolic disturbance. These two abnormalities may contribute to cerebral glucose hypometabolism in T3DM or the brain insulin resistance disease state. A decreased glucose transporters correlated to abnormal hyperphosphorylation of tau in neurodegenerative diseases was reported \[[@B36-ijms-21-03165]\]. Therefore, impairment of insulin signaling not only affects systemic glucose blood levels but also causes various degenerative processes or neuronal cell death or loss \[[@B37-ijms-21-03165]\]. In addition, insulin resistance in T2DM has been defined as "reduced sensitivity in body tissues to the action of insulin" \[[@B38-ijms-21-03165]\]. Similarly, brain insulin resistance can be defined as the failure of brain cells to respond to insulin and its corresponding IRs \[[@B39-ijms-21-03165]\]. Consequently, this leads to insulin deficiency and impaired glucose transport inside the neurons due decreased number of expressed GLUTs in the cell membrane. Furthermore, insulin resistance in the CNS correlates with insulin resistance in the periphery. Therefore, loss of responsiveness to insulin could make neurons more susceptible to neurotoxic insults due to their being devoid of protective effect of insulin \[[@B40-ijms-21-03165]\]. Furthermore, insulin-resistant patients have many increased pathologic features such as apoptosis, neurodegeneration, and the resultant decline in cognition.
The desensitization of the neuronal insulin receptor in brain insulin resistance, similar to the process in T2DM, may play a key role in causing T3DM and its future complications \[[@B41-ijms-21-03165]\]. Besides, T2DM is a metabolic syndrome characterized by insulin resistance, which is also a pathological feature of neurodegeneration or neuroendocrine disorder or T3DM \[[@B34-ijms-21-03165]\]. Thus, glucose homeostasis plays a role in T3DM pathogenesis. Brain glucose uptake or metabolism is impaired in T3DM. Therefore, the combination of T2DM and neurodegenerative brain diseases may be considered as this new classification of diabetes, called T3DM or a neuroendocrine disorder.
4. Type 3 Diabetes and Aβ Protein Pathology {#sec4-ijms-21-03165}
===========================================
Amyloidosis is a pathological condition which consists of the accumulation of fibrillary proteins, characterizing by extracellular amyloid deposits with a clinical variability depending on the affected tissue. Recently, there has been newly emerged evidence regarding the relationship between the pathogenesis of AD and insulin resistance. It is important to consider T2DM as a risk factor essential for the formation of deposits of amyloid-β in patients' brains with dementia. There was a toxic cycle between continuous insulin exposure and Aβ accumulation inside the neurons \[[@B42-ijms-21-03165]\]. According to Farris et al., insulin degrading enzyme (IDE) regulates the levels of insulin, Aβ protein, and amyloid precursor protein (APP) intracellular domain in vivo \[[@B42-ijms-21-03165]\]. This study showed that a rat model of T2DM of mutant IDE was associated with hyperinsulinemia and glucose intolerance, as hallmarks of T2DM and T3DM or brain insulin resistance. This implies that IDE hypofunction may underlie or contribute to some forms of T3DM and T2DM and provide a mechanism for the recently recognized association among hyperinsulinemia, diabetes, and neurodegeneration or neuronal loss \[[@B42-ijms-21-03165]\]. Therefore, in normal subjects, IDE reduces Aβ, regulates insulin and also degrades APP intracellular domain (AICD). Thus, there was a regulatory relationship among insulin, IDE and Aβ. In the case of brain insulin resistance, insulin possibly failed to stimulate the clearance of Aβ, which permits its buildup inside the neurons causing neurodegeneration or neuronal loss, as hallmarks of T3DM or brain insulin resistance \[[@B42-ijms-21-03165]\]. There is a debate about T3DM and brain insulin resistance as to whether it is a consequence or a cause of abnormal Aβ expression and protein processing \[[@B43-ijms-21-03165]\]. In terms of the concept of T3DM being a consequence, Aβ toxicity may cause insulin resistance in the brain. The Aβ disturbs insulin signaling by competing with insulin on its receptors \[[@B44-ijms-21-03165]\], reducing the surface expression of IRs, and reducing the insulin affinity to its relative receptors, and interfering directly with phosphatidylinositol-4, 5- bisphosphate 3-kinase (PI3K)/Akt activation, causing a blockade of its signaling and leading to impaired survival signaling, increased activation of GSK-3β activity, and increased hyperphosphorylation of tau \[[@B45-ijms-21-03165]\].
On the other hand, in terms of the concept of T3DM being the cause, the brain insulin resistance with oxidative stress and neuroinflammation may cause Aβ accumulation, as shown in [Figure 2](#ijms-21-03165-f002){ref-type="fig"}. The studies that incorporate this concept claim that insulin stimulation may increase or accelerate trafficking of Aβ from the Golgi network to the plasma membrane. Therefore, insulin may activate Aβ extracellular excretion and, at the same time, inhibit its intracellular accumulation by activating its degradation by the insulin-degrading enzyme (IDE) \[[@B46-ijms-21-03165]\]. Thus, impaired insulin signaling can disturb both APP processing and Aβ clearance \[[@B47-ijms-21-03165]\]. This leads to increased neurotoxic effects of Aβ on neurons, resulting in possible neurodegeneration and neuronal cell death. T2DM and AD patients have similar amyloid beta deposits both in pancreas and in the brain. Several researchers have suggested this new pathology should be addressed as T3D \[[@B48-ijms-21-03165],[@B49-ijms-21-03165],[@B50-ijms-21-03165],[@B51-ijms-21-03165]\]. Some of the target receptors of T2DM such as the insulin-like growth factor 1 (IGF-1) and peroxisome proliferator-activated receptor gamma (PPARG) are also involved in the regulation of the expression and phosphorylation of tau protein \[[@B51-ijms-21-03165]\].
5. Type 3 Diabetes Regarding Alzheimer's Disease {#sec5-ijms-21-03165}
================================================
Insulin resistance in AD and diabetes can lead to hyperinsulinemia, thereby, saturating insulin-degrading enzymes (IDE) for insulin and Aβ degradation. Recently, many studies indicated that the incidence of AD is higher in T2D patients and obese individuals, implying common mechanisms driving these disorders \[[@B10-ijms-21-03165],[@B52-ijms-21-03165],[@B53-ijms-21-03165]\]. Insulin resistance could be a main feature which is shared among diabetes, obesity, and AD \[[@B54-ijms-21-03165]\]. The neuronal glucose uptake may not depend on insulin totally, thus the concept of insulin resistance in the brain is more related to impaired insulin signaling pathways. The malfunction of insulin signaling pathways and resultant state of hypometabolism observed are considering among factors in altered bioenergetics that connects AD and T2D \[[@B55-ijms-21-03165]\]. The insulin resistant state could lead to compromised neuron functions and cognitive skills, accompanied by an extreme rise in insulin and relatively declined insulin activity in the periphery as important predictors of T2D \[[@B56-ijms-21-03165],[@B57-ijms-21-03165]\]. Consequently, this leads to the development of neuritic plaques, hippocampal atrophy, cognitive performance and lower cerebrocortical glucose metabolism which may closely correlate with memory impairments \[[@B50-ijms-21-03165]\]. A previous study revealed that increased p-Ser312IRS1 manifested in prodromal AD patients that sustained these alterations a decade ago as AD patients \[[@B58-ijms-21-03165]\], suggesting that insulin resistance in AD develops years before clinical manifestations and that neural-derived exosomes carry potential for early AD diagnosis. Due to lack of insulin response, down regulation of insulin receptors, reduced binding of insulin receptors or faulty activation of the insulin signaling cascade cause the defective brain insulin signaling in AD and T2D. The major consequence of this altered cascade is the decreased neuronal glucose uptake that is manifested as impaired neuroplasticity, neurotransmitter deficits, collapse of bioenergetics mechanism and initiation of fateful inflammatory cascade. Overall, the consequences of impaired insulin signaling are attributed to impaired metabolism in the brain that may lead to brain malfunction, providing possible explanations for the connection between diabetes, obesity, and AD \[[@B11-ijms-21-03165]\], as shown in [Table 1](#ijms-21-03165-t001){ref-type="table"}.
Insulin resistance or dysfunction of insulin signaling is a universal feature of T2D, due to altered glucose metabolism and its interdependence on cell death pathways form the basis of linking T3D with AD, as shown in [Figure 3](#ijms-21-03165-f003){ref-type="fig"}. T3D occurs when neurons in the brain become unable to respond to insulin, which is essential for basic tasks, including memory and learning. Some researchers believe insulin deficiency is central to the cognitive decline of AD. Dysfunctional insulin pathways and resistance of insulin is a status of receptor dysfunction, altered receptor expression, deviations in receptor binding and malfunctioned events in the phosphorylation chain or the altered activities related to kinases involved in phosphorylation. At the molecular level, a cell senses insulin through insulin receptors, with the signal propagating through a signaling cascade collectively known as the PI3K/Akt/mTOR signaling pathway. Recent studies suggested that the pathway operates as a bistable switch under physiologic conditions for certain types of cells, and insulin response may well be a threshold phenomenon \[[@B13-ijms-21-03165],[@B59-ijms-21-03165],[@B60-ijms-21-03165]\]. The pathway's sensitivity to insulin may be blunted by many factors such as free fatty acids, causing insulin resistance. It also is based on the finding that insulin resistance may be reversed rapidly by exposing cells to mitochondrial uncouplers, electron transport chain inhibitors, or mitochondrial superoxide dismutase mimetics \[[@B61-ijms-21-03165],[@B62-ijms-21-03165]\].
Interestingly, impaired insulin signaling is present in several transgenic and nontransgenic mouse models of AD. Some previous clinical studies have reported that AD patients could have glucose intolerance, suggesting a bidirectional relationship between the two conditions \[[@B63-ijms-21-03165],[@B64-ijms-21-03165]\]. Reduced levels of IRS-1 associated to the membrane of hippocampal extracts \[[@B65-ijms-21-03165]\] and a decreased activation of IRS-1 and PI3K in the hippocampus and cortex were observed in ten-month-old mice \[[@B66-ijms-21-03165]\]. Markers of insulin resistance were also reported in the hypothalamus of APP/PS1 mice \[[@B67-ijms-21-03165]\] since the IRS-1 phosphorylated in serine 616 in the hippocampus at nine months of age was higher than that of the control group \[[@B68-ijms-21-03165]\], and increased levels of IRS-1 phosphorylated in serine 636 and 312 in the frontal cortex at 13 months \[[@B69-ijms-21-03165]\] were also demonstrated. In combination with peripheral insulin resistance, there was also a report of an increased inhibitory phosphorylation of IRS-1 in serine 612 in the hippocampus of five-month-old tg2576 mice \[[@B66-ijms-21-03165]\]. Remarkably, the central infusion of AβOs lead to peripheral insulin resistance, which was further observed in the APP/PS1 and in the 3xTgAD mouse models of AD \[[@B70-ijms-21-03165]\]. To confirm these concepts, further evidence is still required to investigate the mechanisms whereby AD affects the diabetic phenotype. T3D regarding AD and its approaches for treatment and prevention using naturally synthetic compounds, as shown in [Figure 2](#ijms-21-03165-f002){ref-type="fig"}.
In addition, in the wake of the worldwide increase in T2DM, a major focus of research aims to understand the signaling pathways impacting this disease. Insulin signaling regulates glucose, lipid, and energy homeostasis, predominantly via action on liver, skeletal muscle, and adipose tissue. Cell signaling pathways can be described by a list of biomolecular reactions which occur between the pathway components. T2DM associated with impaired insulin and insulin-like growth factor-1 (IGF1) signaling (IIS) is a risk factor for cognitive impairment and dementia including AD \[[@B71-ijms-21-03165]\]. Importantly, systemic heterozygous inactivation of IGF1R (IGF1R+/−) or neuronal deletion of IGF1R (nIGF1R−/−) could improve the survival in the Tg2576 mouse model of AD while reducing behavioral impairment and Aβ accumulation \[[@B72-ijms-21-03165]\]. Reduced IRS2 signaling throughout the body or in the brain prolongs life span \[[@B73-ijms-21-03165]\] may lead to systemic reduction of IRS2 (IRS2−/−), improves cognitive function, and reduces Aβ deposition and premature mortality in Tg2576 mice with normal blood glucose levels \[[@B72-ijms-21-03165],[@B74-ijms-21-03165]\]. Hence, more recent animal studies have revealed that a reduction in intracellular signaling mediated by IGF1R-IRS2 signaling but not the IR cascade in the CNS exerts neuroprotective effects in AD animal models \[[@B71-ijms-21-03165]\].
6. Therapeutic Approaches to Type 3 Diabetes in Alzheimer's Disease {#sec6-ijms-21-03165}
===================================================================
Insulin resistance is well known as an essential feature of T3D, therefore treatment strategies for T3D, particularly those aimed at improving insulin sensitivity, may also benefit those patients at risk for AD at the early stages. Due to the overlapping yet distinct pathological features among diabetes, insulin resistance and cognitive decline, multitargeted drug therapies along with lifestyle interventions are also explored \[[@B75-ijms-21-03165]\] from the perspective of research in the pharmaceutical industry, including nutraceuticals, antioxidant activity \[[@B76-ijms-21-03165]\], polyphenols, omega-3 fatty acids as well as the brain--gut connections \[[@B77-ijms-21-03165]\].
Among nutraceuticals produce, a brain-permeable compound, curcumin is able to target abnormal protein aggregates \[[@B78-ijms-21-03165]\]. Curcumin may also thwart "proapoptotic signaling pathways in primary hippocampal neuron cultures". Previous research has also shown the benefit of metformin in mice when coupled with curcumin and piperine supplementation, particularly regarding enhanced insulin sensitivity, signaling, and better systemic glucose tolerance \[[@B78-ijms-21-03165]\], promising natural substances for AD patients. However, the anti-inflammatory benefits of fruits and vegetables have been widely publicized for decades, particularly regarding antioxidant action in reducing inflammatory damage \[[@B79-ijms-21-03165]\]. Rodent research has linked various vegetables and fruits as protective "against cognitive and brain neuropathology from dietary oxidative stress" due to innumerable bioactive constituents such as carotenoids, antioxidant vitamins, polyphenols and flavonoids \[[@B80-ijms-21-03165]\]. A various families of flavonoids have been suggested to be the potential therapeutic implications via in vivo models \[[@B81-ijms-21-03165]\]. This has significant potential to advance our understanding of proactive approaches toward preventing AD and inhibiting progression. The essential role of omega-3 fatty acids in brain development and maintenance has been well recognized, particularly in the past ten years, yet only recently "have their effects on brain aging been explored" \[[@B82-ijms-21-03165]\]. Diets rich in omega-3 fatty acids and naturally low in omega-6fatty acids may hold the key for nutritional therapy for AD patients \[[@B83-ijms-21-03165]\]. The ketogenic diet may even diminish and clear beta amyloid plaques within the brain, while convalescing damaged mitochondria and reducing universal inflammation \[[@B84-ijms-21-03165]\]. New research has shown that glycated APOE4 protein and faulty insulin signaling leads not only to impaired energy transport for brain tissues, but also impaired lipid transportation, mainly cholesterol \[[@B84-ijms-21-03165],[@B85-ijms-21-03165]\]. APOE4 accounted for approximately 20% of the general population and \>50% among Alzheimer's cases, is responsible for interrupting how the brain processes insulin \[[@B86-ijms-21-03165]\]. The gene and the peripheral insulin resistance caused by the high-fat diet together induced insulin resistance in the brain \[[@B87-ijms-21-03165]\]. The APOE4 protein produced by the gene can bind more aggressively to insulin receptors on the surfaces of neurons than its normal counterpart, APOE3. APOE4 goes on to do lasting damage to brain cells. After blocking the receptor, the sticky APOE4 protein begins to clump and becomes toxic \[[@B87-ijms-21-03165]\]. Furthermore, once the protein enters the interior of the neuron, the clumps get trapped within the cell's machinery, impeding the receptors from returning to the neuron surface to do their work. The insulin signal processing gets increasingly more impaired, starving brain cells. There is no pharmaceutical intervention that has ever existed that has been more potent in improving overall vasculature throughout the body, than exercise \[[@B88-ijms-21-03165]\]. This also has extensive implications for AD patients and type 2 diabetics due to increases in quality of life, neurochemical messaging within the brain, restorative power over insulin resistance, and the ability to clear Aβ plaques in certain individuals \[[@B88-ijms-21-03165]\]. The concept of the gut--brain axis, the bidirectional communication between gut and brain, contributing significantly to the pathogenesis of AD that has been supported by many experimental and clinical studies \[[@B77-ijms-21-03165]\]. Representatives of some compounds and drugs for the treatment or prevention of T3D regarding AD progression are presented in [Table 2](#ijms-21-03165-t002){ref-type="table"}.
7. Conclusions {#sec7-ijms-21-03165}
==============
The relationship between T3DM and AD is based on the fact that both the processing of AβPP and the clearance of Aβ are attributed to impaired insulin signaling in the brain. Additionally, it focuses on the molecular mechanism of brain insulin resistance that may involve either increased serine phosphorylation of IRS-1 protein (i.e., IRS-1 inhibition) and elevated degradation of IRS protein as common pathological mechanisms, including aggregation of toxic Aβ plaques, tau hyperphosphorylation and autophagy. Increasing the knowledge and awareness of the term T3D has the potential to pave the way for disease treatment, prevention and possibly even deliver a cure. Currently, there have been no particular treatments with established efficacy in counteracting cognitive decline or AD, so the implications of identifying AD as a disorder with an etiology rooted in faulty insulin signaling and irregular energy pathways could be critical in disease management. While the specific mechanisms between AD and all forms of diabetes remain convoluted and unclear, which subsequently may have devastating socioeconomic impacts on public health and healthcare systems, T3D has the potential to provide a plethora of proactive and therapeutic strategies to current patients. For now, it seems that the testing of more anti-T3D drugs with beneficial effects against cognitive impairment has a certain promising future.
Conceptualization, T.T.N., Q.T.H.T., T.K.O.N., T.T.D.N., and V.V.G.; collection the literature, writing---review and editing the manuscript, T.T.N., Q.T.H.T., T.T.D.N., and V.V.G.; substantive supervision and revision over the article, T.T.N., and V.V.G.; project administration, V.V.G.; funding acquisition, V.V.G.; All authors have read and agreed to the published version of the manuscript.
This research was supported by a National Research Foundation of Korea (NRF) Grant, awarded by the Korean government (Ministry of Education, Science and Technology, No. NRF-2019R1G1A109740012).
The authors declare no conflict of interest.
{#ijms-21-03165-f001}
{#ijms-21-03165-f002}
{#ijms-21-03165-f003}
ijms-21-03165-t001_Table 1
######
Causal model for the potential associated with between T3D and AD.
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Upstream Risk Factors Metabolic Precursors Pathways Subclinical Pathology Disease Outcome
------------------------------------------------------------------------------------ --------------------------------------- ------------------------------------- ----------------------- ---------------------------- ---------------------
Social factors: stress, low socioeconomic status, certain ethnic and racial groups Obesity\ **Vascular Processes**\ Cerebral blood flow\ Amyloid precursor proteins Alzheimer's disease
Visceral Adiposity Blood pressure and hypertension\ Atherosclerosis
Hyperlipidemia\
Apolipoprotein E
Poor diet: high in calories, fat and sugar, low in fiver **Inflammatory/Oxidative processes**\ Neurofibrillary\
Inflammation\ Tangles\
Oxidative stress\ Amyloid B deposits
Endothelial function
Physical inactivity\ Hyperglycemia\ **Metabolic processes**\
Genetics and family history Hyperinsulinemia Insulin resistance\
Insulin-degrading enzyme\
Peroxisome proliferative-activated\
receptors
Early childhood exposures in utero and birth weight Brain and hippocampal atrophy\
White matter hyperintensities
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
ijms-21-03165-t002_Table 2
######
Summary of representative of preclinical and clinical studies on the efficacy of antidiabetic, insulin-sensitizing drugs on multiple aspects of AD pathology.
Compound Potential Pathway Study Design Reference
---------------------- --------------------------------------------------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------- -------------------------------------------------
DA5-CH Reduces tau phosphorylation and normalizes theta rhythm Injected intracerebroventricula (ICV), streptozotocin on rat \[[@B89-ijms-21-03165]\]
DA-JC1 Antagonizing circadian rhythm disorders induced by Aβ~31--35~ ICV, amyloid(31--35) AD model \[[@B90-ijms-21-03165]\]
DA5-CH Improved of hippocampal synaptic plasticity and activation of the PI3K/AKT signaling pathway APP/PS1 mouse model of AD \[[@B91-ijms-21-03165]\]
DA-CH3 Reduced ER stress and apoptotic signaling, reduced amyloid plaque load in the brain APP/PS1 mouse model of AD \[[@B92-ijms-21-03165]\]
Insulin Prevention of Aβ oligomer induced synapse loss and insulin receptor reduction, amelioration of PKR-mediated ER stress Rat hippocampal neuronal cultures \[[@B93-ijms-21-03165],[@B94-ijms-21-03165]\]
Insulin AD patients that are not ε4 carriers have reduced sensitivity to insulin, effecting cognitive performance AD patients homozygous or not for the ApoE-ε4 allele and normal subjects intravenously injected \[[@B95-ijms-21-03165]\]
Insulin Improved verbal memory in MCI AD ε4-subjects after acute insulin administration, but not in ε4 carriers AD patients homozygous or not for the ApoE-ε4 allele, MCI patients and most subjects intranasally administrated \[[@B96-ijms-21-03165],[@B97-ijms-21-03165]\]
Insulin Chromic intranasal insulin doses enhanced selective attention, retention of new information and functional status of MCI and early AD subjects AD patients, MCI patients and normal subjects intranasally administrated \[[@B98-ijms-21-03165]\]
Insulin Only women presented improved working memory after treatment Healthy men and woman intranasally administrated \[[@B99-ijms-21-03165]\]
Liraglutide Reduction of tau phosphorylation; protection of insulin reception and synapse loss in a c-AMP dependent manner Cynomolgus monkeys ICV with Aβ oligomer \[[@B100-ijms-21-03165]\]
Liraglutide Improvement of memory deficits in novel object recognition test and fear conditioning Swiss mice injected ICV with Aβ oligomer \[[@B100-ijms-21-03165]\]
Liraglutide Restored memory deficits in object recognition test and Morris water maze; enhanced LTP; reduced microglial activation; diminished Aβ plaque load APP/PSEN1 mice \[[@B101-ijms-21-03165],[@B102-ijms-21-03165]\]
Exendin-4 Decrease in the inhibitory phosphorylation of Ser312IRS1, Ser66IRS1 of INK, while restoring activating Tyr465 IRS1 phosphorylation Rat hippocampal neural cultures \[[@B69-ijms-21-03165]\]
Exendin-4 Improvement of spatial memory in the Morris water maze; reduced amyloid plaque LOAD APP/PS1 mice \[[@B69-ijms-21-03165]\]
Exedin4- Liraglutide eIF2α phosphorylation reduction Rat hippocampal neural cultures, APP/PS1 mice, cynomolgus monkeys injected ICV with Aβ oligomer \[[@B94-ijms-21-03165]\]
GLP-1 Exendin-4 Reduction of neural excitotoxicity Rat hippocampal neural cultures, rats injected on the basal nucleus with ibotenic acid \[[@B103-ijms-21-03165]\]
Rosiglitazone Reversal of memory deficits in objects recognition test and the Morris water maze; Aβ levels reduction AD transgenic mice J20 line \[[@B104-ijms-21-03165]\]
[^1]: These authors contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
*Nocardia* is a genus belonging to the aerobic actinomycetes group of bacteria which are Gram-positive bacilli and showing branching filamentous forms \[[@B1]\]. They are saprophytic ubiquitous bacteria which can be found in several environments such as fresh water and saltwater, soil, dust, decaying vegetation, and decaying fecal deposits from animals \[[@B1]\]. Nevertheless, these environmental bacteria can be opportunistic pathogens and lead to human infectious diseases called "nocardiosis" \[[@B2]\]. Nocardiosis can be discriminated into two groups: invasive infection, mainly caused by*N. asteroides,*presenting commonly as pneumonia in patients who are immunocompromised, have underlying chronic lung disease, and are with a possible dissemination to other organs \[[@B3]\], and cutaneous infection via a cut or abraded skin, which can be manifest clinically as (i) abscess and cellulitis, (ii) lymphangitis, (iii) skin infection secondary to dissemination, and (iv) actinomycetoma. This latter group is the most amazing infection due to their severity characterized by the presence of tumefaction, subcutaneous nodules, destructive granulomata, fistulas, and pus \[[@B2], [@B4]\].
*N. brasiliensis* is the species isolated from the majority (approximately 80%) of cases of cutaneous nocardiosis, especially in actinomycetoma \[[@B2]\]. This species is more commonly isolated in areas with tropical or subtropical climates such as South America, Asia, and Africa. Due to false diagnosis, rural lifestyles, and poor access to care in these countries,*N. brasiliensis* nocardiosis constitutes a real public health problem that can lead, in the absence of treatment, to amputations and death in young populations. On the basis of epidemiological surveys conducted in France, the number of cases of nocardiosis between 2000 and 2007 according to the French Nocardiosis Observatory (OFN) was 607 with*N. farcinica* and*N. nova* being the most frequent species \[[@B5]\]. However, no data currently exists on the phylogenetic relationships between the indigenous*N. brasiliensis* strains of tropical origin and native strains isolated in France. Routine genus/species identification of*Nocardia*was based on macroscopic, microscopic, and biochemical characteristics. The methods described by Boiron et al. \[[@B6]\] were used to determine the decomposition of adenine, casein, hypoxanthine, tyrosine, and xanthine. In addition to the phenotype-based methods, species-level identification is mainly genetically based, nowadays. Classically, 16S rRNA*(rrs)* gene sequencing is generally used for the species-level identification \[[@B7], [@B8]\], but it fails to discriminate among some species of*Nocardia* because it does not have enough polymorphism to differentiate them at the species level. Multilocus sequence analysis (MLSA) using concatenated sequences of several housekeeping genes such as superoxide dismutase A*(sodA)* and heat shock protein 65*(hsp65)* has been increasingly used to provide higher accuracy and discriminatory power in the molecular identification of*Nocardia* spp. \[[@B9], [@B10]\]. Indeed, a recent study seeking to identify new molecular targets shows that the polymorphism observed in the*sodA* gene sequence contains variable regions that allow the discrimination of closely related*Nocardia* species \[[@B9]\].
The aim of the present study was to perform a genetic characterization and assess the phylogenetic relationships of*Nocardia brasiliensis* focusing on using housekeeping*rrs*,*hsp65,* and*sodA* genes, for 36 autochthonous*N. brasiliensis*strains isolated in France and analyzed by the OFN between 2002 and 2012. Phenotypic characterization was also conducted by assessing antimicrobial resistance profiles, metabolic profiles, and culture condition.
2. Materials and Methods {#sec2}
========================
2.1. Bacterial Strains and Culture Media {#sec2.1}
----------------------------------------
A collection of 36 human clinical strains of*N. brasiliensis* was studied ([Table 1](#tab1){ref-type="table"}). All strains were identified as such, at species level by the French Nocardiosis Observatory (OFN) by genetic approach. Moreover, six*Nocardia* reference strains belonging to*N. brasiliensis* clade \[[@B9]\] were also used:*N. brasiliensis* ATCC 19296^T^ (unknown),*N. altamirensis* DSM 44997^T^ (karstic cave),*N. boironii* DSM 101696^T^ (pus sample),*N. iowensis* DSM 45197^T^ (garden soil),*N. tenerifensis* DSM 44704^T^ (rhizosphere), and*N. vulneris* DSM 45737^T^ (human leg wound). Prior to the assays, strains were cultured 72 hours in Bennett medium (made in the laboratory) aerobically at 37°C.
2.2. Growth Test on Culture Media {#sec2.2}
---------------------------------
From 0.5 McF bacterial suspension, bacterial growth was evaluated on three culture media: (i) bromocresol purple (BCP) (Biomérieux, Marcy l\'étoile), (ii) Bennett (made in the laboratory), and (iii) Middlebrook (Biomérieux, Marcy l\'étoile). One hundred microliters from bacterial suspension standardized was inoculated on the different plate of culture media. The plates were incubated at 37°C and the observations were performed at 48, 72, and 96 hours.
2.3. Antimicrobial Susceptibility {#sec2.3}
---------------------------------
The susceptibility of the isolates to different antimicrobials was determined by disk diffusion method with a panel of 31 antibiotics (Biorad, Marnes-la-Coquette France) on Muller Hinton E medium (Biomérieux, Marcy l\'étoile, France). Susceptibility testing was done with amikacin 30 *μ*g, gentamycin 15 *μ*g, tobramycin 10 *μ*g, ciprofloxacin 5 *μ*g, levofloxacin 5 *μ*g, moxifloxacin 5 *μ*g, minocycline 30 *μ*g, doxycycline 30 *μ*g, tigecycline 15 *μ*g, cefotaxime 30 *μ*g, ceftriaxone 30 *μ*g, cefepime 30 *μ*g, cefuroxime 30 *μ*g, amoxicillin 25 *μ*g, amoxicillin + clavulanic acid 20/10 *μ*g, ampicillin 10 *μ*g, ertapenem 10 *μ*g, meropenem 10 *μ*g, imipenem 10 *μ*g, vancomycin 30 *μ*g, pristinamycin 15 *μ*g, erythromycin 15 *μ*g, trimethoprim + sulfamethoxazole 1.25/23.75 *μ*g, rifampicin 30 *μ*g, and linezolid 30 *μ*g.
From visible colonies, bacterial suspension was done in sterile water, using a cotton swab to obtain a concentration of 0.5 McFarland according to the Clinical and Laboratory Standards Institute standard M24-A2 \[[@B12]\]. Seeding was done according to the swab method. In this latter, the bacterial inoculum was spread on the agar using a sterile cotton swab in three different directions. The disks were dispensed with a dispenser and the plates were incubated at 37°C for 72 hours and read manually according to the thresholds defined in the recommendations of the SFM 2013 \[[@B11]\].
2.4. Substrate Degradation {#sec2.4}
--------------------------
The methods of Boiron et al. \[[@B6]\], Goodfellow et al. \[[@B13], [@B14]\], and Goodfellow and Lechevalier \[[@B15]\] were used to determine the decomposition of adenine, casein, and uric acid \[[@B9]\]. Clinical strains of*N. brasiliensis* and the strains of species belonging to the*N. brasiliensis* clade (*N. brasiliensis, N. altamirensis, N. iowensis, N. tenerifensis*,*N. boironii,* and*N. vulneris*) were tested \[[@B9]\]. Strains*N. boironii* DSM 101696^T^,*N. brasiliensis* ATCC 19296^T^, and*N. vulneris* DSM 45737^T^ were incubated at 37°C, and*N. altamirensis* DSM 44997^T^,*N. tenerifensis* DSM 44704^T^, and*N. iowensis* DSM 45197^T^ were incubated at 28°C \[[@B9]\]. The readings were performed at 3, 7, 10, 14, 17, and 21 days.
2.5. Methods of DNA Extraction {#sec2.5}
------------------------------
DNA extraction from*Nocardia* strains was performed with achromopeptidase according to the method reported by Rodríguez-Nava et al. \[[@B10]\]. Colonies were picked off with a loop, and one loopful was suspended in 200 *μ*L of sterile water containing a dozen glass beads and vortexed for 5 minutes. The mixture was then incubated for 15 minutes at 70°C. The suspension supplemented with 3.4 *μ*L of achromopeptidase (Sigma, Steinheim, Germany) at 10 U/mL was incubated at 55°C for 15 minutes. The suspensions were then centrifuged for 5 minutes at 13,000 rpm. The supernatants were stored at −20°C until use.
2.6. Amplification and Sequencing {#sec2.6}
---------------------------------
*Gene rrs.* A 606-bp fragment of the rrs gene was amplified with primers Noc1, 5′-GCTTAACACATGCAAGTCG-3′, and Noc2, 5′-GAATTCCAGTCTCCCCTG-3′, and PCR program and reaction mixture were carried out according the recommendations of Rodríguez-Nava et al. \[[@B10]\].
*Gene hsp65.* A 441-bp fragment of the*hsp65* gene encoding the 65-kDa heat shock protein was amplified with primers described by Telenti et al. (TB11: 5′-ACCAACGATGGTGTGTCCAT-3′ and TB12: 5′-CTTGTCGAACCGCATACCCT-3′) \[[@B16]\]. PCR program and reaction mixture were carried out according to the recommendations of Sánchez-Herrera et al. \[[@B17]\].
*Gene sodA.* A 440-bp fragment of the*sodA* gene was amplified and sequenced with primers SodV1 (5′-CAC CAY WSC AAG CAC CA-3′) and SodV2 (5′-CCT TAG CGT TCT GGT ACT G-3′) where Y = C or T, W = A or T, and S = C or G. The amplification was also done according to the recommendations of Sánchez-Herrera et al. \[[@B17]\].
All resulting PCR products were sequenced and verified (Biofidal, Lyon, France).
The breakpoints for identification based in*sodA* and*hsp65* genes are 99% for each one \[[@B17], [@B18]\]. For the*rrs* gene, a higher breakpoint of 99.6% is used, according to CLSI \[[@B19]\].
2.7. Phylogenetic Analysis {#sec2.7}
--------------------------
The*rrs* gene sequences which we obtained for the 36 clinical isolates of*N. brasiliensis* and the reference strains were aligned manually for the comparative phylogenetic analysis using the Seaview program.
MLSA was performed using*hsp65* and*sodA* sequences of the strains collection. The trimmed aligned sequences were concatenated in the order*sodA-hsp65* to generate an 846 bp sequence using the Seaview program. The Seaview program was also used to infer the evolutionary trees according to the neighbour-joining method \[[@B20]\] and Kimura\'s two-parameter model \[[@B21]\]. The robustness of the tree was performed with a bootstrap of 1000 replicates.
Taking into account the breakpoints for identification at species level of*sodA* and*hsp65* genes individually, the breakpoint for concatenated sequence has been also fixed at 99%.
2.8. DNA Polymorphism of*rrs*,*hsp65,* and*sod*A Genes {#sec2.8}
------------------------------------------------------
The number of haplotypes, the haplotype diversity (Hd), the number of polymorphic sites, and other variables were obtained with DnaSP software \[[@B22]\].
3. Results {#sec3}
==========
3.1. Growth on Culture Medium {#sec3.1}
-----------------------------
The three culture media allowed the growth of clinical strains of*N. brasiliensis*. The Bennett medium showed abundant and rapid growth (48 hours). Middlebrook medium showed strong growth but also it was slightly slower (72 hours). The BCP medium presented interesting results with good rapid growth at 48 hours.*N. brasiliensis* clade tested type strains showed similar patterns to the clinical strains, except that*N. boironii* had a difficult growth on BCP and no growth on Middlebrook; this seems a peculiarity of this species.
3.2. Antimicrobial Susceptibilities {#sec3.2}
-----------------------------------
Eight out of 31 antibiotic molecules tested were active on all the strains\' collection: linezolid, tigecycline, trimethoprim + sulfamethoxazole, moxifloxacin, amikacin, amoxicillin + clavulanic acid, tobramycin, and gentamycin. Regarding the imipenem and pristinamycin molecules, resistance was observed on the majority of clinical isolates of*N. brasiliensis*([Table 1](#tab1){ref-type="table"}).
3.3. Degradation of Substrate {#sec3.3}
-----------------------------
The assimilation test of adenine and uric acid proved negative for all the strains tested of the*N. brasiliensis* clade including clinical and reference ones. The casein degradation test showed that all clinical strains are able to metabolize casein except the clinical strain 12.28. In addition our result showed that some types of strains such as*N. vulneris*,*N. tenerifensis*,*N. boironii,* and*N. iowensis* are also able to degrade casein in the same way as*N. brasiliensis* except*N. altamirensis*. Casein is ultimately a marker that can be used for the phenotypic identification of the*N. brasiliensis* clade and not the*N. brasiliensis* species as it has been believed for many years.
3.4. Phylogeny {#sec3.4}
--------------
Primers Noc1 and Noc2 amplified the expected 606-bp fragment of the*rrs* gene for all the collection strains. Phylogenetic trees ([Figure 1](#fig1){ref-type="fig"}) based upon*rrs* showed homogeneity within clinical strains of*N. brasiliensis*. For this, the*rrs* gene is not relevant to show intraspecies diversity.
In addition, based upon the concatenation of*sodA* and*hsp65* housekeeping genes, the phylogenetic tree generated ([Figure 2](#fig2){ref-type="fig"}) had several distinct genotypes: (i) genotype 1 containing clinical strains, (ii) genotype 2 harboring some clinical strains, and (iii) genotype 3 harboring some clinical strains and*N. brasiliensis* ATCC 19296^T^. For the tropical*N. brasiliensis*HUJEG01 strain, it is observed that it does not belong to any of the 3 genotypes and is positioned alone in the tree between genotypes 1 and 2. This distribution of clinical strains of*N. brasiliensis* in 3 different genotypes shows an intraspecies diversity rather important. To better understand the polymorphism showed by phylogenetic trees, we studied the percentages of the similarities between the sequences. The average percentages of similarities based on the*rrs* gene ([Table 2](#tab2){ref-type="table"}) range from 99.39% to 99.57% between the clinical strains and the 2 reference strains of*N. brasiliensis* (type and tropical strains). According to the CLSI, the similarity percentage needed for identification at species level must be greater than or equal to a threshold of 99.6% \[[@B19]\]. The clinical strains that showed a similarity percentage lower than this threshold for both reference strains of*N. brasiliensis* were anyway considered as belonging to this species because no higher similarity percentage was obtained for any other species. In the same way, the*N. vulneris* type strain was also revealed to be close to clinical strains according to the average of percentage of similarity (98.77%). Between the 2 reference strains of*N. brasiliensis* the percentage of similarity is higher, up to 99.82%. The percentages of similarities based on the concatenation of the*sodA-hsp65* genes ([Table 2](#tab2){ref-type="table"}) decrease and range now from 97.99% to 99.19% between the clinical strains and the 2 reference strains of*N. brasiliensis*. Between the type and reference strains of*N. brasiliensis* the percentage of similarity does not reach 99% this time. The comparison of the 3 genotypes between them (based on the representation of each genotype by 3 clinical strains) by using*sodA-hsp65* genes shows that genotypes 2 and 3 are closer to each other (98.97% of similarity). The average of the percentages of similarity between genotypes 1 and 2 were 97.97%. and 98.28% between genotypes 1 and 3. Finally this value goes up to 98.97% between genotypes 2 and 3. This means that the more distant genotypes between them are 1 and 2 and the closer ones are 2 and 3.
In parallel, an epidemiological study based on the clinical files was carried out, and the data were presented in [Table 1](#tab1){ref-type="table"}. In order to know the link between the genetic diversity and the tropism of the clinical strains, a superposition of data was made between the phylogenetic tree obtained by the concatenation of*sodA* and*hsp65* and the tropism of the clinical strains ([Figure 2](#fig2){ref-type="fig"}). Thus, we can see that in genotypes 2 and 3 we have almost all the clinical strains that have a cutaneous tropism except the 08.188 strain which has a subcutaneous tropism. Regarding genotype 1 it is more heterogeneous with various tropism: (i) pulmonary, (ii) cerebral, and (iii) cutaneous. Regarding the immunocompetence of patients, we have only 4 patients who have immunodepression factors, whose strains are in genotype 1 except the 10.93 strain which is in genotype 3.
3.5. Analysis of*rrs*,*hsp65,* and*sod*A Genes Polymorphism {#sec3.5}
-----------------------------------------------------------
The 36 clinical strains and 2 reference strains of*N. brasiliensis* studied showed (i) for*rrs* gene 16 polymorphic sites sharing 16 haplotypes and showing a Hd of 0.824; (ii) for*hsp65* gene, 27 polymorphic sites and up to 22 different haplotypes with a Hd of 0.927; and, (iii) for*sodA* gene, up to 29 polymorphic sites sharing 14 haplotypes having a Hd of 0.885 ([Table 3](#tab3){ref-type="table"}).
4. Discussion {#sec4}
=============
*Nocardia* spp. are common soil-inhabiting bacteria that frequently infect humans through traumatic injuries or inhalation routes and cause infections, such as actinomycetoma and nocardiosis, respectively.*N. brasiliensis* is the main aetiological agent of actinomycetoma in various countries \[[@B23]\]. The input data used in this study highlight the existence of indigenous cases of cutaneous and subcutaneous (such as actinomycetoma) nocardiosis caused by*N. brasiliensis* in France. Moreover, we can observe that*N. brasiliensis* is also responsible for severe cases of disseminated nocardiosis in immunocompromised patients (pulmonary and cerebral cases).
To determine whether there is an association between clinical tropism of strains and their genetic profile we performed genetic characterization of 36 indigenous cases of*N. brasiliensis* that happened in France.
The three culture media allow the growth of clinical strains of*N. brasiliensis*. However, on Bennett\'s medium more abundant and fast growth (48 hours) was observed. But the downside of this medium is its inaccessibility in the hospital because it is not marketable. Middlebrook medium shows strong growth but also it was slightly slower (72 hours). This medium is very expensive and not accessible to all budgets. However, it is an interesting alternative in isolating*Nocardia* from a complex sample. It is a selective medium of Mycobacteria, which promotes the growth of some*Nocardia* to the detriment of other external bacteria or commensal flora that may be in the biological sample analyzed. The BCP medium, used routinely in hospitals for Gram-negative bacteria, has interesting results with good fast growth (48 hours). It would therefore be advisable to use it as isolation medium for urgent cases, by the speed of growth.
Antibiograms results show resistance of most of the clinical strains to imipenem. This can pose therapeutic problems since it is part of molecules proposed during a phase of a general treatment for nocardiosis \[[@B24]\]. However, all clinical strains of*N. brasiliensis* were sensible to SXT and would be an effective molecule during treatments. The sensibility of*N. brasiliensis* type strain to this antibiotic has already been observed by Gilquin et al. \[[@B9]\].
Our study confirms that all clinical strains of*N. brasiliensis* are capable of degrading casein except 12.28 clinical strain. As shown by Seol et al. the*N. brasiliensis* type strain is able to degrade casein as well \[[@B25]\]. However, the test on the reference strains reveals that*N. vulneris*,*N. tenerifensis*,*N. boironii,* and*N. iowensis* are also capable of degrading casein in the same way as*N. brasiliensis,* as also shown by Gilquin et al. \[[@B9]\]. This type of test is used in some countries without the necessary molecular biology tools to identify*N. brasiliensis*. But, now, they must be aware that with this test we target several species of clade*N. brasiliensis*. So, it is no longer a criterion of identification proper to*N. brasiliensis*.
Phylogenetic tree based on the*rrs* gene sequence of our collection showed a low genetic diversity resulting in low polymorphism sequence. In addition, we can note that*N. vulneris* DSM 45737^T^, identified as a new species by Lasker et al., present a genetic sequence very close to*N. brasiliensis* strains \[[@B26]\] with percentages of similarities on average greater than 98%.
Analysis of the phylogenetic tree ([Figure 2](#fig2){ref-type="fig"}) based on the MLSA by the concatenation of*sodA* and*hsp65* housekeeping genes showed that the isolates are surprisingly classified according to 3 genotypes. These groups were formed upon similarity percentages and existing phylogenetic distances between the sequences of the strains studied. Genotype 1 concerns a well-defined cluster containing 9 clinical strains only which is sustained by a bootstrap of 81%. This genotype hosts all the cases of pulmonary forms (3), the sole cerebral form, and almost all the cases of immunocompromised patients (3 out of 4). Moreover, eight out of twelve strains susceptible to imipenem can be found in this group. There is just one remaining strain in this group not presenting this kind of susceptibility. The reason may be an acquired resistance to this drug due to a previous treatment. This well-defined genotype evokes the possible existence of another species or a strong variability in this case. This may have been caused by environmental pressures in the ecosystem of these isolates which may have resulted in the selection of strains that may have acquired, by mutations or genetic transfer with other microorganisms, new virulence characters different from that of the strain type*N. brasiliensis*. Regarding genotypes 2 and 3, they include clinical strains and*N. brasiliensis* ATCC 19296^T^ type strain. However, the discrimination between these 2 genotypes is less clear than that with genotype 1 because of being in weak bootstrap that is less than 50. The genetic differences do not allow distinguishing them properly and their phenotypic behavior remains similar. Then, it would be interesting to study on another gene capable of generating more divergences, for example,*gyrB* and*rpoB* genes, which have already successfully been used for studying the polymorphism of some other*Nocardia* species \[[@B27], [@B28]\].
Concerning the percentage of similarity between the type and tropical strain of*N. brasiliensis*, it is 99.82% according to the*rrs* and goes down to 98.51% with the concatenation of*sodA* and*hsp65*. The fact of highlighting a greater dissimilarity with the concatenation between the type and tropical*N. brasiliensis* strain shows the advantage of the use of 2 markers like*sodA* and*hsp65* vis-à-vis the*rrs*. The discriminatory power of these two genes may be explained by the presence of more polymorphic sites (*hsp65*: 27;*sodA*: 29) than in the case of*rrs* gene (16) and also by having Hd values higher than that of*rrs* gene (*hsp65*: 0.927;*sodA*: 0.885;*rrs*: 0.824).
It would be interesting to identify the genes involved in the virulence of different genotypes, including those of actinomycetoma. Interesting leads can be considered: (i) as identification of virulence genes expressed using the RNAseq method or (ii) to identify noncoding RNAs \[[@B23]\]. In addition, to genomically distinguish*N. brasiliensis* and*N. vulneris* a specific PCR to*N. vulneris*, using a specific gene of the species, should be developed.
The authors thank Audrey Dubost for her technical support in bioinformatics analysis.
Conflicts of Interest
=====================
The authors declare that they have no conflicts of interest.
{#fig1}
{#fig2}
######
Table of clinical strains including the type of tropism observed in host, their corresponding *sodA/hsp65* genotypes, and drug phenotypes.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Sample date Nature of sampling Patient record Immunosuppressed Genotype \ Tropism Pristinamycin Imipenem Amikacin Trimethoprim + sulfamethoxazole
*sodA/hsp65*
------------- ---------------------------------- ---------------- ------------------ -------------- -------------- --------------- ---------- ---------- ---------------------------------
04/2002 Intraoperative tissue 02.56 No G3 Cutaneous R R S S
12/2003 Pus from cutaneous thigh abscess 04.21 Yes G1 Cutaneous R S S S
2004 Pus from cutaneous abscess 04.101 No G3 Cutaneous R R S S
07/2005 Expectoration then LBA 05.64 Yes G1 Lung R S S S
01/2005 Skin biopsy 05.12 No G1 Cutaneous R S S S
07/2005 Phalanx biopsy 05.63 No G3 Cutaneous R R S S
2005 CSF 05.77 No G1 Brain R R S S
11/2005 \- 3247 No G3 Unknown R S S S
2007 Cutaneous abscess 07.168 No G3 Cutaneous R R S S
10/2008 Wound of forehead 08.178 No G3 Cutaneous R S S S
11/2008 Subcutaneous abscess 08.188 No G3 Subcutaneous R R S S
10/2008 Elbow abscess 2985 No G3 Cutaneous R R S S
03/2008 \- 9044 No G2 Unknown R R S S
03/2009 Pus of leg abscess 09.71 No G2 Cutaneous R S S S
04/2009 Bronchial aspiration 09.106 No G1 Lung R S S S
10/2009 Bronchial aspiration 09.244 Yes G1 Lung R S S S
12/2009 Finger skin 09.280 No G3 Cutaneous R R S S
10/2009 Pus from the lip 10.16 No G3 Cutaneous R R S S
02/2010 Finger abscess 10.35 No G2 Cutaneous R R S S
05/2010 Hand abscess 10.82 No G3 Cutaneous R R S S
05/2010 Toe abscess 10.93 Yes G3 Cutaneous R R S S
09/2010 Leg wound 10.146 No G3 Cutaneous R R S S
11/2010 Pus finger 10.180 No G3 Cutaneous R R S S
11/2010 Leg wound 12786 No G2 Cutaneous R R S S
04/2010 \- 14229 No G2 Unknown R S S S
07/2010 Sepsis hand 45762 No G3 Cutaneous R R S S
2011 Pus 11.44 No G2 Cutaneous R R S S
05/2011 Cutaneous abscess 11.73 No G1 Cutaneous R S S S
05/2011 Hand wound 11.80 No G1 Cutaneous R S S S
08/2011 Cutaneous 11.116 No G2 Cutaneous R R S S
2011 Thigh abscess 11.140 No G2 Cutaneous R R S S
09/2011 Knee wound 11.151 No G3 Cutaneous R R S S
2011 Hand wound 11.172 No G3 Cutaneous R R S S
12/2011 Cutaneous abscess 11.189 No G3 Cutaneous R R S S
2012 \- 12.08 No G1 Unknown R S S S
02/2012 Cutaneous abscess 12.28 No G3 Cutaneous R R S S
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
######
Percentage of similarity expressed in interval and mean for *sodA-hsp65* and *rrs* genes.
*N. brasiliensis* HUJEG01 *N. brasiliensis* ATCC 19296${}^{\overset{\,}{T}}$ *N. vulneris*DSM 45737${}^{\overset{\,}{T}}$
--------------------------------------------------- ------------------------ --------------------------- ---------------------------------------------------- ---------------------------------------------- ---------------- ---------------- ----------------
*N. brasiliensis*ATCC 19296${}^{\overset{\,}{T}}$ Similarity average (%) 98.51 99.82 \- \- \- \-
*N. vulneris*DSM 45737^T^ 98.38 98.73 98.51 98.55 \- \-
*N. altamirensis*DSM 44997^T^ 95.91 98.24 95.41 98.07 95.91 97.47
*N. boironii*DSM 101696^T^ 96.53 98.07 96.28 97.89 96.15 97.29
*N. iowensis* DSM 45197^T^ 94.91 97.89 95.29 97.71 94.54 97.64
*N. tenerifensis* DSM 44704^T^ 96.28 96.67 96.03 96.49 95.91 96.03
Genotype 1 Similarity range (%) (97.77--98.38) \- (97.77--98.26) \- (97.52--98.14) \-
Similarity average (%) 98.01 \- 97.99 \- 97.67 \-
Genotype 2 Similarity range (%) (98.39--98.88) \- (98.88--99.38) \- (98.63--98.88) \-
Similarity average (%) 98.73 \- 99.19 \- 98.80 \-
Genotype 3 Similarity range (%) (98.14--98.88) \- (98.26--99.00) \- (98.01--98.76) \-
Similarity average (%) 98.63 \- 98.76 \- 98.50 \-
All clinical strains Similarity range (%) \- (98.57--100) \- (98.38--99.82) \- (98.01--99.09)
Similarity average (%) \- 99.57 \- 99.39 \- 98.77
######
DNA polymorphism of *rrs*, *hsp65,* and *sodA* genes from clinical *N. brasiliensis* strains isolated in France.
----------------------------------------------------------------------------------------------------------------------------------------------------
*Nocardia* species Genes (bp)^a^ Number of haplotypes\ Number of of polymorphic sites
(Hd, S^2^, SD)^b^
--------------------------------------------------------- ---------------------------- ---------------------------- --------------------------------
*N. brasiliensis* \ *rrs* (569) 16 (0.824, 0.00300, 0.055) 16
(*N* = 38: 36 clinical strains and 2 reference strains)
*hsp65*(401) 22 (0.927, 0.00087, 0.029) 27
*sodA* (406) 14 (0.885, 0.00080, 0.028) 29
----------------------------------------------------------------------------------------------------------------------------------------------------
^a^Resulting fragment size without the primers sequences; ^b^Hd: haplotype (gene) diversity, S^2^: variance of haplotype diversity, and SD: standard deviation of haplotype diversity.
[^1]: Academic Editor: Charles Spencer
| {
"pile_set_name": "PubMed Central"
} |
[^1]: Chun-yang Xu, Sheng-dong Lu. These authors contributed equally to this work
| {
"pile_set_name": "PubMed Central"
} |
Rajendran K, Krishnasamy N, Rangarajan J, Rathinam J, Natarajan M, Ramachandran A. Convalescent plasma transfusion for the treatment of COVID‐19: Systematic review. J Med Virol. 2020;1--9. 10.1002/jmv.25961 31502247
1. INTRODUCTION {#jmv25961-sec-0010}
===============
The recent coronavirus disease 2019 (COVID‐19) epidemic developed into an unprecedented global public health crisis with significant humanitarian consequences. As of 19 April 2020, the World Health Organization has been informed of 2 241 359 confirmed cases of COVID‐19, with 152 551 deaths (6.8%) documented worldwide.[^1^](#jmv25961-bib-0001){ref-type="ref"}
The current treatment of COVID‐19 caused by novel coronavirus SARS‐CoV‐2 has been limited to general supportive care, with provision of critical care as no approved therapies or vaccines are available.[^2^](#jmv25961-bib-0002){ref-type="ref"}
The clinical data for the studies involving COVID‐19 are still scarce and limited to data from China, Spain, Italy, United States of America, Germany, France, The United Kingdom, and other international registries. This will be a problem when predicting treatment outcomes.
Passive immunization therapy has been successfully used to treat infectious diseases back to the 1890s. An individual who is sick with infectious diseases and recovers has blood drawn and screened for particular microorganism neutralizing antibodies. Following identification of those with high titers of neutralizing antibody, convalescent plasma containing these neutralizing antibodies can be administered in individuals with specified clinical disease to reduce symptoms and mortality. Hence, convalescent plasma transfusion (CPT) has been the subject of increasing attention, especially in the wake of large‐scale epidemics.[^3^](#jmv25961-bib-0003){ref-type="ref"} It has recently been suggested by Food and Drug Administration that administration and study of investigational CPT may provide a clinical effect for treatment of COVID‐19 during the public health emergency.[^4^](#jmv25961-bib-0004){ref-type="ref"}
We conducted a systematic review to evaluate available data for the clinical effectiveness of convalescent plasma for the treatment of COVID‐19. This will help to provide clinicians and scientists with an overview of scientific evidence on a potential treatment option and better clinical management of critically ill COVID‐19 patients.
2. METHODS {#jmv25961-sec-0020}
==========
2.1. Protocol and registration {#jmv25961-sec-0030}
------------------------------
This systematic search was carried out in major electronic databases (PubMed, Embase, and Medline) to identify available evidence providing Information on the CPT for treatment of COVID‐19 in accordance with the preferred reporting items for systematic reviews and meta‐analyses guidelines.[^5^](#jmv25961-bib-0005){ref-type="ref"} Due to the urgency of the matter and anticipated long waiting period, we were not able to wait for registration of this systematic review protocol (PROSPERO Submission id number: 179739).
2.2. Eligibility criteria {#jmv25961-sec-0040}
-------------------------
### 2.2.1. Study designs {#jmv25961-sec-0050}
Study designs from the selected publication reported CPT in COVID‐19 patients included clinical trials such as randomized controlled trials, controlled clinical trials, prospective and retrospective comparative cohort studies, case‐control studies; cross‐sectional studies, case series, and case reports.
2.3. Intervention {#jmv25961-sec-0060}
-----------------
We included clinical studies involving assessment of CPT treatment for the COVID‐19 patients.
Study population, timing, and setting:
Published literatures were identified between 1 December 2019 and 19 April 2020 using "convalescent plasma and COVID‐19" as search term without restrictions on the study type of setting.
2.4. Comparators {#jmv25961-sec-0070}
----------------
There were no restrictions on the type of comparator in the studies.
2.5. Outcomes {#jmv25961-sec-0080}
-------------
The outcome of interest was clinical effects, survival benefits, viral load & antibody titer status and adverse events.
2.6. Languages {#jmv25961-sec-0090}
--------------
We included articles without considering any restriction of language to identify potential published studies.
2.7. Publication status {#jmv25961-sec-0100}
-----------------------
We included articles published in scientific journals.
2.8. Information sources {#jmv25961-sec-0110}
------------------------
This systematic search was carried out in major electronic databases (PubMed, Embase, and Medline) to identify available evidence providing information on the CPT for treatment of COVID‐19. In addition, we also searched the reference lists of selected studies.
2.9. Search strategy {#jmv25961-sec-0120}
--------------------
The results of our database searches and records identified from other sources were documented. Removal of duplicates were also done manually and depicted in a PRISMA flow diagram.
2.10. Study selection {#jmv25961-sec-0130}
---------------------
A study screen was done minimum of two authors from the search results spreadsheet, authors independently screened the titles and abstracts of studies using the inclusion criteria. Studies selected at title and abstract levels were further screened with the full text of the article for eligibility to include in our review. The studies exploring preclinical trials such as in vitro trials and studies on animal models and in silico drug screens were excluded.
2.11. Data extraction and data items {#jmv25961-sec-0140}
------------------------------------
A pre‐conceived data extraction sheet was used to extract data from selected eligible studies. Any consensus in case of disagreement was resolved by opinion of a third reviewer. The extracted information included mortality, viral load, viral antibody titers, clinical benefits, and adverse events. Outcomes were extracted in all data forms (eg, dichotomous and continuous) as reported in the included studies. The results of our databases search were documented and described in a PRISMA flow diagram (Figure [1](#jmv25961-fig-0001){ref-type="fig"}).
{#jmv25961-fig-0001}
2.12. Risk of bias in individual studies {#jmv25961-sec-0150}
----------------------------------------
To reduce risk of bias two authors independently assessed the included studies. Overall risk of bias was judged as low risk, unclear risk, and high risk.
3. RESULTS {#jmv25961-sec-0160}
==========
The search identified 110 sources. Following screening of titles and abstracts and removing duplicates, we evaluated eight articles in full text. Among these, we found five relevant articles (one pilot study, one preliminary communication, one novel report, one case report, one descriptive study).[^6^](#jmv25961-bib-0006){ref-type="ref"}, [^7^](#jmv25961-bib-0007){ref-type="ref"}, [^8^](#jmv25961-bib-0008){ref-type="ref"}, [^9^](#jmv25961-bib-0009){ref-type="ref"}, [^10^](#jmv25961-bib-0010){ref-type="ref"} Extracted details for five studies are presented in Table [1](#jmv25961-tbl-0001){ref-type="table"}, including the country of study, number of patients, dosage of CPT, mortality, length of hospital stay during transfusion, critical care interventions, clinical outcome, viral load, and adverse events. The five studies include a total of 27 patients who received CPT therapies for COVID‐19.
######
The efficacy and safety of convalescent plasma transfusion (CPT) in patients with COVID‐19
<table><thead><tr class="header"><th>Author</th><th>Country</th><th>Study period</th><th>Study population</th><th>CPT dosage</th><th>Antiviral (antimicrobial drugs)</th><th>Administrated day</th><th>Status during CPT</th><th>Outcome</th><th>Viral load</th><th>Severe adverse events & treatment complications</th></tr></thead><tbody><tr class="odd"><td>Duan et al<a href="#jmv25961-bib-0006" data-ref-type="ref"><sup>6</sup></a></td><td>China</td><td>23 January 2020 to 19 February 2020</td><td>10, 6 M:4 F, Age (x̃‐52.5 y), Cardiovascular and/or cerebrovascular diseases and HTN (n = 4)</td><td>200 mL within 4 h, antibody titer >1:640</td><td><p>arbidol or/and remdesivir/ribavirin/peramivir (n = 9)</p><p>ribavirin (n = 1)</p><p>Antibacterial/antifungal for coninfecion (n = 8)</p></td><td>Onset to CPT (x̃ ‐16.5 d)</td><td>All at ICU, Mechanical ventilation (n = 3), HFNO (n = 3), Conventional LFNO (n = 2)</td><td><p>Clinical symptoms, paraclinical improved,</p><p>Increase of oxyhemoglobin saturation within 3 d</p><p>CP well tolerated, increase/maintain the neutralizing antibodies,</p><p>Varying degrees of absorption of lung lesions within 7 d</p></td><td>Viral load undetectable (n = 7), Neutralizing antibody increased rapidly up to 1:640 (n = 5), maintained at a high level (1:640) (n = 4)</td><td>No severe adverse effects, Evanescent facial red spot (n = 1)</td></tr><tr class="even"><td>Chenguang Shen et al<a href="#jmv25961-bib-0007" data-ref-type="ref"><sup>7</sup></a></td><td>China</td><td>20 January 2020 to 25 March 2020</td><td>5, Age (range, 36‐73 y), 3M:2F, HTN; mitral insufficiency (n=1)</td><td>400 mL of CP in 2 doses on the same day, antibody titer >1:1000</td><td>interferon alfa‐1b + Lopinavir/ritonavir (n = 4) + favipiravir (n = 1), arbidol + darunavir + Lopinavir/ritonavir (n=1)</td><td>After admission between 10 and 22 d</td><td>All 5 critical severe ARDS on mechanical ventilation, ECMO (n = 1)</td><td>Temp normalized within 3 d (n = 4), SOFA score decreased, and PAO2/FIO2 increased within 12 d (range, 172‐276 before and 284‐366 after), Neutralizing antibody titers increased (range, 40‐60 before and 80‐320 on 7th d), ARDS resolved (n = 4) at 12 d, Weaned from mechanical ventilation (n = 3) within 2 wk</td><td>Decreased and became negative within 12 d</td><td>No severe adverse effects</td></tr><tr class="odd"><td>Bin Zhang et al<a href="#jmv25961-bib-0008" data-ref-type="ref"><sup>8</sup></a></td><td>China</td><td>16 February 2020 to 15 March 2020</td><td>69 y/F, HTN</td><td>900 mL in 3 doses</td><td>arbidol, lopinavir‐ritonavir, interferon alpha</td><td>After admission 19th d</td><td>Critically ill invasive mechanical ventilation</td><td>Extubated and non‐invasion ventilation was given on 34th d, Chest CT persistent absorption of consolidation, discharged on 44th d</td><td>Decreased 55 × 10<sup>5</sup> copies/mL (20th d) ‐ 3.9 × 10<sup>4</sup> copies/mL (30th d) ‐ 180 copies/mL (36th d). Negative (40th, 42th d)</td><td>No severe adverse effects</td></tr><tr class="even"><td></td><td></td><td></td><td>55 y/M, COPD</td><td>200 mL</td><td>arbidol, lopinavir‐ritonavir, interferon alpha‐2b</td><td>After admission 12th d</td><td>Critically ill ARDS invasive mechanical ventilation</td><td>pO2 increased to 97 mm Hg with OI of 198 mm Hg in 1 d, All drugs discontinued except methylprednisolone, Chest images absorption of interstitial pneumonia (13th d‐17th d), Discharged on (19th d)</td><td>Negative (18th d)</td><td>No adverse reactions</td></tr><tr class="odd"><td></td><td></td><td></td><td>73 y/M, HTN & chronic renal f‐ure</td><td>2400 mL in 8 doses</td><td>arbidol, lopinavir‐ritonavir, oseltamivir, ribavirin, interferon alpha‐2b</td><td>After admission 15th d</td><td>Critically ill Acute respiratory failure invasive mechanical ventilation in V‐V ECMO</td><td>Positive anti‐SARS‐CoV‐2 IgG (26th d). Chest x‐rays absorbed infiltrative lesions but pneumothorax, Serum IgM level decreased to normal range (45th d, 46th d), Transferred to unfenced ICU for underlying diseases, multiple organ failure (50th d)</td><td>Negative (45th d, 46th d)</td><td>No adverse reactions</td></tr><tr class="even"><td></td><td></td><td></td><td>31 y/F, pregnant (35 wk & 2 d)</td><td>300 mL</td><td>lopinavir‐ritonavir and ribavirin, Imipenem, vancomycin for coinfection</td><td>After admission 19th d</td><td>Critically ill ARDS, invasive mechanical ventilation in V‐V ECMO</td><td>Removed CRRT, ECMO (27th d), anti‐SARS‐CoV‐2 IgM changed from positive to weakly positive to negative, anti‐SARS‐CoV‐2 IgG was persistently positive (35th d 37th d), Chest CT showed near‐complete absorption of opacities, Trachea cannula removed, nasal oxygen given (40th d), Discharged (46th d)</td><td>Negative (40th d, 43th d)</td><td>No adverse reactions</td></tr><tr class="odd"><td>Jin Young Ahn et al<a href="#jmv25961-bib-0009" data-ref-type="ref"><sup>9</sup></a></td><td>South Korea</td><td>22 February 2020 to 6 March 2020</td><td>71 y/M</td><td>500 mL in 2 doses at 12 h interval</td><td>hydroxychloroquine, lopinavir/ritonavir</td><td>After admission 10th d</td><td>Severe ARDS, mechanical ventilation</td><td>Weaned from the mechanical ventilator, underwent a tracheostomy</td><td>Ct changed 24.98 (10th d) ‐ 33.96 (20th d), Negative (after 26th d)</td><td>No adverse reaction</td></tr><tr class="even"><td></td><td></td><td></td><td>67 y/F, HTN</td><td></td><td></td><td>After admission 6th d</td><td></td><td>Extubated and discharged on 24th d</td><td>Negative (after 20th d). Ct changed 20.51 (5th d) ‐36.33 (9th d)</td><td></td></tr><tr class="odd"><td>Mingxiang Ye et al<a href="#jmv25961-bib-0010" data-ref-type="ref"><sup>10</sup></a></td><td>China</td><td>11 February 2020 to 18 March 2020</td><td>69/M</td><td>600 mL in 3 doses</td><td>arbidol, levofloxacin</td><td>After symptom 33th d</td><td>Myalgia, Chest CT‐patchy areas of GGOs</td><td>Symptoms improved, GGOs resolved 37th d, Cured and ready to discharge.</td><td>Negative</td><td>No adverse reaction</td></tr><tr class="even"><td></td><td></td><td></td><td>75/F</td><td>400 mL in 2 doses</td><td>arbidol</td><td></td><td>Fatigue, shortness of breath, oxygen therapy through nasal catheter, respiratory distress, Multiple consolidation</td><td>Symptoms improved, alleviation of respiratory distress, two‐fold increase in IgM and IgG titers, consolidation gradually reduced, turned into scattered GGOs, Cured and under further clinical monitoring</td><td>Negative</td><td></td></tr><tr class="odd"><td></td><td></td><td></td><td>56/M, Bronchitis</td><td>600 mL in 3 doses</td><td></td><td></td><td>Fever, nonproductive cough, shortness of breath, Chest CT‐Multiple GGOs, reticular opacities, and fibrosis streak,</td><td>Symptoms improved, complete resolution consolidation,gradually resolution of GGOs, IgM and IgG titers increased Discharged</td><td>Not mentioned</td><td></td></tr><tr class="even"><td></td><td></td><td></td><td>63/F Sjogren syndrome</td><td>200 mL</td><td></td><td>After symptom 40th d</td><td>Fever, cough, shortness of breath, decreased exercise tolerance, Chest CT ‐Multiple GGOs with consolidation and fibrosis streak</td><td>Symptoms improved, GGOs tended to reduce, anti‐SARS‐CoV‐2 IgM and IgG, Discharged 46th d</td><td>Negative 41th d</td><td></td></tr><tr class="odd"><td></td><td></td><td></td><td>28/F</td><td>200 mL</td><td></td><td>After symptom 33th d</td><td>Fatigue and myalgia, other symptoms</td><td>Discharged 39th d</td><td>Negative</td><td></td></tr><tr class="even"><td></td><td></td><td></td><td>57/M</td><td>200 mL</td><td></td><td>After symptom 50th d</td><td>Fever, cough, shortness of breath and myalgia, Chest CT‐ Extensive bilateral GGOs, respiratory distress</td><td>Symptoms improved, GGOs resolved, discharged 54th d</td><td></td><td></td></tr></tbody></table>
Abbreviations: ARDS, acute respiratory distress syndrome; COPD, chronic obstructive pulmonary disease; COVID‐19, coronavirus disease 2019; CP, convalescent plasma; CT, computed tomography; Ct, computed tomography; ECMO, extracorporeal membrane oxygenation; GGOs, ground‐glass opacity; HFNO, High‐flow nasal oxygen therapy; HFNC, high‐flow nasal cannula oxygenation; HTN, hypertension; ICU, Intensive care unit; IgG, immunoglobulin G; IgM, immunoglobulin M; LFNO, low‐flow nasal cannula oxygenation; SARS‐CoV‐2, severe acute respiratory syndrome coronavirus 2; SOFA, sequential organ failure assessment.
John Wiley & Sons, Ltd.
This article is being made freely available through PubMed Central as part of the COVID-19 public health emergency response. It can be used for unrestricted research re-use and analysis in any form or by any means with acknowledgement of the original source, for the duration of the public health emergency.
All studies but one (South Korea) were conducted in China. In five studies, the male patients (n = 15) were larger in number than the female patients (n = 12). The age of the patients across the different studies varied from 28 to 75. Comorbidity was observed in some patients who were given CPT including COPD/Bronchitis (n = 2), Cardiovascular and cerebrovascular diseases (n = 1), hypertension (n = 7). Among hypertensive patients, one had mitral insufficiency, another one had chronic renal failure. In addition, one 63‐year‐old female patient presented with Sjogren syndrome. Another 31 years aged female COVID‐19 patient was pregnant with a gestation period of 35 weeks and 2 days.
4. DISCUSSION {#jmv25961-sec-0170}
=============
CPT has a very long history of use in the treatment of infectious disease. Its use has been well documented during the outbreak of many diseases at various periods, including spanish Influenza A (H1N1) infections in 1915 to 1917,[^11^](#jmv25961-bib-0011){ref-type="ref"} severe acute respiratory syndrome (SARS) in 2003,[^12^](#jmv25961-bib-0012){ref-type="ref"} pandemic 2009 influenza A (H1N1),[^13^](#jmv25961-bib-0013){ref-type="ref"} avian influenza A (H5N1),[^14^](#jmv25961-bib-0014){ref-type="ref"} several hemorrhagic fevers such as Ebola,[^15^](#jmv25961-bib-0015){ref-type="ref"} and other viral infections. In addition, studies show convalescent plasma antibodies that can limit the virus reproduction in the acute phase of infection and help clear the virus, which is beneficial to the rapid recovery of the disease.[^16^](#jmv25961-bib-0016){ref-type="ref"}
Previous reviews have stated that the CPT may be considered for critically sick COVID‐19 patients based on the earlier reported studies.[^17^](#jmv25961-bib-0017){ref-type="ref"}, [^18^](#jmv25961-bib-0018){ref-type="ref"} In this systematic review of CPT to the COVID‐19 patients, we identified and critically evaluated five studies that described about 27 patients. All studies reported good outcomes after CPT performance, but all were considered to have risk of bias owing to a combination of non‐randomized evaluations, confounding, predictor description and poor methodological conduct for participant selection, dosage of CPT, and duration of therapy. This heterogeneity did not permit us to perform a meta‐analysis. However, the important strength of this study is a comprehensive search of published clinical study data abstraction. Our review is the first to summarize all such literature in humans with COVID‐19.
4.1. CPT dosage {#jmv25961-sec-0180}
---------------
The doses of CPT used as described by the different studies is varied. A Chinese pilot study showed a minimal use of a single dose of 200 mL convalescent plasma with neutralizing antibody titers \>1:640. Another study by Bin Zhang et al[^8^](#jmv25961-bib-0008){ref-type="ref"} reported a maximum of 2400 mL of convalescent plasma administered to a 73 years old male patient. Due to variability of CPT doses in reports, the optimal dose of CPT for COVID‐19 could not be determined. All 27 survivors received CPT between Day 6 and Day 50 after the onset of symptoms or admission to hospitals.
4.2. Antiviral, antibacterial/antifungal medications addition to CPT {#jmv25961-sec-0190}
--------------------------------------------------------------------
All 27 COVID‐19 patients described in these five studies received more than one antiviral drug including CPT, in addition, 10 patients received antibacterial/antifungal drugs for coinfection.
4.3. ICU admission, mechanical ventilation, length of stay {#jmv25961-sec-0200}
----------------------------------------------------------
Most of the patients are considered critically ill who received ICU admission (n = 21) and most of the patients received mechanical ventilation during the CPT (n = 14). However, six patients received nasal cannula oxygenation in which three received HFNO and two received conventional LFNO. Acute respiratory distress syndrome (ARDS) were reported in 17 patients in which 7 received extracorporeal membrane oxygenation during CPT. The length of stay was not specified but most studies revealed data of discharge from hospital (n = 15).
4.4. Viral load and antibody titer levels after CPT {#jmv25961-sec-0210}
---------------------------------------------------
All five studies found that CPT significantly reduces the viral load and increase the level of neutralizing antibody over time. Viral loads also decreased and became negative between day 1 and 30 days after the CPT. Chenguang Shen et al[^7^](#jmv25961-bib-0007){ref-type="ref"} described that IgG titers of the treated patients increased upto 145 800 and the IgM titers also increased upto 145 800 after CPT.
4.5. Clinical benefits {#jmv25961-sec-0220}
----------------------
After receiving convalescent plasma transfusion, almost all the patients showed improvements of symptoms including their body temperature normalized, varying degrees of absorption of lung lesions, ARDS resolved, weaned from ventilation within 1 day to maximum of 35 days post transfusion.
4.6. Survival {#jmv25961-sec-0230}
-------------
All studies reported unanimously positive findings of zero mortality after patients received CPT in varying doses. However, it was not clearly determined that whether the high percentage of survival was due to the treatment of patients with multiple other agents (including antiviral medications) or CPT treatment or a combinatorial/synergistic effect of both. Bin Zhang et al[^8^](#jmv25961-bib-0008){ref-type="ref"} referred that one patient (73/Male) was transferred to unfenced ICU for further treatment due to underlying diseases and multiple organ failure.
4.7. Severe adverse events and treatment complications {#jmv25961-sec-0240}
------------------------------------------------------
CPT was well tolerated by the participants in all studies. No fatality occurred in SARS CoV2‐infected individuals administered with convalescent plasma. Duan et al[^6^](#jmv25961-bib-0006){ref-type="ref"} mentioned a minor side effect of evanescent facial red spot in one patient administered with convalescent plasma but it was very minimal with no adverse events.
4.8. Limitations {#jmv25961-sec-0250}
----------------
A lack of high‐quality RCT studies and relevant literature paucity limited our analyses. All the reported studies were predominately case reports or series, had no proper control groups, and had a moderate to high risk of bias.
5. CONCLUSION {#jmv25961-sec-0260}
=============
There is a compelling need to control the greatest global health crisis by COVID‐19 outbreak. Currently, there is no reliable therapeutic options for critically ill COVID‐19 contracted patients. Based on the consolidated clinical data derived from five independent studies of 27 patients suggests, in addition to antiviral/antimicrobial drugs, CPT could be an effective therapeutic option with promising evidence on safety, improvement of clinical symptoms, and reduced mortality. We recognize that a definitive conclusion cannot be drawn on optimal doses and treatment time point for the CPT to COVID‐19, a large multicenter clinical trials are urgently needed to tackle this pandemic.
CONFLICT OF INTERESTS {#jmv25961-sec-0280}
=====================
The authors declare that there are no conflict of interests.
AUTHOR CONTRIBUTIONS {#jmv25961-sec-0290}
====================
KR conceived the content, retrieved the data, wrote the manuscript, and approved the final version. KN retrieved the data and approved the final version. JaR, JeR retrieved the data, wrote the manuscript. MN, AR helped in data extraction, revised the manuscript critically, and approved the final version.
The authors would like to thank the support from the Department of Health Research (DHR), Ministry of Health & Family Welfare, Government of India.
| {
"pile_set_name": "PubMed Central"
} |
All sequence files are available from the GenBank database (accession numbers KT804429-42).
Introduction {#sec001}
============
*Leptospira* spp. are helical-shaped bacteria and form a particular group of causative agents for the zoonotic disease Leptospirosis. *Leptospira* spp. are transmitted through infected urine of small mammals or contaminated water via the direct contact to skin lesions or conjunctivae \[[@pntd.0004501.ref001]\]. Small mammals are described as the most important maintenance reservoirs in nature and thus as an essential vector for several pathogenic *Leptospira* spp. \[[@pntd.0004501.ref002], [@pntd.0004501.ref003], [@pntd.0004501.ref004], [@pntd.0004501.ref005]\]. Leptospirosis is considered the most widespread zoonotic disease worldwide, which is of emerging concern \[[@pntd.0004501.ref006]\]. In the past, Leptospirosis was described to be a disease of occupational risk for harvesters, miners, veterinarians and rodent control workers in Europe \[[@pntd.0004501.ref002], [@pntd.0004501.ref007]\]. Nowadays, it is increasingly linked to recreational outdoor activities, such as water sports and adventure travels \[[@pntd.0004501.ref005], [@pntd.0004501.ref008]\]. However, partially due to the broad variety of clinical symptoms, which are nonspecific, the awareness for this disease is not yet present especially in temperate regions \[[@pntd.0004501.ref005], [@pntd.0004501.ref009]\]. The estimated incidence of clinical cases per year is 0.2 / 100,000 in Germany \[[@pntd.0004501.ref010]\]. Severe cases associated with rats have also been reported \[[@pntd.0004501.ref008], [@pntd.0004501.ref011]\]. Recently, human cases, which were linked to contaminated water or soil, occurred in Austria \[[@pntd.0004501.ref012], [@pntd.0004501.ref013], [@pntd.0004501.ref014]\]. Furthermore leptospirosis outbreaks were reported among triathletes and strawberry harvesters in Germany \[[@pntd.0004501.ref015], [@pntd.0004501.ref016]\]. The clinical severity of *Leptospira* spp. infection depends on the virulence of the infecting *Leptospira* serovar as well as on the health status of the patient \[[@pntd.0004501.ref003]\]. The taxonomy of *Leptospira* spp. is complex. To date, ten different pathogenic *Leptospira* species with more than 300 serovars, grouped in 20 serogroups are known \[[@pntd.0004501.ref017]\]. The term serogroup is of taxonomic importance and defines groups with antigenetically related serovars. However identical serovars may belong to different *Leptospira* species \[[@pntd.0004501.ref002]\]. Duplex PCR \[[@pntd.0004501.ref018]\] and detailed sequence typing are used for the characterisation of *Leptospira* spp. strains and genotypes while the microscopic agglutination test (MAT) which is important for the categorisation of serovars, is still the gold standard in routine diagnostics \[[@pntd.0004501.ref019]\]. Most commonly, human clinical cases in Europe are caused by *L*. *interrogans* and/or *Leptospira* spp. serovar Grippotyphosa \[[@pntd.0004501.ref012], [@pntd.0004501.ref013], [@pntd.0004501.ref016], [@pntd.0004501.ref020]\]. A recent study from Poland reported also antibody titres in humans against the serovars Australis, Autumnalis, Hebdomadis, Hardjo, Sejroe, Zanoni, Bataviae, Bratislava, Canicola and Grippotyphosa, belonging to 3 species, *L*. *interrogans*, *L*. *borgpetersenii* and *L*. *kirschneri*\[[@pntd.0004501.ref021]\]. Studies from Germany and France reported high prevalences for *Leptospira* spp. in small mammals which are likely responsible for simultaneous human leptospirosis cases \[[@pntd.0004501.ref016], [@pntd.0004501.ref020]\]. So far there are only a few studies reporting moderate to high prevalences in small mammals, beavers (*Castor fiber*) and wild boars (*Sus scrofa*) from Germany \[[@pntd.0004501.ref005], [@pntd.0004501.ref008], [@pntd.0004501.ref022]\]. Little is known about the prevalence and the geographic distribution of pathogenic *Leptospira* spp. in rodent maintenance hosts in Germany. Possible host-pathogen associations were not further determined thus far.
Therefore, this study's objectives were:
1. Detection of prevalence rates for different pathogenic *Leptospira* spp. in captured small mammals from three selected sites in Germany;
2. Comparison of the detected *Leptospira* spp. and their sequence types (ST) in relation to captured small mammal species and the various study sites.
Materials and Methods {#sec002}
=====================
Study sites {#sec003}
-----------
### Bavarian Site 1 "Angelberger Forst", Bavaria (B1) {#sec004}
The site B1, located near Tussenhausen, Bavaria, is a large mixed forest (641 ha). The anthropogenic influence is low thus interaction between wild and domestic animals and humans is limited \[[@pntd.0004501.ref023]\]. Details of this study area have been described before \[[@pntd.0004501.ref024], [@pntd.0004501.ref025]\].
### Bavarian Site 2 "Bavarian Forest National Park" (B2) {#sec005}
The second study site B2 was divided in three transects in the "Bavarian Forest National Park" (242,000 ha) from 629 to 1420 m a. s. l. It is also situated in Bavaria and is part of the Bohemian Massif. This National Park and the adjacent Czech Sumava National Park form one of the most homogenous and extensively forested landscapes in Central Europe. The study site is located in a forested montane area and is often frequented by visitors for recreational activities. Further characteristics, GPS coordinates of trapping sites along the transects as well as ecologic properties have been described in detail before \[[@pntd.0004501.ref026]\].
### Renaturalised area in the city centre of Leipzig, Saxony (S) {#sec006}
The third site (S) is divided in five areas located around the city centre of Leipzig formerly consecutively named from E to I \[[@pntd.0004501.ref027]\]. The study site belongs to a renaturalised location ([www.neuseenland.de](http://www.neuseenland.de/)), which was created out of a former brown coal mining area. Today, the surroundings of this site are the largest recreational area near Leipzig thus many visitors frequent this site. Detailed descriptions about the areas around Leipzig have been published elsewhere \[[@pntd.0004501.ref027]\].
Sampling of small mammals {#sec007}
-------------------------
Small mammals were trapped with Sherman live animal traps (H. B. Sherman Traps, Inc., Tallahassee, Fla., U.S.A.) in 2012 at site B1, in 2010 at site B2 and from 2010 to 2012 at site S. Traps, baited with apple slices, were placed for at least two consecutive nights per month and site and were checked twice a day. For site B1, 50 traps were set up between July and October in 2012. At site B2 in plot sizes of 18 x 18 m, 16 traps were laid out in a grid. The traps were checked twice on two successive days once a month, from May to October 2010. At site S, small mammals were captured with at least 20 traps per subdivided area per month in August and October in 2010 and from March to June in 2011 (site E, H, I) and further in November 2011 and from March to October in 2012 (site E, F, G, H, I). Collected animals were euthanized in accordance with the German Animal Protection Act and stored at −80°C. Detailed trapping procedures were published elsewhere \[[@pntd.0004501.ref026], [@pntd.0004501.ref027], [@pntd.0004501.ref028]\]. Necropsy was carried out with collection of biometric data of all small mammals and kidneys were collected. All small mammals were morphologically identified with taxonomic keys \[[@pntd.0004501.ref029]\]. Further, a conventional PCR targeting the partial mitochondrial *cytochrome b* gene \[[@pntd.0004501.ref030]\] yielding an amplicon of 354 bp was performed with 37 small mammal DNA samples from site B1 (17.1%) and 36 DNA samples from site S (6.7%) including all bycaught small mammal species, which were not rodents, in order to verify morphological identification and to verify successful DNA extraction.
DNA extraction {#sec008}
--------------
Depending on the initial kidney size of the small mammals, kidney samples weighed 0.01--0.05g and were homogenized either by cutting the samples into small pieces each with a sterile scalpel (site B1) or by the use of the Precellys 24 Tissue Homogenizer (Bertin Technologies, Montigny-le-Bretonneux, France) for which 600--800 μl phosphate buffered saline (PBS, pH = 7.2) and 0.6 g of sterile ceramic beads, (PeqLab Biotechnologie GmbH, Erlangen, Germany) sized 1.4 mm, were added to the samples in advance (sites S and B2). DNA was extracted either with the Maxwell 16 LEV Blood DNA Kit (Promega GmbH, Mannheim, Germany) and the corresponding Maxwell 16 System (site B1) or manually with the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) as recommended by the manufacturers (sites S and B2) after addition of 300 μl lysis buffer and 30 μl proteinase K to each sample and incubation overnight at 56°C in a thermomixer (Eppendorf, Hamburg, Germany). For all samples, quantity and quality of the extracted DNA samples were determined with a spectrophotometer (NanoDrop 2000c respectively NanoDrop ND-1000, Peqlab Biotechnologie GmbH).
PCR methods {#sec009}
-----------
### Real-Time PCR {#sec010}
For the initial screening of all samples, a real-time PCR targeting a partial sequence (242 bp) of the 32-kDa leptospiral major outer membrane lipoprotein *lipl 32* gene was performed as described \[[@pntd.0004501.ref022], [@pntd.0004501.ref031]\] by the use of the Mx3000P QPCR System (Agilent Technologies, Santa Clara, C.A., U.S.A.). Samples with a CT-value below 40 were regarded as positive.
### Duplex PCR {#sec011}
In order to distinguish *L*. *kirschneri* from other pathogenic *Leptospira* spp., samples tested positive by real-time PCR were further determined by a duplex PCR \[[@pntd.0004501.ref018]\]. The duplex PCR was targeting a flagellin-encoding *flaB* gene fragment (563 bp) which exclusively amplifies in *L*. *kirschneri* as well as a preprotein translocase-encoding *secY* gene fragment (285 bp), which also amplifies in several other pathogenic *Leptospira* spp. such as *L*. *interrogans*, *L*. *weilii*, *L*. *noguchii*, *L*. *borgpetersenii*, *L*. *santarosai* and *L*. *meyeri* \[[@pntd.0004501.ref005], [@pntd.0004501.ref018]\].
### Conventional PCR {#sec012}
Samples which were positive only for *secY* were further determined by conventional PCR targeting a gene fragment (504 bp) which is encoding for the DNA gyrase subunit B (*gyr B*) \[[@pntd.0004501.ref032]\]. Eight pathogenic *Leptospira* species have been described to be detected by this PCR method: *L*. *interrogans*, *L*. *borgpetersenii*, *L*. *weilii*, *L*. *santarosai*, *L*. *alexanderi*, *L*. genomospecies 1, now known as *L*. *alstonii*, *L*. *noguchii* and *L*. *kirschneri* \[[@pntd.0004501.ref032], [@pntd.0004501.ref033]\]. PCR products were purified with the NucleoSpin and PCR clean-up kit (MACHEREY-NAGEL, Düren, Germany) according to the manufacturer's recommendations and subsequently commercially sequenced (Interdisziplinäres Zentrum für Klinische Forschung, Leipzig, Germany) with forward and reverse primers used for PCR amplification. Consensus sequences without primer sequences were analysed and aligned with the MegAlign Pro Software (DNASTAR, Inc., Madison, W.I., U.S.A.), compared to available sequences in the GenBank with BLASTn (National Center for Biotechnology Information, Bethesda MD, USA) and deposited in GenBank under following Acc. No.: KT804429-42.
### Multi Locus Sequence Typing (MLST) {#sec013}
A set of seven primer pairs was used which amplify different housekeeping gene loci: *glmU*, *pntA*, *sucA*, *fadD*, *tpiA*, *pfkB* and *mreA* \[[@pntd.0004501.ref017], [@pntd.0004501.ref034]\]. The PCR method was modified for the *fadD* and *glmU* loci as described before \[[@pntd.0004501.ref022]\]. MLST succeeded only for 14 samples tested positive by duplex PCR. Products of all seven primer sets were purified, sequenced and analysed as described above. To receive the particular *Leptospira* sequence type, sequences were trimmed accordingly, and further analysed and compared to sequences at <http://leptospira.mlst.net/>.
### Statistical analysis {#sec014}
Confidence intervals (95%CI) for prevalences in small mammals were determined by the Clopper and Pearson method using the Graph Pad Software Prism (Graph Pad Software Inc., San Diego, Ca., USA). Pearson's chi-squared test was used with a type I error α of 0.05 to test the independence of compared prevalences. Fisher's exact test was used for small sample sizes tested (n\<30). The Bonferroni correction was used to control Pearson's chi-squared tests computed with multiple values.
### Ethics statement {#sec015}
The permission of small mammal trapping was granted by the "Landesdirektion Sachsen" at site S (permission number: AZ 36.11--36.45.12/4/12-001) and by the "Regierung von Schwaben" at site B1 (permission number: 55.1-8646-2/30). Permission for small mammal trapping was not needed at site B2 as this study site was investigated for federal concerns by the Bundeswehr. Collected animals were euthanized in accordance with the German Animal Protection Act.
Results {#sec016}
=======
Trapping of small mammals {#sec017}
-------------------------
Altogether, 1213 small mammals of eleven different species (737 *M*. *glareolus*, 12 *Microtus agrestis*, 431 either *A*. *flavicollis or A*. *sylvaticus*, one *Sorex coronatus*, seven *Sorex araneus*, four *Crocidura russula*, three *Arvicola terrestris*, two *Talpa europaea*, two *Mustela nivalis*, seven *Apodemus agrarius*, seven *Microtus arvalis*) were captured ([Table 1](#pntd.0004501.t001){ref-type="table"}). Altogether, 216 small mammals were captured at site B1 (three species), 456 at site B2 (four species) and 541 at site S (ten species). DNA was extracted from one kidney of all 1213 animals.
10.1371/journal.pntd.0004501.t001
###### Number of collected small mammal species from each study site with the number of small mammals positive for *Leptospira* spp. as detected by real-time PCR in brackets.
{#pntd.0004501.t001g}
Small mammal family Small mammal species Study site Total
--------------------- ---------------------------- -------------- -------------- -------------- ---------------
**Muridae** ***Apodemus flavicollis*** 83 (2) \- 164 (13) **247 (15)**
***Apodemus* spp.\*** \- 184 (12) \- **184 (12)**
***Apodemus agrarius*** \- \- 7 (3) **7 (3)**
***Arvicola terrestris*** \- \- 3 **3**
**Cricetidae**
***Microtus arvalis*** \- \- 7 (3) **7 (3)**
***Microtus agrestis*** \- 11 1 **12**
***Myodes glareolus*** 132 (23) 257 (3) 348 (13) **737 (39)**
**Soricidae** ***Crocidura russula*** \- \- 4 **4**
***Sorex* spp.** 1 4 3 **8**
**Others** ***Talpa europaea*** \- \- 2 **2**
***Mustela nivalis*** \- \- 2 (1) **2 (1)**
**Total** **216 (25)** **456 (15)** **541 (33)** **1213 (73)**
PCR analysis for *Leptospira* spp. {#sec018}
----------------------------------
From altogether 1213 small mammals, 73 tested positive by real-time PCR (5.9%; 95%CI: 4.7--7.4) ([Table 2](#pntd.0004501.t002){ref-type="table"}). Regarding the different sites, 11.6% (95%CI: 8--16.7) of the small mammals from site B1 (n: 25/216), 3.3% (95%CI: 1.9--5.44) from site B2 (n: 15/456) and 6.1% (95%CI: 4.32--8.49) from site S were positive (n: 32/542). Interestingly along the altitude gradient of B2 at two sampling sites at low altitude (both small beech forests at 379 m and 412 m a.s.l., [Fig 1](#pntd.0004501.g001){ref-type="fig"}) most small mammals of B2 were PCR positive (n: 12). However, *Leptospira* spp. were found up to an altitude of 1.298 m a.s.l. (see [Fig 1](#pntd.0004501.g001){ref-type="fig"}). From the altogether 73 PCR-positive samples, 67 could be further determined by duplex PCR. The other six samples did not yield any of the two amplicons most likely due to the high CT value achieved by real-time PCR (\>39). Fifty-four (80.3%; 95%CI: 69--88.3) of these 67 samples were identified as *L*. *kirschneri*. Four of the other 13 (19.7%; 95%CI: 11.8--31) samples were further determined as *L*. *borgpetersenii*, and 9 as *L*. *interrogans* by conventional PCR of the partial *gyr*B gene and amplicon sequencing.
{#pntd.0004501.g001}
10.1371/journal.pntd.0004501.t002
###### Number of the Leptospira species detected in the small mammals per study site[\*](#t002fn001){ref-type="table-fn"}[^\#^](#t002fn002){ref-type="table-fn"}.
{#pntd.0004501.t002g}
Study sites *Leptospira* species Small mammal species[\*](#t002fn001){ref-type="table-fn"}[^\#^](#t002fn002){ref-type="table-fn"}
------------- --------------------------------- -------------------------------------------------------------------------------------------------- -------- -------- ------- ------- -------- --------
**B1** ***Leptospira interrogans*** \- \- \- \- \- \- **-**
***Leptospira borgpetersenii*** \- \- \- 1 \- \- **1**
***Leptospira kirschneri*** 21 \- \- 1 \- \- **22**
**B2** ***Leptospira interrogans*** 2 \- 7 \- \- \- **9**
***Leptospira borgpetersenii*** \- \- 1 \- \- \- **1**
***Leptospira kirschneri*** 1 \- 3 \- \- \- **4**
**S** ***Leptospira interrogans*** \- \- \- \- \- **-**
***Leptospira borgpetersenii*** \- \- \- 2 \- \- **2**
***Leptospira kirschneri*** 12 3 \- 10 1 2 **28**
**Total** **36** **3** **11** **14** **1** **2** **67**
\*numbers may differ from sum of positive samples detected by real-time PCR as some samples could not be further determined.
^\#^from the following small mammal species none was positive for *Leptospira* spp. detected by real-time PCR: *Arvicola terrestris*, *Crocidura russula*, *Talpa europaea*, *Sorex* spp. and *Microtus agrestis*.
*Leptospira borgpetersenii* was exclusively and *L*. *interrogans* was mainly found in *Apodemus* spp. (n: 7/9, 77.8%, 95%CI: 44.3--94.7). Only two *M*. *glareolus* were positive for *L*. *interrogans* ([Table 2](#pntd.0004501.t002){ref-type="table"}). *Leptospira kirschneri* was mainly detected in positive small mammals (34 of 36 positive *M*. *glareolus*, 94.4%; 95%CI: 81--99.4; 14 of 25 positive *Apodemus* spp., 56%, 95%CI: 37.0--73.35). *Apodemus agrarius*, *Mustela nivalis* and *Mi*. *arvalis*, though captured in small numbers, showed high prevalences for *L*. *kirschneri* (42.86%; 95%CI: 15.75--75.02; 50%; 95%CI: 9.5--90.5 and 37.5%; 95%CI: 13.5--69.6 respectively ([Table 1](#pntd.0004501.t001){ref-type="table"})). *Leptospira kirschneri* was obtained in four of the five rodent species (*A*. *agrarius*, *Mi*. *arvalis*, *M*. *glareolus*, *A*. *flavicollis*). In contrast, *L*. *interrogans* and *L*. *borgpetersenii* were detected only in *A*. *flavicollis* and *M*. *glareolus*. All other investigated small mammal species were negative.
Whereas most *L*. *kirschneri*-positive rodents were found at site S (n: 27/54, 50.94%; 95%CI: 37.88--63.88), the majority of other pathogenic *Leptospira* spp.-positive samples were found at site B2 (n: 10/13, 76.92%; 95%CI: 49.06--92.5) (χ^2^ = 30.0823; p\< 0.00001), at a sampling site 379 m a.s.l.
Multi Locus Sequence Typing (MLST) {#sec019}
----------------------------------
From 67 samples tested positive by duplex PCR, for 14 a complete MLST covering all seven housekeeping genes could be determined for rodents captured at all three study sites (ten *M*. *glareolus*, three *Apodemus* spp. and one *A*. *agrarius*). Altogether sequence types for eleven *L*. *kirschneri* and three *L*. *interrogans* positive samples were detected. All *Leptospira interrogans* positive samples were found to be ST 24 ([Table 2](#pntd.0004501.t002){ref-type="table"}). For *L*. *kirschneri* two different sequence types were detected. Four samples were positive for ST 117 and seven for ST 110. While ST 110 could be detected at all three sites, ST 117 was only detected at site S. Here, ST 117 could be detected in *A*. *flavicollis*, *A*. *agrarius* and *M*. *glareolus* whereas ST 110 could only be detected in *M*. *glareolus*. *Leptospira interrogans* ST 24 could be detected in *A*. *flavicollis* and *M*. *glareolus* but exclusively at site B2 ([Table 2](#pntd.0004501.t002){ref-type="table"}).
Discussion {#sec020}
==========
This study focussed on pathogenic *Leptospira* species in small mammals from selected habitats in Germany. Studies on prevalences for *Leptospira* spp. in mammals in Europe are rare and focussed mainly on larger rodent species such as rats (*Rattus norvegicus*) which are considered to be the major source of *Leptospira* infection for humans \[[@pntd.0004501.ref034], [@pntd.0004501.ref035]\]. High prevalences (20--88%) in *Rattus norvegicus* have been reported from different European countries such as Turkey, France and Denmark \[[@pntd.0004501.ref036], [@pntd.0004501.ref037], [@pntd.0004501.ref038]\]. Studies from Germany, Switzerland, the Netherlands, Croatia and Austria showed the occurrence of *Leptospira* spp. in a wide range of different small mammal species including *M*. *glareolus*, *Apodemus* spp., *Mi*. *arvalis*, *Mus musculus*, *Castor fiber* and *Sorex* spp. (2.9--71.4%)\[[@pntd.0004501.ref005], [@pntd.0004501.ref022], [@pntd.0004501.ref039], [@pntd.0004501.ref040], [@pntd.0004501.ref041], [@pntd.0004501.ref042], [@pntd.0004501.ref043]\]. This study's prevalences show a similar wide range in prevalence regarding the investigated rodent species (5.3--42.9%). The highest prevalence in small mammals was detected at site B1, a forest in southern Bavaria in comparison to the other two study sites. A recent German study showed high prevalences of leptospiral DNA in *Mi*. *arvalis* and *A*. *agrarius* (12--14%) which are supposed to be the most common carrier hosts for *L*. *kirschneri* \[[@pntd.0004501.ref005]\]. In this study the highest prevalence was also found in both of these rodent species for *L*. *kirschneri*. *Leptospira kirschneri* was detected in almost all investigated rodent species (*M*. *glareolus*, *Mi*. *arvalis*, *A*. *flavicollis*, *A*. *agrarius*) with the exception of *Mi*. *agrestis* suggesting that this *Leptospira* species has a broad host range and is well adapted to a number of different small mammal species. Additionally, *L*. *kirschneri* was found in *Mustela nivalis* but not in *Sorex* spp. which therefore may play a subordinate role as maintenance host for *L*. *kirschneri*.
Human leptospirosis case reports caused by *L*. *kirschneri* are scarce. A recent study from Poland, however, reported antibody titres against ten serovars belonging to *L*. *kirschneri*, *L*. *borgpetersenii and L*. *interrogans* in several healthy humans \[[@pntd.0004501.ref021]\]. This argues for asymptomatic infections and a less pathogenic potential than in other pathogenic *Leptospira* species. *Leptospira kirschneri* is known to cause unspecific clinical symptoms in dogs including diarrhoea, lethargy and dehydration \[[@pntd.0004501.ref044]\]. Human cases caused by *L*. *kirschneri* may likewise display such unspecific illness and Leptospirosis may well be overlooked or kept undiagnosed. Nevertheless, this *Leptospira* species should be taken into account as a possible cause of disease in mammals other than dogs and humans.
*Leptospira interrogans* is known to cause severe symptoms such as pneumonia, hepatitis and kidney failure in humans and dogs \[[@pntd.0004501.ref045], [@pntd.0004501.ref046]\]. The hazardous impact of *L*. *interrogans* to human health was recently described in France in human cases with symptoms such as lumbar myalgia and pneumonia \[[@pntd.0004501.ref020]\]. Moreover unspecific clinical symptoms such as lethargy and fever in humans with previous outdoor activities (e.g. strawberry harvesters, triathletes) were reported in Germany and Austria \[[@pntd.0004501.ref012], [@pntd.0004501.ref013], [@pntd.0004501.ref015], [@pntd.0004501.ref016]\]. High prevalences (33.3--100%) were detected in *Mi*. *arvalis*, *Mus musculus* and *Rattus norvegicus* which occurred sympatrically to human Leptospirosis outbreaks in France and Germany \[[@pntd.0004501.ref016], [@pntd.0004501.ref020]\]. In former studies *Mi*. *arvalis* was also pointed out to be one of the most important maintenance hosts for *L*. *interrogans* among small mammal species \[[@pntd.0004501.ref047], [@pntd.0004501.ref048]\]. In the current study, however, this highly pathogenic *Leptospira* species was mostly detected in *Apodemus* spp. and exclusively at site B2, at four locations in a national park, which leads to the assumption that *L*. *interrogans* in contrast to *L*. *kirschneri* (at least of ST 110, see below) rather occurs focally.
*Leptospira borgpetersenii* and in particular serovar Hardjo type Hardjobovis is reported as the most causative leptospiral agent for infertility and abortion in cattle from North America \[[@pntd.0004501.ref049]\]. *Leptospira borgpetersenii* strains were described to be associated with *Mus musculus* \[[@pntd.0004501.ref050]\]. In the present study this *Leptospira* species was detected at all three sites but, exclusively and significantly more often in *Apodemus* spp. than in any other small mammal species which suggests that certain *L*. *borgpetersenii strains* have probably a host preference for the genus *Apodemus* in the investigated habitats.
To the authors' knowledge, this study is providing first evidence on sequence types of *Leptospira* spp. in rodents from Germany. *Leptospira kirschneri* ST 110 was the most widespread and most common ST in this study. In Germany former studies reported diseases in humans caused by the serogroup Grippotyphosa, declared as "mudfever" and associated with fieldwork activities, such as strawberry harvesting \[[@pntd.0004501.ref016]\]. *Leptospira kirschneri* ST 117 was formerly found in *A*. *agrarius* and *A*. *flavicollis* collected in Croatia \[[@pntd.0004501.ref042]\]. In our study this ST was also found in both of Apodemus species and additionally in *M*. *glareolus*. In Spain *Leptospira interrogans* ST 24was detected in dogs and wild carnivores such as *Vulpes vulpes* showing clinical signs, thus several carnivore species were suggested to be not maintenance but dead end hosts for *Leptospira interrogans* \[[@pntd.0004501.ref051]\]. Further ST 24 was detected in *A*. *flavicollis* from Croatia \[[@pntd.0004501.ref042]\]. In the present study, this sequence type was found in two different rodent species (*M*. *glareolus*, *A*. *flavicollis)*.
It should be taken into account that the comparison of our results between sites and species is limited due to different DNA extraction methods and as the animals examined were not caught in the same years.
In summary, in our study at three sites in Germany pathogenic *Leptospira* spp. were detected in high prevalences in four of five investigated rodent species. Therefore humans could during leisure time activities get into contact with these pathogenic *Leptospira* spp. if respective transmission conditions are optimal. Regarding the *Leptospira* spp. prevalences this study's results suggest a host preference for *L*. *borgpetersenii* in *Apodemus* spp. Moreover a broad host spectrum was detected for *L*. *kirschneri* which was the most common species detected in this study. Besides this study is proving first evidence of *L*. *kirschneri* ST 110 and 117 as well as *L*. *interrogans* ST 24 in rodents from Germany.
The authors wish to thank Claudia Thiel, Pauline Bleichert, Daniela Eder, Stefan Frey, Mareike Pollaerts, Robin Reiter, Rahime Terzioglu, Florian Goldberg, Harald Weber, Tim Tiedemann, Jennifer Krieg, Daniela Sum, Franziska Eller and Claudia Kehler for their help in lab and field work.
[^1]: The authors have declared that no competing interests exist.
[^2]: Conceived and designed the experiments: CS MP SE. Performed the experiments: MP SE DW SS CK AO. Analyzed the data: DW AO. Contributed reagents/materials/analysis tools: CS MP SE. Wrote the paper: AO SE CS MP.
[^3]: Current address: National Center of Vector Entomology, Institute of Parasitology, University of Zürich, Zurich, Switzerland
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper.
Introduction {#sec006}
============
Parkinson's disease (PD) is a neurodegenerative process that leads to the selective loss of dopaminergic neurons, mainly in the basal ganglia of the brain. Clinical manifestations include movement alterations as well as non-motor symptoms, such as dementia, depression, and autonomic dysfunction \[[@pone.0177163.ref001]\]. Neurons and neural circuits outside the basal ganglia can be affected simultaneously or upstream of the substantia nigra \[[@pone.0177163.ref002]\].
Vision is one of the non-motor systems altered in PD, especially the visual field corresponding to the fovea \[[@pone.0177163.ref003],[@pone.0177163.ref004]\]. Recent studies demonstrated retinal thinning in different macular sectors and retinal nerve fiber layers (RNFL) in PD patients compared with healthy subjects, \[[@pone.0177163.ref005],[@pone.0177163.ref006]\], and alterations in multifocal electroretinograms \[[@pone.0177163.ref007],[@pone.0177163.ref008]\].
Several mechanisms have been proposed for the axonal loss in PD disease, leading to tissue degeneration and ultrastructural changes of the retinal ganglion cells \[[@pone.0177163.ref009]\], but the changes of the choroidal layer have not been thoroughly evaluated. Mechanobiologic response of tissues \[[@pone.0177163.ref010]\] and cells \[[@pone.0177163.ref011]\] depends on the mode of deformation, and the magnitude and temporal profile of the stimulus, as well as the type of tissue or cell and its biologic state. Understanding the particular deformations observed in each tissue and ocular layer in patients with PD might facilitate diagnosis and treatment.
Before the development of OCT, choroidal studies were limited to histopathologic analysis. OCT is a useful tool for choroidal studies; nevertheless, the role of choroidal analysis for ocular pathologies is not yet established. Spectral domain-optical coherence tomography (SD-OCT), mostly with enhanced depth imaging (EDI), has been used to evaluate the macular and peripapillary choroid (mainly in healthy eyes and glaucoma patients) \[[@pone.0177163.ref012],[@pone.0177163.ref013]\], but the relation of OCT measurements with changes in the peripapillary choroid remain unclear. Some studies report a reduction in the mean or regional peripapillary choroidal thickness (PPCT) in primary open-angle glaucoma \[[@pone.0177163.ref014]--[@pone.0177163.ref016]\], but in these studies, choroidal thickness was measured manually using SD-OCT at only a few points and beneath the circumpapillary ring, an area typically used for RNFL evaluation. The automated segmental measurement software used in the present study, however, is better suited for a broader and more objective evaluation of choroidal thickness. Swept-source (SS) OCT, as compared with SD-OCT with EDI, provides better visualization of the choroid \[[@pone.0177163.ref017]\], more accurately measures the deep tissues, detects the posterior limit of the sclera, \[[@pone.0177163.ref018]\], and is applicable for evaluating a broader area of the posterior segment.
Based on the improved ability of this new SS-OCT technology to reveal and automatically measure a wide area of the peripapillary choroid, our first objective was to measure the PPCT in a 26×26 cube-grid centered on the optic disc, which is automatically performed by the Deep Range Imaging (DRI)-OCT Triton (Topcon Corporation, Tokyo, Japan), in a sample of healthy subjects to determine the pattern or distribution of PPCT and to establish objective zones with similar choroidal thicknesses. Our second objective was to study the PPCT differences within these zones in a sample of PD patients compared with age- and sex-matched healthy controls. The third objective was to evaluate the relationship between PPCT alterations and PD severity. The main advantage of the present study is that PPCT was evaluated in a wide area of the peripapillary choroid using an automatic and accurate new method.
Material and methods {#sec007}
====================
Study population and design {#sec008}
---------------------------
This was a prospective, observational, cross-sectional case-control study. The study included patients with definite PD, and age- and sex-matched healthy controls. Based on our preliminary studies, we calculated the necessary sample size to detect differences in choroidal thickness of at least 20 μm as measured by OCT, applying a two-tailed test with an alpha of 5% and a beta of 10%, and a risk ratio of 0.5. Based on this calculation, at least 70 eyes were needed (35 from PD patients and 35 from healthy controls). A total of 40 eyes of 40 PD patients and 80 eyes of 80 healthy controls were evaluated. PD diagnosis was based on the UK Brain Bank Criteria, which included, in the first stage, bradykinesia and one additional symptom, i.e., rigidity, 4--6 Hz resting tremor, or postural instability \[[@pone.0177163.ref019],[@pone.0177163.ref020]\].
Patients with a visual acuity less than 0.1 (Snellen scale), intraocular pressure (IOP) \>20 mmHg, optic neuritis antecedent, no transparent ocular media (nuclear color/opalescence, cortical or posterior subcapsular lens opacity ≥1 according to the Lens Opacities Classification System III system) \[[@pone.0177163.ref021]\] and systemic disease that could affect the eye (e.g., diabetes, neurologic pathologies, hypertension, and endocrine disorders) were excluded from the study. Subjects with refractive errors greater than 5 diopters (D) of spherical equivalent refraction or 3 D of astigmatism were also excluded from the study.
Standard protocol approvals, registrations, and patient consent {#sec009}
---------------------------------------------------------------
The study procedures were performed in accordance with the tenets of the Declaration of Helsinki, and the study protocol was reviewed and approved by the Aragon Ethics Committee For Clinical Research before the study began. Written informed consent to participate in the study was obtained from all subjects.
Main outcome measures {#sec010}
---------------------
All subjects underwent a complete neuro-ophthalmic examination, including assessment of best-corrected visual acuity using the Snellen chart, pupillary reflexes, and ocular motility; examination of the anterior segment, IOP with the Goldmann applanation tonometer, and papillary morphology by funduscopic exam; as well as OCT.
In the PD group, disease severity was assessed using the Unified Parkinson Disease Rating (UPDRS) and the Hoehn and Yahr scales, and disease duration since the PD diagnosis was recorded. The Hoehn and Yahr scale is a commonly used diagnostic tool for quantifying the progression of PD symptoms \[[@pone.0177163.ref022]\]. Stages range from 0 (no signs of disease) to 5 (requiring a wheelchair, or bedridden unless assisted). Clinicians and researchers most commonly use the UPDRS, and the motor section in particular, to follow the longitudinal course of PD in clinical studies \[[@pone.0177163.ref023]\]. The scale includes three sections that evaluate the key areas of disability, and a fourth section that evaluates treatment complications. Treatment for PD was registered using three different categories for clearer classification: "drugs that enhance dopamine levels" (carbidopa, levodopa, and rasagiline), "dopaminergic drugs" (pramipexole, ropirinole, rotigotine), and "other" (amitriptyline, propranolol, clonazepam).
OCT {#sec011}
---
An optic disc 6.0×6.0 mm three-dimensional scan was obtained using the DRI OCT Triton (Topcon Corporation). This scan combines morphometric optic disc parameters and various peripapillary parameters, including RNFL and choroidal thickness. The subjects were seated and properly positioned. All DRI-OCT images were obtained by a single well-trained technician blinded to the presence or absence of PD. The DRI-OCT Triton includes the new SMARTTrack tool that enhances tracking, corrects for motion, and guides the operator to reduce potential errors while acquiring images. Only eyes with good quality scans were included in the analysis. Good-quality SS-OCT images were defined as those with a signal strength ≥40 (maximum = 100), and without motion artifact, involuntary saccade, or overt misalignment of decentration. A total of three eyes (two in the PD group and one in the control group) were excluded due to poor DRI-OCT image quality. These eyes were substituted with two new patients in the PD group and one new healthy subject in the control group. The same investigator performed all of the OCT scans and checked the accuracy of segmentation in each scan and the lack of artifacts. A total of 15 scans in the PD group and 10 in the control group were excluded and repeated.
A 26×26 cube-grid centered on the optic disc was generated to automatically measure choroidal thickness. This grid comprised 676 cubes (200 μm x 200 μm) around the optic nerve head with the 88 central cubes corresponding to the optic nerve head area not analyzed; therefore the DRI-OCT Triton displays choroidal thickness for a total of 588 peripapillary cubes ([Fig 1](#pone.0177163.g001){ref-type="fig"}). The Bruch membrane and choroidal-scleral interface were delineated with the segmentation algorithm implemented by Topcon ([Fig 1](#pone.0177163.g001){ref-type="fig"}).
{#pone.0177163.g001}
A total of 120 eyes were analyzed {#sec012}
---------------------------------
80 healthy controls and 40 PD patients. The healthy control group was randomly divided in two populations: the teaching population (n = 40 controls, used to establish choroidal zones) and the validating population (n = 40 controls, used to compare measurements with PD patients). Sex and age were not significantly different between the control groups or between each of the control groups and the PD group. Only right eyes were selected for the statistical analysis, because choroidal thickness is reported to differ between right and left eyes \[[@pone.0177163.ref023],[@pone.0177163.ref024]\].
Statistical analysis {#sec013}
--------------------
Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS 20.0, SPSS Inc., Chicago, IL). The Kolmogorov-Smirnov test was used to assess the sample distribution of the variables. For quantitative data following a parametric distribution, differences between evaluation groups were compared using the Student´s t-test. For qualitative data, a chi square test was used for comparison. We delimited four areas based on the range of choroidal thicknesses and compared the average of these areas between the PD and control groups. We also compared choroidal thicknesses and volume in the four quadrants and six sectors provided by Triton OCT between PD patients and healthy controls. To calculate volume, we computed the thickness of the four quadrants and the six sectors, and total thickness relative to the effective imaging area; i.e., the total imaging area (peripapillary constant area, 27.04 mm^2^) minus the optic nerve head area (variable area, 3.52 mm^2^). We evaluated the linear agreement between PPCT and the two neurologic scales of PD severity (the UPDRS and the Hoehn and Yahr scales) using the Pearson correlation coefficient. A p value of less than 0.05 was considered statistically significant.
Results {#sec014}
=======
Teaching population evaluation to establish peripapillary choroidal zones {#sec015}
-------------------------------------------------------------------------
The teaching population comprised 40 right eyes from healthy subjects and was used to identify 5 choroidal zones. The mean age of this population was 69.0 ± 7.9 years (range, 62--84 years). Of the 51 subjects, 14 (35%) were women. Mean spherical equivalent was 1.10 ± 1.35 D. Study zones were previously set regarding the thickness of the cubes for each acquisition point. Zone 1 corresponded to the optic nerve head area, and thus was not measured by the OCT and not incorporated into the study. Zone 2 included the cubes with a choroidal thickness of less than 105 μm, zone 3 included cubes with a thickness of 105 to 139 μm, zone 4 included cubes with a thickness of 140 to 174 μm, and zone 5 included cubes with a thickness of 175 μm or greater ([Fig 1](#pone.0177163.g001){ref-type="fig"}).
The mean choroidal thickness in zone 2 was 95.00 ± 8.20 μm and included 120 cubes of the choroidal grid; in zone 3 the mean thickness was 121.84 ± 9.56 μm and included 248 cubes; in zone 4 the mean thickness was 156.58 ± 9.07 μm and included 181 cubes; and in zone 5 the mean thickness was 186.90 ± 8.82 μm and included 31 cubes ([Fig 2](#pone.0177163.g002){ref-type="fig"}).
{#pone.0177163.g002}
Figs [3](#pone.0177163.g003){ref-type="fig"} and [4](#pone.0177163.g004){ref-type="fig"} show the five zones in the teaching population of healthy controls, that are roughly concentric to the optic nerve head, zone 2 mainly (the thinnest of the study zones with a minimum mean PPCT of 78 μm) located nearest the optic nerve head and inferior peripapillary choroid, zone 3 was mainly located in the inferior and nasal peripapillary choroid, zone 4 (around zone 3, especially in the temporal and superonasal areas), and zone 5 (the zone with the maximum mean PPCT of 205 μm, corresponding to the farthest cubes, mainly located in the superior and temporal peripapillary choroid).
{#pone.0177163.g003}
{#pone.0177163.g004}
Validating population and statistical comparison between healthy and PD eyes {#sec016}
----------------------------------------------------------------------------
Once the study zones were established, a statistical comparison between control and PD eyes was performed on a different population of control eyes. A total of 40 right eyes from healthy subjects (independent from those subjects used in the teaching population) and 40 right eyes from PD patients were included in the study. The mean age of the healthy control group was 68.58 ± 7.17 years (range: 61 to 85 years) and the mean age of the PD group was 69.76 ± 6.45 years (range: 62 to 85 years). Of the 40 subjects in each group, 14 (35%) were women. Mean spherical equivalent was 0.13 ± 1.88 D in the control group (range, -2.50 to 2.50) and 0.14 ± 1.76 D in the PD group (range, -2.75 to 2.50). Age, sex, and spherical equivalent did not differ significantly between groups (p\<0.05). The study population characteristics are summarized in [Table 1](#pone.0177163.t001){ref-type="table"}.
10.1371/journal.pone.0177163.t001
###### Descriptive, clinical data, choroidal thicknesses and volume for eyes of patients with Parkinson's disease and from healthy controls (validating population).
{#pone.0177163.t001g}
PD PATIENTS CONTROLS P[\*](#t001fn002){ref-type="table-fn"}
------------------------------------------------------------------------- ---------------- ---------------- ----------------------------------------
Number of eyes 40 40 \-
Age (years) 69.8 ± 6.5 68.6 ± 7.2 0.137
Sex (men/women) 26/14 26/14 0.901
Spherical Equivalent (D) 0.14 ± 1.76 0.13 ± 1.88 0.388
Number of myopic subjects (%) [\*\*](#t001fn003){ref-type="table-fn"} 18 (45%) 17 4(42.5%) 0.404
Number of hyperopic subjects (%)[\*\*](#t001fn003){ref-type="table-fn"} 6 (15%) 6 (15%) 0.945
BCVA (LogMAR) 0.11 ± 0.17 0.03 ± 0.16 0.005
IOP (mm Hg) 16.5 ± 2.0 16.2 ± 2.8 0.439
Disc area (mm^2^) 1.92 ± 0.37 1.89 ± 0.37 0.309
Unified Parkinson Disease Rating Scale 25.1 ± 5.4 \- \-
Hoehn and Yahr Scale 2.7 ± 0.8 \-
Disease duration (years) 6.7 ± 2.3 \- \-
Choroidal thickness (μm)
Zone 2 126.89 ± 11.23 96.53 ± 9.55 \<0.0001
Zone 3 152.75 ± 16.00 121.39 ± 12.10 \<0.0001
Zone 4 190.21 ± 12.16 156.69 ± 12.44 \<0.0001
Zone 5 213.21 ± 11.83 186.64 ± 9.47 \<0.0001
Total 163.01 ± 29.90 131.51 ± 30.22 \<0.0001
Number of peripapillary cubes
70--104 μm thick 0 116 \<0.0001
105--139 μm thick 165 258 \<0.0001
140--174 μm thick 202 175 \<0.0010
≥175 μm thick 221 39 \<0.0001
Comparison of the choroidal thicknesses in the four established zones in the control group and Parkinson's disease patients, and comparison between groups. Comparison of the number of peripapillary cubes with a thickness from 70 to 104 μm (zone 2), from 105 to 139 μm (zone 3); from 140 to 174 μm (zone 4), and ≥ 175 μm (zone 5). Zone 1 corresponds to the optic nerve head and was not included in the analysis.
\* p: level of statistical significance in comparison between the two groups using Student's t-test (except for sex, chi-square test). Data are mean ± standard deviation. Bold text indicates statistically significant results (p\<0.05).
\*\* Subjects were considered myopic when spherical equivalent was \< -1D. Subjects were considered hyperopic when spherical equivalent was \>1D. Abbreviations: PD, Parkinson´s disease; D, diopters; BCVA, best-corrected visual acuity; IOP, intraocular pressure.
"Drugs that enhance dopamine levels" was the most prescribed category (87% of patients) and combination therapy with levodopa and carbidopa was the most frequent treatment (46%). Sixty-five percent of treatments were categorized as "dopaminergic", most of which were used in combination with the drugs included in the "Drugs that enhance dopamine levels" category. A small percentage of patients (8%) were prescribed drugs with no dopaminergic effects.
Comparison of the PPCT between the healthy controls and the PD eyes in the five zones described revealed a statistically thicker choroidal layer in all zones in the PD group (p\<0.0001; [Table 1](#pone.0177163.t001){ref-type="table"} and [Fig 2](#pone.0177163.g002){ref-type="fig"}). The choroid followed a similar pattern in controls and PD; it was thicker in the superotemporal region, followed by the temporal, nasal, and inferior regions, as shown Figs [3](#pone.0177163.g003){ref-type="fig"} and [4](#pone.0177163.g004){ref-type="fig"}. PPCT was higher in PD patients. The choroidal thicknesses did not correlate with disease severity.
Evaluation of choroidal thicknesses in classic tomographic areas {#sec017}
----------------------------------------------------------------
The parapapillary choroidal thicknesses of the quadrants and sectors provided by the OCT analysis were also compared between PD patients and healthy controls, and revealed a significant increase in patients ([Table 2](#pone.0177163.t002){ref-type="table"}).
10.1371/journal.pone.0177163.t002
###### Macular and peripapillary choroidal thickness and volume measured with swept-source deep range imaging optical coherence tomography in patients with Parkinson's disease and healthy controls.
{#pone.0177163.t002g}
-------------------------------------------- ------------------ ---------------- ---------
**Peripapillary choroidal thickness (μm)** **PD PATIENTS**\ **CONTROLS**\ **p**
**Mean (SD)** **Mean (SD)**
Total 153.51 (61.32) 125.01 (54.76) 0.009
*Quadrants (x4)*
Superior 163.01 (64.44) 146.65 (59.42) 0.121
Nasal 151.97 (52.21) 131.60 (55.05) 0.017
Inferior 127.11 (61.49) 100.23 (56.40) 0.007
Temporal 163.80 (75.96) 130.12 (53.58) 0.002\*
*Sectors (x6)*
Superotemporal 164.11 (65.00) 142.98 (56.81) 0.101
Superonasal 168.87 (64.05) 150.28 (64.41) 0.254
Nasal 149.11 (52.76) 128.82 (56.04) 0.023
Inferotemporal 129.71 (69.92) 94.15 (50.01) 0.002\*
Inferonasal 126.01 (57.13) 100.02 (58.93) 0.015
Temporal 164.04 (77.92) 130.00 (56.21) 0.002\*
**Peripapillary choroidal volume (mm^3^)**
Total 3.61 (61.32) 2.94 (1.29) 0.007
*Quadrants (x4)*
Superior 0.96 (0.38) 0.86 (0.35) 0.098
Nasal 0.89 (0.31) 0.77 (0.32) 0.012
Inferior 0.74 (0.36) 0.59 (0.33) 0.005
Temporal 0.96 (0.44) 0.76 (0.31) 0.001\*
*Sectors (x6)*
Superotemporal 0.64 (0.25) 0.56 (0.22) 0.081
Superonasal 0.66 (0.25) 0.59 (0.25) 0.178
Nasal 0.581 (0.21) 0.50 (0.22) 0.019
Inferotemporal 0.51 (0.27) 0.37 (0.20) 0.002\*
Inferonasal 0.49 (0.22) 0.39 (0.23) 0.016
Temporal 0.64 (0.31) 0.51 (0.221) 0.001\*
-------------------------------------------- ------------------ ---------------- ---------
Bold letters indicate p\<0.05. Asterisks mark significant values according to Bonferroni's correction for multiple comparisons (p\<0.005). Abbreviations: PD, Parkinson's disease.
Discussion {#sec018}
==========
The involvement of the choroid in PD is unknown. To our knowledge, Eraslan et al.\[[@pone.0177163.ref025]\] published the only study that discusses the choroid in relation to PD and suggested that choroidal thickness is significantly diminished in PD patients \[[@pone.0177163.ref025]\]. In that study, a smaller sample size was evaluated and the measurements of the choroidal plexus were obtained using SD-OCT EDI. In SD-OCT EDI, the inner and outer borders of the choroid are determined manually by the observer \[[@pone.0177163.ref023]\], and the error in the thickness measurements must be estimated from the obtained images. In the present study, we used SS-OCT, in which the tunable laser source has a lower signal decay versus depth than previous SD-OCT devices. Thus, the chorioscleral border is more clearly delineated due to the reduction in speckle noise, providing more accurate manual and automated three-dimensional images of the retina and choroid \[[@pone.0177163.ref026]\]. In addition, with the SS-OCT device, observers only analyze a circle scan (SS-OCT only evaluates 12 central degrees around the optic nerve and reports manual measurements of the choroidal thickness for various points within this small area) \[[@pone.0177163.ref012]\], while the DRI-OCT used in our study provides a total of 588 automatic measurements of the choroidal thickness, with each measurement covering a broader area around the optic nerve (5.2x5.2 mm, corresponding to approximately 20 central degrees). Our present results revealed a significantly thickened choroid in the peripapillary area, which is opposite to the findings for the retina in PD and to the Eraslan et al. results \[[@pone.0177163.ref025]\].
RNFL thickness measurements by OCT are considered a useful indirect marker of the progression of brain atrophy in patients with PD \[[@pone.0177163.ref006],[@pone.0177163.ref027]\]. Previous studies demonstrated macular thinning of all areas in PD patients compared with controls, \[[@pone.0177163.ref005],[@pone.0177163.ref006]\], an inverse correlation with Hoehn and Yahr and UPDRS severity, and a positive correlation with the Schwab-England Activities of Daily Living scale \[[@pone.0177163.ref027]\]. Therefore, increased neurologic effects and severity of PD progression are linked to thinning of macular tissue \[[@pone.0177163.ref027]\]. Typical ophthalmic findings in PD patients are optic nerve atrophy and peripapillary RNFL thinning \[[@pone.0177163.ref006],[@pone.0177163.ref009]\], but the secondary impact on other deeper ocular layers, such as the choroid, have not been evaluated.
In our study, PD patients showed a 31.5% increase in the choroidal thickness in zone 2, 25.8% increase in zone 3, 21.5% increase in zone 4, and 14.2% increase in zone 5 compared with controls. Therefore, the choroidal thickness in PD patients increased with increasing distance from the optic nerve head. The cause of this mean total increase of 24.0% in the choroidal thickness of PD patients is unknown. We found that all quadrants and peripapillary sectors provided by classic topographic OCT analysis were increased in PD compared with healthy controls. The choroidal thickening in PD was generalized, and affected all the topographic zones around the optical nerve.
A vascular-based influence for PD progression was recently proposed without clear results \[[@pone.0177163.ref028]--[@pone.0177163.ref030]\]. Despite the observation of vascular changes in vascular parkinsonism \[[@pone.0177163.ref028]\], no differences are observed in white matter vascularization in idiopathic PD compared to controls \[[@pone.0177163.ref029]\]. In a recent study, an increased density of string vessels (vessels with collapsed basement membranes, without endothelium, and with no functional circulation) was observed in the brain of PD patients, associated with neuronal loss \[[@pone.0177163.ref030]\]. Vascular ultrastructural alterations in the choroid of PD patients have not yet been described, thus we postulate that the increased choroidal thickness may be due to alterations in the perivascular connective tissue density despite the previously suggested hypoperfusion in these patients \[[@pone.0177163.ref025]\]. Also, SS-OCT technology provides increased depth analysis of the choroid compared to SD-OCT devices, and that also may account for the increased choroidal thickness compared to previous studies. More studies analyzing the macular and peripapillary choroid in PD using new SS-OCT technology and histologic analysis are needed to corroborate our findings.
The choroid is a dynamic structure and its thickness depends on several factors. For example, older age \[[@pone.0177163.ref031],[@pone.0177163.ref032]\], higher IOP \[[@pone.0177163.ref033]\], higher myopia, and longer axial length \[[@pone.0177163.ref034],[@pone.0177163.ref035]\] are associated with a thinner choroid. In the present study, however, we excluded subjects with high IOP, high myopia, and longer axial length. Several factors contribute to the conflicting findings in PPCT with OCT; such as different measurement techniques (most of them performed manually), and the dynamic and variable nature of the choroid \[[@pone.0177163.ref036]\]. New technologies, such as DRI SS OCT, provide not only better delineation of the retinal layers but also a much improved method for measuring a very large number of points in the peripapillary region and other regions farther from the choroid. With these tools, we can determine the true shape of the choroidal layer and possible differences due to various diseases.
Our intention was to first better understand the pattern of PPCT in normal eyes, establish different zones based on the distribution of choroidal thickness, and to compare these areas with eyes of patients with PD. Some histologic studies and others with SD-OCT demonstrated that the choroid tends to be thinner around the optic nerve head compared with the subfoveal choroid \[[@pone.0177163.ref032],[@pone.0177163.ref035],[@pone.0177163.ref037]\], which is consistent with our results. We found that the peripapillary choroid follows a pattern in healthy subjects and PD patients; and is thicker superiorly and thinner near the optic nerve head, especially inferiorly, which is consistent with other studies \[[@pone.0177163.ref038]\]. The choroidal pattern is the same in PD patients, but with thicker measurements. Our findings revealed that the peripapillary choroid was thicker in all zones in PD patients, and the differences between healthy and PD eyes tended to decrease with increasing distance from the optic nerve head.
This study included a teaching population to determine the pattern of the topographic distribution of thickness in the peripapillary choroidal layer, and a validating population to compare the thicknesses of the PD patients with an independent sample of healthy controls. Our research team, as well as others, use this method of utilizing teaching and validating populations to avoid the overestimation effect caused by using the same subjects to create the model and to compare that model with controls \[[@pone.0177163.ref039],[@pone.0177163.ref040]\].
This study has several limitations. First, we did not collect data on axial length; however, we demonstrated that there was no significant difference in refractive error among groups. The axial length and refractive error variables are related, but not interchangeable, because axial length tends to not change after the second decade of life, except in cases with some pathology. Instead, the refractive error may fluctuate or change due to different factors \[[@pone.0177163.ref024]\]. The findings from a recent study by Ikuno et al. \[[@pone.0177163.ref041]\] using OCT, however, suggest that refractive error and axial length may have a similar relationship with OCT measurements. Second, the choroid can be influenced by circadian changes \[[@pone.0177163.ref042]\], and we did not perform OCT examinations at the exact same time of day in all participants; although all of the OCT studies were acquired in the evening (between 16.00 and 20.00). Third, we only measured choroidal thickness in the right eyes of each participant, and inter-eye differences were not assessed. Fourth, the presence of motor symptoms in some of the patients may have affected the masking (i.e., masking was more effective in patients with mild or non-evident symptoms). Finally, we did not establish the reliability of automated choroidal thickness measurements with this new SS-OCT. It would be interesting to investigate the accuracy of this device in another study.
Imaging techniques such as OCT are noninvasive, inexpensive, and fast, and may be useful for monitoring treatment efficacy, detecting progressive neurodegeneration, and improving diagnostic procedures in neurologic diseases (such as making a differential diagnosis between PD and essential tremor) \[[@pone.0177163.ref043]\]. In our study, we did not find that choroidal thickness was correlated with PD severity, but studies with a larger sample size may find significant associations. A subgroup analysis according to PD medication was not possible as most of our patients were prescribed combined treatment and the sample size per subgroup was not large enough to compare the subgroups. More studies are needed to corroborate these results and to evaluate the usefulness of choroidal measurements in PD and the role of this evaluation in clinical practice.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
[^2]: **Conceptualization:** EGM MPB.**Data curation:** EGM MPB MS.**Formal analysis:** EGM LEP MPB.**Investigation:** EGM MPB MS.**Methodology:** EGM LEP TR EV EO BC RA.**Project administration:** EGM MS LEP.**Resources:** LEP VP JML RA.**Supervision:** EGM.**Validation:** EGM.**Visualization:** EGM TR BC.**Writing -- original draft:** EGM MS LEP.**Writing -- review & editing:** EGM LEP MPG RA VP JML EV EO BC TR MS.
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
This case highlights the management of a rare complication of a condition, which is usually benign.
Case presentation
=================
In 2007, a 22-year-old female university student, presented to a local hospital with severe epigastric pain and vomiting. As her amylase level was significantly elevated, she was diagnosed with acute pancreatitis. She was managed conservatively with analgesia and i.v. fluids and made an uneventful recovery. The commonest causes of pancreatitis including drugs, gallstones, corticosteroids, excess alcohol and hypertriglyceridaemia were excluded. However, she was noted to have an elevated calcium level, which was thought to be the cause of her pancreatitis. As her parathyroid hormone (PTH) level was detectable, an initial diagnosis of primary hyperparathyroidism was made. She subsequently underwent two neck explorations, which failed to reveal a parathyroid adenoma. She also had detailed imaging of her pancreas including a magnetic resonance pancreatography, that ruled out any anatomical causes of pancreatitis.
She continued to have recurrent episodes of pancreatitis requiring admission to hospital. In 2010, the episodes of pancreatitis were becoming more frequent, limiting her activities and education. Her serum calcium levels remained persistently elevated, and she was referred to our centre for a second opinion.
Investigation
=============
Her investigations performed at our institution were as follows: elevated adjusted serum calcium level of 2.79 mmol/l (reference range, 2.2--2.58), PTH level of 4.2 pmol/l (0.6--6.0), low 24 h urine calcium of 0.3 mmol/l and a urine calcium:creatinine ratio of \<0.03. A clinical and biochemical diagnosis of familial hypocalciuric hypercalcaemia (FHH) was made. Genetic testing confirmed the diagnosis and showed a G to A (c.1703G\>A) nucleotide substitution in exon 6 of calcium-sensing receptor (*CASR*).
Treatment
=========
She was initially started on 30 mg of cinacalcet daily. Although her calcium levels improved (2.53--2.66 mmol/l) and her frequency of pancreatitis significantly reduced, she was not free from attacks. Her dose of cinacalcet was subsequently increased to 30 mg twice daily and her corresponding calcium levels dropped further to 2.24 mmol/l. After 8 months of being symptom free on the latter dose, she had a further episode of pancreatitis. At this point, her cinacalcet dose was increased to 30 mg three times daily, aiming for a serum calcium level at the lower end of the reference range.
Outcome and follow-up
=====================
The cinacalcet was well tolerated throughout. [Figure 1](#fig1){ref-type="fig"} shows a decrease in the frequency of her pancreatitis attacks as her serum calcium levels dropped with an increase in cinacalcet dose. She has not had an episode of pancreatitis for more than 2 years and her latest calcium levels remain at 2.10 mmol/l. She is being followed up annually with a repeat serum calcium level as well as 24 h urine calcium level before analysis in the clinic.
{#fig1}
Discussion
==========
FHH is usually a benign inherited condition, which is usually associated with minimal sequelae from hypercalcaemia. Pancreatitis has been reported rarely [@bib1] [@bib2]. Other complications include chondrocalcinosis, gallstones and osteoporosis [@bib3]. It is well established that FHH is caused by inactivating mutations in the *CASR* gene, which reduce sensitivity to extracellular calcium. However, the mechanism of pancreatitis in this condition is poorly understood. Although the link between FHH and pancreatitis had previously been questioned [@bib4], more recent papers have associated it with variant mutations of the *CASR* [@bib1] possibly due to a combination of *CASR* mutation and a coexisting SPINK1 mutation, which predisposes to chronic pancreatitis [@bib5]. Theoretically, improving the sensitivity of mutant *CASR* to extracellular calcium would improve calcium homoeostasis and may offer effective treatment for patients with symptomatic FHH.
Calcimimetic agents such as cinacalcet are allosteric activators of CASR, which lower its activation threshold to extracellular calcium [@bib6]. Although commonly used in the management of secondary hyperparathyroidism, inoperable parathyroid adenomas and carcinomas [@bib6], it has also been a successful treatment for patients with symptomatic FHH [@bib3] [@bib7]. Mutations of the *CASR* in FHH also have various degrees of responsiveness to calcimimetics [@bib8]. Therefore, the response to cinacalcet and the dose required are likely to be unpredictable.
Whether or not cinacalcet can be used as a long-term therapy in these patients remains yet to be determined. One long-term case series [@bib9] in which patients with FHH were treated with cinacalcet for up to 3 years showed that not only was it well tolerated but there was also a sustained improvement in serum calcium levels as well as PTH levels.
Our patient responded well to a moderate dose of cinacalcet. This case highlights the effective use of cinacalcet and the potential strategy in managing FHH patients presenting with recurrent pancreatitis. It also draws attention to getting the right diagnosis in order to avoid unnecessary surgical intervention in these patients.
Patient consent
===============
Written informed consent was obtained from the patient for publication.
Author contribution statement
=============================
Dr M Druce is the patient\'s main physician and is a co-author.
Declaration of interest
=======================
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Funding
=======
This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Severe, acute salicylate toxicity remains a common presentation to the Emergency Department (ED) and is associated with a significant degree of mortality \[[@B1]\]. In its unionized form, salicylate can move across cell membranes into tissues to exert toxic effects. In the presence of acidemia, salicylate will shift to this unionized form, which allows it to cross the blood-brain barrier, and cause central nervous system toxicity (cerebral edema, seizures, and coma). Therefore, the presence of acidemia is seen as a poor prognostic indicator. Classically, salicylate toxicity is initially associated with a respiratory alkalosis, secondary to direct stimulation of the medulla, and subsequent tachypnea and hyperpnea as a response to metabolic acidosis. For this reason, intubation and mechanical ventilation have been commonly recommended against in severe salicylate poisoning, as it is believed that this intervention may take away this protective respiratory drive \[[@B2], [@B3]\]. Other sources suggest that intubation may be safely performed, as long as apneic time during induction is minimized, and the patient is hyperventilated adequately on the ventilator \[[@B3]\]. Unfortunately, very little evidence exists on this topic, and there are multiple reasons that a patient with salicylate toxicity may require intubation, including decreased or altered level of consciousness, failure to protect airway, and respiratory distress from pulmonary edema. Taken together, there is very little understanding regarding the approach to intubation of patients with severe salicylate overdose. We present a case of a patient with unknown severe salicylate toxicity, who was intubated upon arrival, but immediately hyperventilated several minutes later once initial blood gas levels revealed the diagnosis. To our knowledge, we are the first to publish pre- and postintubation blood gas data in the context of ventilator settings that should have resulted in hyperventilation and improving acidemia.
2. Case Report {#sec2}
==============
The patient was a 59-year-old male farmer who had been found by his daughter to be confused and tachypneic on the morning of presentation, and she brought him to the local community ED. His vital signs were blood pressure of 116/75 mmHg, heart rate (HR) of 107 beats/minute, respiratory rate (RR) of 20 breaths/minute, initial temperature of 36.3 degrees\' Celsius, and 100% oxygen saturation. Glasgow Coma Scale (GCS) was 15 on initial physician assessment. His weight was estimated at 65 kg. He was treated for presumed pneumonia with moxifloxacin and transferred to our tertiary care hospital for computed tomography (CT) scan of the head, in order to rule out intracranial pathology. During transfer, paramedics assessed the patient and found him to be significantly obtunded, with a GCS of 7.
Upon arrival at our ED, the patient had a HR of 130 beats/min, a RR of 15--20 breaths/min, and a stable blood pressure. He had a temperature of 39 degrees\' Celsius and was diaphoretic and mottled to the chest. GCS remained at 7. Pupils were equal and reactive and the physical exam was otherwise unremarkable. Bedside glucose was obtained and was 5.3 mmol/L. A venous blood gas (VBG) and standard sepsis bloodwork (including blood cultures) were sent upon presentation. At this point, there was concern regarding the rapid decline in level of consciousness without a clear etiology. It was decided that an urgent CT scan of the head was indicated and since the patient would not obey commands and was not lying still, the decision was made to intubate him in order to facilitate the scan. The patient was induced for rapid sequence intubation (RSI) with propofol (50 mg) and rocuronium (50 mg). Following administration of propofol, the patient was ventilated using bag mask ventilation, in order to facilitate gas exchange during induction. He was successfully intubated using a GlideScope® on the first attempt. Nursing notes suggest that the longest possible apnea time was approximately 4 minutes, which reflects the time from rocuronium administration to detection of end-tidal carbon dioxide. A postintubation chest X-ray revealed a possible small consolidation in the left lower lung, but no evidence of pulmonary edema to suggest acute respiratory distress syndrome (ARDS).
Several minutes following intubation, the patient\'s preintubation bloodwork returned. VBG revealed a pH of 7.37, pCO2 of 19, HCO~3~ of 11, and venous base excess of −11.1 mmol/L. This mixed respiratory alkalosis and metabolic acidosis were suggestive for salicylate poisoning, and ventilator settings were adjusted to increase the respiratory rate (increased to 25, as compared to the patient\'s initial 20 breaths/minute) and tidal volume (increased to 620 mL, approximately 10 mL/kg of total body weight, as ideal body weight was unknown), which were the maximal tolerated settings. At this time, a toxicologic work-up was added to the bloodwork, and collateral history of ingestion was obtained, revealing access to 240 mg acetylsalicylic acid (ASA) boluses used for cattle. It was later confirmed that the patient had in fact intentionally overdosed on this ASA, though his family was not aware of this.
The patient was given two bolus ampules of IV NaHCO~3~ and started on a NaHCO~3~ infusion. The serum salicylate level was found to be critically elevated at 7.2 mmol/L. VBG performed twenty minutes following intubation revealed a pH of 6.89, pCO2 of 121, HCO~3~ of 23, and venous base excess of −13.2 mmol/L. A 50 g/L dextrose infusion was started to counter possible neuroglycopenia (which was presumed to be contributing to the decreased level of consciousness), and Nephrology and Critical Care were consulted for urgent hemodialysis. Despite the above interventions, a repeat VBG twenty minutes after the previous showed a pH of 7.03 and pCO2 \>120 (HCO~3~ and venous base excess were not reported). Soon after, the patient suddenly became very bradycardic and arrested. Initial rhythm was consistent with pulseless electrical activity (PEA). Chest compressions were started and the patient received bolus doses of 1 mg epinephrine and 50 mEq NaHCO~3~. Return of spontaneous circulation was briefly achieved, but subsequently lost again several minutes later. The patient again received multiple doses of epinephrine, NaHCO~3~, and 1 g calcium chloride. Bedside ultrasound confirmed absence of cardiac activity. The resuscitation was terminated after thirty minutes, at which point the patient\'s family requested to cease resuscitative efforts. CT of the head was ultimately not completed. The remainder of the toxicologic work-up (acetaminophen, ethanol, ECG) did not reveal any obvious coingestion.
3. Discussion {#sec3}
=============
Acute salicylate poisoning remains an important consideration in the undifferentiated patient with altered mental status, due to its subtle signs and significant mortality. While intubation of these patients may sometimes be necessary, it is commonly recommended against by several sources, due to the belief that it may stifle respiratory compensation and worsen acidemia. That being said, there are no experimental trials to support this belief, and only a few case reports exist (with varying results). In our literature review, we found one case report supporting the notion of worsening acidemia with intubation in severe salicylate overdose \[[@B4]\]. However, we also noted a case report of successful intubation in a patient with an intentional ASA overdose, though the serum concentration of salicylate was unknown \[[@B5]\]. The most comprehensive literature on this topic was a case-series published by the New York City Poison Control Centre, which was a retrospective review of 3,144 cases of salicylate poisoning from 2001 to 2007 \[[@B6]\]. In this 8-year time period, these authors found only 7 cases with available blood gas data of intubation in severe salicylate poisoning. In all of these cases, intubation resulted in acidemia, though the preintubation pH was not always known. Two of these 7 cases resulted in death. Furthermore, full blood gas values are not provided. Therefore, it is unclear how much of this acidemia is secondary to metabolic acidosis from salicylate metabolism and how much is secondary to respiratory acidosis from intubation and reduced respiratory drive. Finally, the time periods at which these pH values were collected are unknown, and evidence demonstrates that peak concentration of serum salicylate in acute overdose may not reach peak values until 6 hours following ingestion, due to bezoar formation \[[@B7], [@B8]\]. In our case, acidemia can be attributed to both respiratory acidosis (evidenced by hypercarbia) and likely metabolic acidosis from salicylate metabolism.
Several sources on management of acute salicylate toxicity also recommend that, if proceeding with intubation, the clinician should try to minimize the time of apnea and try to match the patient\'s minute ventilation with the ventilator \[[@B3], [@B6]\]. However, again there is no data available regarding either of these factors in intubation of these patients. In our case, the time from induction to airway security was less than four minutes, which should have minimized time of apnea. Further to that, the ventilator settings were adjusted in an attempt to maintain the patient\'s respiratory alkalosis. The patient\'s initial respiratory rate was 20, and this was increased to 25 on the ventilator. His tidal volume was unknown but was set at 620 mL (roughly 10 mL/kg of total body weight, as ideal body weight was unknown). There is no data available regarding the degree of hyperpnea seen in salicylate overdose. Evidence seems to indicate that normal tidal volume is roughly 6--8 mL/kg of ideal body weight \[[@B9]\], suggesting that this patient received tidal volumes and minute ventilation that should be consistent with hyperventilation. However, despite this, the patient continued to have worsening hypercapnia. This may support suggestions that mechanical ventilation is simply incapable of maintaining the necessary ventilation to prevent acidemia. It is also worth noting that our patient was paralyzed with rocuronium, which should have improved hyperventilatory effort from mechanical ventilation.
Uniquely, we are the first to demonstrate worsening acidemia despite ventilator settings that should have matched the patient\'s hyperventilation, lending credence to the notion that mechanical ventilation may be unable to provide the respiratory support required to maintain respiratory alkalosis in these patients. Ultimately, this case supports the notion that intubation should remain as a very last resort in the setting of acute salicylate toxicity.
4. Conclusion {#sec4}
=============
Severe salicylate toxicity is a life-threatening condition, and in the absence of an obvious history of ingestion, the diagnosis can only be made with a high degree of suspicion. Furthermore, the sequelae of toxicity on multiple organ systems suggest that the initial presentation can be quite varied. In any case of undifferentiated altered mental status, the instinct of the Emergency Physician is to first focus on the patient\'s ability to maintain their airway. This often results in endotracheal intubation. However, in the case of severe salicylate toxicity, intubation can result in worsening acidemia and cardiac arrest. Sometimes the only clue to the diagnosis is tachypnea and hyperpnea, and therefore all clinicians should be wary of the patient that appears to have rapid minute ventilation, as this patient may be attempting to maintain their neutral pH in the face of mounting acidosis. Further to this, should there be no choice but to intubate a patient in this condition, the clinician must do their best to maintain the large minute ventilation. However, as demonstrated in our case, that may not be feasible through mechanical ventilation.
Conflicts of Interest
=====================
The authors declare that they have no conflicts of interest.
[^1]: Academic Editor: Ricardo Oliveira
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Growth hormone (**GH**) deficiency (**GHD**) can be congenital or acquired. The incidence of congenital GHD has been assessed at from 1/4000 to 1/10 000 [@pone.0016367-Sizonenko1], [@pone.0016367-Vimpani1], [@pone.0016367-Bao1], [@pone.0016367-Lindsay1]. The pituitary stalk interruption syndrome (**PSIS**) is a sign of congenital and permanent GHD [@pone.0016367-Tauber1], [@pone.0016367-Marcu1], [@pone.0016367-Argyropoulou1]. It is diagnosed by magnetic resonance imaging (MRI) and includes the absence of both a visible pituitary stalk and normal posterior lobe hyperintense signals in the sella turcica, together with the presence of a hyperintense nodule in the region of the infundibular recess of the third ventricle. Familial forms of PSIS and associated malformations suggest that its origin is antenatal [@pone.0016367-Pinto1]. It is important to diagnose GHD and start treatment as soon as possible because this deficiency is associated with excess mortality and substantial morbidity [@pone.0016367-Taback1], [@pone.0016367-Mills1]. Moreover, because insufficient height at the onset of puberty leads to short final height, early diagnosis and treatment of GHD are necessary to allow catch-up growth to optimal height before puberty [@pone.0016367-Grumbach1]. Signs of congenital GHD in neonates include hypoglycemia, prolonged jaundice, and microphallus [@pone.0016367-Sizonenko1], [@pone.0016367-Pinto1], [@pone.0016367-Pinto2], [@pone.0016367-Rottembourg1]. In older children, the diagnosis is based on short stature or growth failure. Height for age is the most common criterion for referral for GH evaluation [@pone.0016367-Grote1]. However, the mean ages reported for diagnosis of symptomatic PSIS in various studies range from 4 to 9 years and suggest important diagnostic delay [@pone.0016367-Tauber1], [@pone.0016367-Argyropoulou1], [@pone.0016367-Pinto2], [@pone.0016367-Rottembourg1], [@pone.0016367-Maghnie1].
In 2000, the GH Research Society (**GHRS**) published guidelines based on height for age but also five other auxological criteria (see below), to ensure that children and adolescents with GHD are appropriately identified and treated [@pone.0016367-Consensus1]. A survey has shown that these criteria are not currently applied, probably because the concomitant use of six auxological criteria might be difficult in day-to-day routine practice [@pone.0016367-Grote1]. Moreover the performance (notably sensitivity for early diagnosis) of these guidelines has never been tested.
The objective of this study was therefore to study the diagnostic delay for PSIS with GHD and the sensitivity of the auxological criteria of the GHRS to identify the most useful ones and simplify their routine use.
Results {#s2}
=======
Characteristics of the population {#s2a}
---------------------------------
During the study period, 67 patients seen for growth failure had PSIS and/or GHD: 38 (57%) had GHD with a normal MRI or an isolated hypoplastic anterior pituitary gland, 2 (3%) had GHD and PSIS but had been adopted, six (9%) patients had GHD and PSIS diagnosed in the neonatal period. The study thus included 21 (31%) patients with GHD and PSIS ([Table 1](#pone-0016367-t001){ref-type="table"}), 76% of them boys. One patient was born preterm, and nine were delivered by cesareans (43%) (confidence interval, **CI** = 22--64), including three in breech presentation. One patient had midline abnormalities, including bilateral optic nerve hypoplasia.
10.1371/journal.pone.0016367.t001
###### Patient characteristics.
{#pone-0016367-t001-1}
Isolated GHD (n = 16) MPD (n = 5) TOTAL (n = 21)
----------------------- ----------------------- ------------------------------------- ---------------- ------------------------------------ ----- ------------------------------------
**Neonatal symptoms** n\' Percentage n\' Percentage n\' Percentage
Breech delivery 2 12.5% (CI 0--29) 1 20% (CI 0--55) 3 14% (CI 0--29)
Cesarean delivery 5 31% (CI 8--54) 4 80% (CI 45--100) 9 43% (CI 22--64)
**At diagnosis** Median (range) Median (range) Median (range)
Age (yr) 16 3.2 (1; 13.6) (IQR 2.6; 4.9) 5 5.1 (1; 10.5) (IQR 5; 5.6) 21 3.6 (1; 13.6) (IQR 2.6; 5.5)
Bone age (yr) 12 1.5 (0.5; 9.5) (IQR 1.2; 2.3) 4 2.2 (0.5; 4) (IQR 1.6; 2.9) 16 1.7 (0.5; 9.5) (IQR 1.2; 2.5)
Bone age delay (yr) 12 1.3 (0.5; 4.1) (IQR 1; 1.7) 4 2.8 (0.5; 6.4) (IQR 2; 3.9) 16 1.4 (0.5; 6.4) (IQR 1; 2.6)
Target height (SDS) 16 −0.2 (−1.6; 1.5) (IQR --0.7; 0.3) 5 −0.3 (−1.5; 0.6) (IQR --0.6; 0.4) 21 −0.3 (−1.6; 1.5) (IQR --0.6; 0.4)
Height (SDS) 16 −2.7 (−4.3; −1.3) (IQR --3.7; −2.3) 5 −2.2 (−2.4; −2) (IQR --2.2; −2) 21 −2.5 (−4.3; −1.3) (IQR --3.5; −2)
Height velocity (SDS) 16 −3 (−4.1; 0.3) (IQR --3.3; −1.6) 5 −3.3 (−4.2; 0) (IQR --3.4; −3.2) 21 −3.1 (−4.2; 0.3) (IQR --3.4; −1.6)
Weight (SDS) 16 −2.5 (−4; −0.4) (IQR --3; −1.9) 5 −0.7 (−1.3; 1.1) (IQR --1.2; −0.3) 21 −2.4 (−4; 1.1) (IQR --2.8; −1)
BMI (SDS) 16 −0.9 (−3.7; 2.2) (IQR --1.5; 0.2) 5 1.3 (−0.2; 4) (IQR --0.1; 1.7) 21 −0.23 (−3.7; 4) (IQR --1.1; 0.5)
GH peak (ng/mL) 16 3.2 (1.5; 23) (IQR 2; 6.7) 5 2.1 (0.5; 4.1) (IQR 0.9; 3.1) 21 3 (0.5; 23) (IQR 2; 5.5)
IGF-1 (ZS) 16 −2.9 (−5.1; −2) (IQR --4; −2.4) 5 −4.8 (−5; −4.1) (IQR --4.9; −4.4) 21 −3.1 (−5; −2) (IQR --4.4; −2.7)
CI: confidence interval 95%.
IQR: interquartile range.
GHD: growth hormone deficiency.
MPD: multiple pituitary deficiencies.
SDS: standard deviation score.
ZS: Z-score.
Median age at diagnosis was 3.6 years (range 1--13.6; interquartile range **IQR**: 2.6--5.5), and all patients were prepubertal ([Table 1](#pone-0016367-t001){ref-type="table"}). Sixteen patients (76%) (95%) (CI 58--94) had isolated GHD and five (24%) (CI 6--42) had multiple pituitary deficiencies (**MPD**) with thyroid stimulating hormone deficiency in four and adrenocorticotrophin deficiency in two. The median height was −2.5 SDS (range −4.3; −1.3) (IQR −3.5; −2) and median BMI −0.23 SDS (range −3.7; 4) (IQR −1.1; 0.5). The median height velocity was −3.1 SDS (range −4.2; 0.3) (IQR −3.4; −1.6).
Medical and growth history {#s2b}
--------------------------
Nine families (43%) (CI 22--64) first consulted a private-practice pediatrician about growth failure, and 12 families an outpatient pediatric department (57%) (CI 36--78). One family sought care directly from our team. Eleven patients (52%) (CI 31--73) had undergone laboratory testing for growth retardation before consulting our team, two had had a GH stimulation test, and 3 had had serum IGF-1 measured. Both GH stimulation tests were normal, and serum IGF-1 was less than --2 SDS, but no further diagnostic procedures were performed to rule out GHD. The patient with bilateral optic nerve hypoplasia had had neonatal hypoglycemia and microphallus but was not evaluated for GH secretion until the age of one year, and then for growth failure. His pediatrician had ordered an MRI at 2 months of age because his eyes were not yet following objects. At 5 months of age, his growth rate started to decrease and at one year of age, he was addressed to our department for growth retardation. The PSIS diagnosis was based on the MRI performed at 2 months of age.
No episodes of severe hypoglycemia or adrenal crisis were observed before diagnosis, and no child had any neurological deficiency.
Performance of GHRS criteria {#s2c}
----------------------------
[Table 2](#pone-0016367-t002){ref-type="table"} summarizes the performance of each GHRS criterion. The criterion of height more than 2 SDS below the mean + height velocity over 1 year more than 1 SDS below the mean for chronological age had a frequency at final diagnosis of 100%. Height more than 1.5 SDS below the target height was the most effective criterion: 90% of the patients had met the criterion before diagnosis, at a median age of 1 year (range 0; 9) (IQR 0.5; 1.8), and it was the first criterion to be met for 84% of the patients. Its use could have reduced diagnostic delay by 2.1 years (range 0; 12.6) (IQR 1.5; 2.9). The combined use of these two criteria, height more than 2 SDS below the mean + height velocity over 1 year more than 1 SDS below the mean for chronological age and height more than 1.5 SDS below the target height, might also have reduced diagnostic delay by 2.1 years (range 0; 12.6) (IQR 1.5; 3) for a median age at first validation of one of these criteria, that is, the first visit at which a doctor could have determined that the criterion had been met, was 1 year (range 0; 4.7) (IQR 0.6; 2).
10.1371/journal.pone.0016367.t002
###### Individual analysis of auxological GHRS criteria.
{#pone-0016367-t002-2}
Height \<−3 SDS Height \<−1,5 SDS below the target height Height \<−2 SDS and height velocity \<−1 SDS[\*](#nt107){ref-type="table-fn"} Height \<−2 SDS and height diminution \>0,5 SDS[\*\*](#nt108){ref-type="table-fn"} Normal height and height velocity \<−2 SDS[\*](#nt107){ref-type="table-fn"} Normal height and height velocity \<−1,5 SDS[\*\*\*](#nt109){ref-type="table-fn"} At least one of the 6 criterion
------------------------------------------------------------------------------------------------------------------ ------------------------ ------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------------------------------------ ----------------------------------------------------------------------------- ----------------------------------------------------------------------------------- ---------------------------------
Criterion completed at diagnosis n (%) (CI) 11 (52%) (31--73) 19 (90%) (77--100) 21 (100%) (100--100) 11 (52%) (31--73) 5 (24%) (6--42) 4 (19%) (2--36)
Age at criterion completion (yr) median (range) (IQR) 1 (0,6; 10) (0.7; 2.2) 1 (0; 9) (0.5; 1.8) 2 (1; 9) (1; 3.9) 3 (2; 6) (3; 4.3) 3 (2; 6) (3; 4) 3 (2; 4) (2.7; 3.2) 1 (0; 4) (0.6; 2)
Number of patients who completed the criterion first n (%) (CI) 2 (18%) (0--41) 16 (84%) (67--100) 4 (19%) (2--36) 0 (0%) (0--0) 2 (40%) (3--83) 1 (25%) (0--67)
Potential reduction of diagnostic delay among the patients who completed the criterion (yr) median (range) (IQR) 2 (0; 6.8) (0.1; 3.3) 2.1 (0; 12.6) (1.5; 2.9) 1.5 (0; 9.6) (0; 3) 0 (0; 1.5) (0; 0.3) 2 (0.6; 4.5) (0.7; 2.1) 2.7 (0.5; 6.5) (1.6; 4.2)
Potential reduction of diagnostic delay among all patients (yr) median (range) (IQR) 0 (0; 6.8) (0; 2) 2 (0; 12.6) (0.6, 2.8) 1.5 (0; 9.6) (0; 3) 0 (0; 1.5) (0; 0) 0 (0; 4.5) (0; 0) 0 (0; 6.5) (0; 0) 2.3 (0; 12.6) (1.5; 3.6)
\*over 1 year.
\*\*over 1 year in children older than 2 years of age.
\*\*\*over 2 years.
CI: confidence interval 95%.
IQR: interquartile range.
SDS: standard deviation score.
Late Diagnosis {#s2d}
--------------
Median age at diagnosis was 3.6 years (range 1; 13.6) (IQR 2.6; 5.5). Median age when the auxological criterion was met was 1 year (range 0; 4) (IQR 0.6; 2). The median diagnostic delay was 2.3 years (range 0; 12.6) (IQR 1.5; 3.6), with late diagnosis in 17 patients (81%).
Discussion {#s3}
==========
Main results {#s3a}
------------
We analyzed the diagnostic delay and sensitivity for the GHRS auxological criteria in the largest reported cohort of children seen for PSIS with GHD since the publication of these criteria. We studied the GHRS guidelines rather than other rules, such as the Dutch consensus guidelines or the UK guidelines, because it has been already demonstrated that both of these European guidelines lack specificity [@pone.0016367-Grote2], [@pone.0016367-vanBuuren1], [@pone.0016367-Grote3] or sensitivity [@pone.0016367-Grote2], [@pone.0016367-Grote3]. A Dutch team recently proposed another algorithm to identify children with short stature who require a diagnostic work-up, but this algorithm did not target PSIS with GHD as a key diagnosis [@pone.0016367-Oostdijk1]. We chose to study patients with GHD and PSIS because they comprise a homogeneous population with a permanent GHD, and because the real clinical significance of GHD without PSIS (diagnosed by a low GH response after 2 pharmacological stimulation tests and normal MRI) is a matter of debate today [@pone.0016367-Louvel1].
In all, 71% of patients had a diagnostic delay greater than 1 year. Correct application of the GHRS auxological criteria could have allowed diagnosis of these patients and the beginning of their treatment 2 years earlier. Of the GHRS criteria, the most effective for early and frequent diagnosis was height more than 1.5 SDS below the target height and the criterion met by all patients was height more than 2 SDS below the mean + height velocity over 1 year more than 1 SDS below the mean for chronological age. Height velocity and distance to target height have already been described by other teams as effective markers for detecting other growth disorders, such as Turner\'s syndrome, GHD and celiac disease [@pone.0016367-Grote2], [@pone.0016367-Grote4], [@pone.0016367-vanBuuren2].
Distance to target height and height velocity are still underused in routine practice [@pone.0016367-Grote1]. Interestingly, height velocity is not included in the UK consensus guideline [@pone.0016367-Grote3], [@pone.0016367-Hall1], [@pone.0016367-Fayter1] nor as a growth monitoring indicator in the national French pediatric health notebook. It is not provided by the World Health Organization (WHO) growth charts after 24 months [@pone.0016367-Grote3], not included in any study evaluating the effectiveness of height-screening programs [@pone.0016367-Fayter1], and was used by fewer than 50% of European pediatric endocrinologists in a 2002 survey [@pone.0016367-Grote1].
The specificity of each of the best criteria identified by our study (height more than 1.5 SDS below the target height, as well as height more than 2 SDS below the mean + height velocity over 1 year more than 1 SDS below the mean for chronological age) could not be determined but can be compared indirectly with those of other studies [@pone.0016367-Grote2], [@pone.0016367-Grote4], [@pone.0016367-vanBuuren2]. The specificity of the Dutch guidelines for short stature was tested on a longitudinal growth data of 870 children born in a geographical area of the Netherlands [@pone.0016367-vanBuuren1]. Of the six criteria of the Dutch guidelines, the criteria of height more than −1.3 SDS below the mean and of height more than −1.3 SDS below the target height, which are close to one of our best criteria, had a specificity of 94%.
Although it may be somewhat difficult to use all GHRS criteria in routine practice to detect growth anomalies, our results for patients with GHD and PSIS as well as results from a larger population [@pone.0016367-Grote4] indicate that distance to target height should be used routinely as a warning sign for growth anomalies to select the patients who require further investigation. It should replace height for age which is relatively insensitive for the detection of clinically relevant growth disorders [@pone.0016367-Grote3].
Our work shows that GH peak is not enough to rule out a diagnosis of GHD. Indeed, GHD had been ruled out for 2 (10%) of the 21 patients included during their medical care because of normal GH peaks, despite serum IGF-1 less than --2 SDS. This observation supports the current modification of the use of GH provocative tests in the evaluation of GHD [@pone.0016367-Louvel1], [@pone.0016367-Duche1], [@pone.0016367-Juul1], [@pone.0016367-Ghigo1]. Indeed, they are expensive, labor intensive, occasionally risky, and their results not very reproducible [@pone.0016367-Ghigo1], [@pone.0016367-Badaru1], [@pone.0016367-Lemaire1]. Their use has declined over the past two decades. Serum IGF-1, together with the growth rate, provides high quality diagnoses that are practical, simple and very accurate [@pone.0016367-Lemaire1]. Patients suspected for GHD, with a BMI between --2 and +2 SDS, with very low IGF-1 levels should skip GH provocative tests and should be prescribed a MRI [@pone.0016367-Badaru1], [@pone.0016367-Lemaire1].
Study limitations {#s3b}
-----------------
We used the national growth charts included in the French health notebook, developed in 1979 [@pone.0016367-Sempe1]. In 2006, the WHO multicentre growth reference study published growth charts for healthy breastfed infants living in good hygiene conditions [@pone.0016367-WHO1]. The comparison of the anthropometric measurements of French children with the new WHO growth standards showed similarities for the neonatal measurements but differed substantially thereafter, with French measurements (height, weight and BMI) lower from 1 to 6 months and French height lower but BMI higher from 6 months to 5 years old [@pone.0016367-Peneau1]. The GHRS consensus guidelines do not make it clear which growth charts should be used. Testing the sensitivity of GHRS criteria by using WHO growth charts is thus probably necessary. Our study was limited to a single center, a design that can result in recruitment bias. The presence of such a bias is supported by the mean age at diagnosis of symptomatic PSIS in our cohort, 3.6 years, compared to those reported in the literature, 4 to 9 years [@pone.0016367-Tauber1], [@pone.0016367-Argyropoulou1], [@pone.0016367-Pinto2], [@pone.0016367-Rottembourg1], [@pone.0016367-Maghnie1], [@pone.0016367-Louvel1]. It is thus possible that diagnostic delays are greater in the general population and that application of the GHRS criteria would reduce diagnostic delays still more than it would have in our patients.
Adoption and uncertain paternity are common, limiting utility of the "target height" criterion. That is the reason why it may be useful to consider the use of a combination of our two best criteria: height more than 1.5 SDS below the target height and height more than 2 SDS below the mean + height velocity over 1 year more than 1 SDS below the mean for chronological age.
Unexpected findings {#s3c}
-------------------
We were surprised by the high proportion (14%) of breech presentation vs 4% in the general population in France [@pone.0016367-Carayol1], and by the high proportion of cesareans (43%) vs 25% in the general population in France [@pone.0016367-LeRay1]. Of patients with MPD, 20% were born in breech presentation, and 80% (including all patients with thyrotrophic insufficiency) were born by cesarean delivery. If we incorporate in the analysis the six excluded patients with PSIS diagnosed during the neonatal period, 22% of patients had breech presentations and 56% cesarean births, for all six were born by cesarean deliveries, three in breech presentation. We were not able to identify a selection bias that could explain this unexpected finding. The frequency of breech presentation and cesarean delivery for GHD patients in the literature varies respectively from 7 to 60% and 30 to 40% [@pone.0016367-Hanew1], [@pone.0016367-Arrigo1], [@pone.0016367-AlbertssonWikland1]. TSH and/or ACTH deficiency and/or GHD may play a role in labor or fetal mobility and lead to breech presentation and/or cesarean delivery. Although we certainly do not recommend a pituitary MRI for all newborns by cesarean delivery or with breech presentations, clinicians should be aware of this finding in determining which newborns with hypoglycemia require a diagnostic workup for GHD.
Perspectives {#s3d}
------------
Screening rules based on growth monitoring are currently a topic of debate [@pone.0016367-vanBuuren1], [@pone.0016367-Grote3], [@pone.0016367-Oostdijk1]. Evidence-based strategies must be tested, both for their sensitivity for early diagnosis in case-cohort series of given target diseases (e.g., GHD, celiac disease, and Turner\'s syndrome) and for their specificity in healthy populations [@pone.0016367-vanBuuren2], [@pone.0016367-Hindmarsh1]. The introduction of some of the GHRS criteria (especially height more than 1.5 SDS below the target height and height more than 2 SDS below the mean + height velocity over 1 year more than 1 SDS below the mean for chronological age) would probably be helpful for the early diagnosis of the target disease here, PSIS with GHD. However, the precise specificity of these criteria and their performance for the early diagnosis of other target diseases involving growth monitoring must be tested.
Methods {#s4}
=======
Study design {#s4a}
------------
This single-center retrospective case-cohort study included all patients seen for PSIS with GHD by a senior pediatric endocrinologist (R Brauner) from January 2000 to December 2007. During the study period, the local routine protocol called for the systematic prescription of GH stimulation tests for all patients seen for growth failure and for systematic MRI of the hypothalamic-pituitary area of those with GHD (as defined below). All patients whose computerized hospital chart or discharge codes contained the words "growth hormone deficiency" and "pituitary stalk interruption syndrome" were considered for inclusion. The Institutional Review Committee (Comité de Protection des Personnes Ile de France III) stated that "this research was found to conform to generally accepted scientific principles and research ethical standards and to be in conformity with the laws and regulations of France, where the research experiment was performed." Written informed consent of the patients or their parents was not judged necessary for this kind of retrospective study. The data of some of the patients included in the present were previously used for other purposes [@pone.0016367-Duche1], [@pone.0016367-Lemaire1].
Inclusion criteria {#s4b}
------------------
We included all patients seen consecutively for GHD and PSIS. GHD was diagnosed by a GH peak of 10 ng/mL or less or 20 mIU/L or less after two pharmacological stimulation tests or a very low level of insulin-like growth factor (IGF)-1 (less than −2 standard deviation scores (**SDS**)) [@pone.0016367-Trivin1]. PSIS was diagnosed by MRI, according to the criteria described above. Patients with GHD but with a normal MRI or an isolated hypoplastic anterior pituitary gland were excluded, as well as adopted patients (because perinatal history and parental heights were not available). Patients with a diagnosis of PSIS in the neonatal period were also excluded because their growth rate before diagnosis could not be calculated.
Collected data {#s4c}
--------------
Social, demographic, and medical data were extracted from the medical report: sex, parental height, and perinatal history. Signs observed before diagnosis and medical and growth records were noted. Data related to the GHRS clinical and auxological criteria were also extracted. During the neonatal period, these criteria are hypoglycemia, prolonged jaundice, microphallus, or traumatic delivery. In the post-neonatal period, they include severely short stature, defined as a height more than 3 SDS below the mean; height more than 1.5 SDS below the target height; height more than 2 SDS below the mean and a height velocity during the previous year more than 1 SDS below the mean for chronological age, or a decrease in height SDS of more than 0.5 over 1 year in children older than 2 years; in the absence of short stature, a height velocity more than 2 SDS below the mean over 1 year or more than −1.5 SD over 2 years [@pone.0016367-Consensus1].
Definitions {#s4d}
-----------
Target height was calculated from parental height [@pone.0016367-Tanner1] and expressed in SDS. Microphallus was defined as a penis length of 2.5 cm or less (−2 SDS) [@pone.0016367-Pinto2]. Height, weight, body mass index (**BMI**, weight in kg/height in m^2^) and height velocity were expressed in SDS for chronological age [@pone.0016367-Sempe1], [@pone.0016367-RollandCachera1]. Bone age was evaluated by one of us (RB) according to the Greulich and Pyle method [@pone.0016367-Greulich1]. Bone age delay was defined as the difference in years between chronological and bone ages. Thyroid stimulating hormone deficiency was defined by thyroxin level less than 12 pmol/L and adrenocorticotrophin deficiency by basal blood cortisol at 08.00 h less than 70 µg/L.
Analysis {#s4e}
--------
We first analyzed population characteristics at diagnosis of PSIS with GHD and then studied the medical and growth history, symptoms, and clinical signs through diagnosis. Comparison of each GHRS auxological criterion with growth charts allowed us to establish the age at which each criterion was met, to class each criterion in chronological order of fulfillment, and then to evaluate the diagnostic delay, defined as the difference between the age at which the earliest GHRS criterion was met and the age at diagnosis of PSIS with GHD. We arbitrarily considered a diagnostic delay of one year or more as late diagnosis. Finally, we analyzed each GHRS criterion for how early and with what frequency it was met and arbitrarily defined the most effective criterion as the one that was most sensitive and earliest.
**Competing Interests:**The authors have declared that no competing interests exist.
**Funding:**The authors have no support or funding to report.
[^1]: Conceived and designed the experiments: GGL RB MC. Analyzed the data: GGL RB MC. Wrote the paper: GGL RB MC. Collected data: GGL RB LD.
| {
"pile_set_name": "PubMed Central"
} |
HTLV-1 infection of humanized NOG mice has been demonstrated to recapitulate the development of ATL-like symptoms within several months of infection. Infected human T-cells in these mice start to proliferate vigorously in a couple weeks after infection and the mice die of ALT-like lymphoproliferative disorder. Thus, this mouse model should provide a potent tool to analyze the *in vivo* effect of various candidates for ATL treatment.
Treatment of ATL with the combination of anti-viral agents, zidovudine (AZT) and interferon-alpha (IFN), has been reported to be highly effective, especially to indolent type, but the mechanism of action is totally unknown. We, therefore, examined the efficacy and the *in vivo* mechanism of AZT/IFN treatment in the humanized mouse system.
HTLV-1 infected humanized mice were inoculated daily with AZT and IFN from two to four weeks post infection and the number of infected cells and proviral loads (PVL) were analyzed. Treatment with either AZT or IFN alone attenuated the onset of lymphoproliferative disorder, whereas the combined treatment suppressed the growth of infected T-cells in PBL almost completely and the PVL remained low throughout lifetime. The suppressive effect is infected-cell specific because the number of uninfected human lymphocytes in PBL stayed constant on the administration of drugs.
It is suggested that infected cells expressing higher level of viral gene, most provably Tax, should have been selectively eliminated, since a similar suppressive effect has been obtained in HTLV-1 infected humanized mice treated with an Hsp90 inhibitor, 17-DMAG, which enhances the degradation of Tax.
| {
"pile_set_name": "PubMed Central"
} |
Tse G, Gong M, Li CKH, et al. T~peak~‐T~end~, T~peak~‐T~end~/QT ratio and T~peak~‐T~end~ dispersion for risk stratification in Brugada Syndrome: A systematic review and meta‐analysis. J Arrhythmia. 2018;34:587--597. 10.1002/joa3.12118
1. INTRODUCTION {#joa312118-sec-0005}
===============
Brugada syndrome is a used to describe the combination of specific ECG changes, the Brugada pattern, in addition to life threatening arrhythmias and sudden cardiac death (SCD).[1](#joa312118-bib-0001){ref-type="ref"} Traditionally, it has been considered a congenital ion channelopathy linked to abnormalities in the cardiac sodium channel.[2](#joa312118-bib-0002){ref-type="ref"}, [3](#joa312118-bib-0003){ref-type="ref"} Recently, pathogenic mutations in other ion channels have been described. Mechanisms of arrhythmogenesis can be broadly divided into triggered activity and re‐entry. Of these, re‐entry is thought to be the predominant mechanism underlying increased arrhythmogenicity in Brugada syndrome requiring an increased spatial dispersion of repolarization. Such re‐entrant activity may involve direct electrotonic activation during phase 2 of the cardiac action potential, as shown in pre‐clinical studies using arterially perfused, canine wedge preparations,[4](#joa312118-bib-0004){ref-type="ref"} or circus‐type/spiral wave activity around an anatomical or functional obstacle. Regardless of the precise underlying mechanism for re‐entry, this transmural dispersion of repolarization can be quantified electrocardiographically by the interval from the peak to the end of the T‐wave (T~peak~‐T~end~ interval), (T~peak~‐T~end~)/QT ratio and T~peak~‐T~end~ dispersion.[5](#joa312118-bib-0005){ref-type="ref"}, [6](#joa312118-bib-0006){ref-type="ref"}
However, not all studies have shown an association between higher T~peak~‐T~end~ intervals, (T~peak~‐T~end~)/QT ratio or T~peak~‐T~end~ dispersion with an arrhythmogenic phenotype in Brugada Syndrome. Recently, Mugnai and colleagues conducted one of the largest retrospective studies to date, including a total of 448 patients with spontaneous or drug induced type 1 Brugada pattern.[7](#joa312118-bib-0007){ref-type="ref"} They found no statistically significant difference in all three indices between asymptomatic subjects and patients with syncope and malignant arrhythmias. Morita and colleagues also found in 471 patients no difference in T~peak~‐T~end~ intervals between patients with syncope or VT/VF and those who were asymptomatic.[8](#joa312118-bib-0008){ref-type="ref"} These findings contrast with a meta‐analysis published previously by some members of our group, which extracted and pooled odds or hazard ratios for the relationship between T~peak~‐T~end~ and arrhythmic and/or mortality outcomes in various clinical conditions, including Brugada Syndrome.[9](#joa312118-bib-0009){ref-type="ref"} This demonstrated prolonged T~peak~‐T~end~ interval was associated with an increased risk of ventricular arrhythmias and SCD in Brugada Syndrome.
However, our previous study did not determine the absolute mean values for T~peak~‐T~end~, nor was it possible to include the largest dataset from Mugnai and colleagues. Moreover, it did not investigate the utility of other indices such as (T~peak~‐T~end~)/QT ratio or T~peak~‐T~end~ dispersion. Therefore, we conducted a systematic review with meta‐analysis into the relationships between T~peak~‐T~end~ interval, (T~peak~‐T~end~)/QT ratio and T~peak~‐T~end~ dispersion and arrhythmic and/or mortality endpoints in Brugada Syndrome.
2. METHODS {#joa312118-sec-0006}
==========
2.1. Search strategy, inclusion and exclusion criteria {#joa312118-sec-0007}
------------------------------------------------------
This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISM) statement. PubMed and Embase were searched for studies that investigated the association between T~peak~‐T~end~ or T~peak~‐T~end~ /QT with arrhythmic or mortality endpoints in Brugada syndrome. The following search terms were used for both databases: \["Tpeak‐Tend" or "Tpeak‐end" or "Tp‐e" AND Brugada\]. The databases were searched until 1 May 2018 without language restrictions. The following inclusion criteria were used: (a) the study was a case‐control, prospective or retrospective cohort study in human subjects with a Brugada phenotype, (b) T~peak~‐T~end~ intervals or (T~peak~‐T~end~) /QT ratios were provided; (c) predefined adverse events (appropriate implantable cardioverter‐defibrillator therapy \[ICD\], syncope, ventricular tachycardia/fibrillation \[VT/VF\], SCD, cardiovascular death \[CVD\], major adverse cardiac events \[MACE\]) or all‐cause mortality were reported. In cases of incomplete data from the published studies, the original authors were contacted, but no replies were received.
The Newcastle‐Ottawa Quality Assessment Scale (NOS) was used for quality assessment of the included studies.[10](#joa312118-bib-0010){ref-type="ref"} The NOS system evaluated the categories of study participant selection, results comparability, and quality of the outcomes. Specifically, the following characteristics were assessed: (a) representativeness of the exposed cohort; (b) selection of the non‐exposed cohort; (c) ascertainment of exposure; (d) demonstration that outcome of interest was not present at the start of study; (e) comparability of cohorts based on study design or analysis; (f) assessment of outcomes; (g) follow‐up periods that were sufficiently long for outcomes to occur; and (h) adequacy of follow‐up of cohorts. This scale varied from zero to nine stars, which indicated that studies were graded as poor quality if the score was \<5, fair if the score was 5‐7, and good if the score was \>8. Studies with a score equal to or higher than six were included. The details of the NOS quality assessment are shown in Tables [S1](#joa312118-sup-0001){ref-type="supplementary-material"} and [S2](#joa312118-sup-0001){ref-type="supplementary-material"}.
2.2. Data extraction and statistical analysis {#joa312118-sec-0008}
---------------------------------------------
Data from the different studies were entered in pre‐specified spreadsheets in Microsoft Excel. All potentially relevant studies were retrieved as complete manuscripts, which were assessed fully to determine their compliance with the inclusion criteria. We extracted the following data from the included studies: (a) publication details: last name of first author, publication year and locations; (b) study design; (c) endpoint(s); (d) quality score; and (e) characteristics of the population including sample size, gender, age and number of subjects. Two reviewers (GT and MG) reviewed each included study independently. Disagreements were resolved by adjudication with input from a third reviewer (TL).
Adverse events were defined as ventricular arrhythmias (VT/VF), SCD, cardiovascular death, MACE or all‐cause mortality. If more than one mortality endpoint was described, then SCD was preferentially used for analysis, followed by cardiovascular and all‐cause mortality in this order. Mean differences between event‐positive and event‐negative groups, with 95% confidence intervals (CIs) for T~peak~‐T~end~ interval, (T~peak~‐T~end~)/QT ratio and T~peak~‐T~end~ dispersion were extracted and subsequently combined to generate a pooled estimate.
Heterogeneity between studies was quantified using The Cochran\'s Q value and the *I* ^2^ statistic from the standard chi‐square test, which describes the percentage of the variability in effect estimates resulting from heterogeneity. *I* ^2^ \> 50% was considered to reflect significant statistical heterogeneity. A fixed effects model was used if *I* ^2^ \< 50%. The random‐effect model using the inverse variance heterogeneity method was used when *I* ^2^ \> 50%. To locate the origin of the heterogeneity, sensitivity analysis by excluding one study at a time, and subgroup analyses based on different disease conditions and different endpoints were performed. Funnel plots, Begg and Mazumdar rank correlation test and Egger\'s test were used to detect publication bias.
3. RESULTS {#joa312118-sec-0009}
==========
Figure [1](#joa312118-fig-0001){ref-type="fig"} shows a flow diagram detailing the above search terms with inclusion and exclusion criteria. A total of 29 and 57 entries were retrieved from PubMed and Embase, respectively. Nine studies met the inclusion criteria and were included in our final meta‐analysis.[6](#joa312118-bib-0006){ref-type="ref"}, [7](#joa312118-bib-0007){ref-type="ref"}, [11](#joa312118-bib-0011){ref-type="ref"}, [12](#joa312118-bib-0012){ref-type="ref"}, [13](#joa312118-bib-0013){ref-type="ref"}, [14](#joa312118-bib-0014){ref-type="ref"}, [15](#joa312118-bib-0015){ref-type="ref"}, [16](#joa312118-bib-0016){ref-type="ref"}, [17](#joa312118-bib-0017){ref-type="ref"} In this meta‐analysis, a total of 1740 subjects with Brugada Syndrome were included (mean age 45 years old, 80% male). The mean follow‐up duration was 68 ± 27 months. Of the entire cohort, 40% had a spontaneous Type 1 pattern and 19% were positive for SCN5a mutation. The baseline characteristics of these studies and of the study populations are shown in Table [1](#joa312118-tbl-0001){ref-type="table"}.
{#joa312118-fig-0001}
######
Characteristics of the nine studies included in this meta‐analysis
First author/year Sample size (n) T~peak~‐T~end~ measurement: method and leads Age (SD) No. of males (%) No. of Sp. type 1 patients (%) No. of SCN5a positive patients (%) Endpoints Comparisons No. of patients with adverse events /without adverse events/%/% per year Follow‐up duration (months) Quality score Ref.
------------------- ----------------- ---------------------------------------------- ---------- ------------------ -------------------------------- ------------------------------------ ----------------------- ------------------------------------------------ -------------------------------------------------------------------------- ----------------------------- --------------- -------------------------------------------
Morita 2017 471 Tangent method; V1, V2, V3, V5 47 (19) 447 (95) 118 (25) 27 (15) Syncope or VT/VF Syncope/VT/VF vs asymptomatic 145/326/31/4.09 91 7 [16](#joa312118-bib-0016){ref-type="ref"}
Mugnai 2017 448 End of the T‐wave; V1 to V6 45 (16) 273 (61) 96 (21) 55 (22) Spontaneous VF or SCD AT/SD vs asymptomatic 43/290/13/1.67 93 6 [7](#joa312118-bib-0007){ref-type="ref"}
Kawazoe 2016 143 Tangent method; V1 to V6 46 (12) 140 (98) 84 (59) -- VF VF vs no VF 35/108/24/1.9 105 7 [17](#joa312118-bib-0017){ref-type="ref"}
Zumhagen 2016 78 Tangent method; V1 45 (14) 57 (73) 22 (28) 17 (22) Spontaneous VT/VF VT/VF/aborted SCD vs asymptomatic/syncope 22/54/29/‐ -- 6 [14](#joa312118-bib-0014){ref-type="ref"}
Maury 2015 325 Tangent method; V1 to V4 47 (13) 260 (80) 143 (44) 43 (13) Spontaneous VT/VF AT/SD vs asymptomatic 26/226/10/2.50 48 7 [11](#joa312118-bib-0011){ref-type="ref"}
Letsas 2010 23 End of the T‐wave; V2, V6 43 (15) 19 (83) 10 (43) -- Inducible VT/VF Inducible VT vs no inducible VT 17/6/74/16.15 55 6 [12](#joa312118-bib-0012){ref-type="ref"}
Junttila 2008 200 End of the T‐wave; V2, II 40 (16) 143 (72) 200 (100) 25 (50) Syncope, VT/VF, SCD Syncope/VT/VF/aborted SCD vs asymptomatic 66/134/33/‐ -- 7 [15](#joa312118-bib-0015){ref-type="ref"}
Wang 2007 23 End of the T‐wave; Max from V1 to V6 45 (8) 23 (100) -- -- Spontaneous VT/VF Syncope/VT/VF/inducible VT vs asymptomatic 11/9/55/5.12 43 8 [13](#joa312118-bib-0013){ref-type="ref"}
Castro Hevia 2006 29 Tangent method, Max from V1 to V6 41 (12) 25 (86) 15 (52) -- Spontaneous VT/VF Presyncope/syncope/aborted SCD vs asymptomatic 12/17/41/3.81 43 8 [6](#joa312118-bib-0006){ref-type="ref"}
SCD: sudden cardiac death; VT: ventricular tachycardia; VF: ventricular fibrillation; Sp.: spontaneous.
John Wiley & Sons, Ltd
3.1. T~peak~‐T~end~ {#joa312118-sec-0010}
-------------------
For determining T~end~, the tangent method and the return of the voltage to baseline method were used. T~peak~‐T~end~ intervals from different leads and the maximum of these measurements have been presented by most studies. Regarding maximum T~peak~‐T~end~ intervals, the mean value for the event‐positive group was 98.9 ms (95% CI: 90.5‐107.2 ms) (Figure [2](#joa312118-fig-0002){ref-type="fig"}A) and event‐negative group was 87.7 ms (95% CI: 80.5‐94.9 ms) (Figure [2](#joa312118-fig-0002){ref-type="fig"}B). Five studies reported longer values in the event‐positive compared to event‐negative groups, whereas four studies reported no significant difference (Figure [2](#joa312118-fig-0002){ref-type="fig"}C). T~peak~‐T~end~ intervals were 11.9 ms longer (95% CI: 3.6‐20.2 ms, *P* = 0.005) in event‐positive patients than in event‐negative patients. The Cochran\'s Q value was greater than the degrees of freedom (56 vs 8), indicating that the true effect size was different between studies. *I* ^2^ took a value of 86%, suggesting the presence of substantial heterogeneity. A funnel plot plotting standard errors against differences in means is shown in Figure [S1](#joa312118-sup-0001){ref-type="supplementary-material"}. Begg and Mazumdar rank correlation analysis demonstrated that Kendall\'s Tau took a value of 0.3 with *P* = 0.30, which suggests no significant publication bias. Egger\'s test demonstrated no significant asymmetry (intercept 2.4, *t*‐value 1.2; *P* = 0.25). To identify the source of the heterogeneity, sensitivity analysis was performed by removing one study at a time, but this did not significantly influence the mean difference (Figure [S2](#joa312118-sup-0001){ref-type="supplementary-material"}), suggesting that no single study was responsible for the heterogeneity observed in this meta‐analysis. Subgroup analysis based on the method of T~end~ determination was performed. For the tangent method, the T~peak~‐T~end~ mean difference was 15.5 ms (95% CI: 3.9‐27.2 ms; *P* = 0.009) and *I* ^2^ remained high at 90%. For full recovery of voltage to baseline, the mean difference was 6.0 ms (95% CI: 0.7‐11.4 ms; *P* = 0.006) and *I* ^2^ remained high at 76%. Therefore, different methods of T~end~ determination did not introduce significant heterogeneity to the pooled effect estimate.
{#joa312118-fig-0002}
3.2. (T~peak~‐T~end~)/QT ratio {#joa312118-sec-0011}
------------------------------
Regarding maximum (T~peak~‐T~end~)/QT ratio, the mean value for the event‐positive group was 0.221 (95% CI: 0.208‐0.234) (Figure [3](#joa312118-fig-0003){ref-type="fig"}A) and event‐negative group was 0.210 (95% CI: 0.205‐0.214) (Figure [3](#joa312118-fig-0003){ref-type="fig"}B). Two studies reported higher values in Brugada subjects with positive events compared to those without such events, whereas four studies demonstrated no significance between the groups (Figure [3](#joa312118-fig-0003){ref-type="fig"}C). Pooling of the mean values demonstrated significantly higher (T~peak~‐T~end~)/QT ratios in the event‐positive group than in the event‐negative group (mean difference = 0.019, 95% CI: 0.003‐0.036, *P* = 0.024). The Cochran\'s Q value was greater than the degrees of freedom (19 vs 5), indicating that the true effect size was different between studies. *I* ^2^ took a value of 74%, suggesting significant heterogeneity. A funnel plot plotting standard errors against differences in means is shown in Figure [S3](#joa312118-sup-0001){ref-type="supplementary-material"}. Begg and Mazumdar rank correlation analysis demonstrated that Kendall\'s Tau took a value of 0.07 with *P* = 1, which suggested no significant publication bias. Egger\'s test demonstrated no significant asymmetry (intercept 3.5, *t*‐value 1.1; *P* = 0.31). To identify the source of the heterogeneity, sensitivity analysis was performed by removing one study at a time, but this did not significantly influence the mean difference (Figure [S4](#joa312118-sup-0001){ref-type="supplementary-material"}), suggesting that no single study was responsible for the heterogeneity observed in this meta‐analysis. Subgroup analysis based on the method of T~end~ determination was performed. For the tangent method, the mean difference of (T~peak~‐T~end~)/QT ratio was 0.03 (95% CI: 0.01‐0.05; *P* \< 0.05) and *I* ^2^ was lowered to 55%. For full recovery of voltage to baseline, the mean difference was only 0.004 (95% CI: −0.03 to 0.03 ms; *P* = 0.81) and *I* ^2^ remained high at 74%. Therefore, different method of T~end~ determination appeared to contribute partially to the heterogeneity of the pooled effect estimate. Moreover, statistical significance was achieved when the tangent method was used, but was lost when the return to baseline method was used, which may suggest the former approach may be more sensitive.
{#joa312118-fig-0003}
3.3. T~peak~‐T~end~ dispersion {#joa312118-sec-0012}
------------------------------
Regarding maximum T~peak~‐T~end~ dispersion, the mean value for the event‐positive group was 40.8 ms (95% CI: 26.9‐54.8 ms) (Figure [4](#joa312118-fig-0004){ref-type="fig"}A) and event‐negative group was 29.7 ms (95% CI: 24.5‐34.8 ms) (Figure [4](#joa312118-fig-0004){ref-type="fig"}B). Regarding T~peak~‐T~end~ dispersion, two studies reported longer values in event‐positive group compared to event‐negative groups, whereas three studies found no significant difference (Figure [4](#joa312118-fig-0004){ref-type="fig"}C). Overall, pooling of the data showed that T~peak~‐T~end~ dispersion was significantly higher in the event‐positive than in the event‐negative groups (mean difference = 7.8 ms, 95% CI: 2.1 to 13.4 ms, *P* = 0.007). The Cochran\'s Q value was greater than the degrees of freedom (20 vs 4), indicating that the true effect size was different between studies. *I* ^2^ took a value of 80%, suggesting significant heterogeneity. A funnel plot plotting standard errors against differences in means is shown in Figure [S5](#joa312118-sup-0001){ref-type="supplementary-material"}. Begg and Mazumdar rank correlation analysis demonstrated that Kendall\'s Tau took a value of −2 with *P* = 0.62, which suggests no significant publication bias. Egger\'s test demonstrated no significant asymmetry (intercept −5.4, *t*‐value 0.8; *P* = 0.48). To identify the source of the heterogeneity, sensitivity analysis was performed by removing one study at a time, but this did not significantly influence the mean difference between event‐positive and event‐negative groups (Figure [S6](#joa312118-sup-0001){ref-type="supplementary-material"}), suggesting that no single study was responsible for the heterogeneity observed in this meta‐analysis. Subgroup analysis based on the method of T~end~ determination was performed. For the tangent method, the mean difference of T~peak~‐T~end~ dispersion was 16.2 ms (95% CI: 7.9‐24.5 ms; *P* \< 0.0001) and *I* ^2^ was 65%. For full recovery of voltage to baseline, the mean difference was 0.4 ms (95% CI: −7.3 to 8.2 ms; *P* = 0.91) and *I* ^2^ was reduced to 19%. Therefore, different method of T~end~ determination contributed heterogeneity to the pooled effect estimate. Moreover, statistical significance was achieved when the tangent method was used, but was lost when the return to baseline method was used, which may suggest the former approach may be more sensitive.
{#joa312118-fig-0004}
3.4. Comparisons between patients with and without SCN5A mutations {#joa312118-sec-0013}
------------------------------------------------------------------
SCN5A is the commonest ion channel gene that is mutated in Brugada syndrome.[2](#joa312118-bib-0002){ref-type="ref"}, [3](#joa312118-bib-0003){ref-type="ref"} Separate meta‐analyses were conducted to compare the different T~peak~‐T~end~ parameters between patients with and without SCN5A mutations. Two of the included studies provided sufficient information for such analyses.[7](#joa312118-bib-0007){ref-type="ref"}, [14](#joa312118-bib-0014){ref-type="ref"} No significant difference in T~peak~‐T~end~ (mean difference = 8.2 ms, 95% CI: −6.7 to 23.2 ms, *P* = 0.28; *I* ^2^ = 59%; Figure [S7](#joa312118-sup-0001){ref-type="supplementary-material"}), T~peak~‐T~end~/QT ratio (mean difference = −0.006 ms, 95% CI: −0.023 to 0.011 ms, *P* = 0.47; *I* ^2^ = 24%; Figure [S8](#joa312118-sup-0001){ref-type="supplementary-material"}) or T~peak~‐T~end~ dispersion (mean difference = 5.2 ms, 95% CI: −2.9 to 13.2 ms, *P* = 0.21; *I* ^2^ = 31%; Figure [S9](#joa312118-sup-0001){ref-type="supplementary-material"}) was observed between patients with and without SCN5A mutations.
4. DISCUSSION {#joa312118-sec-0014}
=============
The main findings of our meta‐analysis, which included 1597 Brugada subjects, are (a) T~peak~‐T~end~ intervals, (b) (T~peak~‐T~end~)/QT ratio and (c) T~peak~‐T~end~ dispersion are higher in Brugada subjects with adverse cardiac events (ventricular tachy‐arrhythmias and SCD) when compared to Brugada subjects free from such events.
The presence of pre‐existing electrophysiological heterogeneities is important for mediating the normal, unidirectional spread of action potentials in the heart.[18](#joa312118-bib-0018){ref-type="ref"}, [19](#joa312118-bib-0019){ref-type="ref"} These are attributed to differences in repolarization times of the different cell types, which are responsible for generation of the T‐wave on the electrocardiogram (ECG).[20](#joa312118-bib-0020){ref-type="ref"}, [21](#joa312118-bib-0021){ref-type="ref"} However, exacerbation of such differences has been associated with ventricular tachy‐arrhythmias in different conditions, thereby generating a pro‐arrhythmic phenotype. These include congenital ion channelopathies such as long QT syndrome and Brugada syndrome[22](#joa312118-bib-0022){ref-type="ref"}, [23](#joa312118-bib-0023){ref-type="ref"}, [24](#joa312118-bib-0024){ref-type="ref"} and acquired cardiac diseases such as myocardial infarction.[25](#joa312118-bib-0025){ref-type="ref"}, [26](#joa312118-bib-0026){ref-type="ref"} These heterogeneities can occur locally or across the myocardial wall,[27](#joa312118-bib-0027){ref-type="ref"} potentially causing arrhythmias by inducing unidirectional conduction block and therefore circus‐type or spiral wave re‐entry.[28](#joa312118-bib-0028){ref-type="ref"}, [29](#joa312118-bib-0029){ref-type="ref"} Moreover, a greater epicardial‐endocardial repolarization time difference may increase the propensity of phase 2 re‐entry, which is hypothesized to generate extrasystolic activity in Brugada syndrome.[30](#joa312118-bib-0030){ref-type="ref"} This occurs when sites with an action potential dome to sites which a dome morphology, leading to direct depolarization of the downstream sites.[31](#joa312118-bib-0031){ref-type="ref"} Once an extrasystole is generated, together with a favorable re‐entrant substrate, ventricular tachycardia and fibrillation can result.[32](#joa312118-bib-0032){ref-type="ref"}
A number of electrocardiographic indices have been proposed for stratification of arrhythmic or mortality risk.[33](#joa312118-bib-0033){ref-type="ref"}, [34](#joa312118-bib-0034){ref-type="ref"} Of these, Yan and Antzelevitch were the first to propose the use of the difference between the peak and the end of the T‐wave (the T~peak~‐T~end~ interval) as a measure of transmural dispersion of repolarization.[20](#joa312118-bib-0020){ref-type="ref"}, [35](#joa312118-bib-0035){ref-type="ref"}, [36](#joa312118-bib-0036){ref-type="ref"}, [37](#joa312118-bib-0037){ref-type="ref"} Subsequent clinical studies have demonstrated that, confirmed recently in a systematic review and meta‐analysis from our group,[9](#joa312118-bib-0009){ref-type="ref"} that T~peak~‐T~end~ prolongation significantly elevated the risk of ventricular tachy‐arrhythmias and/or SCD in heart failure, ischemic heart disease, Brugada syndrome, hypertension, and the general population. Recently, Mugnai and colleagues in a total of 448 subjects found no significant differences T~peak~‐T~end~ intervals, (T~peak~‐T~end~)/QT ratio or T~peak~‐T~end~ dispersion between patients with VT/VF requiring anti‐tachycardia pacing or with sudden death, and those who were asymptomatic.[7](#joa312118-bib-0007){ref-type="ref"} Similarly, in a separate population of 471 subjects, Morita and colleagues found no significance difference in T~peak~‐T~end~ intervals between patients with syncope or VT/VF and asymptomatic patients.[16](#joa312118-bib-0016){ref-type="ref"} Publication of these two studies prompted us to conduct this meta‐analysis, which confirms the value of T~peak~‐T~end~ interval, (T~peak~‐T~end~)/QT ratio and T~peak~‐T~end~ dispersion, in distinguishing high‐risk patients from low‐risk patients.
In the Mugnai study, the largest study to date, the percentage of patients with adverse events were the lowest at 13%.[7](#joa312118-bib-0007){ref-type="ref"} Male gender, a spontaneous Type 1 Brugada pattern and SCN5a mutation positive status were significantly associated with ventricular arrhythmias.[38](#joa312118-bib-0038){ref-type="ref"} Therefore, the lower percentage of patients with adverse events can be explained by the lower percentage of Type 1 Brugada patients (21% vs 28%‐100% in the remaining studies) and lower percentage male patients (61% vs 72%‐100%) despite similar percentage with SCN5a positive status (22% vs 13%‐50%). While these differences in patient characteristics affect the likelihood of adverse events occurring, they should not explain the lack of difference in T~peak~‐T~end~ intervals between event‐positive and event‐negative groups in the Morita study[16](#joa312118-bib-0016){ref-type="ref"} or the Mugnai study. Interestingly, Mugnai and colleagues found a non‐statistically significant lower T~peak~‐T~end~ intervals in event‐positive groups. Of the remaining six studies, five studies had reported significantly higher T~peak~‐T~end~ intervals and one study reported no difference.[15](#joa312118-bib-0015){ref-type="ref"} A recent epidemiological study reported a U‐shaped relationship between T~peak~‐T~end~ intervals and increased mortality.[39](#joa312118-bib-0039){ref-type="ref"} Autonomic modulation, which is part of Coumel\'s triad for arrhythmogenesis,[40](#joa312118-bib-0040){ref-type="ref"} is known to modulate the re‐entrant substrate. Increased activity of the parasympathetic nervous system may reduce T~peak~‐T~end~ intervals, which may also be pro‐arrhythmic.[41](#joa312118-bib-0041){ref-type="ref"} By contrast, exercise, during which sympathetic activity is increased, can exacerbate pre‐existing heterogeneities, such as producing conduction slowing[42](#joa312118-bib-0042){ref-type="ref"} and increasing the dispersion of repolarization.[43](#joa312118-bib-0043){ref-type="ref"}
In our previous meta‐analysis pooling together studies that reported odds ratios or hazard ratios, the average cut‐off for T~peak~‐T~end~ was 95.8 ms across different clinical conditions.[9](#joa312118-bib-0009){ref-type="ref"} The present meta‐analysis pooling mean values for event‐positive and ‐negative groups clearly indicates that the 100 ms cut‐off is too high for Brugada syndrome. Our data would support a lower cut‐off value between 88 and 99 ms to be used. This cut‐off will also be method‐dependent for determining T~end~ in the case of the T~peak~‐T~end~ intervals. Previously, it was shown that in a cohort of high‐risk Brugada subjects, only 10 of 16 studies reported a T~peak~‐T~end~ longer than 100 ms, supporting our notion that this cut‐off value may be too high.[44](#joa312118-bib-0044){ref-type="ref"} Moreover, different studies measured T~peak~‐T~end~ from different leads. Some had measured it from all 12 leads and taken the mean values while others have done so for V1 to V3 only. While there is no consensus as to which leads are most appropriate for measurement, obtaining it from all 12 leads is likely to be less useful clinically due to the time‐consuming nature. To simplify T~peak~‐T~end~ determination, we would thus propose measuring it from the right precordial leads given BrS is primarily a right ventricular disorder.
While it may appear that the difference in T~peak~‐T~end~ between high‐risk and low‐risk Brugada patients was only small, at around 12 ms, it should be emphasized that increased transmural dispersion of repolarization is only one mechanism by which re‐entrant arrhythmogenesis is generated. Other mechanisms, such as reduced conduction velocity, increased dispersion of conduction[45](#joa312118-bib-0045){ref-type="ref"} or dynamic substrates such as steep action potential restitution,[46](#joa312118-bib-0046){ref-type="ref"} in which normal T~peak~‐T~end~ interval, T~peak~‐T~end~/QT ratio or T~peak~‐T~end~ dispersion may be observed, also contribute to arrhythmogenesis in Brugada syndrome. Therefore, better risk stratification scores will need to incorporate a combination of repolarization and conduction indices. Moreover, some of these dynamic changes may not be detectable on the ECG and may require additional tests such as non‐invasive ECG imaging (ECGi),[43](#joa312118-bib-0043){ref-type="ref"} or only becomes detectable only under stressful conditions such as exercise.[43](#joa312118-bib-0043){ref-type="ref"}
4.1. Limitations {#joa312118-sec-0015}
----------------
The following limitations of this meta‐analysis should be noted. First, there is marked heterogeneity between the included studies. The method of T~peak~‐T~end~ determination across the studies was split even between the tangent method and full recovery of the voltage to baseline. Subgroup analysis based on the method used did not reduce the heterogeneity observed. Therefore, measurement method was unlikely to have significantly contributed to the heterogeneity observed. Moreover, the Letsas 2010 study[12](#joa312118-bib-0012){ref-type="ref"} used a different endpoint of inducible VT compared to the remaining studies, but its exclusion did not significant affect the mean T~peak~‐T~end~ values for event‐positive group, event‐negative group, and mean difference between these groups. Second, retrospective studies may have more bias than prospective studies. Finally, it should be acknowledged that there is overlap between event‐postiive and event‐negative groups irrespective of the method of measuring T~end~. This would suggest as a single measurement, T~peak~‐T~end~ is unlikely to be useful in its own right. Indeed, accurate risk stratification will require a composite scoring system assessing not only dispersion of repolarization, but that of conduction, clinical symptoms, family history, the type of Brugada pattern, genetic background, electrical and drug provocation testing as well as electrophysiological mapping.[38](#joa312118-bib-0038){ref-type="ref"}, [41](#joa312118-bib-0041){ref-type="ref"}, [45](#joa312118-bib-0045){ref-type="ref"}, [47](#joa312118-bib-0047){ref-type="ref"}, [48](#joa312118-bib-0048){ref-type="ref"}, [49](#joa312118-bib-0049){ref-type="ref"}
5. CONCLUSIONS {#joa312118-sec-0016}
==============
T~peak~‐T~end~ interval, T~peak~‐T~end~/QT ratio and T~peak~‐T~end~ dispersion were higher in high‐risk than low‐risk Brugada subjects, and thus offer incremental value for risk stratification.
CONFLICTS OF INTERESTS {#joa312118-sec-0018}
======================
Authors declare no Conflict of Interests for this article.
Supporting information
======================
######
######
Click here for additional data file.
GT thanks the Croucher Foundation of Hong Kong for the support of a clinical assistant professorship.
| {
"pile_set_name": "PubMed Central"
} |
Studies using magnetic resonance imaging (MRI),^[@bibr6-2325967114550274],[@bibr7-2325967114550274],[@bibr11-2325967114550274],[@bibr26-2325967114550274]^ computed tomography,^[@bibr21-2325967114550274]^ and ultrasonography^[@bibr25-2325967114550274]^ have shown that the semitendinosus (ST) tendon can regenerate after harvest. With increasing numbers of anterior cruciate ligament (ACL) reconstructions, the amount of revision procedures after failed primary reconstructions will increase as well. If total regeneration of the ST tendon really occurs, this might be a potential source for future graft material in conjunction with ACL revision surgery. Studies concerning the histology of the ST tendon appear important since open biopsy specimens obtained in the short term have shown that real tendinous tissue regenerates. However, focal areas of scar tissue^[@bibr7-2325967114550274]^ and histological changes^[@bibr23-2325967114550274]^ compared with the healthy ST tendon can also be found. The literature does not provide evidence that the ST tendon returns to normal, and furthermore, there are no studies where the regenerated tendon is compared with the contralateral nonharvested tendon from the same patients. The purpose of the present medium- to long-term study was to investigate whether the regenerated ST tendon has histology similar to that of the contralateral nonharvested tendon from the same patient, in terms of fiber structure, cellularity, vascularity, and content of glycosaminoglycans (GAGs). The hypothesis of the study was that this would be the case.
Materials and Methods {#section1-2325967114550274}
=====================
Eighteen patients (8 females, 10 males) who underwent ACL reconstruction using ipsilateral ST and gracilis (G) tendon autografts were included in the study. The patients were a subgroup from a previously published study focusing on tendon regeneration as seen on MRI,^[@bibr1-2325967114550274]^ who agreed to undergo a bilateral biopsy procedure.
The median age at reconstruction was 23 years (range, 17-40 years). Percutaneous specimens were obtained from the regenerated tendon and the contralateral nonharvested healthy ST tendon under ultrasonographic guidance at a median of 8.4 years (100.5 months; range, 77-129 months) after the harvesting procedure. Specimens from the nonharvested side served as controls. In all, 36 biopsies were obtained.
Surgical Technique {#section2-2325967114550274}
------------------
At the index operation, the ST/G autograft was harvested through a 3-cm oblique incision over the pes anserinus. The sartorius fascia was incised parallel to the fibers of the fascia just above the thicker and more distally inserted ST tendon. After the vinculae had been cut under visual control, the full lengths of the tendons were harvested with a semiblunt, semicircular open tendon stripper (Acufex; Microsurgical Inc). The femoral bone tunnels were prepared using a standard transtibial or medial portal approach. No harvest site drain was used.
Rehabilitation {#section3-2325967114550274}
--------------
All patients were rehabilitated according to the same accelerated protocol used for all patients at the clinic, permitting immediate full weightbearing and full range of motion.^[@bibr27-2325967114550274]^ No rehabilitation brace was used.^[@bibr4-2325967114550274]^ Closed-chain exercises were started immediately postoperatively. Terminal extension with an external load other than the weight of the operated leg was not permitted during the first 6 weeks postoperative. Running was permitted after 3 months, and contact sports after 6 months at the earliest. During the rehabilitation period, no sprains or ruptures were registered in the posterior part of the thigh.
Biopsy Procedure {#section4-2325967114550274}
----------------
Specimens were obtained from the ST tendon on the operated and nonoperated side of each patient. The biopsy specimens were obtained under ultrasonographic guidance with a free-hand technique using a 1.2-mm Tru-cut Monopty instrument (Bard Inc). This is a metal handle with a preattached disposable biopsy needle. The gun needle moves in 2 stages when fired. During the first stage, the inner stylet punctures the target and, in the second stage, an outer cannula follows the path of the stylet, covering the sample notch and thus capturing the sample. Local anesthesia with adrenaline (5-10 mL) was given subcutaneously. Under ultrasonographic guidance, the ST and G tendons were identified proximally on the thigh and followed to a position approximately 4 cm above the medial joint line with the knee in slight flexion. In this position, the specimens were obtained from the central part of the ST tendon through a small incision. Each specimen was placed separately in a coded tube. The specimens had a depth of approximately 5 mm and a maximum diameter of 1.2 mm.
Clinical Assessment {#section5-2325967114550274}
-------------------
The Tegner activity level and the Lysholm score were used to assess patient function at follow-up.
Evaluation of Histology Using the Light Microscope {#section6-2325967114550274}
--------------------------------------------------
The specimens were fixed in 10% neutral-buffered formalin, embedded in paraffin, and sectioned at 4 to 5 µm, according to routine procedures. The sections were stained with hematoxylin and eosin to evaluate fiber structure, cellularity, and vascularity, and Alcian blue (pH, 2.5)/periodic acid--Schiff (AB/PAS) for the detection of GAG-rich areas. A pathologist and an orthopaedic surgeon, both with a specific interest in and knowledge of tendon pathology, simultaneously examined the tendon specimens using a light microscope (Leica DMRBE). Both examiners were blinded for whether the specimens came from regenerated or nonharvested ST tendon. The specimens were evaluated using a semiquantitative (nonparametric) grading system for the tendon alterations used in multiple previous studies.^[@bibr13-2325967114550274],[@bibr17-2325967114550274],[@bibr18-2325967114550274],[@bibr28-2325967114550274]^ Grading was based on a 4-point scoring system ([Table 1](#table1-2325967114550274){ref-type="table"}). Fiber structure, cellularity, vascularity, and level of GAGs were graded after examining the entire section. The number of cells was estimated in a high-power field (HPF) representative of the section.
######
Semiquantitative Scoring System*^a^*
Grade 0 Grade 1 Grade 2 Grade 3
-------------------- ----------------------------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------- ---------------------------------------------------------
Fiber structure Straight, parallel, packed fibers, with slight waviness Slight separation of fibers, increased waviness Separation of fibers, deterioration of fibers Complete loss of fiber structure and hyalinization
Cellularity \<100 cells/HPF 100-199 cells/HPF 200-299 cells/HPF \>300 cells/HPF
Vascularity Vessels run parallel to the collagen fiber bundles in the septa Slight increase in vessels, including transverse vessels in the tendon tissue Moderate increase in vessels within the tendon tissue Markedly increased vascularity with clusters of vessels
Glycosaminoglycans No alcianophilia Slight alcianophilia between the collagen fibers Moderate increase in alcianophilia Markedly increased alcianophilia
*^a^*A semiquantitative, 4-point scoring system^[@bibr13-2325967114550274]^ was used to evaluate the biopsies. HPF, high-power field.
Statistical Analyses {#section7-2325967114550274}
--------------------
Median (range) values are presented. The Wilcoxon signed-rank test was used for comparisons between the regenerated and nonoperated ST tendon specimens. A value of *P* \< .05 was considered significant. When planning the study, a difference of 1 unit in the classification of fiber structure between regenerated and nonharvested tendons was expected. The required sample size would then be 10 paired specimens to reach a power of 80%, if the standard deviation is 1 unit for the difference between pairs. To allow for lost and nongradable samples, 18 paired specimens were obtained.
Ethics {#section8-2325967114550274}
------
The Ethics Committee at the University of Gothenburg approved the study. All patients gave their informed consent.
Results {#section9-2325967114550274}
=======
The patients had a median Tegner activity level of 6 (range, 5-7) and a median Lysholm score of 87 (range, 47-100) at the time for biopsy procedure.
Bilateral biopsy specimens were obtained in all patients (n = 36 specimens). The patients experienced no pain or discomfort during or after the biopsy procedures. One patient had undergone ACL reconstruction on the contralateral side and was therefore excluded. Also, the regenerated ST tendon in 1 patient and the nonharvested tendon specimen in 4 patients contained insufficient amounts of tissue for evaluation. This left 16 specimens from regenerated tendons and 13 specimens from the contralateral side; thus, 12 patients (24 specimens) were available for paired specimen comparison. The semiquantitative scoring system revealed no significant differences for the fiber structure, cellularity, vascularity, and the amount of GAGs between the regenerated and nonharvested contralateral side ([Table 2](#table2-2325967114550274){ref-type="table"}). The fiber structure in both the regenerated and nonharvested tendons was classified as median grade 1. However, of the 16 specimens from the regenerated tendon, 3 were classified as fiber structure grade 3 in focal areas, and in 3, increased levels of GAGs were detected. Furthermore, 5 specimens from the regenerated tendons had \>200 cells/HPF.
######
Results of the Histological Analysis*^a^*
Regenerated ST Tendon Healthy ST Tendon *P* Value
-------------------- ----------------------- ------------------- -----------
Fiber structure 1 (0-3) 1 (0-2) .20
Cellularity 1 (0-3) 0 (0-2) .11
Vascularity 0.5 (0-2) 0 (0-2) .36
Glycosaminoglycans 0 (0-1) 0 (0-0) .08
Missing value 1 4
*^a^*Values are reported as median (range). The specimen from 1 of the regenerated tendons and 4 specimens from the nonoperated side contained insufficient amounts of tissue, and 1 patient had undergone anterior cruciate ligament reconstruction on the contralateral side. This left 12 patients for the paired specimen comparison. ST, semitendinosus.
In the remaining specimens from the regenerated tendon tissue and in all 13 healthy tendon specimens, no areas of grade 3 fiber structure or GAGs could be detected. In terms of cellularity, 2 nonharvested ST tendons had \>200 cells/HPF ([Figures 1](#fig1-2325967114550274){ref-type="fig"} and [2](#fig2-2325967114550274){ref-type="fig"}).
{#fig1-2325967114550274}
{#fig2-2325967114550274}
Discussion {#section10-2325967114550274}
==========
The principal finding of the present study was that the regenerated ST tendon appeared similar to the contralateral nonharvested ST tendon when evaluated histologically at a median 8.4 years after harvesting.
To our knowledge, this is the only study in which specimens from the regenerated ST tendon are compared with contralateral nonharvested tendon specimens from the same patients. Furthermore, this is the largest study of histological analyses of regenerated ST tendons.
There are few studies addressing the histological appearance of regenerated ST tendon. Previous studies^[@bibr7-2325967114550274],[@bibr8-2325967114550274],[@bibr23-2325967114550274]^ and 1 case report^[@bibr31-2325967114550274]^ on open biopsy specimens support our findings that it is tendinous tissue that regenerates and not just scar tissue. However, the longest time period presented in these studies between harvest and biopsy procedure is less than 2.5 years. Reviewing these articles, the authors^[@bibr22-2325967114550274]^ state that at 2.5 years postoperative, the maturation process is probably not complete. A long-term study presenting the final histological outcome for the entire regeneration process is therefore of particular interest. Eriksson et al^[@bibr7-2325967114550274]^ obtained open biopsies from regenerated ST tendons in 5 patients at a median 20 months after harvesting, and they reported that the regenerated tendons showed the features of a healthy tendon, but that focally there were small scar-like areas with more irregularly oriented collagen, increased fibroblastic proliferation, and capillary formation compared with healthy control tendons.
In the present study, the fiber structure in both the regenerated and nonharvested control tendons generally showed slight separation of and increased waviness in the fibers. However, 3 of the regenerated tendons displayed a grade 3 classification in focal areas in terms of fiber structure. These findings are similar to those reported by Eriksson et al.^[@bibr7-2325967114550274]^
Sulphated GAGs appear in low concentrations in the healthy patellar tendon,^[@bibr2-2325967114550274],[@bibr3-2325967114550274]^ while high levels are found in Achilles tendinopathy,^[@bibr19-2325967114550274]^ patellar tendinosis ("jumper's knee"),^[@bibr15-2325967114550274]^ ruptured tendons,^[@bibr12-2325967114550274]^ and tendons subjected to compression forces.^[@bibr30-2325967114550274]^ The content of GAGs in the present study was not detectable in most regenerated specimens, which could be regarded as a sign of tendon normalization.
The 4-point scoring system used in the present study was initially developed for evaluating alterations in the patellar tendon, with a score of 0 in all the measured items in the healthy patellar tendon. In the present study, the histological score was slightly higher for the nonharvested ST tendon than for the healthy patellar tendon. However, a different morphologic appearance between the patellar tendon and the ST tendon has been described by Hadjicostas et al.^[@bibr10-2325967114550274]^ They report increased cellularity as well as a tendency toward increased vascularity in the ST tendon in 20 cadavers, similar to the findings in the present study.
The way the tendon regenerates has not been clarified. Initially, it was suggested that the tendon regenerates in a proximal to distal direction,^[@bibr6-2325967114550274]^ and Leis et al^[@bibr16-2325967114550274]^ termed it "the lizard tail phenomenon." However, the present view is that the tendon matures uniformly along the harvest site.^[@bibr5-2325967114550274],[@bibr24-2325967114550274],[@bibr25-2325967114550274]^ Using serial ultrasonography, Papandrea et al^[@bibr25-2325967114550274]^ described an initial hematoma followed by an edema with gradual solidification along the entire harvest site. This theory is supported by a systematic review,^[@bibr5-2325967114550274]^ and further proof was recently presented using an animal model for tendon regeneration in Achilles tendons.^[@bibr24-2325967114550274]^ These authors^[@bibr24-2325967114550274]^ described a similar regeneration and maturation uniformly along the length of the regenerated tendon. Okahashi et al^[@bibr23-2325967114550274]^ suggested that the surgical method of "stripping" when harvesting the tendons plays an important role in the regenerative process. Synovial cells possess the ability to differentiate when subjected to mechanical stress. Eriksson et al^[@bibr7-2325967114550274]^ proposed that the hematoma that occupies the harvest defect acts as a scaffold for the subsequent tendon regeneration. Otoshi et al^[@bibr24-2325967114550274]^ concluded that the hematoma scaffold enhances migration of fibroblast precursor cells from the surrounding peritendinous tissue and tendon sheath when examining the regeneration process after Achilles tendon stripping in a rat model. In the present study, a standard stripping technique was used for harvesting the ST/G tendon in all patients, and no drain was used. It might be that the initial hematoma is important for the regeneration of the tendons. By stripping both tendons at surgery, the risk that the tendon from which the specimen was obtained actually was a nonharvested tendon was eliminated.
The regeneration rate for the harvested tendons differs in the literature from 46%^[@bibr29-2325967114550274]^ to 95%.^[@bibr1-2325967114550274]^ The reason for this is unknown. Differences in the harvesting technique, time from harvest to study, and patient factors such as smoking could influence the regeneration rate. Furthermore, aggressive postoperative rehabilitation could cause the weak regenerated tendon structure to rupture in the first months after harvest. This is supported by Nakamae et al,^[@bibr20-2325967114550274]^ who reported 2 cases of unsuccessful regeneration of the semitendinosus tendon at 12 months using 3-dimensional computed tomography. These patients had experienced a sudden sharp pain in the posterior aspect of their thighs when their hamstring muscles were subjected to aggressive load shortly after surgery. This raises the question of whether something happened in the early regeneration process to the patients classified as grade 3 for fiber structure. Trauma with a microbreak in the weak tendon that did not result in a hematoma as seen after tendon stripping might be the cause of loss of fiber structure.
The essential question is whether a regenerated ST tendon can be used for ACL revision surgery in the same way as has been reported for the patellar tendon.^[@bibr14-2325967114550274]^ There is only 1 case report in the literature in which regenerated ST tendon was used for revision surgery.^[@bibr31-2325967114550274]^ However, information about the size and strength of the tendon, in addition to histological data, is necessary to predict the outcome. In animal models, the biomechanical strength in the regenerated ST tendon^[@bibr9-2325967114550274],[@bibr16-2325967114550274]^ and Achilles tendon^[@bibr24-2325967114550274]^ has been described as being inferior to that of the healthy tendon up to 1 year after harvesting but with a trend toward increasing strength over time.^[@bibr16-2325967114550274]^ Since ST tendon regeneration is unpredictable in terms of focal scarring and until studies have been conducted with long-term biomechanical testing in humans, it is the opinion of the authors that regenerated ST tendon cannot be recommended for ACL revision surgery.
The strengths of the present study are its long-term design and the paired biopsies from the patients' regenerated and nonharvested ST tendon, enabling the patient to serve as her or his own control. A limitation of the study is that 5 biopsies contained an insufficient amount of material for analysis. Although this is the largest study in the literature, there is a potential risk that no significant differences were found due to a type II error. Performing the power analyses on the nonparametric primary variable is also a potential weakness. In spite of the nonsignificant difference, some of the ST tendons displayed focal areas of scarring. It is possible that, by obtaining several biopsies from the entire length of the ST tendon, focal scarring would have been found in more of the regenerated ST tendons. There is a small risk that the biopsies were obtained from the G tendon. However, we chose to only include patients in whom both the ST and G tendons were harvested. This means that the biopsy was always obtained from a regenerated tendon. A further limitation is that no intra- or interobserver reliability testing has been performed on the score that was used in this study.^[@bibr13-2325967114550274],[@bibr17-2325967114550274],[@bibr18-2325967114550274],[@bibr28-2325967114550274]^ However, the original score from which the score used in the present study was developed has been tested for intraobserver reliability with satisfactory agreements for different tendons.^[@bibr19-2325967114550274]^ Finally, since no biomechanical tests of the regenerated ST tendon have been performed, for obvious ethical reasons, the true quality of the regenerated tendon is unknown.
Conclusion {#section11-2325967114550274}
==========
The ST tendon regenerates and may regain a histological appearance similar to that of the nonharvested contralateral tendon, as seen in this study, a median of 8.4 years after harvesting. However, in some tendons, loss of fiber structure was found. The hypothesis of the study was thus verified.
One or more of the authors declared the following potential conflict of interest or source of funding: J.K. has received honorariums for lecturing from Linvatec. This study was supported by funding from the Western Sweden County Council Research Fund and the Swedish Centre for Research in Sports.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Complementary feeding practice is a noteworthy factor that determines the nutritional status of children. The transition period from exclusive breastfeeding to 2 years is a critical window for optimal growth and development of the child. During this period, appropriate, safe, adequately nourished and frequent feeding is essential. Innocently, the food provided to a child might be too high or too low in some nutrients, the diversity of food might be adequate or inadequate and micronutrient content including iron could be lower than required \[[@CR1]\].
Suboptimal (inadequate) infant feeding practices are the major reasons for childhood under nutrition in developing countries \[[@CR2], [@CR3]\]. Poor nutritional status of children in most developing countries is due to the presence of overwhelming of poverty, low maternal education, high burden of disease and mal-feeding practices \[[@CR4], [@CR5]\]. Many survey reports \[[@CR4]--[@CR9]\] consistently indicated that underprivileged child feeding practices are correlated with cultural factors such as selection of low-quality complementary foods, taboos, restrictive traditional beliefs and social factors including caregivers' poor knowledge on nutrition and lack of knowledge on food diversity in their surroundings. Ultimately, all of these factors lead to low dietary diversity, low feeding frequency and low food and energy intake for children \[[@CR2], [@CR10]\].
Several studies have shown that dietary diversity score (DDS) is positively associated with overall dietary quality and micronutrient intake of young children and found to be proxy indicator for household food security and in the long run for childhood stunting \[[@CR6], [@CR11]\]. A higher DDS has also been associated with better nutritional status of children in developing countries \[[@CR12], [@CR13]\].
As per the recommendation of World Health Organization/Pan American Health Organization (WHO/PAHO) 2003, breastfed children 6--23 months should receive animal-source foods and vitamin A-rich fruits and vegetables daily. Therefore, four food groups (grain- or tuber-based staple, animal-source food, vitamin A-rich fruit or vegetable) are considered the minimum acceptable number of food groups for breastfed infants. Non-breastfed children should be fed meals four or five times per day, with one to two snacks as desired. Meal frequency is considered a proxy for energy intake from foods other than breast milk. Therefore, for non-breastfed children feeding frequency indicators include both milk feeds and solid or semi-solid feeds \[[@CR14], [@CR15]\].
Updated knowledge of feeding practices will assist the national nutrition programme to monitor the changes in the feeding practices and design interventions to increase the recommended feeding practices and thereby contribute in reducing undernutrition in the country. In general speaking, Ethiopia is known to have low minimum dietary diversity (MDD) as compared to the rest of the world. Some surveys including national demographic health survey of 2011 conducted in Ethiopia revealed that MDD is below 5%. Five years back, the magnitude of MDD, minimum meal frequency (MMF) and minimum acceptable diet in Southern Ethiopia were 3.8, 49.8 and 3.1%, respectively \[[@CR16]\]. Even though there are limited studies done in Ethiopia which were primarily focusing on measuring complementary feeding levels in the rural communities, the current study was unique because it was designed to be conducted in urban area of Ethiopia. The existing surveys are found mostly in Northern and Western parts of Ethiopia \[[@CR17]\]. No survey was investigated on dietary diversity and meal frequency in Southern Ethiopia particularly in the urban areas. There should be an urgent measure to identify reasons why complementary feeding indicators are still low. This study was aimed to measure the proportions of minimum dietary diversity and minimum meal frequency and to investigate factors associated with them among young children aged 6--23 months.
Methods {#Sec2}
=======
Study area, study design and participants {#Sec3}
-----------------------------------------
Community-based cross-sectional study design was carried out in Wolaita Sodo town on March 02--20, 2015. The town is found in Wolaita zone, Southern Ethiopia, and 315 km far away from Addis Ababa. The town is administratively structured by 11 kebeles and has a total population of 110,660, of which 54,275 are males and 56,385 are females. Out of all female population, 25,784 of them are women in the reproductive age group. About 14% of the total population are children 6--59 months of age.
Mothers or caregivers of children 6--23 months of age who reside in Wolaita Sodo town were the source population, whereas mothers/caregivers of children 6--23 months that were drawn from the selected kebeles were considered as the study population. Mothers/caregivers of children 6--23 months of age who have been residents of the town and have ever breastfed in the selected kebeles were included in the study. However, mothers or caregivers who are seriously ill, mental problem or unable to communicate were excluded.
Sample size and sampling procedure {#Sec4}
----------------------------------
The sample size was determined using single proportion population formula by assuming the proportion of introduction of complementary foods as 42.9% in western Ethiopia \[[@CR17]\], 1.5 as design effect, 5% as level of significance and 5% as degree of precision and which was 566. Adding 10% as non-response rate, 623 mothers/caregivers with 6--23 months of children were included in the study. To ensure the adequacy of sample size, Epi-info was used to calculate sample size for factors associated with minimum dietary diversity and minimum meal frequency. Then, the maximum sample size was taken.
Two stage sampling was used to select the participants. Seven administrative units were randomly drawn from 11 administrative units. A census was conducted in these selected kebeles to identify the study participants. The sample size was allocated to the population size proportionately, and sampling interval was calculated. Finally, 623 participants were selected using a systematic sampling method after randomly identified the first household and proceed to the second participant based on the interval.
Data collection and measurements {#Sec5}
--------------------------------
Data were collected using interviewer-administered validated questionnaire in a face-to-face manner from mothers/caregivers of children 6--23 months of age. The questionnaire consists of three parts: socio-demographic characteristics of households, maternal and child health related features and child feeding practices. Twenty-four-hour recall method and food frequency questionnaire were used to assess dietary diversity and meal frequency. Ten first degree in nursing holders as data collector and three second degree holders who had previous experiences as supervisor were involved in the survey.
Structured questionnaire partly adopted from WHO assessment tool for infant and young child feeding (IYCF) practices were used and translated into local language by fluent speakers and back translated to English to validate the consistency. Training was given for data collectors and supervisors on how to interview and maintaining the quality of data. Pre-test was done on 5% of the participants out of the selected areas. Then, the questionnaire was rechecked for its precision and consistency, and necessary modifications were incorporated before commencing the actual data collection. The supervisors and the investigatory were regularly monitored and checked the completeness of the data in daily bases.
### Minimum dietary diversity {#Sec6}
Proportion of children 6--23 months of age who receive foods from four or more food groups during the previous day. The seven food groups used for tabulation of this indicator were as follows: cereals, roots and tubers; legumes and nuts; dairy products (milk, yoghurt and cheese); flesh foods (meat, fish, poultry and liver/organ meats); eggs; vitamin A-rich fruits and vegetables and other fruits and vegetables \[[@CR14]\].
### Minimum meal frequency {#Sec7}
Proportion of breastfed and non-breastfed children 6--23 months of age, who receive solid, semi-solid or soft foods (but also including milk feeds for non-breastfed children) the minimum number of times or more. Breastfed infants of age 6--8 and 9--23 months should obtain a minimum of two or three meals with one to two snacks and three or four meals with one to two snacks per day, respectively. But non-breastfed infants of age 6--23 months should receive milk products at least twice a day \[[@CR8]\].
Statistical analysis {#Sec8}
--------------------
Data were entered to Epi-Data 3.02 and exported, cleaned and analysed by SPSS version 21. Missed data were explored, and normality for continuous variables was checked. Dietary diversity score (DDS) was computed out of seven from seven food groups. Reliability of the tool was done, and Cronbach's Alpha value was 0.76. Household economic status was measured by constructing a wealth index through principal component analysis. The indicator variables used for wealth index construction that fulfil the requirement of factor analysis were telephone, table, chair, refrigerator and electric mitad. Varimax rotation was used. The communality of each variable was greater than 0.53; Kaiser-Meyer-Olkin measure of sampling adequacy was 0.58. The cumulative proportion of variance criteria was met with two components which was 66.80%. Split sample validation was done, and none of communality's of the variable in each split was below 0.5 and finally categorized into poor, medium and rich.
The data were presented in tables and figures by computing the percentages of minimum dietary diversity, meal frequency and acceptable diet. Binary logistic regression was done for the two outcome variables of MDD (1 = met 4 and above food groups, 0 = met less than four food groups) and MMF (1 = met the minimum requirement for age groups, 0 = not fulfil the minimum requirement). The strength of associations and statistical significances between explanatory variables and outcome variables were expressed using OR and 95% of confidence interval, respectively.
All variables in the binary logistic regression with *p* value \<0.1 were moved to multivariable logistic regression and done using backward likelihood ratio to control the possible confounders and to identify predictors of the outcome variables. At this level, model fitness was checked which was Hosmer-Lemonshow as 0.67 and no multicollinearity. The variable was considered to be predictive for each outcome variable at *p* value of less than 0.05.
Results {#Sec9}
=======
Characteristics of participants {#Sec10}
-------------------------------
There were 611 mothers/caregivers with children 6--23 months, which constitute 98.1% of response rate. From all participants, 605 (99%) of them were biological mothers. The mean age of mothers/caregivers was 26.7 years with ± 4.8 years of standard deviation, and the median age was also 26 years. Two third (65%) of the respondents' occupational status were found to be as house wives, and about a quarter of the participants (26%) were accomplished grade 10 and colleges but 11.5% of mothers/caregivers had no education (Table [1](#Tab1){ref-type="table"}).Table 1Socio-demographic, economic and other characteristics of the participants reside in Wolaita Sodo town, Ethiopia, in 2015Variables (*n* = 611)Frequency (%)Household head Male570(93.3) Female41(6.7)Maternal and caregivers age (years) 15--1918(3.0) 20--24166(27.2) 25--29263(43.0) 30--34109(17.8) ≥3555(9.0)Maternal relation with child Mother605(99.0) Caregiver6(1.0)Marital status Single3(0.5) Married577(94.4) Divorced22(3.6) Widowed9(1.5)Household head occupation House wife30(4.9) Government employee184(30.1) Daily worker144(23.6) Private worker253(41.4)Maternal occupation Housewife397(65.0) Private91(14.9) Government123(20.1)Educational status of woman Illiterate70(11.5) Grade 1--8218(35.7) Grade 9--10164(26.8) Above 10159(26.0)Wealth index Poor433(70.9) Middle (medium)72(11.8) Rich106(17.3)Sex of child Male351(57.4) Female260(42.6)Age of child (in months) 6--8112(18.3) 9--11118(19.3) 12--17249(40.8) 18--23132(21.6)ANC follow-up Yes595(97.4) No16(2.6)PNC follow-up Yes469(76.8) No142(23.2)Place of delivery Home without TBA25(4.1) Home with TBA12(2.0) Government health facility523(85.6) Private health facility51(8.3)
Complementary feeding practices {#Sec11}
-------------------------------
The overall children who met the requirement of minimum dietary diversity were 27.3% ranged from 23.7--30.8% at 95% CI and minimum meal frequency for both breastfed and non-breastfed children were 68.9% which lied within 65.2--72.6% at 95% CI. Moreover, those who met the requirement of a minimum acceptable diet were 21.1% (Fig. [1](#Fig1){ref-type="fig"}).Fig. 1Indicators of complementary feeding practice of children 6--23 months of age in Wolaita Sodo town, Ethiopia, in 2015
There was a clear observation that dietary diversity decreases as age increases but as age increases meeting the requirement of meal frequency increases. Significant numbers of 6--8 months of children (50.9%) were met the requirement of a minimum dietary diversity as compared to children 18--23 months of age (22.3%) (Table [2](#Tab2){ref-type="table"}).Table 2Percentage distribution of minimum dietary diversity, meal frequency and acceptable diet disaggregated by age of children 6--23 months in Wolaita Sodo town, 2015Child age categoryMeet minimum dietary diversity (*n* = 611)Meet minimum meal frequency (*n* = 611)Meet minimum acceptable diet (*n* = 611)6--8 months50.9%65.2%33.9%9--11 months28.8%62.7%19.5%12--17 months18.1%69.3%17.2%18--23 months22.3%75.3%18.7%Over all practice with 95% CI27.3% (23.7--30.8%)68.9% (65.2--72.6%)21.1% (17.8--24.3%)
The majority of children (84.6%) consumed grains, roots and tubers and followed by legumes and nuts (Fig. [2](#Fig2){ref-type="fig"}) in the past 24 h prior to the data collection. However, smaller proportions of children were consumed fish (9.7%) and iron-rich food (liver) 1.1%.Fig. 2The proportion of 6--23 months of children who consumed a variety of food groups in Wolaita Sodo town, Ethiopia, in 2015
Factors associated with minimum dietary diversity and meal frequency {#Sec12}
--------------------------------------------------------------------
Among the variables reached to the final model, household head, occupation and child age were statistically associated with minimum dietary diversity. Children from households headed by housewives were 2.3 times more likely to be fed four and above food items or groups per day as compared to children from households headed by private workers \[adjusted odds ratio (AOR) = 2.3; 95% CI (1.01--5.4)\]. In addition, the odds of meeting minimum dietary diversity among children from families headed by government workers were almost four times higher than the odds of minimum dietary diversity among children from families headed by private workers \[AOR = 3.7; 95% CI (2.3--5.9)\]. The odds' of minimum dietary diversity in children 6--8 months of age were nearly five times higher than the odds' of minimum dietary diversity among older children \[AOR = 5.2; 95% CI (2.9--9.1)\].
Likewise, among the variables reached the final step, household head occupation, maternal/caregiver educational level, child sex and age were identified as predictors of minimum meal frequency. The chance of children came from households lead by government workers, who met the requirement of minimum meal frequency, were reduced by 40% \[AOR = 0.6; 95% CI (0.4--0.9)\]. The probability of meeting the requirement of minimum meal frequency was reduced by 50% if the child fed by illiterate mothers/caregivers as compared to mothers or caregivers who attain grade 10 and above \[AOR = 0.5; 95% CI (0.2--0.9)\]. Girls were nearly two times more likely to be fed frequently as compared to boys \[AOR = 1.5; 95% CI (1.02--2.1)\] (Table [3](#Tab3){ref-type="table"}).Table 3Backward multivariable logistic regression used to identify predictors of complementary feeding practices (MDD and MMF) in children 6--23 months at Wolaita Sodo town, Ethiopia, in 2015 with 95% of confidence intervalVariablesMeet MDD (*n* = 611)Meet MMF (*n* = 611)*n*CORAOR*n*CORAORHousehold head Male1540.8(0.4--1.6)^a^3910.8(0.4--1.6)^a^ Female131301Maternal age ≤24 years541.5(0.8--2.7)^a^1320.8(0.5--1.5)^a^ 25--30 years941.4(0.7--2.4)2230.6(0.4--1.1) ≥31 years191661Marital status Single131.7(0.8--3.5)^a^261.5(0.7--3.4)^a^ Married15413951House hold head occupation House wife12*2.4*(*1.1--5.3*)*2.3*(*1.01--5.4*)252.1(0.8--5.8)2.0(0.7--5.6) Government employee31*3.3*(*2.1--5.2*)*3.7*(*2.3--5.9*)112*0.7*(*0.4--0.9*)*0.6*(*0.4--0.9*) Daily worker690.7(0.4--1.2)0.7(0.4--1.2)1071.2(0.8--1.9)1.3(0.8--2.2) Private worker55*1*117711Maternal occupation House wife1201.6(0.9--2.6)2761.1(0.7--1.7) Government employee211.1(0.6--2.1)^a^621.0(0.6--1.8)^a^ Private worker261831Maternal education Illiterate320.9(0.6--1.4)1.5(0.6--2.7)450.8(0.4--1.4)*0.5*(*0.2*--*0.9*) Up to grade 1094*2.3*(*1.3*--*4.4*)0.7(0.4--1.1)2650.9(0.6--1.5)0.8(0.5--1.2) Above grade 10411111111Wealth index Poor1211.3(0.7--2.0)3081.5(0.9--2.3) Middle211.3(0.6--2.6)^a^471.1(0.6--2.1)^a^ Rich251661Child sex Male941^a^22911 Female731.1(0.7--1.5)192*1.5*(*1.1*--*2.1*)*1.5*(*1.02*--*2.1*)Child age (months) 6--857*3.6*(*2.1*--*6.1*)*5.2*(*2.9*--*9.1*)730.6(0.4--1.0)0.6(0.3--1.02) 9--11341.4(0.8--2.4)*1.8*(*1.01*--*3.3*)74*0.6*(*0.3*--*0.9*)*0.5*(*0.3*--*0.9*) 12--17390.7(0.5--1.3)0.9(0.6--1.7)1490.7(0.5--1.2)0.7(0.5--1.2) 18--23371112511ANC follow-up Yes1600.5(0.2--1.3)^a^4080.5(0.1--1.8)^a^ No71131Place of delivery Home delivery151^a^261^a^ Health facilities1520.5(0.3--1.0)3950.9(0.4--1.9)The italic values show statistically significant variables*COR* crude odds ratio, *AOR* adjusted odds ratio, *MDD* minimum dietary diversity, *MMF* minimum meal frequency^a^Variables in the model not reached final step
Discussion {#Sec13}
==========
The survey publicized that the proportions of minimum dietary diversity and minimum meal frequency were 27.3 and 68.9% in all 6--23 months of children, respectively. In this study, minimum dietary diversity and meal frequency were higher than findings from northwest part of Ethiopia (with MDD of 12.6% and MMF of 50.4%) and also higher than results analysed from EDHS 2011 which showed that 10.8 and 44.7% of children meet MMD and MMF, correspondingly \[[@CR16], [@CR17]\]. The proportion of 6 to 23 months of children who met the recommended level of meal frequency in the study were nearly two times more than report from demographic health survey data analysis from Tanzania and Uganda but with nearly similar report with dietary diversity \[[@CR7]\]. And these core indicators were also higher than findings from WHO 2010 reports in Eretria (19 and 44%), Guinea (18 and 30%), India (12 and 44%), Niger (5 and 42%) and Mali (16 and 25%) but less than from Kenya (45 and 58%), Zambia (37 and 55%), Indonesia (65 and 67%) and Morocco (66 and 62%) for MDD and MMF, respectively \[[@CR15]\]. The reason for the high percentage of child feeding practice in the study area may be due to the variation in time of data collection, and nutrition education now a day may play a vital role in increasing the awareness of community for better feeding practices in urban settings. Likewise, the result of the study was almost consistent with survey conducted in Nepal among 6 to 23 months of children which revealed that 76.6 and 30.4% of them obtained MMF and MDD, respectively \[[@CR8]\].
In this study among variables moved to the final model, occupation of household head, child age, maternal education and child sex were found to be statistically associated with complementary feeding practices. Household head occupation being government-employed and house wife, younger children were more likely met the recommended dietary diversity. The finding was supported by report from Nepal \[[@CR8]\]. Probably, housewives spent their entire times with their children and care them in a better manner. Government-employed mothers/caregivers are also more educated and attentive and have plan to prepare diversified diets.
Children from household headed by government-employed mothers and illiterate mothers or caregivers were less likely to meet the minimum requirement of meal frequency. Girls were more frequently take meals as compared to boys. Children in the age of 9--11 months were less likely to meet recommended meal frequency as compared to older children \[[@CR7], [@CR17]\]. But a report from Uganda \[[@CR9]\], maternal education has no association with child feeding practices. Educated mothers or caregivers are easier to be familiarized with new information and knowledge, and they know the importance of child feeding practices as compared to illiterate one who are stagnant and needs longer time to bring behavioural change. This study indicated that government-employed mothers/caregivers fed their children more diversified diet but less frequent. This is due to employed mothers/caregivers stayed at work place separated from their children for long time.
Eventually, this survey had its own strength and limitation. Tremendous efforts were made to assure the quality of the study starting from the period of predata collection to write-up of the report. Reliability of the tool was checked, and appropriate statistical test was performed for different model assumptions. Nevertheless, the study had unforgettable limitations. The study used only 24-h recall method which tells us only one time phenomenon but did not demonstrate dietary habit of the participants and affected by variation of days. Moreover, the study fails to address measuring of mean dietary adequacy, the mean density of nutrient and finally did not show the relationship of these feeding practices to nutritional status of children.
Conclusions {#Sec14}
===========
The study revealed that the percentage of children who meet the recommended level of minimum dietary diversity, meal frequency and acceptable diet were 27.3, 68.9 and 21.1%, respectively. These are good achievements as compared to the national figures but surprisingly insignificant as weighted against countries who are found in a better standard of living condition. Particularly minimum dietary diversity and acceptable diet in the study site were inadequate.
Child feeding practices in the study area were significantly influenced by household occupational status, maternal education, child sex and age, and socioeconomic status of the household. And thus, they were identified as independent factors that made inadequate complementary feeding practices. In contrast to this, there were some important factors that are not statistically significant with child feeding practices.
Despite of the observed better achievement as compared the national proportion of complementary feeding indicators, much is expected to reach the target of the health sector development programme of the country as well as the WHO recommendation level. Emphasis should be undertaken to increase maternal literacy and employment opportunity to have better economic status because they are the proxy and decisive stakeholders for superior accomplishment of reaching the target. It is better to conduct further research that focuses on the relationship of these infants and young child feeding practices with child nutritional status.
AOR
: Adjusted odds ratio
COR
: Crude odds ratio
DDS
: Dietary diversity score
EDHS
: Ethiopia Demographic Health Survey
IYCF
: Infant and young child feeding
MDD
: Minimum dietary diversity
MMF
: Minimum meal frequency
We would like to acknowledge the Wolaita Sodo University for providing the materials used for the work. Our deepest gratitude goes to the data collectors, kebeles leaders and Wolaita Sodo health centre manager for his cooperation starting from the beginning till the end of the data collection time. The last but not the least, our heartfelt thanks goes to Wolaita Sodo town residents particularly the respondents.
Funding {#FPar1}
=======
Wolaita Sodo University funded the research. The funders had no role in the study design, data collection and analysis, decision to publish or preparation of the manuscript.
Availability of data and materials {#FPar2}
==================================
The datasets during and/or analysed during the current study are available from the corresponding author on reasonable request.
Authors' contributions {#FPar3}
======================
TC initiated the research, wrote the research proposal, conducted the research, did the data entry and analysis and wrote the manuscript. SB was involved in the write-up of the methodology of proposal and research work. TM contributed to the write-up of proposal and write-up of the manuscript. WF participated in the write-up of the methodology of proposal and research work. All authors contributed equally. All authors read and approved the final manuscript.
Competing interests {#FPar4}
===================
The authors declare that they have no computing interests.
Consent for publication {#FPar5}
=======================
Not applicable.
Ethics approval and consent to participate {#FPar6}
==========================================
Ethical clearance was obtained from Wolaita Sodo University after the proposal had been approved by the research ethics review committee of the University before conducting the survey; letter from Wolaita Sodo University was given to the town administrator. After getting permission, cooperation letter was taken to each Kebele. Then, information sheet and informed consent were provided to the participants after giving clear and deep instructions about the aim of the study and taken signed written informed consent from the mothers/caregivers. Anonymous data were taken, and the confidentiality of participants' information was secured.
Publisher's Note {#FPar7}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .325}
{#sp2 .326}
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .649}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Myotonic dystrophy type 1 (DM1, OMIM no. 160900) is an autosomal dominant repeat expansion disorder, affecting skeletal and smooth muscle as well as the heart, the endocrine system, the eye and the central nervous system ([@DDW042C1]). The multisystemic manifestation and progression of DM1 are caused by expansion of a (CTG·CAG)*~n~* repeat, located in the 3′-untranslated region (3′ UTR) of the dystrophia myotonica protein kinase (*DMPK*) gene ([@DDW042C2]) and in an overlapping antisense transcription unit in the DM1 locus ([@DDW042C3]). In DM1 families the expanded repeat is unstable, both somatically and intergenerationally, with a bias toward expansion, causing progression of disease symptoms during ageing and over successive generations ([@DDW042C4]).
Several mechanisms may contribute to the molecular pathogenesis of DM1 ([@DDW042C5]). Expanded *DMPK* transcripts are retained in the nucleus, where they form focal complexes in insoluble or diffuse-soluble state by abnormal association with transcription factors and RNA-binding proteins, like members of the muscleblind-like family (MBNL1--3), DEAD-box helicases and hnRNP proteins ([@DDW042C1],[@DDW042C6],[@DDW042C7]). In turn, abnormal phase transitions in RNP complexes lead to sequestering of factors needed for processing of other transcripts with *in trans* consequences for faithful alternative splicing and polyadenylation and expression of miRNAs ([@DDW042C7],[@DDW042C8]). Production of proteins by ribosomes that decode the normally untranslated (CUG)*~n~* repeat tract in *DMPK* mRNA by a newly discovered process, coined repeat-associated non-ATG (RAN) translation, is also possible ([@DDW042C9],[@DDW042C10]). Similar toxic events may occur with antisense transcripts originating from the complementary strand of the DM1 locus, overlapping the 3′ end of the *DMPK* gene. Abnormal RNAs are thus formed with an expanded (CAG)*~n~* repeat, potentially leading to the production of homopolymeric peptides by RAN translation of the (CAG)*~n~* repeat, which may evoke an imbalance in proteostasis ([@DDW042C9],[@DDW042C10]). Finally, it cannot be excluded that problems with DNA replication across the repeat tract or abnormal epigenetic modification of the chromatin region containing the DM1 locus also contribute to pathology ([@DDW042C3],[@DDW042C11]).
Together, alterations in the transcriptome, proteome and replisome may compromise the physiological integrity of cells and tissues in which the mutant *DMPK* and the *DM1-antisense* gene are expressed. Throughout development, growth and adulthood this imbalance may lead to the loss of function and ultimately to cell degeneration, causing the muscle wasting and CNS white matter loss in patients ([@DDW042C4],[@DDW042C12]).
For study of biological mechanisms underlying DM1 pathology and for testing of possible therapeutic strategies in preclinical studies, several animal models are available, including *Drosophila*, zebra fish and mouse ([@DDW042C13],[@DDW042C14]). Predominant focus is thereby oriented toward mechanisms involved in RNA-based disease etiology. Notably, DM1 animal models differ profoundly in nature, structural organization and chromatin context of their transgenic insert and in the length of the (CTG·CAG)*~n~* segment therein. Comparison of pathobiological findings between models and extrapolation to the situation in patients remain therefore difficult. Work of others has already demonstrated that the timing of *DMPK* expression, i.e. the onset of potential RNA toxicity, influences phenotypic severity ([@DDW042C15]). Expression of RNA with an abnormal repeat tract in satellite cells or neuronal progenitor cells may affect proper muscle and brain development ([@DDW042C16]--[@DDW042C19]) and have serious consequences for tissue regenerative capacity in adulthood. The absolute number of expanded RNAs and their structure at any given moment may also be crucial, as these ultimately will influence the extent of toxicity caused by abnormal RNP binding or abnormal properties of RAN translation products ([@DDW042C20]--[@DDW042C22]). The type of gene promoters, whether from endogenous or ectopic origin, that drive transcription during development and ageing, and the structure of the transcripts that entail the repeat segment are therefore critical parameters in animal models and patients.
Here, we report on comparison of expression and measurement of absolute numbers of (CUG)*~n~*-repeat containing RNAs in muscle cells and tissues of four commonly used mouse DM1 models and in cells and biopsies from patients. DM1 mouse models express transgenes with different promoters, different structural organization and different repeat lengths: DM500, DMSXL, Tg26 and *HSA*^LR^ (Table [1](#DDW042TB1){ref-type="table"}). DM500 and DMSXL mice are both descendants of the DM300-328 line, which was subject to intergenerational repeat expansion. These mice carry a complete human DM1 locus ([@DDW042C23],[@DDW042C24]). The *DMPK* transgene in Tg26 mice carries a tandem insert of ∼25 copies of the complete human *DMPK* gene, with a normal-sized (CTG)11 repeat ([@DDW042C25],[@DDW042C26]). In *HSA*^LR^ mice, the transgene is under control of the *ACTA1* promoter and the repeat is embedded in the context of the *ACTA1* gene ([@DDW042C27]). The rationale for quantification of repeat RNA expression in these models is that knowledge about toxic RNA concentration will provide us with more insight in pathophysiological cascades *per se*, especially as more and more anatomical, physiological and behavioral phenotype data become available, enabling relatively easy cross comparisons. Furthermore, some of the DM1 models have already been extensively used for preclinical translational studies in the past decade, but translation of findings in these models has been difficult. Table 1.Characteristics of DM1 mouse models used in this study.Mouse modelTransgeneTransgene copy numberPromoterExpression(CTG)*~n~*Genetic backgroundReferencesDM500 (DM300-328 line)Human DM1 locus (43 kb transgene)1Human *DMPK* (∼11.5 kb region upstream of main TSS)All DM1-related tissues (e.g. skeletal muscle, heart and CNS)500--600\>90% C57BL/6([@DDW042C23])DMSXL (DM300-328 line)Human DM1 locus (43 kb transgene)1Human *DMPK* (∼11.5 kb region upstream of main TSS)All DM1-related tissues (e.g. skeletal muscle, heart and CNS)∼1300\>90% C57BL/6([@DDW042C24])Tg26Human *DMPK* gene (14 kb transgene)∼25Human *DMPK* (∼1.9 kb region upstream of main TSS)All DM1-related tissues (e.g. skeletal muscle, heart and CNS)11FVB/*n*([@DDW042C25])*HSA*^LR^ (LR20b line)Human α-actin gene; CTG repeat inserted in 3′ UTR (7.1 kb transgene)2Human α-actin (∼2.1 kb region upstream of TSS)Skeletal muscle only220--250FVB/*n*([@DDW042C27])WTNo transgenen.a.n.a.n.a.n.a.\>90% C57BL/6n.a.[^3]
We demonstrate that, in comparison with expression of normal and mutant *DMPK* transcripts in patient cells, considerable variation exists in level and developmental timing of transgene expression in DM1 cell and animal models. A remarkable low level of expression with absolute numbers of, at most, a few dozen RNA molecules per cell was observed for *DMPK* transcripts in human samples. Our findings highlight the hitherto unrecognized involvement of low-abundance RNA molecules in DM1 pathophysiology, altering our current view on the RNA gain-of-function theory, which explains the role of repeat RNA in DM1 manifestation. We discuss the possible implications of our findings for future interpretation of data from fundamental and translational studies in which these DM1 models and patient cells will be used.
Results {#s2}
=======
Derivation of myogenic cell lines from DM1 mouse models {#s2a}
-------------------------------------------------------
Characteristics of mouse models included in this study are listed in Table [1](#DDW042TB1){ref-type="table"}. For profiling of transgene expression at the cellular level, we established conditionally immortalized myoblast populations from each model by pooling clones of individual cells derived from the calf muscle complex from double hemizygous mice carrying one transgenic DM1 allele and one H-2K^b^-tsA58 allele ([@DDW042C28]). As there is strong evidence that satellite cells from different inbred mice behave intrinsically differently ([@DDW042C29],[@DDW042C30]), it is important to note that crossings included different genetic backgrounds to generate the double hemizygous animals. The cell populations have therefore distinct mixed genetic backgrounds with contributions of C57BL6, FVB/*n*, CBA/Ca and C57BL/10. We do believe, however, that these differences have no major impact on transcriptome composition and therefore should not overtly confound our comparison.
An important feature of the immortalized myoblasts is that during prolonged passaging in culture under permissive conditions cells have the tendency to undergo polyploidization, as an effect of the presence of the temperature-sensitive SV40 large T-antigen ([@DDW042C31]). This leads to a mix of 2N and 4N cells in populations of the different lineages and to variation in the absolute abundance of individual transcripts per cell ([Supplementary Material, Fig. S1A](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). Hence, all possible care was taken to compare myoblasts from similar passage numbers.
DM500, DMSXL, *HSA*^LR^ and WT myoblast populations had normal morphological appearance and showed apparently normal proliferative capacity and terminal differentiation features upon shift to myogenesis-promoting conditions (Fig. [1](#DDW042F1){ref-type="fig"}). Multinuclear contractile myotubes appeared after 3--5 days of differentiation. The Tg26 cell population had normal morphological appearance, but showed slightly disordered cell alignment and diminished fusion capacity upon induction of differentiation. This may be due to overproduction of certain DMPK protein isoforms ([@DDW042C26],[@DDW042C32],[@DDW042C33]). Figure 1.Morphology of myoblasts derived from DM1 and control mouse models. Conditionally immortalized myoblasts were derived from the GPS muscle complex of DM1 mouse models. Myoblasts differentiated into contractile myotubes under low-serum conditions. Representative images of cultures during proliferation (Day -2), confluency (Day 0) and differentiation (Days 1, 3 and 7) are shown. Bar 100 µm.
To quantify gene expression during proliferation and differentiation of the myoblast populations, RNA was isolated at various time points and analyzed by northern blotting and reverse transcriptase--quantitative polymerase chain reaction (RT-qPCR). Progression of myogenic differentiation *in vitro* was accompanied by a clear increase in skeletal muscle α-actin (*Acta1*) mRNA (Fig. [2](#DDW042F2){ref-type="fig"}A), a well-known differentiation marker encoding a major constituent of the contractile apparatus ([@DDW042C34]--[@DDW042C36]). The increase was less profound in Tg26 cultures, in accordance with their diminished fusion capacity. Whereas *Acta1* was induced, β-actin (*Actb*) expression decreased during the 7-day differentiation period ([Supplementary Material, Fig. S2](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)) ([@DDW042C37]). Figure 2.Expression profile of proliferating and differentiating myoblasts derived from DM1 and control mouse models. Analysis of endogenous *Acta1* (**A**), transgene (**B**) and endogenous *Dmpk* (**C**) mRNA levels by northern blotting. Each hybridization signal was normalized to that of *18S* rRNA and then compared with the normalized value measured for the same (trans)gene in the GPS muscle from that particular DM1 mouse model. Data were obtained from two independent culture series per cell line, for which triplicate cultures per series were pooled and analyzed; bars represent mean + SEM. For profiling of endogenous *Dmpk* (C) in Tg26 samples RT-qPCR was used, because signals of transgenic *DMPK* and endogenous *Dmpk* overlapped on northern blot. Bars represent mean + SEM. Timing of differentiation on the *x*-axis refers to Figure [1](#DDW042F1){ref-type="fig"}.
For reference, primary human muscle cells obtained from a healthy individual and two DM1 patients were also included in our study. Myoblasts with healthy (CTG)5/(CTG)5 or disease-specific (CTG)21/(CTG)200 or (CTG)11/(CTG)760 repeat combinations appeared morphologically diverse, which can be explained by their different origin and culture history (Fig. [3](#DDW042F3){ref-type="fig"}). Multinucleated myotubes were formed during differentiation, but spontaneous contractions were never observed (Fig. [3](#DDW042F3){ref-type="fig"}, data not shown). RNA expression analysis corroborated this observation by showing that *ACTA1* expression was minimal at all time points measured and remained low in comparison with *Acta1* levels in DM500 GPS tissue (Fig. [4](#DDW042F4){ref-type="fig"}A). We, therefore, conclude that human cultures did not attain the same endpoint of terminal differentiation as mouse myoblasts under our *in vitro* conditions. Figure 3.Morphology of primary human DM1 and healthy myogenic cells. Human primary myoblasts aligned and fused under low-serum conditions. Representative images at different time points during proliferation (Day -2), confluency (Day 0), and differentiation (Days 1, 3 and 7) are shown. The number of CTG triplets for the two *DMPK* alleles are indicated for the one healthy (5/5) and two DM1 (21/200 and 11/760) cultures. Bar 100 µm. Figure 4.Expression profile of proliferating and differentiating primary human DM1 myogenic cells. Assessment of *ACTA1* (**A**) and *DMPK* (**B**) transcript levels by northern blot analysis. Signal strength in each sample was normalized to that of *18S* rRNA, as outlined in the legend of Figure [2](#DDW042F2){ref-type="fig"}. To facilitate direct comparison with DM1 mouse model-derived myoblasts, ratios were related to *Acta1* and *Dmpk* levels in DM500 GPS muscle. Data were obtained from two independent culture series per cell line, for which triplicate cultures per series were pooled and analyzed; bars represent mean + SEM. Stacked bar graphs show levels of normal-sized *DMPK* (black part) and expanded *DMPK* (white part) in patient-derived cultures. On average, the ratio expanded versus normal-sized *DMPK* transcripts was 0.63 for 21/200 and 0.84 for 11/760. Timing of differentiation on the *x*-axis refers to Figure [3](#DDW042F3){ref-type="fig"}.
Transgene expression differs between DM1 myoblast models {#s2b}
--------------------------------------------------------
Transgene expression was assayed in the myoblast populations during proliferation and differentiation (Fig. [2](#DDW042F2){ref-type="fig"}B and [Supplementary Material, Fig. S3](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). *DMPK* transgene expression in differentiating DM500 and DMSXL cells transiently increased 3- to 10-fold and then returned to basal levels in proliferating myoblasts. A modest increase in transgene expression was also observed in differentiating Tg26 myoblasts, but this change appeared to be more permanent. In contrast, transcript levels from the *Dmpk* gene hardly varied with differentiation state in any of the five myoblast populations (Fig. [2](#DDW042F2){ref-type="fig"}C). The observed differential regulation of endogenous/transgenic *Dmpk*/*DMPK* expression may be caused by (i) species differences between the structure and function of *Dmpk*/*DMPK* promoters, (ii) differences in the length of the region upstream of the TSS in the human transgenes used (Table [1](#DDW042TB1){ref-type="table"}), (iii) differences between chromosomal insert sites and the endogenous *Dmpk* locus or (iv) different epigenetic alterations across the transgenic loci. Influence of the (CUG)*~n~*-repeat length on RNA stability is unlikely, because temporal profiles of transgenic RNA expression in DM500 and DMSXL cells during myogenic differentiation levels were highly similar.
For *DMPK* expression in human myoblasts, we performed separate quantitative analysis of healthy and expanded transcripts, which migrate differentially on northern gels. Steady-state levels of normal and expanded *DMPK* RNA molecules were approximately similar for both the 21/200 and the 11/760 patient cell cultures (Fig. [4](#DDW042F4){ref-type="fig"}B). Comparison of total *DMPK* expression levels between healthy 5/5 and patient cell cultures showed variation. However, temporal profiles had a similar shape with a peak in expression during early differentiation, alike expression behavior of *DMPK* transcripts from the *DMPK* transgene in mouse myoblasts. This supports our idea that species-specific differences in the *DMPK*/*Dmpk* promoter explain the differential expression pattern of *DMPK* and *Dmpk* transcripts.
For better interpretation of *in vitro* and *in vivo* findings, we compared *DMPK* transgene expression in myogenic cultures to that in parental GPS muscle in the DM1 models. Transgene expression was ∼0.6- to 6-fold higher, 6- to 18-fold higher and 3- to 11-fold lower for DM500, DMSXL and Tg26 myoblast/myotube cultures, respectively (Fig. [2](#DDW042F2){ref-type="fig"}B). Of particular, interest was the expression profile of the expanded *ACTA1* transgene in differentiating myoblasts from the *HSA*^LR^ model, showing that transcripts were expressed only late in differentiation (Fig. [2](#DDW042F2){ref-type="fig"}B). Even in fully differentiated hemizygous *HSA*^LR^ myotubes *in vitro*, expanded *ACTA1* transgene expression remained ∼200-fold lower than in GPS muscle from homozygous *HSA*^LR^ mice.
Transgene expression in distinct DM1 mouse models differs over a 1000-fold range {#s2c}
--------------------------------------------------------------------------------
By extending our analysis to mouse muscle *in vivo*, using the GPS complex as RNA source and *18S* rRNA for normalization, our comparison revealed that *Acta1* expression was remarkably similar between models (Fig. [5](#DDW042F5){ref-type="fig"} and [Supplementary Material, Fig. S4A](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)), despite differences in their genetic background (Table [1](#DDW042TB1){ref-type="table"}). Also *Dmpk* expression showed only minor fluctuations. Whereas *Acta1* and *Dmpk* expression did not vary \>2- to 4-fold between mouse models, the mean level of *Acta1* RNA was ∼2,500-fold higher than that of *Dmpk* RNA. Figure 5.Transgene expression in DM1 mouse skeletal muscle. Expression of *Acta1*, *Dmpk* and transgenes was determined in GPS muscle of each DM1 mouse model. Transcripts were quantified by northern blotting using *18S* rRNA for normalization (*n* ≥ 4). Endogenous *Dmpk* RNA levels in DM500 GPS muscle were used as reference (set to one). Mean + SEM are shown. a.u., arbitrary units; n.d., not determined (*Dmpk* and transgenic signals overlap); n.a., not applicable.
We next compared transgene expression in GPS muscles from DM500, DMSXL and Tg26 mice. Quantification was straightforward, since all models expressed differentially expanded, but otherwise similar and intact *DMPK* transgenes. Transgenic *DMPK* RNA accumulated in muscle from hemizygous DM500 and DMSXL mice to a 10-fold lower level than RNA from the endogenous *Dmpk* genes (Fig. [5](#DDW042F5){ref-type="fig"}). In contrast, hemizygous Tg26 GPS muscle expressed 20- to 25-fold more *DMPK* mRNA than *Dmpk* mRNA \[assuming that *Dmpk* expression in Tg26 mice (FVB background) was similar to that in DM500 mice (C57BL/6 background)\]. Determination of relative levels of transgenic RNA in *HSA*^LR^ muscle appeared more challenging. Use of a (CAG)9 probe to detect (CUG)*~n~* segments, the only sequence shared between transgenic products, proved unreliable to compare RNAs with different repeat lengths (data not shown). Therefore, a balanced mix of *ACTA1* and *DMPK* probes, generated by random-primed labeling on cDNA templates of equal size was used instead (see the Materials and Methods' section). With this approach we found *ACTA1* (CUG)*~n~* RNA over 1000-fold higher expressed than *Dmpk* mRNA in homozygous *HSA*^LR^ mice (Fig. [5](#DDW042F5){ref-type="fig"}). This means that the concentration of expanded transcripts in *HSA*^LR^ muscle is extraordinarily high and similar to that of *Acta1* mRNA. This finding is perhaps not too surprising as both promoter and backbone of the human transgene and the *Acta1* gene share strong homology.
Analysis of *DMPK* RNA levels in human muscle allows direct interpretation of transgene dosage {#s2d}
----------------------------------------------------------------------------------------------
Skeletal muscle samples from healthy humans and DM1 patients were included in our study to extend our comparisons. *DMPK* mRNA expression in the samples varied, but was not \>3-fold higher than the level of *Dmpk* mRNA in mouse GPS tissue (Fig. [6](#DDW042F6){ref-type="fig"}). Expression of normal-sized and of expanded *DMPK* alleles were about equal in each of the patient samples. Since *DMPK* levels in human muscles and *Dmpk* levels in mouse muscle show high similarity, observations about transgene expression dosage effects in DM1 mice may be directly translated to relevance for DM1 patients. Any difference in *DMPK* mRNA content between human samples probably represents variation in fiber type and muscle origin or must be caused by differentiation or disease state. Note that, effects of ageing on repeat length heterogeneity were clearly visible on northern blot: in samples from adult DM1 patients a smear was observed for RNA from the expanded allele, whereas in congenital DM (CDM) patient material a defined signal was apparent (Fig. [6](#DDW042F6){ref-type="fig"}) ([@DDW042C38]). Figure 6.*DMPK* expression in human control and DM1 patient tissue. Northern blot (left) of human control and DM1 muscle tissue using a *DMPK* and a (CAG)9 repeat probe. Healthy *DMPK* mRNA appeared as a defined band in all tissues. Expanded *DMPK* mRNA appeared as a defined band in congenital DM1 tissue (CDM-e and -f), but as a smear (representing somatic mosaicism of repeat size) in adult DM1 tissue (DM1-c and -d). As size markers, we indicated the location of *DMPK* (CUG)700 and (CUG)1300 transcripts. *DMPK* transcript levels, normalized to those of *18S* rRNA, were plotted relative to *Dmpk* transcript levels in DM500 GPS muscle to allow for comparison with DM1 mouse models (right). Stacked bar graphs show levels of normal-sized *DMPK* (black part) and expanded *DMPK* (white part) in patient-derived samples. The ratio expanded versus normal-sized *DMPK* transcripts was 0.33 in DM1-c; 0.92 in DM1-d; 0.84 in CDM-e and 1.79 in CDM-f.
Multi-pronged analysis reveals low copy number of expanded *DMPK* transcripts {#s2e}
-----------------------------------------------------------------------------
For further stoichiometric and pathomechanistic considerations, we decided to determine the absolute number of *DMPK* mRNA copies per cell. Quantification on northern blot, after normalization for signal strength and correction for probe length, revealed that *DMPK/Dmpk* expression was 400- to 2000-fold lower than that of *ACTB/Actb* in proliferating human/mouse myoblasts (Fig. [7](#DDW042F7){ref-type="fig"}A and [Supplementary Material, Fig. S4B](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). Based on known values for the copy number of *ACTB/Actb* mRNA, which ranges from 350 to 8000 per cell ([@DDW042C39]--[@DDW042C42]) ([Supplementary Material, Table S1](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)), we inferred that the absolute number of *DMPK* transcripts must be in the range of 1--20 per cell. In patient myoblasts, this population consists of approximately equal numbers of normal-sized and expanded *DMPK* transcript molecules (Figs. [4](#DDW042F4){ref-type="fig"} and [7](#DDW042F7){ref-type="fig"}A). Figure 7.*DMPK* RNA copy number in human and mouse myoblasts. The absolute number of *DMPK* transcripts per cell was calculated based on experimental data from northern blotting (A), RT-qPCR (B) and RNA-sequencing (C). (**A**) Stacked bar graph showing *DMPK/Dmpk*:*ACTB/Actb* transcript ratios determined from signal strengths on northern blots for human myoblast lines 5/5, 21/200 and 11/760, and mouse myoblasts lines WT and DM500 ([Supplementary Material, Fig. S4B](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). (**B**) *DMPK* transcript copy number in 11/760 and DM500 myoblasts was determined on the basis of known amounts of synthetic *DMPK* RNA fragments in RT-qPCR ([Supplementary Material, Fig. S5](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). Note that, normal-sized and expanded *DMPK* transcripts could not be measured independently. (**C**) *DMPK*:*GAPDH* ratio based on RNA-sequencing signal of healthy and DM1 TA and quadriceps (Quad) tissue ([www.dmseq.org](www.dmseq.org)). Each data point represents a single tissue. Mean + SEM (A and B), or mean only (C) are shown. Descriptions d-2 and d0 refer 938 to Figures [1](#DDW042F1){ref-type="fig"} and [3](#DDW042F3){ref-type="fig"}.
To validate this rough estimate, *DMPK* RNA copy number was quantified using two *in vitro* transcribed *DMPK* RNAs as standard references in RT-qPCR ([Supplementary Material, Fig. S5](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). We found 20--25 *DMPK* transcripts per proliferating 11/760 myoblast and 45--50 *DMPK* molecules in myoblasts just prior to the onset of myogenic differentiation (Fig. [7](#DDW042F7){ref-type="fig"}B). This number includes both healthy and expanded *DMPK* transcripts. On average about four expanded *DMPK* transcripts per cell were detected in DM500 myoblasts.
Thirdly, we used RNA-sequencing data of healthy and DM1 skeletal muscle (available via [www.dmseq.org](www.dmseq.org)) to estimate *DMPK* transcript copy number. We calculated the *DMPK*:*GAPDH* ratio based on RNA-sequencing signal and found that *DMPK* mRNA molecules were 60- to 160-fold less abundant than *GAPDH* mRNA (Fig. [7](#DDW042F7){ref-type="fig"}C). Based on current estimates for the prevalence of *GAPDH* RNA, which is in the order of 250--2900 molecules per cell ([@DDW042C39],[@DDW042C42]--[@DDW042C45]) ([Supplementary Material, Table S2](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)), *DMPK* transcript copy number must be between 2 and 50 per cell.
Finally, we determined the number of fluorescent *in situ* hybridization (FISH)-detectable RNP complexes in mouse and human myoblasts using a Cy3-labeled (CAG)7 probe. Initial analyses demonstrated that the number of FISH signals slightly varied between myoblast pools and experiments and was influenced by culture conditions, such as cell density, feeding regime or changes in the culture medium (data not shown). This variation in foci number appeared suppressible by maintaining a very strict scheme for cell culture. On average, 5--9 foci per cell were detected in DM500 myoblasts (Fig. [8](#DDW042F8){ref-type="fig"}A and B). The number of foci differed between diploid and tetraploid DM500 myoblasts, with averages of ∼5 and ∼9 foci per cell, respectively ([Supplementary Material, Fig. S1B](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)). For reasons unknown, we also consistently observed signals in WT mouse myoblasts (Fig. [8](#DDW042F8){ref-type="fig"}A and B). These signals could not be distinguished from expanded *DMPK* (CUG)*~n~* mRNA signals and may also be present in other cell types from mouse. In cells from human origin, we did not observe overt background signals, so the problem may be species specific. To deal with this issue, we decided to apply a correction by defining a background interval in Figure [8](#DDW042F8){ref-type="fig"}. After this correction, an average of two transgenic expanded (CUG)*~n~*-specific foci per cell remained in DM500 myoblasts. Foci counts in Tg26 and *HSA*^LR^ myoblasts were within the background range, as expected, based on the idea that the (CTG)11 repeat in Tg26 RNA cannot yield sufficient signal and our finding that the *HSA*^LR^ transgene is barely expressed in proliferating myoblasts. In 11/760 patient myoblasts 3--4 foci per nucleus were detected, whereas no signal was observed in 21/200 and 5/5 myoblasts (Fig. [8](#DDW042F8){ref-type="fig"}C and D), obviously caused by lack of FISH-signal strengths by the limited length of the repeat target. Figure 8.(CUG)*~n~* foci number in DM1 mouse and human myoblasts. Representative RNA FISH images using a (CAG)7 oligo probe (**A** and **C**) and quantification (**B** and **D**) of (CUG)*~n~* foci in DM1 mouse and human myoblasts. (**A** and **B**) RNA FISH signal in WT and DM500 myoblasts. In WT mouse myoblasts foci were observed which cannot be specific for the transgenic expanded (CUG)*~n~* transcripts. As these signals may also be present in cell lines derived from other mouse lineages, including the transgenic models, we defined a level of uncertainty, plotted as a shaded area in (B). Thus, an average of two expanded (CUG)*~n~*-specific foci per cell remained in DM500 myoblasts and no specific foci were identified in Tg26 and *HSA*^LR^ myoblasts. (C and D) RNA FISH on human healthy (5/5) and DM1 (21/200 and 11/760) myoblasts. Foci were only detected in 11/760 cells. Each data point represents foci number in one nucleus; mean is plotted in graph. Bar 20 µm. \*\*\**P* \< 0.001.
All data combined, we conclude that mouse and human myoblasts each contain up to half a dozen foci. We have to keep in mind that FISH protocols may detect transcripts with an efficiency of only 30--50% ([@DDW042C41]), for example, due to inaccessibility of the transcripts to the probe or loss of RNA molecules from the fixed cells during washing. Thus, the actual number of RNP complexes that appear as foci may be 2- to 3-fold higher. Given our estimates for the absolute number of expanded transcripts, our data suggest that every RNP complex that forms a FISH-visible aggregate is nucleated by one or only few expanded (CUG)*~n~* transcripts.
Discussion {#s3}
==========
DM1 is considered a prototypical RNA-dominant disorder, because its neurodegenerative and myopathic manifestation is thought to be based on processes wherein repeat-containing RNAs play a crucial role. Much of our current knowledge on the presumed toxic role of RNA in DM1 pathophysiology originates from direct comparison of findings in mutation carriers with findings in transgenic animals or genome edited cells. This is often done with simple bypassing of the fact that models may differ profoundly in nature, structural organization and chromatin context of the transgene and in length of the contained (CTG·CAG)*~n~* repeat ([@DDW042C13],[@DDW042C14]). A major unsolved aspect of DM1 etiology is how repeat length and dose and nature of abnormally expanded RNA transcripts affect onset and complexity of disease manifestation and its rate of progression and severity. Quantitative studies of transcript production could thus help to explain differential experimental findings with distinct DM1 models and add to conceptual progress. Here, we used gene expression profiling to compare relative abundance and absolute copy number of expanded repeat RNAs between muscle cells from patients and mouse models that are among the most commonly used in the DM1 field.
Our analyses unveiled differences in timing of production during myogenic differentiation as well as in the abundance of (CUG)*~n~* transcripts between DM1 mouse model and patient muscle cells. Expression of *DMPK* transgenes in DM500, DMSXL and Tg26 mice is at a basal level in proliferating myoblasts and peaks early in differentiation in a manner similar to the profile of endogenous *DMPK* expression in human myoblasts. Increase in *DMPK* expression at the start of myogenic differentiation was earlier reported for human cells ([@DDW042C46]) and also seen for C2C12 mouse myoblasts ([@DDW042C47]). Our group ([@DDW042C25]) and others ([@DDW042C32]) have demonstrated that the actual onset of *DMPK* expression is already seen in somites in the developing embryo, well before the actual commitment to specific muscle cell fate and the onset of myogenesis.
DM500/DMSXL- and Tg26-derived myoblasts carry transgenes of different size that may lack regulatory elements upstream of the *DMPK* TSS, but the proximal promoter sequence and the enhancer sequence in Intron 1 (elements responsive to MyoD via conserved E-boxes ([@DDW042C48])), are present. Thus, regulatory sequences from the human locus may drive 'near-natural' behavior of these transgenes and control the peak-shaped up-regulation with transient increase in transcriptional activity during early differentiation. The 'human-like' regulation of the *DMPK* transgene renders these mouse myoblasts useful models for study of toxic effects of normal or expanded *DMPK* RNA in early development. Others have demonstrated that early-onset expression of repeat-containing RNA may influence phenotypic severity by affecting proper tissue development in animals *in vivo* ([@DDW042C16]--[@DDW042C19]).
In *HSA*^LR^ myoblasts, expression of repeat-containing RNA was hardly detectable until late in differentiation *in vitro*. Conspicuously, the *Acta1* gene, whose human counterpart served as the recipient body for the transgene with (CTG)250 repeat insertion, showed an earlier onset of expression in differentiating mouse cells. The explanation for this differential behavior may thus be that the proximal promoter is present in the transgene, but that a *cis*-regulatory module in a region that activates transcription in differentiating myoblasts, \>20 kb downstream of the basal promoter ([@DDW042C49]), is lacking.
We found pronounced differences in the abundance of (CUG)*~n~*-repeat RNAs from transgenes and RNA products from the endogenous *Dmpk* gene and between expanded RNAs from cell lineages and the transgenic mouse tissues from which they originate. For proliferating cells, comparison was only meaningful for Tg26, DM500 and DMSXL myoblasts, as *HSA*^LR^ myoblasts essentially lacked expression of expanded (CUG)*~n~* RNA. As explained by the presence of multiple copies of the *DMPK* gene in the transgenic insert in Tg26 cells, *DMPK* levels were relatively high. Transcripts from the *DMPK* transgene in DM500 and DMSXL cells were ∼10-fold lower than transcripts from the *Dmpk* gene in these myoblasts. Even if we take into account that these observations reflect the production rate from one *DMPK* transgene versus two *Dmpk* genes in the DM500 cell model, transgenic RNA production is still ∼5-fold lower. Intrinsic differences in regulatory capacity of *DMPK* and *Dmpk* promoters may partly explain this observation, but we cannot exclude involvement of progressive epigenetic alterations, like DNA methylation or heterochromatization of the transgene or the *DMPK* promoter. Brouwer *et al.* ([@DDW042C50]) observed decreased transgene expression with age in the DM300-328 lineage, the ancestral mouse model from which DM500 and DMSXL mice and thus DM500 and DMSXL cells originate. *Cis* effects of repeat presence may have contributed to methylation state alterations in the *DMPK* transgene ([@DDW042C50]). Also transient propagation of mice in the homozygous state may have triggered partial gene silencing even before the myoblasts were derived *in vitro* ([@DDW042C51]--[@DDW042C53]). Finally, we cannot exclude effects of experimental handling, for example, by the use of dimethylsulfoxide during repeated cell freezing, which can affect the epigenetic state and is a known modulator of differentiation programming ([@DDW042C54]). Although we did not systematically study expression levels over time, we have the impression that *DMPK* expression in DM500 cells decreases with increasing passage number \[compare data shown in ([@DDW042C55])\]. This could point to the influence of epigenetic silencing, explaining also variability in observations of different research groups in the DM1 field.
Also in muscle tissue obtained from hemizygous DM500 and DMSXL mice *DMPK* transcripts were 10-fold less abundant than *Dmpk* RNAs, confirming the observation by Huguet *et al*. ([@DDW042C56]). The overall low level of transgene expression may explain why disease manifestation in these models is relatively mild, even in homozygous animals. Conversely, however, one can also argue that the presence of these few expanded RNAs is already sufficient to cause mild myotonia, slow progressive muscle weakness, nuclear (CUG)*~n~* foci and splicing abnormalities ([@DDW042C1],[@DDW042C14],[@DDW042C57]), for which compelling evidence was provided in various studies.
Our most conspicuous observation was that *ACTA1* (CUG)*~n~* expression in GPS muscle of homozygous *HSA*^LR^ mice was over 1000-fold higher than that of *Dmpk*, and remarkably similar to *Acta1* expression. The high (CUG)~n~ RNA expression may explain why the *HSA*^LR^ mouse lineage has by far the most severe DM1-like phenotype of all models, including myotonia, extensive splice abnormalities and a conspicuous abnormal density of intranuclear foci, all limited to skeletal muscle ([@DDW042C1],[@DDW042C14],[@DDW042C27]).
The use of expression ratios between transgenic and endogenous transcripts (e.g. *Dmpk*, *Acta1*, *Actb* and *18S* RNA) helped us to compare repeat RNA levels between muscle cells and mouse and human tissues, but we reasoned that the knowledge of absolute transcript copy numbers in muscle would be mechanistically more informative. Absolute numbers of molecules can only be meaningfully assessed if studied on a per cell basis. Analysis of *DMPK* expression in the three human myoblast lineages with no, intermediate or long repeat expansions confirmed that transcripts from the wild-type and mutant allele are about equally abundant. We found no evidence for an inverse correlation between repeat the length and RNA expression level ([@DDW042C46],[@DDW042C58]). Because endogenous *DMPK/Dmpk* transcripts appeared equally abundant in human and mouse myoblasts, quantitative comparison between patients and DM1 models was straightforward. Estimates for RNA expression ratios of *ACTB* and *GAPDH*, data from northern blot and RNA-sequencing analysis and absolute RT-qPCR experiments all point to the presence of at most 50 *DMPK* transcripts per cell. We thus propose that in patient cells \<25 are expanded (CUG)*~n~* transcripts. In myoblasts from the DM500 and DMSXL mouse models around four RNA molecules per cell originate from the *DMPK* transgene.
Estimates for *DMPK* transcript numbers correspond remarkably well with the observed foci number in DM500 cell nuclei. Also in patient myoblasts the number of foci per nucleus and the number of *DMPK* transcripts per cell are in the same order of magnitude. Other research groups, studying different cell types from DM1 patients, have reported similar amounts of foci from their FISH experiments ([@DDW042C59]--[@DDW042C62]). Different and variable amounts of foci were observed for *MyoD*-transduced DM1 fibroblasts, which may be best explained by the distinct MyoD induction of *DMPK* expression, caused by differential vector systems or *MyoD* promoters ([@DDW042C60],[@DDW042C63]). We, therefore, predict that RNP complexes that assemble in (CUG)*~n~* foci contain, on average, one to maximally six expanded transcripts. Along the same line of arguments, we propose that RNP aggregate formation is nucleated by one or very few expanded RNA molecules, thereby acting as individual entities. Phase transition of RNP complexes to insoluble aggregates (which appear as foci after FISH or MBNL antibody staining) may therefore not require further fusion with additional naked or RNP-decorated RNAs, but could be merely a protein-based event.
Many mechanistic avenues are now awaiting further exploration as our findings have important implications for the RNA-gain-of-function hypothesis that is currently in use for explaining DM1 features. How low numbers of only 1--25 abnormally folded RNA scaffolds can have a negative impact on ribostasis regulation---via titration of protein molecules from the total available pools of RNP proteins---is only one of the mechanistic questions that must be answered. The extent of loss of function of proteins may determine the gain of function toxicity of mutant RNA---and ultimately control the extent of cell stress caused by the DM1 mutation. Whether there is a direct relationship with the magnitude of temporal or permanent sequestration from the cellular pool of these RNP proteins is still difficult to answer. Stochastic events could be involved, as formation of an abnormal type of---microscopically visible or invisible diffusive---RNP aggregate on a repeat-containing transcript ([@DDW042C6],[@DDW042C7]) might cause temporal or permanent perturbation of a specific nuclear pathway for mRNA processing or transport. Dominant effects of low-abundance triplet repeat RNA on RNA splicing and polyadenylation may be expected if effects on inhomogeneity and compartmentalization of protein factor pools (e.g. MBNL isoforms) in different nuclear trajectories for mRNA RNP processing play a decisive role. Finally, amplification of toxic effects by rare triplet repeat RNAs may occur otherwise: various studies have demonstrated that activation of stress-signaling cascades that involve GSK3β ([@DDW042C64]), PKC ([@DDW042C65]), double-stranded RNA-dependent protein kinase PKR ([@DDW042C66]) or sensors of foreign RNA that normally initiate immune responses ([@DDW042C67]) is involved in disease manifestation in DM1. Whatever the exact course of cellular events in these scenarios, even with low-expressed, abnormally expanded repeat RNAs the probability of exceeding the toxic threshold increases over time for any permanently expressing cell in the tissue population. Ultimately, this may result in accumulation of stress in an increasing number of cells, and progressive cell loss and loss of function of tissue during ageing.
In a parallel study, we have recently found that expression of another possible player, a DM1 antisense transcript carrying a (CAG)*~n~* repeat, is even 10-fold lower than levels observed for *DMPK* (Gudde *et al*. manuscript in preparation). Thus also here, the question remains how such an extremely rare transcript could contribute to disease manifestation. One unifying answer may be that the rare *DMPK* sense RNA and the even rarer antisense transcript both serve as templates for RAN translation ([@DDW042C9],[@DDW042C10]). Homopolymeric protein products thus formed could possibly contribute to disease manifestation via initiation of aggregation of metastable proteins, initiating a prion-like cascade of events ([@DDW042C68]--[@DDW042C70]). Again, further analysis of steady-state levels of *DMPK* (CUG)*~n~* and antisense transcripts in cycling or resting cells and study of the frequency of use of RAN translation on these rare RNAs is necessary to better understand this possible distinct aspect of DM1 etiology.
Our findings highlight that careful choice of cellular and animal model systems that take structural properties of transgenes and transgenic products into account in combination with quantitative modeling is imperative for such studies. For this work not only knowledge of the 'per cell' presence of individual endogenous and mutant RNA molecules is important, but also stoichiometric considerations on the binding of MBNL1--3, Staufen, DDX, HnRNP or other RNP proteins ([@DDW042C7]) by one (CUG)*~n~* repeat tract are needed to better understand nucleation and phase transition events in abnormal aggregates that could form around expanded DMPK mRNAs. Whatever the molecular mechanism involved, based on our findings, we propose that DM1 is caused by anomalous behavior of only very few mutant RNA molecules per muscle cell.
Materials and Methods {#s4}
=====================
Human material {#s4a}
--------------
Human skeletal muscle samples were obtained from pre-consented post-mortem donors for research purposes in accordance with local guidelines in The Netherlands more than 18 years ago. No additional approval by an ethics committee was required at that time. Muscle autopsies were obtained from patients with a confirmed clinical and DNA diagnosis of adult-onset DM1 \[male DM1-c, 55 years, psoas muscle (CTG)5/(CTG)exp; female DM1-d, 65 years, quadriceps muscle (CTG)16/(CTG)exp\] and congenital DM1 \[male CDM-e and -f, 14 days, sternocleidomastoideus and gastrocnemius muscle (CTG)12/(CTG)1300\]. All tissues were snap frozen immediately after collection and stored at −135 to 80°C until further use. As control, skeletal muscle samples from healthy anonymous donors \[healthy-a and -b (CTG)11/(CTG)12 and (CTG)5/(CTG)11, respectively\] from our own repository were included in our study. These samples were collected long before the current guidelines for written consent were enforced and no detailed information could be traced.
Mice {#s4b}
----
Mice were housed and procedures performed with approval of the Animal Ethics Committee of Radboud University Nijmegen (Permit number: RU-DEC 2014-099). Characteristics of the mouse models are summarized in Table [1](#DDW042TB1){ref-type="table"}. DM500 and DMSXL mice both originated from the DM300-328 lineage ([@DDW042C23]). Due to intergenerational triplet instability the repeat in the *DMPK* gene grew to 500--600 and ∼1300 CTG triplets in DM500 and DMSXL mice, respectively ([@DDW042C24]). For isolation of immortalized myoblasts, each of the DM1 mouse models was crossed with hemizygous H-2Kb-tsA58 transgenic mice (ImmortoMouse^®^, Charles River Laboratories) ([@DDW042C71]), harboring the gene for thermolabile TAg from SV40. Nine-day old pups carrying one *DMPK* or *ACTA1* transgene copy and one *H-2Kb-tsA58* transgene copy were selected and used for myoblast generation.
Cell culture {#s4c}
------------
Conditionally immortalized myoblasts from the gastrocnemius--plantaris--soleus (GPS) muscle of DMSXL, Tg26 and *HSA*^LR^ mice after crossing with H-2Kb-tsA58 mice were derived as described ([@DDW042C28],[@DDW042C55],[@DDW042C71]). Individual myoblast lineages were obtained by ring cloning and selected for myotube formation ability. Myoblast populations used in this study were established by forming equal mixtures of five cell clones for DMSXL, seven cell clones for Tg26 and four cell clones for *HSA*^LR^. DM500 and WT myoblasts were derived previously ([@DDW042C55]).
Myoblasts were grown on 0.1% (w/v) gelatin-coated culture dishes in the proliferation medium containing Dulbecco\'s modified Eagle\'s medium (DMEM) (Gibco) supplemented with 20% (v/v) fetal bovine serum (PAA Laboratories), 4 m[m]{.smallcaps} [l]{.smallcaps}-glutamine (Gibco), 1 m[m]{.smallcaps} pyruvate (Sigma), 50 µg/ml gentamicin (Gibco), 20 units/ml γ-interferon (BD Biosciences) and 2% (v/v) chicken embryo extract (Sera Laboratories International) at 7.5% CO~2~ and 33°C. Differentiation to myotubes was induced by placing myoblasts, grown to confluency on Matrigel (BD Biosciences), in the differentiation medium containing DMEM supplemented with 5% (v/v) horse serum and 50 µg/ml gentamicin at 7.5% CO~2~ and 37°C. Differentiation conditions were maintained for a maximum of 7 days. Spontaneous contractions started to appear around Day 3.
Human myoblast cultures isolated from skeletal muscle of fetuses, one healthy control (5/5) and two DM1 affected lines \[21/200 with (CCG·CGG)*~n~* interruptions in the 3′ end of the expanded (CTG·CAG)*~n~* repeat; data not shown\] and 11/760 were a gift of Dr Furling *et al.* ([@DDW042C46]). Myoblasts were maintained on 0.1% (w/v) gelatin-coated culture dishes in proliferation medium containing Ham\'s F10 medium (Gibco) supplemented with GlutaMAX, 20% bovine growth serum (Thermo Scientific) and 25 µg/ml gentamicin at 7.5% CO~2~ and 37°C. Differentiation to myotubes was induced by placing confluent myoblast cultures in the differentiation medium containing DMEM supplemented with 4 m[m]{.smallcaps} [l]{.smallcaps}-glutamine, 1 m[m]{.smallcaps} pyruvate, 10 μg/ml insulin (Sigma), 100 μg/ml apo-transferrin (Sigma) and 25 μg/ml gentamicin.
Phase contrast images of cell cultures were taken with a Zeiss Axiovert 35 M light microscope, 10×/0.30 objective.
Fluorescence-activated cell sorting {#s4d}
-----------------------------------
Immortalized WT mouse myoblasts were detached from the culture surface using trypsin, washed in phosphate-buffered saline (PBS) and fixed in 70% (v/v) ethanol in PBS overnight at −20°C. After two times PBS wash, cells were stained in 20 µg/ml propidium iodide, 0.2 mg/ml RNase A, 0.1% Triton X-100 in PBS for 15 min at 37°C. Fluorescence-activated cell sorting analysis was performed on the Beckman Coulter Epics Altra cell sorter. Immortalized DM500 myoblasts were cultured in proliferation medium at 37°C for 24 h, detached and incubated with 3 µg/ml Hoechst 33342 at 37°C for 30 min and then sorted. Diploid and tetraploid DM500 cells were seeded on gelatin-coated glass cover slips, cultured for another 24 h at 37°C and analyzed by FISH.
RNA isolation {#s4e}
-------------
RNA from cultured cells was isolated using the Aurum Total RNA Mini Kit (Bio-Rad), according to the manufacturer\'s protocol. RNA from muscle tissue was isolated using TRIzol reagent (Invitrogen), according to the manufacturer\'s protocol.
Northern blotting {#s4f}
-----------------
Northern blotting was performed according to standard procedures. RNA was subjected to electrophoresis in a 1.2% agarose gel under denaturing conditions. Depending on sample availability, 5--12 µg RNA was loaded per lane. RNA was transferred to Hybond-XL nylon membrane (Amersham Pharmacia Biotech) by capillary transfer in 10× SSC and hybridized with random-primed ^32^P-labeled cDNA-based probes or ^32^P-end-labeled oligonucleotides: a 2.6 kb *DMPK* cDNA (covering the entire ORF and 3′ UTR) ([@DDW042C55]) and the complete 1.9 kb human *18S* rRNA cDNA were used as templates in random priming reactions. A (CAG)9 oligo ([@DDW042C55]), a *Acta1* oligo (5′-ACCCTGCAACCACAGCACGATTGTCGATTG-3′) ([@DDW042C72]) and a mixture of three oligos complementary to both *Acta1* and *ACTA1*(5′-GCGGTGGTCTCGTCTTCGTCGCACAT-3′, 5′-TGGCATACAGGTCCTTCCTGATGTCGATGTC-3′, 5′-GCCTCGTCGTACTCCTGCTTGGTGATCC-3′) were 5′-end labeled. The *Acta1/ACTA1* oligo mix and the *DMPK* cDNA probe were used to quantify *ACTA1* and *DMPK* mRNA levels, respectively, in human samples. In murine samples, the single Acta1 oligo, *DMPK* cDNA and (CAG)9 oligo probes were used to quantify *Acta1*, *Dmpk* and transgenic *DMPK* (DM500 and DMSXL) and transgene (Tg26 and *HSA*^LR^) mRNA levels, respectively. Blots were washed and exposed to X-ray film (Kodak, X-OMAT AR). Quantification of signals was performed by phosphor-imager analysis (Molecular Imager FX, Bio-Rad) and analyzed with Quantity One (Bio-Rad) and FIJI ([@DDW042C73]) software. *18S* rRNA levels were used for normalization.
To be able to compare transgene expression in the DM1 mouse models, carrying different transgenes (i.e. *DMPK*- or *ACTA1*-type), a mix of *ACTA1* cDNA probe (1.3 kb, Exon 2--7, 27% G-content) and *DMPK* cDNA probe (1.3 kb, Exon 3--12, 34% G-content) was used. Both genes are ∼90% identical between mouse and human, so simultaneous detection of endogenous and transgenic transcripts was possible. \[α-^32^P\]-dCTP label was diluted with non-radioactive dCTP for labeling of the *ACTA1* probe to obtain comparable signal strengths for simultaneous phosphor-imager detection of *Acta1/ACTA1* and *Dmpk/DMPK* on the same blot. Signals were corrected for label dilution and G-content in the two probes. Since *ACTA1* and *Actb* are 84% identical, the *ACTA1* probe was also used for detection of *Actb*. A similar method was used to compare *DMPK*/*Dmpk* and *ACTB*/*Actb* expression. A 1.3 kb *ACTB* cDNA probe was used (Exon 1--6, 27% G-content), which has a 90% identical sequence with the mouse variant.
*In vitro* transcription {#s4g}
------------------------
DNA templates corresponding to *DMPK* regions Exon 1--6 (0.6 kb) and Exon 11--15 (1.0 kb) were generated by PCR using primers 5′-GAATTTAATACGACTCACTATAGGGAGAGGCGGCTCCAGCAGC-3′ and 5′-CCGCAGCTTGA-GGCAAGAG-3′, and 5′-GAATTTAATACGACTCACTATAGGGAGATGGAGGCCATCCG-3′ and 5′-GTCCTGTAGCCTGTCAGCGA-3′, respectively (T7 promoter underlined). DNA purity and sequence identity were confirmed by sequencing. For *in vitro* transcription, 200 ng of the templates was used in the MEGAscript T7 Transcription Kit (Ambion) following manufacturer\'s instructions. RNA products were purified using the Aurum Total RNA Mini Kit (Bio-Rad). Transcript length, purity and concentration were verified by gel electrophoresis and absorbance at 260/280 nm (NanoVUE spectrophotometer, GE Healthcare Life Sciences). Known amounts of RNA fragments *DMPK* Exon 1--6 and Exon 11--15 were mixed with WT mouse total RNA (as carrier RNA) and used as standards in RT-qPCR.
Reverse transcriptase--quantitative polymerase chain reaction {#s4h}
-------------------------------------------------------------
Using 500 ng RNA template per reaction and random hexamers as primers, cDNA synthesis was performed with SuperScript™ II Reverse Transcriptase (Invitrogen) or iScript™ cDNA Synthesis Kit (Bio-Rad). For RT-qPCR, 3 µl of 10--100-fold diluted cDNA preparation was mixed in a final volume of 10 µl containing 5 µl 2× Sybr Green mix (Roche Applied Science) or 5 µl iQ™ SYBR^®^ Green Supermix (Bio-Rad) and 4 pmol of each primer: *DMPK* Exon 1--2: 5′-ACTGGCCCAGGACAAGTACG-3′ and 5′-CCTCCTTAAGCCTCACCACG-3′; *DMPK* Exon 15(5′): 5′-AGAACTGTCTTCGACTCCGGG-3′ and 5′-TCGGAGCGGTTGTGAACTG-3′; *ACTA1*: 5′-CACCTCCAGCACGCGACTT-3′ and 5′-CGATGGCAGCAACGGAAGTTGT-3′; *Dmpk*: 5′-TTTTGAAGGTGATCGGGCGTG-3′ and 5′-CCTCTCTTCAGCATGTCCCACTTA-3′; *18S* rRNA: 5′-GTAACCCGTTGAACCCCATT-3′ and 5′-CCATCCAATCGGTAGTAGCG-3′. Samples were analyzed using the CFX96 Real-time System (Bio-Rad). A melting curve was obtained for each sample in order to confirm single product amplification. cDNA samples from no template control and no reverse transcriptase control (RT-) were included as negative controls. Transgene (*DMPK* and *ACTA1*) and endogenous *Dmpk* mRNA levels were normalized to *18S* rRNA expression.
Fluorescent *in situ* hybridization {#s4i}
-----------------------------------
Human myoblasts (5/5, 21/200 and 11/760) and mouse myoblasts (DM500, Tg26, *HSA*^LR^ and WT) grown on gelatin-coated glass cover slips were washed once with PBS and fixed in 4% (w/v) formaldehyde, 2 m[m]{.smallcaps} MgCl~2~ in PBS for 10 min at room temperature. Cells were stored under 70% ethanol at 4°C. After two times PBS wash, cells were pre-hybridized in 40% (v/v) deionized formamide in 2× SSC for 20 min at room temperature, followed by overnight hybridization with a 0.1 ng/μl Cy3-(CAG)7 probe (2′-*O*-methyl phosphorothioate-modified) in 40% deionized formamide, 10% (w/v) dextran sulfate, 0.1% Triton X-100, 1 mg/ml herring sperm DNA, 100 μg/ml yeast tRNA, 0.2% (w/v) BSA, 2 m[m]{.smallcaps} VRC, 2× SSC in a humidified chamber at 37°C. After two times PBS wash, cells were counterstained with 0.33 μg/ml DAPI in PBS for 10 min at room temperature, followed by a PBS wash, dehydration in methanol and mounting in Mowiol. Images were acquired using the Zeiss Axiophot2 Fluorescence microscope or Olympus IX-71 wide field fluorescence microscope.
RNA-sequencing {#s4j}
--------------
RNA-sequencing data from healthy and DM1-affected human tissue samples \[tibialis anterior (TA), quadriceps and heart\] were obtained from [www.dmseq.org](www.dmseq.org). RNA-sequencing signals of full-length *DMPK* and *GAPDH* were used as an independent measure for transcript abundance.
Statistical analysis {#s4k}
--------------------
Endogenous and transgenic mRNA levels were compared between time points and models, using a one-way analysis of variance (ANOVA) test, followed by a post-test for linear trend or Tukey\'s multiple comparison test. All values in graphs are presented as mean±SEM. Foci counts were compared between different myoblast lines with a one-way ANOVA test and Tukey\'s multiple comparisons post-test. Data are visualized as scatter plot, each dot representing one observation, and the mean is shown. Statistical analyses were performed with GraphPad Prism version 5.01 for Windows.
Supplementary Material {#s5}
======================
[Supplementary Material is available at *HMG* online.](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw042/-/DC1)
Funding {#s6}
=======
This work was supported by the Prinses Beatrix Spierfonds (grant number W.OR10-04; with contribution from the Stichting Spieren voor Spieren) and by ZonMw (TOP grant NL91212009). Funding to pay Open Access charges was provided by both organizations and Radboudumc.
Supplementary Material
======================
###### Supplementary Data
We thank Geneviève Gourdon (Paris, France) for the DM500 and DMSXL mouse lines, Charles Thornton (Rochester, NY, USA) for the *HSA*^LR^ line and Denis Furling (Paris, France) for human myoblasts. We thank the founders and contributors of the Myotonic Dystrophy Deep Sequencing Data Repository available via[www.dmseq.org](www.dmseq.org). We thank members of the Department of Cell Biology for discussions.
*Conflict of Interest statement*. None declared.
[^1]: A.G.B and W.J.A.A.v.d.B. contributed equally to this work.
[^2]: B.W. and D.G.W jointly supervised this work.
[^3]: TSS, transcription start site; n.a., not applicable.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-nutrients-12-01609}
===============
Cardiovascular (CV) disease is the most common cause of morbidity and mortality in patients with chronic kidney disease (CKD) \[[@B1-nutrients-12-01609],[@B2-nutrients-12-01609]\].
The high CV risk may be due, at least in part, to the excess of vascular calcifications (VC) observed even in very young dialysis patients, who lack the typical CV risk factors such as hypertension, dyslipidemia, and smoking \[[@B3-nutrients-12-01609],[@B4-nutrients-12-01609],[@B5-nutrients-12-01609]\]. The presence of abdominal aortic calcifications is significantly associated with all-cause and cardiovascular mortality in maintenance hemodialysis (HD) patients \[[@B2-nutrients-12-01609]\].
Although arterial medial calcification is the most represented, both intimal and medial calcifications coexist in CKD patients \[[@B3-nutrients-12-01609]\]. Medial and intima calcification are active processes that share some triggering factors: mineral bone disorders, inflammatory status, humoral factors, and phenotypic switch of resident or circulating precursor \[[@B3-nutrients-12-01609],[@B6-nutrients-12-01609],[@B7-nutrients-12-01609],[@B8-nutrients-12-01609],[@B9-nutrients-12-01609],[@B10-nutrients-12-01609]\]. Chronic kidney disease--mineral bone disorder (CKD-MBD) is considered one of the main factors associated with increased cardiovascular morbidity and mortality in CKD patients \[[@B11-nutrients-12-01609]\]. These patients frequently show both impaired bone mineralization and ectopic vessel mineralization, a feature which is the result of at least two factors: the "bone--vessel axis" and the "calcification paradox".
Albeit some pathways involved in the crosstalk between vasculature and bone (the "bone-vascular axis") are still unknown, it is clear that osteotropic hormones, including parathyroid hormone and calcitriol, physiologically regulate both vascular and skeletal mineralization as well as stem cell renewal \[[@B9-nutrients-12-01609],[@B12-nutrients-12-01609]\]. Cellular, endocrine and metabolic signals that flow bidirectionally between vasculature and bone are necessary for the health of both. Diabetes and CKD in particular are associated with an important derangement of the bone--vascular axis which usually determines simultaneous alterations in both.
VC is frequently associated with and evolves in parallel with decreased bone mineral density and deranged bone turnover. Bone and vessel mineralization share common pathways, and this parallel is known as the "calcification paradox" \[[@B13-nutrients-12-01609]\]. Moreover, treatments targeting bone disorders might have adverse consequences for cardiovascular health.
This creates the need to identify treatments that have a positive impact on the factors or pathways involved in the crosstalk between vasculature and bone. Over the last decade, as a growing body of evidence has seemed to point to an involvement of vitamin K deficiency in VC, the hypothesis that vitamin K supplementation could be a tool to prevent the rapid progression of VC in CKD patients is gaining momentum \[[@B14-nutrients-12-01609]\].
2. Vitamin K and Gla Proteins {#sec2-nutrients-12-01609}
=============================
There are two main natural forms of Vitamin K: K1 (or phylloquinone, PK) contained in green vegetables, and K2 (including several different vitamers called menaquinones, MKs) mostly derived from fermented foods and intestinal bacteria (e.g., cheeses and the Japanese soybean product known as "natto") \[[@B15-nutrients-12-01609]\]. There are up to 12 different types of MKs, from MK-4 to MK-15; the most common MKs in humans are the short-chain MK-4 and MK-7 to MK-10, which are respectively produced by systemic conversion of phylloquinone to menaquinones and synthesized by bacteria \[[@B16-nutrients-12-01609]\]. The pivotal biological role of vitamin K is to act as a cofactor for the carboxylation (and thereby activation) of vitamin-K-dependent proteins (VKDPs). Vitamin-K-dependent carboxylase is an enzyme that catalyzes the addition of carbon dioxide to the specific glutamic acid residues of a limited number of proteins leading to the formation of g-carboxyglutamic (Gla) residues.
Seventeen members of the Gla protein family, involved in various biological processes, have currently been identified. In particular, the Gla family includes (i) seven proteins belonging to the coagulative cascade: prothrombin, factor VII, factor IX, factor X, protein C, protein S and protein Z; (ii) four proteins that regulate bone and vascular mineralization: matrix Gla protein (MGP), osteocalcin (OC), growth arrest-specific protein 6 (Gas6), Gla-rich protein (GRP); (iii) two proline-rich Gla proteins, two transmembrane Gla proteins, periostin and periostin-like factor \[[@B17-nutrients-12-01609],[@B18-nutrients-12-01609],[@B19-nutrients-12-01609],[@B20-nutrients-12-01609]\].
Measuring Vitamin K plasma levels is difficult due its low circulating concentration, the non-polar nature of the molecule and the interference of lipids. Moreover analytical variability, diet, inflammation, and the coexistence of chronic disease may further influence plasma levels of vitamin K subtypes \[[@B21-nutrients-12-01609]\]. Functional tests can be performed to estimate vitamin K blood levels indirectly. In particular, the measurement of undercarboxylated proteins, OC and MGP, has proved to be more sensitive than prothrombin time in detecting subclinical vitamin K deficiency. Vitamin K deficiency status does not allow VKDPs to acquire their carboxylated form: the gamma-carboxyglutamate (Gla) domain is responsible for the high-affinity binding of calcium ions, thus allowing coagulation factors OC and MGP to interact with negatively charged phospholipid membranes. Moreover, adequate calcium binding is a critical physiologic step in bone mineralization counteracting VC \[[@B22-nutrients-12-01609]\].
Since some Gla-proteins are involved in bone metabolism and vascular health, their reduced carboxylation leads to bone metabolism impairment and an increase in the VC formation process \[[@B23-nutrients-12-01609]\]. Mild vitamin K deficiency does not usually induce clinically evident changes in coagulation, which is consistent with the observation that functional tests (e.g., prothrombin time) are altered only when the activity of vitamin K-dependent coagulation is reduced by 50% \[[@B24-nutrients-12-01609],[@B25-nutrients-12-01609]\]. On the other hand, due to differences between hepatic and extra-hepatic vitamin K metabolism, extra-hepatic VKDP carboxylation is most affected by subclinical deficient states \[[@B25-nutrients-12-01609]\]. The current understanding is that vitamin K2 is mainly transported to the extra-hepatic tissues and controls the carboxylation of VKDPs in bone and blood vessels, while vitamin K1 is central for the carboxylation of vitamin-K-dependent coagulation factors in the liver \[[@B25-nutrients-12-01609]\].
Considering that each Gla-protein has specific biological functions, no single biomarker is considered a gold standard in the assessment of vitamin K deficiency.
OC is synthesized by the osteoblasts under the control of vitamin D, and its active carboxylated form (cOC) is mainly involved in bone mineralization, since it binds and incorporates calcium ions in the hydroxyapatite crystals in bone matrix. Thus, a high undercarboxylated OC (ucOC) level is an expression of poor vitamin K levels and intake. Moreover ucOC can be released during osteoclastic resorption, and is thought to be responsible for functions related to energy metabolism \[[@B17-nutrients-12-01609],[@B26-nutrients-12-01609]\].
Low levels of ucOC are associated with higher bone mineral density and reduced fracture risk \[[@B27-nutrients-12-01609]\]. It is a fact that ucOC mirrors total OC, and expressing ucOC as a percentage of the total OC (%ucOC) may therefore be considered a more reliable index of vitamin K storage than the absolute value of ucOC. As showed by Booth et al., a value of %ucOC \>20% is congruent with subclinical vitamin K deficiency in vitamin K depletion and repletion studies \[[@B28-nutrients-12-01609]\].
As OC distribution seems to change with age and gender, reflecting the degree of bone formation, the differences in its distribution in bone matrix may be responsible, at least in part, for the altered remodeling of bone associated with gender and aging \[[@B29-nutrients-12-01609],[@B30-nutrients-12-01609],[@B31-nutrients-12-01609]\]. In addition, glutamic acid carboxylation at position 17 is probably fundamental for the spatial and structural conformation of the osteocalcin, allowing a correct interaction with hydroxyapatite crystals \[[@B18-nutrients-12-01609]\].
It has been observed that mice knocked out for the OC gene are affected by hyperostosis, suggesting that osteocalcin is crucial for bone formation. Vitamin K2 seems not to be involved only in OC carboxylation, but also in the reduction of bone resorption through the increase in osteoprotegerin (OPG) production and the inhibition of RANK ligand (RANKL) expression \[[@B32-nutrients-12-01609]\]. MGP is a 12-KDa gamma-carboxyglutamic acid-containing protein synthesized by vascular smooth muscle cells (VSMC) and chondrocytes; it is currently considered one of the most potent endogenous inhibitors of vascular calcification in vivo \[[@B33-nutrients-12-01609],[@B34-nutrients-12-01609],[@B35-nutrients-12-01609]\].
The protein undergoes two types of post-translational modification before maturation: (i) phosphorylation of up to three serine residues and (ii) gammacarboxylation of up to five glutamate residues. While the role of carboxylation, which depends on Vitamin K, is better understood and determines MGP's bioactivity as a calcification inhibitor, the function of phosphorylation is not yet clarified; recent data, however, suggest a role in regulating MGP secretion into the extracellular milieu \[[@B36-nutrients-12-01609],[@B37-nutrients-12-01609]\]. The aforementioned reactions do not proceed in parallel, and, according to the state of carboxylation and/or phosphorylation, at least four different molecules can be found in circulation: (i) dephosphorylated uncarboxylated MGP (dp-ucMGP); (ii) dephosphorylated carboxylated MGP (dp-cMGP); (iii) phosphorylated uncarboxylated MGP (p-ucMGP); (iv) phosphorylated carboxylated MGP (p-cMGP). The active form is both phosphorylated and carboxylated (p-cMGP) and its synthesis is stimulated by vitamin D \[[@B38-nutrients-12-01609]\].
Findings about ucMGP levels in the bloodstream of CKD patients are contradictory. Low, rather than high, circulating ucMGP levels were a powerful predictor of cardiovascular calcification, and ucMGP was decreased within the serum of CKD patients: this feature could reflect an accumulation of ucMGP within the calcified vessel wall \[[@B39-nutrients-12-01609],[@B40-nutrients-12-01609]\]. Krueger speculated that uncarboxylated Gla proteins are less secreted from the cells, maybe in attempt to prevent inactive proteins from entering the tissues. A reduced phosphorylation of the three serine residues of MGP seems equally important as it would result in reduced secretion of MGP \[[@B16-nutrients-12-01609],[@B17-nutrients-12-01609],[@B18-nutrients-12-01609],[@B19-nutrients-12-01609],[@B36-nutrients-12-01609],[@B41-nutrients-12-01609]\].
The fully inactive form of the protein, dephosphorylated uncarboxylated MGP (dp-ucMGP), does not interact with calcium or BMP-2 and closely reflects the vitamin K status at the vascular level \[[@B42-nutrients-12-01609],[@B43-nutrients-12-01609]\], thus representing a potential biomarker for cardiovascular endpoints. Dp-ucMGP levels show an inverse correlation with vitamin K status, hence decreasing with vitamin K supplementation and increasing with vitamin K antagonist \[[@B44-nutrients-12-01609]\].
True deficiency and functional shortage of Vitamin K (as occurs in the case of warfarin treatment) through the reduction of active MGP and OC may therefore lead to vascular calcification and impaired bone mineralization.
Some members of the Gla protein family (GRP and MGP) are components of circulating calciprotein particles (CPP) and extracellular vesicles (EVs) in which they play an important role in regulating the calcification process. Circulating CPP are colloidal nanoparticles diffused in the blood and mainly composed of Ca and P precipitates, fetuin-A, GRP and probably MGP. They are considered a possible mechanism by which ectopic mineralization is prevented. Fetuin, GRP, and other soluble proteins contained in CPP act like mineral carriers with a role in the stabilization, transport and recycling of water-insoluble minerals in the blood. A pivotal event in the onset of VSMC calcification is the release of extracellular EVs capable of efficiently nucleating hydroxyapatite in the absence of calcification inhibitors like MGP. It is likely that γ-carboxylated GRP, fetuin- A, and MGP represent an effective mechanism to regulate the steps of mineral formation both at systemic and tissue levels, thus preventing pathological calcification \[[@B20-nutrients-12-01609],[@B45-nutrients-12-01609],[@B46-nutrients-12-01609]\]. If the content of fetuin-A, GRP or MGP is insufficient or Gla proteins are predominantly undercarboxylated, CPPI would be transformed to crystalline mineral core (CPPII) that cannot be excreted through the liver \[[@B20-nutrients-12-01609]\]. A high proportion of CPPII induces endothelial damage and, by deposition in the extracellular matrix, an enhancement of VC \[[@B20-nutrients-12-01609],[@B47-nutrients-12-01609]\].
Therefore, an insufficient level of calcification inhibitors, or a lower activation of Gla-proteins, whatever the cause, predisposes to vascular calcification \[[@B47-nutrients-12-01609],[@B48-nutrients-12-01609]\]. The subsequent release of extracellular vesicles, containing less fetuin-A or GRP, could play an additional role in modulating the VSMC calcification process \[[@B20-nutrients-12-01609],[@B49-nutrients-12-01609]\].
3. Vitamin K Deficiency, MGP and OC Metabolism in CKD Patients {#sec3-nutrients-12-01609}
==============================================================
Observational and interventional clinical studies suggest that CKD patients often develop subclinical vitamin K deficiency \[[@B50-nutrients-12-01609],[@B51-nutrients-12-01609],[@B52-nutrients-12-01609]\], a condition that may determine a higher risk of morbidity and mortality in this population. Dietary recommendations aimed at restricting potassium and phosphate intake (e.g., reduced consumption of leafy green vegetables and dairy products, rich in vitamin K1 and K2, respectively) may play a pivotal role in promoting this deficit \[[@B17-nutrients-12-01609],[@B21-nutrients-12-01609],[@B50-nutrients-12-01609],[@B53-nutrients-12-01609]\].
Vitamin K deficiency is also supposedly due to the uremic milieu and to the exhaustion caused by the high volume of vitamin K required by VKDPs to inhibit calcification.
CKD patients are exposed to the risk of subclinical Vitamin K deficiency, which entails a reduced activation of MGP and OC, both playing a crucial role in vascular and bone health. Thus, it is not unexpected that in CKD patients Vitamin K deficiency can contribute to the high VC burden as well as to the derangement of bone metabolism.
However, in CKD, especially in patients that are not on dialysis, studies on the effect of vitamin K deficiency on OC (expressed as ucOC or %ucC cut off) and bone health are still limited.
MGP could counteract the progression of vascular calcification in CKD patients by binding hydroxyapatite crystals; this interferes with their deposition and promotes macrophage-mediated clearance \[[@B19-nutrients-12-01609]\]. In addition, the interaction between active MGP and bone morphogenetic protein-2 (BMP-2) results in the inhibition of VSMC osteoblast transformation \[[@B54-nutrients-12-01609],[@B55-nutrients-12-01609]\], a central mechanism in the development of vascular calcification \[[@B56-nutrients-12-01609]\].
Vitamin K deficiency leads to the formation of uncarboxylated MGP (ucMGP), whose concentration on arterial walls progressively increases along with the severity of the vascular calcifications \[[@B57-nutrients-12-01609]\]. Deposits in the blood vessels could explain the low circulating ucMGP levels usually found in patients with end-stage renal failure on chronic hemodialysis \[[@B40-nutrients-12-01609],[@B58-nutrients-12-01609],[@B59-nutrients-12-01609]\], and may provide the rationale for the inverse relationship between ucMGP serum concentration and vascular calcification observed in some of these studies.
In CKD patients, an increase in the fully inactive form of MGP, dp-ucMGP, has been shown in CPP1 and matrix vesicles. This entails a faster transformation of CPP1 into CPP2, and this accelerated transformation is associated with VC \[[@B48-nutrients-12-01609]\]. Moreover, a reduced amount of fetuin-A and active GRP has been related to a more severe mineral calcification in the soft tissue \[[@B20-nutrients-12-01609]\].
The concentration of circulating dp-ucMGP progressively increases across CKD stages \[[@B38-nutrients-12-01609],[@B60-nutrients-12-01609],[@B61-nutrients-12-01609]\]; its levels have been associated with the severity of aortic calcifications \[[@B38-nutrients-12-01609],[@B61-nutrients-12-01609],[@B62-nutrients-12-01609]\] and vascular stiffness \[[@B63-nutrients-12-01609],[@B64-nutrients-12-01609]\] in many studies, albeit not all (Table 1) \[[@B43-nutrients-12-01609],[@B65-nutrients-12-01609]\]. Moreover, high dp-ucMGP represents an independent predictor of mortality in renal transplant recipients \[[@B66-nutrients-12-01609]\] but not in other groups of CKD patients \[[@B38-nutrients-12-01609],[@B43-nutrients-12-01609]\]. Interestingly, dp-ucMGP levels in blood \[[@B67-nutrients-12-01609]\] and kidney tissue \[[@B68-nutrients-12-01609]\] also predict the impairment of renal function both in the general population and in CKD patients, suggesting that vitamin K deficiency may increase the risk of progression towards end-stage renal disease. Even though vitamin K2 supplementation can lower dp-ucMGP levels in patients with CKD \[[@B43-nutrients-12-01609],[@B69-nutrients-12-01609],[@B70-nutrients-12-01609],[@B71-nutrients-12-01609]\], the effects of this treatment on clinical outcomes still need to be fully clarified.
In CKD patients, vitamin K levels can be further influenced by the frequent use of the anticoagulant warfarin \[[@B72-nutrients-12-01609]\]. Warfarin is a 4-hydroxycoumarin derivative that selectively inhibits the oxidoreductase responsible for the regeneration of inactive and dietary vitamin K, thus preventing the hepatic gamma-glutamyl carboxylation of coagulation factors \[[@B73-nutrients-12-01609]\]. In the liver, however, the recycling of inactive vitamin K also depends on the activity of DT-diaphorase, a reducing enzyme not affected by warfarin, that activates vitamin K and thus limits the effect of the drug. On the other hand, DT-diaphorase activity is extremely low in VSMC and other extra-hepatic tissues \[[@B74-nutrients-12-01609]\]. Thus, the carboxylation of extra-hepatic VKDP, including MGP, heavily relies on the activity of vitamin K oxidoreductase, and is therefore more susceptible to warfarin treatment.
Consistently with these data, simultaneous administration of warfarin and vitamin K to Wistar Kyoto rats does not affect the synthesis of coagulation factors but blocks MGP carboxylation, inducing diffuse medial vascular calcification, which can be only partially relieved by a vitamin K-rich diet \[[@B38-nutrients-12-01609]\]. Similarly, in an animal model of CKD, a lower therapeutic dose of warfarin has promoted calcium deposition in major blood vessels and induced arterial stiffening \[[@B75-nutrients-12-01609]\]. Interestingly, the same treatment did not induce medial calcification in rats without kidney dysfunction, suggesting that vitamin K deficiency potentiates the effect of other pro-calcifying factors in CKD patients. In keeping with experimental evidence, administration of warfarin in CKD patients has been associated with a steep increase in the concentration of circulating dp-ucMGP, while stopping the treatment has consistently reduced dp-ucMGP levels \[[@B44-nutrients-12-01609],[@B62-nutrients-12-01609],[@B76-nutrients-12-01609]\].
From a clinical standpoint, patients on hemodialysis treated with warfarin had a 3.77 odds ratio of developing severe aortic calcifications compared to those not taking the drug \[[@B77-nutrients-12-01609]\]. Long-term warfarin use was also associated with increased coronary and extra-coronary calcifications \[[@B78-nutrients-12-01609],[@B79-nutrients-12-01609],[@B80-nutrients-12-01609]\]. A cross-sectional study across 1838 Japanese dialysis centers identified warfarin use as an independent predictor of calciphylaxis (i.e., calcific uremic arteriolopathy) \[[@B81-nutrients-12-01609]\], a severe complication of end-stage renal disease. These data raise concerns about the risk--benefit ratio of the use of warfarin in the hemodialysis setting for conditions such as atrial fibrillation; indeed, a retrospective cohort study of 41,425 patients on hemodialysis highlighted that warfarin use was associated with a higher mortality in this group, even after stratification and covariate adjustment \[[@B82-nutrients-12-01609]\]. The same group also reported an increased risk of stroke in hemodialysis patients with pre-existing atrial fibrillation treated with warfarin compared to those not assuming the drug. This suggests that, in this setting, warfarin might not be as beneficial as in the general population \[[@B83-nutrients-12-01609]\].
Several studies show that hemodialysis patients' low vitamin K levels are likely related to the dietary regimen prescribed and to overall poor nutritional intake. Moreover, given its lipophilic characteristics, vitamin K is not supposed to be removed or absorbed by the dialysis membrane.
Cranenburg et al. evaluated the vitamin K1 and K2 intake and the vitamin K status of 40 haemodialysis patients and found a low intake (median 140 μg/day) in dialysis per day and during the weekend, compared to healthy adults (mean K1 and K2 intake 200 and 31 μg/day, respectively).. Undercarboxylated VKDPs (dp-ucMGP and ucOC) were found to be high in 33 HD patients, pointing to hepatic as well as extrahepatic deficient vitamin K status. On the other hand, high non-carboxylated MGP in the same population strongly suggested vascular vitamin K deficiency \[[@B52-nutrients-12-01609]\]. Voong et al. reported that 29% and 93% of maintenance HD patients met the criteria for sub-clinical vitamin K deficiency based on low levels of phylloquinone and high levels of ucOC, respectively, suggesting vitamin K deficiency at the bone level \[[@B51-nutrients-12-01609]\]. Westenfeld et al. demonstrated that haemodialysis patients had 4.5-fold higher dephosphorylated, uncarboxylated MGP and 8.4-fold higher uncarboxylated osteocalcin levels compared with controls, confirming that most hemodialysis patients have a functional vitamin K deficiency \[[@B71-nutrients-12-01609]\]. Fusaro et al. carried out an observational study to assess the prevalence of vitamin K deficiency and the relationship between vitamin K status, vertebral fractures, vascular calcification, and survival in 387 patients on hemodialysis for ≥1 year. Total OC and ucOC levels were higher in patients with CKD than in healthy controls, and vitamin K1 deficiency was an independent predictor for vertebral fractures in prevalent hemodialysis patients (the median total OC level was 29% lower in patients with one or more vertebral fractures) \[[@B84-nutrients-12-01609]\].
Peritoneal dialysis (PD) patients and maintenance HD patients displayed a similar degree of vitamin K depletion. The results of a cross-sectional study carried out on 21 PD patients demonstrate that a significant proportion of patients have subclinical vitamin K deficiency, defined by low serum phylloquinone concentrations (\<0.4 nmol/L) and elevated %ucOC (\>20%), respectively, at 23.8% and 100% \[[@B85-nutrients-12-01609]\].
Jansz et al. analyzed the effect of kidney transplantation and phosphate binder use on vitamin K status; they found lower dp-ucMGP levels in kidney transplant recipients compared to patients on HD or peritoneal dialysis, which seem to suggest vitamin K levels improve after kidney function is restored. An important exception is represented by sevelamer monotherapy, which is associated with significantly higher dp-ucMGP levels; this suggests that sevelamer has a negative effect on patients' vitamin K status \[[@B86-nutrients-12-01609]\].
4. Effects of CKD-MBD Therapy on Vitamin K and Vitamin K Dependent Proteins {#sec4-nutrients-12-01609}
===========================================================================
The mainstays of treatment of CKD-MBD are (i) lowering increased parathyroid hormone levels with vitamin D receptor activators (VDRA) and/or calcimimetics and (ii) decreasing serum phosphorus level with dietary interventions and phosphate binders. However, due to the wide variation in serum levels of mineral metabolism markers and to the variability of therapeutic response among patients, CKD-MBD therapy needs to be tailored to the individual patient.
Calcifediol (25OH) is the precursor of calcitriol and is regarded as the best indicator of vitamin D status, although a significant number of studies have shown no effect on parathyroid hormone (PTH) suppression; in experimental studies performed in calcifediol-deficient mice, calcifediol administration reduced the extent of calcification, but did not induce any changes in systemic calcium and phosphate levels \[[@B87-nutrients-12-01609]\].
Actually, calcitriol may cause detrimentally elevated levels of serum calcium and phosphate \[[@B88-nutrients-12-01609]\]. At high doses, it activates specific vitamin D receptors (VDR) expressed in vascular smooth muscle cells (VSMCs), and promotes VC both in vitro and in vivo, while at more physiological doses vitamin D could even be protective \[[@B3-nutrients-12-01609],[@B89-nutrients-12-01609],[@B90-nutrients-12-01609]\]. These results suggest that vitamin D derivatives have different effects on calcification. In this scenario, the active vitamin D metabolite may favor the calcification process through the stimulation of calcium and phosphate, while its precursor form (calcifediol) may be able to provide protection against calcification via calcium- and phosphate-independent mechanisms \[[@B3-nutrients-12-01609],[@B91-nutrients-12-01609]\]. Calcimimetics, by binding with calcium-sensing receptors, increase the sensitivity to extracellular calcium, lowering PTH synthesis and secretion. Unlike VDRA, they do not increase calcium and phosphorus levels and prevent VC \[[@B92-nutrients-12-01609]\]. Fusaro et al. performed a secondary analysis \[[@B84-nutrients-12-01609]\] of the VIKI (VItamin K Italian) study, an observational study designed to assess the prevalence of vitamin K deficiency in hemodialysis patients and to investigate the effects of ongoing treatment for CKD-MBD on OC and MGP levels. They showed that vitamin D analogs increase the levels of OC and MGP, while calcimimetics, alone or combined with calcium acetate, increase only MGP levels; in addition, the combination of vitamin D analogs and calcimimetics proved to be most effective in inducing a further increase of total OC \[[@B93-nutrients-12-01609]\], while increased total MGP levels are found only in patients treated with calcimimetics, alone or combined with calcium acetate \[[@B93-nutrients-12-01609]\]. Moreover, in this study, lower OC levels were associated with vertebral fractures or vascular calcification, while higher baseline total OC levels were combined with a lower progression rate of abdominal aortic calcification; OC thus appears to protect against VC, although it promotes mineralization of bone tissue. Although it is likely that the effect on OC and MGP of both vitamin D analogs and calcimimetics may be indirectly mediated by the PTH decrease, it cannot be excluded that a direct gene/transcriptional (direct) effect may be at work not only for vitamin D analogs but also for calcimimetics \[[@B93-nutrients-12-01609]\].
Therefore, patients with CKD and hyperparathyroidism could disclose low serum OC, but this does not necessarily entail that they are vitamin-K-deficient, and could be the expression of the PTH activity. In CKD patients, hyperphosphatemia is commonly treated with dietary restrictions and intestinal phosphorus binders. A range of phosphate binders currently available for long-term use include calcium-based phosphate binders (calcium carbonate and acetate) and calcium-free binders (aluminum hydroxide, lanthanum carbonate, magnesium carbonate, sevelamer hydrochloride, and sevelamer carbonate). As current phosphate binders are all effective in lowering phosphorus, the main considerations for selecting one over another include absorbability (ideally non-absorbed), compliance, adequate gastrointestinal tolerability, and cost or cost-effectiveness.
Is not yet clear weather calcium-based phosphate binders are associated with harmful effects. Non-calcium-based phosphate binders demonstrate a reduction of VC progression and significant survival benefits over a 3-year interval compared to calcium-based ones \[[@B94-nutrients-12-01609],[@B95-nutrients-12-01609],[@B96-nutrients-12-01609],[@B97-nutrients-12-01609],[@B98-nutrients-12-01609]\]. However, the results in this regard are contradictory \[[@B99-nutrients-12-01609],[@B100-nutrients-12-01609]\].
VC frequently progress despite adequate phosphate levels, probably following chronic inflammation, increased calcium load from calcium-based binders and, not least, the capacity of phosphate binders to also bind vitamin K. Therefore, the advantage of lowering phosphate blood levels may be blunted by a worsening of vitamin K deficiency, a possible off-target effect of phosphate-binding therapy. Studies looking at this problem are still limited, and, most importantly, the results show noticeable differences in the association between various phosphate binders and vitamin K deficiency.
Neradova et al. evaluated the interaction of vitamin K2 (menaquinone-7; MK-7) with five different phosphate binders, in the presence or absence of phosphate \[[@B101-nutrients-12-01609]\]. In this in-vitro study sucroferric-oxyhydroxide and sevelamer carbonate were the only binders that did not interact with vitamin K2. Instead, calcium acetate/magnesium carbonate bound vitamin K2 strongly, both in the absence and presence of phosphate. The binding of lanthanum carbonate and vitamin K2 depended on the absence of phosphate, suggesting a competitive interaction between phosphate and vitamin K2 for this compound. No significant binding was observed in the solution containing vitamin K2 and phosphate. Calcium carbonate significantly bound vitamin K2 in a solution containing phosphate, while no significant binding was observed without phosphate. A nominally lower concentration of K2 was shown in the mixture with sevelamer carbonate as well, but this decline was not statistically significant. Finally, the addition of sucroferric-oxyhydroxide did not lead to any decline of vitamin K2, irrespective of the presence or absence of phosphate.
The results of this study suggest that: (i) the interaction between any phosphate binder and vitamin K2 depends on the specific physical and chemical properties of the binder; (ii) there is no direct interaction between vitamin K2 and phosphate; (iii) only in the case of calcium carbonate can the phosphate facilitate the link of this binder with vitamin K2 ([Figure 1](#nutrients-12-01609-f001){ref-type="fig"}).
Sevelamer carbonate, because of its binding to bile acids, was expected to impair the absorption of fat-soluble vitamins like vitamin K2. However, Westenfeld et al. \[[@B71-nutrients-12-01609]\], in agreement with the findings by Neradova et al. \[[@B101-nutrients-12-01609]\], found no association between sevelamer administration and vitamin K (menaquinone) levels, although only a small proportion of patients were prescribed sevelamer. On the contrary, another in-vitro study shows extensive vitamin K binding by sevelamer hydrochloride \[[@B102-nutrients-12-01609]\].
5. Conclusions {#sec5-nutrients-12-01609}
==============
Although several studies seem to assess the involvement of functional vitamin K deficiency in the pathogenesis of vascular calcifications, as yet there are no studies in CKD and dialysis population demonstrating that vitamin K supplementation can prevent the development of vascular calcifications as well as the associated cardiovascular morbidity and mortality.
Undercarboxylated Gla-proteins are used as markers for vitamin K levels in the tissues from which they are produced, and their high levels are considered equivalent to functional vitamin K deficiency, although the threshold below which vitamin K status can be considered optimal is not yet known. A significant contribution to the understanding of the role of vitamin K deficiency could derive from future studies on vitamin-K-dependent GRP and calcification propensity score (T50) \[[@B103-nutrients-12-01609]\].
The current understanding is that vitamin K deficiency in CKD patients is mainly related to dietary recommendations, while the exact role of the uremic state itself, and the high demand of vitamin K in the "pro-calcifying" uremic environment, remain unknown ([Table 1](#nutrients-12-01609-t001){ref-type="table"}). Future studies will better define the effects of ongoing treatment for mineral and bone disorders on OC and MGP metabolism.
Conceptualization, M.C. and G.C.; Methodology, M.C., G.C., M.A.P. and A.G.; Writing: M.C., G.C., P.C. and M.A.P. Original Draft Preparation, M.C., G.C., L.G., M.A.P, A.G. and P.C. Review, editing and figure preparation: L.G.; Supervision, M.C. and G.C. All authors have approved the submitted version (and a version substantially edited by journal staff that involves the author's contribution to the study) and agree to be personally accountable for the author's own contributions and for ensuring that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and documented in the literature. All authors have read and agreed to the published version of the manuscript.
This research received no external funding.
Authors declare no conflict of interest.
AOC
aortic calcification
CAC
coronary artery calcification
CKD
chronic kidney disease
dp-ucMGP
not phosphorylated, undercarboxylated matrix Gla protein
HD
hemodialysis
OR
odds ratio
ucMGP
undercarboxylated matrix Gla protein
VKA
vitamin K antagonist
CF-PWV
Carotid-femoral pulse wave velocity
{#nutrients-12-01609-f001}
nutrients-12-01609-t001_Table 1
######
Association between vitamin K status and cardiovascular outcome in CKD populations.
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Study Design/Patients Vitamin K Assessment Outcome Results
----------------------------------------------------------- ------------------- ---------------------- ------------------------- ------------------------------------------------------------------------------------------
Puzantian et al., 2018 \[[@B64-nutrients-12-01609]\] Prospective\ dp-ucMGP dp-ucMGP\ Increase with CKD severity (*p* \< 0.0001) Positive correlation (β = 0.21; *p* = 0.019).
137 CKD 1--5 CF-PWV
Thamratnopkoon et al., 2017 \[[@B61-nutrients-12-01609]\] Cross-sectional\ dp-ucMGP Vascular Calcification\ OR 1002, *p* = 0.004
83 CKD 3-5 Score (Kauppila)
Kurnatowska et al., 2016 \[[@B60-nutrients-12-01609]\] RCT, DB\ dp-ucMGP dp-ucMGP after\ Decreased dp-ucMGP\
38 CKD 4--5 Vit K2 for 9 m Vs control (10.7% vs. no change)
Meuwes et al., 2015 \[[@B65-nutrients-12-01609]\] Cross-sectional\ dp-ucMGP\ CAC Score No association
64,9% on Dialysis PIVKA-II
Delanaye et al., 2014 \[[@B62-nutrients-12-01609]\] Cross-sectional\ dp-ucMGP Vascular Calcification\ R = 0.17\
160 HD patients Score (Kauppila) *p* = 0.0049
Schlieper et al., 2011 \[[@B43-nutrients-12-01609]\] 188 HD patients dp-ucMGP Vascular calcification\ No association
(X-ray and ultrasound)
Schurgers et al., 2010 \[[@B104-nutrients-12-01609]\] Prospective\ dp-ucMGP AOC Associated with AOC\
107 CKD 2--5 (*p* \< 0.001)
Cranenburg et al., 2009 \[[@B58-nutrients-12-01609]\] Cross-sectional\ ucMGP CAC Inverse correlation\
40 HD (r = 0.41, *p* = 0.009)
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
The World Health Organization describes the healthcare waste as discarded, untreated materials from healthcare activities, which have the potential of transmitting infectious agents to humans \[[@R1]\]. Dental instruments and materials are exposed to blood and saliva during various dental procedures. Hence, biomedical waste (BMW) management in dental practice is equally critical as in the medical fraternities. Proper BMW disposal is essential for the safety of the dental personnel and the general public at large. Justifiably, the majority of countries control the dental waste under medical waste management regulations. Dental laboratories are an integral part of dental practice. The well-equipped, efficient laboratory is vital across the dental specialties including Prosthodontics, Restorative dentistry, pedodontics, and orthodontics. Most of the restorative dental procedures require the dental laboratories support to complete the planned treatment. During the process of indirect restoration fabrication, the dental laboratories generate various hazardous wastes potentially detrimental to the health and environment. The dental laboratory solid waste classified as infectious waste, non-infectious toxic waste, and domestic waste. The waste suspected to contain the pathogen in sufficient concentration causing disease in susceptible hosts is considered as infectious waste \[[@R2],[@R3]\]. The Dental prosthesis, occlusal bite blocks, occlusal records, and orthodontic appliances routinely come in contact with human saliva and blood \[[@R4]\]. The researchers demonstrated the presence of bacteria and fungi over the dental impression on their arrival to the laboratory \[[@R5]\]. Some studies even indicate the existence of bacteria on the denture polishing pastes and wheels. The other infectious waste includes silicones used for maxillofacial prosthesis, gloves and plastic containers used for transporting the dentures and appliances from clinics. The non-infectious toxic wastes are devoid of human fluids contamination but potentially toxic in nature. It includes the acrylic resin scraps, wasted metal alloys, metal dust, porcelain, and gypsum waste. Other non-infectious toxic wastes are amalgam alloys and acids used in electrolytic polishing of metal frameworks. Domestic type wastes are comprised of Paper cups, plastics, sand papers, and household wastes.
The Dental laboratories broadly belong to two categories. The first group operates as an integral part of dental clinics or hospital. The second group is independent of dental hospitals; work as separate establishments to serve the dental fabrication need of clinics. The laboratories of later categories are routinely ignored by regulating bodies from the government. The dental laboratory solid waste is frequently disposed of through the municipal solid waste. The reason for improper disposal is multifactorial; it includes a lack of knowledge, improper attitude of the dental technician and inadequate facilities. The existing dental literature regarding the waste management are conducted predominantly in dental clinics and critically deficient on waste disposal practices at dental laboratories. The knowledge and attitude of personnel, disposal practices and disposal facilities at different categories of dental laboratories need to be explored further. The result of the study will help to identify and initiate the corrective measures for acceptable dental solid waste disposal. Hence, this study was designed to assess the knowledge, attitude and practices about dental solid waste management among dental staff at different categories dental laboratories in Abha city, Kingdom of Saudi Arabia.
MATERIAL AND METHODS
====================
The study was conducted in the Abha- Khamis Mushiath city Kingdom of Saudi Arabia during the first semester of 2014. The Abha City is the largest city in the southern part of the Kingdom of Saudi Arabia. The city has several dental treatment facilities; it includes the College of Dentistry dental clinics, three government dental specialty treatment hospitals, twelve private specialty dental clinics and four private, independent dental laboratories. The approval for the research proposal was obtained from institution research ethics committee. The study population included the dental technicians across all the laboratories. The study was a cross-sectional study, with stratified sampling. Total of 90 dental technicians were working in all the dental laboratories, out of which 80 consented to participate in the study. The participation rate was at 88.9%. The exclusion criteria were the subjects not willing to participate in the study. The study group comprised of four groups with twenty subjects from each group. The Group I consisted of the dental technicians working in a dental teaching hospital; Group II included the laboratory technicians working at government dental hospital. The technician at private dental clinics laboratories considered as Group III and Group IV were technicians at independent dental laboratories. A written consent was obtained from all the study participants. The data collection was done through the anonymous, pre-designed, pre-tested and structured questionnaire. The pretesting was carried out on 20 subjects among target populations to determine the variation in the language, terminology, interpretation of question and response options in the questionnaire. According to the participant's feedback, the required modification was incorporated in the questionnaire. The internal consistency of the survey instrument was ascertained by Cronbach\'s alpha coefficient (0.891) analysis. The question consisted 29 closed-ended questions to assess the knowledge, practice, disposal facilities and education regarding biomedical waste management among the dental technicians. The questionnaire was self-administered, the purpose of the study was explained to all members and collected back immediately after the completion. The resultant data was analysed using SPSS software version 19 for proportions to interpret the results.
RESULTS
=======
Table **[1](#T1){ref-type="table"}** indicates the improper BMW disposal risk was recognized by only 45% (9) in Group IV, and 55% (11) among Group III. The risk awareness among Group II was at 85% (17) and Group I was 75% (15). The knowledge regarding the different BMW categories, was at 40% (8), 11% (55), 25% (5), 15% (3) for Group I, II, III, IV respectively. The information regarding BMW colour coding and segregation at source among Group I was at 65% (13) and 55% (11) respectively. The Group IV subjects had the least information on colour coding and segregation at 5% (1) and 10% (2). The knowledge of disinfection methods, disposal options was also at lowest among the Group IV at 10% (2) and 5% (1). The Group III was slightly better at 15% (3) and 5% (1). The information on national and institutional BMW disposal regulations was surprisingly less among all the studied groups. Only 5-25% of the participants across the groups were aware of BMW regulation applicable to dental practice.
The positive attitude towards BMW disposal was relatively high and uniform among all groups (Table **[2](#T2){ref-type="table"}**). It was at an average of 75-90%. The majority of respondents agree that good BMW management and handling is the integral part of their work.
The third segment of the questionnaire (Table **[3](#T3){ref-type="table"}**) was about the BMW disposal practice at dental laboratories.
The result indicated color specific bags for BMW disposal was utilized by 20% of Group IV, followed by Group III (25%), Group I (45%) and Group II (65%). The 20% (4) subjects from Group IV practiced proper disposal of heavy metals and amalgam and 35% (7) of subjects from Group III. The favorable heavy metal and amalgam disposal was practiced by 65% (13) for Group II and 55% (11) from Group I. Only 25% (5) amongst Group IV and 35% (7) from Group III dental technicians were utilizing puncture-proof container to dispose of sharp objects. 20% (4) and 25% (5) practiced the disinfection of solid waste prior disposal among Group IV and Group III respectively. Though the solid waste disinfection practice was better at Group I (45%) and Group II (86%), still at inadequate proportion.
The results of the study revealed (Table **[4](#T4){ref-type="table"}**) inadequate disposal facilities at Group III and Group IV working place. The 15% (3) of Group IV respondents had coloured container while only 5% (1) had puncture proof containers for waste disposal at the working place. The availability of these containers was also insufficient at Group III respondents working area with corresponding values of 25% (5) and 30% (6). The disposal facilities at Group I and Group II was significantly better at 65-85%. The monitoring agency visits for these facilities was only 5% at Group IV, followed by group III (45%), Group I (45%) and Group II (65%).
Though 25% of the study respondents had learned about BMW disposal in their curriculum, there were very few reorientation programs to update the knowledge (Table **[5](#T5){ref-type="table"}**). The Dental Technician at Group IV had no reorientation programs conducted or attended. The Group II subjects had better exposure to reorientation program with 55% (11) respondents attended these reorientation scientific activities.
DISCUSSION
==========
The dental practice generates the infectious, non-infectious toxic and domestic waste. The potentially infectious dental solid wastes are blood/saliva soaked paper towels, gauze, cotton roll, latex gloves, syringes, dental floss, and surgical blades. The dental laboratories handle the potentially infectious objects like dental bridges and prosthesis, matrix bands, dental impressions, wax, interocclusal records. Toxic wastes produced at laboratories include dental amalgam and heavy metal waste. Proper handling and disposal of the potentially infectious and toxic waste is critical for the safety of patients, professionals, public, and the environment. The study reports from Nabizadeh R *et al.* \[[@R6]\] suggests, the dental waste constituted of 71.15% domestic waste, 21.40% potentially infectious waste, 7.26% chemical and 0.18% toxic waste. They observed highest dental solid waste was produced by denture maker (37.96%), followed by a general dentist (34.95), practical dentist (20.69) and specialist dentist (6.40%). A study by Kizlary E *et al*. \[[@R7]\] at Greek reported dental BMW had 94.7% infectious waste, 2% non-infectious and 3.3% domestic waste by weight. The higher percentage of infectious waste in this study was due to the inclusion of hazardous waste like metal and amalgam as infectious waste. According to the authors, BMW production rate was at 513 g/practice/day, and 486 g/practice/day was the potentially infectious waste. The similar results were reported by Cristina Dutra Vieira *et al*. \[[@R8]\] according to them the dental solid waste included 24.3% infectious waste, 48.1% non-infectious waste and 26.7% domestic-type waste. The overall projection of these studies indicates the dental waste composition is similar to WHO biomedical waste report from other medical specialties. It shows the proper dental solid waste management requires the equal importance like in medical fraternity.
The previous studies \[[@R9]\] have suggested; the prosthodontic clinics produced the highest dental BMW among all the dental subspecialties, followed by restorative and orthodontic dentistry clinics. The denture fabrication, impression making procedures are the major part of prosthodontic clinics. The impression materials amount to 30% of total solid waste generated in dental practice. It could be the reason for the maximum amount of waste generation from this sub-speciality. Effective production laboratory support is crucial for prosthodontic, restorative and orthodontic specialties. The adequate knowledge; proper disposal practice is essential from dental technicians for overall efficient BMW management. The results of the study showed the knowledge regarding bio-medical waste management was inadequate across the groups. The Group IV had considerable ignorance level with only 15% knew about different BMW categories and color coding. The knowledge level among Group I (35%) and Group (40%) though better, still inadequate. Sudhir KM *et al.* \[[@R10]\] observed the similar result regarding the knowledge level. The solid waste segregation at the source is necessary for the volume reduction of potentially infectious waste, decrease in waste treatment cost and recycling of paper, plastic materials. The results of the present study indicate the knowledge regarding segregation was also inadequate among Group IV (10%) and Group III (35%). The Group II were better at 70% knowledge level. The previous studies have demonstrated the presence of potentially harmful microorganism in the dental solid waste 2.The dental casts and impressions received at laboratories also had the microbes over them \[[@R11]\]. Hence, it is strongly advised to disinfect the hazardous waste prior to their disposal. It is helpful in preventing the dissemination of multi-resistant bacteria into the environment. Experience on appropriate disinfection methods for BMW was unsatisfactory among all groups, particularly among Group IV (10%) and Group III (15%).
The Saudi Arabian government has formulated the Bio-Medical Waste (Handling and Management) Rules in 1998 and later revised in February 2001. The law incorporates all persons who generate, collect, transport, treat and dispose of the biomedical waste in any form \[[@R12]\]. The rules regulate the handling and disposal of BMW including human anatomical waste, blood, body fluids, medicines, soiled, liquid & biotechnology waste. Previous studies concluded the insufficient information about national and institutional regulations resulting in BMW discharge into the wastewater system and household disposal sites \[[@R13],[@R14]\]. This procedure endangers the human health and environment to the potential risk. The study respondents from all the evaluated group were ignorant (5-10%) about the existing regulation of the biomedical waste management and handling law in Saudi Arabia.
The research results exhibited the good BMW disposal attitude among all the evaluated groups. Similar to the previous study reports \[[@R15]\], the majority of the respondents (80-85%) across all the groups agreed on the lesser generation, better segregation and disinfection of dental solid waste. Total of 85-95% of the respondents is of the opinion, the continuous update on BMW disposal and treatment plant necessity at an institute.
The disposing of the waste in color coded, and puncture proof containers were least observed at Group IV (25%) and Group III (35%) respondents. The Group I and Group II were slightly better at 65%. Similar disposal practice result was observed by the researchers like Taiwo, JO *et al.* \[[@R16]\] and Charania1 ZK *et al.* \[[@R17]\].
The dental amalgam is still widely used as a direct restorative material and used in the laboratory for die preparation. The laboratory utilizes the alloys from nickel, chromium and silver for the crown, bridge, implant and removable prosthesis fabrication. Improper handling and disposal of dental amalgam is known to cause the neurotoxicity, soft tissue toxicity, allergenicity and ecological damage \[[@R18]\]. The heavy metals like chromium known to cause liver, Kidney and respiratory damages \[[@R19]\]. The heavy metal landfill is attributed to soil and underground water contamination \[[@R20]\]. The results of the present study like other studies \[[@R21]\] indicated only 15% of Group IV, and 35% of Group III were disposing of amalgam and metal in the proper method. The Group I (55%) and Group II (65%) technicians followed the better disposal methods. The disinfection of waste prior to disposal was practiced by 20-25% subjects in group IV and Group III.
The results of the study notice the poor infrastructure for dental solid waste disposal. Disposal facilities like color coded container, puncture-proof containers, and facility to dispose of the metal were severely inadequate at Group IV (5-15%) and Group III (15-30%) working place. The disposal facilities were substantially better with Group I (65-75%) and Group II (80-85%). According to the feedback, the government monitoring agencies visit to evaluate the practice was almost non-existent at Group IV working place with only 5% respondents confirmed the visit.
The researchers strongly advocate the continuous training and reorientation programs to update, emphasize the proper BMW handling \[[@R22]\]. The results indicate the technician at Group IV had no reorientation program to update the knowledge while 55% of Group III had experienced the reorientation programs. The study also shows the insufficient BMW handling information in the dental technicians educational curriculum.
The results of the study exposed the gap in the knowledge and practice among all the four categories of respondents. The Group IV and Group III respondents performed poorly in all waste handling and disposal parameters. The independent dental laboratories and private dental clinics had poor adherence to dental solid waste disposal guidelines. The government monitoring agencies are required to monitor and regulate these dental facilities more stringently to reinforce the safe BMW practices.
The authors of the opinion, more studies are required to understand the waste disposal management especially at independent dental laboratories across multiple cities. The results of the research will help to understand the trend and develop the measures to improve the disposal practices.
CONCLUSION
==========
Within the limitation of the study, it can be concluded the there was considerable variation in the knowledge, facilities, handling and disposal of BMW among dental technicians evaluated. The study concludes the need to update the curriculum, regular orientation training programs and strict implementation of guidelines for BMW management. There is an urgent need to upgrade the disposal facilities at independent dental laboratories and private dental clinical to correct the deficient practices. The monitoring agencies needs to supervise the strict implementation of BMW regulations at private sector establishments.
Declared none.
CONFLICT OF INTEREST
====================
The authors confirm that this article content has no conflict of interest.
######
Biomedical waste knowledge among different groups.
BMW[\*](#T1F1){ref-type="table-fn"} knowledge Group I Group II Group III Group IV
----------------------------------------------- --------- ---------- ----------- ----------
Risks of improper BMW disposal 15(75) 17(85) 11(55) 9(45)
Different categories of BMS waste 8(40) 11(55) 5(25) 3(15)
Colour coding of Waste disposal 13(65) 11(55) 4(20) 1(5)
Importance of Segregation at source 11(55) 14(70) 7(35) 2(10)
Specific disinfection methods before disposal 8(40) 11(55) 3(15) 2(10)
Proper disposal option for various waste 9(45) 13(65) 1(5) 1(5)
Waste disposal regulation in Saudi Arabia 5(25) 2(10) 5(25) 2(10)
BMW regulating agency in Saudi Arabia 5(25) 2(10) 1(5) 1(5)
BMW regulations awareness at your work place 8(40) 9(45) 5(25) 1(5)
BMW: Biomedical waste
######
Attitude regarding BMW[\*](#T2F1){ref-type="table-fn"} among surveyed dental technician.
Favourable attitude Group I Group II Group III Group IV
--------------------------------------------------------------------------------------------- --------- ---------- ----------- ----------
Believe and follow reducing BMD generation at source 17(85) 17(85) 18(90) 15(75)
Requirement of disinfection of waste before disposal 18(90) 16(80) 18(90) 17(85)
Proper segregation and disposal -- part of our responsibility 18(90) 17(85) 18(90) 16(80)
Continuous update on BMW disposal is essential for health professional 18(90) 17(85) 19(95) 13(65)
Do you think effluent treatment plant essential for dental colleges to treat infected water 19(95) 17(85) 19(95) 17(85)
BMW: Biomedical waste
######
Biomedical waste disposal practice in surveyed groups.
Correct BMW[\*](#T3F1){ref-type="table-fn"} disposal practice Group I Group II Group III Group IV
------------------------------------------------------------------------------ --------- ---------- ----------- ----------
Disposal in colour specified container 9(45) 13(65) 5(25) 4(20)
Sharp objects disposal in puncture proof container 13(65) 11(55) 7(35) 5(25)
Reported injury due improperly disposed waste 4(20) 1(10) 3(15) 4(20)
Proper disposal of mercury containing waste 11(55) 13(65) 7(35) 3(15)
Proper disposal of substance containing silver,nckel,chromium and lead waste 9(45) 13(65) 7(35) 4(20)
Always disinfect BMW before disposal 9(45) 13(65) 5(25) 4(20)
BMW: Biomedical waste
######
Available disposal facilities at the work place.
Disposal facilitates Group I Group II Group III Group IV
-------------------------------------------------------------------------------- --------- ---------- ----------- ----------
Colour specified container in laboratory 13(65) 17(85) 5(25) 3(15)
Puncture proof container to dispose sharp objects 15(75) 16(80) 6(30) 1(5)
Proper methods to dispose chemicals used at laboratory 13(65) 17(85) 3(15) 3(15)
Heavy Metal waste disposal 13(65) 16(80) 3(15) 1(5)
Government BMW[\*](#T4F1){ref-type="table-fn"} monitoring agency visit anytime 9(45) 13(65) 9(45) 1(5)
BMW: Biomedical waste
######
BMW education among the dental technicians.
BMW education Group I Group II Group III Group IV
--------------------------------------------------------------- --------- ---------- ----------- ----------
Learnt BMW disposal in curriculum 5(25) 13(65) 6(30) 5(25)
Attended reorientation programme on BMW 2(10) 11(55) 2(10) 0(0)
Any reorientation programme conducted at your facility 3(15) 8(40) 2(10) 0(0)
Have you received special training in BMW disposal management 3(15) 7(35) 3(15) 1(5)
BMW: Biomedical waste
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Liver X receptors (LXRs) were originally identified as ligand-dependent transcriptional activators that induce target genes involved in lipid metabolism. The subfamily consists of two isoforms: LXR*α* and LXR*β*. Gene transcription is modulated by LXRs, which heterodimerize with the retinoid X receptor and bind to LXR-response elements in the transcriptional regulatory regions of their target genes \[[@B1]\]. Recently, LXRs have been reported to regulate macrophage inflammatory responses, phagocytosis, and apoptosis \[[@B2], [@B3]\]. LXRs inhibit the transcription of proinflammatory cytokines such as interleukin-1*β* (IL-1*β*), IL-6, and tumor necrosis factor-*α* (TNF-*α*) via their promoters or enhancers \[[@B3]\]. One study showed that LXRs mediate the regulation of Th17 cell differentiation and autoimmunity \[[@B4]\]. Furthermore, LXR has been demonstrated to be involved in the upregulation of various genes, including apoptotic inhibitor of macrophage and arginase II \[[@B5]\]. Therefore, LXRs have emerged as important regulators of inflammatory gene expression in several inflammatory diseases \[[@B6]--[@B8]\].
Systemic lupus erythematosus (SLE) is a chronic autoimmune disease with various clinical manifestations and autoimmune serologic markers. The pathogenesis of SLE is unclear, but several causes, such as genetic background, environmental factors, and disturbance in both innate and adaptive immunity, have been proposed as contributing factors for the development of the disease \[[@B9]\]. Disturbances in apoptotic cell clearance, hyperactive immune cells, and an abnormal production of autoantibodies are observed as major pathological features of SLE \[[@B9], [@B10]\]. In particular, uncleared apoptotic cells and their accumulation in tissues have been suggested to contribute most to the inflammation in SLE \[[@B11]\]. Some molecules, such as growth arrest-specific 6 and protein S, enhance the recognition and susceptibility of apoptotic cells to phagocytosis \[[@B12], [@B13]\]. These interact with receptor tyrosine kinases of the TAM (Tyro-3, Axl, and Mer) family \[[@B14]\]. The loss of regulation of inflammation and delayed clearance of apoptotic materials are associated with the development of a lupus-like syndrome in TAM knockout mice \[[@B15]\]. In particular, Mer signaling has been reported to increase the transcriptional activity of LXR to promote the resolution of acute sterile inflammation \[[@B16]\]. Therefore, LXRs might play an important role in the regulation of inflammatory gene expression in SLE. However, the association between LXR activation or expression and pathogenesis of SLE has not been well addressed.
We had previously reported that LXR*α* gene (*NR1H3*) promoter polymorphisms are associated with SLE in Koreans \[[@B17]\]. Specifically, the -1830 T \> C polymorphism within *NR1H3* promoter region was associated with clinical manifestations of SLE; increased B cell proliferation and decreased *NR1H3* mRNA expression were observed in patients with -1830 TC genotype compared to those with the -1830 TT genotype. Therefore, in this study, we assessed cytokine expression in different LXR*α* polymorphism in monocyte-derived macrophages from patients with SLE. Furthermore, we evaluated the effect of LXR activation on proinflammatory cytokine secretion induced by several Toll-like receptor (TLR) agonists.
2. Materials and Methods {#sec2}
========================
2.1. Cell Culture {#sec2.1}
-----------------
U937 cells (human myelomonocytic leukemia cell line) were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS) at 37°C in a 5% CO~2~ incubator. THP-1 cells (human acute monocytic leukemia cell line) were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated FBS and 0.05 mM 2-mercaptoethanol at 37°C in a 5% CO~2~ incubator. Macrophages were obtained after 72 h of culture of human monocytes (U937 or THP-1) in RPMI 1640 medium (Gibco by Life Technologies, Grand Island, NY) supplemented with PMA (40 nM or 80 nM). Cells were cultured at a density of 1 × 10^6^ cells/mL in 24-well plates (Corning, NY), and the cells were transfected with 1 *μ*g control pGL3 or LXR*α* promoter constructs, using FuGENE HD (Promega, Madison, WI), Lipofectamine 2000 (Thermo scientific, Fremont, CA), and ultra TRAX transfection agent (GeneDireX, Taoyuan, Taiwan) according to the manufacturer\'s instructions. After incubation for 6 h, the medium was replenished with 500 *μ*L of fresh medium with 20% FBS, and the cells were incubated for another 18 h at 37°C in a 5% CO~2~ incubator. Twenty-four hours after transfection, cells were preincubated with LXR agonist (3 *μ*M GW3965 or 5 *μ*M T0901317) at the indicated concentrations for 24 h prior to the addition of TLR ligands: 100 ng/mL ultrapure lipopolysaccharide (LPS; Calbiochem, San Diego, CA), 1 *μ*g/mL CL097 (tlrl-c97, InvivoGen, San Diego, CA), and 1 *μ*M ODN TTAGGG (tlrl-ttag151, InvivoGen) for 24 h.
2.2. Ex Vivo Cell Culture {#sec2.2}
-------------------------
Twelve patients with SLE, who were involved in the previous study, were enrolled again \[[@B17]\]. Among them, 6 patients had LXR*α* -1830 TT and 6 patients had TC genotype. All patients satisfied at least four of the criteria laid out by 1982 revised American College of Rheumatology criteria for SLE \[[@B18]\]. Supplementary [Table 1](#supplementary-material-1){ref-type="supplementary-material"} shows the clinical characteristics and laboratory findings of enrolled 12 SLE patients. This study was approved by the Institutional Review Board of Ajou University Hospital (IRB No. AJIRB-BMR-EXP-14-186). Informed consent was obtained from all subjects. All experiments were performed in accordance with relevant guidelines and regulations.
PBMCs from buffy coats of patients were isolated using Ficoll-Paque PLUS gradient (GE Healthcare Life Sciences, Pittsburgh, PA). The purity of CD14^+^ cells was \>90%, as assessed by flow cytometry. CD14^+^ cells were cultured for 5 days at 1 × 10^6^ cells/mL in 6-well plates containing serum-free DMEM media (Gibco, Carlsbad, CA) in the presence of M-CSF (100 ng/mL; R&D Systems, Minneapolis, MN).
LXR agonist, on day 2, was coincubated with either activators or inhibitors of TLR7 and TLR9 for 24 h. Cells were then harvested by centrifugation. Supernatants were collected and immediately stored at -20°C before being tested by enzyme-linked immunosorbent assay (ELISA). Pellets were resuspended in phosphate-buffered saline (PBS), and proteins were extracted for western blot analysis.
2.3. Preparation of Plasmid DNA and Transfection {#sec2.3}
------------------------------------------------
Structures, composed of the LXR*α* -1830 T \> C sequence, were assembled carrying each allele. A 500 bp fragment (from -2121 to -1622) of the LXR*α* gene was PCR-amplified using either -1830 T homozygous or -1830 C homozygous genomic DNA as a template and the following primers: forward primer: 5′-CGGCGG**GGTACC**ACATCTATGCCAGCCCTGTTTCAG-3′ (the bold characters represent the KpnI site); reverse primer: 5′-CCGCCG**CTCGAG**ACTGAGCCCCAGCGGCTTTC-3′ (the bold characters denote the XhoI site). Each PCR product was subcloned separately into the KpnI-XhoI site of the pGL3-Basic luciferase reporter vector (Promega, Madison, WI).
2.4. RNA Extraction and Quantitative Real-Time PCR {#sec2.4}
--------------------------------------------------
Total RNA was extracted from cells, using an RNeasy Mini kit according to the manufacturer\'s instruction (Qiagen, Valencia, CA); cDNA was synthesized from total RNA using GoScript Reverse Transcription System kit (Promega, Madison, WI) and 18-residue oligo (dT) (Bioneer, Seoul, Korea). After annealing at 25°C for 5 min and extension at 70°C for 15 min, the product was stored at -20°C until use. The real-time PCR amplification was performed using a Rotor-Gene SYBR Green PCR kit (Qiagen, Valencia, CA). The following PCR conditions were used: heating to 95°C for 5 min, then 40 cycles of 95°C for 5 s, 58°C for 10 s, and 72°C for 30 s. The primers used were as follows: human LXR*α* (F): 5′-AGGGCTGCAAGGGATTCTTCC-3′, (R): 5′-TCTGACAGCACACACTCCTCCC-3′, TNF-*α* (F): 5′-TGCCTATGTCTCAGCCTCTTC-3′, (R): 5′-GGGCCATAGAACTGATGAGAG-3′, IL-1*β* (F): 5′- TCCCAGACAACCACCTTCTC-3′, (R): 5′- ACTGGGCAGACTCAAATTCC -3′, IL-6 (F): 5′- TCCTCATTCCCTCAACTTGG-3′, (R): 5′- GTCAGCAGGCTGGCATTT-3′, IL-10 (F): 5′-TTA CCT GGA GGA GGT GAT GC-3′, (R): 5′-TGG GGG TTG AGG TAT CAG AG-3′, ATP binding cassette 1 (ABCA1) (F): 5′-GAACTGGCTGTGTTCCATGAT-3′, (R): 5′-GATGAGCCAGACTTCTGTTGC-3′, *β*-actin (F): 5′-CAAGAGATGGCCACGGCTGC-3′, (R) 5′-TCCTCTGCATCCTGTCGGC-3′.
2.5. Total Protein Extracts and Immunoblot Analysis {#sec2.5}
---------------------------------------------------
Total protein extracts were prepared as described \[[@B17]\]. Equal amounts of protein were resolved by SDS-PAGE and analyzed with anti-LXR*α* (1 : 1,000, ab135039, Abcam, Cambridge, MA), anti-TNF-*α* (1 : 1,000, ab183896, Abcam), anti-interferon-*γ* (IFN-*γ*; 1 : 1,000, EPR1108, Abcam), anti-ABCA1 (PA1-16789, Thermo scientific, Fremont, CA), and anticytoskeletal actin (1 : 10,000, A300-491A, Bethyl Laboratories, Montgomery, TX) antibodies. The secondary antibody used with each was goat anti-mouse antibody (1 : 2,000, AbFrontier, Seoul, Korea), except for anticytoskeletal actin (1 : 20,000). Following transfer and blotting, the proteins of interest were visualized by enhanced chemiluminescence (Pierce, Rockford, IL) and analyzed.
2.6. Measurement of Cytokine Production {#sec2.6}
---------------------------------------
The concentration of cytokines in cell culture supernatants was analyzed by commercial ELISA, specific for human IL-1*β*/IL-1F2 (DY201-05, R&D, Minneapolis, MN), human TNF-*α* (DY210-05, R&D), human IFN-*γ* (DY285-05, R&D), and human COX-2 (DYC4198-2, R&D) according to the manufacturer\'s protocol.
2.7. Data Analysis {#sec2.7}
------------------
Statistical analyses were performed using the IBM SPSS software ver. 23.0 (IBM Corp., Armonk, NY). The data are shown as mean ± standard deviation (SD) or median and interquartile range, as appropriate. Differences in cytokine levels and LXR*α* levels were determined by Student\'s *t*-test. A *p* value \< 0.05 was regarded as indicator for statistical significance.
3. Results {#sec3}
==========
3.1. Monocyte-Derived Macrophage Differentiation Upregulates LXR*α* Expression whereas LXR Agonists Downregulate Proinflammatory Cytokines in Monocyte-Derived Macrophages {#sec3.1}
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------
To confirm the expression of LXR*α* in monocyte-derived macrophages, we measured mRNA and protein abundance of LXR*α* and LXR target gene *ABCA1* for 72 h after phorbol 12-myristate 13-acetate (PMA) treatment in U937 and THP-1 cells. Monocytes (U937 and THP-1 cells) were differentiated to macrophages after 72 h of PMA treatment (Figures [1(a)](#fig1){ref-type="fig"} and [1(b)](#fig1){ref-type="fig"}). [Figure 1(c)](#fig1){ref-type="fig"} shows a significant increase in LXR*α* and *ABCA1* mRNA abundance after macrophage differentiation with PMA. The data for protein levels were consistent with those of mRNA levels ([Figure 1(d)](#fig1){ref-type="fig"}).
To determine whether an LXR*α* agonist would influence LXR*α* and *ABCA1* expression in monocyte-derived macrophages, we treated the cells with LXR agonists, T0901317 and GW3965, for 24 h. [Figure 2(a)](#fig2){ref-type="fig"} shows an increase in LXR*α* and *ABCA1* mRNA abundance upon treatment of macrophages with LXR agonists, although statistically significant difference was found only with the treatment of T0901317 in THP-1 cells. Furthermore, LXR agonists decreased mRNA expression of proinflammatory cytokines such as TNF-*α*, IFN-*γ*, IL-1*β*, and IFN-*α*, however that was significantly different only in TNF-*α* and IFN-*γ* in GW3965-treated U937 cells and in IL-1*β* in GW3965-treated THP-1 cells ([Figure 2(b)](#fig2){ref-type="fig"}).
3.2. Expression of LXR*α* and Proinflammatory Cytokines in Human Monocyte-Derived Macrophages according to Genotype (-1830 T \> C) {#sec3.2}
----------------------------------------------------------------------------------------------------------------------------------
To verify the involvement of LXR*α* -1830 T \> C genotype in the expression of LXR*α* and proinflammatory cytokines in human monocyte-derived macrophages, we evaluated the endogenous LXR*α* -1830 T \> C genotype in monocyte cells (U937 and THP-1 cells). We confirmed apriori that the genotypes of U937 and THP-1 cells were TT. The reporter constructs with the promoter sequence carrying each allele (-1830 T \> C) were transfected into U937 or THP-1 cell lines (Supplementary [Figure 1](#supplementary-material-1){ref-type="supplementary-material"}). The expression of LXR*α* in the monocytes transfected with pGL3-Basic vector only (control), TT type vector, and TC type vector were not different (data not shown). The transfected monocytes were differentiated to macrophages after 72 h of PMA treatment, and the cells were treated with T0901317 or GW3965 for 24 h. We confirmed that mRNA expression of LXR*α* and *ABCA1* between 24 h and 72 h treatment of PMA was similar and was at its best after treatment with LXR agonists at 24 h (data not shown). The mRNA expression of LXR*α* and *ABCA1* was decreased in TC genotype-transfected U937 cells compared to that in TT genotype-transfected cells ([Figure 3(a)](#fig3){ref-type="fig"}). Further, mRNA expressions of proinflammatory cytokines, including TNF-*α* and IFN-*γ*, were increased in TC genotype-transfected U937 cells compared to that in TT genotype-transfected cells. The data for mRNA levels of LXR and proinflammatory cytokines in transfected THP-1 cells were similar to those in the U937 cells ([Figure 3(b)](#fig3){ref-type="fig"}); the protein levels were consistent with the measured mRNA levels ([Figure 3(c)](#fig3){ref-type="fig"}). The protein levels of LXR*α* and ABCA1 were decreased in TC genotype-transfected U937 cells compared to that in TT genotype-transfected cells. Further, the protein expression of TNF-*α* and IFN-*γ* was increased in TC genotype-transfected U937 cells compared to that in TT genotype-transfected cells. The corresponding data in transfected THP-1 cells were similar to those in the U937 cells.
3.3. Effect of LXR Activation on Proinflammatory Cytokine Production in Human TLR-Stimulated Monocyte-Derived Macrophages {#sec3.3}
-------------------------------------------------------------------------------------------------------------------------
TLRs have been widely implicated as the pathogenic drivers in SLE, and TLR2, TLR4, TLR7, and TLR9 have been shown to be expressed at higher levels in B cells, peripheral blood mononuclear cells (PBMC), or kidney tissues \[[@B19]--[@B22]\]. Moreover, LXR agonists have been shown to regulate TLR-induced macrophage cytokine secretion with TLR agonists \[[@B23]--[@B26]\]. Therefore, we evaluated the effect of LXR activation on proinflammatory cytokine secretion induced by several TLR agonists. PMA-treated U937 cells were cultured in the presence of T0901317 or GW3965 for 24 h prior to stimulation with TLR ligands LPS (for TLR4), CL097 (for TLR7/8), and ODN (for TLR9). The proinflammatory cytokines were not altered in the macrophages stimulated with LPS in the presence of LXR agonists. Further, secretion of proinflammatory cytokines was slightly different in the macrophages stimulated with CL097 in the presence of LXR agonists, though not statistically significant ([Figure 4(a)](#fig4){ref-type="fig"}). The secretions were significantly augmented for only TNF-*α* in macrophages stimulated with ODN in the presence of LXR agonists ([Figure 4(b)](#fig4){ref-type="fig"}).
To verify the involvement of LXR*α* -1830 T \> C genotype in the expression of proinflammatory cytokines in monocyte-derived macrophages treated with TLR agonists, U937 cells were transfected with reporter constructs harboring either of the genotypes. Interestingly, the levels of TNF-*α* and cyclooxygenase-2 (COX-2) were increased in TC genotype-transfected U937 cells compared to those in TT genotype-transfected macrophages stimulated with CL097 ([Figure 5(a)](#fig5){ref-type="fig"}). Furthermore, the levels of IL-1*β*, TNF-*α*, and COX-2 were increased in TC genotype-transfected U937 cells compared to those in TT genotype-transfected macrophages stimulated with ODN ([Figure 4(b)](#fig4){ref-type="fig"}).
3.4. TLR7 And TLR9 Inhibitors Attenuate Proinflammatory Cytokine Production in TLR-Stimulated Human Monocyte-Derived Macrophages, Especially in TC Genotype-Transfected U937 Cells {#sec3.4}
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
The levels of proinflammatory cytokines were increased in both TC and TT genotype-transfected cells stimulated with CL097, but relatively more in TC genotype-transfected U937 cells compared to those in TT genotype-transfected macrophages. Treatment with TLR7 inhibitor significantly decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in TC genotype-transfected CL097-stimulated cells, whereas only the level of IL-1*β* was decreased in TT genotype-transfected U937 cells ([Figure 5(a)](#fig5){ref-type="fig"}).
The data obtained upon treatment of the cells with TLR9 inhibitor were similar to those obtained for the TLR7 inhibitor ([Figure 5(b)](#fig5){ref-type="fig"}). The TLR9 inhibitor decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in U937 cells treated with PMA for 24 h prior to stimulation with TLR ligand ODN. The levels of proinflammatory cytokines were increased in both TC and TT genotype-transfected cells, stimulated with ODN, but especially increased in TC genotype-transfected U937 cells compared to those in TT genotype-transfected macrophages. Furthermore, treatment with TLR9 inhibitor (more than 1 *μ*M) significantly decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in TC genotype-transfected U937 cells stimulated with LXR ligand (T0901317), and those of TNF-*α*, COX-2, and IFN-*γ* in TT genotype-transfected U937 cells ([Figure 5(c)](#fig5){ref-type="fig"}).
3.5. Expression of Proinflammatory Cytokines, Based on LXR*α* -1830 T \> C Genotype, in LXR Agonist-Treated PBMC-Derived Macrophages of Patients with SLE {#sec3.5}
---------------------------------------------------------------------------------------------------------------------------------------------------------
The PBMC from patients with SLE were differentiated to macrophages after treatment with macrophage colony-stimulating factor (M-CSF) for 72 h, which were then treated with T0901317 and GW3965 for 24 h. The data for the levels of proinflammatory cytokines in PBMC-derived macrophages of patients, according to the *LXRα* -1830 T \> C genotypes, were similar to those for the macrophage cell lines transfected with reporter constructs harboring either of the genotypes ([Figure 6(a)](#fig6){ref-type="fig"}). The basal levels of proinflammatory cytokines including IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* were increased in patients with TC genotype compared to those in patients with TT genotype. After treatment with LXR agonists, the level of proinflammatory cytokines in macrophages was decreased for both TT and TC genotypes in PBMCs of patients with SLE, but more strikingly so in TC genotype individuals.
TLR7 and TLR9 ligands augment proinflammatory cytokines in LXR agonist-treated macrophages derived from PBMCs of patients with SLE (harboring LXR*α* -1830 T \> C genotypes; Figures [6(b)](#fig6){ref-type="fig"} and [6(c)](#fig6){ref-type="fig"}). The TLR7 inhibitor decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in macrophages treated with M-CSF for 24 h prior to stimulation with LXR agonist (T0901317) and CL097 (TLR7/8) in both LXR*α* -1830 TT and TC genotypes. Also, treatment with TLR9 inhibitor significantly decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in macrophages stimulated with LXR agonist (T0901317) and ODN (TLR9) in both genotypes. Furthermore, immunoblot analysis was performed with antibodies specific to LXR*α* and ABCA1 in M-CSF-treated macrophages with TLR ligands ([Figure 6(d)](#fig6){ref-type="fig"}). TLR7 and TLR9 ligands decreased LXR*α* and ABCA1 levels, and treatment with TLR7 or TLR9 inhibitors could recover the levels.
4. Discussion {#sec4}
=============
In this study, we showed that the expression of LXR*α* is increased in human monocyte-derived macrophages compared to that in unstimulated monocytes. Levels of proinflammatory cytokines, such as IL-1*β* and TNF-*α*, are decreased with increased expression of LXR*α*. Interestingly, the production of proinflammatory cytokines depends on the expression of LXR*α* -1830 T \> C genotypes, consistent with our previous findings \[[@B17]\]. Decreased LXR*α* expression with increased proinflammatory cytokine expression was observed in human monocyte-derived macrophages transfected with the TC genotype of LXR*α* -1830 T \> C compared to that in cells transfected with the TT type. These data are consistent in human PBMC-derived macrophages, isolated from patients with SLE, according to respective genotype. Furthermore, increased expression of proinflammatory cytokines in the TC genotype of LXR*α* -1830 is related to TLR7 and TLR9 expression.
An important role of macrophages, in the control of inflammation, is the removal of dying cells \[[@B27]\]. This function could be induced through nuclear receptors, such as PPAR*γ*, PPAR*δ*, and LXR; activation of these signaling pathways could suppress inflammation. Several studies showed that LXR activation exhibited potent anti-inflammatory activities \[[@B8], [@B28]--[@B31]\]. A previous study evaluated the effect of LXR agonists on innate immunity responses in human primary lung macrophages and in a preclinical rodent model of lung inflammation \[[@B31]\]. The authors demonstrated an LXR-dependent reduction in lung neutrophils in a rodent model of lung inflammation. However, this inhibition was not associated with suppression of NF-*κβ*/AP-1 DNA binding. Therefore, these results suggest that anti-inflammatory activity of LXR agonists was not via inhibition of the NF-*κ*B pathway. A recent study identified the lipid transporter ABCA1 as a critical mediator for anti-inflammatory effects of LXR \[[@B30]\]. The activation of LXR inhibited the signaling pathway from TLR2, 4, and 9 to their downstream NF-*κ*B and MAPK effectors through ABCA1-dependent changes. In the present study, we showed the anti-inflammatory effects of LXR in human monocyte-derived macrophages. The expression of LXR*α* was increased in human monocyte-derived macrophages while levels of proinflammatory cytokines were decreased with LXR*α* expression, consistent with previous reports \[[@B8], [@B28]--[@B31]\].
However, LXR activation is known to potentiate proinflammatory cytokine secretion in LPS-activated human macrophages, and this is suggested to be related to increased expression of TLR4 \[[@B24], [@B32], [@B33]\]. A recent study showed that LXR activation leads to a dramatic increase in proinflammatory cytokine secretion driven by TLR1/2, TLR2/6, and TLR7/8 \[[@B24]\]. The authors showed that the LXR pathway is upregulated in rheumatoid arthritis synovial macrophages and activation of LXRs by ligands in synovial fluid augments TLR-driven cytokine secretion. However, the data for LXR activation, in the present study, with TLR agonists was not similar to that. Although proinflammatory cytokines were not augmented in macrophages stimulated with several TLR receptor ligands in the presence of LXR agonists, only the secretion of TNF-*α* was elevated in macrophages stimulated with TLR9 ligands in the presence of LXR agonists. Therefore, our results suggest that LXR activation does not potentiate proinflammatory cytokine secretion (except for TNF-*α*) through the upregulation of TLR7 and TLR9 expression in patients with SLE. Interestingly, the inflammatory cytokine response was dependent on the duration of pretreatment with LXR agonist in primary human macrophage; short-term pretreatment reduced the inflammatory response to TLR4 ligand; however, pretreatment longer than 48 h, with LXR agonist, significantly enhanced TLR4 ligand response \[[@B33]\]. Our results could possibly be the outcome of pretreatment with LXR agonist for 24 h. However, further studies would be required to resolve the mechanism by which LXR activation promotes TNF-*α* production by interaction with TLR9.
We previously evaluated the functional effects of the LXR*α* -1830 T \> C polymorphism and reported that the -1830 T allele-containing reporter construct had higher promoter activity than that containing the corresponding C allele \[[@B17]\]. Proliferation of B cells of the LXR*α* -1830 TC type was found to have increased beyond those of the TT type or of LXR agonist-treated B cells from patients with SLE. Furthermore, *NR1H3* mRNA expression levels were lower in *NR1H3* -1830 TC type B cells than in TT type cells. To investigate the effects of the *NR1H3*-1830 T \> C polymorphism on LXR expression and inflammation in macrophages, we transfected the reporter promoter gene construct, carrying the LXR*α* -1830 TT and TC genotypes, into THP-1 and U937 cells. We also treated cells with T0901317 and GW3965 for confirmation of the effects of LXR agonists. The results were similar to our previous data with B cells. The expression of LXR*α* and ABCA1 was decreased in TC genotype-transfected macrophages compared to that in TT genotype-transfected cells. Further, the expression of proinflammatory cytokines was increased in TC genotype-transfected cells compared to that in the TT genotype-transfected cells. These data are consistent with those of an ex vivo study on patients with SLE. Basal levels of proinflammatory cytokines were higher in macrophages from patients with the TC genotype compared to those with the TT genotype, and similar patterns were identified after treatment with LXR agonists. These data suggest that the TC genotype of LXR*α* -1830 leads to low expression of LXR and insufficient effect of LXR on proinflammatory cytokines compared to that of TT genotype. The low expression of LXR might be associated with susceptibility or disease activity in SLE. Proinflammatory cytokine levels were significantly decreased in TC genotype-transfected macrophages after treatment with TLR7 or TLR9 inhibitors. These data are consistent with those of an *ex vivo* study on PBMCs from patients with SLE with respect to the TC and TT genotypes of LXR*α* -1830. The results suggest that antagonists targeting TLR signaling could be effective for the treatment of SLE, especially in patients with the TC genotype.
5. Conclusions {#sec5}
==============
Our results imply that expression of LXR*α* according to LXR*α* -1830 T \> C genotypes may contribute to the inflammatory responses by inducing inflammatory cytokines in SLE. In particular, the LXR*α* -1830 TC genotype leads to low expression of LXR and insufficient control on proinflammatory cytokine secretion.
This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI16C0992).
Data Availability
=================
The data used to support the findings of this study are included within the article.
Conflicts of Interest
=====================
The authors declared no conflicts of interest with respect to the authorship and/or publication of this article.
Supplementary Materials {#supplementary-material-1}
=======================
######
Supplementary Table 1: clinical characteristics of SLE patients (*n* = 12). Supplementary Figure 1: pGL3-Basic vector circle map and preparations of *LXRα* -1830 T \> C reporter constructs. The liver X receptor A (*LXRA)* gene promoter was separately subcloned into a KpnI-XhoI site of pGL3-Basic luciferase reporter gene vector. Additional description: luc+, cDNA encoding the modified firefly luciferase; *Amp^r^*, gene conferring ampicillin resistance in *E. coli*; f1 ori, origin of replication derived from filamentous phage; ori, origin of replication in *E. coli*. Arrows within luc+ and the *Amp^r^* gene indicate the direction of transcription; the arrow in the f1 ori indicates the direction of ssDNA strand synthesis.
######
Click here for additional data file.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
{ref-type="fig"}). The TLR9 inhibitor decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in U937 cells treated with PMA for 24 h prior to stimulation with TLR ligand ODN. The levels of proinflammatory cytokines were increased in both TC and TT genotype-transfected cells, stimulated with ODN, but especially increased in TC genotype-transfected U937 cells compared to those in TT genotype-transfected macrophages. Furthermore, treatment with TLR9 inhibitor (more than 1 *μ*M) significantly decreased the levels of IL-1*β*, TNF-*α*, COX-2, and IFN-*γ* in TC genotype-transfected U937 cells stimulated with LXR ligand (T0901317), and those of TNF-*α*, COX-2, and IFN-*γ* in TT genotype-transfected U937 cells ([Figure 5(c)](#fig5){ref-type="fig"}).](MI2019-6217548.005){#fig5}
{#fig6}
[^1]: Academic Editor: Calogero Caruso
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
In July 2014, the University of Hawai'i hosted the First Island Biology Symposium, where nearly 450 island biologists from all over the world and from very different island biology subdisciplines (Island Biogeography, Phylogeography, Ecology, Palaeoecology, Conservation Biology, etc.) joined together for the first time ([@PLV148C17]). The aim of the event was to launch a regular series of International Island Biology conferences, bringing together researchers who, by the very nature of the islands they study, face geographical barriers to communication and collaboration. This event was broadly recognized as a big success and the Second Symposium, following the agreed policy of rotating the ocean of the archipelago hosting the next event, will be held in Angra do Heroismo, Azores, in summer 2016 (<http://www.islandbiology2016.uac.pt/>).
The year of the first Island Biology conference coincided with the 40th anniversary of the publication of an outstanding book that inspired a generation of island biologists: Sherwin Carlquist\'s *Island Biology*, published in 1974 by Columbia University Press. Unfortunately, unforeseen circumstances at the last minute prevented Sherwin Carlquist from attending the conference and presenting a keynote address.
This special issue brings together some of the most interesting botanical contributions presented at the Symposium. The 18 contributions that we compiled deal with very different themes and come from very different archipelagos across the world and disciplines within plant science. Indeed, the broad coverage of themes relevant to island biology and of different islands, archipelagos and ocean regions was a key feature of the conference. It was a strong indication that island biology is flourishing, and given the importance of Sherwin Carlquist\'s work across many research fields, it showed that his legacy lives on and grows. We will first provide a short overview of the life and work of the inspiring scientist to whom we dedicate the special issue, Sherwin Carlquist, before summarizing the main findings of each contribution in the special issue.
Sherwin Carlquist\'s life and work {#s1a}
----------------------------------
Sherwin John Carlquist was born in 1930 in California. From a young age, he was fascinated with islands, very likely due to reading some books about the Galápagos archipelago borrowed from Los Angeles School Library where Sherwin\'s mother worked. He received his undergraduate degree (1952) and a Ph.D. in Botany (1956) from the University of California, Berkeley. Following a postdoctoral stay at Harvard University, he began his teaching career at the Claremont Graduate School. In 1977, he also began teaching at Pomona College and continued working at both institutions until 1992. From 1984 to 1992, Carlquist was the resident Plant Anatomist at Rancho Santa Ana Botanic Garden. His last post was as an adjunct professor at University of California at Santa Barbara from 1993 to 1998.
Although trained as a wood anatomist, the early opportunity of doing research in Revillagigedo (Mexico) and Hawai'i strengthened the allure of oceanic island floras to the young Carlquist. During his life, he has worked in many other island groups around the world, including Guadalupe, Society Islands, Samoa, Fiji, New Caledonia, Vanuatu, New Guinea, New Zealand, Australia, Taiwan, Japan, Borneo, Ryukyu Islands, Ogasawara (Bonin) Islands and the Subantarctic islands of New Zealand.
His scientific production includes \>300 scientific articles, the majority being related to wood anatomy, as well as \>10 books, 4 of them dealing with islands: *Island Life* ([@PLV148C5]), *Hawaii: A Natural History* ([@PLV148C6]), *Island Biology* ([@PLV148C7]) and *Tarweeds and Silverswords: Evolution of the Madiinae (Asteraceae)* ([@PLV148C8]). While all of these are very important contributions to our knowledge of island biology in general, and for Hawaiian natural history in particular, one of them, *Island Biology* ([@PLV148C7]), became a major milestone in the field, comparable with classic books such as Darwin\'s *On the Origin of Species*, ([@PLV148C9]), Wallace\'s *Island Life* ([@PLV148C35]) or MacArthur and Wilson\'s *The Theory of Island Biogeography* ([@PLV148C19]). In that book, Carlquist displayed, in an outstandingly organized structure, his encyclopaedic knowledge about the evolutionary trends and phenomena occurring among insular plants (and animals) and the selection pressures that drive them. The resulting set of evolutionary innovations is characteristic of island taxa and is known as the *island syndrome* ([@PLV148C1]). Four of Carlquist\'s most important contributions to the development of the biology of island plants were as follows: *Long-distance dispersal to islands and secondary loss of dispersability on islands*. Following the classic ideas of Darwin, Wallace and Hooker, Carlquist was convinced of the importance of long-distance dispersal as a way to colonize oceanic islands and of the role of these as stepping stones between continents. In *Island Biology*, he formulated his famous 24 principles of dispersal and evolution ([@PLV148C23]), addressing the evidence for and implications of long-distance dispersal. He also focussed his attention on the several ways island plant (and animal) lineages, after arriving on an island via long-distance dispersal, lost or shifted their dispersal abilities when the ancestral trait was no longer adaptive. The increases in fruit size and weight, making the original fruit design no longer functional for dispersal (e.g. the Polynesian genus *Fitchia* of the Asteraceae---a subject on which Carlquist wrote his Ph.D. dissertation) or the flightlessness achieved by both birds and insects, are outstanding examples of this phenomenon.*Adaptive radiation*. Using his outstanding knowledge of island plants, Carlquist provided a comprehensive synthesis of information on the most important adaptive radiations of vascular plants on islands worldwide. He dedicated at least seven chapters of *Island Biology* to this subject, including a general introduction to the theme and detailed examples of evolutionary processes in Hawaii, Macaronesia, Galápagos, Juan Fernández, New Caledonia, New Zealand, islands off the coast of Western Australia and other islands.*Reproductive biology on islands*. In his book, Carlquist displayed his profound knowledge of floristic sexual systems providing plentiful examples of different reproductive systems among island plants (especially on Hawai'i). He emphasized a general insular rule promoting outcrossing, i.e. self-compatible continental ancestors evolve different types of outcrossing on islands (dichogamy, herkogamy), via anemophily, heterostyly and all the possible stages in the transition from hermaphroditism to dioecy. He also elaborated upon hybridization in insular floras and its significance.*Insular woodiness*. Carlquist proposed that insular secondary woodiness is a result of evolution from continental herbaceous ancestors. He hypothesized that a release from seasonality occurred on islands---due to the buffer effect of the surrounding ocean---leading recurrently in different archipelagos (Hawai'i, Galápagos, Macaronesia, etc.) and across many plant families to the *in situ* evolution of a woody habit from herbaceous plants. His view was in contrast to the traditional explanation by European island biologists for whom insular woodiness was not a derived feature but instead a basal characteristic of insular palaeoendemic species that became extinct on continents due to climatic or geological events after establishing on the islands. Only the emergence of phylogenetic analyses, a couple of decades after Carlquist\'s claims, made definitively clear that his interpretation was correct.
Long-distance dispersal {#s1b}
-----------------------
The topics of more than a third of the papers covered in this special issue are related to the first three themes indicated above. Three papers advance our knowledge of different aspects of long-distance dispersal and its importance to island plant biology. **[Alsos, Ehrich, Eidesen, Solstad, Westergaard, Schönswetter, Tribsch, Birkeland, Elven and Brochmann (2015)](Alsos, Ehrich, Eidesen, Solstad, Westergaard, Schönswetter, Tribsch, Birkeland, Elven and Brochmann (2015))** present the first comprehensive study of long-distance dispersal to oceanic islands based on combined population genetic and floristic similarity analyses. The authors studied 25 representative vascular plant species of five Arctic Ocean islands (Greenland, Iceland and Jan Mayen) or island groups (Svalbard and Faroe), with the aim of shedding light on the origin and timing of their present floras\' composition. They were able to detect that most plant species colonized those islands after the Last Glaciation through multiple long-distance dispersal events from several source regions, including Europe, Asia and North America, some of them travelling as far as 3000 km. The authors also found that the relative intensity of the founder effect was similar at the species and gene level, broadly corresponding with the predictions of the Island Equilibrium Theory ([@PLV148C19]), and indicating that species and genetic diversities on islands are shaped by similar processes. They report that insect-pollinated species show a strong founder effect that increases with island isolation and decreases with island size, whereas only a weak founder effect was found for wind-pollinated outcrossing species. Finally, they found that colonization patterns among the study islands were largely congruent, indicating that despite the importance of stochasticity, long-distance dispersal is mainly determined by the availability and geographic configuration of dispersal vectors. Another study showing that dispersal of species to oceanic islands from continental sources were not single events that occurred at random in the past is that by **[Wolf, Rowe, Der, Schilling, Visger and Thomson (2015)](Wolf, Rowe, Der, Schilling, Visger and Thomson (2015))**. By examining the morphological and genetic variation among bracken fern species (*Pteridium*) within the Galápagos Islands, these authors demonstrate that these processes can be dynamic and ongoing. The Galápagos archipelago sits on the equator ∼1000 km west of South America. Phylogeographic analyses show that *P. esculentum* from South America and *P. aquilinum* from the northern hemisphere have repeatedly colonized the islands in the past; *P. esculentum* populations have established numerous times, whereas *P. aquilinum* has not, but has resulted in the formation of hybrid combinations with *P. esculentum* (homoploid and allotetraploid) referred to as *P. caudatum*. This leads to speculation as to the origins, diversifications and evolutionary changes in other widely dispersed species inhabiting oceanic islands. In another study, **[Vargas, Arjona, Nogales and Heleno (2015)](Vargas, Arjona, Nogales and Heleno (2015))** analysed the plant traits that allow long-distance dispersal and their effectiveness. These authors performed floristic and syndrome analyses on the native recipient flora of an Atlantic Ocean archipelago (Azores), finding that diplochorous species, i.e. those in which seed dispersal occurs by a sequence of two or more steps or phases, each involving a different dispersal agent, are over-represented relative to species in mainland Europe, but not when compared with their most likely propagule\'s source (mainland Portugal). They further analysed inter-island colonization patterns carried out in three oceanic archipelagos with well-studied floras (Azores, the Canaries and Galápagos) showing, as expected, a general trend of a higher number of islands colonized by vascular plant species with one or two long-distance dispersal syndromes than by unspecialized species. Nevertheless, statistical significance for differences in colonization was limited in some cases due to the low proportion of diplochorous species existing anywhere. Contrary to expectations, the authors assert that only a very marginal advantage for long-distance dispersal of species bearing multiple syndromes was observed.
Adaptive radiation {#s1c}
------------------
Two papers in the special issue deal with adaptive radiation. Examples of adaptive radiation in oceanic islands due to colonization of new habitats distinct from the parent populations (e.g. cladogenesis) are abundant. However, diversification following colonization of similar habitats and accumulation of mutations eventually resulting in speciation (e.g. anagenesis) is also a common pattern among island floras ([@PLV148C29]). **[Takayama, López-Sepúlveda, Greimler, Crawford, Peñailillo, Baeza, Ruiz, Kohl, Tremetsberger, Gatica, Letelier, Novoa, Novak and Stuessy (2015)](Takayama, López-Sepúlveda, Greimler, Crawford, Peñailillo, Baeza, Ruiz, Kohl, Tremetsberger, Gatica, Letelier, Novoa, Novak and Stuessy (2015))** investigated the level of variation within and among species of five genera in the Juan Fernández Archipelago that demonstrate both cladogenetic and anagenetic speciation. Their key finding is that anagenetically derived species have populations with high levels of genetic variation yet no geographic genetic structure, whereas cladogenetic populations have less diversity within and among them. Their results also show important corollaries between island ages and levels of diversity. By examining multiple lineages, they demonstrate that these findings will be widely beneficial to the study of island evolution. Floras of many islands or island archipelagos are well studied and the species well known. However, it is not uncommon for new species to periodically be discovered. In some cases, new species are found during field surveys ([@PLV148C28]), and others may occur during taxonomic revision. The flora of the Cape Verde Islands consists of ∼740 species (∼12 % endemic) and has been well studied. **[Romeiras, Monteiro, Duarte, Schaefer and Carine (2015)](Romeiras, Monteiro, Duarte, Schaefer and Carine (2015))** demonstrated through molecular phylogenetic analyses of three endemic plant lineages on the Cape Verde Islands (genera *Cynanchum*, *Globularia* and *Umbilicus*) that additional species within each genus are in need of recognition. This study points out that the number of species in many of the plant lineages within this archipelago should be re-evaluated and the number of species within the flora, along with the level of endemism, is likely to increase substantially.
Reproductive biology {#s1d}
--------------------
Plant reproductive traits are dealt with in two papers, one by **[Lord (2015)](Lord (2015))** and the other by **[Watanabe and Sugawara (2015)](Watanabe and Sugawara (2015))**. The former reports the most comprehensive survey to date of reproductive traits of the floras of Subantarctic Islands---islands in the Southern Oceans such as the Falkland, Kerguelen or Macquarie islands. Southern Ocean islands have seen a steady increase in research interest, often related to large-scale habitat restoration and invasive species control efforts. Equally, both macroecological and experimental work on the reproductive biology of island plants have re-gained momentum, years after such questions had been at the core of Carlquist\'s work. [@PLV148C18] tests three interrelated classical hypotheses: whether self-compatibility is higher on islands as a result of self-fertile taxa being favoured as colonizers (Baker\'s law; [@PLV148C43]); whether wind pollination is more common as a result of a pollinator-poor environment and whether gender dimorphism (dioecy) is also more frequent, resulting from selection against inbreeding risk in small island populations. The study reports indeed very high levels of self-compatibility (92.6 %, the highest percentage reported to date from any island flora), but found no clear trend for increased wind pollination or dioecy, partly because of the confounding factors of cold climates and trait frequencies in source floras. This highlights the importance of comprehensive data sets that include many islands across complete latitudinal and climate gradients and information from source floras for testing hypotheses in island biology ([@PLV148C17]). The study finds, however, strong indication that for many plant species, insect pollination is likely important, at least as part of a mixed strategy together with wind pollination, and thus, the work highlights the value of more detailed work on plant--pollinator interactions on these very isolated and cold Southern Ocean islands. Populations on oceanic islands are often small, and self-compatibility that can promote successful colonization may lead to inbreeding and the consequences associated with it. Island lineages have repeatedly evolved away from self-compatibility towards systems that promote outcrossing, including monoecy, dioecy, gynodioecy, androdioecy, heterostyly and other strategies. The evolution of heterostyly on oceanic islands is compounded by the complexity of the genetic system involved in maintaining it and, as such, is thought to be a rarity on remote island systems. [@PLV148C38] investigated the frequency of heterostyly among oceanic island lineages and then explore the evolution of this trait within *Psychotria* (Rubiaceae), a genus of ∼2000 species, widespread throughout the tropics and exhibiting variation for the heterostylous habit. Their study sheds light on the rarity of heterostyly on islands.
Plant conservation {#s1e}
------------------
Many contributions in the special issue are highly relevant to plant conservation on islands. **[Birnbaum, Ibanez, Pouteau, Vandrot, Hequet, Blanchard and Jaffré (2015)](Birnbaum, Ibanez, Pouteau, Vandrot, Hequet, Blanchard and Jaffré (2015))** present a very comprehensive data set on the small-scale distribution of over 700 native tree species on New Caledonia, which is in many ways unique for its very high species diversity, its complex geological and therefore biogeographic history and its habitats characterized by extreme, ultramafic soil substrates. This study reports two main conclusions. First, the habitat and environmental distribution of most of the species is broad, i.e. they occur across broad elevational, climatological and geological gradients, or to put it differently, they generally have broad ecological niches and are not specialists of one particular soil substrate or climate. Second, most species are highly spatially clustered, which the authors interpret as a result of reduced dispersability of island plants; an alternative or complementary explanation might be that this is at least partly a result of anthropogenic land use and resulting habitat fragmentation. Remnant native species distributions on many oceanic islands---or on islands that separated from continents millions of years ago---are nowadays confined to very small habitat fragments ([@PLV148C16]). The results of [@PLV148C4] have important implications for plant conservation on such islands. It might be possible to conserve threatened species in a wide range of environmental conditions, but they will not themselves be able to overcome habitat fragmentation. Remnant fragments of high biodiversity must be strictly protected, and gene exchange between the last remnant stands restored through active conservation management. An interesting case study on the conservation biology of a particular threatened plant species in the Azores is provided by **[Silva, Dias, Sardos, Azevedo, Schaefer and Moura (2015)](Silva, Dias, Sardos, Azevedo, Schaefer and Moura (2015))**. These authors integrated information on the spatial distribution (including modelling of the environmental niche), population demography, conservation genetics and threats to a chamaephyte, *Veronica dabneyi*, assembling a comprehensive data basis for developing a conservation strategy for the species.
Understanding contemporary ecology and being able to plan appropriate conservation strategies requires an understanding of the past, and of how humans have shaped the modern biota. This is especially so for islands, where anthropogenic effects have often been recent and profound, causing extinctions and significant biological invasions ([@PLV148C7]). In this special issue, **[Shiels and Drake (2015)](Shiels and Drake (2015))** show that a key factor in the rarity and extinctions of once-dominant endemic palms (*Pritchardia* spp.) of the Hawaiian Islands is predation of their seeds by rats---*Rattus exulans*, introduced by native Hawaiians at their arrival 750--800 years ago ([@PLV148C40]), and *R. rattus* introduced in the last 240 years. Identifying seeds as the most vulnerable stage of the palms\' life history and rats as their main predator is a key step towards their conservation. Most tropical forest trees rely on animals for seed dispersal ([@PLV148C12]), yet the faunas of most tropical, oceanic islands have become depleted through extinction, extirpation and population decline following human colonization ([@PLV148C39]). As a result, many island plant species may currently suffer from dispersal limitation ([@PLV148C22]). Another contribution in this issue that shows the threats of seed dispersal disruptions is by **[McConkey and Drake (2015)](McConkey and Drake (2015))**, who demonstrate that flying foxes (*Pteropus tonganus*) in Tongan rain forests are the sole effective dispersers of 57 % of the plant species whose fruits they consume, and that they are especially important for large-seeded species. Combined with previous work which found that flying foxes are effective dispersers only when they forage at high densities ([@PLV148C20]), these new results reinforce the flying fox\'s role as the last remaining functional disperser for many tree species. The importance of seed dispersers for plant recruitment in islands is also highlighted by **[Wandrag, Dunham, Miller and Rogers (2015)](Wandrag, Dunham, Miller and Rogers (2015))**. These authors compared the forest seed banks of a Micronesian island (Guam) that lacks seed dispersers with those of two islands where dispersers still exist. A key finding was that species in the seed bank were more likely to be found near adult conspecifics on the island without dispersers than on those with dispersers. The seed bank represents the viable pool of seeds in the soil from which new plants can recruit into plant communities, and its composition is determined by a combination of seed traits (e.g. germination, longevity and dormancy), the spatial pattern of seed dispersal and the actions of post-dispersal threats (e.g. pathogens and seed predators) ([@PLV148C10]). Studying the seed bank\'s composition in disturbed sites/islands can provide us with useful information on the ultimate consequences of the loss of dispersers, predators, herbivores, etc. on plant species distribution and abundance.
Habitat restoration {#s1f}
-------------------
In recent years, the study of species interactions using network theory has proved to be an increasingly good framework to predict consequences of disturbance at a community level ([@PLV148C3]) and to advance conservation management. By using plant--pollinator networks and islands as model systems, **[Kaiser-Bunbury and Blüthgen (2015)](Kaiser-Bunbury and Blüthgen (2015))** identify a series of quantitative metrics to describe changes in network patterns that have implications for conservation. They distinguish between the 'diversity' and 'distribution' of interactions on three hierarchical levels (species, guild/group and network). Some of the metrics may be suitable indicators of anthropogenic changes in pollinator communities, and may allow assessment of the structural and functional robustness and integrity of ecosystems. They, thus, claim that a conservation network approach may be beneficial for advancing adaptive management. **[González-Castro, Yang, Nogales and Carlo (2015)](González-Castro, Yang, Nogales and Carlo (2015))** used fleshy fruited plants and frugivorous birds in the Canary Islands to test which variables best explained the interactions in a frugivory network: the phenotypic match between bird and fruit traits, or the relative abundance of birds and fruits. The most important explanatory variables were phenotypic, especially the overlap between fruit size and bill width, but fruit and bird abundance also played a significant role. This study provides a useful model for examining mutualistic networks in more diverse and complex systems. Ecological networks are a good tool to understand how alien species infiltrate the receptive communities as well as how and to what extent they can impact and modify the structure of such communities. **[Traveset, Chamorro, Olesen and Heleno (2015)](Traveset, Chamorro, Olesen and Heleno (2015))** show that alien plant species integrate easily into the native communities in the Galápagos and that the impact on the overall network structure is rather low, except for an increase in network selectiveness (i.e. species become more selective in their partner\'s choice, interacting with species that tend to be less abundant). Following a study on the emergence of novel communities in Galápagos ([@PLV148C31]), these authors find that aliens represent a high fraction (\>50 %) of the total number of interactions, and that they prevail in all habitats and seasons. In spite of the low overall effect on network structure, they suggest that aliens may well have an important effect on community functioning as they may influence the reproductive success of native species. When native species are lost, their functional role in the community may or may not be replaced by alien species. In a study carried out in Hawai'i, **[Pejchar (2015)](Pejchar (2015))** found that introduced birds can act as seed dispersers of native plants replacing---but only to a limited extent---a native frugivorous bird ('ōma'o, *Myadestes obscurus*) in areas in which it has become extinct. This author claims that alien birds are imperfect substitutes for native species and that seed dispersal patterns actually are notably altered following local extinctions of 'ōma'o, and suggests reintroducing it to suitable habitats within its historical range.
Species losses and gains result in major shifts in ecosystem composition and function ([@PLV148C37]), and this presents challenges for restoring island ecosystems. Island biologists are increasingly considering reintroducing ecologically important species to suitable habitats within their historic range and to neighbouring islands with depauperate communities of vertebrate seed dispersers. **[Hansen (2015)](Hansen (2015))** is one such study advocating 'rewilding' introductions of large non-native tortoises to small areas of islands where large native herbivores have been extinct for hundreds of years: flightless birds such as moa-nalo in Hawai'i and giant native tortoises on Rodrigues and Round Island in the Mascarene Islands. On these islands, attempts to restore native ecosystems are difficult because of multiple invasive alien plants that can out-compete native plants. In his study, Hansen presents preliminary data to show that native woody plant growth benefits when non-native large tortoises graze sites. On this basis, he proposes carefully monitored introductions of large tortoises worldwide to islands where large herbivores are extinct as a tool to assist the restoration of degraded island ecosystems.
Palaeoecological studies can help establishing whether plant species have been introduced to islands. In the Galápagos Islands, pollen evidence from before human settlement revealed that plant species that had been considered introduced to the islands were in fact native ([@PLV148C33]). In contrast, pollen evidence from the Subantarctic Auckland Islands, south of New Zealand, shows that a tree, *Olearia lyallii*, is not native to the islands (**[Wilmshurst, McGlone and Turney 2015](Wilmshurst, McGlone and Turney 2015)**). This species first appears in the pollen record in the early 19th century, when it was probably inadvertently introduced by sealers from the Snares Islands, further to the north. Historical paintings and photographs during a period of settlement on the Auckland Islands in the mid-19th century show areas of forest clearance and burning, and the authors of this contribution claim that these are the areas where the tree has become locally dominant up to the present.
Concluding Remarks and Future Directions {#s2}
========================================
The First International Island Biology conference in 2014 in Honolulu and the resulting special issue demonstrate that islands are reasserting their role as ideal model systems for the latest research questions and methods in ecology, evolutionary biology, biogeography and conservation. The key research questions related to long-distance dispersal, adaptive radiation and plant reproductive systems, broadly covered 40 years ago by Sherwin Carlquist in his classic book, *Island Biology*, are now being addressed with much more comprehensive data and new methodologies. While island biology might indeed be entering a new golden era of research ([@PLV148C11]), we are also at the brink of losing these unique biological systems for research and humanity. Literally, thousands of endemic island species survive as only a few individuals or small and fragmented populations---and for most of them, very little is known about their basic biology. What is unequivocally clear, however, is that most of these species will disappear from the wild in this century unless we markedly intensify our conservation efforts.
Contributions by the Authors {#s3}
============================
A.T. led the writing but all authors contributed similarly to this article.
Conflict of Interest Statement {#s4}
==============================
None declared.
We want to thank the chief editors of *AoB PLANTS* for inviting us to prepare this special issue and the University of Hawai\'i for providing support for the 1st International Conference on Island Biology. We are indebted to all reviewers of the manuscript for their good work. We dedicate the special issue to Sherwin Carlquist for being a source of inspiration and for the wonderful legacy he is leaving to the present and future island biologists worldwide.
[^1]: **Associate Editor:** J. Hall Cushman
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
*Erysipelothrix* are gram-positive, rod-shaped, facultative anaerobic bacteria. Based on phylogenetic relatedness, *Erysipelothrix* spp. can be categorized into *Erysipelothrix rhusiopathiae* (serovars 1a, 1b, 2, 4, 5, 6, 8, 9, 11, 12, 15, 16, 17, 19, 21, and N), *Erysipelothrix tonsilarium* (serovars 2, 7, 10, 14, 20, 22, and 23), *Erysipelothrix* sp. strain 1 (serovar 13), *Erysipelothrix* sp. strain 2 (serovar 18), and *Erysipelothrix inopinata*, a novel species that was recently isolated from sterile-filtered vegetable broth.^[@bib1],[@bib2],[@bib3],[@bib4]^ These *Erysipelothrix* species are ubiquitous in nature and can cause diseases in a variety of animals including swine, humans, poultry, sheep, cattle, and wild animals. The diseases caused by these species are called erysipelas in animals and erysipeloid in humans.^[@bib5]^ Humans can be infected with pathogens from the pigs.^[@bib6]^
Swine erysipelas is caused by *E. rhusiopathiae*, which enters the bloodstream through the tonsils and other lymphoid tissues of the alimentary canal. It is estimated that 30%--50% of healthy pigs carry *E. rhusiopathiae* in the tonsils. The carriers serve as a reservoir for acute erysipelas outbreaks and may not have clinical signs.^[@bib7]^ The clinical presentation of swine erysipelas are usually manifested as acute septicemia or chronic disease characterized as endocarditis and polyarthritis.^[@bib8]^
Swine erysipelas appears worldwide and causes economic loss to the swine industry.^[@bib9]^ Within the Chinese swine industry, the occurrence of erysipelas has significantly decreased since the 1990s due to intensification and improved management of the industry. It only occasionally occurs in small farms. However, since 2010, we have observed an increase in the number of cases of erysipelas submitted to the Huazhong Agricultural University Clinical Microbiology Laboratory. In addition, other research groups have reported similar trends in China.^[@bib10]^ Whether this increased occurrence of swine erysipelas is due to the development of antimicrobial resistance or extension of a specific lineage is not clear, because information regarding the molecular characteristics and antibiogram of *E. rhusiopathiae* remains limited. To the best of our knowledge, molecular epidemiology of *E. rhusiopathiae* has not been conducted in China.
In the present study, 48 strains of *E. rhusiopathiae* isolated from diseased pigs in China in 2013--2014 were subject to antimicrobial susceptibility testing, virulence gene typing, and pulsed-field gel electrophoresis (PFGE) genotyping to improve our understanding of the virulence traits and molecular profile of *E. rhusiopathiae* in China.
Materials and methods
=====================
The 48 *E. rhusiopathiae* strains were collected from liver, spleen, and blood samples of pigs with a clinical history of acute septicemia by the Clinical Microbiology Laboratory at Huazhong Agricultural University during 2012--2013. These samples were from Southern and Central China, including Zhejiang, Hunan, Hubei, Anhui, Guangdong, Henan, and Jiangsu provinces. Each sample was collected from different pigs and strains with typical morphological characteristics of *E. rhusiopathiae* were picked from each plate. The strains were identified as *E. rhusiopathiae* on the basis of polymerase chain reaction and cellular morphology; growth in gelatin; positive reactions in triple sugar iron agar slants for glucose, lactose, and arabinose fermentation by arginine dihydrolase; and negative reaction for catalase, oxidase, urease, nitrate reduction, mannose, and sucrose fermentation, as described by Wang *et al.*^[@bib11]^
Pulsed-field gel electrophoresis (PFGE)
---------------------------------------
The total DNA of *E. rhusiopathiae* was purified using protocol described previously.^[@bib12]^ DNA was digested with a mixture of *XbaI*, 0.1% bovine serum albumin, and buffers (Takara Biotechnology, Dalian, China) at 37 °C for 8 h. Digested DNA plugs was loaded in 1% agarose gels and run in an contour-clamped homogeneous electric field (CHEF-DRIII; Bio-Rad, Shanghai, China) for 22 h at 14 °C and 6 V with pulse times from an initial 2.2 s to a final 64 s. The gel was subsequently stained by ethidium bromide for detecting PFGE patterns.
Antimicrobial susceptibility testing
------------------------------------
Susceptibility to ampicillin, erythromycin, cefotaxime, norfloxacin, cefazolin, sulfadiazine, amikacin, polymyxin, tetracycline, doxycycline, lincomycin, and levofloxac in was evaluated by microdilution technique, according to the protocol of the Manual of Antimicrobial Susceptibility Testing described by Stephen *et al.*^[@bib13]^ Quality control was performed with *Staphylococcus aureus* (ATCC29213) and *Streptococcus pneumonia* (ATCC49619). The breaking point of the minimum inhibitory concentration (MIC) was cited from the Clinical and Laboratory Standards Institute.^[@bib14]^
Detection of virulence-associated genes
---------------------------------------
The presence of capsule synthesis gene (cpsA-C), neuraminidase (nanH.1 and nanH.2), hyaluronidase (hylA-C), surface protective antigen (spa), adhesion, rhusiopathiae surface protein (rspA and rspB), and nine putative virulence genes, including patatin-like phospholipase A and B (locus tag: ERH_0072 and ERH_00334, respectively; GenBank assembly accession: GCA_000270085.1), phospholipase/carboxylesterase family protein A and B (locus tag: ERH_0083 and ERH_0347, respectively; GenBank assembly accession: GCA_000270085.1), lysophospholipase A, B, and C (locus tag: ERH_0148, ERH_1214, and ERH_1433, respectively; GenBank assembly accession: GCA_000270085.1), cardiolipin synthetase, and phospholipase D was investigated using primers described in [Supplementary Table S1](#xob1){ref-type="supplementary-material"}. Polymerase chain reaction (PCR) was carried out in a 25 μL mixture containing 1 μL DNA, 0.5 μL each primer, 2.5 μL 10× Ex Taq buffer, 2 μL dNTP mixture, 19.25 μL deionized distilled water, and 0.25 μL Ex Taq DNA polymerase (Takara, Dalian, China). Each PCR test was repeated for three times.
Data analysis
-------------
The PFGE patterns were analyzed with BioNumerics software version 6.5 (Applied Maths, Kortrijk, Belgium). A dendrogram was calculated using the unweighted pair group method using arithmetic averages (unweighted pair group method with arithmetic mean (UPGMA)), dice coefficient, and 1.5% optimization with 1.5% band position tolerance. Strains with similarities \>70% had eight band differences and were clustered in a clonal complex. Comparisons between different complexes were tested by Fisher\'s exact test (two-tailed), if a significant result between different complexes was found, comparisons between each complex and all other complexes combined were subsequently tested as well. A *P* value \< 0.05 was considered significant.
Results
=======
Molecular characterization of *E. rhusiopathiae*
------------------------------------------------
PFGE resulted in 32 distinct patterns from 48 *E. rhusiopathiae* strains ([Figure 1](#fig1){ref-type="fig"}). Dendrogram analysis showed that, at 30% divergence, four clonal groups (A--D) were identified. Twenty-five patterns were observed in clonal complexes A and B, suggesting a high diversity of genetic characterization in these two predominant clonal complexes. Clonal complex A accounted for 48% (23/48) of all strains, with a similarity of 78%--96%. Clonal complex B accounted for 35% (17/48) of strains, with a similarity of 84%--97.5%. Only two strains were classified into clonal group C and showed 83% similarity. Clonal group D included six strains, which showed five different PFGE patterns. Strains with similar PFGE patterns were recovered from different provinces and at different times. Strains 1231 and 13002 shared the same PFGE pattern, but strain 1231 was isolated from Hunan province in 2012 and strain 13002 from Hubei province in 2013. Similarly, strain 1210 isolated from Jiangsu province in 2012 also had the same PFGE pattern as that of strain 13007 isolated from Hubei province in 2013. PFGE pattern of strain 13001, 13005, and 13006 was isolated from samples from Hubei and Henan provinces during the same period.
Antimicrobial susceptibility
----------------------------
The MICs of 13 antibiotics for the 48 strains are shown in [Table 1](#tbl1){ref-type="table"}. All the stains were susceptible to ampicillin, erythromycin, and cefotaxime, whereas all were resistant to kanamycin, cefazolin, sulfadiazine, and amikacin. Differences in antimicrobial susceptibility were detected for gentamicin (47/48 resistant), tetracycline (29/48 resistant), doxycycline (18/48 resistant), lincomycin (37/48 resistant), norfloxacin (37/48 resistant), and levofloxacin (34/48 resistant). With regard to multi-resistance profiles, all strains grouped into 15 resistance phenotypes were resistant to at least five of the ten antimicrobials tested. Thirteen strains were resistant to all six antibiotics mentioned above. Most of them were in clonal complex A (*n* = 8), followed by clonal complexes D (*n* = 3) and B (*n* = 2), and none was found in clonal complex C. There was no difference in antimicrobial susceptibility pattern among the clonal complexes.
Virulence-associated genes
--------------------------
A total of 21 virulence-associated genes were screened by PCR in this study ([Supplementary Table S2](#xob1){ref-type="supplementary-material"}). The number of virulence genes detected in *E. rhusiopathiae* ranged from 16 to 21. Strains with all tested virulence genes were evenly distributed among the clonal complexes. Two strains with the fewest virulence genes (*n* = 16) were in clonal complex A, which was the most predominant clonal complex. Capsule polysaccharide synthesis gene (cps-A, B and C), spaA, rspA, nanH. 2, patatin-like phospholipase, lysophospholipase, phospholipase/carboxylesterase, and phospholipase/carboxyl esterase family genes were detected in all *E. rhusiopathiae* strains. Various virulence gene detection rates were observed for the rest of the genes. Among them, phospholipase D and adhesin genes were detected in fewest strains, with detection rates of 63% (30/48) and 77% (37/48), respectively. The virulence gene profiles followed no specific pattern and were independent of the four clonal complexes.
Discussion
==========
*E. rhusiopathiae* is an opportunistic pathogen that can cause acute septicemia or chronic endocarditis and polyarthritis in pigs.^[@bib7]^ Diseases caused by *E. rhusiopathiae* generally occur sporadically in China. Nevertheless, we observed an increased frequency of *E. rhusiopathiae* isolation from diseased pigs after 2010. This increasing trend highlights the importance of understanding the molecular epidemiology of *E. rhusiopathiae*. However, there is no previous report about the molecular epidemiology of *E. rhusiopathiae* in China. We therefore genotypically and phenotypically characterized 48 *E. rhusiopathiae* strains from seven provinces of Southern and Central China, to provide a basis for surveillance of this disease in the future.
PFGE is one of the most discriminative typing methods, which provides important insight into population structures of many pathogens, including *E. rhusiopathiae*.^[@bib15],[@bib16]^ Our data showed 32 distinct PFGE patterns from 48 *E. rhusiopathiae* strains. These PFGE patterns were classified into four clonal complexes (A--D). The majority of strains and patterns fell into clonal complexes A and B. The high variation within these predominant clonal complexes indicates that most of the swine erysipelas is not closely related. Therefore, *E. rhusiopathiae* in China may originate from different lineages and sources instead of expansion of a single clonal lineage across different regions. This corresponds to the pathogenic trait of *E. rhusiopathiae* as an opportunistic pathogen that causes sporadic disease in immunocompromised hosts.^[@bib13]^ However, three PFGE patterns from strain 1231, 13002, 1210, 13007, 13001, 13005 and 13006 were found in different provinces during different times. These clones may spread through trade between these regions or they may possess advantageous traits that facilitate their spread and recurrence.
The antibiotic susceptibility test showed that our strains were all susceptible to ampicillin, erythromycin and cefotaxime, a third-generation cephalosporin. As cefotaxime are not drug intended for use in food animals, ampicillin and erythromycin should be the first choice for the treatment of swine erysipelas in China. Susceptibility of our strains to ampicillin is in line with previous reports in Japan and North America.^[@bib12],[@bib17],[@bib18]^ However, resistance to erythromycin was found in 5.8%--6.1% of strains isolated from Japan but not from our collection of *E. rhusiopathiae* strains.^[@bib17],[@bib18]^ The difference in erythromycin resistance in these countries may be due to frequent use of erythromycin for treatment of porcine bacterial respiratory diseases in Japan but not in the swine industry in China.^[@bib19]^ All *E. rhusiopathiae* strains were resistant to cefazolin, sulfadiazine, amikacin and kanamycin and showed high resistance to gentamicin 98% (47/48), followed by tetracycline 60% (29/48), doxycycline 38% (18/48), lincomycin 77% (37/48), norfloxacin 77% (37/48) and levofloxacin 71% (34/48). Resistance to these antibiotics was in agreement with previous reports by other research groups in Japan, North America and Brazil.^[@bib12],[@bib20],[@bib21]^ Only few strains in Brazil was reported to be resistant to fluoquinolones and cefazolin has not been reported before. However, all the *E. rhusiopathiae* strains in this study were resistance to cefazolin and \>70% of strains were resistant to norfloxacin and levofloxacin, which are fluoquinolones. The high resistance to these antibiotics indicates over- or misuse of these classes of antibiotics in bacterial infections in swine production in China.
Also, to understand better the virulence potential of our study strains, we characterized 21 virulence-associated genes, including nine putative phospholipase genes and 12 well-characterized and accepted genes: major surface protective antigen (spaA), capsule polysaccharide gene (cpsA-C), rhusiopathiae surface protein (rspA and rspB), hyaluronidase (hylA-C), neuraminidase (nanH.1 and nanH.2), and adhesin, and nine phospholipase genes predicted to be potentially involved in intracellular survival.^[@bib14]^ The results showed a high detection rate of the above genes in our *E. rhusiopathiae* strains. Eleven genes (cpsA--C, spaA, rspA, rspB, hylA, nanH. 2, patatin-like phospholipase B, lysophospholipase C, and phospholipase/carboxylesterase family genes A and B) were detected in all strains and 30% (15/48) of strains possessed all the virulence genes tested in this study. Phospholipase D and adhesin genes were detected in the fewest strains, with detection rates of 63% (30/48) and 77% (37/48), respectively. The low detection rate of these two genes may have been because they are not essential for the pathogenesis of *E. rhusiopathiae*, or because of the presence of genes with similar function in the genome. Different detection rates were observed for virulence genes hylC, hylA, nanH.1, patatin-like phospholipase A, lysophospholipase A and B, adhesin, and cardiolipinsynthetase; however, there was no correlation between these genes and clonal complexes.
The results of this study could not provide a definitive answer for the current increasing trend of swine erysipelas in China. However, the high variation in phylogenetic characteristics of strains from different regions shows that the increasing trends are caused by strains of various clonal lineages from multiple sources. One explanation for this result is failure to use vaccines. There are three kinds of killed bacterins and two kinds of live-attenuated vaccines for swine erysipelas available in China. However, because of the low occurrence of swine erysipelas after the 1990s, many of the swine farms have not included *E. rhusiopathiae* vaccine in routine regimens. Another explanation may be concurrent virus infection, such as with porcine parvovirus, porcine reproductive and respiratory syndrome virus, and porcine circovirus type 2. These viruses impair the immune system of pigs and either provide a chance for opportunistic pathogens like *E. rhusiopathiae* to infect pigs or reactivate latent infection.
This work was supported by Hi-Tech Research and Development Program of China (863 Program) (NO 2012AA101601), the National Basic Research Program of China (973 Program) (NO 2012CB518803) from the Ministry of Science and Technology of China and the Program of Introducing Talents of Discipline to Universities (NO B12005).
[Supplementary information](#sup1){ref-type="supplementary-material"} of this article can be found on the *Emerging Microbes and Infections*\'s website: <http://www.nature.com/emi>.
Supplementary Information {#sup1}
=========================
######
Click here for additional data file.
{#fig1}
###### Distribution of MICs for the 50 *E. rhusiopathiae* strains
Number of strains with MIC (μg/mL)
-------------- ------------------------------------ ---- ---- --- --- --- ---- ---- ---- ---- ---- ---- ---- ---- ----- -----------
Kanamycin 48 64 48 (100%)
Gentamicin 1 2 21 13 11 4 47 (98%)
Ampicillin 33 12 3 0.5 0 (0%)
Erythromycin 13 3 25 7 1 0 (10%)
Cefataxime 8 8 20 9 3 64 0 (100%)
Cefazolin 48 32 48 (100%)
Norfloxacin 1 10 15 13 8 1 16 37 (77%)
Levofloxacin 2 12 23 6 2 2 1 4 34 (71%)
Sulfadiazine 48 512 48 (100%)
Amikacin 48 64 48 (100%)
Tetracycline 3 3 3 3 7 5 9 10 5 16 29 (60%)
Doxycycline 2 5 5 1 1 3 13 6 12 16 18 (38%)
Lincomycin 2 4 5 1 1 2 20 9 3 1 2 37 (77%)
Resistant strains include intermediate resistant and resistant strains from CLIS.^[@bib13]^
[^1]: These authors contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
The PhD (doctor of philosophy) pathway is full of highs and lows, and should terminate with a valuable contribution to knowledge. During this pathway, one should demonstrate one's ability to conduct independently scholarly research, and the ability to solve research problems to reach the "Cinderella moment". To keep the research steps in the correct orbit, toolboxes are needed to navigate the PhD journey. Most biomedical students decide to go through the PhD journey either to improve their academic careers, seek the intellectual challenge or to satisfy their eager about a certain topic. The reason that spurred one to pursue the PhD will determine the shape and fate of your PhD thesis and contribution to knowledge. To better understand the PhD process and the pre-and PhD tools which are needed during the journey, the panacea toolboxes are discussed in this manuscript.
***Pre-PhD toolbox:*** Before one takes a step into the PhD world, one should make sure first it is really what one want. Second, you should have list of skills to make the way unclouded through the PhD journey. Communication skills such as academic writing, reading papers and how to evaluate a paper are very important skills that would give the biomedical student the ability to touch the strengths and weaknesses of each paper related to his topic, and how to find the gaps in the previous research and how to fill the gaps in his future research. Independent work and creativity are also needed. [Table-I](#T1){ref-type="table"} shows the top 10 skills that the biomedical student should have to make the way easier, as one have several tasks over a tight time through the PhD journey.
######
Top 10 skills of a pre-PhD student.
**No.** **Skill** **Description**
--------- ------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------
**1** **Creativity** Seven key questions (who, what, why, when, where, how and how much) cover the basics and help you to understand and figure out a situation.
**2** **Research methods** Qualitative, quantitative and cohort studies. Sampling techniques. All of those and many others should be familiar to the biomedical students.
**3** **Search medical databases** Skills of searching effectively the databases are needed to achieve highly sensitive and specific results.^1^
**4** **Reading a paper** How to read and interpret a paper in your field is highly-demanded.^2^
**5** **Writing a paper** Improve your writing skills as much as you can.^3^
**6** **Computer skills** Microsoft's word, PowerPoint and Excel skills are all desired during your PhD journey.
**7** **Presenting results** How to present your results to academics and non-academics is an art, which needs training and previous knowledge.^4^
**8** **Independent work** One should be used to organize the daily research without having to be alerted.
**9** **Documenting and reporting** Learn how to keep everything documented and reported would help you so much during the PhD degree.
**10** **Statistical knowledge** It is always preferred to have some knowledge about statistics and the corresponding softwares.
***PhD toolbox:***\"If we knew what it was we were doing, it would not be called research, would it?\" It is a transition time that you spend during the PhD period to go day by day to higher orbitals armed with more and more skills, which will enable you to "hit the nail on the head" in a specific area and topic of interest, and should terminate with a valuable contribution to knowledge. First of all and before you decide to go through the PhD journey, one should choose a supervisor who is interested and has done some work in that particular area and who is productive as well. To improve one's skills during the PhD period, start first with reading and evaluating the literature to enable you to identify the previous problems, how to analyze and solve them. The top 10 tips for improving the PhD skills and for passing successfully the journey are highlighted in [Table-II](#T2){ref-type="table"}.
######
Top 10 tips of a PhD student.
**No.** **Tip** **Description**
--------- ------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------
**1** **Time management** Plan you days and weeks and try to stick to your plans.
**2** **Reading and analyzing the literature** This will help you a lot to define the problems and to solve the problems.
**3** **Patience** You should have a high patience threshold.
**4** **Attending seminars and conferences** To extend your knowledge and creativity. Be always active and feel free to ask questions.
**5** **Practice writing** To increase and improve your ability to write your first paper and the thesis.
**6** **Talk about your project** Talk about your projects and the problems you face to your colleagues and your supervisor. This will always open new vistas.
**7** **Take courses** Courses such as statistics, communication skills, English, or any course you feel that you need it to improve your career.
**8** **Join international and professional societies** This will keep you in the loop in your topic and field.
**9** **Talk to the sales representatives** You will always learn new diagnostic things from them.
**10** **Take some days off and enjoy the formal holidays** It will make your brain always fresh to re-think and evaluate the projects' troubles.
| {
"pile_set_name": "PubMed Central"
} |
{.779}
{.780}
{.781}
{.782}
{.783}
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Trust in the doctor and the healthcare system is important for patient satisfaction, compliance and partnership towards successful aging and better disease management. [@b53-cia-0102-175] defined trust as "one's willingness to rely on another's actions in a situation involving the risk of opportunism." Recent work on trust has increasingly focused on conceptualizations regarding distrust ([@b44-cia-0102-175]; [@b4-cia-0102-175]; [@b45-cia-0102-175]). Distrust entails "the belief that a person's values or motives will lead one to approach all situations in an unacceptable way" ([@b44-cia-0102-175]). Distrust is not mistrust or no-trust, the contradictory notion of trust. Distrust is a qualified conditional trust in doctors and/or the healthcare delivery system on the part of the patient. The latter may be burdened by cost, beset by anxiety, having to cope with difficulties of navigating the managed care system, and confused by the complexities of modern medicine. In the midst of such a multifaceted healthcare delivery system, positive and legitimate distrust can co-exist with positive trust during patient--physician encounters. This area of positive distrust has received minimal attention in the medical literature ([@b27-cia-0102-175]; [@b12-cia-0102-175]; [@b37-cia-0102-175]), when compared with the numerous studies relating to patient--physician trust ([@b50-cia-0102-175]; [@b19-cia-0102-175]; [@b20-cia-0102-175]; [@b42-cia-0102-175]; [@b51-cia-0102-175]; [@b21-cia-0102-175]; [@b49-cia-0102-175]; [@b15-cia-0102-175]), that followed the sentinel work of [@b1-cia-0102-175] patient--physician trust scale. When it comes to the elderly, however, there appears to be a paucity of research on trust or distrust ([@b32-cia-0102-175]; [@b34-cia-0102-175]; [@b48-cia-0102-175]), despite the fact that the elderly account for over 30% in medical resource utilization as a group in the US. Moreover, with increasing longevity and the growing numbers of the elderly worldwide, the issue of patient--doctor trust and distrust (PDTD) in this group of patients clearly merits research. In this exploratory study we will focus on the concept of trust and distrust as perceived by a convenient sample of older patients with chronic diseases who had interacted with their doctors and healthcare delivery systems over a long period of time. We will then review the literature as it relates to the dynamics of trust--distrust in the day-to-day patient--doctor encounter and define a set of hypothesized predictors of PDTD. We hope these predictors will serve as a basis to develop a PDTD scale (PDTDS).
Importance of patient trust--distrust determinants
==================================================
It is important to understand the concept of PDTD. We would therefore like to expound on the trust--distrust concept based on various theories of trust and distrust and accordingly, derive hypothesized predictors of trust--distrust. Traditionally, patients have relied on trust in medical professionals to minimize the stress and uncertainty associated with their illness. If in addition, patients have to worry about their physician's control, given the increasing strictures of managed care and the perceptions related to the trustworthiness of the Health Maintenance Organization (HMO), it may become a major factor in how patients trust their physicians ([@b14-cia-0102-175]). In the last four to five years, state regulators have reported a 50% rise in complaints about HMOs by patients and physicians, particularly regarding healthcare service denials or delays, and most of these complaints reflect the public's increasing distrust of managed care rather than a true decline in quality healthcare ([@b28-cia-0102-175]). Obviously, increasing trust of patients in the entire healthcare delivery system, inclusive of managed care, is critical.
This trust--distrust bi-dimensional but mutually complementing perspective may provide a better and more insightful framework to understand the dynamics of patient--doctor trust-relations than those expressed in existing trust scales ([@b1-cia-0102-175]; [@b50-cia-0102-175]; [@b42-cia-0102-175]; [@b21-cia-0102-175]; [@b49-cia-0102-175]; [@b15-cia-0102-175]). Distrust is not mistrust, nor the opposite of trust, but a complimentary dimension that can enable doctors, nurses, managed care executives, and even governments who subsidize healthcare, to understand the specific and even positive role of distrust in patient--doctor trust. High levels of patient--doctor trust can coexist with high levels of patient--doctor distrust. Given the current complex US healthcare delivery system, patients are bound to experience high levels of trust and distrust with healthcare providers. Moreover, the perceived complexity, ambiguity, and vulnerability of healthcare delivery inputs, its processes and outcomes, and patient--physician encounters are bound to be a mix of high trust and distrust states that need to be carefully studied, predicted, and managed.
Measuring PDTD in older populations is important, especially, to better understand patient perceptions and design interventions to influence both doctor and patient behaviors. In chronic disease management, trust and distrust are important if patients are to adhere to care plans in partnership with their doctors.
Methodology
===========
As an initial and experimental approach to the understanding of patient trust--distrust in doctors, we analyzed the results from an earlier study of patient trust in doctors where distrust was only a component of a scale that measured patient trust and satisfaction with doctors. This scale was administered to a convenient sample of 515 patients with chronic diseases. The scale (see [Table 6](#t6-cia-0102-175){ref-type="table"}) was designed to assess four major trust factors: Trust 1 (cooperation and caring attributes by doctors); Trust 2 (quality and hospital reputation); Trust 3 (patient's confidence in doctors); and Trust 4 (patient's distrust and fear in the healthcare delivery system).
Based on these preliminary results we undertook to investigate in depth the trust--distrust literature both in the management and the medical sociology fields and accordingly, derive a set of hypothesized predictors which we believe can be used as the basis for developing a PDTDS.
Results
=======
Our preliminary study involved a mixed population of 200 breast cancer survivors, 174 hospitalized patients, and 141 ambulatory care patients. The demographics of the study population are presented in [Table 1](#t1-cia-0102-175){ref-type="table"}. We then compared the age--trust relationship and patient satisfaction ([Figure 1](#f1-cia-0102-175){ref-type="fig"}). As observed in [Figure 1](#f1-cia-0102-175){ref-type="fig"}, whereas the first three constructs of the trust instrument (Trust 1, Trust 2, and Trust 3) moved in tandem with patient satisfaction, the fourth component that measured trust--distrust (Trust 4) significantly departed from the other three trust components and the satisfaction construct. Additionally, when the patient data was classified into age-groups, elderly (aged 65 years and above) and younger (aged less than 65), there were significant differences (p=0.005) in trust and distrust levels between the elderly and younger patients ([Table 2](#t2-cia-0102-175){ref-type="table"}). To further investigate and analyze this phenomenon, we chose the first sample of 200 female breast-cancer patients. In this sample, we studied two major groups: 101 African-American and 69 Caucasian patients. These results are provided in [Table 3](#t3-cia-0102-175){ref-type="table"}. As observed in [Table 3](#t3-cia-0102-175){ref-type="table"}, while the first trust components are statistically equivalent across both groups, the fourth component of trust--distrust shows significantly (p=0.028) higher levels of distrust among African-American patients. These results caused us to study the conceptual and theoretical foundations of patient trust--distrust and their determinants.
Conceptual and theoretical foundations of patient trust--distrust in doctors
============================================================================
While there is widespread agreement on the importance of trust--distrust in human conduct, there also appears a bewildering diversity in defining the construct of trust ([@b17-cia-0102-175]). Trust researchers have developed different trust constructs in response to disparate sets of questions regarding social phenomena ([@b5-cia-0102-175]). There has been remarkably little effort, however, to integrate these different perspectives ([@b22-cia-0102-175]; [@b5-cia-0102-175]). The formidable variety in approaches to trust is largely a function of the diverse theoretical perspectives and research interests of scholars engaged in trust research ([@b22-cia-0102-175]). For instance, personality theorists view trust as an individual attribute or difference; sociologists and economists study trust as an institutional phenomenon or arrangement, and social psychologists conceptualize trust as behavior in a situational context: eg, an expectation of another party in a transaction ([@b44-cia-0102-175]; [@b22-cia-0102-175]). Whereas earlier trust literature in the management field contrasts trust with distrust as polar opposites, later developments reveal a complimentary approach to trust and distrust. Trust and distrust are separate dimensions that can coexist and mutually reinforce each other. It is therefore necessary to review these two streams of literature for a better understanding of the concepts of trust and distrust. [Table 4](#t4-cia-0102-175){ref-type="table"} summarizes the discussions to follow.
Trust and distrust as polar opposites
=====================================
Psychological view of trust and distrust
----------------------------------------
The earliest views on trust reflect a psychological approach. [@b29-cia-0102-175] defined trust as an individual's confidence in another person's intentions and motives, and the sincerity of that person's word. Following this approach, [@b36-cia-0102-175] argued that trusting individuals: (a) expect their interests to be protected and promoted by those they trust; (b) feel confident about disclosing negative personal information about themselves; (c) feel assured of full and frank information sharing; and (d) are prepared to overlook apparent breaches of trust relationship. [@b9-cia-0102-175] viewed trust as an individual's confidence in the intentions and capabilities of the trust partner and the belief that he or she would behave as hoped. [@b9-cia-0102-175] also viewed *distrust* as confidence about a relationship partner's undesirable behavior, stemming from the knowledge of his or her capabilities and intentions. Our first research hypothesis in this regard is:
*Hypothesis 1:* The higher the patients' sense that their interests are not being protected or promoted by doctors and nurses, the higher their distrust is with doctors and other healthcare providers.
Behavior theory of trust and distrust
-------------------------------------
Examining trust and distrust from a rational (predictive) choice perspective, behavior decision theorists define trust as co-operative conduct and distrust as nonco-operative conduct in a mixed-motive game situation, and see trust and distrust as polar opposites ([@b8-cia-0102-175]). Some earlier social psychological studies also considered trust and distrust as conflicting psychological states, and hence as unstable and transitory, and reckoned trust and distrust as opposing attributes ([@b24-cia-0102-175]). Normatively, therefore, trust was viewed as something good, and distrust as something bad or as a psychological disorder. Distrust was considered to reflect psychological imbalance and inconsistency, both adverse conditions that must be avoided ([@b9-cia-0102-175]). Our second research hypothesis in this regard is:
*Hypothesis 2:* The higher the patients' sense of nonco-operative conduct and conflicting interests on the part of doctors and nurses, the higher their distrust is with doctors and other healthcare providers.
Personality dispositional view of trust and distrust
----------------------------------------------------
For personality researchers who view trust as an individual difference, trust and distrust exist at opposite ends of a single trust--distrust continuum ([@b39-cia-0102-175]). They are mutually exclusive and opposite conditions. In general, low trust expectations are indicative of high distrust from this point of view ([@b46-cia-0102-175]; [@b47-cia-0102-175]). The central focus of these theories is how individuals develop their propensities to trust, and how these predilections affect their thoughts and actions regarding persons ([@b38-cia-0102-175], [@b39-cia-0102-175], [@b40-cia-0102-175]). According to these theories, factors exist within individuals that predispose them to trust or distrust others, especially when they do not know them. [@b38-cia-0102-175], [@b39-cia-0102-175], [@b40-cia-0102-175] argues that trust is a stable belief based on individuals' extrapolations from their early life-experiences. Trust develops during childhood as an infant seeks and receives help from its benevolent caregiver. Children of trusting parents trust others more easily than children of distrusting parents, and children with trusting siblings are better predisposed to trust. The more novel, complex, and unfamiliar the situations, the more influence such predispositions bear on trusting (or distrusting) behavior. According to [@b16-cia-0102-175]. people with trusting dispositions co-operate better, whereas people with distrusting predispositions tend to avoid co-operative activities, fearing exploitations. The latter, have fewer positive interaction experiences that beget trust; the former have more and progressively increase their trust. In this sense, trust begets trust, and distrust perpetuates distrusting predilections.
Cognition-based trust researchers, however, would argue that trust relies on rapid, cognitive cues or first impressions, as opposed to personal dispositional characteristics of trust ([@b24-cia-0102-175]; [@b30-cia-0102-175]). Especially, during first new patient--doctor encounters, parties may have to develop trust based on initial cognitive cues and first impressions. In such situations, individuals may have to rely either on one's predispositions to trust or on institution-based trust-development cues. Our third research hypothesis in this regard is:
*Hypothesis 3:* The higher the patients' predispositions to distrust the complex, unfamiliar, and costly healthcare system, the higher their distrust is with doctors and other healthcare providers.
Expectation theory of trust and distrust
----------------------------------------
Expectation theory defines trust as "a generalized expectancy held by an individual that the word, promise, oral or written statement of another individual or group can be relied upon" ([@b40-cia-0102-175]). Trust is "a set of expectations shared by all those involved in an exchange" ([@b54-cia-0102-175]). Trust is based on an individual's expectations that others will behave in ways that are helpful or at least not harmful ([@b10-cia-0102-175]). [@b54-cia-0102-175] definition of trust as a preconscious expectation suggests that vulnerability is only salient to trustors after a trustee has caused them harm. In reciprocal terms, distrust is understood as the expectation that others will not act in one's best interests, even engaging in potentially harmful behavior ([@b13-cia-0102-175]). Our fourth research hypothesis in this regard is:
*Hypothesis 4:* The higher are the patients' expectations regarding doctors, nurses, hospitals and managed healthcare, the higher their distrust is with doctors and other healthcare providers.
Criticism of trust and distrust as polar opposites
==================================================
Research within management literature has focused on trust primarily in terms of "rational prediction" ([@b24-cia-0102-175]) wherein agents conceive distrust as a highly risky situation that must be reduced or avoided by rational choices that predict distrust. Such "predictive" accounts of trust "appear to eliminate what they say they describe", thus disregarding or removing core elements of trust ([@b24-cia-0102-175]). Under this view, trust exists only in an uncertain and risky environment; that is, trust cannot exist in an environment of certainty ([@b3-cia-0102-175]).
The expectation-approach views trust as a disposition that would be most predictive in situations where individuals are relatively unfamiliar with one another. Trust, in this tradition, is viewed as a calculated decision to cooperate with specific others, based on information about others' personal qualities and social constraints: a context that very much reflects the patient--doctor trust encounter situation. Under this view, trust reflects an aspect of predictability, that is, it is an expectation; it cannot exist without some possibility of error. That is, trust can exist with some distrust. For instance, when patients say they trust a doctor, they do not necessarily make a statement whether the doctor is good or bad; but they reflect the notion of trust as a prediction of the doctor's behavior in a given context ([@b3-cia-0102-175]).
[Table 4A](#t4-cia-0102-175){ref-type="table"} summarizes various polar theories of trust and distrust. Key assumptions of these theories are: (a) trust and distrust are mutually exclusive and opposite unidimensional conditions; that is, trust and distrust are polar opposites; (b) trust is good and distrust is bad; (c) the social context of trust and distrust is either irrelevant or of low consequence ([@b23-cia-0102-175]). Most of these models are "undersocialized" and omit the role of concrete personal and social relationships and structures of such relations.
The major problem for these divergent views on trust and distrust is that scholars (a) have given limited attention to the *role of social context* in trust and distrust research, and (b) have considered trust--distrust as a one-dimensional construct. In the latter case, scholars have considered interpersonal relationships within organization or exchange situations as one-dimensional, with a single dimension or component of relationship to determine the quality of the entire relationship ([@b23-cia-0102-175]).
Trust and distrust as complimentary constructs
==============================================
Trust and distrust are reciprocal terms. Both trust and distrust are separate but linked dimensions. They are not polar opposites on a single continuum such that low trust means high distrust and high trust means low distrust. Trust and distrust both entail certain expectations, but whereas trust expectations anticipate beneficial conduct from others, distrust expectations anticipate injurious conduct ([@b23-cia-0102-175]). Both involve movements toward certainty: trust concerning expectations of things hoped for and distrust concerning expectations of things feared. Hence, both states can coexist ([@b35-cia-0102-175]); they are functional equivalents ([@b25-cia-0102-175]).
Organizational psychology theory of trust and distrust
------------------------------------------------------
Institution-based trust means that one believes the necessary impersonal structures are in place to enable one to act in anticipation of a successful future endeavor ([@b54-cia-0102-175]; [@b43-cia-0102-175]). [@b54-cia-0102-175] describes how certain specific institutional or social structures and arrangements generate trust. Institution-based distrust means that one believes the necessary impersonal structures are not in place. For instance, rational bureaucratic organizational forms could be trust-producing mechanisms for situations where the scale and scope of economic activity overwhelm interpersonal trust relations. Public auditing of firms, Securities and Exchange Commission (SEC) regulations, Federal Trade Commission (FTC) mandates and other government vigilance programs may increase customer trust in those companies. Institution-based trust researchers maintain that trust reflects the security one feels about a situation because of guarantees, safety nets, or other structures ([@b54-cia-0102-175]; [@b43-cia-0102-175]). Thus, the safe and structured atmosphere of a classroom may enable students to develop high levels of initial trust ([@b24-cia-0102-175]; [@b43-cia-0102-175]). Tough screening and high professional experience levels of new recruits may help senior employees to trust then implicitly.
Trusting intention at the beginning of a relationship may be high because of institution-based trust stimulators. Institution-based trust literature speaks of two such stimulators: situation normality and structural assurances. *Situation-normality*: defined as the belief that successful interaction is likely because the situation is normal ([@b11-cia-0102-175]) or customary ([@b2-cia-0102-175]), or that everything is in proper order ([@b24-cia-0102-175]). *Structural assurances*: defined as the socially learned belief that successful interaction is likely because of such structural safeguards or contextual conditions as promises, contracts, regulations, legal recourse, and guarantees are in place. The current healthcare crisis as a result of lack of insurance, high prices of prescription drugs in the US and fragmentation of care are instances of breakdown of situation normality and structural assurances such that high levels of trust and distrust could coexist. A fifth researchable hypothesis in this regard is:
*Hypothesis 5:* The higher the patients' sense of situation abnormality and lack of structural assurances in modern health delivery system, the higher their distrust is with doctors and other healthcare providers.
Sociological theory of trust and distrust
-----------------------------------------
Sociologists recognize the importance of trust and distrust as mechanisms for reducing social complexity and uncertainty, and, accordingly, view them as functional equivalents or substitutes. [@b25-cia-0102-175] argues that both trust and distrust function to allow rational actors to understand, contain, and manage social uncertainty and complexity, but they do so by different means. Trust reduces social complexity and uncertainty by disallowing undesirable conduct from consideration and replacing it with desirable conduct. Conversely, distrust functions to reduce social complexity and uncertainty by allowing undesirable conduct and by disallowing desirable conduct in considering alternatives in a given situation. In the latter case, distrust becomes a "positive expectation of injurious action" ([@b25-cia-0102-175]). Distrust simplifies the social world, allowing the individual to move rationally to take protective action based on these positive expectations of harm. Social structures appear most stable where there is a healthy dose of both trust and distrust to generate a productive tension of confidence ([@b23-cia-0102-175]). [@b25-cia-0102-175] even argues that "trust cannot exist apart from distrust, and trust cannot increase without increases in distrust. Increases in trust or distrust -- apart from increases in the other -- may do more harm than good." An over-trusted person can often exploit the over-trusting person. "Apart from a genuine openness to the possible necessity of distrust, benign and unconditional trust appears to be an extremely dangerous strategy for managing social relations" ([@b23-cia-0102-175]). Our sixth research hypothesis in this connection is:
*Hypothesis 6:* The higher the patients' sense of social complexity and uncertainty brought about by undesirable behaviors of doctors, managed healthcare and other healthcare providers, the higher their distrust is with doctors and other healthcare providers.
Social psychology theory of trust and distrust
----------------------------------------------
Human psychology functions in a social context. Hence, if the social context of an exchange situation or an organizational relationship is properly focused and fully brought into the social equation, then it is quite possible that an individual who trusts a partner on some attributes (eg, scientific knowledge, technical skill) may distrust that partner on other features (eg, social skills, ethical conduct, compassion skills), and both these states can coexist. According to social psychologists ([@b7-cia-0102-175]), positive-valent and negative-valent attitudes can coexist, and thus, trust which involves confident positive expectations and distrust which implies confident negative expectations regarding trusting partners, can operate simultaneously in the same individual, although from different viewpoints ([@b23-cia-0102-175]).
[@b52-cia-0102-175] noted that high positive affectivity (eg, active, strong, excited, enthusiastic, and elated) was not synonymous with low negative affectivity (eg, calm, relaxed, and placid). Similarly, low positive affectivity (eg, sleepy, dull, drowsy, and sluggish) was not synonymous with high negative affectivity (eg, distressed, scornful, hostile, fearful, nervous, and jittery). These and other studies ([@b6-cia-0102-175]) clearly indicate that positive-valent and negative-valent constructs are separable. The two constructs may systematically and negatively correlate, but their antecedents and consequences may be separate and distinct ([@b6-cia-0102-175]). The factors related to positive affect are distinct from those surrounding negative affect ([@b52-cia-0102-175]). These considerations indicate that the bases of trust and distrust may be different and separable. That is, trust is not the opposite of distrust; there may not be a singular trust--distrust continuum. High trust may be opposed to low trust; and high distrust may be antithetical to low distrust. The two states, even though ambivalent, could coexist. Our seventh research hypothesis may be stated as follows:
*Hypothesis 7:* The higher the patients' level of negative-valent attitudes regarding doctors, nurses, hospitals, and managed healthcare, the higher their distrust is with doctors and other healthcare providers.
Interdependence theory of trust and distrust
--------------------------------------------
Recent definitions of trust imply interdependent behavioral expectations. Thus, [@b17-cia-0102-175] defines trust as one party's optimistic expectations of the behavior of another, when the party must make a decision about how to act under conditions of vulnerability and dependence. According to [@b33-cia-0102-175] and [@b31-cia-0102-175], vulnerability is an important constituent of trust. That is, in the absence of risk or vulnerability trust is not necessary, since outcomes are not of consequence to trustors. Sabel's definition of trust assumes vulnerability: "trust is the mutual confidence that no party to an exchange will exploit the other's vulnerability" ([@b41-cia-0102-175]). According to [@b26-cia-0102-175], vulnerability accompanies trust. They define trust as "the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control the other party." [@b54-cia-0102-175] definition of trust as a preconscious expectation suggests that vulnerability is only salient to trustors after a trustee has caused them harm. Following this important trend, we will incorporate the domain of vulnerability in the trust--distrust scale, since so much of modern medicine in all its complexity, speed on innovation, and cost-conscious managed care involves vulnerability. [@b53-cia-0102-175] defines trust as "one's willingness to rely on another's actions in a situation involving the risk of opportunism." In contrast, distrust entails "the belief that a person's values or motives will lead one to approach all situations in an unacceptable way" ([@b44-cia-0102-175]).
In fact, trust-research "appears to be premised on the general idea that actors (ie, individuals, groups or organizations) become, in some ways, vulnerable to one another as they interact in social situations, relationships and systems" ([@b5-cia-0102-175]). As organizational arrangements become more complex (as in the current healthcare environment), actors' vulnerability to one another could become broader and deeper, and trust may be one of the best mechanisms actors have to cope with these new conditions ([@b5-cia-0102-175]). Often, patients are unfamiliar with physicians, surgeons, nurses, and hospitals. Gathered information in this regard may not be complete or totally reliable for establishing affective bonds with one another. Patient trust may be an effective surrogate in this regard. Our eighth related research hypothesis is:
*Hypothesis 8:* The higher the patient's sense of unfamiliarity and vulnerability with the complexity of modern health delivery system, the higher their dependence upon and distrust with doctors and other healthcare providers.
Complimentary theories of trust and distrust
============================================
[Table 4B](#t4-cia-0102-175){ref-type="table"} summarizes various complimentary theories of trust and distrust. They make some key assumptions: (a) Trust and distrust are mutually inclusive and complementary bi-dimensional conditions; that is, trust and distrust can coexist and reinforce each other; (b) Trust is good and positive and distrust is also good and positive, although based on different expectations; trust relates to beneficial expectations; distrust involves hazardous expectations; life experiences involves both, and often at the same time; (c) Trust--distrust is embedded in the complex, unfamiliar, and vulnerable social context of human relationships.
Discussion
==========
The importance of PDTD cannot be underestimated as it relates to compliance and patient satisfaction. There have been recent changes in the experiences of Medicare beneficiaries as a result of decline in the quality of interactions between patients and their doctors, a breakdown in continuity and integration of care and difficulties with access to care despite improvements in medical technology ([@b32-cia-0102-175]).
Expectations of care by the elderly include trust and the need for a sense of personal touch. Trust is complex in the older person given that they could be satisfied but not trust providers or they could trust providers but not be satisfied ([@b18-cia-0102-175]). A recent study on how patients' trust relates to their involvement in medical care ([@b48-cia-0102-175]) identifies age as an important predictor with older patients being more compliant, deferential, passive, and trusting of their doctors as compared with younger patients. Our preliminary studies and those of other research workers appears to support that the perceptions of PDTD in the elderly are different from the rest of the patient population. It is therefore necessary to have an ability to measure PDTD as a basis for developing interventions that can positively affect both patient and doctor behaviors during the clinical encounter. We are proposing a set of eight hypothesized predictors, based on the trust distrust theories that could serve as a basis for developing a PDTD scale.
[Table 5](#t5-cia-0102-175){ref-type="table"} synthesizes patient--physician interpersonal relations as a function of Low versus High, Trust and Distrust. Each quadrant suggests clear implications to physicians, doctors and other healthcare givers, as well as to patients. It is a challenge for all healthcare givers to generate in their patients lower levels of fear, skepticism, and cynicism such that costs of patient monitoring and fragmentation of care is significantly reduced. Analogously, healthcare providers must do everything within their power and skills to generate in their patients high levels of hope, faith, confidence, assurance, and also welcome high patient cooperation.
Based on the trust--distrust literature reviewed earlier and the various factors of trust--distrust hypothesized, we present a tentative patient's trust--distrust measurement instrument in [Table 6](#t6-cia-0102-175){ref-type="table"}. Accordingly, [Table 7](#t7-cia-0102-175){ref-type="table"} indicates which theory reflects which scale statement. Following [Table 4](#t4-cia-0102-175){ref-type="table"}, [Table 8](#t8-cia-0102-175){ref-type="table"} projects which statement is best positioned to fall into one of the four quadrants. Both [Table 7](#t7-cia-0102-175){ref-type="table"} and [Table 8](#t8-cia-0102-175){ref-type="table"} ensure nomological (conceptual--theoretical) validity of the trust--distrust scale. Finally, [Table 9](#t9-cia-0102-175){ref-type="table"} sketches costs versus benefits of various patient--physician trust--distrust encounters. The bottom line in healthcare is to have a profit margin so that ongoing research education and development of innovative modes of healthcare is possible.
Concluding remarks
==================
Distrust of doctors and the healthcare system may be a significant barrier to seeking proper medical care, enforcing effective preventive care and following treatment regimens. Hence, conceiving, formulating, and implementing various strategies to reduce patient distrust and mistrust are an important component of delivering modern healthcare.
**Disclosure**
To the best of our knowledge, there exists no known conflict of interest among the authors.
Figure and Tables
=================
{#f1-cia-0102-175}
######
Sample sociodemographics
**Variables** **Study 1** **Study 2** **Study 3** **Entire study**
---------------------- --------------- --------------- --------------- ------------------
Sample size 200 (100.0 %) 141 (100.0 %) 174 (100.0 %) 515 (100.0 %)
Gender
Female 200 (100.0 %) 108 (76.6 %) 78 (44.8 %) 386 (75.0 %)
Male 0 (0.0 %) 33 (23.4 %) 96 (55.2 %) 129 (25.0 %)
Ethnicity
African American 101 (50.5 %) 119 (84.4 %) 140 (80.5 %) 360 (69.9 %)
Caucasian 69 (34.5 %) 15 (10.6 %) 22 (12.6 %) 69 (20.6 %)
Others 30 (15.0 %) 7 (5.0 %) 12 (6.9 %) 49 (9.5 %)
Marital status
Married 105 (52.5 %) 40 (28.4 %) 32 (18.4 %) 177 (34.4 %)
Separated/divorced 27 (13.5 %) 32 (22.7 %) 39 (22.4 %) 98 (19.0 %)
Widowed 50 (25.0 %) 38 (26.9 %) 18 (10.3 %) 106 (20.6 %)
Never married 18 (9.0 %) 31 (22.0 %) 85 (48.9 %) 134 (26.0 %)
Age (years) 58.5 ± 11.7 59.3 ± 18.2 50.5 ± 13.9 56.1 ± 14.9
Education (years) 13.0 ± 2.7 13.0 ± 3.0 11.3 ± 3.2 12.4 ± 3.1
Highest degree
None 28 (30.4 %) 26 (18.4 %) 69 (39.7 %) 128 (23.9 %)
High School 36 (39.1 %) 76 (53.9 %) 96 (55.2 %) 208 (40.4 %)
Associate's 13 (14.1 %) 13 (9.2 %) 2 (1.1 %) 28 (5.4 %)
Bachelor's 13 (14.1 %) 18 (12.8 %) 5 (2.9 %) 36 (7.0 %)
Master's/Doctoral 2 (2.2 %) 8 (5.7 %) 2 (1.1 %) 2 (2.3 %)
Occupation
Employed 131 (65.5 %) 92 (65.2 %) 72 (41.4 %) 295 (57.3 %)
Unemployed 69 (34.5 %) 49 (34.8 %) 102 (58.6 %) 220 (42.7 %)
Income (In \$US)
≤ \$20 000 36 (40.0 %) 54 (42.5 %) 137 (80.6 %) 227 (44.1 %)
\$20 001--60 000 30 (33.3 %) 61 (48.0 %) 27 (15.9 %) 118 (22.9 %)
\$60 001--100 000 16 (17.8 %) 8 (6.3 %) 4 (2.3 %) 28 (5.4 %)
≥\$100 000 8 (8.9 %) 4 (3.2 %) 2 (1.2 %) 14 (2.7 %)
Health insurance
Insured 199 (99.5 %) 141 (100.0 %) 126 (72.4 %) 466 (90.5 %)
Uninsured 1 (0.5 %) 0 (0.0 %) 48 (27.6 %) 49 (9.5 %)
**Note:** Values are mean ± standard deviation or n and %. Percentages are derived from column totals and adjusted for missing data.
######
Comparison between trust and satisfaction among elderly (age ≥65 years) and younger (age \<65 years) groups
**Variables** **Younger group (Age \<65 years) \[n = 364\]** **Elderly group (Age ≥65 years) \[n = 141\]** **t** **p**
---------------------------------------------- ------------------------------------------------ ----------------------------------------------- -------- -------
Trust 1 (cooperation, caring, vulnerability) 4.62 ± 0.54 4.68 ± 0.53 −1.149 0.251
Trust 2 (quality & hospital reputation) 4.70 ± 0.55 4.81 ± 0.43 −1.976 0.049
Trust 3 (confidence) 4.48 ± 0.69 4.47 ± 0.70 0.159 0.874
Trust 4 (distrust & fear) 4.28 ± 0.90 4.03 ± 0.92 2.808 0.005
Total trust 4.52 ± 0.54 4.49 ± 0.47 0.523 0.601
Satisfaction 4.62 ± 0.64 4.69 ± 0.50 −1.199 0.231
**Note:** Values are mean ± standard deviation.
######
Characteristics of the female breast cancer patients (Study 1: n=170)
**Variable** **African-American (n=101)** **Caucasian (n=69)** **p value**
------------------------------------------------- ------------------------------ ---------------------- -------------
Nr of patients 101 (59.4) 69 (40.6) .......
Age (years) 60.0 ± 10.5 54.3 ± 10.7 0.001
Education (years) 12.1 ± 2.4 14.5 ± 2.6 0.001
Marital status 0.022
Never married 9 (9.7) 4 (8.7)
Married 36 (44.4) 54 (71.7)
Separated/divorced 21 (19.4) 4 (4.3)
Widowed 35 (26.4) 7 (15.2)
No. of children 3.2 ± 1.9 2.1 ± 1.4 0.001
Occupation 0.634
Unemployed 50 (92.0) 10 (100.0)
Employed 4 (8.0) 17 (0.0)
Annual household income (In \$US) 0.001
Up to \$20 000 33 (29.2) 2 (0.0)
\$20 001--40 000 17 (66.7) 3 (66.7)
\$40 001--60 000 2 (0.0) 4 (33.3)
\$60 001--80 000 0 (4.2) 6 (0.0)
\$80 001--100 000 0 (4.2) 5 (4.2)
\$100 000+ 1 (4.2) 6 (4.2)
Year of breast cancer diagnosis 0.298
1970--1993 18 (17.8) 6 (8.7)
1994--2003 83 (82.2) 63 (91.3)
Type of breast cancer treatment 0.184
None 13 (16.9) 1 (2.2)
Chemotherapy 10 (11.3) 6 (11.1)
Radiation therapy 3 (4.2) 3 (6.7)
Tamoxifen (Nolvadex) 8 (9.9) 5 (11.1)
Combination/other 67 (57.7) 54 (68.9)
Type of breast cancer surgery 0.018
None 4 (5.6) 1 (2.2)
Lumpectomy 43 (50.0) 31 (43.5)
Mastectomy 46 (36.1) 20 (23.9)
Mastectomy & breast reconstruction 5 (4.2) 14 (23.9)
Combination/other 3 (4.2) 3 (6.5)
Measures of trust & satisfaction
Trust1 -- cooperation, caring & vulnerability 4.7 ± 0.5 4.6 ± 0.4 0.328
Trust2 -- quality & hospital reputation 4.6 ± 0.6 4.6 ± 0.6 0.976
Trust3 -- confidence 4.8 ± 0.4 4.9 ± 0.3 0.181
Trust4 -- distrust & fear 4.2 ± 1.0 4.5 ± 0.7 0.028
Total trust 4.6 ± 0.5 4.6 ± 0.4 0.400
Satisfaction 4.8 ± 0.5 4.8 ± 0.4 0.452
######
A synthesis of theories and definitions of trust and distrust
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Theory (Authors)** **Approach** **Definition of trust** **Definition of distrust** **Implications for patient--physician encounters**
---------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Table 4A: Trust and distrust as polar opposites**
**Psychology**\ Trust as an individual trait Trust is one's confidence in another's positive intentions and promises. Distrust is one's confidence about one's undesirable behavior. Foster trusting and avoid distrusting confidence of patients.
([@b29-cia-0102-175]; [@b9-cia-0102-175]; [@b36-cia-0102-175])
**Behavioral**\ Trust as a rational predictive choice of a partner. Devoid of real social context, trust is a function of incentives. Trust is cooperative conduct in a conflicting interpersonal encounter. Distrust is a non-cooperative conduct in a mixed-motive game situation. Distrust is a psychological disorder. Normatively, trust is good, distrust is bad. Nurture trust to solve intractable conflict situations and to promote effective collaboration.
([@b9-cia-0102-175]; [@b24-cia-0102-175])
**Personality disposition**\ Trust is a personal pre-dispositional attribute Trusting pre-dispositions indicate low expectations and cooperate better. Distrusting predispositions indicate high expectations and cooperate less with the trusted. Distrust is a psychological disorder that needs to be corrected. Trust--distrust transcends the social context.
([@b38-cia-0102-175], [@b39-cia-0102-175]; [@b46-cia-0102-175]; [@b47-cia-0102-175])
**Expectation**\ Trust as a generalized expectancy Trust is a set of expectations that the trusted will behave in a helpful manner as expected by the trustor. Distrust is a set of expectations that the trusted will not behave helpful as expected by the trustor. Assure patients that you will act always in their interests, thus converting distrust to trust.
([@b40-cia-0102-175]; [@b54-cia-0102-175]; [@b10-cia-0102-175])
**Table 4B: Trust and distrust as complimentary constructs**
**Organizational psychology**\ Trust as an organizational phenomenon supported by institutional mechanisms. Trust as believing in the institutional systems (normal situations and structural assurances) that support trust. Distrust as believing in the institutional systems (abnormal situations and structural non-assurances) that support distrust. Complexity, undesirability and vulnerability of modern healthcare outcomes can weaken situation normality and structural assurances that, in turn, could result in high distrust levels.
([@b11-cia-0102-175]; [@b2-cia-0102-175]; [@b24-cia-0102-175]; [@b54-cia-0102-175]; [@b43-cia-0102-175])
**Sociology**\ Trust--distrust as a mechanism for reducing social complexity and uncertainty. Trust and distrust coexist as functional equivalents or substitutes for reducing social complexity. Trust is a positive expectation of beneficial action; distrust is a positive expectation of injurious action. Do not over-trust. Total, unconditional trust could be dangerous for managing social relations.
([@b25-cia-0102-175]; [@b23-cia-0102-175])
**Social psychology**\ Trust--distrust as a continuum of a psychological state that is unstable and transitory. Trust as positive-valent and distrust as negative-valent attitudes can coexist. Trust involves confident positive expectations and distrust involves confident negative expectations regarding trusting partners. Trust is a necessary ingredient for social order; hence, focus on nurturing trust. Be sensitive to sources of patient distrust and manage them carefully.
([@b7-cia-0102-175]; [@b23-cia-0102-175])
**Interdependence**\ Trust--distrust as interdependent behavioral expectations amidst complexity and vulnerability. Trust is a function of one's dependence upon and vulnerability regarding the other party. Distrust is also a function of one's dependence upon and vulnerability regarding the other party. Trust--distrust investment should not be too high, or too low, but geared to meet all situations within the complexities and risks of modern healthcare systems.
([@b44-cia-0102-175]; [@b26-cia-0102-175]; [@b53-cia-0102-175])
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
######
Patient--physician interpersonal relations as a function of low and high, trust and distrust
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Patient trust:** **Patient distrust**
-------------------- -------------------------------------------------------------------------------- --------------------------------------------------------------------------------------------------------
**Low:**\ **High:**\
Low fear\ High fear\
Low skepticism\ High skepticism\
Low cynicism\ High cynicism\
Low monitoring\ High monitoring\
Low vigilance High vigilance
**High:**\ **Patient--physician:**\ **Patient--physician:**\
High hope,\ High value congruence;\ Sustained trust and distrust; trust constantly verified;\
High faith,\ common objectives;\ Strong reason to be confident in certain areas and diffident in others;\
High confidence\ frequent interactions;\ Relationships are multiplex, multifaceted, highly segmented and bounded; like in strategic alliances;\
High assurance\ Pooled positive and trust-reinforcing experiences; few defense mechanisms;\ Significant amounts of information shared under strict confidentiality;\
High initiatives Conversations are rich, deep, personal and occasionally complex;\ Collaboration opportunities pursued but risks assessed;\
Hence, reason to be mutually confident;\ Vulnerabilities continuously monitored and protected.
No reason for suspicion;\
High willed pooled interdependence and cooperation;\
All opportunities for sharing information pursued;\
New trust-building initiatives sought.
**Low:**\ **Patient--physician:**\ **Patient--physician:**\
Low hope,\ Casual acquaintance;\ Undesirable eventualities expected and feared;\
Low faith,\ Careful, bounded, arms-length discrete transactions;\ Conversations are cautious, guarded, and often laced with cynicism\
Low confidence\ No pooled trust-reinforcing experiences;\ Pooled negative distrust-reinforcing experiences; bureaucratic checks;\
Low assurance\ Conversations simple and casual;\ No reason for mutual confidence;\
Few initiatives\ No closeness or intimacy;\ Strong reason for watchfulness;\
Low resistance No threats to confidentiality as little information of consequence is shared;\ Significant resources for monitoring;\
No reason to fear or be confident;\ Harmful or exploitative motives not ruled out;\
Limited interdependence and cooperation;\ Interdependence difficult over time or at best, carefully managed;\
Just professional courtesy. Offensive self-defense.
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Note:** Adapted from [@b23-cia-0102-175].
######
Patient--physician trust--distrust scale statements
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
\[\*Bracketed numbers indicate the most likely quadrant the statement fits under [Table 5](#t5-cia-0102-175){ref-type="table"}\].
I have very strong positive hopes about modern medicine and what doctors can do for me. \[1\]\*I have tremendous faith in physicians, doctors, nurses and other healthcare providers. \[1\]I have every reason to suspect the profitability motives of the health insurance companies (eg, HMO, HAP). \[4\]My unfamiliarity with our complex healthcare delivery system makes me very distrustful of what my doctors can do for me. \[2\]The cost-controlling devices of managed healthcare (eg, HMOs) make me very skeptical about the treatment-efficiency of my doctors. \[2\]At times, I am very cynical about the morality of our healthcare delivery system. \[2\]The government should very carefully monitor our entire healthcare system. \[2\]The current complex healthcare system makes me doubt the competence of my doctors, nurses and other caregivers. \[2\]I deeply distrust managed care (eg, BCBS, HMOs, HAP) that controls my doctor's care for me. \[4\]I am losing faith in our health delivery system that is controlled by health insurance companies. \[3\]Our information privacy laws make me feel very uneasy when disclosing vital information about myself to my doctors. \[4\]I am very scared that when I get sick I will be totally dependent upon doctors, nurses and hospitals. \[4\]I am afraid my interests and health objectives do not meet those of my primary physician. \[3\]I am very afraid that my interests and health objectives do not meet those of my specialty doctors. \[4\]Given our profit-oriented managed healthcare systems (eg, HMOs), I have every reason for suspecting the quality of care my doctors can deliver to me. \[2\]My conversations with my doctors are rich, deep, personal and very straightforward. \[1\]Because of my fears and anxiety about my disease, I am not fully ready to cooperate with my doctor. \[2\]I have tremendous confidence in my doctor's technical and professional skills in handling my case. \[1\]I have full faith in my doctor, in his/her abilities, skills and decisions. \[1\]I am very confident about my doctor's sincerity in treating my sickness. \[1\]I do not give my best cooperation in listening and following my doctor's advice. \[3\]I am very obedient to whatever my doctor will ask me to do regarding my health problem. \[1\]I am afraid my doctor will exploit my vulnerability concerning my ill-health and not really care. \[4\]I am very hesitant about disclosing negative information about myself to my doctor. \[4\]I am distrustful of my doctor's interests and intentions regarding my treatment. \[3\]At times, I am scared about my doctor as to what he/she will say, decide and do about my disease. \[3\]I deeply distrust doctors, nurses, and hospitals, in general. \[4\]The hospital administration is very careful in its choice of nurses and other support staff. \[1\]I feel very comfortable in the hospital because of the very cooperative and understanding staff. \[1\]My hospital has the best reputation for medical excellence. \[1\]Because modern medicine is so sophisticated, I totally depend upon my doctor's knowledge and skills. \[2\]Thanks to regular government quality control, I am very trustful of my doctor and his/her treatment. \[2\]Despite my unfamiliarity with doctors, nurses and hospitals, I feel very confident about my treatment. \[1\]I trust my doctor to put my medical needs above all other considerations when treating my medical problem. \[3\]. I love my doctors and nurses so much for the sensitivity with which they communicate with my family. \[1\]My doctor is a very caring person and I feel very happy about it. \[1\]I feel that my conversations with my doctors are very careful, bounded, guarded and discrete. \[3\]I have no reason to hope for high levels of mutual confidence as far my doctors are concerned. \[4\]I am very watchful and vigilant as to what doctor will say, diagnose, and treat about case. \[4\]I feel no closeness to my doctors and feel forced to deal with them with just professional courtesy. \[3\]I pursue all opportunities for sharing all my health information with my doctors with utmost openness. \[1\]I feel my doctor cannot do much for me because of the severity of my illness. \[4\]I feel the hospital can do only so much for me owing to my health insurance carriers. \[4\]I feel I cannot rule out harmful motives of my doctors as far as my health is concerned. \[4\]It is too risky for me to totally collaborate with my doctor during my office visits. \[3\]It is part of my personality that I deeply distrust doctors, nurses and hospitals in general. \[4\]I naturally distrust my doctors once I know that they do not care. \[2\]I am a very trusting person when it comes to healthcare, doctors and nurses. \[1\]Given the complexity of modern healthcare, I cannot but trust doctors and nurses. \[2\]I am afraid to trust my doctors, nurses and hospitals when I encounter a serious disease. \[3\]There is great sense of bonding with my doctor because of his/her gentleness and compassion. \[1\]I am very satisfied with my treatment because of the excellent teamwork skills of my doctors. \[1\]I am very happy and content with my doctor. \[1\]I would not change my doctor for anybody in the world. \[1\]I would gladly recommend my doctor to anybody. \[1\]I am very satisfied with my doctor. \[1\]I love my doctors and nurses because they treat me as a person with dignity, feelings and beliefs. \[1\]I am very satisfied with the entire staff of doctors, nurses, and support people. \[1\]I am very satisfied with my nurse. \[1\]
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Abbreviations:** BCBS, Blue Cross, Blue Shield; HAP, Health Alliance Plan; HMO, Health Maintenance Organization.
######
Distribution of trust--distrust scale statement by theories of trust--distrust
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
**Theory (Authors)** **Approach** **Definition of trust** **Definition of distrust** **Trust--distrust scale items**
---------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------------------- -----------------------------------
**Table 7A Trust and distrust as polar opposites**
**Psychology**\ Trust as an individual trait Trust is one's confidence in another's positive intentions and promises. Distrust is one's confidence about one's undesirable behavior. 1, 18, 19, 20, 34, 49
([@b29-cia-0102-175]; [@b9-cia-0102-175]; [@b36-cia-0102-175])
**Behavioral**\ Trust as a rational predictive choice of a partner. Devoid of real social context, trust is a function of incentives. Trust is cooperative conduct in a conflicting interpersonal encounter. Distrust is a non-cooperative conduct in a mixed-motive game situation.\ 5, 16, 17, 21, 24, 29, 35, 36, 41
([@b9-cia-0102-175]; [@b24-cia-0102-175]) Distrust is psychological disorder.
**Personality Disposition**\ Trust is a personal pre-dispositional attribute Trusting pre-dispositions indicate low expectations and cooperate better. Distrusting predispositions indicate high expectations and cooperate less with the trusted. 2, 6, 22, 25, 27, 46, 47, 48, 50
([@b38-cia-0102-175], [@b39-cia-0102-175]; [@b46-cia-0102-175]; [@b47-cia-0102-175])
**Expectation**\ Trust as a generalized expectancy. Trust is a set of expectations that the trusted will behave helpful as expected by the trustor. Distrust is a set of expectations that the trusted will not behave helpful as expected by the trustor. 3, 4, 14, 15, 23, 44,
([@b40-cia-0102-175]; [@b54-cia-0102-175]; [@b10-cia-0102-175])
**Table 7B: Trust and distrust as complimentary constructs**
**Organizational psychology**\ Trust as an organizational phenomenon supported by institutional mechanisms. Trust as believing in the institutional systems that support trust. Distrust as believing in the institutional systems that support distrust. 3, 5, 9, 11, 15, 28, 30, 43
([@b11-cia-0102-175]; [@b24-cia-0102-175]; [@b2-cia-0102-175]; [@b54-cia-0102-175]; [@b43-cia-0102-175])
**Sociology**\ Trust--distrust as a mechanism for reducing social complexity and uncertainty. Trust and distrust coexist as functional equivalents or substitutes for reducing social complexity. Trust is a positive expectation of beneficial action; distrust is a positive expectation of injurious action. 7, 11, 26, 38, 42, 43
([@b25-cia-0102-175]; [@b23-cia-0102-175])
**Social psychology**\ Trust--distrust as a continuum of psychological state that is unstable and transitory. Trust as positive-valent and distrust as negative-valent attitudes can coexist. Trust involves confident positive expectations and distrust involves confident negative expectations regarding trusting partners. 8, 16, 37, 39, 40, 44, 45,
([@b7-cia-0102-175]; [@b23-cia-0102-175])
**Interdependence**\ Trust--distrust as interdependent behavioral expectations amidst complexity and vulnerability. Trust is a function of one's dependence upon and vulnerability regarding the other party. Distrust is also a function of one's dependence upon and vulnerability regarding the other party. 9, 10, 12, 31, 32, 49
([@b26-cia-0102-175]; [@b44-cia-0102-175]; [@b53-cia-0102-175])
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
######
Distribution of scale statement in the trust--distrust quadrants
**Patient distrust**
--------------------------- ---------------------------------------------------------------------- ------------------------------------------------------------------- --------
**Low** **High**
**High** **Quadrant 1** **Quadrant 2**
1,2, 16, 18, 19, 20, 22, 28, 29, 30, 33, 35, 36, 41, 48 \[15 items\] 4, 5, 6, 7, 8, 15, 17, 31, 32, 47, 49 \[11 items\] 26
**Low** **Quadrant 3** **Quadrant 4**
10, 13, 21, 25, 26, 34, 37, 40, 45, 50 \[10 items\] 3, 9, 11, 12, 14, 23, 24, 27, 38, 39, 42, 43, 44, 46 \[14 items\] 24
**Total number of items** **25** **25** **50**
######
Profile of patient--physician trust levels: costs versus benefits
--------------------------------------------------------------------------------------------------------------------------------------------------
**Physician's trust level** **Patient's trust level**
----------------------------- ------------------------------------ ---------------------------------------- --------------------------------------
**Low** **High**
**Low** **Costs** Both patient and physician:\ High agency costs for the patient:\
low mutual cooperation,\ high trust investment costs;\
low mutual honesty,\ high affect and emotion costs;\
low mutual benevolence high health-loss probability;\
very few options;\
low monitoring ability.\
For the doctor: no significant costs
**Benefits** Both patient and physician:\ Almost none to patients;\
low involvement;\ Significant benefits to doctors.
low interdependence;\
low investments, and low benefits.
**Risks** Physician-opportunism\ Patient abuse;\
Low physician commitment Patient exploitation,\
Patient dissatisfaction;\
Patient may switch & not return.
**High** **Costs** High agency costs for the doctor:\ Both for patient and physician\
high trust investment costs;\ low agency costs such as:\
high affect and emotion costs;\ bonding costs\
high loss probability;\ monitoring costs\
very few options;\ warranty-guarantee costs;\
low monitoring ability.\ search costs
For the patient: no significant costs
**Benefits** Almost none to doctors;\ Both for patient and physician:\
Significant benefits to patients. high commitment;\
high mutual cooperation,\
healthy interdependence;\
high mutual honesty,\
high mutual benevolence\
high satisfaction.
**Risks** Doctor abuse;\ Sustaining high mutual trust;\
Doctor exploitation,\ High dependence;\
Doctor dissatisfaction;\ Stifled creativity due to over-trust;\
Doctor may refuse treatment.\ Few other options due to over-trust.
Patient's opportunism.\
Patient's betrayal.
--------------------------------------------------------------------------------------------------------------------------------------------------
| {
"pile_set_name": "PubMed Central"
} |
{.74}
{.75}
{.76}
{.77}
{.78}
{.79}
{.80}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
The magnitude of linkage disequilibrium (LD) can be used to estimate effective population size [@pone.0069078-Sved1]--[@pone.0069078-Waples1]. In general, low populations sizes are expected to give rise to relatively high levels of LD, and similarly high population sizes to low LD levels. An important feature of this means of estimation is that measurement at a single point in time can provide information on effective size. Furthermore closely-linked loci give information on population sizes over historical periods of time, while loosely-linked loci estimate population sizes in the immediate past [@pone.0069078-Hill1], [@pone.0069078-Hayes1].
Much recent attention has been paid to the use of unlinked loci for estimating population size, for which the term 'Linkage Disequilibrium' will inappropriately be used. There are three major advantages of studying unlinked loci. First, the majority of pairs of loci are unlinked. Secondly, these are the only locus pairs for which it is easy to estimate the recombination frequency, 50%. Finally, in the study of pest populations, and in the area of conservation, it is usually the most recent population sizes that are of interest, for which unlinked loci are the most relevant.
The principal problem in studying unlinked loci comes from the sample sizes needed to obtain accurate LD estimates. The expected disequilibrium is a function of , where is the effective population size, assumed constant, and , where is the sample size [@pone.0069078-Weir1]. Unless sample sizes are large, the latter can overwhelm the former.
A second complication comes from the usual necessity to use diploid data. Most LD theory is based on haplotypes rather than diploid genotypes, which typically cannot be observed. Although the recognition of haplotypes may seem inappropriate for unlinked loci, the same distinction applies as for linked loci, because the information on population size comes from genes with the same parental origin rather than genes inherited from different parents. The passage from zygotic to to gametic parameters can be made using either the maximum likelihood estimator of Hill [@pone.0069078-Hill2], or, as will be used here, the Burrows estimator as elaborated by Weir [@pone.0069078-Weir2].
In preliminary investigations of the size of Queensland fruit fly populations, we found very low estimates for populations that are believed to be large. We traced this discrepancy to an excess of homozygous genotypes, believed to be due to the presence of null alleles at some of the microsatellite loci used in the study.
Because of these complications, the problem of finding an adequate estimator of is fraught with potential biases. Waples and Do [@pone.0069078-Waples2] have, however, shown that their LDNe program works well in estimating from simulated data. The program uses empirically derived correction factors rather than investigating the underlying reasons for the biases. The purpose of the present paper is to produce an analytical solution to account for the biases. We derive two sets of formulae that do this, depending on whether a 'within-locus disequilibrium factor' is used or not, and compare the application of these two sets to simulated and real data.
Materials and Methods {#s2}
=====================
Queensland Fruit Fly Samples {#s2a}
----------------------------
Two data sets are analysed in the paper.
1. East coast Australian populations. The data are from 55 samples from towns in the state of NSW in the years 2002--2004 [@pone.0069078-Gilchrist1]. Some of these sample come from areas where the flies are endemic, and in other cases the outbreaks appear to be only temporary.
2. NorthWest. These flies were collected during the years 1999--2003 from Northern West Australia and the Northern Territory [@pone.0069078-Cameron1].
The data in the two cited papers have previously been summarised only in terms of single locus statistics. The present paper involves a two-locus analysis, which requires additional information from the original data sets. The original data sets are provided in Supporting Information, [Data S1](#pone.0069078.s005){ref-type="supplementary-material"} and [Data S2](#pone.0069078.s006){ref-type="supplementary-material"}.
Computer Simulation {#s2b}
-------------------
All simulations reported in the paper are forward Monte-Carlo simulations under the Wright-Fisher model. Parents were chosen randomly in each case, thereby allowing selfing and not assuming permanent mate bonding, an important aspect of population structure [@pone.0069078-Weir1]. Most simulations involved a starting population with either 16 or 32 loci, each locus having the number of alleles chosen randomly between 2 and 8. Alleles were assigned randomly at different loci, assuming no systematic LD. Populations were simulated for 20 generations, followed by sampling without replacement of 32 individuals from the final population, and calculation of LD levels. Simulations were written in C, and are available on request.
Theory {#s2c}
------
Most of LD theory applies to gametes rather than genotypes. Fortunately a simple method, the 'Burrows composite LD coefficient', is available for handling genotypes. This coefficient has been defined by Cockerham and Weir [@pone.0069078-Cockerham1] in terms of sums of genotype frequencies. It is convenient to introduce here a slightly different but simpler way of relating genotype frequencies to gamete frequencies. See [Table 1](#pone-0069078-t001){ref-type="table"} for a listing of symbols used.
10.1371/journal.pone.0069078.t001
###### Symbols used in the text.
{#pone-0069078-t001-1}
------------- -----------------------------------------------------------------------------------------
N~e~ Effective population size
S Number of diploid individuals in a sample
n~11~ Number of genotypes in a sample with aa at first locus and bb at second locus
n~12~ Number of aa b-- genotypes where -- refers to non-b allele at the second locus
n~21~ Number of a-- bb genotypes
n~22~ Number of a-- b-- genotypes
n~a~, n~b~ Number of a and b alleles respectively
p~a~, p~b~ Allele frequencies in gametic and composite table, = n~a~/2S and n~b~/2S
p~ab~ Frequency of the ab haplotype
D Gametic disequilibrium coefficient = p~ab~ -- p~a~p~b~
r^2^ Gametic correlation = D^2^/\[p~a~(1-- p~a~)p~b~(1-- p~b~)\]
M Number of ab haplotypes in composite haplotype table = 4n~11~+2n~12~+2n~21~+ n~22~
p~ab~(comp) Frequency of ab in composite haplotype table = M/4S
D(comp) Disequilibrium coefficient from composite haplotype table = p~ab~(comp) -- p~a~p~b~
Δ Burrows' disequilibrium coefficient = 2D(comp)
r^2^(comp) r^2^ value from composite haplotype table = D^2^(comp)/\[p~a~(1-- p~a~)p~b~(1-- p~b~)\]
Composite r^2^ parameter = 4r^2^(comp)
Estimate of from sample
with single-locus disequilibrium =
?^2^(comp) ?^2^ calculated from composite haplotype table
p~n~ Frequency of null alleles at a locus
α Half the difference between coupling and repulsion heterozygote frequencies
------------- -----------------------------------------------------------------------------------------
[Figure 1](#pone-0069078-g001){ref-type="fig"} shows the principle for populating a 'composite haplotype table'. Each genotype in Part (i) contributes the four possible gametes to the composite haplotype table in Part (ii). In the case of double heterozygotes, where the phase is usually unknown, each of the four possible haplotypes is represented. For all other genotypes the haplotypes are known, but each genotype nevertheless contributes four haplotypes. Note the use of rather than for the diploid sample total to emphasise the distinction between number in a population () and number in a sample (). The normal haploid table cannot be written down from the genotypes in [Figure 1](#pone-0069078-g001){ref-type="fig"}, but the total would be , and, for example, the number of genes = . The marginal totals in the composite table are double these.
{#pone-0069078-g001}
[Figure 2](#pone-0069078-g002){ref-type="fig"} shows a numerical example of the composite haplotype table for one sample of size 32 from the Eastern Australia fruit fly data set, where one microsatellite, , has 3 alleles and a second, , has 4. Again the total in the haplotype table of Part (ii) of [Figure 2](#pone-0069078-g002){ref-type="fig"} is 4x the total in the genotype table of Part (i), rather than 2x as would be found in a table where all haplotypes were known.
{#pone-0069078-g002}
The usual LD coefficient can be calculated for the numbers in the composite haplotype table of [Figure 1](#pone-0069078-g001){ref-type="fig"}, and given the designation . It is:
The LD coefficient of Cockerham and Weir [@pone.0069078-Cockerham1], , is defined in terms of frequencies and , and given as the sum of two coefficients, :
It can be seen from the definitions of and from [@pone.0069078-Cockerham1], ignoring the sample-size correction N/(N−1), that this LD coefficient is double the value of given above.
The intuitive justification for the composite haplotype table is most readily seen in the case of random mating (which is not assumed in the definition of ). In a genotype such as , the true haplotypes will be either and or alternatively and . Under random mating, whichever are the 'false' haplotypes are expected to occur at frequencies that are simply the products of the relevant gene frequencies. The frequencies contributed by the false haplotypes will dilute, but not bias, the haplotype frequencies. It is readily shown that this dilution will be simply a factor of 2. For example, following [Figure 1](#pone-0069078-g001){ref-type="fig"}, the frequency of the haplotype in the composite table, , is the true frequency of the haplotype, , except for the contribution from the double heterozygotes. The true contribution ought to be , whereas it is in fact . Thus the difference between these two is the difference between and , giving.
Under the assumption of random mating, it can be seen that.where D is the usual LD parameter, equal to . Therefore
Subtracting from each side,
The LHS of this equation is, by definition, the disequilibrium coefficient from the composite table, . So the equation is simply
Since this is an expectation under the assumption of random mating, the equation can be written as:where the expectation is taken over replicate populations of the same sample size.
The LD measure introduced by Hill and Robertson [@pone.0069078-Hill3] is . An equivalent parameter can be calculated from the composite haplotype table. The marginal frequencies are the same as for the regular gamete table. So from (1) it follows that the expectation of calculated from the composite table is
It is convenient to define a coefficient where, under random mating, the composite estimates the gametic , rather than one-quarter of the latter. As pointed out above, the LD coefficient of Cockerham and Weir [@pone.0069078-Cockerham1] does this. Therefore we define the statistic aswhich from (), () and () is calculated as
The above definition of ignores an extra factor introduced by Weir [@pone.0069078-Weir2]. This factor arises from the potential covariance of the two alleles at the locus and similarly at the locus. These covariances are implemented through a 'single-locus disequilibrium factor', at the locus and at the locus, which essentially measure deviations from expected homozygosity. The modified definition of , , is
Because of difficulties in implementing this disequilibrium factor, its discussion is deferred to a later section under this label.
### for the composite haplotype table {#s2c1}
Owing to double-counting of genes, the composite gamete table has the property that all marginal totals are multiples of 2, while the overall total is a multiple of 4. Nevertheless a regular can be calculated for such tables, and the resulting values for a × table has close to the expected distribution for degrees of freedom ([Appendix S1](#pone.0069078.s002){ref-type="supplementary-material"}). It has the advantage of having more power than the values calculated from the genotype table, owing to the large number of zero and unit values in the genotype table. Its use in independence tests may, however, be limited by its sensitivity to null alleles (see below).
### Weighting of values {#s2c2}
The calculation of LD for a microsatellite data set involves two levels of summation. There will usually be many loci, say , and each of the pairs yields a separate estimate of . However within each locus pair, say locus and locus , there will be separate calculations for each pair of alleles. These two levels may be labelled as 'between locus pairs' and 'within locus pairs'. Each needs to be separately treated in terms of weighting of the values.
#### Between locus pairs {#s2c2a}
It is often the case that, through missing readings, different locus pairs will have reduced numbers of observations. The sample size for loci and may be designated as . Furthermore some loci will have large numbers of alleles and therefore provide more information than loci with small numbers of alleles. Waples and Do [@pone.0069078-Waples2] have suggested the weighting for the different values, where and are the number of alleles at the and loci respectively. The overall estimate of then becomes
A recent publication [@pone.0069078-Peel1] suggests a slightly different weighting compared to that of Waples and Do [@pone.0069078-Waples2], which would make a small difference to the overall estimate.
#### Within locus pairs {#s2c2b}
values for alleles at locus and at locus can be simply averaged to provide the value. However this has the undesirable property that rare alleles exert a disproportionate influence on the overall value. This effect that can be ameliorated by omitting low frequency alleles [@pone.0069078-Waples2]. A more systematic way of avoiding this problem is to weight alleles according to their frequency. In the case where the frequencies of alleles and are respectively and , a suitable weighting is [@pone.0069078-Zhao1]. The overall value then becomes
Since = 1, this value does not need to be normalised. And since the marginal frequencies are the same for the regular and composite tables, the same weighting applies to both.
It is interesting to contrast this weighting proportional to gene frequencies to the normal weighting of allele pairs for a × table. The with degrees of freedom can be expressed as the sum of × individual values each with 1 df, if the values are weighted by rather than . Thus the weighting gives rare alleles higher weight than common ones. Zhao et al [@pone.0069078-Zhao1] have compared these two measures, amongst others, for their use in QTL mapping, and recommend a standardised weighting for this case. However the higher weighting for rare alleles, as suggested from , performs poorly as just a simple measure of LD ([Appendix S2](#pone.0069078.s003){ref-type="supplementary-material"}).
Because of the different weighting for and , there is no simple relationship between the two statistics. In general, however, significant values of will lead to low estimates of and non-significant values of will be associated with high estimates. See [@pone.0069078-Zaykin1] for a more detailed examination of the statistic.
### The estimation of {#s2c3}
The theory for estimating from unlinked loci has been developed by Weir [@pone.0069078-Weir2], Weir and Hill [@pone.0069078-Weir1] and Hill [@pone.0069078-Hill1]. The effective size refers to a model Wright-Fisher population, and departures from this model, such as permanent pair bonding, make a difference of a factor of 2 in estimates [@pone.0069078-Weir1]. Such pair bonding is, of course, unlikely in fruit fly populations. A model assuming discrete generations as considered here is, however, necessarily an approximation to real populations that are likely to have overlapping generations.
Taking no account, for the moment, of the effect of sample size, the key equation relating the expected LD level to iswhere is the recombination frequency. This reduces tofor unlinked loci, . The expectation for here assumes a balance between increase of due to finite population size and loss due to recombination. All of the equations below assume this balance between drift and recombination. [Equation (8](#pone.0069078.e157){ref-type="disp-formula"}) is derived using the ratio of expectations of rather than the expectation of the ratio (see Hill [@pone.0069078-Hill4]). However computer simulation shows that it works well for loosely linked or unlinked genes, those of interest in the present study. It is unbounded for low values of , when the expression given by Sved and Feldman [@pone.0069078-Sved2]:seems to work better. However for , the RHS of [equation (10](#pone.0069078.e165){ref-type="disp-formula"}) reduces to , which is double the value of [equation (9](#pone.0069078.e159){ref-type="disp-formula"}) and clearly inaccurate at this end of the scale.
[Equations (8](#pone.0069078.e157){ref-type="disp-formula"})--(10) assume the measurement of haplotype or gamete frequencies. As previously indicated, diploid data may be taken into account using the composite LD measure. It follows from [equations (1](#pone.0069078.e070){ref-type="disp-formula"}) and (4) that the expectation for this measure is identical to that of (8):
Sample size is a critical issue in determining LD levels [@pone.0069078-Weir2], [@pone.0069078-Weir1], [@pone.0069078-Hill1]. This is especially the case for unlinked loci, where the levels of and cannot be zero even if there is no association of loci in the population being sampled. The usual procedure in estimating true LD levels in the population is simply to subtract the level of expected for zero LD with a particular sample size. As pointed out in [@pone.0069078-Sved3], however, there is one circumstance where this procedure will not work. With complete LD in the population, , as commonly found for the most tightly linked SNPs, the subtraction will falsely suggest levels less than 1.
The effect on the equation for gametes (8) is to increase the expected value of by a factor of , where is the haploid sample size. The statistic in this case is shown as to indicate that it is an estimate that includes the effects of sampling
In fact the exact expectation for should include the term rather than , equivalent to noting that the exact expectation of is rather than 1 [@pone.0069078-Haldane1]. Weir [@pone.0069078-Weir2] takes this factor into account in working with the 'unbiased' rather than 'biased' value of .
As shown in [equations (1](#pone.0069078.e070){ref-type="disp-formula"}) and (3) of [Appendix S1](#pone.0069078.s002){ref-type="supplementary-material"}, the expectation for the composite , or equivalently the composite LD coefficient , involves the factor , rather than applicable to haploid data. This factor is very close to 1. Similarly the sampling correction factor for for a diploid sample of size is close to :or for unlinked loci:
The estimate for comes from inverting [equation (13](#pone.0069078.e194){ref-type="disp-formula"}), where is calculated according to [equation (6](#pone.0069078.e116){ref-type="disp-formula"}) and each is calculated according to [equation (7](#pone.0069078.e131){ref-type="disp-formula"})
### The effect of null alleles {#s2c4}
Use of the composite disequilibrium index depends critically on the ability to distinguish heterozygous and homozygous genotypes. Unfortunately the presence of any null alleles makes this distinction difficult. Genotypes such as , will be incorrectly scored as . Homozygous null genotypes are not easily detected, since it is difficult to distinguish between absence of a band and simple failure of the PCR reaction in the rare cases expected for homozygotes.
The expected effect of null alleles on the composite LD statistic can be quantified as in [Appendix S3](#pone.0069078.s004){ref-type="supplementary-material"}. This shows that a null allele at one of the two loci at a frequency alters the expectation of [equation (1](#pone.0069078.e070){ref-type="disp-formula"}) to:
The statistic is increased by the factor .
Although this effect may be small, it can readily be shown to overwhelm the calculations when the expected LD value is small due to high effective population size. In the case of an infinitely large population, the true value of is expected to be just the sampling correction, which is approximately . A null allele at one of the two loci is expected to increase this value to . Applying [equation (13](#pone.0069078.e194){ref-type="disp-formula"}), the estimated value of is then found by subtracting the usual sampling contribution, giving
Applying numerical values to [equation (15](#pone.0069078.e210){ref-type="disp-formula"}), for a sample size and null frequency , the equation yields a value for of 259. The actual population in this case should be infinitely large, so that a null allele frequency as low as 2% can have a strikingly large effect. A null allele at frequency 0.1, still difficult to detect, leads to a estimate of 45.
### Simulations with null alleles {#s2c5}
Simulations with null alleles have been carried out to test these expectations. These are 2-locus simulations with heterozygosities ranging from 50% to 87%. Under these conditions, [equation (15](#pone.0069078.e210){ref-type="disp-formula"}) may slightly over-estimate the effect of null alleles. For example, in the above case with and , simulation yields a value of compared to expectation of 259, while yields compared to 45.
Simulation can also be used to check on more realistic cases where the value of comes from multiple loci, rather than a two-locus simulation. These show that even low levels of null alleles at a single locus may have measurable effects. For example with 32 loci each with 5 alleles, the presence of just one locus amongst these having a null allele frequency of 10% can have a detectable effect, reducing the expected value of from infinitely large to less than 1,000. Much the same result is found for 5 loci each with a null frequency of 2%. Simulations also indicate that 8 out of 16 loci having null alleles at a particular frequency has much the same effect as one out of two loci in the simulations and calculations given above.
### Correcting the effect of null alleles through permutation {#s2c6}
A general formulation for the estimation of may be given as follows:
Here is the estimate derived from the data, and is the true measure of LD in the population, which is the quantity of interest in estimating . The analysis above has shown that in the absence of null alleles, the correction factor is attributable purely to sampling, and is . The analysis on null alleles has shown that these will act as disturbing factors, whose effect can conveniently be subsumed into the correction factor in [equation (16](#pone.0069078.e223){ref-type="disp-formula"}).
A randomising procedure can be suggested that will ameliorate the effect of null alleles. If the genotypes at each locus are independently randomly permuted amongst individuals, such as in the exact test of significance of LD, eg. [@pone.0069078-Rousset1], there can be no underlying LD. So the mean value of given by the average of many such randomly permuted samples is a direct estimate of the correction factor in (16) taking into account the actual genotype structure. If is the estimated value of in such permuted samples, then [equation (16](#pone.0069078.e223){ref-type="disp-formula"}) becomes
From [equation (9](#pone.0069078.e159){ref-type="disp-formula"}), the estimate of is then simply
Both and can be given with or without the sampling correction factor . In the data tables below, the factor has been subtracted from both in order to use [equation (14](#pone.0069078.e198){ref-type="disp-formula"}) to estimate the value of with no permutation. However for the value of with subtracted, the sampling factor cancels out and could have been omitted.
The permutation approach can be tested by simulation. This is shown in the first four lines of [Table 2](#pone-0069078-t002){ref-type="table"}. All, except for the final two rows, involved 16 loci simulated for 20 generations, followed by sampling of 32 individuals. The first row shows the average value for a range of population sizes from 32 to 1028. The second row shows the estimated values using [equation (14](#pone.0069078.e198){ref-type="disp-formula"}), with each of the values calculated directly from the composite haplotype table according to [equations (1](#pone.0069078.e070){ref-type="disp-formula"}) and (4). The values are in good agreement with expectation.
10.1371/journal.pone.0069078.t002
###### Observed statistics from simulations with and without incorporating single-locus disequilibrium.
{#pone-0069078-t002-2}
Actual *N~e~* 32 64 128 256 512 1024
-------------------------------------------------------------------------------- --------- --------- --------- --------- --------- ---------
0.00993 0.00511 0.00255 0.00129 0.00065 0.00032
\(2\) *N~e~* (from [equation 14](#pone.0069078.e198){ref-type="disp-formula"}) 34 65 131 259 516 1036
\(3\) *N~e~* (null alleles) 26 41 59 76 89 97
\(4\) *N~e~* (null alleles+permute) 33 64 127 249 494 1025
\(5\) (diseq. included) 0.01067 0.00598 0.00352 0.00225 0.00163 0.00133
\(6\) *N~e~* (diseq. included) 31 56 95 148 203 249
\(7\) *N~e~* (diseq. included+permute) 35 68 134 265 523 1040
\(8\) *N~e~* (null alleles - diseq. included) 31 56 96 147 206 248
\(9\) *N~e~* (nulls - diseq. corr.+permute) 35 68 136 274 559 1127
\(10\) s.d. 0.00655 0.00397 0.00285 0.00231 0.00205 0.00193
\(11\) s.d. (diseq. included) 0.00468 0.00272 0.00186 0.00146 0.00126 0.00117
\(12\) s.d. (32 loci) 0.00454 0.00277 0.00195 0.00153 0.00134 0.00124
\(13\) s.d. (diseq. included, 32 loci) 0.00299 0.00167 0.00108 0.00081 0.00067 0.00059
All used sample size S = 32.
The effect of introducing null alleles is shown in row (3). The simulations here involved choosing 8 of the 16 loci, and replacing 5% of alleles with null alleles in these. The values calculated using equation () are drastically reduced, especially for the higher population sizes. However the permutation correction in row (4) essentially brings the estimated values back to their expected value.
In the case of an infinitely large population, simulation is not necessary to justify the permutation approach for correcting for null alleles. The loci would be in linkage equilibrium in such a population, with a true value of of zero. The only contributing factor to the observed value of must be the correction factor, attributable to null alleles, plus the usual sampling factor of approximately . Additional permutation of genotypes in a sample from a population with zero LD will not have any effect, so the estimates with and without permutation will be identical and equal to .
The case of an infinitely large population also serves to show that the permutation approach will NOT work in removing biases due to non-random mating. For example, a sample might consist of individuals from two independently randomly mating populations, where the substructure has not been recognised. Such a sample will give a reduced estimate of due to the induced LD [@pone.0069078-Nei1] even though there may be no LD within each of the two contributing populations. However permuting the sample cannot resolve this issue. It can be seen that the value of from the composite table will be zero, except for the normal sampling component of approximately , assuming no null alleles. The application of [equation (17](#pone.0069078.e231){ref-type="disp-formula"}) would then falsely indicate that the LD within populations was real and attributable to small population size. A valid correction could be produced if the sub-samples from the two populations could be independently permuted, which is possible in computer simulation but not with real data where the substructure is unknown.
Taking account of all types of departure from random mating thus appears difficult. But Waples and England [@pone.0069078-Waples3] have considered the case of migration into a random mating population, and shown that there is little effect on estimates in this case.
### Including the single-locus disequilibrium factor {#s2c7}
As mentioned above, a homozygosity correction term was suggested by Weir [@pone.0069078-Weir2], as shown in [equation (5](#pone.0069078.e091){ref-type="disp-formula"}). The effects of this term are shown in row (5) of [Table 2](#pone-0069078-t002){ref-type="table"}, the value, and row (6), the value. The latter shows a substantial bias in values, especially for the larger population sizes. The size of this discrepancy seems surprising, since, under random mating, the mean value of the homozygosity correction should be zero, and only a small correction should result. However there is a bias due to the fact that, in a finite-size sample, the expectation of frequency is less than . This is most evident where there is a single allele, giving , but where the frequency of the genotype must be zero.
The obvious way of eliminating this bias would seem to be the use of \] as the expected frequency of homozygotes. But simulation shows that this substantially over-corrects the bias. It is, however, possible, just as in the case of correcting the bias for null alleles, to use a permutation correction. This involves calculation of from equation (), random permutation of genotypes in the sample, and calculation of in permuted samples. The procedure may be summarised as:
From [equation (19](#pone.0069078.e274){ref-type="disp-formula"}), the estimate of is
Simulation in row (7) of [Table 2](#pone-0069078-t002){ref-type="table"} shows that this correction works well for all values.
The homozygosity deviation factor, , was not specifically designed in [@pone.0069078-Weir2] to take into account null alleles. It seems particularly vulnerable to their effect, since may be substantially over-estimated. However simulation shows that this factor dramatically improves rather than worsens the effect of null alleles. In contrast to the bias of the considered previously that lacks the disequilibrium correction, row (8), which introduces null alleles at the same frequency of 5% in half of the loci, gives almost the same value as row (6) where there are no null alleles. As previously, the bias due to the factor can be eliminated by subtracting the permutation using equation (), as shown in row (9).
A second advantage of the disequilibrium factor is that it reduces the variance of estimates. The estimates given in [Table 2](#pone-0069078-t002){ref-type="table"} are based on large numbers of replicates. However the variability between individual simulation runs is high. Estimated standard deviations of and are given in rows (10) and (11). Both standard deviations are high in relation to the mean, but that associated with is especially so. Of course the magnitude of the standard deviations is heavily dependent on the choice of number of loci and heterozygosity levels. Doubling the number of loci from 16 to 32 substantially reduces standard deviations, row (12) and row (13), but the relativities between the two terms are maintained.
In summary of [Table 2](#pone-0069078-t002){ref-type="table"}, only the original estimate from [equation (14](#pone.0069078.e198){ref-type="disp-formula"}), where lacks the single-locus disequilibrium factor, gives unbiased estimates. Nevertheless there is a strong reason for including the faxtor, provided that the bias in values is compensated, either by permutation as above, or by empirical correction as implemented in the computer program LDNe [@pone.0069078-Waples2]. Weir's insight in introducing this factor is vindicated by the increased accuracy of estimation and lowered sensitivity to null alleles.
Results and Discussion {#s3}
======================
Results for the East coast populations are given in [Table 3](#pone-0069078-t003){ref-type="table"}. Populations with low sample numbers, 15 or less, were omitted from the analysis, leaving 40 out of the original 52 samples. The table includes mostly samples from outbreak areas where the flies were not normally found, but also ten samples where the flies are endemic, including one from Queensland, the home range of the flies. The expectation is that these ten are samples from large populations.
10.1371/journal.pone.0069078.t003
###### Summary of *N~e~* estimated by various procedures for East coast outbreak populations of *B.tryoni*, with the most likely estimate shown by ⇓.
{#pone-0069078-t003-3}
S No homozygote correction Homozygote correction Likelihood Significance
------------------------------------------- ----- -------------------------- ----------------------- ------------------------- ----- ----- --------------------------------------- ---------------------------------------
Albury03 27 60 ∞ ∞ ∞ ∞ [\*](#nt103){ref-type="table-fn"}
Barooga03 33 40 30 40 20 20 [\*\*\*](#nt105){ref-type="table-fn"}
Condobolin02 42 40 ∞ ∞ ∞ ∞ [\*](#nt103){ref-type="table-fn"}
Coota02 43 110 ∞ 450 340 510 [\*\*\*](#nt105){ref-type="table-fn"}
Corowa02 22 20 120 180 100 ∞
Cowra 20 20 230 150 180 ∞
Deniliquin02 40 30 40 40 30 ∞ [\*\*\*](#nt105){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Deniliquin03 53 40 100 150 70 90 [\*](#nt103){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Deniliquin04 73 50 130 160 70 110 [\*](#nt103){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Dubbo02 26 30 180 130 160 ∞ [\*\*](#nt104){ref-type="table-fn"}
Forbes02 34 40 250 180 170 ∞ [\*\*](#nt104){ref-type="table-fn"}
Grenfell02 31 130 ∞ ∞ ∞ ∞ [\*\*\*](#nt105){ref-type="table-fn"}
Hay02 26 20 30 20 20 140 [\*\*\*](#nt105){ref-type="table-fn"}
Hay03 28 40 230 120 50 80 [\*](#nt103){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Henty02 20 20 120 60 50 190 [\*\*](#nt104){ref-type="table-fn"}
LakeCarg02 74 30 40 50 30 70 [\*\*](#nt104){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Leeton03 82 70 110 160 70 80 [\*\*](#nt104){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Narrandera04 25 30 ∞ 770 130 510 [\*](#nt103){ref-type="table-fn"}
Parkes02 20 30 130 100 80 500
Parkes03 41 30 140 140 190 310 [\*\*](#nt104){ref-type="table-fn"}
Temora02 20 20 120 160 150 ∞
TheRock02 20 30 410 170 100 ∞
Tumut 20 20 670 470 270 ∞ [\*](#nt103){ref-type="table-fn"}
Wagga02 57 70 790 ∞ ∞ ∞
Wagga03 162 210 660 740 610 860
Wahgunyah 24 20 90 70 50 ∞ [\*](#nt103){ref-type="table-fn"} [\*](#nt103){ref-type="table-fn"}
Wilcannia02 43 20 50 60 30 50 [\*\*\*](#nt105){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Wodonga 42 30 110 110 100 130 [\*\*\*](#nt105){ref-type="table-fn"}
WWyalong03 24 120 ∞ ∞ 110 ∞ [\*](#nt103){ref-type="table-fn"}
Young02 49 110 170 380 400 440 [\*\*](#nt104){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Coffs02[¶](#nt102){ref-type="table-fn"} 18 40 70 60 70 ∞
Foster02[¶](#nt102){ref-type="table-fn"} 34 40 ∞ ∞ ∞ ∞ [\*\*\*](#nt105){ref-type="table-fn"} [\*\*\*](#nt105){ref-type="table-fn"}
Grafton03[¶](#nt102){ref-type="table-fn"} 29 40 290 280 510 ∞ [\*\*\*](#nt105){ref-type="table-fn"}
Maclean02[¶](#nt102){ref-type="table-fn"} 34 50 600 280 360 ∞ [\*\*\*](#nt105){ref-type="table-fn"}
NSW03[¶](#nt102){ref-type="table-fn"} 42 90 380 ∞ ∞ ∞
QLD03[¶](#nt102){ref-type="table-fn"} 42 70 430 290 530 ∞ [\*\*\*](#nt105){ref-type="table-fn"}
Sawtell02[¶](#nt102){ref-type="table-fn"} 34 120 ∞ ∞ ∞ ∞
SWRocks02[¶](#nt102){ref-type="table-fn"} 33 40 ∞ ∞ ∞ ∞ [\*\*\*](#nt105){ref-type="table-fn"}
Syd03[¶](#nt102){ref-type="table-fn"} 42 130 ∞ ∞ ∞ 630 [\*](#nt103){ref-type="table-fn"}
Taree03[¶](#nt102){ref-type="table-fn"} 30 40 ∞ ∞ ∞ ∞
Non-outbreak population.
Significant at 5% level.
Significant at 1% level.
Significant at 0.1% level.
The results are based on 29 microsatellites, a total of 29×28/2 = 406 locus pairs. Because of missing readings, not all pairs are present in all populations.
Amongst the 29 loci, 5 pairs are known to be closely linked, 51 pairs to be loosely linked, and 197 to be unlinked [@pone.0069078-Zhao2]. For the remaining 153 locus pairs, one or both chromosomes are unknown. Average values of for the four classes are 0.0434, 0.0153, 0.0084 and 0.0096 respectively. As expected, average values are higher for the known linked loci.
Values of were calculated from the composite haplotype tables, and values (column 3) were then calculated from these values using [equation (14](#pone.0069078.e198){ref-type="disp-formula"}). All populations, including the eight non-outbreak populations, show very low estimated population sizes. All are highly significantly different from infinite population size. The major conclusion from the above analysis, however, is that the existence of either null alleles or population sub-structure can cause cause values to be substantially under-estimated.
A direct test for null alleles is given in [Table 4](#pone-0069078-t004){ref-type="table"}. The signal for null alleles is, eg. [@pone.0069078-Chapuis1], excess of homozygotes over expectation. In a data set with multiple populations, a non-parametric test can be carried out based on number of populations where there is such an excess. [Table 4](#pone-0069078-t004){ref-type="table"} shows the results, revealing at least 10 out of 29 microsatelltes with significant excess of homozygotes, which, in the lack of systematic homozygote excess, can likely be attributed to null alleles rather than to population structure.
10.1371/journal.pone.0069078.t004
###### Excess of homozygosity for different microsatellites.
{#pone-0069078-t004-4}
Rank Microsatellite Number of populations
------ ---------------- ----------------------- ----
1 Bt2.9a 36 39
2 Bt6.1a 33 36
3 Bt15 36 40
4 Bt4.1a 36 40
5 Bt1.7a 35 40
6 Bt2.6a 33 40
7 Bt2.6b 31 38
8 Bt3.2b 30 37
9 Bt1.6a 31 39
10 Bt32 30 39
11 Bt10 30 40
12 Bt7.9a 29 39
13 Bt6.12a 27 40
14 Bt5.10a 27 40
15 Bt8.5a 26 40
16 Bt11 25 40
17 Bt7.2b 23 39
18 Bt1.1a 20 40
19 Bt9.1a 20 40
20 Bt14 18 40
21 Bt8.6a 18 40
22 Bp78 18 40
23 Bt17 17 40
24 Bt4.3a 16 40
25 Bt4.6a 15 38
26 Bt6.8a 15 40
27 Bt8.12a 15 40
28 Bt6.10b 14 40
29 Bt5.8a 9 38
Returning to [Table 3](#pone-0069078-t003){ref-type="table"}, column 4 shows the values of using values corrected using [equation (17](#pone.0069078.e231){ref-type="disp-formula"}). The correction factor in this case comes from 200,000 simulated populations for each outbreak sample. The values clearly have a more realistic mixture of population sizes than the estimates based on the raw values. Positive values of greater than 1,000 are listed as infinite, as also are the estimates associated with negative estimates. Lower values of have been rounded to the nearest 10.
The disequilibrium factor is introduced in column 5. This column is marked as giving the most likely estimate of . As expected, all of the really small population size estimates come in the outbreak populations rather than in the endemic populations.
The values in columns 3--5 are based on the unlinked locus pairs, including the 153 additional pairs likely to be loosely linked or unlinked. The values in column 6 are the equivalent corrected estimates based on all locus pairs. These can be directly compared to the values of given by the LDNe program [@pone.0069078-Waples2], also based on all locus pairs. There is good agreement for the smallest population sizes, although the LDNe program shows infinite sizes in a number of cases where the values of in column 5 are finite.
values in column 5, using unlinked loci, differ very little from values on column 6 using all loci. The expectation is that the use of linked loci will lead to under-estimation of . Many, but not all, values in column 6 are slightly below those in column 5, but the differences are not large. This result seems fortuitous, given that linkage relationships are not as well established for many organisms, necessitating the use of all locus pairs.
The final two columns of [Table 3](#pone-0069078-t003){ref-type="table"} show two different tests of significance, each based on the unlinked plus likely unlinked sub-sample of locus pairs. The first is the usual genotype likelihood test of LD [@pone.0069078-Rousset1], based on permutation of genotypes, with log likelihoods of the genotype tables summed over all relevant locus pairs. The second is a likelihood test based on permutation of genotypes, with likelihoods calculated on the composite haplotype tables. This test seems much more sensitive. Partly this is because, as indicated above and illustrated in [Figure 2](#pone-0069078-g002){ref-type="fig"}, the composite haplotype table is much denser than the genotype table, where all the zero and unit values do not contribute to the likelihood. However the second test is influenced by LD, but also by null alleles. The significant values are mostly associated with low population sizes, but there are exceptions to this in both directions. In general, the significance tests seem to be of limited value in judging whether population sizes are infinite or not.
The results from North-West samples [@pone.0069078-Cameron1] are given in [Table 5](#pone-0069078-t005){ref-type="table"}. The results show a comparable proportion of high population numbers compared to the East coast populations of [Table 3](#pone-0069078-t003){ref-type="table"}. Less has been known about these populations, but these results would suggest that, with the exception of the final two samples from Broome and Derby in West Australia, these are well-established outbreaks in most cases.
10.1371/journal.pone.0069078.t005
###### Estimated *N~e~* values for North-West population samples.
{#pone-0069078-t005-5}
S No homozygote correction Homozygote correction Likelihood Significance
------------- ---- -------------------------- ----------------------- ------------------------- ----- ----- --------------------------------------- ---------------------------------------
K-Ke2002 22 30 160 270 90 ∞ [\*\*\*](#nt108){ref-type="table-fn"}
K-Ke2003 39 20 60 90 100 ∞ [\*\*\*](#nt108){ref-type="table-fn"}
K-Kl2000 77 70 240 290 160 190
K-Kl2001 50 60 190 210 170 ∞
K-Kl2002 44 30 60 100 70 80 [\*\*](#nt107){ref-type="table-fn"} [\*\*\*](#nt108){ref-type="table-fn"}
K-Kl2003 50 50 ∞ ∞ ∞ ∞ [\*\*](#nt107){ref-type="table-fn"}
K-Km2002 27 20 420 280 90 50 [\*](#nt106){ref-type="table-fn"} [\*\*\*](#nt108){ref-type="table-fn"}
N-DWN02 40 20 50 80 90 780
N-DWN03 20 60 ∞ ∞ ∞ ∞ [\*\*\*](#nt108){ref-type="table-fn"}
N-DWN99 20 ∞ ∞ ∞ ∞ ∞
N-DWNBUSH02 30 40 ∞ ∞ ∞ ∞
N-DWN-KTH03 19 60 ∞ ∞ ∞ ∞
N-GOVE02 17 ∞ ∞ ∞ ∞ ∞
N-KAK02 40 40 80 120 120 440 [\*\*\*](#nt108){ref-type="table-fn"}
N-KTH03 20 30 100 230 ∞ ∞
N-KTHGO02 28 80 ∞ 440 470 ∞ [\*\*](#nt107){ref-type="table-fn"}
N-mDK02 27 40 300 180 270 ∞
N-mDKA02 20 80 ∞ ∞ 150 ∞
N-mKKu03 36 30 100 120 80 200 [\*\*](#nt107){ref-type="table-fn"}
N-nDWN02 50 70 140 210 320 ∞ [\*](#nt106){ref-type="table-fn"}
N-nDWN03 20 90 ∞ ∞ ∞ 100 [\*\*\*](#nt108){ref-type="table-fn"}
N-nKTH03 20 30 170 270 420 ∞
Q-AT02 21 40 ∞ ∞ ∞ ∞
Q-ATH99 21 110 ∞ ∞ ∞ 340
Q-CT00 23 140 ∞ ∞ ∞ ∞ [\*](#nt106){ref-type="table-fn"}
Q-CT99 17 50 90 280 ∞ ∞
Q-LR00 24 80 ∞ ∞ ∞ 110
Q-MB02 21 40 ∞ ∞ ∞ ∞
Q-Qld00 94 110 260 260 390 ∞ [\*\*\*](#nt108){ref-type="table-fn"}
Q-QLD01 55 70 280 280 630 300
Q-QLD02 40 40 220 250 160 ∞ [\*](#nt106){ref-type="table-fn"}
Q-QLD03 42 40 250 110 140 ∞
W-Brm01 21 20 30 40 30 80
W-Der01 17 10 10 10 10 10 [\*\*\*](#nt108){ref-type="table-fn"} [\*\*](#nt107){ref-type="table-fn"}
Significant at 5% level.
Significant at 1% level.
Significant at 0.1% level.
Summary of the Findings {#s3a}
-----------------------
The Burrows composite index can be equivalently derived from a 'composite haplotype table' in which all genotypes sampled contribute four possible haplotypes.
Although the composite haplotype table has marginal totals that are even numbers due to double counting, a valid × can be calculated for the table. The value calculated from this table, , needs to be multiplied by a factor of 4 to give , a valid estimator of .
The expected value calculated for the table is in the absence of LD. This contrasts with the sampling correction of for calculated when haplotypes can be recognised.
The overall calculation of involves summation of values from different locus pairs. Within locus pairs, it involves summation of values for each pair of alleles. The weighting for the former is taken from [@pone.0069078-Waples2], while a weighting proportional to gene frequencies is proposed for the latter.
The results when this formula are applied to data from Queensland fruit fly give low values in all samples, including ones from known large endemic populations. Null alleles are suggested as a cause for this discrepancy, and shown to be frequent in the data.
The effect of a null allele at frequency is shown to increase the composite value by the fraction . Although this effect seems small, it will nevertheless overwhelm the calculations for large population sizes.
The value can be corrected for null alleles using a comparison between the calculated value and an equivalent value calculated when genotypes in the sample are permuted at random. This correction is verified by simulation.
The single-locus disequilibrium factor suggested by Weir [@pone.0069078-Weir2], equivalent to a homozygosity correction, is introduced into the calculation. This alters the value of to . Use of is shown to bias the values due to the difficulty of calculating the single-locus disequilibrium factor using in a finite population.
Simulation shows that this bias can be rectified using the same permutation approach as for null alleles.
, and calculated from , have lower variances than , and calculated from .
Simulation shows that the values are almost unaffected by null alleles, in sharp contrast to the values.
The estimates of from both East coast and NorthWest populations are, as expected, mostly low for outbreak populations and high for endemic populations.
The calculations are based on loci known to be unlinked, but are not substantially changed when all locus pairs are considered. Linkage information is usually not available for non-laboratory organisms, and this result shows that lack of such information may not be critical in calculating based on LD.
Although the LDNe program [@pone.0069078-Waples2] is empirically based, it uses the single-locus disequilibrium factor, and appears to work well both with and without null alleles.
Supporting Information {#s4}
======================
######
**The effect on the estimate of r2 from ?2 weighting compared to allele frequency weighting when introducing a single new mutant.**
(TIF)
######
Click here for additional data file.
######
***x*** **^2^ and** ***r*** **^2^ for the composite haplotype table.**
(PDF)
######
Click here for additional data file.
######
***x*** **^2^ as a measure of LD for multiple alleles.**
(PDF)
######
Click here for additional data file.
######
**The effect of null alleles on r^2^.**
(PDF)
######
Click here for additional data file.
######
**Microsatellite data for East coast samples of** ***Bactrocera tryioni*** **(Queensland Fruit Fly).**
(TXT)
######
Click here for additional data file.
######
**Microsatellite data for NorthWest samples of** ***Bactrocera tryioni*** **(Queensland Fruit Fly).**
(TXT)
######
Click here for additional data file.
Discussions with Ian Franklin, Bill Hill, Bill Sherwin and Robin Waples are gratefully acknowledged
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
[^2]: Conceived and designed the experiments: JAS ECC ASG. Performed the experiments: ECC ASG. Analyzed the data: JAS ECC ASG. Contributed reagents/materials/analysis tools: JAS ECC ASG. Wrote the paper: JAS.
| {
"pile_set_name": "PubMed Central"
} |
(J Am Heart Assoc. 2016;5:e003524 doi: [10.1161/JAHA.116.003524](10.1161/JAHA.116.003524))
Introduction {#jah31684-sec-0005}
============
Autonomic dysfunction is a typical phenomenon in patients with chronic heart failure (HF).[1](#jah31684-bib-0001){ref-type="ref"} Measures of more advanced autonomic dysfunction are indicators of the severity of HF and adverse prognosis.[2](#jah31684-bib-0002){ref-type="ref"}, [3](#jah31684-bib-0003){ref-type="ref"}, [4](#jah31684-bib-0004){ref-type="ref"}, [5](#jah31684-bib-0005){ref-type="ref"}, [6](#jah31684-bib-0006){ref-type="ref"} They may include heart rate variability,[4](#jah31684-bib-0004){ref-type="ref"}, [7](#jah31684-bib-0007){ref-type="ref"} heart rate recovery,[3](#jah31684-bib-0003){ref-type="ref"} baroreflex sensitivity,[5](#jah31684-bib-0005){ref-type="ref"} measurement of norepinephrine spillover,[2](#jah31684-bib-0002){ref-type="ref"}, [7](#jah31684-bib-0007){ref-type="ref"} and muscle nerve sympathetic activity.[8](#jah31684-bib-0008){ref-type="ref"} However, the techniques needed to identify these are complex, and their application requires considerable infrastructure and time and are, therefore, not well suited for serial testing in clinical practice. Less sophisticated tests of autonomic function are not available in HF patients yet. However, an easy and quick test for assessing autonomic function that provides prognostic information in HF patients would be highly desirable.
In contrast to the situation in HF, a battery of simple tests based on the heart rate (HR) and blood pressure response to a variety of maneuvers has been in clinical use for decades for the assessment of diabetic autonomic neuropathy.[9](#jah31684-bib-0009){ref-type="ref"} One of these tests assesses the HR response to standing using continuous ECG monitoring of HR and RR intervals, respectively. The normal response to standing is characterized by a rapid baroreceptor‐mediated rise in HR, which peaks around the 15th beat (shortest RR interval), followed by a relative bradycardia with a maximum RR interval around the 30th beat after getting up. The ratio of the longest (\~30th beat) and shortest (\~15th beat) RR intervals ("30/15 ratio") is used to describe this HR response. However, after the 30th beat, the HR remains elevated compared to the supine baseline in the presence of an intact sympathicovagal balance, and patients with diabetic neuropathy are characterized not only by a low 30/15 ratio but also by a lack of or a severely overall blunted rise in HR from the supine to the standing position.[10](#jah31684-bib-0010){ref-type="ref"} Thus, the change in HR following getting up from the supine to the upright position (∆HR) might be a simple test to assess the degree of autonomic dysfunction in patients with chronic HF, which may also predict prognosis, but this has not yet been investigated.
Therefore, we have prospectively measured ∆HR in patients with chronic HF included in the Trial of Intensified Medical Therapy in Elderly Patients with Congestive Heart Failure (TIME‐CHF)[11](#jah31684-bib-0011){ref-type="ref"}, [12](#jah31684-bib-0012){ref-type="ref"} to describe the association of ∆HR with clinical characteristics in an elderly HF cohort and to evaluate its prognostic value.
Methods {#jah31684-sec-0006}
=======
Study Population and Protocol {#jah31684-sec-0007}
-----------------------------
This is a post‐hoc analysis of TIME‐CHF (isrctn.org identifier ISRCTN43596477). Design[11](#jah31684-bib-0011){ref-type="ref"} and main results[12](#jah31684-bib-0012){ref-type="ref"} of TIME‐CHF have been published previously. In brief, TIME‐CHF was a randomized, controlled multicenter trial comparing an N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP)‐guided vs a symptom‐guided management strategy in patients with chronic HF aged ≥60 years with symptoms corresponding to New York Heart Association (NYHA) ≥II, HF‐related hospitalization within 12 months prior to inclusion, and an age‐adjusted elevated NT‐proBNP plasma concentration (\>400 ng/L in those \<75 years, \>800 ng/L in those ≥75 years). Patients with both reduced (n=499) and preserved (n=123) left ventricular ejection fraction (LVEF) were included between January 2003 and December 2006. The study was approved by the local ethics committees, and all participants provided written informed consent.
For the present analysis, patients in sinus rhythm and without any pacemaker (bradypacing, cardiac resynchronization; patients with defibrillators and pure VVI backup pacing were not excluded) throughout the trial were eligible (n=327). HR was measured in the supine position after at least 1 minute of supine rest. HR in the upright position was measured immediately after getting up by pulse palpation or an automated blood pressure monitor, and ΔHR was calculated as the difference between HR in the upright and supine positions. Thus, a positive value for ΔHR indicates a rise in HR upon standing, whereas a negative value indicates a fall in HR after getting up. This measurement of the HR at rest and after getting up was part of the study protocol, and data were collected in a prospective manner. The present analysis is based on the 321 patients with complete data on HR in the supine and upright positions. Patients were followed in the outpatient clinics after 1, 3, 6, 12, and 18 months. Medical treatment was prescribed in accordance with predefined escalation rules either to reduce symptoms to NYHA ≤II or also to reduce NT‐proBNP below the age‐specific target level (\<400 ng/L in patients \<75 years, \<800 ng/L in those ≥75 years) as described in the design paper of TIME‐CHF.[11](#jah31684-bib-0011){ref-type="ref"} The primary endpoint of TIME‐CHF was 18‐month survival free of any hospitalization. Secondary endpoints included survival and survival free of HF‐hospitalization at 18 months. For the present analysis, survival free of HF‐hospitalization was the primary endpoint, and the other two were secondary endpoints.
Statistical Analysis {#jah31684-sec-0008}
--------------------
Descriptive statistics are expressed as mean±standard deviation, or median (interquartile range) for continuous variables and as numbers and percentages for categorical variables. Distribution of continuous data was assessed using Kolmogorov‐Smirnov tests. First, the association between ΔHR and outcomes was examined using univariate Cox regression. Then, best cutoff for ΔHR to predict death or HF hospitalization based on log‐rank testing was performed. Characteristics of patients with ΔHR above and below this cutoff were compared by chi‐squared tests for categorical variables and by t tests or Mann‐Whitney U tests for continuous variables, as appropriate. Kaplan‐Meier curves were constructed for calculating the time‐dependent occurrence of events in patients with ΔHR above and below this cutoff, and for comparison between these groups the log‐rank test was used. To test the independence of the association between ΔHR (as a continuous and dichotomized variable) and outcomes, multivariate Cox regression was performed after testing the proportional hazard assumption. A stepwise backward model was used. To account for the number of events, the number of covariates was limited to those with the strongest association with the dependent variable based on Wald score in addition to the variable of interest ΔHR. For the model with death or HF hospitalization as the dependent variable, the following covariates were included in the model: age, ischemic HF etiology, NT‐proBNP (log~10~‐transformed), estimated glomerular filtration rate (eGFR), hemoglobin, peripheral arterial obstructive disease (PAOD), NYHA functional class, presence and extent of rales (4 categories), and β‐blocker use at baseline. For the model with death as the dependent variable, the following covariates were included in the model: ischemic HF etiology, NT‐proBNP (log~10~‐transformed), eGFR, hemoglobin, NYHA class, and presence and extent of rales. The results of these models are presented in the tables. We also tested whether adjustment of this final model for sex, body mass index, and diabetes changed the findings, and the results of these analyses are mentioned in the text. Interactions between ΔHR and patient characteristics were analyzed using Cox regression with ΔHR and a second covariate and the interaction term forced into a model. In particular, the interactions between ΔHR and age, resting heart rate, NT‐pro‐BNP‐guided therapy (ie, allocation to the NT‐proBNP‐guided arm), LVEF stratum (LVEF \<45% vs ≥45%), diabetes, and baseline β‐blocker use were tested 1 at a time. To illustrate the prognostic value of the multivariate model, we constructed a 5‐point score including the following items: ΔHR ≤3 bpm, ischemic HF etiology, eGFR \<47 mL/min per 1.73 m^2^, no β‐blocker use at baseline, and the presence of rales at baseline. One point was allocated for each item. Thus, the maximal score was 5 points, and the minimal score was 0 points. Because there were few patients with 0, 4, and 5 score points, the following 3 risk categories were built for comparison using Kaplan‐Meier curves and log‐rank tests and Cox regression, respectively: 0 or 1 point, 2 points, and 3, 4, or 5 points.
To also assess the prognostic value of changes in ΔHR over time we performed a landmark analysis in which patients with available ΔHR measurements at baseline and 6 months and without event during the first 6 months were included. Within this population, the association between the change in ΔHR from baseline to month 6 and death or HF hospitalization during the following 12 months was assessed. Bootstrapping (1000 bootstrap samples) was used to calculate 95% confidence intervals (95% CI) and *P*‐values in Cox regression analyses. The level of statistical significance was set at a 2‐tailed probability value ≤0.05. For interactions a *P*\<0.1 was considered relevant. Statistical analysis was performed using the IBM^®^ SPSS^®^ for Windows^®^ software (version 22.0, SPSS^®^ Inc, Chicago, IL).
Results {#jah31684-sec-0009}
=======
Patient Characteristics {#jah31684-sec-0010}
-----------------------
The study population consisted of 321 patients (57% men) with a mean age of 76±8 years and a mean LVEF of 34±12% (Table [1](#jah31684-tbl-0001){ref-type="table-wrap"}). The mean HR in the supine position in the entire population was 74±13 bpm. The mean HR after getting up was 78±13 bpm, and the mean ∆HR was 5±6 bpm.
######
Baseline Characteristics of the Entire Study Population and Patients With ∆HR \>3 bpm vs ≤3 bpm
All (n=321) ∆HR \>3 bpm (n=211) ∆HR ≤3 bpm (n=110) *P* Value
--------------------------------------------------------------------------------------- -------------------------------------- ------------------------------------ ------------------------------------- -----------
Age, y 76±8 76±8 77±7 0.17
Male sex 182 (57%) 119 (56%) 63 (57%) 0.91
Body mass index, kg/m^2^ 25±4 25±4 26±4 0.13
Ischemic heart failure etiology 219 (68%) 136 (64%) 83 (75%) 0.06
LVEF ≥45% 53 (16%) 33 (16%) 20 (18%) 0.64
NT‐proBNP‐guided therapy 156 (49%) 101 (48%) 55 (50%) 0.73
Medical history
Hypertension 226 (70%) 150 (71%) 76 (69%) 0.70
Diabetes 119 (37%) 69 (33%) 50 (45%) 0.03
Stroke 25 (8%) 13 (6%) 14 (13%) 0.05
Chronic obstructive lung disease 67 (21%) 42 (20%) 25 (23%) 0.85
Peripheral arterial obstructive disease 63 (20%) 37 (18%) 26 (24%) 0.65
Smoking 50 (16%) 33 (16%) 17 (15%) 1.0
Clinical characteristics
Heart rate, bpm 74±13 73±12 75±13 0.16
∆HR, bpm 5±6 8±5 −1±4 \<0.001
Supine systolic blood pressure, mm Hg 120 (110‐132) 120 (109‐130) 124 (110‐136) 0.11
Upright systolic blood pressure, mm Hg 113 (100‐129) 112 (100‐126) 115 (100‐133) 0.13
NYHA (II/III/IV) 92 (29%)/193 (60%)/36 (11%) 59 (28%)/129 (61%)/23 (11%) 33 (30%)/64 (58%)/13 (12%) 0.88
LVEF (%) 34±12 34±13 34±12 0.97
Orthopnea (no/\<20°/20°‐30°/\>30°) 103 (32%)/118 (37)/76 (24%)/23 (7%) 69 (33%)/75 (36%)/52 (23%)/14 (7%) 34 (31%)/43 (39%)/24 (22%)/9 (8%) 0.85
Edema (no/ankle/\<1/2 lower leg/\>1/2 lower leg) 209 (65%)/51 (16%)/31 (10%)/28 (9%) 149 (71%)/31 (15%)/18 (8%)/12 (6%) 60 (55%)/20 (18%)/13 (12%)/16 (15%) 0.01
Rales (no/basal/\<1/3 lung/\>1/3 lung) 176 (55%)/98 (31%)/40 (12%)/6 (2%) 118 (56%)/61 (29%)/26 (12%)/5 (2%) 88 (53%)/37 (34%)/14 (13%)/1 (1%) 0.68
Jugular venous pressure (normal/\>4 cm H~2~O/positive hepatojugular reflux/congested) 112 (36%)/88 (28%)/64 (20%)/49 (15%) 118 (56%)/61 (29%)/26 (12%)/5 (2%) 88 (53%)/37 (34%)/14 (13%)/1 (1%) 0.47
NT‐proBNP, ng/L 3920 (1773‐7068) 3645 (1709‐6799) 4638 (2078‐7538) 0.07
Potassium, mmol/L 4.2±0.5 4.2±0.5 4.2±0.5 0.91
Hemoglobin level, g/L 130±18 131±18 127±18 0.07
Serum creatinine, μmol/L 104 (85‐136) 104 (85‐139) 105 (83‐135) 0.82
eGFR, mL/min per 1.73 m^2^ 55±20 55±21 54±19 0.47
QRS width, ms 114 (97‐136) 114 (96‐136) 114 (99‐136) 0.82
Data are given as numbers and percentages, mean±standard deviation, or median (interquartile range). eGFR indicates estimated glomerular filtration rate; LVEF, left ventricular ejection fraction; NT‐proBNP, N‐terminal‐pro‐B‐type natriuretic peptide; NYHA, New York Heart Association.
Univariate Association Between ∆HR and Outcomes {#jah31684-sec-0011}
-----------------------------------------------
There were 100 (31%) patients who experienced the primary endpoint of HF hospitalization or death. There were 61 deaths (19%), and 187 (58%) patients experienced the endpoint of death or any hospitalization. When expressed as a continuous variable, a higher ∆HR was associated with a lower risk of death or HF hospitalization (hazard ratio 0.95 \[95% CI 0.92‐0.99\] per 1 bpm increase, *P*=0.01) and death (hazard ratio 0.95 \[95% CI 0.90‐1.00\] per 1 bpm increase, *P*=0.05). There was no significant association between ∆HR and all‐cause hospitalization or death (hazard ratio 0.98 \[95% CI, 0.96‐1.01\] per 1 bpm increase, *P*=0.25). The optimal threshold for ∆HR to identify subjects experiencing death or HF hospitalization was a ΔHR ≤3 bpm with only marginally less discriminative value for both ≤2 and ≤4 bpm (data not shown). As shown in Figure [1](#jah31684-fig-0001){ref-type="fig"}, patients with ∆HR ≤3 bpm had significantly worse HF hospitalization‐free survival, survival, and hospitalization‐free survival compared to those with ∆HR \>3 bpm. Thus, in the following, patients with ΔHR ≤3 bpm and those with ΔHR \>3 bpm are compared for descriptive purposes.
{#jah31684-fig-0001}
Baseline Characteristics of Patients With ΔHR ≤3 bpm Vs ΔHR \>3 bpm {#jah31684-sec-0012}
-------------------------------------------------------------------
As shown in Tables [1](#jah31684-tbl-0001){ref-type="table-wrap"} and [2](#jah31684-tbl-0002){ref-type="table-wrap"}, patients with ∆ HR ≤3 bpm were more likely to have diabetes, more often had peripheral edema, and were less likely to be on digoxin than those with ∆HR \>3 bpm. Otherwise, there were no significant differences in baseline characteristics and HF medication.
######
Medication at Baseline in the Entire Study Population and Patients With ∆HR \>3 bpm vs ≤3 bpm
All (n=321) ∆HR \>3 bpm (n=211) ∆HR ≤3 bpm (n=110) *P* Value
------------------------------------------------------------------ ---------------- --------------------- -------------------- -----------
β‐Blocker 261 (81%) 171 (81%) 90 (82%) 0.87
Target dose, median (IQR)[a](#jah31684-note-0005){ref-type="fn"} 25 (6.25‐37.5) 25 (6.25‐50) 12.5 (2.3‐37.5) 0.26
ACE inhibitor or ARB 303 (94%) 199 (94%) 104 (95%) 0.93
Target dose, median (IQR)[a](#jah31684-note-0005){ref-type="fn"} 50 (25‐67) 50 (25‐67) 50 (25‐67) 1.00
Mineralocortcoid receptor blocker 135 (42%) 92 (44%) 43 (39%) 0.44
Dose 0 (0‐25) 0 (0‐25) 0 (0‐25) 0.12
Loop diuretic 298 (93%) 197 (93%) 101 (92%) 0.61
dose, median (IQR)[b](#jah31684-note-0006){ref-type="fn"} 40 (20‐80) 40 (20‐80) 40 (20‐80) 0.73
Digoxin 23 (7%) 20 (10%) 3 (3%) 0.03
ACE inhibitor or ARB indicates angiotensin‐converting enzyme inhibitor or angiotensin receptor blocker.
Indicates percentage of target dose patients were receiving.
A dose of 10 mg of torasemide is equivalent to 40 mg of furosemide.
Multivariate Association Between ∆HR and Outcomes {#jah31684-sec-0013}
-------------------------------------------------
In Tables [3](#jah31684-tbl-0003){ref-type="table-wrap"} and [4](#jah31684-tbl-0004){ref-type="table-wrap"}, univariate and multivariate predictors of HF hospitalization‐free survival and survival are shown with models with ΔHR as a dichotomized variable (ΔHR ≤3 bpm vs \>3 bpm). As shown in Table [3](#jah31684-tbl-0003){ref-type="table-wrap"}, ∆HR ≤3 bpm was an independent predictor of HF hospitalization or death along with ischemic HF etiology, lower eGFR, presence and extent of rales at baseline, and no β‐blocker use at baseline. When used as a continuous variable, ΔHR (HR 0.95 \[95% CI, 0.91‐0.99\] per bpm increase; *P*=0.02) was also an independent predictor of HF hospitalization or death along with the same covariates as in the model with ΔHR as a dichotomized variable (data not shown). Adjustment of the multivariate models for sex, body mass index, and diabetes did not significantly change the results (data not shown).
######
Multivariate Predictors of HF Hospitalization or Death in a Model With ΔHR as a Dichotomized Variable
Univariate Multivariate
-------------------------------------------------- ------------------- -------------- ------------------ -------
ΔHR ≤3 bpm 1.79 (1.19‐2.75) 0.004 1.75 (1.18‐2.61) 0.004
Age, per year 1.06 (1.03‐1.08) 0.002
Ischemic HF etiology 3.28 (2.00‐6.62) 0.001 2.75 (1.69‐5.77) 0.001
Log~10~ NT‐proBNP 2.94 (1.81‐4.78) 0.001
Heart rate, per bpm 1.02 (0.999‐1.03) 0.06
eGFR, per mL/(min·1.73 m^2^) 0.98 (0.96‐0.99) 0.001 0.98 (0.96‐0.99) 0.001
Hemoglobin, per g/dL 0.98 (0.97‐0.99) 0.001
Hypertension 1.56 (1.03‐2.68) 0.05
COPD 1.62 (1.01‐2.42) 0.03
PAOD 1.90 (1.19‐2.92) 0.002
NYHA class 1.61 (1.17‐2.24) 0.004
Edema[a](#jah31684-note-0008){ref-type="fn"} 1.33 (1.12‐1.56) 0.002
Orthopnea[a](#jah31684-note-0008){ref-type="fn"} 1.49 (1.17‐2.00) 0.001
Rales[a](#jah31684-note-0008){ref-type="fn"} 0.60 (0.40‐0.93) 0.01 1.36 (1.06‐1.75) 0.01
β‐Blocker use at baseline 1.45 (0.99‐2.21) 0.06 0.63 (0.40‐1.00) 0.04
Spironolactone use at baseline 0.41 (0.05‐1.04) 0.09
Digoxin use at baseline 1.33 (1.12‐1.56) 0.002
95% CI indicates 95% confidence interval; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; HF, heart failure; HR, hazard ratio; LVEF, left ventricular ejection fraction; NT‐proBNP, N‐terminal‐pro‐B‐type natriuretic peptide; NYHA, New York Heart Association; PAOD, peripheral arterial obstructive disease.
Semiquantitative 4‐point scale (hazard ratio for increase by 1 class).
######
Multivariate Predictors of Death in a Model With ΔHR as a Dichotomized Variable
Univariate Multivariate
---------------------------------------------- ------------------ -------------- ------------------- -------
ΔHR ≤3 bpm 2.09 (1.23‐3.46) 0.002 2.01 (1.16‐3.47) 0.003
Age, per year 1.04 (1.01‐1.08) 0.02
Ischemic HF etiology 4.16 (1.89‐9.15) \<0.001 3.78 (1.84‐11.75) 0.001
Log~10~ NT‐proBNP 3.33 (1.82‐6.08) \<0.001
eGFR, per mL/(min·1.73 m^2^) 0.98 (0.97‐0.99) 0.004
Hemoglobin, per g/dL 0.98 (0.97‐0.99) 0.002
NYHA class 1.74 (1.16‐2.62) 0.008
Edema[a](#jah31684-note-0010){ref-type="fn"} 1.33 (1.06‐1.67) 0.01
Rales[a](#jah31684-note-0010){ref-type="fn"} 1.93 (1.47‐2.52) \<0.001 2.05 (1.50‐2.75) 0.003
PAOD 2.19 (1.29‐3.73) 0.004
95% CI indicates 95% confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio; LVEF, left ventricular ejection fraction; NT‐proBNP, N‐terminal‐pro‐B‐type natriuretic peptide; NYHA, New York Heart Association; PAOD, peripheral arterial obstructive disease.
Semiquantitative 4‐point scale (hazard ratio for increase by 1 class).
As shown in Table [4](#jah31684-tbl-0004){ref-type="table-wrap"}, ∆HR ≤3 bpm was an independent predictor of death along with ischemic HF etiology and presence and extent of rales. Patients with ΔHR ≤3 bpm had a more than doubled risk of death compared to those with ΔHR \>3 bpm (HR 2.01 \[95% CI, 1.16‐3.47\] *P*=0.003). Adjustment of the multivariate models for sex, body mass index, and diabetes did not significantly change the results (data not shown). When expressed as a continuous variable, ∆HR failed to remain in the model as an independent predictor of death (HR 0.95 \[95% CI, 0.90‐1.00\] per 1 bpm increase; *P*=0.05). In this model, ischemic HF etiology, lower eGFR, and the presence of rales were independent predictors of death (data not shown).
In Figure [2](#jah31684-fig-0002){ref-type="fig"}, the prognostic value of the 5‐point score built from the multivariate model including ΔHR is shown. Patients with a score of 3 points or more had a more than 8‐fold risk of HF hospitalization or death and death compared to those with 0 or 1 point.
{#jah31684-fig-0002}
There were no significant interactions between allocation to treatment strategy (NT‐proBNP--guided vs symptom‐guided), resting heart rate, baseline β‐blocker use, or diabetes with the prognostic impact of ∆HR on outcomes. There was an interaction between age if expressed as a dichotomized variable (\<75 vs ≥75 years) and the prognostic impact of ∆HR on HF hospitalization‐free survival (*P*=0.07 for ∆HR as a continuous variable and *P*=0.07 for ∆HR as a dichotomized variable) and survival (*P*=0.04 for ∆HR as a continuous variable and *P*=0.09 for ∆HR as a dichotomized variable) in that low ∆HR predicted poor outcome in younger but not in older patients. There was, however, no interaction between age expressed as a continuous variable and the prognostic impact of ∆HR. There was also an interaction between the LVEF stratum (LVEF \<45% vs ≥45%) and the association of ΔHR (both as continuous or categorical variables) with the endpoint of death or HF hospitalization (*P*=0.09 and 0.07, respectively) in that the association between lower ΔHR and worse outcomes (death or HF hospitalization) was somewhat stronger for patients with preserved LVEF.
Changes in HR Over Time and Impact on HF‐Hospitalization‐Free Survival {#jah31684-sec-0014}
----------------------------------------------------------------------
There were 229 patients alive at 6 months and without HF hospitalization during the first 6 months who had available data on ∆HR at baseline and month 6. The mean change in ∆HR from baseline at month 6 was 0±9 bpm (*P*=0.57 for overall comparison baseline vs month 6). There were 107 patients with a decrease in ∆HR, 27 patients with unchanged ∆HR, and 95 patients with an increase in ∆HR. A more positive change in ∆HR from baseline to month 6 was not significantly associated with a lower risk of HF hospitalization or death (hazard ratio 0.96 \[95% CI, 0.92‐1.002\] per 1 bpm increase; *P*=0.07), but patients with an increase in ∆HR from baseline to month 6 by \<2 bpm (optimal cutoff; n=138) had a significantly higher risk of HF hospitalization or death than those with an increase in ∆HR from baseline to month 6 by ≥2 bpm (n=91; hazard ratio 2.13 \[95% CI 1.12‐5.00\]; *P*=0.027; Figure [3](#jah31684-fig-0003){ref-type="fig"}).
{#jah31684-fig-0003}
Patients with an increase in ∆HR from baseline to month 6 by \<2 bpm were characterized by a higher ∆HR (7±6 vs 2±5 bpm; *P*\<0.001) and a lower eGFR (55±21 vs 60±18 mL/min per 1.73 m^2^; *P*=0.03) at baseline compared to those with an increase in ∆HR from baseline to month 6 by ≥2 bpm. Otherwise, there were no significant differences in baseline characteristics between groups including medication. Patients with an increase in ∆HR from baseline to month 6 by \<2 bpm were less likely to be prescribed a higher dose of spironolactone or to be started on spironolactone at baseline as part of the study protocol (12/138 vs 17/91; *P*=0.03), and the same tended to be the case at the month 1 visit (9/138 vs 13/91; *P*=0.05). Otherwise, there were no significant changes in the titration of HF medication from baseline to month 3 (data not shown). Given the possible impact of congestion on ΔHR, we looked at the association between changes in edema and changes in ΔHR. However, the proportion of patients with a reduction in edema severity did not differ between patients with a change in ∆HR from baseline to month 6 by \<2 bpm vs ≥2 bpm (reduction in edema 31 vs 23, unchanged edema 98 vs 57, worsening of edema: 9 vs 9; *P*=0.47).
Discussion {#jah31684-sec-0015}
==========
We showed that ΔHR provided prognostic information in patients with chronic HF, particularly in younger patients. The association between ∆HR and a HF hospitalization‐free survival remained statistically significant in the multivariable analysis. We also showed that ΔHR is modifiable and that changes in ΔHR over time also predicted outcomes. These findings may imply that this easy‐to‐perform and easy‐to‐repeat test may be a novel prognostic marker in patients with chronic HF with potential clinical applicability.
∆HR is a simple test that can be performed at the doctor\'s office within a short time, and its concept as a tool to assess the autonomic response to a physiological stimulus is plausible and biologically intuitive. We must acknowledge that the exact pathophysiological correlate of ∆HR is unknown, and given the methods used in our study, it is likely that we assessed neither the maximum baroreflex‐mediated rise in HR nor the nadir of the relative bradycardia thereafter. Rather, our measurement likely reflects an average HR throughout this period of HR dynamics after getting up or even the more or less stable HR that is established after \~2 minutes at least in healthy subjects.[10](#jah31684-bib-0010){ref-type="ref"} Still, we assume that ΔHR as assessed in the present study reflects the integrity or defect, respectively, of the mechanisms responsible for an increased sympathetic outflow to the heart after a person gets up.
We have not demonstrated how well ΔHR reflects established measures of autonomic tone in HF, including HR variability,[4](#jah31684-bib-0004){ref-type="ref"} HR recovery,[3](#jah31684-bib-0003){ref-type="ref"} norepinephrine spillover,[2](#jah31684-bib-0002){ref-type="ref"} baroreflex sensitivity, or muscle sympathetic nerve activity.[8](#jah31684-bib-0008){ref-type="ref"} Further studies will be required to better define the pathophysiological correlates of ΔHR. It might, for example, be considered that ΔHR depends on the presence and extent of congestion because congestion could limit the venous capacitance on standing and thereby attenuate the stimulus for the rise in HR with standing. Indeed, patients with lower ΔHR had more severe edema at baseline. However, we were unable to demonstrate an association between changes in edema severity and changes in ΔHR over time. This may, however, be due to the fact that changes in ΔHR were assessed in only a subgroup of patients surviving until month 6 without event, and thus, the sickest patients were excluded from this analysis. There was an association between changes in ΔHR over time and treatment with spironolactone, which may be regarded as indicative that neurohumoral antagonism and/or reduction of congestion had a favorable impact on ΔHR.
We have shown that the prognostic value of ΔHR is independent of established markers of HF severity, in particular eGFR.[13](#jah31684-bib-0013){ref-type="ref"} Furthermore, the subgroup analysis suggested that ΔHR is modifiable and that its changes over time are also associated with prognosis. Thus, these observations raise the possibility that ΔHR could be used as a biomarker that is measured serially to assess the effect of treatment. This seems to be particularly promising in patients younger than 75 years. Still, the hypothesis‐generating nature of our study needs to be emphasized, and the findings need confirmation in other cohorts. There was also an interaction between the LVEF stratum and the ability of ΔHR to predict outcomes in that the association between ΔHR and death or HF hospitalization was somewhat stronger in patients with preserved LVEF. Cautious interpretation of this finding is required, however. The group of patients with preserved LVEF in TIME‐CHF was small, and this was particularly true for this post‐hoc analysis for which patients with atrial fibrillation were not eligible.
The spectrum of ΔHR values was relatively narrow, and even in patients with good prognosis, the rise in HR on standing was small. This is not a surprising finding in this population of elderly patients with advanced HF, however. The original study by Ewing et al[10](#jah31684-bib-0010){ref-type="ref"} had shown that the magnitude of HR response to standing depends on age and disease status (diabetes, neuropathy). One might argue that this relatively narrow range of ΔHR in HF may limit the applicability of the parameter in practice. Thus, the clinically most meaningful ΔHR cutoff will have to be defined. On the other hand, we were able to show that not only absolute ΔHR cutoff but also individual changes in ΔHR are important.
Our study has some limitations in addition to those already discussed. First, the number of patients in this post‐hoc analysis was limited. Thus, our results will need confirmation in a larger population of HF patients, and it will be important to investigate whether treatment responses depend on ΔHR and how patients with low ΔHR must be treated to improve their outcomes. Second, we studied a population with a high proportion of patients on β‐blocker therapy, and the effect of β‐blocker therapy on ΔHR is unknown. An early pathophysiological study has revealed that the above‐described HR response to standing is under vagal control.[10](#jah31684-bib-0010){ref-type="ref"} However, adjustment for baseline β‐blocker dosage did not alter the results (data not shown), and there was no interaction between baseline β‐blocker use and the association between ΔHR and outcomes. Importantly, we were able to demonstrate that ΔHR is a prognostic predictor in a population treated with β‐blockers in a high proportion of patients, which represents the situation in real life. Third, ΔHR assessment by manual pulse palpation or an automated blood pressure monitor may be less accurate than continuous ECG monitoring. On the other hand, we showed that ΔHR as a true bedside test has the potential to predict outcomes. Still, the optimal method of ΔHR assessment will have to be defined. Finally, we have not performed a formal assessment of the reproducibility of ΔHR, which will be required before the parameter can be used in clinical practice.
Conclusions {#jah31684-sec-0016}
===========
ΔHR as a simple clinical parameter and presumed marker of autonomic function provides independent prognostic information in patients with chronic HF. ΔHR is modifiable, and changes in ΔHR over time are also predictive of outcomes. These findings imply that this easy‐to‐perform and easy‐to‐repeat test may be a novel prognostic marker in patients with chronic HF with potential clinical applicability.
Sources of Funding {#jah31684-sec-0017}
==================
This work was supported by the Horten Research Foundation (Lugano, Switzerland; 55% of the study\'s budget), as well as by smaller unrestricted grants from AstraZeneca Pharma, Novartis Pharma, Menarini Pharma, Pfizer Pharma, Servier, Roche Diagnostics, Roche Pharma, and Merck Pharma.
Disclosures {#jah31684-sec-0018}
===========
None.
Supporting information
======================
######
**Appendix S1.** TIME‐CHF Investigators.
######
Click here for additional data file.
[^1]: An accompanying Appendix S1, which lists the TIME‐CHF Investigators, is available at <http://jaha.ahajournals.org/content/5/8/e003524/DC1/embed/inline-supplementary-material-1.pdf>
[^2]: Dr Maeder and Dr Zurek contributed equally to this work and should be considered first authors.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .389}
{#sp2 .390}
{#sp3 .391}
{#sp4 .392}
{#sp5 .393}
{#sp6 .394}
{#sp7 .395}
{#sp8 .396}
{#sp9 .397}
{#sp10 .398}
{#sp11 .399}
{#sp12 .400}
{#sp13 .401}
{#sp14 .402}
{#sp15 .403}
{#sp16 .404}
{#sp17 .405}
{#sp18 .406}
{#sp19 .407}
| {
"pile_set_name": "PubMed Central"
} |
1. Cachexia as an Energy-Wasting Syndrome {#sec1}
=========================================
Cachexia, from the Greek: "*kakos*" and "*hexis*," meaning "bad condition," is a multiorgan syndrome associated with cancer and other systemic diseases such as sepsis and renal failure and characterized by at least 5% body weight loss due to muscle and adipose tissue wasting and inflammation \[[@B1]\]. Abnormalities associated with cachexia include alterations in carbohydrate, lipid, and protein metabolism \[[@B2]\]. Cancer cachexia has been characterized as a syndrome associated with loss of muscle with or without loss of fat mass. Other disorders associated with cachexia are anorexia, inflammation, insulin resistance, and increased muscle protein \[[@B2]\]. Another defining characteristic is that cachexia cannot be fully reversed by conventional nutritional support and leads to progressive functional impairment \[[@B3]\]. Thus, it can be concluded that cachexia is caused by an energy imbalance which is the result of both decreased food intake, due to marked anorexia, and increased energy expenditure caused by a highly hypermetabolic state. Blum et al. \[[@B4]\] in a recent meta-analysis of cancer cachexia conclude that current data support a modular concept of cancer cachexia with a variable combination of reduced nutritional intake and catabolic/hypermetabolic changes.
Cachexia occurs in the majority of terminal cancer patients and is responsible for the deaths of 22% of cancer patients \[[@B5]\]. Importantly, survival of cancer patient suffering from different types of neoplasias is dependent on the amount of weight loss \[[@B6]\]. Therefore, cachexia represents an important factor in the treatment of a cancer patient, affecting not only survival, but also the efficacy of anticancer treatment, quality of life, and medical costs. Thus there is a strong pressure to better understand the mechanisms that drive cachexia in order to offer cancer patients more effective care.
Here we will discuss the impact of nonmuscle tissues, only the ones that have a certain role; that is, kidney and lung do not seem to have a role in cancer cachexia. Indeed recent developments suggest that tissues/organs such as adipose (both brown and white), brain, liver, gut, and heart are directly involved in the cachectic process and may be responsible for muscle wasting. This suggests that cachexia is indeed a multiorgan syndrome ([Figure 1](#fig1){ref-type="fig"}).
2. Brain {#sec2}
========
Although a recent study involving 1853 cancer patients \[[@B7]\] did not find common genetic causes in appetite loss in cancer patients, cytokines, neuroendocrine changes, and tumour mediators are the main signals involved in appetite depression in cachexia. Additional factors contributing to the anorectic state are altered taste perception, therapy-induced side effects \[[@B8]\], depressed motor activity, possible mechanical interference on the gastrointestinal tract, and, of course, psychological factors \[[@B9]\]. Indeed, patients with cachexia often experience psychological distress as a result of the uncertainties of the disease, its diagnosis, its treatment, and its anticipated and final outcome \[[@B9]\]. This psychological state, which often involves depression, is bound to affect food intake. Both the limbic system and the brain stem participate in the regulation of appetite and energy balance. Thus, morphologically defined regions of the hypothalamus, the arcuate nucleus (ARC), the paraventricular nucleus (PVN), the dorsomedial nucleus (DMH), the ventromedial nucleus (VMH), the lateral hypothalamic area (LHA), and the perifornical area (PFA), appear to play a major role in the regulation of body weight. There are two primary neuron types within the ARC that integrate signals of nutritional status and influence energy homeostasis: a subpopulation of neurons in the medial ARC expresses the orexigenic neuropeptides (neuropeptide Y (NPY) and agouti-related peptide (AgRP)). More laterally there is a second subpopulation that inhibits food intake via the expression of cocaine- and amphetamine-regulated transcript (CART) and proopiomelanocortin (POMC), which is processed to melanocyte stimulating hormone (MSH). Specific neuropeptides are involved in the signalling of the neuronal circuits within these regions of the hypothalamus, for instance, corticotrophin-releasing hormone (CRH), thyrotropin-releasing hormone (TRH), NPY, brain-derived neurotrophic factor (BDNF), orexin, and melanin-concentrating hormone (MCH). The brain stem also plays an important role in the regulation of energy balance. Reciprocal connections are present, in an extensive way, between the hypothalamus and brain stem, particularly at the level of the*nucleus tractus solitarii* (NTS). This nucleus is in close anatomical proximity to the area postrema, a circumventricular organ that has an incomplete blood brain barrier. Like the ARC, the NTS is therefore located in an ideal place to respond to peripheral circulating signals but in addition also receives vagal afferent signals from the gastrointestinal tract and afferents from the glossopharyngeal nerves.
Therefore, brain mediators involved in the control of food intake, appetite, satiation, taste, and smell of food, are responsible for the anorexia of the cancer patient, making the brain a main organ responsible for one of the components of the altered energy balance in cancer patients \[[@B10]\] ([Figure 1](#fig1){ref-type="fig"}). Although anorexia represents a very important factor in the development of cachexia, it has to be pointed out that in many cases the use of total parenteral nutrition does not stop the loss of body weight \[[@B11]\]. In addition to anorexia, the hypothalamus via the melanocortin system may contribute to muscle wasting via neuronal output, as suggested by different animal studies \[[@B12], [@B13]\].
3. Gut {#sec3}
======
Gut-barrier dysfunction is a syndrome characterized by both breakdown and leakage of the gut epithelial barrier, leading to systemic inflammation due to the entry of bacterial cell wall components (endotoxin or lipopolysaccharide), or intact bacteria into the circulation. Gut-barrier dysfunction is often observed during the course of cancer cachexia \[[@B14]\] ([Figure 1](#fig1){ref-type="fig"}) and is partially connected to radiochemotherapy treatment. Additionally, tumour growth or macrophage infiltration at the level of the intestinal wall may affect gastrointestinal permeability, either locally or throughout the intestine via alterations in epithelial tight junctions \[[@B15]\]. From this point of view, tight junction proteins, such as ZO-1 and occludin, show decreased expression in tumour-rich regions of the intestine and colon in humans \[[@B16]\]. Decreases in tight junction proteins would increase permeability and allow passage of large molecules such as lipopolysaccharide (LPS) into the lymphatic circulation. Changes in mucin secretion and profiles in gastrointestinal carcinomas may become a source of inflammation in the course of cancer cachexia \[[@B17]\]. Indeed, gut-barrier dysfunction may lead to endotoxemia and, therefore, increased inflammation in cancer patients \[[@B14]\].
In addition to the gut-barrier dysfunction syndrome, recent studies support a role for gut microbiota in cancer cachexia. Indeed, decreased levels of bacteria, which have immunomodulating properties, are decreased during experimental cancer cachexia \[[@B18]\]. The existence of a gut-microbiota-skeletal muscle axis has been reported \[[@B19]\]; in fact, gut microbiota generates metabolites that can reach skeletal muscle and influence energy expenditure in the muscle cells \[[@B20]\].
Other aspects related with the gut may be beneficial for cachectic patients. Ghrelin is the first identified circulating hunger hormone that influences body weight regulation via a vagal pathway \[[@B21]\]; it is a gastric hormone initially identified in the rat stomach, in 1999, as an endogenous ligand for the growth hormone secretagogue receptor (GHSR) \[[@B21]\]. Under fasting conditions, the stomach, through endocrine cells located in the*antrum*, secretes ghrelin into the bloodstream. The hormone acts as a "hunger" mediator that signals the gastrointestinal fuel status from the periphery to the central nervous system in order to stimulate food intake and to adjust energy balance through decreasing energy expenditure \[[@B22]\]. Ghrelin also binds to the GHSR1a splice-variant that is enriched in the hypothalamus, as well as other brain regions. In the hypothalamus, at the level of the ARC, ghrelin contributes to enhanced food intake by activating orexigenic mediators such as NPY, gamma-aminoisobutyrate (GABA), and AgRP, and inhibiting anorexigenic mediators such as POMC, MSH, and CART \[[@B23]\]. The GHSR1a receptor is also present on vagal afferents \[[@B24]\] and, therefore, there is also strong evidence that ghrelin has a peripheral effect on satiety by affecting the mechanosensitivity of upper gastrointestinal vagal afferents, making them less sensitive to distension which can result in overeating. Ghrelin also has potent effects on fat storage. Ghrelin activates white adipocytes \[[@B25]\] while doing the opposite to brown adipocytes, therefore contributing to decreased energy expenditure \[[@B26]\]. These effects are related with the capacity of the hormone to stimulate growth hormone (GH) release from the anterior pituitary \[[@B27]\]. In addition, ghrelin increases insulin-like growth factor 1 (IGF-1) by stimulating its own receptor. These two factors are major signals being related with the regulation of energy homeostasis. The effects of ghrelin on GH and IGF-1 may be linked to the capacity of the hormone to prevent increases in protein degradation, through different components of the proteasome such as MuRF1 and MAFbx, at the level of skeletal muscle. This is particularly relevant to cancer cachexia since muscle wasting occurs mainly through activation of the ubiquitin-dependent proteolytic system. Ghrelin levels are elevated in cancer cachectic patients with neuroendocrine \[[@B28]\], gastric \[[@B29], [@B30]\] and lung \[[@B31]\] tumours. These elevated levels could represent a counterregulatory mechanism to fight anorexia associated with tumour growth. It is, in fact, an endocrine response to the so-called "ghrelin resistance" found in cancer patients. Indeed, this is, in part, the reason for the high doses of ghrelin used in clinical studies to counteract anorexia in cancer. In clinical practice the use of ghrelin and ghrelin agonists has led to promising results in cancer cachectic patients. Ghrelin treatment improves physical performance and muscle force indicating that the peptide is the best candidate for muscle wasting treatment either alone or in combination with other drugs or nutritional strategies \[[@B32]--[@B37]\]. Therefore, future research is needed to search for the optimal combination. Cancer cachexia is a multiorgan syndrome affecting not only skeletal muscle but also adipose tissues, heart, intestine, kidney, and liver. In fact, the final cause of death in cachectic cancer patients is, apart from the primary tumour itself, either sudden death (heart arrhythmias, hypoventilation), thromboembolic events (platelet aggregation), cardiorenal alterations (kidney dysfunction), or compromised immune function (immunosuppression). Ghrelin has a beneficial effect in all of the referred tissues. The only concern in treating cachectic cancer patients relates to the fact that ghrelin may contribute to tumour cell proliferation \[[@B38]\]. Indeed, ghrelin may increase the levels of growth factors, such as GH and IGF-1 that stimulate tumour growth. Additionally, ghrelin itself may have mitogenic potential. As far as we know, no*in vivo* data has examined the differences in tumour growth following ghrelin or GHS treatment. Long-term, large-scale clinical trials are required to determine whether ghrelin treatment promotes tumour growth.
In addition to the role of the gut ghrelin in the control of food intake, other mediators are bound to be involved in the anorexia associated with the cachexia syndrome. From this point of view, melanocortin-4 receptor \[[@B39]\] or prostaglandins \[[@B40], [@B41]\] may work independently or alongside ghrelin.
4. Liver {#sec4}
========
The liver plays a key role in regulating whole-body metabolism. Claude Bernard introduced the idea that the liver is the "glucostat" of the organism, by regulating glucose production and levels in mammals. Indeed, the liver can be regarded as the central node of supply and utilization of fuel by the tissues, the direction and flux of which are mediated by the endocrine system \[[@B42]\].
During catabolic conditions, opposing patterns of protein metabolism are observed between skeletal muscle and liver. While skeletal muscle is under negative nitrogen balance, mainly due to enhanced protein degradation, the liver exhibits important changes in the patterns of protein synthesis such as increased production of acute-phase proteins \[[@B42]\]. The enhanced muscle proteolysis drives a large release of amino acids from skeletal muscle, such as alanine and glutamine \[[@B43]\]. While glutamine is taken up by tumour cells to sustain both the energy and nitrogen demands of the growing mass, alanine is mainly channeled to the liver for both gluconeogenesis and protein synthesis \[[@B43]\]. Indeed, liver fractions from tumour-bearing animals show increased production of acute-phase proteins including C-reactive protein (CRP), serum amyloid A (SAA), *α*1-antitrypsin, fibrinogen, and complement factors B and C3 and a decrease in the synthesis of transferrin and albumin, leading to hypoalbuminemia \[[@B44]\]. An acute-phase response is also observed in cancer patients. In addition, in cancer patients with advanced cancer and during the fasting state, the total albumin synthesis rate is unchanged, compared with controls, despite much lower albumin concentrations \[[@B45]\]. Although the function of these proteins is far from being clear, it is known that CRP contributes to the activation of complement factors, enhancement of phagocytosis, and regulation of cell immunity; *α*1-acid glycoprotein inhibits platelet aggregation and phagocytosis and may be involved in spacing collagen fibres; haptoglobin binds to and clears haemoglobin from plasma; *α*1-antitrypsin and *α*2-macroglobulin regulate serine-proteases; and ceruloplasmin is probably involved in copper transport. Recently, a link between SAA, in synergy with interleukin-6 (IL-6), and activation of muscle proteolysis has been described \[[@B46]\] ([Figure 2](#fig2){ref-type="fig"}).
During cancer, the patient\'s inflammatory response ([Figure 1](#fig1){ref-type="fig"}) is linked to weight loss and poor performance status. Indeed, many inflammatory mediators are able to influence different metabolic pathways related to cachexia \[[@B2], [@B43]\]. The liver is an important contributor to the inflammation observed in cancer. Indeed, CRP seems to be a very important prognostic parameter \[[@B47]--[@B49]\].
The liver can also contribute to energy inefficiency. Indeed, Dumas et al. found that the efficiency of oxidative phosphorylation in liver mitochondria was decreased in a rat model of peritoneal carcinosis, suggesting that this may also contribute to hypermetabolism, elevated energy expenditure, in cancer-bearing states \[[@B50]\]. These alterations were associated with the content and fatty acid composition of cardiolipins \[[@B51]\]. Indeed, the phospholipid composition and especially cardiolipins are crucial for the mitochondrial energy metabolism. Indeed, cardiolipin is known to provide essential structural and functional support to several proteins involved in oxidative phosphorylation.
Moreover, it has been described that a higher number of CD68 immunoreactive macrophages have been found in liver cross sections of patients with pancreatic cancer and cachexia, suggesting that a crucial interaction between the tumor, peripheral blood mononuclear cells (PBMCs), and the liver, may play a central role in the development and regulation of cachexia \[[@B52]\].
5. Heart {#sec5}
========
Cancer is associated with severe heart alterations. Indeed, tumours implanted in experimental animals result in a decrease of the heart weight \[[@B53], [@B54]\], accompanied by functional cardiac changes, similar to those found in congestive heart failure. According to Schünemann et al. "cancer fatigue syndrome reflects clinically non-overt heart failure," clearly attributing a main role for heart abnormalities in the fatigue of cancer patient \[[@B55]\]. Tian et al. suggested that cardiac alterations in a mouse cancer cachexia model include marked fibrosis, disrupted myocardial ultrastructure, and altered composition of contractile proteins, such as troponin I and Myosin Heavy Chain-*α* (MHC-*α*) \[[@B56]\]. Similarly, Mühlfeld et al. using the well-established cachectic tumour rodent model Lewis lung carcinoma observed changes in heart innervation with the total number of axons in the left ventricle being reduced as a consequence of tumour burden \[[@B57]\]. This altered innervation was associated with a reduced expression of nerve growth factor \[[@B57]\]. The impairment of heart function observed in tumour-bearing animals seems to be specifically related to cardiac remodelling. Indeed, Tian et al. using the mouse C26 tumour model showed increased cardiac B-type natriuretic peptide (BNP) and c-fos expression together with decreased Peroxisome Proliferator Activator Receptor-*α* (PPAR-*α*) and its responsive gene Carnitine Palmitoyl Acyl Transferase-1*β* (CPT-1*β*) and a switch from "adult" isoforms (MHC-*α*, GLUT4) to "foetal" isoforms (MHC-*β*, GLUT1) \[[@B58]\]. The heart atrophy seems to be to some extent related with increased cardiac muscle proteolysis, since protein ubiquitination and expression of MuRF-1 and atrogin-1 are elevated \[[@B58]\]. However, Cosper and Leinwand suggest that the cardiac proteolysis is rather caused by increased autophagy \[[@B59]\], in a converse manner to what happens in skeletal muscle. Interestingly, inhibition of NF-*κ*B protects against tumour-induced cardiac atrophy, at least in experimental animals \[[@B60]\]. Cardiac atrophy in experimental cancer cachexia has recently been related with a high affinity activin type 2 receptor (ActRII) that mediates the signalling by a subset of Transforming Growth Factor-Beta (TGF-*β*) family ligands including myostatin, activin, and GDF11. Blocking pharmacologically this receptor reverses cancer-induced atrophy of the heart \[[@B61]\].
In addition to cardiac atrophy, Drott and Lundholm observed an increase in oxygen consumption, most likely related with the anaemia that very often is present in cancer patients, in the heart of an experimental cancer rodent model \[[@B62]\]. Important ultrastructural changes were also observed, such as an increase in the ratio of myofibrils/mitochondria and sarcomeric alterations, similar to those observed during cardiac failure. The increased oxygen consumption can to some extent be associated with increased energy expenditure, thus making the heart an additional organ involved in generating energy inefficiently ([Figure 1](#fig1){ref-type="fig"}). Indeed, heart rate seems to be elevated in cancer patients \[[@B63]\]. In fact, this parameter seems to be a very effective measure of cancer death risk. The mechanisms that could explain the association between heart rate and cancer mortality are unclear. Heart rate increase might be a marker of chronic stress and anxiety, which represent a natural consequence of the disease.
6. Adipose Tissues {#sec6}
==================
6.1. White Adipose Tissue (WAT) and Muscle Wasting {#sec6.1}
--------------------------------------------------
Previous studies have emphasized the cross talk between adipose tissue and skeletal muscle. Indeed, signals released from both tissues, that is, Tumour Necrosis Factor-*α* (TNF-*α*), IL-6, and interleukin-15 (IL-15), may participate in a reciprocal manner in the regulation of adipose and muscle tissue mass \[[@B64], [@B65]\]. Irisin, a protein produced both in skeletal muscle and adipose tissue in response to exercise, is able to stimulate browning of adipose tissue \[[@B66]\]. The release of irisin seems to be promoted by the transcriptional coactivator PPAR-*α* coactivator-1 alpha (PGC-1*α*) \[[@B67]\]. Very interestingly, the blockade of myostatin drives browning of adipose tissue through activation of the PGC-1*α*-irisin pathway \[[@B68]\]. Alterations in the balance of the signals could well be associated possibly with obesity, diabetes, or cachexia \[[@B64], [@B65]\]. Indeed, the adipocyte releases TNF-*α* and other cytokines that have a direct effect on muscle metabolism. Similarly, skeletal muscle releases IL-6, IL-15, and other signals that interfere with fat metabolism \[[@B64], [@B65]\]. Loss of fat mass is a key feature of cancer cachexia and the mechanism that drives this is multifactorial. On the one hand, lipolysis is activated in the adipocyte, which reduces its cellular volume \[[@B69]\]. Lipolysis may be favoured by a dramatic decrease in perilipin \[[@B70]\], a protein that acts as a protective coating from different lipases. The intense lipolysis is accompanied by changes in the expression of genes that regulate energy turnover, cytoskeleton, and extracellular matrix, suggesting high tissue remodelling. Altogether, this results in not only a net loss of the triglyceride depot, but also a change in the phenotype of the fat cell. On the other hand, fat depletion associated with cancer is linked with a decrease uptake of VLDL and chylomicron triacylglycerol due to a decrease in lipoprotein lipase (LPL) activity \[[@B2]\]. The increased fat removal is accompanied by a decrease in the rate of*de novo* lipogenesis in the adipocyte \[[@B2]\]. Interestingly, during cachexia a concomitant inhibition of adipogenesis takes place, possibly triggered by PPAR-*α* \[[@B71]\].
Both hormonal changes, insulin resistance and hyperglucagonemia, and release of proinflammatory cytokines seem to be responsible for the changes in adipocyte metabolism. In addition, both in experimental animals and humans, a zinc-*α*2 glycoprotein (ZAG) has been associated with the increased lipolytic rate \[[@B72]\]. ZAG, released by the tumour, seems to be the mediator able to activate the triacylglycerol lipase responsible for the increased lipolysis associated with cancer cachexia.
Das et al. found that genetic ablation of adipose triglyceride lipase (ATGL) in the mouse resulted in a prevention of increased lipolysis and, therefore, reduction in WAT, associated with tumour burden \[[@B73]\]. Interestingly, ablation of hormone-sensitive lipase (HSL) leads to similar but less marked effects \[[@B73]\]. Interestingly, the lipolytic ablation resulted in a preservation of skeletal muscle mass, suggesting that the breakdown of fat precedes that of skeletal muscle proteins and implicating that some signal(s) generated during the breakdown of adipocyte triacylglycerols may actually activate muscle proteolysis ([Figure 2](#fig2){ref-type="fig"}). The ablation of the mentioned lipase was also associated with a lack of activation of the main proteolytic system involved in muscle wasting during cancer, which is that of the ubiquitin-proteasome pathway \[[@B73]\]. In addition to the changes related to adipose tissue itself, infiltration of adipose tissue in skeletal muscle could contribute to wasting in this tissue. From this point of view Stephens et al. \[[@B74]\] have reported increased presence of intramyocellular lipid droplets in*rectus abdominis* muscle of cancer patients, which seems to be related to body weight loss in these patients.
6.2. Brown Adipose Tissue (BAT) and Energetic Inefficiency {#sec6.2}
----------------------------------------------------------
The metabolic energetic inefficiency, linked to hypermetabolism, found in the cancer patient, seems to be responsible, together with the reduced food intake, for the negative energy balance found in the patient \[[@B1]\]. Hypermetabolism seems to be related with inflammation \[[@B43]\]. Many futile cycles are responsible for hypermetabolism, including increased Cori Cycle activity between the liver and the tumour \[[@B75]\], liver glycolysis/gluconeogenesis \[[@B76]\], muscle protein synthesis/degradation \[[@B77]\], and adipose tissue triacylglycerol recycling \[[@B78]\]. An alternative mechanism contributing to hypermetabolism is the mitochondrial uncoupling proteins, originally described in BAT. Until quite recently, BAT has been considered as a thermogenic organ in rodents. By burning fat, BAT provides fatty acids which are further oxidized but, instead of serving for mitochondrial ATP synthesis, the energy associated with the oxidative process is released as heat due to the existence in the inner mitochondrial membrane of uncoupling proteins that permeabilize? the mitochondrial membrane to the electrochemical H^+^ gradient that drives ATP synthesis \[[@B79]\]. In 2009, Virtanen et al. showed that BAT was present in adult humans in the upper back, in the neck, between the collarbone and shoulder, and also along the spine, suggesting also a function for BAT in humans \[[@B80]\]. Recent data suggest the existence of two different types of brown adipose tissue cells: in addition to the "classical" uncoupling protein 1, UCP1 positive, derived from a myf-5 cellular lineage, so-called "beige" adipose cell exists with very low UCP1 expression and is derived from a non-myf-5 cellular lineage. Beige cells have their own gene pattern expression, different from both white and brown cells, and respond preferably to irisin \[[@B81]\]. Until now, no information is available concerning a possible role of these cells in cancer cachexia but certainly the topic deserves future attention.
Since BAT plays a key role in thermogenesis and energy balance, it may potentially contribute to the physiologic perturbations associated with cachexia ([Figure 1](#fig1){ref-type="fig"}). Several reports in experimental animals already point out a clear activation of BAT during cancer cachexia. Recently, Tsoli et al. have demonstrated increased BAT thermogenesis in cachectic tumour-bearing mice due to increased UCP1 or lipid oxidation (CPT-1*α* and peroxisomal bifunctional enzyme (PBE), one of the four enzymes of the peroxisomal beta-oxidation pathway) \[[@B82]\]. The changes observed seem to be related to an activation of STAT-3, possibly via IL-6. Unfortunately, no information is available on the role of brown fat in human cancer cachexia; therefore future research on this aspect is strongly encouraged.
In addition, there is another aspect to bear into consideration related to patients with cancer. Since the observation of the existence of BAT in adult humans, a population of cells, within WAT, which has the same characteristics as brown adipocytes, have been described \[[@B83], [@B84]\]. These cells are known as BRITE (brown into white) and seem to appear in WAT under certain conditions that seem to involve the COX-2 prostaglandin pathway. Both BAT and BRITE cells are more sensitive to insulin than WAT ones; therefore they consume glucose at a higher rate. Since both insulin and IGF-1 have been reported to fuel some tumours, having more or less BAT may affect the overall systemic insulin sensitivity and thereby have an indirect influence of tumour progression.
7. Conclusions and Future Directions {#sec7}
====================================
Since human skeletal muscle represents almost 50% of body weight, research on cancer wasting has for a long time been mainly devoted to this skeletal tissue. However, cancer cachexia is indeed a multiorgan syndrome affecting many types of cells, including adipose tissues, heart, liver, gastrointestinal tract, and brain. It has been recently reported that mediators released in nonmuscle tissues, during the cachexia syndrome, may actually be directly responsible for the activation of the metabolic alterations, such as increased protein degradation \[[@B85], [@B86]\], apoptosis \[[@B87], [@B88]\], and altered regeneration \[[@B89]\], leading to skeletal muscle wasting.
The implications of this are important since, on the one hand, the metabolic alterations affecting all cellular types may be very relevant to the understanding of the cachexia syndrome and, secondly, the development of new therapeutic approaches may benefit from this knowledge: for instance, it may be relevant to interfering with lipolysis or with acute-phase protein synthesis to block muscle proteolysis. Therefore, future studies on this field are needed and should concentrate on unrevealing the different mediators released by nonmuscle tissues that may influence muscle metabolism and, therefore, wasting. Furthermore, another important aspect that future research should contemplate is to establish the chronological involvement of the different organs/tissues in cancer cachexia.
This work was supported by a grant from the Ministerio de Ciencia y Tecnología (SAF-26091-2011).
Disclosure
==========
Each author has participated sufficiently, intellectually, or practically in the work to take public responsibility for the content of the paper, including the conception, design, and data interpretation. All authors have read and approved the final paper.
Conflict of Interests
=====================
All authors of this research have no conflict of interests related with employment, consultancies, stock ownership, honoraria, paid expert testimony, patent applications/registrations, and grants or other sources of funding.
{#fig1}
{#fig2}
[^1]: Academic Editor: Yona Keisari
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Patients with psoriasis have increased prevalence of classical coronary risk factors, including hypertension, hypercholesterolemia, and diabetes mellitus and increased risk of cardiovascular disease, e.g., myocardial infarction and stroke [@pone.0036342-Wakkee1], [@pone.0036342-Azfar1], [@pone.0036342-Menter1], [@pone.0036342-Gelfand1], [@pone.0036342-Mehta1], [@pone.0036342-Ahlehoff1], [@pone.0036342-Ahlehoff2]. Screening practices, treatment, and clinical control aimed at coronary risk factors and disease may be inadequate in patients with other chronic diseases [@pone.0036342-Wenger1], [@pone.0036342-MacLean1] and very recent results from a highly selected population of patients participating in phase III randomized trials with ustekinumab, a therapeutic anti-interleukin (IL)-12/IL-23p40 monoclonal antibody, demonstrated marked undertreatment and underdiagnosis of coronary risk factors in patients with psoriasis [@pone.0036342-Parsi1], [@pone.0036342-Kimball1]. Undertreatment of coronary risk factors may contribute to the increased risk of cardiovascular disease in patients with psoriasis. To examine the current practice regarding the pharmacological treatment of coronary risk factors in patients with psoriasis in a real-world setting we examined the pharmacological management of these risk factors in patients with severe psoriasis treated with biologic agents between 2007 and 2009.
Methods {#s2}
=======
Ethics {#s2a}
------
The study was approved by The Danish Data Protection Agency and the DERMBIO steering committee. Data at the individual case level was made available by the national registers in anonymized form. Registry studies do not require ethical approval in Denmark. The corresponding author had full access to the data and take responsibility for its integrity.
Data sources {#s2b}
------------
In Denmark all citizens are provided with a unique social security number at birth enabling linkage across registries. Medical history of patients with severe psoriasis selected for treatment with biologic agents was retrieved from a nationwide Danish registry (DERMBIO), a nationwide Danish database of patients with psoriasis treated with biologic agents. [@pone.0036342-Gniadecki1] Registration of patients with psoriasis treated with biologic agents is mandatory. From DERMBIO information registered on the date of subject database entry, psoriasis activity and severity index score, baseline history of hypertension, hypercholesterolemia and diabetes mellitus, and previous systemic antipsoriatic treatment and biologic treatment were recorded. Patients with psoriasis identified with hypertension, hypercholesterolemia, and diabetes mellitus were age- and sex matched with 4 individuals with corresponding coronary risk factors (hypertension, hypercholesterolemia, and diabetes mellitus) from the general population. All medications dispensed from pharmacies were obtained from the National Prescription Registry (the Danish Registry of Medicinal Product Statistics), wherein all dispensed prescriptions from Danish pharmacies has been recorded since 1995. Comorbidity, including ischemic heart disease (International classification of diseases 10th revision \[ICD-10\] codes: I20--I25), cerebrovascular disease (ICD-10: I60--I69), peripheral arterial disease (ICD-10: I70--I74), hypertension (ICD-10: I10), hypercholesterolemia (ICD-10: E78), and type II diabetes mellitus (ICD-10: I11) was obtained from the Danish National Patient Register in which all hospital admissions, diagnoses, and invasive procedures have been recorded since 1978. Individual-level linkage with these nationwide administrative registries of hospitalizations and dispensed prescriptions were used to assess treatment of coronary risk factors. Results were summarized as means with standard deviations for continuous data and counts and percentages for categorical data. Differences in pharmacological treatment were assessed by chi-square test. A two-sided p-value \<0.05 was considered statistically significant.
Outcome measures {#s2c}
----------------
Prescriptions claimed up to 6 months prior to study inclusion and 6 months after inclusion for drugs with therapeutic indications aimed at coronary risk factors and cardiovascular diseases including platelet inhibitors (anatomical therapeutic chemical classification \[ATC\] code: B01AC), beta-blockers (C07), angiotensin-converting enzyme inhibitors (ACE-Is)/angiotensin II receptor blockers (ARBs) (C09), calcium antagonist (C08), loop diuretics (C03C), thiazide diuretics (C03A), cholesterol-lowering drugs (C10A), and glucose-lowering drugs (A10).
Results {#s3}
=======
A total of 693 patients (mean age 46.1±12.7 years, 65.7% male) with severe psoriasis treated with biologic agents were identified and baseline characteristics of the study population are presented in [Table 1](#pone-0036342-t001){ref-type="table"}. Coronary risk factors, including hypertension, hypercholesterolemia, and diabetes mellitus were identified in 16.6%, 9.2%, and 6.7%, respectively
10.1371/journal.pone.0036342.t001
###### Baseline characteristics of patients with psoriasis.
{#pone-0036342-t001-1}
Patients with psoriasis n = 693
------------------------------------------------------------------------ ---------------------------------
Age/years (mean, standard deviation) ±12.7
Female (%) 238 (34.3)
Male (%) 455 (65.7)
PASI[\*](#nt101){ref-type="table-fn"} score (mean, standard deviation) 13.0±8.1
Psoriatic arthritis 85 (12.3)
Established atherosclerotic disease[\*\*](#nt102){ref-type="table-fn"} 6 (0.9)
**Coronary risk factors**
Hypertension 94/565 (16.6)
Hypercholesterolemia 52/565 (9.2)
Diabetes mellitus 38/565 (6.7)
**Previous systemic psoriasis treatment**
Methotrexate 469 (67.7)
Psoralen+UVA[\*\*\*](#nt103){ref-type="table-fn"} 125 (18.0)
Cyclosporine 161 (23.2)
Retinoids 116 (16.7)
Climate therapy 60 (8.7)
**Biologic agents**
Adalimumab 279 (40.3)
Alefacept 2 (0.3)
Efalizumab 39 (5.6)
Etanercept 264 (38.1)
Infliximab 64 (9.2)
Tocilizumab 2 (0.3)
Ustekinumab 42 (6.1)
Information not recorded 1 (0.1)
PASI; psoriasis activity and severity index.,
Established atherosclerosis; ischemic heart disease, cerebrovascular disease, peripheral vascular disease.
UVA; ultraviolet A light.
Patients with psoriasis and coronary risk factors received significantly less evidence based pharmacotherapy compared to age, sex, and coronary risk factor matched individuals from the general population ([Tables 2](#pone-0036342-t002){ref-type="table"} and [3](#pone-0036342-t003){ref-type="table"}). In psoriatic patients with hypertension 27.7% received no pharmacotherapy. Along this line, patients with dyslipidemia only received cholesterol-lowering medication in 55.8% of cases. Patients with diabetes mellitus received ACE-Is)/ARBs and cholesterol-lowering medication in 42.1% and 23.7% of cases, respectively. Pharmacologic treatment in patents with \>1 cardiovascular risk factor and secondary prophylaxis in the small subset of patents with manifest atherosclerotic disease showed a similar pattern of infrequent use of evidence-based cardiovascular pharmacotherapy ([Tables 4](#pone-0036342-t004){ref-type="table"} and [5](#pone-0036342-t005){ref-type="table"}). Assessment of pharmacological treatment 6 months after the index date indicated similar patterns of drug use, however beta blocker use generally decreased, e.g., 12.8% and 2.1% for hypertension at baseline and after 6 months, respectively ([Tables 3](#pone-0036342-t003){ref-type="table"}, [4](#pone-0036342-t004){ref-type="table"}, [5](#pone-0036342-t005){ref-type="table"}).
10.1371/journal.pone.0036342.t002
###### Medical management of coronary risk factors in patients with psoriasis and age- and sex matched controls.
{#pone-0036342-t002-2}
Patients with psoriasis Controls P value
---------------------------------------------- ------------------------- ------------- ---------
**Hypertension** **n = 94** **n = 376**
ACE-Is/ARBs[\*](#nt104){ref-type="table-fn"} 50 (53.2) 235 (62.5) 0.09
Beta-blockers 12 (12.8) 153 (40.7) \<0.001
Calcium antagonists 25 (26.6) 154 (41.0) 0.01
Thiazide diuretics 16 (17.0) 120 (31.9) 0.004
Any anti-hypertensive drug 68 (72.3) 327 (87.0) \<0.001
**Hypercholesterolemia** **n = 52** **n = 208**
Cholesterol-lowering drugs 29 (55.8) 136 (65.4) 0.19
**Diabetes mellitus** **n = 38** **n = 152**
ACE-Is/ARBs[\*](#nt104){ref-type="table-fn"} 16 (42.1) 94 (61.8) 0.03
Platelet inhibitors 9 (23.7) 44 (29.0) 0.52
Cholesterol-lowering drugs 9 (23.7) 87 (57.2) \<0.001
ACE-1; angiotensin-converting enzyme inhibitor, ARB; Angiotensin II receptor blocker.
10.1371/journal.pone.0036342.t003
###### Medical management of coronary risk factors.
{#pone-0036342-t003-3}
Pharmacotherapy at baseline (%) Pharmacotherapy at 6 months (%)
---------------------------------------------- --------------------------------- ---------------------------------
**Hypertension n = 94**
ACE-Is/ARBs[\*](#nt105){ref-type="table-fn"} 50 (53.2) 47 (50.0)
Beta-blockers 12 (12.8) 2 (2.1)
Calcium antagonists 25 (26.6) 21 (22.3)
Thiazide diuretics 16 (17.0) 14 (14.9)
Any anti-hypertensive drug 68 (72.3) 58 (61.1)
**Hypercholesterolemia n = 52**
Cholesterol-lowering drugs 29 (55.8) 29 (55.8)
**Diabetes mellitus n = 38**
ACE-Is/ARBs[\*](#nt105){ref-type="table-fn"} 16 (42.1) 14 (36.8)
Platelet inhibitors 9 (23.7) 9 (23.7)
Cholesterol-lowering drugs 9 (23.7) 12 (31.6)
ACE-1; angiotensin-converting enzyme inhibitor, ARB; Angiotensin II receptor blocker.
10.1371/journal.pone.0036342.t004
###### Cardiovascular prophylaxis in patients with \>1 coronary risk factor (n = 40).[\*](#nt106){ref-type="table-fn"}
{#pone-0036342-t004-4}
Pharmacotherapy at baseline (%) Pharmacotherapy at 6 months (%)
------------------------------------------------ --------------------------------- ---------------------------------
ACE-Is/ARBs[\*\*](#nt107){ref-type="table-fn"} 19 (47.5) 25 (62.5)
Beta-blockers 6 (15.0) 2 (5.0)
Platelet inhibitors 10 (25.0) 11 (25.5)
Cholesterol-lowering drugs 17 (42.5) 21 (52.5)
Coronary risk factor; hypertension, hypercholesterolemia, diabetes mellitus.
ACE-I; angiotensin-converting enzyme inhibitor, ARB; Angiotensin II receptor blocker.
10.1371/journal.pone.0036342.t005
###### Secondary cardiovascular prophylaxis in patients with established atherosclerotic disease (n = 6).[\*](#nt108){ref-type="table-fn"}
{#pone-0036342-t005-5}
Pharmacotherapy at baseline (%) Pharmacotherapy at 6 months (%)
------------------------------------------------ --------------------------------- ---------------------------------
ACE-Is/ARBs[\*\*](#nt109){ref-type="table-fn"} 3 (50.0) 2 (50.0)
Beta-blockers 2 (33.3) 0 (0.0)
Cholesterol-lowering drugs 4 (66.7) 4 (66.7)
Platelet inhibitors 5 (83.3) 3 (50.0)
Previous hospitalization with a diagnosis of ischemic heart disease, cerebrovascular disease, or peripheral vascular disease.
ACE-I; angiotensin-converting enzyme inhibitor, ARB; Angiotensin II receptor blocker.
Discussion {#s4}
==========
In a nationwide study of patients with psoriasis treated with biologic agents the proportion of patients with coronary risk factors treated with cardio-protective drugs was generally low and this finding was consistent for all examined coronary risk factors. Also, in the subset of patients with \>1 identified cardiovascular risk factor and in those with established atherosclerosis, including ischemic heart disease, cerebrovascular disease, and peripheral artery disease the proportion of patients receiving cardio-protective treatment was low. The results add to recent evidence indicating that patients with psoriasis are subject to inadequate pharmacological treatment of coronary risk factors and the observed undertreatment may contribute to the increased risk of adverse cardiovascular events seen in these patients.
In general, adequate treatment of coronary risk factors, including hypertension, hypercholesterolemia, and diabetes mellitus remains a challenge in clinical practice [@pone.0036342-Malik1], [@pone.0036342-Zafrir1], [@pone.0036342-Kones1]. It is well-established that psoriasis is associated with increased risk of cardiovascular disease, and increased prevalence of risk factors and shared autoimmune and inflammatory pathways between psoriasis and cardiovascular disease may contribute to this association [@pone.0036342-Wakkee1], [@pone.0036342-Azfar1], [@pone.0036342-Menter1], [@pone.0036342-Gelfand1], [@pone.0036342-Mehta1], [@pone.0036342-Ahlehoff1], [@pone.0036342-Ahlehoff2]. This has sparked an ongoing discussion of the need for increased awareness of this risk by treating physicians (general practitioners, dermatologists, cardiologists etc.) and for dedicated guidelines on coronary risk factor management in patients with psoriasis [@pone.0036342-Friedewald1]. Moreover, recent studies have examined the risk of cardiovascular events in patients with psoriasis according to the Framingham cardiovascular risk prediction score and documented that a high proportion of patients with psoriasis were at substantially increased risk and making them potential candidates for pharmacological cardiovascular primary prophylaxis [@pone.0036342-Kimball1], [@pone.0036342-Gisondi1]. In agreement with these findings, previous results from our group demonstrated that patients with severe psoriasis and/or psoriatic arthritis carried an absolute and relative risk of cardiovascular disease comparable to that of patients with diabetes mellitus [@pone.0036342-Ahlehoff1]. Indeed, recent Danish guidelines recommended that cardiovascular primary prophylaxis in patients with severe psoriasis and/or psoriatic arthritis should correspond to the treatment offered to patients with diabetes mellitus [@pone.0036342-Rapport1]. These Danish guidelines largely mirror recent guidelines of the European League Against Rheumatism on the management of cardiovascular risk in patients with inflammatory arthritis, including psoriatic arthritis [@pone.0036342-Peters1].
Regardless of augmented scientific focus on the association between psoriasis and increased cardiovascular risk, however, a very recent study of patients with psoriasis participating in phase III randomized trials with ustekinumab found that patients with coronary risk factors were subject to suboptimal pharmacological treatment [@pone.0036342-Kimball1]. The present results from a real-world clinical setting are in agreement with this observation and suggest that such undertreatment may pose a general problem in patients with psoriasis. While the low proportion of patients treated with beta-blockers was expected since beta-blockers may cause psoriatic exacerbations [@pone.0036342-Basavaraj1], the explanation for low use of other cardio-protective drugs is less apparent. Indeed, the reasons for lack of initiation and/or adherence to cardiovascular pharmacotherapy may be more dependent on inadequate knowledge and implementation of current guidelines by physicians, physician- or patient-driven worries about poly-pharmacy, reliance on non-pharmacological treatments, and other less-defined physician and patient preferences, than on evidence-based factors, e.g., treatment denied because of adverse drug reactions. Irrespective of the underlying reasons, it is an apparent paradox that psoriasis patients selected for treatment with expensive biologic agents and regularly evaluated by physicians (as is well-established in Denmark) seemingly elude treatment of coronary risk factors. Less than one percent of patients in the present study had established atherosclerotic disease and the results thus mainly reflect inadequate cardiovascular primary prophylaxis. On the other hand, our previous study of the prognosis following first-time myocardial infarction in Danish patients with psoriasis indicated that the post myocardial infarction pharmacological treatment (secondary prophylaxis) was adequate and non-differential between patients with and without psoriasis, respectively [@pone.0036342-Ahlehoff3]. However, that study was not powered to address differences in treatment related to psoriasis disease severity and did not specifically focus on the initiation and adherence to cardiovascular secondary prophylaxis.
In 2008 the US National Psoriasis Foundation recommended that all patients with psoriasis should be referred to a general practitioner for cardiovascular risk factor assessment, including family history, annual blood pressure measurement, biennial screening for obesity, and lipid and glucose screening every 2nd or 5th year for patients with or without classical coronary risk factors, respectively [@pone.0036342-Friedewald1], [@pone.0036342-Kimball2]. Of note, it has also recently been suggested that patients with psoriasis are inadequately screened for coronary risk factors [@pone.0036342-Parsi1]. The current report of pharmacological undertreatment of risk factors may therefore in fact underestimate the true magnitude of this problem. In addition, it has been suggested that current risk assessment tools, e.g., the Framingham risk score underestimate the true risk in patients with chronic inflammatory disease, e.g., systemic lupus erythematosus [@pone.0036342-Esdaile1]. This is further supported by numerous studies demonstrating increased carotid intima-media thickening, endothelial dysfunction, and coronary calcification in patients with psoriasis in the absence of established atherosclerotic disease [@pone.0036342-Balci1], [@pone.0036342-Ludwig1], [@pone.0036342-DeSimone1]. Thus the distinction between cardiovascular primary and secondary prophylaxis in these patients may be argued to be somewhat arbitrary.
The following limitations should be acknowledged. The study included patients with severe psoriasis treated with biologic agents and the results may not apply to other subsets of patients with psoriasis, e.g., patients treated with topical treatment or non-biologic systemic anti-inflammatory treatment. Information on non-pharmacological treatment of cardiovascular risk factors was unavailable. Also, information on blood pressure, cholesterol levels, blood glucose levels were absent and we were therefore unable to calculate formal cardiovascular risk scores, e.g., Framingham risk score. We had no information about the reasons for lack of cardiovascular pharmacotherapy.
Important strengths of the present study included the nationwide coverage, the direct comparison with age, sex, and risk factor matched controls, and the contemporary real-world clinical setting. The use of nationwide registries of hospitalization data and dispensed prescriptions from all pharmacies in Denmark where healthcare is readily accessible and essentially free of charge minimized selection bias related to sex, age, socioeconomic status, healthcare insurance status and job situation.
Conclusion {#s4a}
----------
This nationwide study of patients with severe psoriasis treated with biologic agents demonstrated that a high proportion of patients with identified coronary risk factors received little or no cardio-protective pharmacotherapy. The results indicate a need for increased awareness of cardiovascular risk and cardiovascular risk factor management in these patients. Randomized clinical trials of intensified cardiovascular risk factor management in these patients are urgently needed.
Data were extracted from the DERMBIO database by Mikkel Abildtoft, Zitelab Aps.
**Competing Interests:**OA has received an unrestricted grant from DERMBIO. DERMBIO database maintenance expenditure is covered by unrestricted equal grants from the manufacturers of biologics (Abbott, Janssen-Cilag, Merck-Serono \[2007--2009\], MSD, Pfizer). This does not alter the authors\' adherence to all the PLoS ONE policies on sharing data and materials.
**Funding:**The study was financially supported by unrestricted grants from the Axel Muusfeldts Foundation, DERMBIO, (Dermatologisk Database for Biologisk Behandling af Psoriasis) and the Danish Psoriasis Association. DERMBIO database maintenance expenditure is covered by unrestricted equal grants from the manufacturers of biologics (Abbott, Janssen-Cilag, Merck-Serono \[2007--2009\], MSD, Pfizer). The database is administered by an independent steering committee of specialists in dermatology representing the dermatological departments and private practices. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
[^1]: Conceived and designed the experiments: OA JL GG CTP PRH. Analyzed the data: OA JL GG LS CTP PRH. Contributed reagents/materials/analysis tools: LS CTP GG LI RG SL TD. Wrote the paper: OA PRH. Interpretation of data: OA LS GG JL SK LI SL RG TD CTP PRH. Critical revision of the manuscript: OA LS GG JL SK LI SL RG TD CTP PRH. Decision to submit: OA LS GG JL SK LI SL RG TD CTP PRH.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .50}
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION {#sec1-1}
============
Robot-assisted minimally invasive surgery (RAS-MIS) allows less experienced laparoscopic surgeons to perform higher quality operative procedures.\[[@ref1][@ref2]\] These techniques result in a shorter hospital stay and recovery time, reduced blood loss, and fewer complications compared to open surgery.\[[@ref3]\] However, some limitations of RAS include lack of tactile feedback, a fixed-port system, longer operative times, and prohibitive costs.\[[@ref4][@ref5]\] Detailed surveys that assessed the practice patterns and opinions of urologic surgeons have led to interesting insights about RAS.\[[@ref6][@ref7][@ref8]\] Therefore, the present study aimed to evaluate the perceived utility of RAS in Saudi Arabia.
METHODS/PROCEDURES {#sec1-2}
==================
An Internet-based, 59-item questionnaire was sent to 238 practicing urologists in the KSA who attended the 2015 Saudi Urological Association Meeting in the first quarter of 2015. The survey was E-mailed to urology physicians and trainees, and the questionnaire was available on the study website for 3 months to give respondents the opportunity to complete it online. Reminder E-mails were sent after 2 months to encourage those who missed the first invitation to complete the survey. Only one response per computer was allowed to avoid duplicate responses.
The survey comprised five sections inquiring about demographics and individual and institutional surgical practice patterns of MIS with a focus on RAS. Perceptions of MIS, in general, were also assessed, specifically those regarding RAS oncology procedures for prostate, bladder, and kidney treatment. The first section covered baseline characteristics, including geographical region, age, gender, current level of training, and years of practice. The second section assessed training, the number of MIS procedures performed in general, and the number of laparoscopic and RAS procedures performed specifically. The third section assessed institutional aspects, including staff and MIS or RAS fellowship programs. The fourth section evaluated the importance of robotic surgery training for career goals and the importance of developing a robotic surgery program. The fifth and final section was specialty-specific and assessed perceptions of RAS subtypes, including robot-assisted radical prostatectomy (RARP), robot-assisted radical cystectomy (RARC), and robot-assisted radical nephrectomy (RARN) or robot-assisted partial nephrectomy (RAPN).
Data were analyzed using a commercially available Statistical Package for the Social Science (IBM SPSS Statistics V22.0, Chicago, IL, USA). Descriptive data were presented as number of responses and percentages. Fisher\'s exact test was used for comparing discrete variables, and a two-tailed *P* \< 0.05 indicated significant differences among groups.
RESULTS {#sec1-3}
=======
Demographics and practice patterns {#sec2-1}
----------------------------------
The survey response rate was 40%, with 95 surveys completed. Most respondents (93.7%) were males with different levels of training, and \>45% had been in urology practice for at least 10 years. Nearly 53% of respondents were formally trained in laparoscopic surgery, 46% were formally trained in MIS beyond residency training, and only 21% were formally trained in RAS. More than half (66.8%) of respondents had either performed or assisted in laparoscopic surgery, with 43% performing 2 or more laparoscopic operations. Forty percent of respondents had used the surgical robot console in their training courses, and 38% had participated as console surgeons, including 11.6% who had performed at least two RAS operations as console surgeons. Of those who had only been exposed to RAS, 58% stated that they would perform the procedure in the future.
A total of 66.8% of respondents cited at least two surgeons performing laparoscopic surgery at their institutions. Fellowship-trained staff at respondents\' institutions totaled 79% in laparoscopic surgery and 23% in RAS. Nearly 37% of respondents\' institutions offered MIS fellowship programs; most of them were genitourinary or multiple programs that included at least two MIS fellowships. These programs were incorporated with an endourology (42%) or urologic oncology (16%) fellowship. With respect to RAS, 33.7% of respondents had a dedicated RAS support team in place, and 25% of respondents planned to hire new faculty to establish an RAS program at their institution \[[Table 1](#T1){ref-type="table"}\].
######
Demographic characteristics and clinical practice of respondents (*n*=95)
Impact of robot-assisted surgery subtypes {#sec2-2}
-----------------------------------------
Almost 24.2% of respondents felt that RAS training should be included to meet their career goals, and 68.4% would pursue a separate RAS fellowship if given the opportunity. A total of 66.3% would take additional courses to integrate RAS into their practice, with almost 70% choosing an MIS fellowship \[[Figure 1](#F1){ref-type="fig"}\]. Respondents were asked about the usefulness of RAS programs to their departments or clinical practice. Nearly 29.5% stated that RAS programs would strengthen the department academically and/or financially, whereas about 28% stated that RAS programs would strengthen only MIS at their departments. About 45.3% of respondents cited the need for an institutional robot system and identified administrative disinterest with lack of institutional support as the main obstacles to the development of a robotic program. Another 16% stated lack of academic evidence or operation room allocation as the main obstacle to the development of a robotic program \[[Figure 2](#F2){ref-type="fig"}\].
{#F1}
{#F2}
Robot-assisted radical prostatectomy {#sec2-3}
------------------------------------
Twenty-two percent of respondents had performed RARP, including 48.1% who performed \<50 RARP surgeries/year, 3.7% who performed 50--100 surgeries/year, and 9.3% who performed \>100 operations/year. Less than one-third (28.4%) had upgraded their operative skills to RARP. The more traditional open radical prostatectomy (ORP) was still being performed (65.3%) at most respondents\' institutions with 89% having performed \<10 surgeries/year and 11% having performed \>10 surgeries/year. However, many institutions had witnessed a decrease in ORP case volume per year. Laparoscopic radical prostatectomy (LRP) was still being performed in 20% of respondents\' institutions.
Nearly 34% of respondents cited RARP as the gold standard for prostatectomy; 21% believed it to be as good as open or laparoscopic procedures, 12.6% described it to be better, while 14.7% believed it was too early to form judgments regarding RARP \[[Figure 3](#F3){ref-type="fig"}\]. Nevertheless, most of the respondents (73.7%) would recommend RARP over ORP (11.6%) or LRP (14.7%) for themselves or their family \[[Figure 4](#F4){ref-type="fig"}\].
{#F3}
{#F4}
Robot-assisted radical cystectomy {#sec2-4}
---------------------------------
Only 3.2% of respondents had personally performed RARC, and this procedure had not been performed at all in 65% of their institutions. The more traditional open radical cystectomy (ORC) was performed by 36% of respondents and laparoscopic radical cystectomy (LRC) by 7.4%. Nearly 42% believed that 100--200 RARC procedures should be performed to be comfortable with the technique while most of them (51.4%) cited that \<50 cases would be sufficient. Only 23% of respondents declared RARC to be the gold standard for cystectomy; 40% believed RARC to be as good as ORC or LRC, 11.6% described it to be better than open or laparoscopic approaches, while 11.6% believed it too early to form judgments regarding RARC \[[Figure 3](#F3){ref-type="fig"}\]. Nevertheless, 50% of respondents would recommend RARC over either ORP (40%) or LRP (9.5%) for themselves or their family \[[Figure 4](#F4){ref-type="fig"}\].
After RARC and pelvic lymph node dissection, most respondents (53.7%) believed that urinary diversion should be performed either open or laparoscopically while 46.3% reported that urinary diversion should be performed robotically as well. Only 6% of respondents believed proximal lymph node dissection to be challenging using the robot while most of them (78%) were not sure. Similarly, 73% were not sure if there was an advantage to an extended pelvic lymph node dissection with the robotic da Vinci^®^ Surgical System (Intuitive Surgical Inc., Sunnyvale, CA, USA) while only 16.8% believed that it was advantageous.
Robot-assisted radical nephrectomy {#sec2-5}
----------------------------------
RARN and RAPN were performed by 20% and 23.2% of respondents, respectively. Of 56 respondents, 37.5% had performed \<50 surgeries/year while 9% had performed 50--100 operations/year. The more traditional open radical nephrectomy (ORN) and laparoscopic radical nephrectomy (LRN) were still being performed in 75.8% and 65.3% of respondents\' institutions, respectively, at a volume of \>10 cases/year for ORN (55%) and LRN (57%). Nearly one-third of respondents (31.6%) had upgraded their operative skills in ORN to LRN, whereas 76.8% and 63.2% felt that at least 50 procedures were needed to be comfortable with RARN and RAPN, respectively.
Forty-nine percent reported that the greatest advantage of RARN was ease of performance while 36% cited a quality-of-life advantage. Almost 76% of respondents believed that RARN would not prolong the warm ischemia time, but 40% felt that the console surgeon should rely more on laparoscopic assistance. Almost 24.2% reported that robotic arms would have difficulty reaching the bladder cuff during nephroureterectomy while 50% of respondents were not sure. Advantages of RARN or RAPN over laparoscopic techniques included benefits of 3D vision (26.3%), EndoWrist^®^ action (24.2%), and instrument mobility (17.9%). Twenty-five percent of respondents believed it was too early to judge RARN, 21% believed it to be as good as open or laparoscopic procedures, and only 16.8% cited RARN to be the gold standard of radical nephrectomy \[[Figure 3](#F3){ref-type="fig"}\]. Nevertheless, most respondents (56.8%) would recommend RARN over ORN (8.4%) or LRN (34.7%) for themselves or their family \[[Figure 4](#F4){ref-type="fig"}\].
Impact of location, level of training, and duration of practice {#sec2-6}
---------------------------------------------------------------
The region of Riyadh had significantly higher numbers of faculty and consultants (67%) than Makkah (58%), Jeddah (43%), Al-Madinah (50%), and other areas (15%, *P* = 0.027). A significantly larger number of faculty/consultant surgeons in Riyadh were trained in MIS than their colleagues elsewhere (65.9% vs. 31.5%, *P* = 0.001), but numbers were comparable to their colleagues in terms of formal RAS training (22% vs. 20.4%, *P* = 0.85). Despite the fact that both groups had been formally trained in laparoscopic surgery (61% vs. 46.3%, *P* = 0.21), faculty/consultant surgeons had performed more laparoscopic surgeries (97.5% vs. 77.7%, *P* \< 0.001) and used the robotic console more frequently during training (47.4% vs. 24.1%, *P* \< 0.001). Urologists with different training levels had similar perceptions of RAS subtypes, the importance of robotic surgery training for career goals, and the development of a robotic surgery program.
Significantly larger number of respondents in practice for 10 or more years reported having had training in laparoscopic surgery (56% vs. 42%, *P* = 0.02) and MIS beyond residency (69.8% vs. 27%, *P* \< 0.001) but had comparable RAS training with those practicing for \<10 years (23.3% vs. 19.2%, *P* = 0.80). Those in practice for \>10 years reported significantly greater use of surgical robots in training (61.2% vs. 30.8, *P* = 0.04). Both groups comparably stated that they would pursue a fellowship in RAS if given the opportunity (70% vs. 67%, *P* = 0.79) and that taking additional courses would be sufficient to incorporate RAS into their practice (70% vs. 44.3%, *P* = 0.52). Both groups were also comparable in the numbers and perception of RAS subtypes, importance of robotic surgery training for career goals, and development of a robotic surgery program.
DISCUSSION {#sec1-4}
==========
Currently, RAS is most frequently used in urological and gynecologic oncological surgery and has been successfully adopted in complex procedures involving the prostate, kidney, and urinary bladder.\[[@ref9][@ref10][@ref11]\] RAS has gained worldwide popularity as a significant adjunct to laparoscopy, adding to the armamentarium for MIS. The three-dimensional magnified vision, depth perception, EndoWrist^®^ technology with seven degrees of freedom, and precision with intuitive movement provided by RAS make intracorporeal dissection and suturing considerably easier.
However, the learning curve associated with MIS and laparoscopy has been identified as a training obstacle for postgraduate trainees and attending urologists.\[[@ref12]\] Shay *et al*. found that urologists were more likely to perform laparoscopy if they had been trained during their residency than if they had not.\[[@ref13]\] Therefore, it is advocated that optimal MIS training with a focus on RAS should take place in fellowships and postgraduate courses, which have the benefit of virtual reality simulators.\[[@ref14]\] Although there are ten da Vinci^®^ robots in the KSA, the low volume of robotic caseload negatively impacts resident\'s teaching and compromises their training.\[[@ref15]\] Currently, there are six robotic systems distributed among four major hospitals in Riyadh, two in the Eastern Province hospitals, and two in the Western Region hospitals. There are only nine indexed case reports from the KSA with no reports from any other Middle Eastern countries.
Approximately one-half of our respondents in the KSA were formally trained in laparoscopic surgery and MIS while only 21% were formally trained in RAS. However, most of those who were not exposed to RAS stated that they intended to perform the procedure in the future. Forty percent of our respondents had used a robot console during training courses, but 63% had not participated as a console surgeon. These figures are more promising than reports from a previous survey, wherein respondents in the Middle East and Asia were significantly less likely to have had formal training in RAS (11% vs. 40%) and use of robotic consoles in training courses (20% vs. 60%) than those in Europe and North America.
Yuh *et al*. surveyed 291 urologists to compare the status of urologic laparoscopy and RAS worldwide. Nearly 80% of respondents had performed MIS, and 64% had prior formal training.\[[@ref6]\] Duchene *et al*. found a favorable attitude toward robotics among postgraduate urologists, where 80% believed that RAS would increase the volume of potential procedures. Moreover, 45% believed that RAS would become the standard of care within the next decade.\[[@ref14]\] Similarly, 94% of gynecologic oncology fellows in the United States felt comfortable using the robot, and respondents confirmed that they would use this technology in their future practice.\[[@ref16]\] Respondents from western countries performed significantly more RAS, as expected.\[[@ref6]\] Based on these data, it could be argued that the rate of RAS adoption is growing as a worldwide trend.
Seventy-two percent of our respondents felt that RAS training should be required as a part of their career goals and believed that it would strengthen their departments both academically and financially. However, the absence of a robotic system together with administrative disinterest and lack of support were the most common deterrents for incorporating RAS. These findings support the results of Guru *et al*. where 78% of respondents similarly felt that RAS training should be required as a part of their career goals.\[[@ref7]\] Yuh *et al*. also cited the absence of a robotic system as an obstacle to the development of RAS programs in most institutions.\[[@ref6]\] Widespread adoption of robotic surgery is challenged by the high cost (\$1.5 to \$2.2 million) of purchasing and maintaining the da Vinci^®^ Surgical System. An annual maintenance cost of about \$138,000 has been reported, with average costs per case ranging from \$1500 to \$2000.\[[@ref16]\] Other challenges may include lack of operating room support and availability of skilled technical staff.
In the current study, RARP and RARN were performed by \<25% of respondents, whereas only 3.2% had performed RARC. Moreover, after performing RARC, most respondents believed that urinary diversion should be performed either through open or laparoscopic surgery. Less frequent performance of RARC over other robotic procedures showed that RARC remains a controversial means of bladder extirpation. While RARC has been successfully attempted at select centers, its precise long-term benefits are not well established. Of 297 worldwide surveyed urologists, only 10% have performed either RARC or RARN and 21% have performed RARP.\[[@ref7]\]
However, 70% of surgeons in the Guru *et al*.\'s study had upgraded their skills from RARP to RARC versus only 12.6% of our local respondents. RARP (9%), RARC (6%), and RARN (16%) rates in the Middle East and Asia were significantly lower than these RAS procedure rates (38%, 15%, and 32%, respectively) in Europe and North America. However, nearly one-third of our respondents had upgraded their ORN skills to LRN and RARN, which was comparable to the 30% reported by Guru *et al*.\[[@ref7]\] Only 6% of our respondents believed proximal lymph node dissection after RARC to be challenging using the robot while 78% were not sure. About 73% were not sure if there was an advantage to an extended pelvic lymph node dissection using the da Vinci^®^ system. These data were not in agreement with Yuh *et al*. where about half of the respondents believed that RAS proximal lymph node dissection with the da Vinci^®^ system was not challenging.\[[@ref6]\] These differences may be due to the small number of RARC procedures performed by our respondents. In agreement with the literature, our respondents felt that the greatest advantages of RARC and RARN were ease of performance and quality of life while cancer treatment was cited less often.\[[@ref17][@ref18]\] Most of our respondents believed that RARN would not prolong warm ischemia time as previously reported.\[[@ref19]\]
Limitations of our Internet-based survey may include selection bias and recall bias. Selection bias is possible by respondents who may have been more inclined to take the survey due to their interest in MIS, RAS, or new technology in general, which limits the generalizability of our results. Recall bias may have occurred with surgeons overestimating the number of procedures performed. Several responses were left blank on some surveys, but the vast majority of data were complete. Another limitation was a low response rate although an even lower response rate of 9.5% was reported by Duchene *et al*.\[[@ref14]\] Furthermore, the design of the present study did not allow for further comparative or subanalysis of data. This study does, however, address the literature gap regarding practice patterns of MIS and RAS in Saudi Arabia.
CONCLUSION {#sec1-5}
==========
Despite the fact that most of our respondents did not consider RAS to be the surgical gold standard for prostatectomy, cystectomy, and nephrectomy, most of them would recommend RAS for themselves or their family.
Financial support and sponsorship {#sec2-7}
---------------------------------
Nil.
Conflicts of interest {#sec2-8}
---------------------
There are no conflicts of interest.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
Proper sedation is an important component in the care of critically ill patients requiring mechanical ventilation. Deep sedation levels are associated with several negative outcomes, such as increased time on mechanical ventilation,^([@r1])^ delirium,^([@r2])^ memory disturbances,^([@r3])^ and higher mortality in the short^([@r4])^ and long term.^([@r5])^
The deleterious effects of deep sedation can be minimized by employing a strategy of sedation protocols that target lighter sedation levels^([@r6],[@r7])^ and the daily interruption of sedative infusion.^([@r8],[@r9])^ The effects of these strategies have been assessed in two systematic reviews in which the included study control groups consisted of patients who received \"usual\" care in relation to the sedation of patients on mechanical ventilation. The first systematic review that included observational and randomized studies showed that most of the studies suggested a reduction in the duration of mechanical ventilation and ICU and hospital length of stay.^([@r10])^ The second systematic review included only randomized studies and pooled their results into a meta-analysis, which indicated that there were reduced ICU and hospital length of stay and reduced mortality with the use of both sedation reduction strategies.^([@r11])^ Another meta-analysis also suggested that the two sedation minimization strategies were not associated with higher incidences of post-traumatic stress in the long term,^([@r12])^ which was a fear that had been raised when the first study on daily sedation interruption was published.^([@r13])^
Therefore, protocols targeting either a light sedation level or daily sedative infusion interruption should be adopted to reduce the deleterious effects of excessive sedation.^([@r14])^ However, the use of these strategies is still far from universal,^([@r15])^ and it is unclear whether one of the two is more effective than the other.
The objective of this study was to systematically review studies that compared a light target sedation protocol with daily sedation interruption.
METHODS
=======
Search strategy
---------------
Searches of the Medline (via PubMed), Scopus and Web of Science databases were performed. The studies were obtained by combining the following keywords: (\"sedation\" OR \"sedatives\") AND (\"critical care\" OR \"intensive care\" OR \"critically ill\") AND (\"daily interruption\"). Additional studies were sought based on the references of included studies and personal files. There was no language restriction. The searches were limited to randomized clinical studies performed on adults and published up to February 4, 2016. Titles and abstracts were assessed for eligibility. The full texts of potentially relevant articles were analyzed. The eligibility assessment was conducted by the authors, and disagreements were resolved by consensus. Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines were used as a guide.^([@r16])^ The systematic review was recorded in the PROSPERO database (CRD 42014014121). As the study is a literature review, there was no need for Ethics Committee approval.
Study selection
---------------
Studies that met the following criteria were included: those comparing a protocol with a predefined sedation scale target with daily sedative infusion interruption; and those assessing any of the following outcomes: mortality in intensive care, duration of mechanical ventilation, days free of mechanical ventilation in 28 days and ICU length of stay.
Data extraction
---------------
The authors extracted the following data independently using a specific form: country where the study was conducted, year of publication, study design, number of patients included in each study group, description of the sedation protocol and the manner in which daily sedation interruption was conducted, ICU and hospital mortality, duration of mechanical ventilation, days free of mechanical ventilation in 28 days, ICU and hospital length of ICU stay, delirium, accidental extubation rates and extubation failure (reintubation within 48 hours). Authors of included studies were contacted by e-mail to obtain information about missing data from the publications.
Evaluation of study quality
---------------------------
Study quality was assessed by the Cochrane risk of bias assessment tool for clinical studies. The risk of bias was assessed as \"low\", \"uncertain\" or \"high\" in the following areas: generation of random sequence, allocation concealment, blinding of participants and professionals, blinding of outcome assessors, incomplete outcomes, selective outcome reporting, and other sources of bias. Disagreements were resolved by consensus.
Outcomes
--------
The primary outcome was mortality in the ICU. Secondary outcomes were duration of mechanical ventilation, days free of mechanical ventilation in 28 days, hospital mortality, ICU and hospital length of stay, prevalence of delirium, and accidental extubation and extubation failure rates (reintubation within 48 hours after extubation).
Statistical analysis
--------------------
A random effects model was used due to the variability among studies regarding samples and how the interventions were applied. The differences between groups were expressed as odds ratios (OR) for categorical variables and as mean differences (MD) for continuous variables, both with 95% confidence intervals (95%CI). The reference group for the analysis was always \"sedation protocol.\" Heterogeneity was assessed using the I^2^ statistic and was classified as low (\< 25%), moderate (25 - 50%) or high (\> 50%). The analyses were performed using *R* software version 3.3.1, with *R* Studio version 0.99.902, and the meta package (version 4.4.0) developed by Guido Schwazer (<http://cran.rproject.org/web/packages/meta/meta.pdf>).
RESULTS
=======
Study characteristics
---------------------
A total of 279 references were identified by the search strategies, eight full-text articles were assessed for eligibility. In total, seven studies^([@r17]-[@r23])^ were included; one was excluded, as it did not report any of the outcomes of interest^([@r24])^ ([Figure 1](#f1){ref-type="fig"}).
Figure 1Study flowchart.
The characteristics of the included studies are described in [table 1](#t1){ref-type="table"}. In general, the studies were small, and only one was a multicenter study. The goal was light to moderate sedation in all studies. The Sedation-Agitation Scale (SAS), the Ramsay scale and the Richmond Agitation Sedation Scale (RASS) were used in three, three and two studies, respectively. Only one of the studies had a deeper sedation level as its lower target (Ramsay 5).^([@r19])^ Descriptions of the sedation protocols and daily sedation interruption can be found in [table 2](#t2){ref-type="table"}.
######
Study characteristics
Study Country Number of centers Number of patients (protocol/daily interruption) Sedation target
-------------------------------- -------------------------- ------------------- -------------------------------------------------- --------------------------
Mehta et al.^([@r17])^ Canada 1 33/32 SAS 3 - 4
de Wit et al.^([@r18])^ United States 1 38/36 RASS -2 - -3
Anifantaki et al.^([@r19])^ Greece 1 48/49 Ramsay 3 - 5
Strom et al.^([@r20])^ Denmark 1 70/70 Ramsay 3 - 4
Yiliaz et al.^([@r21])^ Turkey 1 25/25 Ramsay 3 - 4
Mehta et al.^([@r22])^ Canada and United States 16 209/214 SAS 3 - 4 or RASS -3 - 0
Nassar Junior e Park^([@r23])^ Brazil 1 30/30 SAS 3 - 4
SAS - Sedation Agitation Scale; RASS - Richmond Agitation Sedation Scale.
######
Sedation protocol and daily interruption performed in each study
Study Sedation protocol Daily sedation interruption
-------------------------------- ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Mehta et al.^([@r17])^ Midazolam and morphine (or fentanyl, if CrCl \< 10mL/min) reduced every 15 - 30 minutes if SAS 1-2. Boluses were administered if there was agitation, and sedative and analgesic doses were increased. SAS was reassessed every 1 - 2 hours The infusion of sedatives and opioids was maintained identically to the protocol, but sedatives and analgesics were turned off after 9 hours, and the patients were assessed for their ability to obey three out of four commands (open your eyes, follow the investigator with your eyes, shake hands and wiggle your toes). If the doctor felt that the patient needed to be sedated, sedation was reinitiated at half the dose. In this case, the protocol continued, targeting SAS 3-4. If it was decided that the patient would not receive any more sedatives, they were only resumed if the patient was at SAS 6 - 7
de Wit et al.^([@r18])^ Analgesia with morphine or fentanyl (if renal failure or hemodynamic instability) in bolus. If boluses were frequent, continuous infusion began. Sedation followed the same pattern, with the use of midazolam or lorazepam. Where there was a need for continuous infusion, lorazepam or propofol were used if there was renal or hepatic failure and lorazepam and midazolam if there was hemodynamic instability. The analgesics and sedatives of patients with RASS 1 or 2 points below the target were reduced by 25 - 50% every 4 hours. If the RASS was more than two points below the target, the drugs were discontinued The sedatives and opioids were turned off 48 hours after the beginning of mechanical ventilation. Patients were considered awake if they could follow three of four commands (open your eyes, follow the researcher, put out your tongue and shake hands). The resumption of sedatives was at the discretion of the investigators. Sedatives were restarted at half the dose if the patient was awake, agitated or had a change in vital signs (RR \> 35ipm; SaO~2~ \< 90%; HR \> 140bpm or change of 20% in either direction; SBP \> 180mmHg or \< 90mmHg). The team had to target RASS -2 to -3 and performed sedative infusion in the absence of the investigators
Anifantaki et al.^([@r19])^ Sedatives (midazolam or propofol) and opioids (remifentanil) were adjusted to maintain Ramsay 3-5. The adjustments were performed every 2 minutes until the target was reached. Sedative infusion was turned off after patient recruitment, but the remifentanil infusion was maintained at a rate of 0.05 - 0.25mg/hour. If the patient was agitated, presented respiratory distress, hemodynamic instability or neurological deterioration (e.g., increased ICP), sedatives and analgesics were reinitiated at half the previous dose
Strom et al.^([@r20])^ Analgesia with morphine. If discomfort was experienced, the team searched for reversible causes. If delirium was suspected, haloperidol was administered. If agitation was still present, propofol was initiated for 6 hours. After this period, the propofol was discontinued. If there was a need to start sedatives three times, the patient was sedated in the same manner as the daily awake group Morphine and propofol to maintain Ramsay 3 - 4, assessed every 2 - 3 hours. Sedation was stopped and awakening assessed daily. In this regard, the patient had to be able to complete three of four tasks: open his eyes, follow with his eyes, shake hands, put out his tongue. After awakening, the sedative was reinitiated at half dose to maintain Ramsay 3 - 4. After 48 hours, propofol was replaced with midazolam
Yiliaz et al.^([@r21])^ Fentanyl for pain control, with target of BPS ≤ 6 and midazolam for agitation control with a target of Ramsay 3 - 4. Additional sedatives (diazepam, propofol and dexmedetomidine) could be used if the Ramsay target was not reached Sedation interruption was employed at any time (without further details)
Mehta et al.^([@r22])^ Adjustment of opioid infusion and sedatives for achieving the target, as in the 2008 study Daily sedation interruption was employed. If the patient could follow three out of four commands, the infusion was kept off at the discretion of the doctor and nurse. If there was a need for sedation or agitation or discomfort, then the doses instituted were half of the previous doses
Nassar Junior e Park^([@r23])^ Maintain without sedation. Analgesia with fentanyl. If the patient was agitated (SAS ≥ 5), the team searched for the causes of agitation, and delirium was treated with haloperidol. If the patient remained agitated, sedation was initiated with midazolam or propofol Daily sedation interruption was employed until the patient could follow commands (open your eyes, follow with your eyes, shake hands and open your mouth). Sedatives and opioids were reinitiated at half the dose if agitated (SAS ≥ 5)
CrCl - creatinine clearance; SAS - Sedation Agitation Scale; RASS - Richmond Agitation Sedation Scale; RR - respiratory rate; SaO~2~ - arterial oxygen saturation; HR - heart rate; SBP: systolic blood pressure; ICP - intracranial pressure; BPS - Behavioral Pain Scale.
Quality assessment
------------------
Generally, all studies had a low risk of bias, except for the blinding of participants and professionals, which was absent in all studies. The risk of bias was considered \"uncertain\" for the blinding of outcome assessors, as they could know into which sedation strategy patients were randomized. This risk was considered uncertain because the analyzed outcomes were goals (e.g., mortality) or because this assessment was not described in the study. Allocation concealment was not adequately described in one study;^([@r19])^ thus, the risk of bias was considered \"uncertain\" ([Table 3](#t3){ref-type="table"}).
######
Risk of bias assessment
Study Generation of random sequence Concealment of allocation Blinding of participants and professionals Blinding of outcome assessors Incomplete outcomes Selective outcome reporting Other sources of bias
---------------------------------- ------------------------------- --------------------------- -------------------------------------------- ------------------------------- --------------------- ----------------------------- -----------------------
Mehta et al.^([@r17])^ Low Low High Uncertain Low Low Low
de Wit et al.^([@r18])^ Low Low High Uncertain Low Low Low
Anifantaki et al.^([@r19])^ Low Uncertain High Uncertain Low Low Low
Strom et al.^([@r20])^ Low Low High Uncertain Low Low Low
Yiliaz et al.^([@r21])^ Low Low High Uncertain Low Low Low
Mehta et al.^([@r22])^ Low Low High Uncertain Low Low Low
Nassar Junior and Park^([@r23])^ Low Low High Uncertain Low Low Low
Outcomes
--------
ICU mortality was assessed in seven studies with a total of 892 patients. There were no differences in ICU mortality between the sedation protocol and daily sedation interruption groups (OR = 0.81, 95% CI 0.60 - 1.10; I^2^ = 0%) ([Figure 2](#f2){ref-type="fig"}). Hospital mortality was assessed in six studies, with a total of 832 patients. There were no differences in hospital mortality between the sedation protocol and daily sedation interruption groups (OR = 0.89, 95% CI 0.66 - 1.19; I^2^ = 0%) (Figure 1S - <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>).
Figure 2Mortality in the intensive care unit.OR - odds ratio; 95%CI: 95% confidence interval; W - weight of study.
The duration of mechanical ventilation was assessed in 6 studies, which included 769 patients. The sedation protocols were not associated with any reduction in the duration of mechanical ventilation when compared to daily sedation interruption (MD = -1.52 days, 95%CI -3.60 - 0.56 days; I^2^ = 86.1%) ([Figure 3](#f3){ref-type="fig"}). Three studies, totaling 266 patients, analyzed the free days of mechanical ventilation in 28 days. The sedation protocols were associated with an increase free days of mechanical ventilation, but this result was marked by significant heterogeneity (MD = 6.70 days; 95%CI 1.09 - 12.31 days; I^2^ = 87.2%) (Figure 2S - <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>).
Figure 3Time on mechanical ventilation.SD - standard deviation; MD - mean difference; 95%CI - 95% confidence interval; W - weight of study.
Intensive care unit length of stay was assessed in six studies, including 769 patients. Hospital length of stay was also assessed in six studies and included 851 patients. There were no differences in the duration of ICU length of stay between the sedation protocols and daily sedation interruption (MD = -2.41 days, 95%CI -6.37 - 1.54 days; I^2^ = 89.9%) (Figure 3S - <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>). However, the sedation protocols were associated with a shorter duration of hospital length of stay (MD = -5.05 days, 95%CI -9.98 - -0.11 days, I^2^ = 69%) (Figure 4S - <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>).
Accidental extubation and extubation failure were assessed in four studies, involving 661 patients. The sedation protocols were not associated with higher rates of accidental extubation (OR = 1.02, 95%CI 0.55 to 1.90; I^2^ = 0%) (Figure 5S- <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>) or extubation failure (OR = 0.64, 95%CI 0.36 - 1.14; I^2^ = 0%) (Figure 6S-. <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>) compared to daily sedation interruption. The occurrence of delirium was assessed in only three studies, for a total of 596 patients. Delirium was not more common in patients allocated to sedation protocols than in those allocated to daily sedation interruption (OR = 1.45, 95%CI 0.77 - 2.73; I^2^ = 42.6%) (Figure 7S - <http://www.rbti.org.br/content/imagebank/pdf/0103-507X-rbti-28-04-0444-suppl01-en.pdf>).
DISCUSSION
==========
This systematic review and meta-analysis suggests that there are no differences between sedation protocols that target light sedation levels and daily sedation interruption strategies regarding mortality, duration of mechanical ventilation and length of ICU stay. With the use of sedation protocols targeting lighter levels of sedation, the number of free days of mechanical ventilation was higher and the hospital stay was shorter. However, these findings were based on a small number of studies, in the case of time free of mechanical ventilation, and were marked by high heterogeneity within the two results.
The minimization of sedation is imperative when considering the deleterious effects of excessive sedation.^([@r4],[@r5])^ Sedation protocols^([@r25])^ and daily sedation interruption^([@r8])^ have been studied for over 15 years and have shown significant benefits in terms of outcomes^([@r10],[@r11])^ and safety with respect to adverse events, such as accidental extubation, extubation failure^([@r9],[@r25])^ and long-term psychological outcomes.^([@r12])^ The results of our meta-analysis do not suggest significant differences between the two approaches in regard to important outcomes. In addition, our meta-analysis suggests low occurrences of accidental extubation and extubation failure, which are two common fears when addressing sedation reduction strategies.^([@r26],[@r27])^
Another meta-analysis that aimed to assess the effectiveness of daily sedation interruption conducted a sub-analysis that compared it with the use of sedation protocols with regard to the duration of mechanical ventilation. The comparison also revealed no differences between the two approaches regarding this outcome.^([@r28])^ Our meta-analysis included two studies that were not included by this other review. Anifantaki et al.\'s study^([@r19])^ was excluded because the authors considered that the protocol group represented \"usual care\" of the unit where the study was conducted. Unlike the Cochrane review, we decided to include the first study because the protocol effectively describes targeting a light sedation level.^([@r19])^ In our opinion, \"usual care\" refers to the decision of the unit to infuse sedatives according to physician\'s orders and without set targets, as effectively occurred in the other studies included^([@r8],[@r9],[@r29])^ and in the other meta-analysis.^([@r28])^ Strom et al.\'s study^([@r20])^ was excluded because the authors contemplated interruption of sedation in both groups. However, we decided to include this study because daily sedation interruption became part of the protocol only for patients who were still uncomfortable after administration of haloperidol and four attempts to suspend propofol infusion, used to control agitation.^([@r20])^ Therefore, we considered that sedation interruption was part of the protocol and not the protocol itself, as occurred in the other group.
Despite the apparent equivalence of daily sedation interruption and the protocols regarding outcomes and safety, three additional factors should be addressed. The first and foremost factor refers to nursing workload. While one of the studies included in this meta-analysis suggested the need for one professional more per patient to meet the demands of the same in the sedation protocol group,^([@r20])^ a Brazilian study showed no differences in nursing workload between groups during the first five days of mechanical ventilation.^([@r23])^ The second factor relates to the expectations and preferences of the professionals responsible for patient care. While approval of the sedation protocol was similar for physicians and nurses who participated in another study included in the review,^([@r22])^ nurses\' approval of daily sedation interruption was much lower than that of doctors. Nurses considered that the sedation protocol was easier to use and allowed greater patient comfort.^([@r27])^ The third factor relates to the experience and knowledge of a strategy, which, as would be expected, is associated with a greater likelihood of employing that strategy.^([@r30])^
The results of this meta-analysis should be interpreted with caution. Despite an extensive database search, only seven studies met the previously established inclusion criteria. Six of the seven studies were single-center and included few patients. Therefore, their results may not be valid in centers with different profiles. The variation in study characteristics is evident from the high statistical heterogeneity found in the analysis of outcomes involving time (i.e., duration of mechanical ventilation, ICU stay and hospital stay). Despite having a common target of sedation, both the protocols and the daily interruption strategy were performed differently from one study to another ([Table 2](#t2){ref-type="table"}), which could lead to discrepancies in results. We believe that data from ongoing studies (NCT01728558, NCT02040649) should shed more light on our findings in the coming years.
CONCLUSION
==========
Sedation protocols and daily sedation interruption appear to be equivalent as strategies targeting lighter sedation levels, although it should be noted that the target of sedation should be the primary goal of management in most patients under mechanical ventilation.
**Conflicts of interest:** None.
**Responsible editor:** Jorge Ibrain de Figueira Salluh
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Proteases are a large group of enzymes that can cleave proteins during a multitude of physiological reactions in all organisms ([@bib28]). A unique group of cysteine proteases called caspases functions to execute the apoptotic cell death program ([@bib69]; [@bib60]; [@bib57]). Caspases are synthesized as inactive zymogens (or proenzymes) and work in a controlled proteolytic cascade to activate themselves and one another ([@bib49]; [@bib52]). Initiator caspases are generally activated through dimerization, facilitated at multiprotein complexes such as the apoptosome (cytochrome c-Apaf-1--caspase-9; [@bib70]; [@bib50]; [@bib52]). Once activated, initiator caspases cleave and activate effector caspases, such as caspase-3 and -7 ([@bib8]; [@bib49]), which in turn cleave a variety of cellular protein substrates, ultimately leading to apoptosis ([@bib40]; [@bib15]; [@bib41]). Caspases are also regulated by inhibitory proteins, such as the inhibitor of apoptosis proteins, which can bind to and inhibit caspases in both insects and mammals ([@bib21]; [@bib61]; [@bib44]). In addition to the efforts invested in revealing and understanding these global pathways of caspase regulation during apoptosis, other studies imply that caspase activation is not necessarily an all-or-nothing process ([@bib26]; [@bib47]; [@bib48]). Indeed, low, transient, or subcellularly restricted levels of caspase activity have been recently reported to promote vital cellular processes ([@bib33]; [@bib32]; [@bib39]; [@bib53]). Furthermore, different cells, under varying physiological and pathological conditions, often display substantial differences in their sensitivity to apoptotic stimuli, which may reflect distinct apoptotic potentials. However, it is still poorly understood why restrictive levels of caspase activity fail to induce apoptosis and what factors determine the apoptotic potential of the cell.
The *Drosophila melanogaster* genome encodes seven distinct caspases, out of which the initiator caspase-9 ortholog Dronc and the effector caspase-3 ortholog Drice are the major apoptotic caspases ([@bib35]; [@bib10]; [@bib67]). Mutations in these caspases cause pleiotropic defects in developmental cell death and stress-induced apoptosis as well as the apoptosis-like process of spermatid individualization ([@bib9]; [@bib12]; [@bib62]; [@bib65], [@bib66]; [@bib5]; [@bib43]). In contrast, only minor physiological roles in apoptosis have been demonstrated for the initiator-like atypical caspase Strica/Dream and the effector-like caspases Dcp-1 and Decay ([@bib36]; [@bib38]; [@bib43]; [@bib66]; [@bib6]; [@bib13]). The caspase-8--like initiator Dredd appears not to be involved in cell death but rather in the innate immune response ([@bib37]; [@bib59]), whereas no role in apoptosis has been thus far demonstrated for the effector-like caspase Damm/Daydream ([@bib23]; [@bib38]). It is unclear why these caspases display different roles in apoptosis. For example, although Dcp-1 is highly homologous to Drice (67% identity; [@bib57]) and loss of *dcp-1* can aggravate *drice* mutant phenotypes ([@bib38]; [@bib43]; [@bib66]), *dcp-1* mutant flies are quite healthy, displaying only mild defects during starvation-induced autophagy and cell death in midoogenesis ([@bib36]; [@bib29]). Likewise, in mammals, knocking out caspase-3 causes decreased apoptosis and pleiotropic morphological defects, whereas caspase-7 knockout mice exhibit only mild antiapoptotic defects ([@bib34]; [@bib64]; [@bib30]). Whether these functional differences are the consequence of tissue specificity, distinct cellular levels of activity, or different execution efficiencies remains to be investigated.
Using a transgenic reporter of caspase-3--like (DEVDase) activity ([@bib63]), we set up an in vivo system for monitoring and comparing the activity levels and execution efficiencies of the two main effector caspases in *Drosophila*, Drice and Dcp-1, during apoptosis. We show that after irradiation-induced apoptosis, Drice and Dcp-1 are the only DEVDases that become activated in wing imaginal discs (WDs), although Drice is more efficient than Dcp-1 in processing this reporter and an endogenous protein substrate. The apoptosome components, Dronc and its adaptor protein Ark (the Apaf-1 homolog), constitute the initiator activity required for Drice and Dcp-1 activation, which can be blocked by expression of caspase inhibitory proteins. Importantly, in a series of genetic and transgenesis studies, we demonstrate that whereas both Drice and Dcp-1 can induce apoptosis, Drice can execute apoptosis more effectively than Dcp-1. Dcp-1, on the other hand, functions to fine tune the rate of cell death, at least in part through further activation of Drice in a positive amplification loop. Building on the differential efficiencies of these caspases to induce apoptosis, we show that the apoptotic potential of the cell is significantly affected by the total level of caspase activity, which in turn is directly proportional to the levels of the procaspases and their ratios. Finally, we demonstrate that the total level of caspase activity must reach a critical threshold in order for the cell to undergo apoptosis; short of that threshold, apoptosis cannot occur, and the cell will recover.
Results
=======
Detection of effector caspase--like DEVDase activity during apoptosis in vivo
-----------------------------------------------------------------------------
To monitor effector caspase activity in vivo, we took advantage of a genetic reporter previously designed for detection of caspase activity during the vital process of dendritic pruning in *Drosophila* sensory neurons ([@bib63]; [@bib53]). This reporter, dubbed *CD8::PARP::Venus* (*CPV*), encodes an artificial effector caspase substrate composed of a 40-aa-long fragment of the human poly(ADP-ribose) polymerase (PARP) protein (including the caspase-3 consensus site DEVD) flanked by the extracellular/transmembrane domain of the mouse CD8 protein at the N-terminal side and the YFP Venus at the C terminus ([Fig. 1 A](#fig1){ref-type="fig"}). The activity is detected by staining with the anticleaved human PARP antibody, whereas the fluorescence of the Venus protein indicates which cells express the reporter. The *CPV* reporter has not been characterized during apoptosis, and the exact caspases (or other proteases) that may cleave it have not been determined. Therefore, we established a genetic assay for monitoring the *CPV* reporter during irradiation-induced apoptosis in *Drosophila*. The *CPV* was expressed in the pouch region of the WD using the *spalt* (*sal*)-Gal4 driver. Transgenic larvae were then γ-irradiated and subsequently allowed a 4-h recovery before their WDs were removed and stained to visualize cleaved human PARP. Whereas nonirradiated animals displayed almost no reporter processing activity, strong activity was detected after apoptosis induction ([Fig. 1 B](#fig1){ref-type="fig"}, a and b, respectively). Furthermore, coexpression of the apoptosome components Ark and Dronc, using the *sal*-Gal4 driver, also induced a high level of *CPV* processing activity ([Fig. 1 B](#fig1){ref-type="fig"}, c), consistent with the idea that this combination is a strong inducer of apoptosis ([@bib56]).
{#fig1}
Apoptosis is associated with massive proteolytic activity ([@bib11]). To validate that *CPV* is cleaved only (immediately) after the DEVD consensus site, we introduced a point mutation to this reporter, which changes the invariably conserved aspartic acid residue to glycine (DEVD to DEVG). Transgenic flies expressing this modified construct (dubbed *CP^G^V*) were treated as described above for the visualization of *CPV* processing activity. However, this mutation completely abrogated processing of the reporter ([Fig. 1 B](#fig1){ref-type="fig"}, d). Therefore, the proteolytic cleavage of *CPV* during apoptosis reflects DEVDase activity only.
The effector caspases Drice and Dcp-1 are both activated during γ-irradiation--induced apoptosis
------------------------------------------------------------------------------------------------
Apoptosis is mainly mediated by caspases, but other proteases may also be involved in this process ([@bib54]). To test whether *CPV* is cleaved by effector caspases during apoptosis, the effector caspase inhibitor baculovirus gene *p35* and the *Drosophila* caspase inhibitor gene *diap1* were each expressed in the pouch region of WDs. After irradiation, *CPV* processing activity was completely blocked in the presence of p35 or largely suppressed in the presence of Diap1, suggesting that effector caspases are indeed responsible for this activity ([Fig. 2, A \[a--c\] and B](#fig2){ref-type="fig"}). The residual activity in the Diap1-expressing WDs is attributed to the relatively short half-life of this protein (*t*~1/2~ = 30 min; [@bib27]).
{ref-type="fig"}. *CPV* processing activity was detected in wild-type (wt; a) and also in *dcp-1* (d) and *drice* (e) mutants and was blocked upon expression of p35 (b) or in the *dcp-1* and *drice* double mutants (f) or largely attenuated by expression of *diap-1* (c). The open boxes group images of the same disc detected in different fluorescent channels. Bar, 50 µm. (B) Quantification of the relative caspase activity levels in the WDs described in A. Bars represent the mean cPARP staining area/reporter expression domain (represented by the Venus expression region). Error bars represent standard errors. P-values of the activity differences between various genotypes are indicated above the bars. All p-values were calculated using Fisher's exact test.](JCB_201107133R_Fig2){#fig2}
Out of the four structurally related effector-like caspases in *Drosophila*, only Drice and Dcp-1 have been previously associated with apoptosis in vivo, although only *drice* mutants were shown to block irradiation-induced apoptosis in the WD ([@bib43]; [@bib66]). Therefore, we used a null allele of *drice* (*drice^Δ1^*; [Fig. S1](http://www.jcb.org/cgi/content/full/jcb.201107133/DC1){#supp1}; see [Fig. 6 E](#fig6){ref-type="fig"}) to examine whether it is the only effector caspase responsible for the *CPV* processing activity in irradiated WDs. Unexpectedly, *CPV* processing activity persisted in WDs from irradiated *drice* mutants, suggesting the involvement of another effector caspase ([Fig. 2, A \[e\] and B](#fig2){ref-type="fig"}). Indeed, whereas a protein-null allele of *dcp-1* (*dcp-1^prev^*; see [Fig. 6 E](#fig6){ref-type="fig"}) also failed to block the *CPV* processing activity in this system, this activity was almost completely abrogated in the double mutants ([Fig. 2, A \[d and f\] and B](#fig2){ref-type="fig"}). Therefore, Drice and Dcp-1 are both activated in WDs after irradiation-induced apoptosis. Notably, quantification of the processing activity in each of the mutants suggests that Dcp-1 activity is slightly lower than that of Drice ([Fig. 2 B](#fig2){ref-type="fig"}). However, comparative analysis of transcript expression by real-time quantitative PCR (RT-qPCR) indicates that both caspases are expressed at comparable levels in this tissue, suggesting that the Dcp-1 activity in this system may be less efficient at cleaving this reporter (Fig. S1).
The apoptosome components Ark and Dronc are exclusively required for Drice and Dcp-1 activation after irradiation-induced apoptosis
-----------------------------------------------------------------------------------------------------------------------------------
The apoptosome components Ark and Dronc are associated with most of the developmental and stress-induced apoptotic events in *Drosophila*. To examine whether these apoptosome components are also required for Drice and Dcp-1 activation in our system, *CPV* processing activity was monitored in WDs from *ark^82^* and *dronc^I24/I29^* mutant flies. Indeed, most of the activity was strongly attenuated in these backgrounds, although some residual levels of activity remained ([Fig. 3, A \[a--c\] and B](#fig3){ref-type="fig"}). As both are null allelic backgrounds for their respective mutations, we asked whether this residual activity might be mediated by the initiator-like atypical caspase Strica, which has been suggested to be functionally redundant with Dronc in certain cellular paradigms ([@bib6]). However, no significant attenuation of the activity was detected in the *strica*-null mutant ([Fig. 3, A \[d\] and B](#fig3){ref-type="fig"}). Moreover, WDs from *dronc* and *strica* double mutants exhibited similar levels of residual activity as those in the *dronc* mutant alone ([Fig. 3, A \[e\] and B](#fig3){ref-type="fig"}). An alternative possibility for the retained residual activity in the *dronc* and *ark* mutant backgrounds may be the persistence of maternal mRNA of these genes. Indeed, *CPV* processing activity was almost completely abrogated in WDs from *dronc^I24/I29^* mutants, which also express an RNAi construct against *dronc* ([Fig. 3, A \[f\] and B](#fig3){ref-type="fig"}). Therefore, the Ark-Dronc apoptosome is exclusively required for Drice and Dcp-1 activation after irradiation-induced apoptosis.
{ref-type="fig"}, and the figure panels are presented accordingly. WDs from wild-type (wt; a), *ark* (b), *dronc* (c), and *strica* (d)-null mutants, *dronc* and *strica* double mutants (e), and *dronc* mutants that also express an RNAi against *dronc* (f) are shown. The open boxes group images of the same disc detected in different fluorescent channels. Bar, 50 µm. (B) Quantification of the relative caspase activity levels in WDs described in A. All calculations were performed as in [Fig. 2 B](#fig2){ref-type="fig"}. Error bars represent standard errors.](JCB_201107133R_Fig3){#fig3}
Dcp-1 activity is insufficient to induce apoptosis in WDs after irradiation
---------------------------------------------------------------------------
Previous works have suggested that irradiation-induced apoptosis in WDs is blocked in *drice* mutants ([@bib43]; [@bib66]). Therefore, we asked whether the Dcp-1 activity detected in WDs after irradiation is dispensable for cell death. To address this, WDs were first costained with anticleaved PARP (anti-cPARP) and TUNEL to visualize *CPV* processing activity and dying cells, respectively. In addition, discs of the same genotypes were also labeled with the vital dye acridine orange (AO), which is known to detect apoptotic corpses in *Drosophila* ([@bib3]). 4 h after 50 Gray (Gy) irradiation, effector caspase activity and cell death were dramatically induced in WDs from wild-type and *dcp-1* mutants ([Fig. 4, A and B](#fig4){ref-type="fig"}, a--a′′′, b--b′′′, and d--d′′′). In contrast, cell death was not induced in irradiated discs from *drice* mutants or *drice* and *dcp-1* double mutants, although, as aforementioned, Dcp-1 activity could be readily detected in the *drice* (single) mutant ([Fig. 4, A and B](#fig4){ref-type="fig"}, c--c′′′ and e′--e′′′). Therefore, in the absence of Drice, Dcp-1 activity in WDs is insufficient to induce apoptosis after irradiation.
{ref-type="fig"}. The specific genotypes indicated at the top relate to all the panels in a column. The first three columns are of the same imaginal disc. Note that caspase activity is confined to the reporter expression area, whereas cell death is detected in the entire WD. The open boxes group images of the same disc detected in different fluorescent channels. nonirrad, nonirradiated; wt, wild type. (B) WDs from irradiated larvae of the same genotypes as in A were stained with the vital dye AO to detect apoptotic corpses. (C) Similar WDs treated as in A and B and stained with the anticleaved caspase-3 antibody (cCasp. 3; lot no. 32; Cell Signaling Technology) to visualize the active forms of the effector caspases. Bars, 50 µm.](JCB_201107133R_Fig4){#fig4}
Because these experiments are performed by expressing relatively high levels of an artificial caspase substrate (the *CPV* reporter) within a large area of the WD, one concern is that this reporter may function as a competitive inhibitor of caspases and thus affect cell death. Therefore, we quantified and compared the levels of cell death induced in WDs from flies with the *UAS-CPV* construct alone (control) or flies containing the driver construct *sal*-Gal4 and the *UAS-CPV*. Importantly, the rates of cell death were almost identical between the control and *CPV*-expressing WDs at both 2 and 4 h after 50 Gy irradiation, suggesting that at least under these conditions, this reporter does not significantly affect the processing of endogenous death-associated substrates by caspases ([Fig. S2](http://www.jcb.org/cgi/content/full/jcb.201107133/DC1){#supp2}).
It has been recently suggested that the anticleaved caspase-3 antibody (also known as CM1) may reflect Dronc activity in *Drosophila*, but the specific cleaved caspases identified by this antibody have not been determined ([@bib16]). Furthermore, it has remained controversial whether this antibody can detect active Dcp-1 in vivo ([@bib68]; [@bib45]; [@bib43]). As the anticleaved caspase-3 antibody is widely used as a marker of caspase activation and apoptosis in *Drosophila*, we decided to examine four antibodies from two different sources raised against the same epitope (see also in Materials and methods). Interestingly, staining WDs with these antibodies revealed that they can recognize both active Drice ([Fig. 4 C](#fig4){ref-type="fig"}, d) and active Dcp-1 ([Fig. 4 C](#fig4){ref-type="fig"}, c) in vivo, and this staining was completely absent in the double mutants ([Fig. 4 C](#fig4){ref-type="fig"}, e). Nevertheless, the specificity of these antibodies toward active Dcp-1 varied among the different antibodies, which in all cases were much more specific toward active Drice ([Fig. S3](http://www.jcb.org/cgi/content/full/jcb.201107133/DC1){#supp3}).
Dcp-1 activity fine tunes the rate of cell death after irradiation
------------------------------------------------------------------
The idea that endogenous Dcp-1 activity cannot induce cell death in WDs after irradiation coupled with the fact that Drice activity alone is sufficient to induce apoptosis under the same conditions raised the question about the role of Dcp-1 in this system. One possibility is that Dcp-1 activity increases the total levels of effector activity in the cell, thus making cell death more efficient. To investigate this possibility, we quantified the levels of cell death in WDs from wild-type and *dcp-1* and *drice* mutants at different time points after 50 Gy irradiation. In wild type, the number of dying cells started to accumulate ∼1.5 h after irradiation; this number reached its peak after 3 h and plateaued in the following hours ([Fig. 5, A \[a\] and B](#fig5){ref-type="fig"}). Interestingly, although the *drice* mutants displayed the expected complete block in cell death, the *dcp-1* mutants displayed a significant delay of ∼1 h in the rate of cell death, reaching the wild-type plateau level 4 h after irradiation ([Fig. 5, A \[b\] and B](#fig5){ref-type="fig"}). To examine the correlation between the rates of cell death and caspase activity, we then quantified the levels of the *CPV* processing activities in these WDs. In accordance with the measured rates of cell death, both caspases became activated at ∼1.5 h after irradiation ([Fig. 5 C](#fig5){ref-type="fig"}). However, whereas Drice activity reached a peak and a plateau after 3 h, Dcp-1 activity reached that same plateau level ∼4 h after irradiation, suggesting that Dcp-1 may be less efficient at cleaving the *CPV* reporter than Drice or that Drice is more efficiently activated than Dcp-1 ([Fig. 5 C](#fig5){ref-type="fig"}). We conclude that whereas Drice activity is the trigger of cell death in this system, Dcp-1 activity fine tunes the rate of this process.
{ref-type="fig"}. (D) Visualization of cell death by TUNEL labeling in WDs from wild-type (a) and *dcp-1* (b) mutants at 4 h after 20 Gy irradiation. Bar, 50 µm. (E) Quantification of the relative levels of cell death in the WDs described in D. Cell death levels are calculated as in [Fig. 2 B](#fig2){ref-type="fig"}. (B, C, and E) Error bars represent standard errors.](JCB_201107133_Fig5){#fig5}
One way by which Dcp-1 may contribute to the acceleration of Drice-triggered apoptosis may be through a positive amplification loop. According to this model, after apoptosis induction, the Ark-Dronc apoptosome activates Drice and Dcp-1, and, in turn, these effector caspases cleave death-associated substrates, whereas Dcp-1 may also cleave and further activate Drice. In support of this model, it has been previously shown that Dcp-1 can cleave Drice in a cell-free system, but not vice versa, similar to the way Dronc cleaves Drice ([@bib24]; [@bib57]). To test this possibility in vivo, we performed Western blot analysis on protein extracts from wild-type, *drice*, and *dcp-1* mutant adult flies using an anti-Drice antibody. Basal cleavage of Drice is detected in wild-type flies but not in the *dcp-1* mutants, suggesting that Dcp-1 may promote cleavage of Drice also in vivo, albeit the significance of this cleavage (nonclassical p10 and p20) is unclear ([Fig. S4 A](http://www.jcb.org/cgi/content/full/jcb.201107133/DC1){#supp4}). To test whether Dcp-1 may indeed promote amplification of Drice activation in vivo, we used the anticleaved caspase-3 antibody to quantify the levels of active Drice in WDs from wild-type and *dcp-1* mutants at different time points after 50 Gy irradiation. Interestingly, a significant delay of ∼1 h is detected in the rate of Drice activation in the *dcp-1* mutant versus wild type (Fig. S4 B). As the anticleaved caspase-3 antibody mostly detects active Drice in vivo (Fig. S3), this delay is mainly attributed to the lack of Drice activation by Dcp-1. Collectively, these findings provide evidence that Dcp-1 may function, at least in part, through activating Drice in a positive amplification loop.
The finding that Dcp-1 activity makes cell death more efficient prompted us to explore other conditions in which Drice activity alone may be suboptimal. By gradually decreasing the dose of γ-irradiation, the rate of cell death was reduced accordingly. At 4 h after an irradiation dose of 20 Gy, the rate of cell death in the wild type was reduced to about half that observed after a dose of 50 Gy (compare D \[a\] and E with B in [Fig. 5](#fig5){ref-type="fig"}). Under these conditions, the rate of cell death in the *dcp-1* mutants (i.e., Drice activity only) was further reduced by 50% compared with wild type ([Fig. 5, D and E](#fig5){ref-type="fig"}). In contrast, the level of Drice activity alone at 4 h after 50 Gy irradiation could induce a similar level of cell death as that in wild type (albeit in a delay; [Fig. 5 B](#fig5){ref-type="fig"}). Therefore, the role of Dcp-1 activity in tuning the rate of cell death may be more significant when facing relatively weaker stresses.
Dcp-1 can induce cell death but less efficiently than Drice
-----------------------------------------------------------
The fact that Dcp-1 activity cannot induce apoptosis in irradiated WDs raised the question of whether it is capable of inducing cell death at all. To answer this question, we generated transgenic fly lines expressing either the full-length *drice* or *dcp-1* genes at identical levels. For this, the complete ORFs of *drice* and *dcp-1* were each subcloned under the same regulatory regions (promoter 5′ and 3′ untranslated regions \[UTRs\]) of the *drice* gene, generating the *drice:drice* and *drice:dcp-1* constructs, respectively ([Fig. 6 A](#fig6){ref-type="fig"}). Each construct was then targeted to the same genomic point using the ϕC31 integrase system, thus ensuring identical expression levels of these transgenes. By crossing these transgenes to the *drice* and *dcp-1* double mutant, we validated that both can induce *CPV* processing activity after irradiation ([Fig. 6 B](#fig6){ref-type="fig"}, e and f). Next, we asked whether these transgenes could restore irradiation-induced cell death in the *drice* mutants (where Dcp-1 is the only endogenous source of effector activity), in *drice* mutants that also lack one allele of *dcp-1*, and in the double mutants. Interestingly, under these conditions, both transgenes were able to induce cell death in the *drice* mutants, as detected by TUNEL labeling ([Fig. 6 B](#fig6){ref-type="fig"}, a and b), but only *drice:drice* could induce cell death in the *drice* mutants that are also heterozygous for the *dcp-1* allele ([Fig. 6 B](#fig6){ref-type="fig"}, c and d) and in the double mutants ([Fig. 6 B](#fig6){ref-type="fig"}, g and h). Of note, the same transgenes were also able to reverse the lethality of the *drice* mutant flies (see in the text for [Fig. 7](#fig7){ref-type="fig"}), indicating that the TUNEL-labeled WD cells are bona fide dying cells. Therefore, Dcp-1 has the potential to induce cell death by itself, but it is less efficient in doing so than Drice.
{ref-type="fig"}. Bars, 50 µm. (C) Quantification of the relative levels of cell death in WDs from transgenic flies expressing the Drice or Dcp-1 transgenes described in A and B in either of the following mutant backgrounds: *drice* homozygous mutants (I), *drice* homozygous and *dcp-1* heterozygous mutants (II), and *drice* and *dcp-1* double homozygous mutants (III). All calculations were performed as in [Fig. 5 B](#fig5){ref-type="fig"}. Error bars represent standard errors. (D) A representative image of a WD from *drice* mutants, which also expresses a *diap1*-targeting RNAi transgene in the pouch region. Flies were treated as in [Fig. 1 B](#fig1){ref-type="fig"}. Bar, 50 µm. (E) Drice and Dcp-1 display distinct efficiencies in cleaving lamin Dmo in vivo. Cleavage of the lamin Dmo protein was assessed by Western blotting of protein extracts from dissected WDs of the indicated genotypes using the anti--lamin Dmo antibody (top). The same membrane was reblotted with the anti--Dcp-1 antibody (middle). A similar membrane was blotted with anti-Drice antibody (bottom). Samples were either nonirradiated (−) or irradiated by a dose of 50 Gy and allowed to recover for 5--7 h (+). wt, wild type. Molecular mass is indicated in kilodaltons.](JCB_201107133_Fig6){#fig6}
{#fig7}
Procaspase levels and ratios affect cellular sensitivity to apoptosis by determining whether and at what rate caspase activity levels may reach a critical execution threshold
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Next, we assessed the differences in the abilities of the two caspases to induce apoptosis by quantifying the levels of cell death induced by each transgene. Consistent with the idea that Drice is more efficient in inducing apoptosis than Dcp-1, *drice:drice* induced more than twice as many cell death events than *drice:dcp-1* in the background of the *drice* mutant ([Fig. 6 C](#fig6){ref-type="fig"}, I). Further reduction of one or two copies of endogenous *dcp-1* in this mutant abolished the ability of *drice:dcp-1*, but not *drice:drice*, to induce cell death, implying the existence of a threshold effect ([Fig. 6 C](#fig6){ref-type="fig"}, II). Finally, in the double mutant background, only *drice:drice* could induce cell death but in a more reduced rate than in the single *drice* mutant ([Fig. 6 C](#fig6){ref-type="fig"}, III).
In living cells, Diap1 binds to and inhibits the activation of both Drice and Dcp-1, whereas upon induction of apoptosis, Reaper family proteins bind to Diap1, relieving its inhibition of caspases ([@bib14]; [@bib58]). Therefore, we hypothesized that reducing the level of *diap1* may lower the execution threshold of the cell. To test this, a *diap1*-RNAi transgene was expressed in the pouch region of the WD in the background of the *drice* mutant. After larval irradiation, these discs displayed dramatic induction of cell death in the pouch region but not in the surrounding tissue ([Fig. 6 D](#fig6){ref-type="fig"}). Therefore, reducing the levels of *diap1* elevates the levels of the activated Dcp-1 in the cell, leading to apoptosis. Finally, to test the possibility that Diap1 may be a stronger inhibitor of Dcp-1 than Drice, the *diap1*-RNAi transgene was expressed in the pouch region of WDs in either the *drice* or *dcp-1* mutants, and the WDs were tested for cell death induction as a result of spontaneous activation of the caspases (without irradiation). However, cell death was only detected in the pouch area of the *dcp-1* mutant but not the *drice* mutant, which is consistent with the idea that the differences in the execution efficiencies of Drice and Dcp-1 are not the consequence of differential inhibition of these caspases by Diap1 ([Fig. S5](http://www.jcb.org/cgi/content/full/jcb.201107133/DC1){#supp5}). Collectively, these results demonstrate that the levels of the proeffector caspases can tip the scales between cell survival and death and that it must go over a critical threshold to induce apoptosis.
Drice and Dcp-1 display distinct efficiencies in cleaving a protein substrate in vivo
-------------------------------------------------------------------------------------
A plausible reason for the relatively low efficiency of Dcp-1 to induce apoptosis is that it may cleave death-associated substrates less effectively than Drice. One such in vitro target of Drice and Dcp-1 is the *Drosophila* B-type lamin, lamin Dmo ([@bib19]; [@bib57]). To test whether Drice and Dcp-1 can also cleave this protein in vivo, Western blot analysis was performed on extracts prepared from wild-type, *dcp-1*, and *drice* mutant WDs both before and 5--7 h after irradiation. Using a specific antibody for lamin Dmo, cleavage of this protein was detected in all of the examined genotypes after irradiation, although this activity was dramatically reduced when *drice* was inactive (i.e., Dcp-1 activity only; [Fig. 6 E](#fig6){ref-type="fig"}). Furthermore, in accordance with the ability of transgenic Dcp-1 to rescue cell death after irradiation in the *drice* mutant, expression of transgenic Dcp-1 (on top of the endogenous Dcp-1 in the *drice* mutant) also increased the cleavage of lamin Dmo ([Fig. 6 E](#fig6){ref-type="fig"}). Therefore, Dcp-1 is less efficient than Drice at cleaving both the artificial *CPV* reporter and the death-associated substrate lamin Dmo in vivo.
Increasing Dcp-1 levels can compensate for the loss of *drice* during developmental cell death
----------------------------------------------------------------------------------------------
Cell death plays a major role during metamorphosis of the fly by removing many larval structures, thus paving the way for the generation of new adult tissues. Indeed, whereas *drice* mutants survive to the third instar larval stage, most of them die during pupal development ([@bib43]; [@bib66]). As transgenic Dcp-1 could restore irradiation-induced cell death in the *drice* mutants (albeit in a reduced rate than transgenic Drice), we asked whether it can also compensate for the loss of Drice during pupal development. Therefore, we evaluated the ability of these transgenes to increase the rate of larva-to--adult fly survival in different combinations of *drice* and *dcp-1* mutants. As shown in [Figs. 7](#fig7){ref-type="fig"} and [8 C](#fig8){ref-type="fig"}, only ∼20--30% of the *drice* mutants survived to adulthood, whereas the survival rate of the *dcp-1* mutants was similar to that of wild type. Further loss of one or two copies of endogenous *dcp-1* in the background of the *drice* mutants significantly reduced the survival rate or caused complete pupal lethality, respectively ([Fig. 7](#fig7){ref-type="fig"}). Importantly, all of the allelic combinations that restored irradiation-induced cell death, such as the transgenic Drice or Dcp-1 in the *drice* mutant background or transgenic Drice in the double mutants ([Fig. 6 C](#fig6){ref-type="fig"}), could also reverse pupal lethality and restore wild-type levels of survival, irrespective of the differences in the rates of cell death ([Fig. 7](#fig7){ref-type="fig"}). On the other hand, inability to restore irradiation-induced cell death (e.g., *drice:dcp-1* in the double mutant background; [Fig. 6 C](#fig6){ref-type="fig"}) fully correlated with a failure of these transgenes to reverse fly lethality ([Fig. 7](#fig7){ref-type="fig"}). Therefore, Drice is also a more effective executioner caspase than Dcp-1 during developmental cell death.
{ref-type="fig"}. Bars, 50 µm. (C) Quantification of the relative levels of cell death in WDs from transgenic flies expressing the Drice or Dcp-1 transgenes described in A and B in either of the following mutant backgrounds: *drice* homozygous mutants (I), *drice* homozygous and *dcp-1* heterozygous mutants (II), and *drice* and *dcp-1* double homozygous mutants (III). All calculations were performed as in [Fig. 5 B](#fig5){ref-type="fig"}. Note that the transgenes in (a and b) were inserted into a similar genomic locus as those described in [Fig. 6 (B and C)](#fig6){ref-type="fig"} but a different genomic locus than the transgenes in c--g. Error bars represent standard errors. (D) Quantification of the survival rates from larva to adult after expression of the *dcp-1:drice* or *dcp-1:dcp-1* transgenes in the background of *drice* homozygous mutants. Calculations were performed as in [Fig. 7](#fig7){ref-type="fig"}. Data are derived from a single prolonged experiment. wt, wild type.](JCB_201107133_Fig8){#fig8}
Distinct tissue specificities of Drice and Dcp-1 also affect their differential requirements during developmental cell death
----------------------------------------------------------------------------------------------------------------------------
RT-qPCR analysis (Fig. S1) and microarray data and EST tissue information deposited in public domains suggested that *drice* expression levels are much higher than *dcp-1* levels in almost all tissues investigated (unpublished data). To investigate the contribution of tissue specificities to the differences between the roles of the two caspases during developmental cell death, we generated new transgenes that contain the coding regions of *drice* and *dcp-1*, this time under the regulatory elements of *dcp-1* ([Fig. 8 A](#fig8){ref-type="fig"}). Similar to the *drice:drice* and *drice:dcp-1* transgenes and in agreement with the findings that the two caspases are expressed at similar levels in WDs (Fig. S1), the *dcp-1:drice* and *dcp-1:dcp-1* transgenes were also able to restore irradiation-induced cell death in WDs of the *drice* mutants, with *dcp-1:drice* being more effective than *dcp-1:dcp-1* ([Fig. 8, B \[a and b\] and C](#fig8){ref-type="fig"}). Furthermore, this similarity is extended to the double mutant backgrounds, showing that the *dcp-1:drice* transgene, but not the *dcp-1:dcp-1*, could restore irradiation-induced cell death in *drice* mutant flies that also contain either one or two copies of the *dcp-1* mutant allele ([Fig. 8, B \[c--h\] and C](#fig8){ref-type="fig"}). Therefore, the fact that a different promoter gives almost identical results reinforces our findings that Dcp-1 is a less efficient inducer of cell death than Drice and further demonstrates the existence of a threshold effect.
However, as opposed to *drice:drice* and *drice:dcp-1* transgenes, the *dcp-1:drice* transgene only partially rescued pupal lethality of the *drice* mutants, whereas the *dcp-1:dcp-1* transgene failed to rescue this lethality and instead further aggravated this phenotype (compare [Fig. 8 D](#fig8){ref-type="fig"} with [Fig. 7](#fig7){ref-type="fig"}). These results suggest that *dcp-1* is either not expressed or expressed at very low levels in at least some of the tissues where Drice activity is required during pupal development. In addition, Dcp-1 may also be uniquely expressed in some other tissues where high Dcp-1 activity, but not Drice activity, is deleterious to the cells. Altogether, these results suggest that the differential requirement for Drice and Dcp-1 during developmental cell death is also the consequence of tissue specificities.
Discussion
==========
The threshold effect of the effector caspase activity
-----------------------------------------------------
One of the most important questions in the fields of developmental apoptosis and cancer biology is what makes different cells more resistant or more sensitive to apoptotic stimuli. In the present study, we uncovered two factors important for establishing the overall apoptotic potential of the cell: the levels of the proeffector caspases and their execution efficiencies. We showed that the *Drosophila* effector caspases Drice and Dcp-1 differ in their competencies to induce apoptosis in vivo, with Drice being more effective than Dcp-1 (also illustrated in [Fig. 9 A](#fig9){ref-type="fig"}). Comparative transgenesis experiments indicate that these differences are the consequence of distinct intrinsic properties of the two caspases. Importantly, these experiments further demonstrate that cellular lethality is exerted only when caspase activity reaches a critical threshold. Short of that lethal level, cells fail to induce apoptosis, whereas surpassing that level increases the rate of cell death, thus further sensitizing the cells to apoptosis (see the model in [Fig. 9 B](#fig9){ref-type="fig"} and also below).
{#fig9}
The levels of the proeffector caspases affect the sensitivity of the cell to apoptotic stimuli
----------------------------------------------------------------------------------------------
Our experiments demonstrate that the level of activity of an effector caspase is directly proportionate to the level of expression of the respective inactive zymogene. Therefore, in addition to the established posttranslational regulatory mechanisms of caspases, which are crucial for the initiation of apoptosis, the expression levels of the proeffector caspases are an important factor in determining the sensitivity of the cells to apoptotic stimuli. This sensitivity is reflected in the rate by which the total level of effector activity meets the execution threshold of the cell. The higher the levels of the procaspases, the faster the cells reach that execution threshold level and the more sensitive the cells are to apoptotic stimuli (illustrated in [Fig. 9 B](#fig9){ref-type="fig"}). However, if the total levels of procaspases are too low, the activity rate may fail to reach that particular threshold, likely because the metabolic processes that normally sustain the cell overwhelm the catabolic processes, allowing for recovery and survival ([Fig. 9 B](#fig9){ref-type="fig"}). Interestingly, in an extreme case of apoptosis regulation at the level of procaspase expression, the onset of cell death in the tail-spiked cell of *Caenorhabditis elegans* was reported to be induced by an increase in the expression level of the caspase-encoding gene *ced-3* ([@bib42]); however, whether or not this is caused by a threshold effect is still unclear.
How does the level of effector caspase activity affect the rate of cell death? The length of the time between the apoptotic stimulus and the point when the cell becomes apoptotic varies between different cell types and distinct apoptotic stimuli. It is believed that this time period mainly reflects the time it takes for the cells to process and transduce the apoptotic signal and eventually translate it to the action of the caspases ([@bib46]; [@bib25]). However, our work suggests that the levels of effector caspase activity also vary between different cells as a consequence of the levels and ratios of the two main proeffector caspases, and this factor plays an important role in determining the time length between the activation of caspases and the eventual death of the cell, where high levels of effector activity shorten that period, thus increasing the rate of cell death. A plausible explanation for this correlation is that in order for the cell to start dying, a critical mass of death-associated substrates must be cleaved by the effector caspases, such that the more caspase molecules available, the faster this mass is cleaved. To date, ∼1,000 cellular proteins are thought to be cleaved by effector caspases during apoptosis, many of which are believed to gain old or new activities ([@bib40]; [@bib15]; [@bib41]). However, it is still unknown what caspase substrates are critical in apoptosis and what levels of cleaved substrates are required for induction of apoptosis. Future approaches to study that critical subset may further help in understanding how precisely the level of caspase activity affects the rate of cell death.
The *Drosophila* effector caspases differ in their abilities to induce apoptosis in vivo
----------------------------------------------------------------------------------------
We have presented several lines of evidence demonstrating that Drice is more efficient in inducing apoptosis than Dcp-1. First, despite the fact that both Drice and Dcp-1 are normally expressed at similar levels in the WD, only Drice activity could induce apoptosis after irradiation. Second, although when expressed under the *drice* regulatory regions both of the procaspase transgenes could restore irradiation-induced apoptosis and survival of the *drice* mutants, the rate of cell death was doubled with the Drice transgene. Moreover, only the Drice transgene could rescue irradiation-induced cell death and survival of the double mutants. Finally, similar differences in the efficiencies of the two caspases to induce apoptosis after irradiation were also observed when the procaspase transgenes were expressed under the *dcp-1* regulatory regions.
What are the reasons for the differences in the in vivo execution efficiencies of Drice and Dcp-1? One idea is that Diap1 may inhibit Dcp-1 more efficiently than Drice. However, this possibility is unlikely, as reduction of the same endogenous levels of Diap1 led to spontaneous induction of cell death only in the *dcp-1* mutant and not the *drice* mutant, demonstrating that despite similar levels of Diap1-free endogenous caspases, only activated Drice has the ability to induce apoptosis in this setup. Another possibility is that Dronc may cleave and activate Drice more efficiently than Dcp-1. In support of this idea, recombinant Dronc was reported to only poorly process Dcp-1, as compared with Drice, in a cell-free system ([@bib24]). Consistently, cleavage of the endogenous Drice was more prominent than that of endogenous Dcp-1 in irradiated WDs (unpublished data). However, an alternative or additional mechanism is that Drice may be more efficient than Dcp-1 at cleaving some of the more critical death-associated substrates. Indeed, Drice was able to process both the *CPV* reporter and an endogenous death-associated substrate, lamin Dmo, more efficiently than Dcp-1 in irradiated WDs. Whereas both possibilities support a model in which Drice and Dcp-1 function in a similar manner to induce apoptosis (i.e., possessing overlapping repertoires of death-associated protein substrates) but with different efficiencies, our data additionally suggest that these caspases may also have some distinct protein substrates. The findings that when expressed under the regulatory regions of *dcp-1* transgenic Drice partially rescued the survival of the *drice* mutants while the Dcp-1 transgene further increased their lethality imply that Dcp-1 may also cleave some distinct protein substrates. This is further supported by the notion that Dcp-1 can cleave Drice, whereas Drice cannot cleave Drice or Dcp-1 (see also below). Therefore, Drice and Dcp-1 may differ in their activation efficiencies by Dronc, their ability to effectively cleave similar critical death-associated substrates, and their repertoires of distinct substrates.
The coarse and fine modes of effector activity control
------------------------------------------------------
Both *Drosophila* and mammals possess major and minor effector caspases. Our work suggests that having both coarse (Drice; major) and fine (Dcp-1; minor) modes of effector activity control may allow the cells more flexibility in determining their apoptotic potential by setting their optimal execution threshold and rate of cell death. The finding that the rate of cell death in the absence of *dcp-1* was dramatically reduced after a relatively weak stress (i.e., 20-Gy irradiation) as compared with a stronger stress (i.e., a 50-Gy dose) suggests that under conditions resembling the normal situation in the wild, the minor caspase plays a more prominent role in determining whether or not the effector activity will even meet the execution threshold. In higher stresses, however, there is no need for a cell to rely on Dcp-1 activity to reach the execution threshold, as its activity is obviated by the high activity of Drice. Interestingly, in addition to *ced-3*, which is essential for apoptosis in *C. elegans* ([@bib69]), the worm genome also contains other caspase genes, but no documented roles in apoptosis have been thus far assigned for these caspases ([@bib55]). Likewise, *Drosophila* and mammals also contain additional caspases, which do not appear to have a major role in apoptosis ([@bib51]). Therefore, it is attractive to consider that similar to Dcp-1, at least some of the additional caspases in these organisms may also function to fine tune the rate of cell death, providing maybe even a finer tuner.
Our RT-qPCR analysis suggests that the WD is an exceptional tissue in relation to the similar levels of *dcp-1* and *drice* expression, as microarray data and EST tissue information deposited in public domains (FlyAtlas and FlyBase, respectively) suggest that *drice* expression levels are much higher than *dcp-1* levels in almost all tissues examined (unpublished data). Thus, fine tuning the rate of cell death may be the main role of Dcp-1 in apoptosis. However, this is probably not the only role of Dcp-1, as this caspase appears to have more prominent roles in certain cellular paradigms. For example, Dcp-1 was reported to be essential for the sporadic and nutrient deprivation--induced germline cell death during midoogenesis, with Drice playing only a minor role, if any, in this system ([@bib36]; [@bib6]). Furthermore, genetic analyses revealed additional cellular paradigms in which the role of Dcp-1 in cell death is more prominent or that Dcp-1 inactivation significantly enhances *drice* mutant phenotypes ([@bib38]; [@bib43]; [@bib66]; [@bib6]). Although it is largely unknown why different cells/tissues need to acquire distinct apoptotic potentials, it is not hard to envision that certain cells may need to be more resistant or more sensitive to apoptosis. For instance, cell types with low regeneration potential and with limited numbers, such as neurons, or cells with direct interface with the outside world and that are in higher risk of facing different stresses (e.g., eyes or sperm) may be more resistant to apoptosis and hence may express relatively low levels of Drice. On the other hand, dying cells that send signals to the surrounding tissue area (e.g., signals for compensatory proliferation) may need to express higher levels of Dcp-1 to fine tune their dying rate and thus control the level of the signal they send. Therefore, different cells in the organism may use distinct levels and ratios of Drice and Dcp-1 to optimize their sensitivity to apoptosis induction.
Drice and Dcp-1 synergize their activities during apoptosis
-----------------------------------------------------------
Our data also point to a synergistic effect of Drice and Dcp-1 on the overall effector activity levels in the cell. This effect is clearly detected when comparing the overall level of effector caspase activity in wild type (which reflects the combined activities of Drice and Dcp-1) and the sum of the single activity levels of Drice and Dcp-1 (measured in *dcp-1* and *drice* mutant backgrounds, respectively). During the time window of 1--2 h after irradiation, wild-type activity levels were significantly higher than the simple summation of the levels in the mutants, implying the existence of a positive amplification loop of effector caspase activation. In a later window of time 3--4 h after irradiation, activity levels in wild type reached a peak and plateaued ([Fig. 5 C](#fig5){ref-type="fig"}). Because the levels of Drice activity alone also reached a similar peak and plateaued within the same time window, we attribute the formation of this plateau to either staining saturation or limited substrate availability but not the actual levels of caspase activity during that time frame. Interestingly, it has been previously shown that Dcp-1 can cleave Drice in a cell-free system but not vice versa, similar to the way Dronc cleaves Drice ([@bib24]; [@bib57]). Likewise, we showed cleavage of Drice, but not Dcp-1, in extracts from wild-type adult flies, which was blocked in the *dcp-1* mutants (Fig. S4 A). Moreover, monitoring the kinetics of Drice activation (using the anticleaved caspase-3 antibody) in wild-type and *dcp-1* mutant WDs displayed a significant delay in the rate of Drice activation when Dcp-1 was lacking (Fig. S4 B). This delay cannot be attributed to merely the lack of staining of activated Dcp-1, as even if we were to assume that this antibody has identical specificities toward activated Drice and activated Dcp-1 (although, as aforementioned, this is not the case), it would be expected that without an amplification loop, the maximal contribution of activated Dcp-1 to the overall staining level shall not surpass that of activated Drice (i.e., a maximum of twofold increase). However, the lack of Dcp-1 caused much more than a twofold reduction in the levels of staining (i.e., 3--7 fold), indicating that this reduction is mainly a result of less activated Drice in the absence of Dcp-1. Altogether, these findings provide an indirect support to a model whereby Dcp-1 may, either directly or indirectly, promote further activation of Drice through a positive amplification loop (see also the model in [Fig. 9 A](#fig9){ref-type="fig"}). It remains to be determined whether this cleavage of Drice by Dcp-1 in vivo leads to activation of the former and whether this may account for the aforementioned synergistic effect.
In conclusion, we have demonstrated that cells can continue living in the presence of a considerable level of effector caspase activity, as long as this level does not reach a critical execution threshold. Current models of the conserved core apoptotic machinery between *Drosophila* and mammals commonly illustrate the effector caspase activation step as an all-or-nothing process ([@bib20]). However, recent studies of caspase-dependent nonapoptotic cellular processes suggest that to avoid excessive caspase activation and apoptosis, effector caspase activity may be low or restricted in space and time in these cells ([@bib17]; [@bib32]; [@bib39]). Furthermore, it is well established that cancer cells can escape cellular lethality by manipulating a variety of steps and components in the apoptotic machinery, ultimately affecting the activity levels of the effector caspases ([@bib31]; [@bib22]). The current study provides insight into why low or restricted caspase activity levels fail to induce apoptosis (i.e., the threshold effect) and suggests that the critical step of caspase activation is more complex than has been previously appreciated.
Materials and methods
=====================
Fly strains and expression vectors
----------------------------------
Flies containing the *sal-gal4* driver (provided by K. Basler, University of Zurich, Zurich, Switzerland) were crossed to flies with the reporter transgene *CPV* ([@bib63]), and the progeny were used as wild-type controls. The fly mutant alleles used in this study are *dronc^I24^* and *dronc^I29^* ([@bib65]), *drice^Δ1^* ([@bib43]), *dcp-1^prev^* ([@bib36]), *ark^82^* ([@bib2]), and *strica^4^* ([@bib6]). The *UAS-CPV* and *sal-gal4* transgenes were recombined to the *drice* and *dronc* mutant alleles. The following *UAS* fly lines were also used: *UAS-p35* (provided by H. Steller, Howard Hughes Medical Institute, Strang Laboratory of Cancer Research, The Rockefeller University, New York, NY), *UAS-pro-dronc* and *UAS-ark* (provided by H.D. Ryoo, New York University School of Medicine, New York, NY), and *UAS-diap1* (provided by T. Volk, Weizmann Institute of Science, Rehovot, Israel). The *dronc*-RNAi line was obtained from the Vienna Drosophila RNAi Center (transformant ID no. 23035), and *diap1*-RNAi was a gift from P. Meier (Institute of Cancer Research, London, England, UK).
To generate the *UAS-CD8::parp^DEVG^::venus* (*UAS-CP^G^V*) construct, the *CD8::parp::venus* was subcloned into pBluescript II SK plasmid, and then a point mutation was introduced using the QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies) with the primers forward 5′-GGCGATGAGGTGGGTGGAGTGGATGAA-3′ and reverse 5′-TTCATCCACTCCACCCACCTCATCGC-3′, which changed the adenine residue (A722) into guanine, leading to a change of the conserved aspartic acid at position P1 into glycine.
The rescue transgenic lines *drice:drice* and *drice:dcp-1* were generated as follows: first, a 564-bp fragment from the *drice* genome encompassing its promoter and 5′ UTR was PCR amplified using the forward primer 5′-GGCAATTGCCTCTTTGAGAGTGTGACCG-3′ and reverse primer 5′-CCAAGATCTGGCTAAGTTCTCTCCTTGAG-3′ with added MfeI and BglII restriction sites, respectively. Second, a 708-bp fragment of *drice* 3′ UTR was also amplified by genomic PCR using the forward primer 5′-CCGGGCGGCCGCTGGCTAATGGTATGGATCAA-3′ and reverse primer 5′-GCGGTACCAGGGTCAACAGCAAACAGCCAA-3′ with added NotI and Acc65I restriction sites, respectively. Both the promoter/5′ UTR and 3′ UTR fragments were respectively cloned in a sequential order into the EcoRI + BglII and NotI + Acc65I sites of the pattB plasmid (a gift from J. Bischof, University of Zurich, Zurich, Switzerland), giving rise to the pattB-*drice*-5′-3′ plasmid. The *drice* coding region was amplified from a cDNA clone (a gift from H. Steller) using the forward primer 5′-CCAAGATCTATGGACGCCACTAACAATGG-3′ and reverse primer 5′-GGGGGGCCCTCAAACCCGTCCGGCTGGTG-3′ with added BglII and PspOMI restriction sites and was subsequently respectively cloned into the BglII and NotI restriction sites of the pattB-*drice*-5′-3′ plasmid. The *dcp-1* coding region was amplified from a cDNA clone (a gift from H. Steller) using the forward primer 5′-CGGGGATCCATGACCGACGAGTGCGTAAC-3′ and reverse primer 5′-CGGCGCGGCCGCCTAGCCAGCCTTATTGCCGT-3′ with added BglII and NotI restriction sites and was subsequently respectively cloned into the BamHI and NotI restriction sites of the pattB-*drice*-5′-3′ plasmid. The two transgenic fly lines, *drice:drice* and *drice:dcp-1*, were generated using the ϕC31-mediated site-specific transgenesis technique, which allows insertion of transgenes into known sites of the *Drosophila* genome ([@bib7]; [@bib18]). Specifically, these transgenes were inserted into the attP18 site on the X chromosome. Transcriptional expression of the transgenes was confirmed by RT-qPCR analysis on RNA from WDs of the two transgenic fly lines.
The rescue transgenic lines *dcp-1:drice* and *dcp-1:dcp-1* were generated as follows: first, a 703-bp fragment from the *dcp-1* genome encompassing its promoter and 5′ UTR was amplified using the forward primer 5′-AACAGATCTCTGTTTTTAATGTTAGATTAG-3′ and reverse primer 5′-GCGAATTCCTTGCGCCCCTTTTCTTGC-3′ with added BglII and EcoRI restriction sites, respectively. Second, a 1,090-bp fragment of *dcp-1* 3′ UTR was also amplified by genomic PCR using the forward primer 5′-AACTCGAGGAAGAGATCTCCCTTCGAAG-3′ and reverse primer 5′-AATCTAGAGTAAGCGGCTCCATCCATGGG-3′ with added XhoI and XbaI restriction sites, respectively. Both the promoter/5′ UTR and 3′ UTR fragments were respectively cloned in a sequential order into the BamHI + EcoRI and XhoI + XbaI sites of the pattB plasmid, giving rise to the pattB-*dcp-1*-5′-3′ plasmid. The *dcp-1* and *drice* coding regions were each amplified from the cDNA clones as described above but this time using primers with added NotI and SalI restriction sites. Each fragment was then cloned into the NotI and XhoI restriction sites in the pattB-*dcp-1*-5′-3′ plasmid. These transgenes were inserted into the attP40 site on chromosome 2L using the ϕC31 system as before.
Immunofluorescence staining, TUNEL, and AO labeling and antibodies
------------------------------------------------------------------
Third instar larvae were subjected to γ-irradiation (50- or 20-Gy doses) and allowed to recover for several hours at room temperature, and their WDs were dissected and stained for either cleaved human PARP, cleaved caspase-3, TUNEL (see below), or AO (see below) using standard protocols ([@bib4]; [@bib3]). The corresponding antibodies used were rabbit polyclonal anticleaved human PARP (1:500, Ab2317; Abcam), rabbit polyclonal anticleaved caspase-3 (1:75, Asp175; Cell Signaling Technology), and rabbit polyclonal anticleaved caspase-3 (1:500, Ab13847; Abcam).
For TUNEL labeling, WDs were dissected and fixed for 20 min at room temperature in 4% PFA in PBS. The samples were then rinsed in PBS, washed twice, 10 min per wash, in 1× BSS (5× BSS: 270 mM NaCl, 200 mM KCl, 37 mM MgSO~4~, 12 mM CaCl~2~, 2 mM H~2~O, 24 mM tricine, 1.8% glucose, and 8.5% sucrose), and washed three times, 5 min per wash, with PBTw (0.1% Tween 20 in PBS). The samples were refixed in 4% PFA for 20 min, washed in PBTw five times, 5 min per wash, incubated in equilibration buffer (ApopTag kit; Millipore) for 1 h, and incubated again in reaction buffer (TdT enzyme; ratio 7:3; ApopTag kit) at 37°C overnight. On the next day, the TdT reaction mix was replaced with stop buffer (diluted 1:34 in dH~2~O; ApopTag kit) and incubated at 37°C for 3--4 h. Then, the samples were washed three times, 5 min per wash, blocked in BTN solution (1× BSS, 0.3% Triton X-100, and 5% normal goat serum) at room temperature for 1 h, and incubated with antidigoxigenin antibody solution (diluted 47:53 in blocking solution; ApopTag kit) overnight in the dark at 4°C. On the following day, the samples were washed four times in 1× BSS, 20 min per wash, and mounted in Fluoromount-G (SouthernBiotech).
For AO staining, WDs were incubated in a 0.6-µg/ml solution of AO (diluted in PBS) for 5 min, washed briefly in PBS, and mounted in a drop of PBS. Images were taken on a confocal microscope (LSM510 Meta Inverted Axio Observer; Carl Zeiss) using an EC Plan Neofluar 20×/0.50 M27 lens. Fluoromount-G was used as the imaging medium. All images were captured using the LSM510 operating software (Carl Zeiss).
Detection of effector caspase activity in vivo using the *CPV* reporter
-----------------------------------------------------------------------
In living cells, the reporter is localized to the plasma membrane via its CD8 domain. Upon cleavage by caspases, Venus is released to the cytoplasm, and, in principle, this process can be revealed by live imaging in large-diameter cells. However, in tissues of small-diameter cells, it is more difficult to distinguish between the membranal and cytoplasmic Venus, and, thus, the detection requires further staining of the tissue with an anti-cPARP antibody, which specifically detects the cleaved C-terminal fragment of human PARP.
Quantification of images
------------------------
Quantification of staining was performed by measuring the area of the positively stained pixels (i.e., cPARP or TUNEL) and dividing it in the area of the Venus expression for cPARP or the total disc area for TUNEL using the ImageJ program (National Institutes of Health; [@bib1]). Statistical analysis was performed using a one-way analysis of variance test followed by Fisher's protected least significant difference posttest for multiple comparisons using the StatView Program (Abacus Concepts). For each experiment, 12--20 WDs were analyzed for each genotype or time point. Significance level was considered as either P \< 0.05 or P \< 0.001, as indicated above the bars in the figures.
Quantification of *drice* and *dcp-1* mRNA expression levels
------------------------------------------------------------
WDs from wild-type and mutants flies were dissected, and total RNA was extracted using the RNeasy Micro Kit (QIAGEN). Reverse transcription was performed with 1 µg of total RNA. First-strand synthesis used Oligo(dT) primers (Promega) and SuperScript II Reverse Transcriptase (Promega) in the presence of RNase-free DNase to eliminate DNA contamination and Protector RNase inhibitor (Roche). Measurements were normalized to mitochondrial large-ribosomal RNA (*MtlrRNA1*: forward 5′-AAAAAGATTGCGACCTCGAT-3′ and reverse 5′-AAACCAACCTGGCTTACACC-3′) or *RP49* (*Dmel/RpL32*: forward 5′-GACCATCCGCCCAGCATAC-3′ and reverse 5′-CCATTTGTGCGACAGCTTAGC-3′). The primers for *drice* mRNA were forward 5′-CCACTAACAATGGAGAATCCGCC-3′ and reverse 5′-GCCGCTGCTACCCGCTCCTC-3′. The primers for *dcp-1* mRNA were forward 5′-ACCGACGAGTGCGTAACCAGAA-3′ and reverse 5′-ACAAGAGACTCCGGCGTACAGC-3′. Products were amplified by the KAPA SYBR FAST quantitative PCR kit (Kapa Biosystems) using the LightCycler 480 quantitative PCR machine (Roche). Results of expression levels were calculated as the mean from three independent experiments with at least two biological repeats with two duplicates each.
Survival assay
--------------
Third instar larvae of each genotype were collected in vials with fresh food at 25°C. Eclosing adults were counted, and fly viability was determined as the number of eclosed adults divided in the total larvae collected. Survival rate was normalized to wild-type levels, and the larvae were collected during a period of 3 wk.
Western blotting
----------------
50 WDs from larvae with the desired genotype were dissected in PBS and collected into cold lysis buffer (50 mM Tris, pH 8, 150 mM NaCl, 1% NP-40, and a protease inhibitor mix). Then, these WDs were homogenized and incubated for 15 min on ice and centrifuged for 20 min, and the supernatant was collected and mixed with a sample buffer. Gel running and subsequent blotting was performed using standard methods, and the blots were incubated with a guinea pig anti--Dcp-1 antibody (1:1,000; a gift from P. Meier), a mouse anti--lamin Dmo antibody (1:500; a gift from Y. Gruenbaum, The Hebrew University of Jerusalem, Jerusalem, Israel), or a rabbit anti-Drice antibody (1:2,000; a gift from P. Friesen, University of Wisconsin-Madison, Madison, WI).
Online supplemental material
----------------------------
Fig. S1 shows that *drice* and *dcp-1* transcripts are expressed at comparable levels in WDs. Fig. S2 shows that the *CPV* reporter does not affect the rate of cell death. Fig. S3 shows that different anticleaved caspase-3 (CM1) antibodies detect both active Drice and Dcp-1, although they are more specific toward active Drice. Fig. S4 shows that Dcp-1 may further activate Drice in a positive amplification loop. Fig. S5 shows that Diap1 does not inhibit Drice and Dcp-1 with different efficiencies. Online supplemental material is available at <http://www.jcb.org/cgi/content/full/jcb.201107133/DC1>.
We are grateful to J.M. Abrams, K. Basler, A. Bergmann, P. Friesen, Y. Gruenbaum, B. Hay, K. McCall, P. Meier, H.D. Ryoo, H. Steller, T. Volk, and D.W. Williams for providing additional stocks and reagents, S. Kredo-Russo for helping with the RT-qPCR, V. Kalchenko for helping with the ImageJ program, Y. Kaplan for helping with coimmunoprecipitation experiments, and S. Pietrokovski and E. Schejter for useful discussions. We thank the Arama laboratory members for encouragement and advice, O. Reiner and L. Ravid for critically reading the manuscript, and S. Brown for English editing.
This research was supported in part by the Israel Science Foundation (grant no. 308/09) and Minerva Foundation, with funding from the German Federal Ministry of Education and Research, Israel Cancer Research Fund, Israel Cancer Association, Yeda-Sela Center for Basic Research, and the M.D. Moross Institute for Cancer Research. E. Arama is also supported by grants from the Y. Leon Benoziyo Institute for Molecular Medicine, the Jeanne and Joseph Nissim Center for Life Sciences Research, and Lord Mitchell. E. Arama is the Incumbent of the Corinne S. Koshland Career Development Chair.
Abbreviations used in this paper:AOacridine orangecPARPcleaved PARPPARPpoly(ADP-ribose) polymeraseRT-qPCRreal-time quantitative PCRUTRuntranslated regionWDwing imaginal disc
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Immunoglobulins (Ig) are glycoproteins produced exclusively by activated B-lymphocytes and plasma cells that mediate humoral response against pathogens. Each B-cell clone presents an antigen-specific membrane-bound immunoglobulin that, together with CD79A and CD79B molecules, comprise the B-cell receptor (BCR). After stimulation by antigens, B-cells secrete immunoglobulins (antibodies) with the same antigen-binding sites than the membrane-bound molecules.
All Ig share a similar basic structure composed of four polypeptide chains: two identical heavy chains and two identical light chains. The heavy chain has one variable domain (V~H~) and three or four constant domains (C~H~1, C~H~2, C~H~3, and C~H~4). Each light chain exhibits one variable domain (V~L~) and one constant domain (C~L~). The variable region (V~H~ and V~L~) is responsible for antigen recognition and binding while the constant regions (C~H~ and C~L~) primarily mediate the Ig effector functions, which includes complement activation and Fc Receptor binding ([@B1], [@B2]).
In humans, the immunoglobulin heavy chain gene (*IGH*) is located in chromosome region 14q32 and consists of four groups of gene segments: the variable heavy (*IGHV*), diversity heavy (*IGHD*), joining heavy (*IGHJ*), and constant heavy (*IGHC*). The *IGHC* group includes *IGHM, IGHD, IGHG3, IGHG1, IGHEP1, IGHA1, IGHGP, IGHG2, IGHG4, IGHGE*, and *IGHA2* gene segments and pseudogenes. Immunoglobulin light chains are encoded by two different genes: lambda (*IGL*) at 22q11 and kappa (*IGK*) at 2p11.2 ([@B3], [@B4]).
During B-cell development, the *IGH* gene undergoes a somatic rearrangement, in which only one *IGHV*, one *IGHD*, and one *IGHJ* gene segment are combined to form the Ig variable region V~H~. In contrast, during clonal expansion after activation of the B-cell, *IGHC* gene segments go through a process called class-switch recombination, which determines the Ig class and subclass: IgM, IgD, IgG (IgG1, IgG2, IgG3, and IgG4), IgA (IgA1 and IgA2), and IgE. The human humoral immune response is mainly mediated by Ig gamma (IgG), which is subdivided into four subclasses, IgG1, IgG2, IgG3, and IgG4, ordered by decreasing abundance in peripheral blood ([@B5]). The constant regions of these four subclasses are encoded by the gene segments *IGHG1, IGHG2, IGHG3*, and *IGHG4*, respectively, the first three being the ones focused on this study. Each *IGHG* gene segment consists of three exons that encode the constant heavy domains (*CH1, CH2*, and *CH3*) and exon *H*, which encodes the hinge between the CH1 and CH2 domains ([@B5]).
Most of the human IgG diversity in populations has only been characterized by serological methods, which defined the immunoglobulin allotypes at the protein level. Ig allotypes are polymorphic epitopes (resulting from nucleotide variation) on the Ig constant domain that provide binding sites for antibodies ([@B6]). Certain IgG allotypes have been associated with susceptibility to cancer, autoimmune and infectious diseases ([@B7]--[@B9]).
Although the genetic variability of some *IGHG* gene segments has been characterized ([@B10], [@B11]), it has never been systematically sequenced at the nucleotide level across populations. Thus, the diversity of these gene segments is probably underestimated. Additionally, this genomic region is not well-covered in genome-wide studies and genomic databases for two reasons: first, DNA samples used are often extracted from B-cell lines, which are not suitable for analyzing this region due to the somatic rearrangement within this locus; second, the high sequence similarity of these segments imposes technical difficulties for sequencing and genotyping ([@B12]).
Here, we analyzed the diversity of *IGHG1, IGHG2*, and *IGHG3* in seven Brazilian populations: five Amerindian populations that have been genetically isolated for centuries and two urban populations. By analyzing deep sequencing data, we found 28 novel *IGHG* alleles, characterized the linkage disequilibrium of variants within these segments and analyzed the relationship among alleles. Additionally, we provided compelling evidence of the occurrence of gene conversion between different gene segments and evidence of purifying selection shaping *IGHG* diversity.
Methods {#s2}
=======
Characterization of the Study Populations
-----------------------------------------
This study was approved by the Brazilian National Human Research Ethics Committee (CONEP), protocol number CAAE 02727412.4.0000.0096, in accordance to the Brazilian Federal laws. We analyzed a total of 357 individuals from seven Brazilian populations, of which five are Amerindian: Guarani Kaiowa (GKW, *n* = 46), Guarani Ñandeva (GND, *n* = 48), Guarani Mbya (GRC, *n* = 51), Kaingang from Ivaí (KIV, *n* = 52), and Kaingang from Rio das Cobras (KRC, *n* = 52); and two are urban populations: Japanese-descendants (BrJAP, *n* = 57) and Euro-descendants from Curitiba (CTBA, *n* = 51). Their detailed geographic location and sample sizes are found in [Figure 1](#F1){ref-type="fig"} and [Table S1](#SM3){ref-type="supplementary-material"}.
{#F1}
The Amerindians samples were collected between late 1980s and early 1990s. According to public data from the Brazilian Institute of Geography and Statistics (IBGE), there are approximately 900,000 Amerindians individuals in Brazil, distributed across 693 official indigenous lands (<https://www.ibge.gov.br>). The Guarani speak a Tupi-Guarani language, which belongs to the Tupi language family. The Kaingang speak Jê, which belongs to the Macro-Jê language stock. Analyzing mtDNA segments and the proposed time of origin of Tupi-Guarani and Jê linguistic families, Marrero and colleagues ([@B13]) estimated the Guarani population split in three partialities (Guarani Kaiowa, Guarani Ñandeva and Guarani Mbya) 1,800 years ago, while the different Kaingang populations would have split more recently, around 200 years ago. Since then, they are believed to have remained isolated from each other and other urban populations due to strong cultural and language barriers ([@B14]). A former study from our group estimated that the gene flow of these Amerindian populations with non-Amerindians was low, being 0% in Guarani Kaiowa, 4% in Guarani Mbya, 14% in Guarani Ñandeva and 7% in Kaingang ([@B15]).
The two urban samples were from Curitiba, the capital of Paraná State and the 5th largest city in Brazil. As a result of the Brazilian history of European colonization 500 years ago, and especially the more recent European migrations since the 19^th^ Century, the population of Curitiba is of predominantly European ancestry. According to the public data from IBGE, 78.7% of the inhabitants of Curitiba self-declared themselves as white, 16.7% as admixed, 3% as black, 1.4% as Asian, and 0.2% as Amerindian (<https://www.ibge.gov.br>). The population here referred as CTBA only included Euro-descendant individuals. Therefore, we excluded all individuals with known miscegenation with Amerindian and/or other non-European ancestries. Paraná State also hosts the second largest Japanese community in Brazil, one of the largest outside Japan. The Japanese migration started in the twentieth century with the Treaty of Friendship, Commerce and Navigation between Brazil and Japan. All Japanese-descendent individuals of this study (BrJAP) were born in Brazil, with either both parents or all four grandparents born in Japan. They reported no history of admixture with non-Japanese ethnicities.
DNA Extraction
--------------
Genomic DNA was extracted from peripheral blood samples by standard salting-out ([@B16]) or by the phenol-chloroform-isoamyl method ([@B17]). High-quality DNA has been stored at −80°C since the extraction. DNA integrity was evaluated by 1% agarose gel electrophoresis and purity was accessed by spectrophotometry.
Sequencing and Allele Identification
------------------------------------
We aligned all previously known *IGHG* alleles and designed primers to amplify each segment specifically. To define the best set of primers we used the following approaches: (i) we ruled out unspecific amplification by verifying that all amplicons did not exhibit any variant that was specific of other segments; (ii) we certified that the genotype distribution of all single nucleotide variable sites, in each amplicon, were in accordance to Hardy-Weinberg equilibrium (*p* \> 0.05).
Polymerase chain reaction (PCR) was performed for *IGHG1, IGHG2*, and *IGHG3* as follows: 1X Buffer (Invitrogen); 0.2 mM dNTP (Life Technologies); 1.5 mM MgCl~2~ (Invitrogen, Carlsbad, CA, USA); 0.3 μM of each primer; 0.05 U/μL Taq polymerase Platinum® (Invitrogen, Carlsbad, CA, USA); and 2 ng/μL genomic DNA. The segments were amplified in a Mastercycler ep Gradient S thermocycler (Eppendorf, Hamburg, Germany), with a first step at 94°C for 2 min and 10 cycles of 94°C for 15 s, T~m~A °C for 15 s and 72°C for 60 s, followed by 25 cycles of 94°C for 15 s, T~m~B °C for 15 s and 72°C for 60 s, with a final extension step of 72°C for 60 s (primer sequences, location, and amplification temperatures are available in [Table S2](#SM4){ref-type="supplementary-material"} and [Figure S1](#SM1){ref-type="supplementary-material"}). Amplicons were visualized by 1% agarose gel electrophoresis with 1% UniSafe Dye® (Uniscience, Sao Paulo, Brazil). Afterwards, PCR products were purified with 0.8 U/μL of exonuclease I enzyme (Fermentas, Waltham, MA, EUA) and 0.14 U/μL of alkaline phosphatase (ThermoFisher Scientific, Waltham, MA, EUA).
Sequencing was performed using Big Dye® Terminator Cycle Sequencing Standard v3.1 (Life Technologies, Carlsbad, CA, USA), according to manufacturer\'s instructions. The sequencing reactions were performed in a Mastercycler ep Gradient S thermocycler (Eppendorf, Hamburg, Germany) with a first step at 95°C for 60 s and 25 cycles of 95°C for 10 s, 50°C for 5 s, and 60°C for 4 min, followed by capillary electrophoresis in a 3500xl Genetic Analyzer Sequencer (Life Technologies, Carlsbad, CA, USA).
After sequencing, the alleles were identified according to the known alleles described at IMGT database (International ImMunoGeneTics Information System) ([@B18]). IMGT database provides public access to an integrated information system specialized in immunoglobulins (Ig), T cell receptors (TCR), and major histocompatibility complex (MHC) genes and molecules. All data submitted to the IMGT database are manually checked by experts in the field, which assure the deposit of high-quality data.
The nucleotide sequence of each individual was aligned with consensus sequences with Mutation Surveyor® DNA Variant Analysis Software v5.0.1 (Softgenetics), and their variable sites were annotated. Alleles that were different from the ones listed in the IMGT database were considered novel and were subsequently confirmed by sequencing and/or molecular cloning as described below.
The novel alleles that were observed in homozygosis (*IGHG1*^\*^*07, IGHG1*^\*^*08, IGHG2*^\*^*09, IGHG2*^\*^*13, IGHG3*^\*^*21, IGHG3*^\*^*22, IGHG3*^\*^*26*) were confirmed by direct re-sequencing from a different PCR product. Novel alleles observed in heterozygosis without phasing ambiguities due to the presence of only one heterozygous position (*IGHG1*^\*^*06, IGHG1*^\*^*09, IGHG1*^\*^*10, IGHG1*^\*^*12, IGHG1*^\*^*13, IGHG1*^\*^*14, IGHG2*^\*^*07, IGHG2*^\*^*10, IGHG2*^\*^*12, IGHG3*^\*^*20, IGHG3*^\*^*27, IGHG3*^\*^*28*) were also confirmed by re-sequencing. The new variants with ambiguous phasing (*IGHG1*^\*^*11, IGHG2*^\*^*08, IGHG2*^\*^*11, IGHG2*^\*^*14, IGHG2*^\*^*15, IGHG3*^\*^*23, IGHG3*^\*^*24, IGHG3*^\*^*25, IGHG3*^\*^*29*) were confirmed by molecular cloning. In this case, the segments were re-amplified and ligated into a PTZ57R/T vector (Fermentas, Waltham, MA, EUA) with terminal deoxynucleotidyl transferase (TdT) enzyme. Afterwards, recombinant plasmids were obtained and purified from multiple transformed colonies and sequenced as described above. Novel alleles were verified based on sequences from at least two independent colonies containing each allele.
Data Analysis
-------------
Allelic frequencies were obtained by direct counting using GenAlEx v6.502 software ([@B19]). Hardy-Weinberg equilibrium was tested for each gene segment in all populations by Guo and Thompson\'s method ([@B20]), performed in Arlequin v3.5.2 software ([@B21]). *IGHG* haplotypes from different gene segments were estimated via ELB algorithm and this information was used for Gm allotype haplotype inference, according to the correspondence between nucleotide variants and allotypes described by Lefranc et al. ([@B6]). Linkage disequilibrium (LD) between single nucleotide variants of each gene segment was estimated with Haploview software ([@B22]).
Allele networks were performed with variants from each gene segment through the median-joining (MJ) algorithm ([@B23]) with Network v5.0 software. Allele frequencies were compared using the exact test of population differentiation ([@B24]) and population-pairwise F~ST~ ([@B25], [@B26]) with Arlequin v3.5.2 software ([@B21]). Principal component analysis (PCA) was performed using the Minitab 17 Statistics Software ([@B27]) for graphical representation of the genetic differences and similarities in the major components of variation among populations. The PCA was performed using inferred allotype haplotype frequencies to compare the frequencies from the study population with others that were previously described serologically. These haplotypes were classified according to Lefranc et al. ([@B6]), and detailed information is available in [Table S3](#SM5){ref-type="supplementary-material"}.
Neutrality tests were performed using the Tajima\'s *D* ([@B28]), Fu and Li\'s *D*^\*^, *F*^\*^*, D*, and *F* ([@B29]) and Fay and Wu\'s *H* ([@B30]) in DnaSP software ([@B31]). Homologous gene segments from rhesus monkey were used as outgroup *(Macaca mulatta;* accession number: NW_001121238, AY292519, AY292512).
Results {#s3}
=======
One Novel Single Nucleotide Variant and 28 Novel *IGHG* Alleles Have Been Discovered
------------------------------------------------------------------------------------
Within all three gene segments in the seven populations analyzed, we found a total of 49 exonic variable sites, of which 26 were non-synonymous substitutions. Based on the Grantham scale ([@B32]), which ranges from 5 to 215 according to the physicochemical distance between amino acid pairs, amino acid replacements were from low to moderate (15 \< D \< 103) ([Table 1](#T1){ref-type="table"}). Of the single variable sites, 21 have not been reported in any of the previously described alleles at the IMGT database ([Table 1](#T1){ref-type="table"}, in bold). We also found a novel synonymous *IGHG3* single nucleotide variant at the position chr14:106235856 (GRCh37.p13 primary assembly) in the CTBA population. This new variant was submitted to the dbSNP database ([@B34]) under reference SNP ID number *rs155533833* (NC_000014.8:g.106235856G\>A).
######
Variable sites found in IGHG1, IGHG2, and IGHG3 gene segments.
**Gene segment** **Exon** **rsID** **Location**[^**a**^](#TN1A){ref-type="table-fn"} **IMGT numbering**[^**b**^](#TN1B){ref-type="table-fn"} **Eu numbering**[^**c**^](#TN1C){ref-type="table-fn"} **Nucleic acid substitution** **Amino acid substitution** **Allotype** **Grantham\'s *D***[^**d**^](#TN1D){ref-type="table-fn"} **Frequency**[^**e**^](#TN1E){ref-type="table-fn"}
------------------ ---------- ------------------- --------------------------------------------------- --------------------------------------------------------- ------------------------------------------------------- ------------------------------- ----------------------------- -------------- ---------------------------------------------------------- ----------------------------------------------------
*IGHG1* CH1 rs11552998 106209340 19 140 G\>A 0.006
**rs17850096** 106209289 40 157 G\>C 0.001
rs1071803 106209119 120 214 A\>G K\>R Gm17\>Gm3 26 0.143
CH2 **rs587690960** 106208471 22 260 A\>G 0.003
**rs377538050** 106208364 84.3 296 A\>T Y\>F 22 0.001
**rs193160354** 106208327 91 308 C\>T 0.011
**rs1043109** 106208326 92 309 C\>G L\>V 32 0.011
**rs1043249** 106208306 98 315 T\>C 0.011
CH3 **rs11557940** 106208107 5 349 C\>T 0.001
rs1045853 106208086 12 356 T\>G D\>E Gm1\>nGm1 45 0.118
rs11621259 106208082 14 358 C\>A L\>M Gm1\>nGm1 15 0.118
rs17841087 106207933 86 407 C\>T 0.114
rs113804727 106207862 110 431 C\>G A\>G nGm2\>Gm2 60 0.270
**rs370028332** 106207858 112 432 G\>C 0.001
rs8011686 106207843 117 437 G\>A 0.003
rs12879979 106207822 124 444 T\>C 0.106
*IGHG2* CH1 **rs189328740** 106111071 15 136 C\>T 0.023
**rs587648672** 106111069 16 137 A\>G E\>G 98 0.023
**rs773818177** 106111067 17 138 A\>G S\>G 56 0.023
rs11557955 106110966 82 171 A\>G 0.157
rs11627594 106110914 92 189 C\>A P\>T 38 0.105
CH2 rs8009156 106110137 45.1 282 G\>A V\>M Gm(.)\>Gm23 21 0.103
rs11160859 106110057 91 308 T\>C 0.163
**rs113678609** 106110056 92 309 G\>C V\>L 32 0.003
CH3 **rs587682450** 106109825 9 353 A\>C 0.017
rs4983499 106109752 38 378 G\>T A\>S 99 0.003
**rs368359789** 106109708 79 392 G\>C K\>N 94 0.001
rs1049810 106109702 81 394 A\>G 0.054
rs28371022 106109573 117 437 G\>A 0.106
*IGHG3* CH1 rs2983777 106237642 30 151 C\>A 0.001
**rs12050095** 106237624 40 157 G\>A 0.025
CH2 **rs138869693** 106236202 35 271 C\>T 0.006
**rs145035200** 106236195 38 274 C\>A Q\>K 53 0.006
rs74093865 106236143 82 291 C\>T P\>L nGm21\>Gm21 98 0.797
rs60746425 106236141 83 292 C\>T R\>W nGm16\>Gm16 101 0.048
rs12890621 106236128 84.3 296 A\>T Y\>F 22 0.123
**rs201027762** 106236035 110 327 C\>G A\>G 60 0.006
rs141959627 106236000 124 339 A\>G T\>A 58 0.006
CH3 rs189025987 106235895 1.4 341 A\>G 0.001
**rs147594653** 106235874 4 348 G\>A 0.001
**\*rs155533833** 106235856 10 354 C\>T 0.001
rs113169458 106235783 39 379 G\>A V\>M nGm15\>Gm15 21 0.047
rs77307099 106235767 44 384 G\>A S\>N Gm11\>nGm11 46 0.799
rs78376194 106235766 44 384 C\>T Gm11\>nGm11 0.799
**rs587739524** 106235758 45.2 387 C\>G P\>R 103 0.003
rs149653267 106235742 79 392 C\>G N\>K 94 0.052
rs139413052 106235729 84 397 A\>G M\>V Gm14\>nGm14 21 0.048
rs4042056 106235614 115 435 G\>A R\>H Gm5\>nGm5 29 0.085
rs1051112 106235611 116 436 T\>A F\>Y Gm5\>nGm5 22 0.847
*In bold, variant sites that have not been observed in any allele listed in the IMGT database. ^\*^Novel single nucleotide variation submitted to dbSNP database*.
Coordinate at chromosome 14 location (GRCh37.p13 primary assembly).
Amino acid position according to IMGT database (International ImMunoGeneTics Information System) ([@B18]).
According Edelman et al. ([@B33]).
Physicochemical distances between amino the amino acids involved in the substitution, according Grantham ([@B32]). The higher the value, the greater the differences, ranging from 5 to 215.
*Frequency of the alternative allele, merging all the samples of this study*.
A total of 28 novel *IGHG* alleles have been found in our study: nine in *IGHG1* ([Table 2](#T2){ref-type="table"}), nine in *IGHG2* ([Table 3](#T3){ref-type="table"}), and ten in *IGHG3* ([Table 4](#T4){ref-type="table"}). All novel alleles have been confirmed either by sequencing or by molecular cloning followed by sequencing. Novel alleles have been submitted to IMGT Nomenclature Committee ([@B18]), which verified the accuracy of our data and assigned official names (reports \#2018-2-0824 and \#2018-5-1113).
######
*IGHG1* alleles previous described and the 9 novel *IGHG1* alleles identified in this study.
**Exon** **CH1** **CH2** **CH3**
----------------- ------------------------------ ------------------------------------------------------- --------- ----- ------ --------- ------ ------ ----- ------ ----- --------- ------ ------ ----- ------ ------ ----- ----- ----- --------------------------------------
**IMGT unique numbering** 19 40 120 22 84.3 85,1 91 92 98 5 12 14 86 101 110 112 117 124
**Eu numbering**[^**a**^](#TN2A){ref-type="table-fn"} 140 157 214 260 296 301 308 309 315 349 356 358 407 422 431 432 437 444
**Amino acid change** A S K\>R T Y\>F R V L\>V N Y D\>E L\>M Y C\>I A\>G L T S
**Exonic position** 68 119 289 89 196 212 233 234 254 26 47 51 200 243 271 275 290 311
**Consensus nucleotide** G G A A A C C C T C T C C G C G G T
**Allele name** **GenBank accession number** **Allotype**[^**b**^](#TN2B){ref-type="table-fn"} [**\#**](#TN2C){ref-type="table-fn"}
*IGHG1\*01* -- 17,1 . . . . . . . . . . . . . . . . . . 0
*IGHG1\*02* -- 17,1 . . . . . T . . . . . . . . . . . . 391
*IGHG1\*03* -- 3 . . G . . T . . . . G A T . . . . C 81
*IGHG1\*04* -- 17,1,27 . . . . . T . . . . . . . A . . . . 0
*IGHG1\*05* -- 17,1 A . . . . T . . . . . . . . . . A . 2
***IGHG1\*06*** **MG920252** **3** . . G . . T . . . . G A T . . C . C 1
***IGHG1\*07*** **MG920245** **17,1,2** . . . . . T . . . . . . . . G . . . 189
***IGHG1\*08*** **MG920246** **3,1** . . G . . T . . . . . . . . . . . . 18
***IGHG1\*09*** **MG920247** **17,1** A . . . . T . . . . . . . . . . . . 2
***IGHG1\*10*** **MG920248** **17,1** . . . G . T . . . . . . . . . . . . 2
***IGHG1\*11*** **MG920249** **17,1** . . . . . T T G C . . . . . . . . . 7
***IGHG1\*12*** **MG920250** **17,1** . . . . . T . . . T . . . . . . . . 1
***IGHG1\*13*** **MG920251** **17,1** . . . . T T . . . . . . . . . . . . 1
***IGHG1\*14*** **MG920253** **17,1** . C . . . T . . . . . . . . . . . . 1
Novel alleles (in bold) have been confirmed by sequencing and/or molecular cloning. Their official names have been assigned by IMGT nomenclature committee. IMGT, International ImMunoGeneTics Information System ([@B18]). Dots represent the consensus nucleotide.
According to Edelman et al. ([@B33]).
llotypes were inferred according to Lefranc et al. ([@B6]).
*Number of copies observed in this study*.
######
*IGHG2* alleles previous described and the 9 novel *IGHG2* alleles identified in this study.
**Exon** **CH1** **CH2** **CH3**
----------------- ------------------------------ ------------------------------------------------------- --------- ------ ------ ----- ----- ------ ------ ------ --------- ----- ------ --------- ------ ------ ----- ----- --------------------------------------
**IMGT unique numbering** 15 16 17 19 82 92 95 96 45.1 91 92 9 38 79 81 117
**Eu numbering**[^**a**^](#TN3A){ref-type="table-fn"} 136 137 138 140 171 189 192 193 282 308 309 353 378 392 394 437
**Amino acid change** S E\>G S\>G A P P\>T N\>S F\>L V\>M V V\>L P A\>S K\>N T T
**Exonic position** 56 58 60 68 161 213 223 227 150 230 231 38 111 155 161 290
**Consensus nucleotide** C A A C A C A C G T G A G G A G
**Allele name** **GenBank accession number** **Allotype**[^**b**^](#TN3B){ref-type="table-fn"} [**\#**](#TN3C){ref-type="table-fn"}
*IGHG2\*01* (..) . . . . . . . . . . . . . . . . 0
*IGHG2\*02* 23 . . . G G A . . A C . . . . . A 71
*IGHG2\*03* (..) . . . G . . . . . . . . . . . . 557
*IGHG2\*04* (..) . . . G . . G G . . . . . . . . 0
*IGHG2\*05* (..) . . . G . . . . . . . . . . G . 0
*IGHG2\*06* (..) . . . G G . . . . C . . T . . A 2
***IGHG2\*07*** **MH025828** **(..)** . . . G . . . . . C . . . . . . 4
***IGHG2\*08*** **MH025829** **(..)** . . . G G . . . . C . . . . G . 30
***IGHG2\*09*** **MH025830** **(..)** T G G G . . . . . . . . . . . . 16
***IGHG2\*10*** **MH025831** **(..)** . . . G . . . . . . . C . . . . 12
***IGHG2\*11*** **MH025832** **(..)** . . . G G . . . . C C . . . G . 1
***IGHG2\*12*** **MH025833** **(..)** . . . G G . . . . . . . . . . . 3
***IGHG2\*13*** **MH025834** **(..)** . . . G . . . . . C . . . . G . 8
***IGHG2\*14*** **MH025835** **(..)** . . . G G . . . . C C . . . . . 1
***IGHG2\*15*** **MH025836** **(..)** . . . G G . . . . C . . . C G . 1
Novel alleles (in bold) have been confirmed by sequencing and/or molecular cloning. Their official names have been assigned by IMGT nomenclature committee. IMGT, International ImMunoGeneTics Information System ([@B18]). Dots represent the consensus nucleotide.
*According to Edelman et al. ([@B33])*.
*Allotypes were inferred according to Lefranc et al. ([@B6])*.
*Number of copies observed in this study*.
######
*IGHG3* alleles previous described and the 10 novel *IGHG3* alleles identified in this study.
**Exon** **CH1** **H1** **H2** **H3** **H4** **CH2**
----------------- ------------------------------ ------------------------------------------------------- --------- --------- --------- ---------- ---------- ---------- -------- -------- -------- -------- -------- --------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- --------------------------------------
**Eu numbering**[^**a**^](#TN4A){ref-type="table-fn"} **118** **151** **157** **176** **192** **193** **-** **-** **-** **271** **274** **291** **292** **296** **309** **327** **339**
**IMGT unique numbering** **1.4** **30** **40** **84,3** **95** **96** **10** **10** **13** **35** **38** **82** **83** **84,3** **92** **110** **124**
**Amino acid change** **S** **S** **S** **S\>Y** **S\>N** **L\>F** **P** **P** **R** **P** **Q\>K** **P\>L** **R\>W** **Y\>F** **L\>V** **A\>G** **T\>A**
**Exonic position** **2** **101** **119** **175** **223** **227** **29** **29** **36** **122** **129** **181** **183** **196** **234** **289** **324**
**Consensus nucleotide** **T** **C** **G** **C** **G** **G** **A** **G** **A** **C** **C** **C** **C** **A** **C** **C** **A**
**Allele name** **GenBank accession number** **Allotype**[^**b**^](#TN4B){ref-type="table-fn"} [**\#**](#TN4C){ref-type="table-fn"}
*IGHG3\*01* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . . 2
*IGHG3\*03* 5,6,11,24,26 . . . . . . . abs . . . . . . . . . . . 0
*IGHG3\*04* 5,10,11,13,14,26,27 C . . . . . . abs abs . . . . . . . . . . 0
*IGHG3\*05* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . . 0
*IGHG3\*06* 5,10,11,13,14,26,27 . A . . . . . . . . . . . . . . . . . 0
*IGHG3\*07* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . . 0
*IGHG3\*08* 5,14,26,27 . . . . . . . . . . . . . . . . . . . 0
*IGHG3\*09* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . G . . 0
*IGHG3\*10* 5,10,11,13,14,26,27 . A . . . . . . . . . . . . . . . . . 1
*IGHG3\*11* 5,10,11,13,14,26,27 . . . . . . . . . A C . . . . T . . . 76
*IGHG3\*12* 5,10,11,13,14,26,27 . . . . . . . . abs A C . . . . T . . . 7
*IGHG3\*13* 5,6,10,11,14,26,27 . . . . . . . . . . . . . . . . . . . 0
*IGHG3\*14* 21,26,27,28 . . . . . . . . . . . . . T . . . . . 529
*IGHG3\*15* 21,26,27,28 . . . . . . . . . . . . . T . . . . . 1
*IGHG3\*16* 21,26,27,28 . . . . . . . . . . . . . T . . . . G 5
*IGHG3\*17* 10,11,13,15,27 . . . . A C . abs G . . . . . . . . . . 0
*IGHG3\*18* 10,11,13,15,16,27 . . . A . . . abs G . . . . . T . . . . 0
*IGHG3\*19* 10,11,13,15,16,27 . . . . . . . abs G . . . . . T . . . . 31
***IGHG3\*20*** **MG920256** **21,26,27,28** . . . . . . . . . . . . . T . . . . . 2
***IGHG3\*21*** **MG920255** **5,10,11,13,14,26,27** . . A . . . . . . . . . . . . . . . . 18
***IGHG3\*22*** **MG920254** **21,27** . . . . . . . . . . . . . T . . . . . 26
***IGHG3\*23*** **MH025837** **10,11,13,16,27** . . . . . . . abs G . . . . . T . . . . 1
***IGHG3\*24*** **MG920257** **26,27,28** . . . . . . . . . . . . . . . . . . . 2
***IGHG3\*25*** **MG920258** **21,26,27,28** . . . . . . . . . . . T A T . . . . . 4
***IGHG3\*26*** **MG920259** **5,10,11,13,14,26,27** . . . . . . . . . . . . . . . T . G . 4
***IGHG3\*27*** **MG920260** **26,27,28** . . . . . . . . . . . . . T . . . . . 1
***IGHG3\*28*** **MG786813** **5,10,11,13,14,26,27** . . . . . . . . . A C . . . . T . . . 1
***IGHG3\*29*** **MG920261** **21,26,27,28** . . . . . . . . . . . . . T . . . . . 1
**Exon** **CH3**
----------------- ------------------------------ ------------------------------------------------------- --------- --------- --------- ---------- ---------- ---------- ---------- ---------- --------- ---------- ---------- --------- --------- ---------- --------- ---------- ---------- ---------- --------------------------------------
**IMGT unique numbering** **1,4** **4** **10** **39** **44** **44** **45,2** **79** **81** **84** **88** **89** **90** **98** **100** **101** **115** **116**
**Eu numbering**[^**a**^](#TN4A){ref-type="table-fn"} **341** **348** **354** **379** **384** **384** **387** **392** **394** **397** **409** **410** **411** **419** **421** **422** **435** **436**
**Amino acid change** **S** **V** **S** **V\>M** **S\>N** **S\>N** **P\>R** **N\>K** **SIN** **M\>V** **K\>R** **SIN** **SIN** **Q\>E** **SIN** **I\>V** **R\>H** **F\>Y**
**Exonic position** **2** **23** **41** **114** **130** **131** **139** **155** **161** **168** **205** **209** **212** **234** **242** **243** **283** **286**
**Consensus nucleotide** **A** **G** **C** **G** **G** **C** **C** **C** **G** **A** **A** **C** **C** **C** **C** **A** **G** **T**
**Allele name** **GenBank accession number** **Allotype**[^**b**^](#TN4B){ref-type="table-fn"} [**\#**](#TN4C){ref-type="table-fn"}
*IGHG3\*01* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . 2
*IGHG3\*03* 5,6,11,24,26 . . . . . . . . . G G A . G T G . . 0
*IGHG3\*04* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . 0
*IGHG3\*05* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . 0
*IGHG3\*06* 5,10,11,13,14,26,27 G . . . . . . G . . . . . . . . . . 0
*IGHG3\*07* 5,10,11,13,14,26,27 G . . . . . . G . . . . . . . . . . 0
*IGHG3\*08* 5,14,26,27 . . . . A T . . . . . . . . . . . . 0
*IGHG3\*09* 5,10,11,13,14,26,27 . . . . . . . . . . . . T . . . . . 0
*IGHG3\*10* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . 1
*IGHG3\*11* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . 76
*IGHG3\*12* 5,10,11,13,14,26,27 . . . . . . . . . . . . . . . . . . 7
*IGHG3\*13* 5,6,10,11,14,26,27 . . . . . . . G A . . . . G . . . . 0
*IGHG3\*14* 21,26,27,28 . . . . A T . . . . . . . . . . . A 529
*IGHG3\*15* 21,26,27,28 . . . . A T . G . . . . . . . . . A 1
*IGHG3\*16* 21,26,27,28 . . . . A T . . . . . . . . . . . A 5
*IGHG3\*17* 10,11,13,15,27 . . . A . . . G . G . . . . . . A A 0
*IGHG3\*18* 10,11,13,15,16,27 . . . A . . . G . G . . . . . . A A 0
*IGHG3\*19* 10,11,13,15,16,27 . . . A . . . G . G . . . . . . A A 31
***IGHG3\*20*** **MG920256** **21,26,27,28** . . . . A T G . . . . . . . . . . A 2
***IGHG3\*21*** **MG920255** **5,10,11,13,14,26,27** . . . . . . . . . . . . . . . . . . 18
***IGHG3\*22*** **MG920254** **21,27** . . . . A T . . . . . . . . . . A A 26
***IGHG3\*23*** **MH025837** **10,11,13,16,27** . . . . . . . G . G . . . . . . A A 1
***IGHG3\*24*** **MG920257** **26,27,28** . . . . A T . G . . . . . . . . . A 2
***IGHG3\*25*** **MG920258** **21,26,27,28** . . . . A T . . . . . . . . . . . A 4
***IGHG3\*26*** **MG920259** **5,10,11,13,14,26,27** . . . . . . . . . . . . . . . . . . 4
***IGHG3\*27*** **MG920260** **26,27,28** . A . . A T . . . . . . . . . . . A 1
***IGHG3\*28*** **MG786813** **5,10,11,13,14,26,27** . . T . . . . . . . . . . . . . . . 1
***IGHG3\*29*** **MG920261** **21,26,27,28** G . . . A T . . . . . . . . . . . A 1
Novel alleles (in bold) have been confirmed by sequencing and/or molecular cloning. Their official names have been assigned by IMGT nomenclature committee. IMGT, International ImMunoGeneTics Information System ([@B18]). Dots represent the consensus nucleotide; abs, absent.
*According to Edelman et al. ([@B33])*.
*Allotypes were inferred according to Lefranc et al. ([@B6])*.
*Number of copies observed in this study*.
Interestingly, some new alleles of all gene segments were observed at high frequency (*f* \> 0.10; [Table 5](#T5){ref-type="table"}). The highest frequencies for novel alleles were observed for *IGHG1*^\*^*07* in GKW (*f* = 0.478; 34 individuals), *IGHG1*^\*^*08* in BrJAP (*f* = 0.155; 15 individuals), *IGHG2*^\*^*08* in BrJAP (*f* = 0.202; 23 individuals), *IGHG2*^\*^*09* in GRC (*f* = 0.137; 9 individuals), *IGHG3*^\*^*21* in BrJAP (*f* = 0.158; 16 individuals), and *IGHG3*^\*^*22* in GRC (*f* = 0.157; 15 individuals).
######
One third of the novel *IGHG* alleles were observed in high frequencies (0.05 \< *f* \< 0.48).
**GKW** **GND** **GRC** **KIV** **KRC** **BrJAP** **CTBA**
----------------- -------------------------- --------- --------- --------- --------- --------- ----------- -----------
*IGHG1\*02* 0.522 0.438 0.725 0.770 0.606 0.600 0.228
*IGHG1\*03* 0.094 0.020 0.020 0.029 0.707
*IGHG1\*05* 0.022
***IGHG1\*06*** 0.011
***IGHG1\*07*** 0.478 0.469 0.235 0.200 0.365 0.136 0.033
***IGHG1\*08*** 0.010 0.155
***IGHG1\*09*** 0.018
***IGHG1\*10*** 0.018
***IGHG1\*11*** 0.064
***IGHG1\*12*** 0.010
***IGHG1\*13*** 0.009
***IGHG1\*14*** 0.010
**Sample size** **46** **48** **51** **52** **52** **57** **47**
**HW** ***p*****-value** **1** **1** **1** **1** **1** **0.232** **0.146**
*IGHG2\*02* 0.083 0.020 0.038 0.029 0.114 0.436
*IGHG2\*03* 0.957 0.885 0.735 0.942 0.952 0.579 0.489
*IGHG2\*06* 0.021
***IGHG2\*07*** 0.033 0.010
***IGHG2\*08*** 0.010 0.019 0.010 0.202 0.032
***IGHG2\*09*** 0.011 0.010 0.137
***IGHG2\*10*** 0.021 0.098
***IGHG2\*11*** 0.009
***IGHG2\*12*** 0.018 0.011
***IGHG2\*13*** 0.061 0.011
***IGHG2\*14*** 0.009
***IGHG2\*15*** 0.009
**Sample size** **46** **48** **51** **52** **51** **57** **51**
***Allele*** **HW** ***p*****-value** **-** **1** **1** **1** **1** **0.951** **0.519**
*IGHG3\*01* 0.020
*IGHG3\*10* 0.010
*IGHG3\*11* 0.094 0.010 0.038 0.020 0.588
*IGHG3\*12* 0.069
*IGHG3\*14* 1.000 0.844 0.833 0.952 0.971 0.474 0.186
*IGHG3\*15* 0.009
*IGHG3\*16* 0.010 0.039
*IGHG3\*19* 0.272
***IGHG3\*20*** 0.018
***IGHG3\*21*** 0.158
***IGHG3\*22*** 0.063 0.157 0.010 0.029
***IGHG3\*23*** 0.009
***IGHG3\*24*** 0.018
***IGHG3\*25*** 0.035
***IGHG3\*26*** 0.039
***IGHG3\*27*** 0.010
***IGHG3\*28*** 0.010
***IGHG3\*29*** 0.009
*Novel alleles (in bold) have been confirmed by sequencing and/or molecular cloning. Their official names have been assigned by IMGT nomenclature committee. IMGT, International ImMunoGeneTics Information System ([@B18]); KIV, Kaingang from Ivaí; KRC, Kaingang from Rio das Cobras; GRC, Guarani Mbya; GKW, Guarani Kaiowa; GND, Guarani Ñandeva; BrJAP, Japanese-descendant from Curitiba; CTBA, Euro-descendants from Curitiba*.
Because most of the previous studies only described the immunoglobulin heavy chain diversity serologically, we inferred the serological Gm allotypes from our nucleotide sequence data, based on the nucleotide sequence description for each previously reported allotype ([@B6]), to allow comparison with previously reported variants. For example, the most frequent allele haplotype (alleles that are in the same chromosome and inherited together in a block) was the one comprising the gene segments *IGHG1*^\*^*02, IGHG2*^\*^*03, IGHG3*^\*^*14* (*f* = 0.182 to 0.740), which encodes the Gm haplotype "C" Gm21,26,27,28;17,1;(.), the most frequent lgG allotype haplotype in our populations (*f* = 0.21 to 0.77; [Table 6](#T6){ref-type="table"}). The correspondence between allele haplotype and allotype haplotypes are in the [Table S4](#SM6){ref-type="supplementary-material"}. More than one *IGHG* allele haplotype can define a single Gm allotype haplotype, as is the case of the Gm haplotype "B" Gm5,10,11,13,14,26,27;3;(.), that is encoded in our data by the allele haplotype *IGHG3*^\*^*11,IGHG2*^\*^*03,IGHG2*^\*^*03*, by *IGHG3*^\*^*11,IGHG1*^\*^*14,IGHG2*^\*^*03*, or by *IGHG3*^\*^*11,IGHG1*^\*^*03,IGHG2*^\*^*08*. In order to simplify the interpretation of the data, Gm haplotype identifiers (from A to M) were used as suggested by Lefranc et al. ([@B6]).
######
Gm allotype haplotypes frequencies inferred from nucleotide sequencing.
**GKW** **GND** **GRC** **KIV** **KRC** **BrJAP** **CTBA**
--- ------------------------------ --------- --------- --------- --------- --------- ----------- ----------
A 5,10,11,13,14,26,27;3;23 0.083 0.01 0.02 0.02 0.444
B 5,10,11,13,14,26,27;3;(.) 0.01 0.277
C 21,26,27,28;17,1;(.) 0.522 0.375 0.588 0.77 0.608 0.384 0.211
D 21,26,27,28;17,1,2;(.) 0.478 0.468 0.235 0.2 0.363 0.134 0.011
I 10,11,13,15,16,27;17,1;(.) 0.286
J 5,10,11,13,14,26,27;3,1;23 0.01 0.107
K 5,10,11,13,14,26,27;3,1;(.) 0.045
21,26,27,28;3;23 0.01
5,10,11,13,14,26,27;17,1;23 0.009
10,11,13,16,27;17,1;(.) 0.009
21,27;17,1,2;(.) 0.022
21,27;17,1;(.) 0.063 0.157 0.01 0.011
5,10,11,13,14,26,27;17,1;(.) 0.022
Allotype haplotype ID are as described by Lefranc et al. ([@B6]).
*KIV, Kaingang from Ivaí; KRC, Kaingang from Rio das Cobras; GRC, Guarani Mbya; GKW, Guarani Kaiowa; GND, Guarani Ñandeva; BrJAP, Japanese-descendants; CTBA, Euro-descendants from Curitiba*.
Lower *IGHG* Diversity Was Observed in Amerindians and Frequencies Differed Significantly Among Populations
-----------------------------------------------------------------------------------------------------------
*IGHG* allelic frequencies varied across populations ([Table 5](#T5){ref-type="table"}). A small number of highly frequent alleles were observed for all gene segments in Amerindian populations. Even though Guarani populations share a more recent common ancestor, allelic frequencies significantly differed among them (*p* \< 0.01), with low to moderate F~ST~ values (0.02--0.10) ([Table 7](#T7){ref-type="table"}). Allelic frequencies did not differ between the two Kaingang populations (*p* = 0.065; F~ST~ = 0.03). More conspicuous differences were found between the Japanese-descendant and Euro-descendant populations compared to each other, and between each of these two populations compared to the Amerindian populations, with F~ST~ values ranging from 0.11 to 0.52, indicating moderate to high genetic differentiation.
######
Genetic differentiation for *IGHG1, IGHG2*, and *IGHG3* among populations.
**GKW** **GND** **GRC** **KIV** **KRC** **BrJAP** **CTBA**
----------- --------- --------- --------- --------- --------- ----------- ----------
**GKW** \*\* \*\*\* \*\*\* ns \*\*\* \*\*\*
**GND** 0.02828 \*\*\* \*\*\* \*\* \*\*\* \*\*\*
**GRC** 0.10738 0.07720 \*\*\* \*\*\* \*\*\* \*\*\*
**KIV** 0.10494 0.11120 0.05722 ns \*\*\* \*\*\*
**KRC** 0.01042 0.03816 0.06889 0.03220 \*\*\* \*\*\*
**BrJAP** 0.21496 0.15144 0.11437 0.18492 0.19168 \*\*\*
**CTBA** 0.51577 0.38142 0.41023 0.51270 0.50134 0.28576
*Upper diagonal: the statistical significance of the exact test of population differentiation between pairs of population. \*\*p \< 0.001--0.01; \*\*\*p \< 0.001; ns p \> 0.05. Lower diagonal: F~ST~ values between populations. Wright\'s F~ST~ scale: 0--0.05 (in very light gray), 0.05--0.15 (in light gray), 0.15--0.25 (in medium gray), and above 0.25 (in dark gray) indicate little, moderate, great, and very great genetic differentiation between populations. KIV, Kaingang from Ivaí; KRC, Kaingang from Rio das Cobras; GRC, Guarani Mbya; GKW, Guarani Kaiowa; GND, Guarani Ñandeva; BrJAP, Japanese-descendants; CTBA, Euro-descendants from Curitiba*.
The principal component analysis (PCA) grouping was consistent with ancestry and geography ([Figure 2](#F2){ref-type="fig"}). Amerindians and Asians formed two separated groups close to each other. Europeans and admixed populations of predominantly European ancestry grouped together, while Africans were more distant.
![Principal component analysis using Gm allotype haplotype frequencies was consistent with geography and ancestry. For comparisons with previously described population, we inferred the Gm allotype frequencies based on the observed nucleotide sequences, according to Lefranc et al. ([@B6]). Circles represent population data from the literature and squares represent populations from the present study. All frequencies reported in the literature are listed in [Table S3](#SM5){ref-type="supplementary-material"}. AFR, African populations; AMER, Amerindian populations; ASIA, Asian populations; EUR, European populations; EUR-BR, Euro-descendant populations from Brazil; ADM-BR, Admixed population from Brazil; KIV, Kaingang from Ivaí; KRC, Kaingang from Rio das Cobras; GRC, Guarani Mbya; GKW, Guarani Kaiowa; GND, Guarani Ñandeva; BrJAP, Japanese-descendants; CTBA, Euro-descendants from Curitiba.](fimmu-10-01161-g0002){#F2}
Genotypic distributions for all gene segments were in accordance with Hardy-Weinberg equilibrium in all population samples (0.08 \> *p* \>1).
Distinct Linkage Disequilibrium Patterns Among Populations
----------------------------------------------------------
Linkage disequilibrium (LD) patterns differed among populations ([Figure S2](#SM2){ref-type="supplementary-material"}). Interestingly, each Guarani population exhibited a distinct LD pattern despite their close relationship. In GKW, only five variable sites were observed in all three gene segments, of which three were in absolute LD (*D*′ = 1, *r*^2^ = 1). In contrast, more variable sites (21 and 24) were observed for the other two Guarani populations. In addition, many variants that were in LD in GND were not observed in LD in GRC.
The G1m3 allotype (*rs1071803*) and the G2m23 allotype (*rs800915*6) were in strong LD in all Amerindian populations (*D*′ = 1; *r*^2^ \> 0.87), as well as in CTBA (*D*′ = 1; *r*^2^ = 0.43), and BrJAP (*D*′ = 0.73; *r*^2^ = 0.92) in which fewer SNPs were observed in strong LD.
Sequence Analysis Suggests That Gene Conversion Between Frequent Alleles of Different Gene Segments Generated Novel Alleles
---------------------------------------------------------------------------------------------------------------------------
Median-Joining network ([Figure 3](#F3){ref-type="fig"}) shows that the most frequent alleles *IGHG1*^\*^*01, IGHG2*^\*^*03*, and *IGHG3*^\*^*14* were central nodes in the network, with few nucleotides differing between them and the other alleles. The loops indicate possible recombination sites.
{#F3}
Alignment of all the known alleles of the *IGHG1, IGHG2*, and *IGHG3* gene segments suggests that some novel alleles discovered in this study could have been generated by gene conversion between alleles of different gene segments ([Figure 3](#F3){ref-type="fig"}). For example, the novel allele *IGHG1*^\*^*11*, present in BrJAP (*f* = 0.064), could have been generated by gene conversion between the most frequent *IGHG2* allele (*IGHG2*^\*^*03; f* = 0.579) and the most frequent *IGHG1* allele (*IGHG1*^\*^*02*; *f* = 0.60). In addition, gene conversion between the frequent *IGHG2*^\*^*03* and *IGHG3*^\*^*14* alleles (*f* = 0.735 and *f* = 0.833, respectively) could explain the origin of allele *IGHG2*^\*^*09* (*f* = 0.14).
Neutrality Tests Suggest Evidence of Natural Selection Shaping *IGHG* Polymorphism
----------------------------------------------------------------------------------
Neutrality tests performed by Tajima\'s *D*, Fu and Li\'s *D* and *F* were non-significant for most populations. However, Fay and Wu\'s test resulted in significant negative values for most populations, which may indicate positive selection at an adjacent site ([Table 8](#T8){ref-type="table"}).
######
Fay and Wu\'s test was significant in the majority of the study populations.
**GKW** **GND** **GRC** **KIV** **KRC** **BrJAP** **CTBA**
--------- ----------------- ---------------------------------------------- -------------------------------------------- -------------------------------------------- ---------------------------------------------- ---------------------------------------------- ------------------------------------------ --------------------------------------------
*IGHG1* Tajima\'s D -- -- -- -- -- -- --
Fu and Li\'s D -- -- -- -- -- -- --
Fu and Li\'s F -- -- -- -- -- -- --
Fay and Wu\'s H -- −2.455[^\*^](#TN8A){ref-type="table-fn"} −3.400[^\*\*^](#TN8B){ref-type="table-fn"} −3.384[^\*\*^](#TN8B){ref-type="table-fn"} −3.490[^\*\*^](#TN8B){ref-type="table-fn"} −2.36[^\*^](#TN8A){ref-type="table-fn"} --
*IGHG2* Tajima\'s D −1.680[^\*\*\*^](#TN8C){ref-type="table-fn"} -- -- -- −1.716[^\*\*\*^](#TN8C){ref-type="table-fn"} -- --
Fu and Li\'s D −2.696[^\*\*^](#TN8B){ref-type="table-fn"} -- -- -- -- -- --
Fu and Li\'s F −2.793[^\*\*^](#TN8B){ref-type="table-fn"} -- -- -- -- -- --
Fay and Wu\'s H −1.768[^\*\*^](#TN8B){ref-type="table-fn"} −2.745[^\*^](#TN8A){ref-type="table-fn"} −4.971[^\*\*^](#TN8B){ref-type="table-fn"} −5.241[^\*\*\*^](#TN8C){ref-type="table-fn"} −5.518[^\*\*\*^](#TN8C){ref-type="table-fn"} -- --
*IGHG3* Tajima\'s D -- -- -- -- −1.825[^\*\*\*^](#TN8C){ref-type="table-fn"} -- --
Fu and Li\'s D -- -- -- -- -- -- −3.615[^\*\*^](#TN8B){ref-type="table-fn"}
Fu and Li\'s F -- -- -- -- -- -- −3.309[^\*\*^](#TN8B){ref-type="table-fn"}
Fay and Wu\'s H -- −4.310[^\*\*^](#TN8B){ref-type="table-fn"} -- −7.253[^\*\*\*^](#TN8C){ref-type="table-fn"} −5.670[^\*\*\*^](#TN8C){ref-type="table-fn"} −6.307[^\*^](#TN8A){ref-type="table-fn"} −7.372[^\*\*^](#TN8B){ref-type="table-fn"}
Statistical significance was tested by coalescent simulations with 10,000 repetitions:
p \< 0.01−0.05;
p \< 0.001−0.01;
p \< 0.001;
*--p\>0.05. KIV, Kaingang from Ivaí; KRC, Kaingang from Rio das Cobras; GRC, Guarani Mbya; GKW, Guarani Kaiowa; GND, Guarani Ñandeva; BrJAP, Japanese-descendants; CTBA, Euro-descendants from Curitiba*.
Deviation of neutrality was also tested by analyzing synonymous and non-synonymous substitution rates across all the known and novel alleles of all gene segments ([Tables S5](#SM7){ref-type="supplementary-material"}--[S7](#SM9){ref-type="supplementary-material"}). Overall, the rate of synonymous substitutions (dS) was significantly higher than the rate of non-synonymous (dN) substitutions (dN/dS \< 1) for *IGHG1* and *IGHG2* (*p* = 0.01 and 0.032, respectively) ([Table 9](#T9){ref-type="table"}), consistent with purifying selection.
######
Codon-based test indicates purifying selection shaping *IGHG1* and *IGHG2* variation.
***IGHG1*** ***IGHG2*** ***IGHG3***
-------------------------------- ------------- ------------- -------------
Purifying selection (dN \< dS) *p* = 0.010 *p* = 0.032 ns
Positive selection (dN \> dS) ns ns ns
*ns, p \> 0.05*.
Discussion {#s4}
==========
Our main goal was to deliver an unprecedented and comprehensive nucleotide sequencing-based characterization of the *IGHG* gene segments in populations of different ancestries. Before this study, only 30 *IGHG* alleles have been described for *IGHG1, IGHG2*, and *IGHG3* together ([@B18]). Here, we report the discovery of 28 novel alleles, of which 16 were in a single population sample of Japanese descendants (*n* = 57) and seven in one population sample of Euro-descendants (*n* = 51). It is interesting that even in Amerindian populations, which exhibited a limited diversity, seven new alleles were found. This is clear evidence that the diversity of *IGHG* is far from being fully described and possibly a much larger number of novel alleles will be discovered as more populations are interrogated. We focused on the segments that code for the most abundant Ig in serum. Considering the homology and high sequence similarity, a different strategy would be needed for the precise characterization of *IGHG4* due to the high frequency of duplications observed for this gene segment ([@B35]).
Some of the new alleles were highly frequent. The novel allele *IGHG3*^\*^*22*, frequent in Guarani Mbya (GRC, *f* = 0.157), exhibited a lower frequency in Guarani Ñandeva (GND, *f* = 0.063), and was absent in Guarani Kaiowa (GKW). These three populations share a more recent common ancestor and the differences observed can be explained by its demographic history and genetic drift. Demographic factors played a major role in shaping the diversity of other genes important for immune responses in these same Amerindian populations ([@B36]). Genetic drift, particularly founder effect and bottleneck, may explain the lower diversity of *IGHG* in Amerindians and the fluctuation of their allelic frequencies. On the other hand, the *IGHG3*^\*^*22* allele was observed only in one Kaingang individual. This fact suggests gene flow from Guarani to Kaingang. Although GRC and KRC remain isolated due to strong cultural barriers, their immediate vicinity did result in a low degree of admixture ([@B14]).
*IGHG3*^\*^*11* is the most common *IGHG3* allele in Euro-descendants (CTBA, *f* = 0.588) and was observed at lower frequency in Amerindians: GND (*f* = 0.094), GRC (*f* = 0.010), KIV (*f* = 0.038), KRC (*f* = 0.020), being absent in GKW. This allele corresponds to the allotype G3m5,10,11,13,14,26,27 which has been previously shown to be highly frequent in Europeans but absent in non-admixed Amerindians ([@B37]--[@B42]). Also, similar allele distribution was observed for *IGHG1*^\*^*03* in the study populations. These observations are consistent with previous studies from our group, which estimated the admixture rate of Guarani and Kaingang by analyzing *HLA* class II genes. In that study, the estimated admixture rate with non-Amerindians was 14.3% for GND, 3.7% for GRC, 7.2% for Kaingang, and no admixture for GKW ([@B15]). The Gm allotype haplotype frequencies inferred from DNA sequencing in our study (in which the most common haplotypes were C and D) were similar to those found in former reports that characterized serologically the Guarani and Kaingang populations from Santa Catarina State, Brazil ([@B42]), and other native American populations ([@B41], [@B43], [@B44]).
The new allele *IGHG3*^\*^*21* was frequent in BrJAP (*f* = 0.158), but absent in the other populations. According to the nucleotide sequence, it encodes the haplotype Gm5,10,11,13,14,26,27, whose frequency was previously reported as 15.2% in a study with Japanese families ([@B45]). In that same study, the haplotypes C (Gm21;17,1;(.) -- 40.7%), D (Gm21;17,1,2;(.) -- 16.4%), I (Gm11,13,15,16;17,1;(.) -- 27.7%), and J (Gm5,11,13;3,1;23 -- 15.2%) exhibited similar frequencies to the ones inferred from DNA sequencing in BrJAP, which were 38.4%, 13.4%, 28.6%, and 10.7%, respectively ([Table 6](#T6){ref-type="table"}). The novelty of our results is showing, for the first time, the characterization of the variants at DNA level that are responsible for the occurrence of these Gm haplotypes in Japanese populations.
Strong linkage disequilibrium (LD) ([Figure S2](#SM2){ref-type="supplementary-material"}) was observed in most Amerindian populations, as expected for these historically small populations that suffered strong genetic drift and multiple founder effects since the arrival of the first Americans to the continent and during their migration from the North to the South in the American continent. Interestingly, the patterns of LD differed among Guaranis, despite their shared ancestry. GKW exhibited a reduced number of variable sites, while GRC exhibited a reduced LD in comparison to GND. These differences could also be explained by genetic drift, as certain haplotypes that stochastically increased their frequencies in a population after their divergence may not have increased in the others.
In contrast, the Japanese-descendant and Euro-descendant populations have higher nucleotide and allele diversity and fewer SNPs in LD. Even so, SNPs from different gene segments are in LD in these urban populations. In BrJAP, SNPs of allotypes G1m17 (*rs1071803*) and G2m(.) (*rs8009156*) are in LD (*D* = 0.92; *r*^2^ = 0.73) and are present in the allotype haplotypes C and D, reported as the most common in Japanese populations ([@B45]).
In the MJ networks ([Figure 3](#F3){ref-type="fig"}), *IGHG1*^\*^*02, IGHG2*^\*^*03* and *IGHG3*^\*^*14* were connected with most alleles and were present at high frequencies in all populations. This pattern suggests that most of the other known alleles could have been originated from them. In the *IGHG2* MJ network, one loop shows two paths where substitutions at position 161 of exon CH3 and 230 of CH2 occurred to generate the *IGHG2*^\*^*05*, ^\*^*07*, and ^\*^*13* alleles. It can be hypothesized that a mutation occurred in one of them, for example, *IGHG2*^\*^*03* at position 161 of exon CH3, generated *IGHG2*^\*^*05* and this allele, likewise, might have mutated at position 230 of exon CH2 originating allele *IGHG2*^\*^*13*. As independent mutations in the same positions are extremely unlikely, the fact that the *IGHG2*^\*^*07* allele has a variant in the same position (230 of CH2 exon) indicates that gene conversion between alleles *IGHG2*^\*^*13* and *IGHG2*^\*^*03* originated the *IGHG2*^\*^*07* allele. Moreover, we suggest that the novel alleles *IGHG1*^\*^*11* and *IGHG2*^\*^*09* resulted from gene conversion between two frequent alleles of different gene segments. Overall, our data point to a major role of recombination and gene conversion originating new *IGHG* alleles, which is consistent with the tandem positioning and high sequence similarity of these segments, which favor unequal crossing-over ([@B46]).
Kaingang from Ivaí and Kaingang from Rio das Cobras presented low genetic differentiation (F~ST~ = 0.032), and similar allele frequencies ([Table 7](#T7){ref-type="table"}), most probably because of their recent common origin and gene flow due to the absence of cultural barriers, in addition to their geographical proximity. The F~ST~ values between the Guarani populations were low to moderate, which is an evidence of genetic drift affecting the *IGHG* diversity in these populations. These results are compatible with previous reports for mtDNA in the same populations, which indicated that divergence of the three Guarani populations occurred at around 1,800 years before present (ybp), much earlier than the separation of the Kaingang populations that was estimated at of 207 ybp ([@B13]).
The PCA results ([Figure 2](#F2){ref-type="fig"}) were consistent with geography and ancestry and showed that our data are consistent with data obtained by serologic methods, previously reported in the literature. The exception was India, which grouped with Europeans and Euro-descendants. In fact, PCA grouping does not necessarily mean common ancestry, as it can also result from migration or stochastic factors, or convergent evolution by natural selection. The grouping solely reflects the similarities of the *IGHG* allelic frequencies in these populations.
The results of most neutrality tests suggested that natural selection is not the major factor responsible for shaping *IGHG* diversity in the study populations. In other words, for *IGHG* the impact of genetic drift due to demographical processes is possibly stronger than the signal left by natural selection. As is known, Amerindians have a long history of migrations and isolation, and went through severe bottlenecks after the European colonization ([@B14]). Still, in GKW and KRC for *IGHG2* and KRC and CTBA for *IGHG3*, the results of Tajima\'s *D*, and Fu and Li\'s *D* and *F* tests indicated diversity sweeps due to bottlenecks or purifying selection.
Analyzing all the currently known *IGHG* alleles, including the 28 novel alleles that we here described, we found that the codon-based dN/dS test showed significant results for purifying selection ([Table 9](#T9){ref-type="table"}) for *IGHG1* (*p* = 0.01) and *IGHG2* (*p* = 0.03). We observed that synonymous (dS) substitution rates were higher than non-synonymous (dN) substitution rates. It was previously demonstrated that Gm1 allotypes have a different impact on the IgG1 ability to bind the Fc gamma receptor (FcγR)-like proteins from viruses. Antibodies with G1m1,2,17 allotype exhibit lower affinity to the viral FcγR-like protein of the human cytomegalovirus (HCMV), which decreases susceptibility to this infection ([@B47]). Similarly, the FcγR-like protein from herpes simplex virus (HSV) binds with lower affinity to antibodies carrying the G1m3 allotype due to certain residues in the CH1 and CH3 domains ([@B9]). In the light of our results, it is plausible to suggest that emerging amino acid replacements that favored binding to viral proteins were negatively selected as a result of their deleterious effect for the individuals carrying the mutations. Higher binding to these viral proteins would favor viral evasion from immune responses and increase the susceptibility to certain viral infections. Moreover, purifying selection against non-synonymous changes could have limited the diversification of *IGHG1* and *IGHG2*.
The Fay and Wu *H* test was significant with negative values for almost every population and gene segment analyzed. This could be interpreted as a result of an excess of derived variants at high frequencies in the gene genealogies. Fay and Wu ([@B30]) suggested that this may be a unique pattern produced by hitchhiking of variants in the vicinity that are being favored by positive selection. *IGHG* gene segments are located downstream of the *IGHV, IGHD*, and *IGHJ* gene segments that encode the immunoglobulin variable regions, which specifically bind to antigens ([@B2], [@B4]). Therefore, we suggest that selection for variants in the variable region may be impacting the diversity of the constant region by hitchhiking mutations in the *IGHG* gene segments. This hypothesis is corroborated by the findings of Tanaka and Nei ([@B48]), who demonstrated that the non-synonymous mutation rate was higher than the synonymous rate in the gene segments that code for the Ig variable region. Their results were consistent with diversity-enhancing selection or overdominant selection driving the nucleotide diversity in the variable region.
Conclusion {#s5}
==========
Antibodies are pivotal for human survival, at both the individual and the population levels. It is surprising that despite decades of compelling evidence about the importance of the immunoglobulin gene variation for human immunity and the not so recent advent of sequencing technologies, most of the knowledge about *IGHG* is still based on serologic typing. As we see here, the fact that the regions encoded by *IGHG* are called "constant" does not mean these segments are not highly polymorphic. In fact, we found 16 novel alleles in a population sample of only 57 Japanese descendants. The *IGHG* genomic region is not well-covered in genome-wide association studies and whole genome sequencing databases. The homology and high sequence similarity of *IGHG* segments impose technical difficulties for sequencing, particularly at large scale. Besides, the somatic recombination events characteristic of the *IGH* locus makes DNA from B-cell lines, used in so many studies, not suitable for *IGHG* sequencing. Our study is the first to sequence systematically these segments at the nucleotide level in populations. We here present a full characterization of *IGHG1-3* diversity in seven Brazilian populations, linkage disequilibrium, haplotypes and evidence of purifying selection and genetic drift. Understanding the *IGHG* normal variation in populations and its evolution may be the key to better comprehend how the immune system fights invading organisms and non-self-antigens and also may contribute to the development of new vaccines.
Ethics Statement {#s6}
================
This study was carried out in accordance with the recommendations of Brazilian National Human Research Ethics Committee (CONEP) with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Brazilian National Human Research Ethics Committee (CONEP).
Author Contributions {#s7}
====================
DA designed the study. VC-S, DA, LV, RD, and HI performed DNA sequencing and genotyping. VC-S analyzed the data. RW, VC-S, RD, HI, and LV performed molecular cloning and validation of novel alleles. MP-E, DA, DM, and RW contributed with reagents. VC-S, DA, MP-E, DM, and MB drafted the manuscript. All authors significantly contributed with ideas and critically reviewed this manuscript.
Conflict of Interest Statement
------------------------------
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Kirsten M. Anderson for kindly reviewing this manuscript.
**Funding.** This work was supported by grants from the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), PRONEX, Fundação Araucária de Apoio ao Desenvolvimento Científico e Tecnológico do Paraná, and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES).
Supplementary Material {#s8}
======================
The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fimmu.2019.01161/full#supplementary-material>
######
Location of IGHG1, IGHG2, and IGHG3 primers. Arrows indicate primers and their direction. All primer sequences are listed in [Table S2](#SM4){ref-type="supplementary-material"}. Primers used for amplification and sequencing are shown in blue, green, and red; primers only used for sequencing are represented in gray. This representation is not to scale.
######
Click here for additional data file.
######
Linkage disequilibrium among SNPs within *IGHG3, IGHG1*, and *IGHG2*. The maximum values of D\'/ LOD are represented in red and the minimum values are in gray and white. Within each rhomb are represented values of *r*^2^ (×100); when it is empty (no values shown) it indicates *r*^2^ = 100 (absolute linkage). KIV, Kaingang from Ivaí; KRC, Kaingang from Rio das Cobras; GRC, Guarani Mbya; GKW, Guarani Kaiowa; GND, Guarani Ñandeva; BrJAP, Japanese-descendants; CTBA, Euro-descendants from Curitiba.
######
Click here for additional data file.
######
The seven population samples analyzed in this study.
######
Click here for additional data file.
######
Description of amplification and sequencing primers.
######
Click here for additional data file.
######
Allotype haplotype frequencies of populations from various continents available in the literature.
######
Click here for additional data file.
######
Correspondence between haplotypes of *IGHG* alleles ad haplotypes of Gm allotypes and their observed frequencies.
######
Click here for additional data file.
######
Statistical significance of codon-based test of selection between sequences of *IGHG1* gene segment.
######
Click here for additional data file.
######
Statistical significance of codon-based test of selection between sequences of *IGHG2* gene segment.
######
Click here for additional data file.
######
Statistical significance of codon-based test of selection between sequences of *IGHG3* gene segment.
######
Click here for additional data file.
[^1]: Edited by: Harry W. Schroeder, University of Alabama at Birmingham, United States
[^2]: Reviewed by: Kay L. Medina, Mayo Clinic, United States; Gestur Vidarsson, Sanquin Research, Netherlands; Celso Teixeira Mendes-Junior, University of São Paulo, Brazil
[^3]: This article was submitted to B Cell Biology, a section of the journal Frontiers in Immunology
[^4]: †Present Address: Danillo G. Augusto, Department of Neurology, University of California, San Francisco, San Francisco, CA, United States
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper and its Supporting Information files.
Introduction {#sec001}
============
Primary open angle glaucoma (POAG) is a chronic and progressive eye disease characterized by loss of retinal ganglion cells, thinning of the retinal nerve fiber layer, and subsequent visual field loss. If left untreated, it can eventually lead to blindness. Previously, an elevated intraocular pressure (IOP) was deemed to be the key factor in the pathophysiology of glaucoma. However, there is a variant of POAG in which the patients have a normal IOP (normal tension glaucoma \[NTG\])\[[@pone.0204939.ref001]\]. One potential explanation for this is that the relationship between IOP and intracranial pressure (ICP) is the key factor, rather than IOP itself \[[@pone.0204939.ref002],[@pone.0204939.ref003]\].
Alterations in IOP, ICP, or both can lead to a change in the pressure gradient across the lamina cribrosa (LC)---known as the trans lamina cribrosa pressure difference (TLCPD)---and cause it to bulge, therefore damaging the nerve fibers. It has been shown with lumbar punctures (LPs) that patients with glaucoma have a lower ICP than the healthy population and that those with NTG have the lowest \[[@pone.0204939.ref004]--[@pone.0204939.ref006]\]. As further evidence, patients with normal pressure hydrocephalus who received shunts, which can significantly lower ICP, had a 40 fold increase in the rate of NTG when compared with the general population \[[@pone.0204939.ref007]\]. There is, however, controversy over this theory \[[@pone.0204939.ref008]\].
LPs are invasive and thus limit a researcher's ability to conduct experiments with a sufficient number of subjects. In addition, LP opening pressures are only a proxy of the actual ICP as they are measured at the lower part of the spine and typically only in the lateral decubitus position. They are especially not likely to be representative of the pressure behind the LC. This is important because it is possible that NTG is related to an inability to maintain a certain pressure in the upright position, rather than the lateral decubitus position. Interestingly, a recent study by Linden et al. \[[@pone.0204939.ref008]\] aimed to use LPs to estimate ICP at the LC as accurately as possible by accounting for the hydrostatic gradients between the auditory meatus and the LC. LPs were done and ICP was measured continuously using a pressure transducer in various body positions but no differences were found in ICP between a small group of NTG patients and controls. A noninvasive device that measures the pressure at the level of the brain rather than in the lower part of the spinal canal could potentially further this field tremendously.
Otoacoustic emissions are sounds that originate in the cochlea and can be easily used to measure cochlear function noninvasively \[[@pone.0204939.ref009],[@pone.0204939.ref010]\]. One particular type of emission is the distortion product otoacoustic emission (DPOAE), which is emitted by the inner ear in response to two tones at specified levels and frequencies. These emissions are thought to depend on ICP because there is a connection between the cranium and inner ear via the cochlear and endolymphatic aqueducts. When ICP fluctuates, the pressure on the stapes also changes, affecting the transmission of sound in the middle ear. Previous research has shown that DPOAE phase shifts can accurately represent changes in ICP \[[@pone.0204939.ref011]--[@pone.0204939.ref014]\]. As such, a DPOAE measurement potentially contains all the desired properties of an ideal ICP measurement. One limitation is that the recorded DPOAE phase has an unknown---subject specific---offset. As a result, if the relationship between ICP and the DPOAE phase would be linear, changes in phase would only convey changes in pressure. In case of a nonlinear relationship (e.g. the phase changes only for pressures above or below a certain value), however, an absolute pressure measurement should be feasible as well. In this study we explore the clinical value of the DPOAE phase shift for ICP assessment and apply the technique to glaucoma patients.
The aim of this study was to (1) determine the relationship between DPOAEs and body position and to (2) compare this relationship between POAG, NTG, and controls. Finally, we aimed to (3) calibrate this relationship using published data regarding LP-based measurements of ICP versus body position.
Methods {#sec002}
=======
Study population {#sec003}
----------------
Subjects with healthy eyes who responded to our advertisement and glaucoma patients who were selected from the Groningen Longitudinal Glaucoma Study database \[[@pone.0204939.ref015]\], received an information letter and informed consent form. The ethics board of the University Medical Center Groningen (UMCG) approved the study protocol. All participants provided written informed consent. The study followed the tenets of the Declaration of Helsinki.
In order to be eligible to participate in this study, subjects in all groups had to meet the following inclusion criteria: 50 to 70 years of age and presence of detectable DPOAEs in at least one ear. Additionally, for healthy controls: IOP of 21 mmHg or lower, no eye disease, and no family history of glaucoma (determined by a questionnaire). To exclude eye disease, we performed optical coherence tomography (OCT-HS100; Canon, Tokyo, Japan; considered normal if the mean retinal nerve fiber layer and retinal ganglion cell layer thickness in macular area were above the 5th percentile), frequency doubling technology (FDT; Carl Zeiss, Jena, Germany; no reproducibly abnormal test locations allowed in C20-1 screening mode), and a measurement of visual acuity (visual acuity at least 0.8 in both eyes). For POAG: diagnosed glaucoma and IOP over 21 mmHg before the onset of IOP lowering treatment. For NTG: diagnosed glaucoma and IOP of 21 mmHg or lower before the onset of IOP lowering treatment and at any time during follow-up. Glaucoma was defined according to Heeg et al \[[@pone.0204939.ref015]\]. We required a reproducible (same hemifield and at least partially overlapping) visual field defect (Humphrey Field Analyzer 30--2 SITA fast; Carl Zeiss, Jena, Germany; criterion: \'glaucoma hemifield test\' outside normal limits) in at least one eye that had to be compatible with glaucoma and without any other explanation. Subjects taking acetazolamide were excluded in this study as this drug has been shown to change ICP \[[@pone.0204939.ref016]--[@pone.0204939.ref018]\].
DPOAE parameters {#sec004}
----------------
DPOAEs were measured using hardware (Elios) and software (Echosoft version 2.4.2) developed by Echodia (St. Beauzire, France). To ensure the highest magnitude responses at the 2*f*1-*f*2 emission, a fixed rate of *f*2/*f*1 = 1.20 with tones at frequencies *f*1 = 1000 Hz and *f*2 = 1200 Hz and levels L1 = L2 = 72 dB SPL were used. All DPOAE measurements were completed in a sound-isolated audiometric room in the otorhinolaryngology clinic.
Measurement protocol {#sec005}
--------------------
Blood pressure (Omron Model M6 Comfort, Omron Healthcare Co., Ltd.) and upright IOP (iCare Pro tonometer, Icare Finland Oy) were measured. Subjects were then secured onto the tilt table (Ironman iControl 400 Disk Brake Inversion System, Paradigm Health and Wellness Inc.). Body mass index (BMI) was determined using self-reported height and weight.
DPOAEs were measured at the following body positions (assuming 90° as upright): 45, 30, 20, 10, 0 (supine), -10, and -20°. At each position, 30 seconds were allowed for normalization of the emission, and presumably the ICP \[[@pone.0204939.ref012],[@pone.0204939.ref019]\]. This was followed by 5 DPOAE measurements that were performed over approximately 20 seconds. IOP was measured again at the supine position. The ear probe was then removed and subjects were allowed a short break before repeating the test. In this way we were able to determine test-retest variability with (between test) and without (within test) replacing the probe.
Data analysis {#sec006}
-------------
Groups were described with mean and standard deviation (SD) for normally distributed variables; means were compared using one-way analysis of variance (ANOVA). For variables with a skewed distribution, we used median and interquartile range (IQR) for descriptive statistics and the Kruskal-Wallis test for comparing medians of groups. Proportions were compared using a chi-square test.
The stimulus and data collection protocols are already described in detail by Avan et al \[[@pone.0204939.ref020]\]. At each body position, 5 DPOAE measurements were taken, which took approximately a total of 20 seconds of measurement time. For the within test test-retest variability, the first 2 and last 2 of these 5 measurements were averaged for each body position and compared. For the between test test-retest variability, the average phase over the full 20 seconds was compared for the first and second test, between which the ear probe had been removed and replaced. Test-retest variability was presented as the SD of differences of both measurements.
Due to the subject specific offset of the DPOAEs, normalization of the data was required. For each subject, an average phase over all body positions was calculated and the individuals' data were normalized using this value. Individual measurements with a signal to noise ratio (SNR) of less than 3 dB were excluded. If fewer than 2 measurements with a SNR over 3 dB were available at more than one body position (out of 5 measurements per body position), those subjects were removed entirely. All analyses were performed using R (version 3.0.2; R Foundation for Statistical Computing, Vienna, Austria). A p-value of 0.05 or less was considered statistically significant.
Results {#sec007}
=======
Of the 35 healthy subjects who agreed to participate, 1 was excluded because of an abnormal eye exam and 1 because of lack of emissions. Of the 43 glaucoma patients who agreed to participate, 7 were excluded because of lack of emissions. Due to signals below the noise floor, we could not analyze the data of 3 of 35 controls, 1 of 22 POAG patients, and 3 of 21 NTG patients. Therefore a total of 62 participants were included in the analysis (30 healthy controls, 17 POAG, and 15 NTG). [Table 1](#pone.0204939.t001){ref-type="table"} shows the demographics and ocular characteristics of the study population. The groups had approximately the same mean age, did not differ regarding gender, and there were no significant differences in any of the other demographic criteria. The visual field defects were similar in both the better eye and the worse eye for POAG and NTG. IOP (reported as the mean of both eyes) increased more from upright to supine for the patient groups than for the controls, although this was only statistically significant between controls and POAG (p = 0.01).
10.1371/journal.pone.0204939.t001
###### Demographics of the study population (n = 62; mean ± SD unless stated otherwise).
{#pone.0204939.t001g}
-------------------------------------------------------------------------------------------------------------------------------------------
Group Healthy\ POAG\ NTG\ P value
(n = 30) (n = 17) (n = 15)
------------------------------------------------------------ --------------------- ----------------------- ---------------------- ---------
Gender (% female) 43% 35% 60% 0.36
Age (yrs) 58.4 ± 6.4 61.6 ± 4.1 62.1 ± 4.7 0.05
BMI (kg/m^2^) 25.9 ± 3.1 25.7 ± 3.5 24.8 ± 3.9 0.58
SBP (mmHg) 131.4 ± 11.1 139.8 ± 18.9 131.1 ± 14.9 0.13
DBP (mmHg) 83.5 ± 9.6 87.0 ± 9.9 83.2 ± 10.3 0.44
VF MD of better eye (dB; median \[IQR\]) \- -2.5 (-6.8 to -0.7) -3.3 (-4.2 to -2.3) 0.56
VF MD of worse eye (dB; median \[IQR\]) \- -12.3 (-15.5 to -4.9) -9.8 (-4.7 to -21.8) 0.74
IOP0 (mmHg; median \[IQR\]) \- 30.0 (28.0 to 34.0) 17.5 (15.2 to 19.7) \-
Upright IOP (mmHg; median \[IQR\]) 15.1 (14.2 to 15.8) 15.6 (14.9 to 16.4) 15.4 (12.7 to 16.0) 0.34
Difference in IOP supine to upright (mmHg; median \[IQR\]) 1.5 (0.9 to 2.0) 2.6 (1.9 to 3.4) 1.9 (0.9 to 3.8) 0.01
-------------------------------------------------------------------------------------------------------------------------------------------
*SD* standard deviation, *BMI* body mass index, *SBP* systolic blood pressure, *DBP* diastolic blood pressure, *VF MD* standard automated perimetry mean deviation, *IOP0* intraocular pressure before onset of IOP lowering treatment.
[Fig 1](#pone.0204939.g001){ref-type="fig"} shows the DPOAE phase as a function of body position for the 30 healthy subjects, averaged for the first and second test. There is a clear relationship between phase and body position, especially when tilting to the 10° body position and lower. The relationship is not strictly monotonic, as the phase seems to increase again for the 30° body position and higher. This finding is likely robust, as it could be observed in both the first and the second test if analyzed separately. Initially we also measured -30° below horizontal, but that position had to be abandoned due to complaints and discomfort for the participants. In the 12 healthy subjects, 9 POAG, and 2 NTG patients in whom we measured this position (-30°), we found a clear further increase in phase compared to the -20° position of (mean ± standard error) 15.4 ± 3.7, 19.4 ± 9.9, and 24.6 ± 14.8° for healthy, POAG, and NTG, respectively.
{#pone.0204939.g001}
The observed relationship between DPOAE phase and body position was also shown in POAG and NTG ([Fig 2](#pone.0204939.g002){ref-type="fig"}). The mean overall phase shifts between the two most extreme body positions (45° to -20°) were 73.6, 80.7, and 66.3° for healthy, POAG, and NTG, respectively. [Fig 3](#pone.0204939.g003){ref-type="fig"} shows the corresponding scatter plots. One NTG subject was not included in this plot because he/she did not feel comfortable tilting to the furthest position. Despite the trend for a smaller mean overall phase shift in NTG patients, an ANOVA for the overall phase shift revealed that there were no significant differences between the healthy subjects and either of the patient groups (p = 0.73).
{#pone.0204939.g002}
{#pone.0204939.g003}
In order to be able to interpret the DPOAE phase data, we merged our data of phase as a function of body position presented in [Fig 1](#pone.0204939.g001){ref-type="fig"} with published data regarding ICP as a function of body position, based on LP measurements by Linden et al. \[[@pone.0204939.ref008]\]. [Fig 4A](#pone.0204939.g004){ref-type="fig"} shows their healthy subject data, adapted by interpolation to our body positions (interpolation was not possible for -20°); [Fig 4B](#pone.0204939.g004){ref-type="fig"} presents the merged data, showing DPOAE phase as a function of ICP. Above an ICP of approximately 3 mmHg, phase was linearly related to ICP, with a slope of 4 degree/mmHg. Below 3 mmHg, there was no clear relationship between phase and ICP.
![DPOAE phase as a function of ICP.\
(A) ICP as a function of body position for healthy subjects (n = 11) from Linden et al \[[@pone.0204939.ref008]\] adapted by interpolation to our body positions. (B) Merged data.](pone.0204939.g004){#pone.0204939.g004}
As previously mentioned, the ear probe was removed between the first and second test to examine the test-retest variability of the device. [Fig 5](#pone.0204939.g005){ref-type="fig"} shows scatter plots of the within test (panel A) and between test (panel B) test-retest variability of absolute phase at supine and also the SD of differences at all body positions (panel C). Nine subjects were removed from the between tests comparison, 6 due to a low SNR (for criteria see [Methods](#sec002){ref-type="sec"} section) in one of the tests and 3 NTG subjects that were not able to complete the second test due to nausea (see [Discussion](#sec008){ref-type="sec"} section). The smallest between test variability, with a SD of differences of 22.2°, occurred at the 10° body position. The within test variability was low and fairly stable across all body positions, with a SD of differences ranging from 5.3° to 10.2° (corresponding to approximately 2 mmHg of ICP; see [Fig 4B](#pone.0204939.g004){ref-type="fig"}).
{#pone.0204939.g005}
Discussion {#sec008}
==========
There is a nonlinear relationship between DPOAE phase and body position that is similar in healthy subjects and patients with POAG and NTG. Above an ICP of approximately 3 mmHg, there is a linear relationship between DPOAE phase and ICP with a slope of 4 degree/mmHg; below 3 mmHg, there is no clear relationship between phase and ICP.
Previous studies have demonstrated that DPOAE phase shifts can accurately represent changes in ICP. As one step of a larger experiment, Büki et al. \[[@pone.0204939.ref012]\] measured DPOAEs in 12 healthy subjects who were tilted on a table between upright and -30° and compared this to data from 5 patients with hydrocephalus and found that "ICP is the key element for all auditory modifications associated with posture". In another study \[[@pone.0204939.ref013]\], DPOAEs were measured in 12 healthy subjects titled from upright to -45°. At the same DPOAE frequency as used in the current study, they demonstrated DPOAE phase shifts of about 54° invoked by posture change. The overall shift found in the current study was of the same order of magnitude (\~70° for a body position of -20 to +45°). De Kleine et al. \[[@pone.0204939.ref019]\] provided an equation to calculate changes in ICP from changes in body position. They predicted an ICP change of \~10 mmHg for -10 vs 45°. The DPOAE phase shift for this body position change was approximately 50° in our study. As can be seen in [Fig 4](#pone.0204939.g004){ref-type="fig"}, these data are in agreement with each other. In the first study to directly compare DPOAEs with LPs, Bershad et al. \[[@pone.0204939.ref011],[@pone.0204939.ref021]\] found that large changes in ICP (\>15 mmHg) between opening and closing pressures were significantly associated with changes in DPOAE phase. In another study \[[@pone.0204939.ref014]\], DPOAEs were measured in 8 subjects undergoing CSF infusion testing, and it was shown that for ICP changes of \~12 mmHg or more over baseline, DPOAE phase changes were significant. In the current study, we have reproduced the finding that changes in DPOAE phase can represent changes in ICP. However, our data suggest that, with appropriate probe placement and sufficient averaging, much smaller changes in ICP could be detected (see below).
For many of the subjects there was a noticeable increase in phase that occurs toward the upright position, so that the minimum phase occurred at 30° and not at 45° (see Figs [1](#pone.0204939.g001){ref-type="fig"} and [2](#pone.0204939.g002){ref-type="fig"}). There are two physiological possibilities for this phenomenon. One is that ICP has its minimum somewhere halfway between supine and upright rather than at upright. In an invasive ICP study \[[@pone.0204939.ref022]\] it was demonstrated that this occurred in approximately 5% of subjects. In the current study, however, an observable minimum in phase at body positions lower than 45° occurred in over 50% of subjects. Another possibility is that when there is a negative ICP, as is often the case in the upright position, the stapes is pulled inward into the oval window and may elicit a response in DPOAE phase similar to when it is pushed outward. In this case the minimum phase would occur at a body position intermediate between upright and supine, when the stapes is in a neutral position.
There were some limitations in the current study: First, DPOAEs were not measured at the upright position which was due to a limitation of the tilt table. However, within the range of body positions included in this study, the nonlinear relationship between phase and body position could clearly be uncovered. It is also important to note that 3 NTG subjects could not complete the second part of the test due to nausea and dizziness. No other subjects had any problems with the test. Interestingly, this supports the concept that vascular factors and impairment of ocular blood flow might play a role in the pathogenesis of NTG \[[@pone.0204939.ref023]\]. Indeed 2 out of the 3 subjects had low blood pressure (106/73 mmHg and 107/79 mmHg). Finally, there was a large intersubject variation in the overall DPOAE phase shift. At first sight, this seems to hamper the use of DPOAE phase for ICP assessment. However, it should be realized that ICP itself also shows a significant variability \[[@pone.0204939.ref022],[@pone.0204939.ref024]\].
What information regarding ICP can be obtained from the DPOAE measurements? Importantly, due to a subject-specific offset, there is no one-to-one relationship between DPOAE phase and ICP. For an ICP above 3 mmHg, there is a linear relationship between DPOAE phase and ICP, which implies that a change in phase can be converted into a change in ICP, where 4° corresponds to 1 mmHg ([Fig 4B](#pone.0204939.g004){ref-type="fig"}). This suggests that it is possible to monitor ICP changes in a subject or patient in the supine position, in which ICP should be amply above 3 mmHg ([Fig 4A](#pone.0204939.g004){ref-type="fig"}). The accuracy (expressed as standard deviation of differences) is approximately 2 mmHg without probe movement, and 5 mmHg with probe movement or replacement. This is already clinically useful; a more stable and reproducible probe positioning could further improve the accuracy. Below 3 mmHg, the phase does not change further, and may even start to change in the opposite direction. This nonlinear behaviour offers potentially an opportunity to obtain information regarding absolute ICP from DPOAE measurements. In the case of a low ICP, one would expect the curve of phase as a function of body position ([Fig 1](#pone.0204939.g001){ref-type="fig"}) to move leftward, meaning a more extreme tilt position is required to elicit a DPOAE response. Related to that, lower ICP patients would have a smaller overall phase shift when changing body position from upright to head down. On the other hand, in the case of a high ICP, one would expect the curve to move rightward with a larger overall phase shift.
There are many theories as to why glaucoma may occur in patients with a normal eye pressure. One theory is that NTG patients simply have thin corneas, yielding erroneously low IOP measurements. In this study, however, the corneal thickness for POAG and NTG were similar (544 μm and 554 μm, respectively), signifying that another explanation is necessary. What do our results suggest regarding ICP in NTG? The overall phase shift seemed smaller in the NTG patients than in the healthy subjects or POAG patients ([Fig 3](#pone.0204939.g003){ref-type="fig"}), but the difference was not significant. Also, the phase as a function of body position curve of the NTG patients ([Fig 2](#pone.0204939.g002){ref-type="fig"}) did not show a clear leftward or rightward shift when compared to that of the healthy subjects or POAG patients. A shift of more than 5° along the body position axis seems unlikely, which suggests that the mean difference in ICP between our NTG patients, healthy subjects, and POAG patients is less than 1--2 mmHg. As such, our results agree with the findings of Linden et al \[[@pone.0204939.ref008]\], who reported no difference in ICP between NTG patients and healthy subjects, and seem to disagree with Berdahl et al. \[[@pone.0204939.ref004],[@pone.0204939.ref005]\] and Ren et al \[[@pone.0204939.ref006]\], who reported a significant difference in ICP of 3.1 and 3.4 mmHg, respectively between NTG patients and healthy subjects. There is little research on invasive measurement of ICP in glaucoma \[[@pone.0204939.ref004]--[@pone.0204939.ref006],[@pone.0204939.ref008],[@pone.0204939.ref025]\], so it is difficult to determine the cause of the disagreement. One possible reason for the discrepancy proposed by Linden et al \[[@pone.0204939.ref008]\] is that there is no difference between the studies in regards to ICP in NTG, but there is, however, a difference in the control groups. A normal ICP at supine is considered to be between 7--15 mmHg \[[@pone.0204939.ref024]\] yet, for the 3 studies that showed a difference between groups, the healthy controls had mean ICPs of \~13 mmHg which is at the higher end of the normal range. In the study by Ren et al \[[@pone.0204939.ref006]\] there was a relationship between visual field defects and TLCPD, but in this case the IOP was measured in the upright position while the ICP was measured in the lateral decubitus position. Interestingly, in all studies there was no significant relationship between ICP and severity of visual field defects, suggesting something beyond reduced ICP is responsible for the pathophysiology of NTG. Finally, it could be the case that the role of ICP in glaucoma is limited to a small subgroup of NTG patients, that is, those with an IOP at the lower end of the normal range (\~10 mmHg; sometimes referred to as low tension glaucoma). These patients were not included in the current study (median sitting IOP 15.4 mmHg; [Table 1](#pone.0204939.t001){ref-type="table"}), and also not in the study by Linden et al \[[@pone.0204939.ref007]\], where the mean sitting IOP was 15.1 mmHg. However, the mean IOP in the studies Berdahl et al \[[@pone.0204939.ref003]\] and Ren et al \[[@pone.0204939.ref005]\] were not clearly lower (14.3 and 16.1 mmHg, respectively). Of note, Ren et al included Chinese patients, whereas in the current study and in the study by Linden et al the patients were of Caucasian origin. When looking at the supine position for healthy subjects in the current study, the ICP determined by DPOAE phase increases by 5.5 mmHg and yet IOP increases by only 1.5 mmHg. This suggests that a large change in TLCPD in the supine position, e.g., during sleep, may be part of a normal physiological pattern.
In conclusion, we did not find evidence that NTG patients have a reduced ICP. Beyond glaucoma, however, DPOAEs can be used to monitor ICP changes noninvasively, which may further clinical care and the understanding of ICP variation and regulation tremendously. First, however, future research should focus on a robust and reproducible probe placement in order to minimize variability.
Supporting information {#sec009}
======================
###### Underlying information.
This is the underlying information for the current study.
(XLSX)
######
Click here for additional data file.
We would like to thank Paul Avan for very stimulating discussions regarding this study and Echodia for the use of their Elios device and for technical support. We would also like to thank L. van de Waardt, E. Chen Yao, and W. Nieboer for their assistance with data collection.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
| {
"pile_set_name": "PubMed Central"
} |
Prostate cancer (CaP) is the second most common malignancy in men worldwide, with 910 000 new cases diagnosed in 2008 ([@bib7]). CaP has a predilection to metastasise to the bone marrow stroma (BMS), and development of CaP bone metastases almost invariably result in CaP-related mortality ([@bib9]). The metastatic process is a complex, multistep process, which can be modelled with modified *in vitro* invasion chambers utilising human primary BMS ([@bib16]; [@bib25]; [@bib13]). These *in vitro* models not only allow the determination of the mechanism of CaP metastasis to the BMS but enable the elucidation of how therapeutic agents may interfere with the metastatic process.
Recent evidence links lipid metabolism and statin use with the behaviour of CaP. Large scale epidemiological data showing lower rates of CaP progression in patients taking statins ([@bib23]) have been consolidated by reports showing that CaP incidence in screened ([@bib20]) and non-screened ([@bib2]) populations is reduced in men taking these drugs and that individuals with lower cholesterol levels had lower rates of high risk disease ([@bib22]). Further clinical data from histological analysis of large case after prostatectomy series showed less aggressive features in men taking statins ([@bib17]) and in patients undergoing radiotherapy for CaP indices of CaP treatment failure were reduced in men taking statins, especially in those with high risk features ([@bib10]).
The basis for these observations is poorly understood but the effects probably relate to the pleiotropic actions of 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) inhibitors on cellular behaviour, cell-cell interaction and cellular motility in relation to lipid metabolism ([@bib27]). To better understand this process, we have studied the effects of different statins in our in-house, well-categorised models of CaP behaviour, evaluating the differential effects of specific statins on cellular binding, migration and early cellular survival in human BMS.
Materials and methods
=====================
Cell culture
------------
PC-3 (ATCC-LGC, Teddington, UK) cells were cultured in HAM\'S-F12 media supplemented with 7% foetal calf serum (FCS) and 2 mℳ L-glutamine at 37 °C, 5% CO~2~ in air. PC3-GFP were cultured as for PC-3 except for addition of 0.15 mg ml^−1^ hygromycin B. Human BMS was obtained from volunteers undergoing surgery for benign disease and cultured according to [@bib6]. Briefly, 2 × 10^6^ cells ml^−1^ in long-term bone marrow culture medium (LTBCM) (Iscove\'s modified Dulbecco\'s medium at 350 mOsm, 10% foetal calf serum, 10% horse serum, 5 × 10^−7^ ℳ hydrocortisone) were grown at 33 °C in 5% CO~2~ in air for 4--5 weeks until haematopoietically active areas were observed. Bone marrow endothelial cells (BMEC) were cultured in LTBCM conditioned by BMS on fibronectin-(50 mg ml^−1^ in PBS) treated flasks. All cell lines were verified by the Paterson Institute for Cancer Research tissue typing service.
Statins and metabolites
-----------------------
Atorvastatin, rosuvastatin (Discovery Fine Chemicals Ltd, Dorset, UK), mevastatin, and simvastatin (Sigma-Aldrich, Poole, UK) were dissolved in dimethyl sulphoxide (DMSO) and pravastatin (Sigma-Aldrich) was dissolved in water at concentrations of 100 mℳ. Activated simvastatin was also assessed. Briefly, 4 mg of simvastatin was dissolved in 100 *μ*l of ethanol to which 150 *μ*l of 0.1 N NaOH was added and incubated at 50 °C for 2 h. The pH was brought to 7.0 by HCl and diluted to a final concentration of 100 mℳ before storing at 4 °C. All statins were used at final non-toxic concentrations as defined by trypan blue exclusion. All metabolites were purchased from Sigma-Aldrich. Geranylgeranyl pyrophosphate (GGPP) and farnesyl pyrophosphate (FPP) were dissolved in methanol at a final concentration of 1 mg ml^−1^. Cholesterol was dissolved in chloroform at 100 mℳ. Mevalonate was made up in water at a stock concentration of 10 mg ml^−1^ (40 mℳ).
Binding assay
-------------
Prostate epithelial cell (PEC) binding to BMS was determined as previously described ([@bib25]); 5 × 10^4^ PC3-GFP cells, treated with statin or DMSO vehicle control, were added to confluent BMS in a 96-well plate. Wells were washed and fluorescence was read using a FLUOstar OPTIMA spectrometer (BMG Labtech, Aylesbury, UK). Binding was expressed as percent of pre-washed total fluorescence.
Invasion assay
--------------
Prostate epithelial cell invasion towards BMS was measured as previously described ([@bib13]); invasion towards statin-pretreated BMS or in the presence of statins was assessed. The FluoroBlok (BD Biosciences, Oxford, UK) tumour cell invasion system used phenol red-free RPMI 1640 media and FluoroBlok cell culture inserts coated with 100 *μ*l of Matrigel (BD Biosciences) (1 : 25 dilution with RPMI 1640) above chemoattractant (BMS or tissue culture plastic (TCP)). 2 × 10^5^ PC3-GFP in RPMI 1640/0.1% BSA were added to each FluoroBlok insert and the co-cultures incubated at 37 °C, 5% CO~2~ for 18 h. Bone marrow stroma was pretreated daily with the different statins. PC-3 cells were treated with statins 30 min before addition. Final readings utilised the FLUOstar OPTIMA spectrometer. Recovery assays were conducted as above or in the presence of a confluent BMEC cell layer barrier ([@bib13]). PC3-GFP cells were treated with 10 *μ*ℳ GGPP, 100 *μ*ℳ mevalonate, 16 *μ*ℳ cholesterol or 10 *μ*ℳ FPP immediately before the additions of statins and adding to the invasion chamber.
Co-culture colony assay
-----------------------
Prostate epithelial cell growth in BMS co-culture was measured as previously described ([@bib16]); confluent primary human BMS, pretreated with statins for 60 min, was seeded with 500 PC-3 cells. Statins were refreshed daily for 8 days before 4% paraformaldehyde fixation. Co-cultures were permeabilised using ice-cold methanol and blocked with 10% rabbit serum, followed by 0.3% hydrogen peroxide. Co-cultures were incubated with mouse anti-human pan-cytokeratin at 1 : 200 (Sigma-Aldrich) followed by biotinylated rabbit anti-mouse 1 : 400 (DAKO Ltd., Cambridge, UK). A complex of avidin DH and biotinylated horseradish peroxidase H (Vector Laboratories, Peterborough, UK) was then added and developed with DAB substrate.
Clonogenic assay
----------------
Five hundred PC-3 cells were treated at day 1 and day 7 with statins, before methanol acetone fixation and staining with crystal violet on day 14; colonies (\>32 cells) were then counted using a standard microscope graticule.
Statistics
----------
Values are presented as mean±s.e. Assays were compared using the two-tailed Student\'s *t-*test with significance set at *P*\<0.05.
Results
=======
Reduction of BMS invasion by lipophilic statins
-----------------------------------------------
We have previously shown that BMS is the most potent chemoattractant for metastatic PECs and that the metastatic process can be modelled using primary human BMS co-cultures ([@bib13]; [@bib4]). We therefore addressed three different questions: do statins affect the BMS microenvironment ([Figure 1A](#fig1){ref-type="fig"}), the PECs directly ([Figure 1B](#fig1){ref-type="fig"}) or the interaction between BMS and PECs ([Figure 1C](#fig1){ref-type="fig"}). Statins did not affect the ability of BMS to induce PEC invasion; although atorvastatin and simvastatin treatment of BMS reduced invasion as compared with the controls (84±8.8% *P*=0.18806 and 80±10.57% *P*=0.14755), this did not reach significant levels ([Figure 1A](#fig1){ref-type="fig"}). All statins, except pravastatin, induced a similar and marked reduction in PEC invasion towards BMS (*P*\>0.05), averaging 66.68% of control (range 54--77% *P*\<0.05). Bone marrow stroma pretreatment ([Figure 1C](#fig1){ref-type="fig"}) did not provide additional benefit to treating PECs alone (58.86% *P*\>0.05 of PC-3\'s invading compared with controls), suggesting a direct effect on the PC-3 cells and not on the BMS 'soil\'.
The effect of statins was more significant in the presence of a confluent BMEC/Matrigel barrier, which models the blood BMS barrier. Both 5 *μ*ℳ rosuvastatin and 1 *μ*ℳ simvastatin ([Figure 1D](#fig1){ref-type="fig"}) completely inhibited direct invasion towards BMS (9.37% *vs* 100%, *P*=0.0201 and 25.6% *vs* 100%, *P*=0.0496 for rosuvastatin and simvastatin *vs* BMS+vehicle control, respectively), with 5 *μ*ℳ rosuvastatin reducing invasion to levels similar to TCP (*P*=0.21212). Activation of simvastatin according to manufacturer\'s instructions had no effect on the ability of simvastatin to inhibit transendothelial invasion towards BMS (13.5%±7.1% and 14.8%±12.8%, respectively; *P*=0.142) ([Supplementary Figure 1](#sup1){ref-type="supplementary-material"}).
Effect of statins on PEC binding to BMS
---------------------------------------
When malignant PECs enter the BMS, they migrate towards and bind to niches within the BMS before proliferating and forming viable colonies ([@bib3]). We sought to assess the inhibitory effect of statins on this process. Incubation of PECs with statins for up to 120 min did not alter the ability of PECs to bind to BMS ([Figure 2](#fig2){ref-type="fig"}).
Reduction of PEC colony formation in BMS by lipophilic statins
--------------------------------------------------------------
Previous studies have documented the antiproliferative effect of high-dose statin treatment of PECs. However, the antiproliferative effect of statins on PEC within the lipid-rich environment of the BMS is unknown. We therefore compared the effect of statins on PEC colony formation and proliferation in isolation with clonogenic assay or in BMS co-culture.
Pravastatin, at doses up to 100 *μ*ℳ, had no effect on PC-3 colony formation or cellular proliferation in either clonogenic or BMS co-culture assays ([Figure 3A](#fig3){ref-type="fig"}). However, the lipophilic statins and rosuvastatin all had a significant effect on colony formation in both assays and cellular proliferation. Lipophilic statins and rosuvastatin significantly reduced the number of colonies to a similar degree in clonogenic assays and BMS co-culture ([Table 1](#tbl1){ref-type="table"}), with simvastatin being the most potent (reduction of 75.44% and 49.44%, respectively). Bone marrow stroma colonies were smaller than controls (0.19±0.025 mm^2^, 0.21±0.03 mm^2^, 0.21±0.03 mm^2^, 0.19±0.03 mm^2^; *P*\<0.05) resulting in a reduction of the total PEC area (12.8±3.34 mm^2^, 20.7±4.5 mm^2^, 18.68±4.1 mm^2^, 14±3.8 mm^2^) in co-cultures treated with rosuvastatin, atorvastatin, mevastatin and simvastatin, respectively ([Figure 3B](#fig3){ref-type="fig"}). There was no difference (*P*\>0.05) in the colony size, number or overall PEC load between the four statins.
Typically, PC-3 colonies in BMS co-culture formed large diffuse colonies of migrating PECs within the BMS. Cells displayed a mesenchymal morphology, being teardrop shaped, with ruffled leading edges of lamellipodia in the absence of therapeutic agents or in the presence of DMSO vehicle control ([Figure 3C](#fig3){ref-type="fig"}, control). The lipophilic statins induced significant morphological changes in the BMS colonies. Prostate epithelial cell colonies were more tightly packed, containing fewer cells than controls. At a higher magnification they displayed differing morphology compared with the mesenchymal morphology seen in control co-cultures, with PECs being more rounded, without obvious lamellipodia at the cell\'s leading edge. Although some cells retained their teardrop appearance, the leading edge did not possess lamellipodia and the tails were distorted, appearing longer, wider and curved ([Figure 3C](#fig3){ref-type="fig"}). These morphological effects were not seen with pravastatin.
Mechanism of statin inhibition of invasion
------------------------------------------
Statins act by blocking the conversion of HMG-CoA to mevalonic acid through inhibition of HMG-CoA reductase ([Figure 4A](#fig4){ref-type="fig"}). We therefore sort to recover the invasion by supplying the downstream metabolites mevalonate, cholesterol or the prenylating agents GGPP and FPP. Treating PC-3-GFP cells with either GGPP or mevalonate had no significant effect on their ability to invade towards BMS (*P*=0.259 and 0.619, respectively). Both cholesterol and FPP reduced PC3-GFP invasion towards BMS, but this reduction was not significant (75.51%±2.25% *P*=0.158 and 54.9%±8.24% *P*=0.056, respectively). PC3-GFP cells were treated with mevalonate, GGPP, FPP or cholesterol before treatment with either 1 *μ*ℳ simvastatin or 5 *μ*ℳ rosuvastatin, and invasion through a BMEC/ Matrigel barrier towards untreated BMS was assessed. The addition of mevalonate to the invasion assay completely restored/enhanced the invasion towards BMS (66.67%±13.26% *P*=0.1126 and 117.42%±17.79% *P*=0.0033 for rosuvastatin and simvastatin, respectively). However, of the addition of downstream metabolites, only GGPP was able to completely restore rosuvastatin-treated invasion towards BMS (82.42±5.52% *P*=0.229). Addition of GGPP to simvastatin-treated cells resulted in an enhanced invasion as compared with the statin-free controls (122±7.83% *P*=0.0185). Addition of either FPP or cholesterol did not affect the ability of rosuvastatin or simvastatin to inhibit invasion towards BMS. Activation of simvastatin had no effect on the recovery profile, with both GGPP (111.5±8.59% *P*=0.9079) and mevalonate (67.04±8.03% *P*=0.0547) restoring the invasive ability of PC3-GFP cells towards BMS (*P*=0.1391 and *P*=0.2267 comparing activated to not activated simvastatin in the presence of GGPP or mevalonate, respectively) ([Supplementary Figure 2](#sup1){ref-type="supplementary-material"}).
Discussion
==========
This is the first study to report the comparative effects of different statins on CaP cellular migration towards and within human BMS. The results demonstrate a clear effect on CaP migration towards and through BMS and on malignant PEC\'s ability to grow clonally in that location. However, this effect was limited to the lipophilic statins and was not seen with the hydrophilic statin, pravastatin. The differential effect of the two basic subtypes of this class of drug has been well described; lipophilic statins diffuse across cellular membranes and exert their metabolic effects in the liver and other tissues; hydrophilic statins require active transport across the cell membrane in order to exert their actions intracellularly. This action of hydrophilic agents is therefore predominantly hepatic and not peripheral ([@bib27]; [@bib8]) and this fact has significant consequences for the effects observed in neoplasms such as breast and other cancers ([@bib5]; [@bib14]; [@bib15]). This differential action was clearly evident in the results presented herein and may be an important consideration when interpreting data from population studies of the effect of statins or when planning statin-based prevention trials in CaP.
Epidemiological studies have shown that although statin use does not affect CaP incidence, it does reduce the risk of clinical progression and CaP-related mortality; the mechanisms of this are unknown. Statins can induce apoptosis in a range of malignant epithelia *in vitro* but the concentrations of 10--20 *μ*ℳ used far exceed those achievable physiologically in humans. We utilised concentrations of statins with reduced PEC toxicity, allowing assessment of their potential role in invasion and early metastasis. The statin concentrations we used were higher than the serum levels detected with current statin dosing regimes, for example, peak plasma concentration of simvastatin in patients receiving 40 mg per day is 7.2 nℳ (3 ng ml^−1^) ([@bib21]). However, in trials with lymphoma or myeloma patients ([@bib29]), the maximum tolerated dose of simvastatin was 15 mg per kg per day, which would increase the circulating serum levels reaching peripheral organs significantly. A regime of 45 mg per kg per day lovastatin in patients with solid tumours induced peak serum levels 3.9 *μ*ℳ ([@bib28]).
Using validated human BMS co-culture models ([@bib13]), we have shown that lipophilic statins all affect the ability of PECs to invade towards and through BMS ([Figures 1](#fig1){ref-type="fig"} and [3](#fig3){ref-type="fig"}) to initiate and develop tumour colonies. BMS pretreatment ([Figure 1A](#fig1){ref-type="fig"}) had no effect on statins\' ability to promote or inhibit this process. [Figure 1B](#fig1){ref-type="fig"} (PC-3 treatment alone) and [Figure 1C](#fig1){ref-type="fig"} (pretreatment of BMS and PC-3) showed similar reductions in invasion. Utilising the more complex model of metastasis using BMECs and basement membrane, which recapitulates the blood BMS barrier, both rosuvastatin and simvastatin completely inhibited directed invasion towards BMS. These data suggest that statins affect PECs directly but have no effect on their microenvironment. The exact mechanism underlying statins\' ability to block CaP migration is unknown but there are several possible aetiological factors.
A recent large cohort study of cancer-free males over the age of 35, showed that men with desirable/borderline cholesterol levels (\<200 mg dl^−1^--\<240 mg dl^−1^) had a lower risk of developing high grade CaP but no risk reduction in their overall incidence of CaP ([@bib19]). This supports other studies where low cholesterol was not associated with organ confined or low grade CaP but was associated with reduced risk of high grade cancer ([@bib22]). A further epidemiological study also showed that low cholesterol was inversely associated with advanced disease (OR=0.42, 95% CI=0.13--1.36). The numbers in this study were small but the senior author ([@bib22]) and others ([@bib26]) have gone on to propose that statins affect CaP\'s intracellular cholesterol metabolism, which is known to be dysregulated. The uptake of statins by the liver and the requirement for liver-specific esterases, for example, in the case of the pro-drug simvastatin, limits statin availability in the peripheral circulation ([@bib18]). It is therefore hypothesised that statins act by blocking HMG-CoA reductase in the liver and not in tumours, and that the reduction in cholesterol affects tumours at extrahepatic sites. Here we have shown that simvastatin can act directly on the PC3-GFP cells in the absence of activation, ([Supplementary Figure 1](#sup1){ref-type="supplementary-material"}) suggesting that simvastatin may act within the periphery. Cholesterol is an essential component of lipid rafts, cellular membrane microdomains that regulate signalling cascades originating from membrane-bound receptors such as tyrosine kinases. Treatment of CaP overnight or glioma cells for 48 h with simvastatin induced a 70% reduction in lipid raft cholesterol, an effect reversed by addition of exogenous cholesterol ([@bib32]; [@bib30]). This reduction inhibited the Akt1 serine-threonine kinase/ protein kinase-B signalling pathway and induced apoptosis, an effect that could be reversed by cholesterol replenishment. Corroborating *in vivo* studies showed that elevated cholesterol levels increased pAkt1 signalling and promoted tumour growth, with a coincident reduction in apoptosis.
The studies above demonstrate a role of cholesterol in CaP progression by promoting tumour growth and preventing apoptosis. However, in the study presented here we are specifically modelling the effects of statins on the invasive and migratory abilities of CaP cell towards and within human BMS. Within these validated invasion co-culture models, cholesterol was unable to recover the invasive phenotype after statin treatment. We have shown previously that invasion towards BMS occurs predominantly within the first 4 h ([@bib12]). In this study, PC-3 cells were pretreated for 30 min before assay, suggesting that a reduction in cholesterol synthesis by CaP cells is not the factor inhibiting invasion. This observation is supported by co-culture assay data; the co-cultures were grown in the presence of FCS and horse serum, both of which contain cholesterol. It is also noteworthy that invasion was towards haematopoietically active adipocyte-rich human BMS, which contains high cholesterol levels. All the co-cultures treated with lipophilic statins showed proliferation of PC-3 cells but formed smaller colonies of tightly packed epithelial cells, differing in their morphology to the controls, losing the ruffled leading edges of their lamellipodia and displaying distorted trailing edges, indicating motility impairment. Taken together these models suggest that the inhibition of invasion towards/through BMS, but not proliferation, is not related directly to circulating and local tissue cholesterol levels.
A critical factor may be the impairment of the transmigrational mechanisms of cancer cells by the effects of statins, in particular, their effect on the mevalonate and isoprenoid axis. Tumour cell exposure to statins downregulates mevalonate expression through HMG-CoA reductase inhibition. This affects membrane integrity, cell signalling, protein synthesis and cell cycle progression. Disruption of these processes by statins may result in inhibition of tumour initiation, growth and metastasis ([@bib1]). Mevalonate is the precursor of the isoprenoids FPP and GGPP, essential for the translocation of signalling proteins, such as the G-proteins Ras and Rho to the plasma membrane, where they mediate signal transduction by controlling cell differentiation, proliferation and migration ([@bib11]). Addition of GGPP, unlike cholesterol or FPP, to our invasion assays completely restores the ability of PC-3 cells to invade towards the BMS, suggesting that the lipophilic statins act by preventing the translocation of nascent signalling proteins, which require geranylgeranylation, to the cell surface. This would lead to a reduction of signalling proteins at the cell surface and thereby preventing the cell from detecting the invasive stimulus from the BMS or BMS adipocytes ([@bib3]).
Further studies support the notion that the mevalonate/G-protein axis is important. [@bib31] demonstrated that 10 *μ*ℳ lovastatin inhibited EGFR dimerisation and signalling via AKT and its downstream targets, 4E-binding protein 1 and S6 kinase 1, in a GGPP-dependent manner. Lovastatin also induced cytoskeletal disorganisation associated with increased inactive RhoA expression, which lacked the GGPP tag. Co-incubation with mevalonate or GGPP restored RhoA activity and EGF signalling. This suggests that inhibition of invasion towards and through BMS is more likely to arise from isoprenoid loss rather than loss of endogenous cholesterol production by the CaP cells. This latter factor may be critical; the ability of the cancer cell to move is fundamental to metastasis and its inhibition prohibits metastatic CaP survival in BMS ([@bib16]). *In vitro* studies have demonstrated the critical importance of the Rho/Rac axis in mesenchymal to amoeboid transition and invasion/migration ([@bib24]); these steps are fundamental to progression in prostate and other cancers. Thus, important elements of metastatic behaviour are potentially inhibited by lipophilic statins and these anti-migrational effects alone may be responsible for the CaP-related observations presented herein and reported clinically ([@bib23]; [@bib10]).
Here we have used validated models of malignant prostate epithelial invasion to examine the effects of statins on the migratory pathway in CaP metastasis. We have shown for the first time in this laboratory setting that statins act directly on malignant PEC and block the formation of GGPP from HMG-CoA. Loss of GGPP leads to a significant reduction in the ability of malignant PEC to invade towards and through BMS and so reduce their ability to form colonies within the BMS. This statin effect is not universal across all classes of statins as the hydrophilic statin pravastatin had no effect on the PC-3 cell line in our models. In summary, statins appear to act on two key components on CaP metastasis; reduce tumour growth, possibly through the reduction of cholesterol but also, as described here, through inhibiting geranylgeranyl prenylated pathways required for transendothelial migration and BMS invasion.
[Supplementary Information](#sup1){ref-type="supplementary-material"} accompanies the paper on British Journal of Cancer website (http://www.nature.com/bjc)
This work is published under the standard license to publish agreement. After 12 months the work will become freely available and the license terms will switch to a Creative Commons Attribution-NonCommercial-Share Alike 3.0 Unported License.
Supplementary Material {#sup1}
======================
######
Click here for additional data file.
######
Click here for additional data file.
######
Click here for additional data file.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
###### Effect of stains on prostate epithelial colony formation in BMS co-culture
**Mean colony number** **s.e.** **Fold reduction** ***P*****-value**
------------------- ------------------------ ---------- -------------------- -------------------
No drug control 118 12.84
DMSO low control 150.7 19.97
DMSO high control 125 13.54
Rosuvastatin 61.2 13.76 2.04 0.0095
Atorvastatin 94.6 13.91 1.59 0.055
Mevastatin 85.8 13.56 1.76 0.0315
Simvastatin 63.2 4.72 2.38 0.0117
Pravastatin 125.8 27.79 0.94 0.7928
Abbreviations: BMS=Bone marrow stroma; DMSO=dimethyl sulphoxide; PC-3 colony formation after statin treatment in BMS co-culture assay. PC-3 cells were treated with 1 *μ*ℳ atorvastatin, mevastatin, simvastatin, 5 *μ*ℳ rosuvastatin or 10 *μ*ℳ pravastatin daily for 8 days in BMS co-culture. Colonies \>32 cells were counted. [Table 1](#tbl1){ref-type="table"} shows the mean number of colonies, the fold reduction in numbers as compared with the relevant vehicle controls and the significant difference between treatments and their respective vehicle controls.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#S0001}
============
Low birth weight (LBW) is considered as birth weight less than 2500gm and in Bangladesh, 45% child born with LBW \[[@CIT0001], [@CIT0002]\]. It has been speculated that low birth weight predisposes children to a high risk of diabetes, heart diseases and other chronic conditions later in life \[[@CIT0003]--[@CIT0008]\]. Data cited in National Food and Nutrition policy (1997) document shows that LBW ranges from 30-50 percent, however from different other studies it has been revealed that LBW prevalence rate is about 30 percent \[[@CIT0009], [@CIT0010]\]. It is assumed that, under nutrition, both before and during pregnancy, causes intrauterine growth retardation and is one of the major reasons for the high LBW. Between 1990 and 2004, underweight levels among children fell from 67% to 48% and child stunting fell from 66% to 43% \[[@CIT0011], [@CIT0012]\], but the levels are still unacceptably high. Some studies, from Bangladesh, showed that mean birth weight increase with an increase in mother\'s age from 14 to 31 years, while after 32 years birth weight decrease and the highest mean birth weight occurred in women between 26 and 31 years of age \[[@CIT0002]\]. There are studies both in support and to deny the relationship between maternal socioeconomic and anthropometric characteristics and birth weight \[[@CIT0013]--[@CIT0015]\].
Breastfeeding is considered as the pivotal factor between life and death for the vast majority of children in developing countries, but pattern of breast feeding and exclusive breast feeding is more important, which is ignored often by most mothers. Breast milk is a natural resource that has a major impact on infant\'s health, growth and development and it is recommended for at least the first two years of a child\'s life \[[@CIT0016]\]. Breast milk contains all desirable nutrients in right quantity that a baby needs and are easily digestible \[[@CIT0017]\]. However it is important to ensure exclusive breastfeeding for the first six months of life, only 35.0% of infants worldwide were exclusively breastfed during the first six months of life and 27.0% under six months infant were bottle fed \[[@CIT0016]\]. It is evident that inappropriate breastfeeding practices are associated with severe malnutrition in the under five children, lack any advantage in terms of weight gain and are associated with growth faltering \[[@CIT0018]\]. A study conducted in Khulna, Bangladesh, stated that the prevalence of breastfeeding as 96% \[[@CIT0019]\], but the prevalence of exclusive breastfeeding and the effect of breastfeeding on infant nutritional status were not clearly stated. Early initiation of breastfeeding generally lengthens the duration of breastfeeding and also prevents postpartum hemorrhage \[[@CIT0019]\], however the pattern of initiation of breastfeeding was not clearly reported in Bangladesh.
In developed countries during the last two decades a large number of studies have been conducted to identify the effect of maternal status and breastfeeding practices on infant development. In Bangladesh, few studies have been conducted to find out the effect of maternal status and breastfeeding practices on infant nutritional status and most of them are weak due to inappropriate study design and lack of statistical validity. We think it is important to investigate the relationship between these in a view to prevent infant morbidity and mortality. The objectives of this analysis was to determine how maternal socioeconomic and anthropometric status effect on birth weight, particularly in determining low birth weight (\<2500 gm), and how breastfeeding practice, particularly the type and effective time of initiation and termination of breastfeeding effect on infant growth and development.
Methods {#S0002}
=======
Subjects {#S20003}
--------
A total of 510 mother-infant pairs were included and the study was continued for 6 months. Mothers and their infants attending the selected hospitals and maternal clinics in Kushtia during the study period were regarded as the study population. Infants' of up to 24 months of age was selected and infants above 24 months of age were excluded from the study. The mothers were selected beyond any ages. Some selection and exclusion criteria were followed as women whose health status was normal and who gave normal live birth were included and women with any severe health complication and caesarian cases were excluded from the study.
Study design {#S20004}
------------
The cross-sectional study was conducted during the period of September 2011 to February 2012. A detailed questionnaire was used to collect data on maternal and infant characteristics. Maternal variable as maternal age, maternal education level and maternal anthropometry like maternal height, weight and mid upper arm circumference (MUAC) was used as maternal variable. Family income was also used as a variable. Whenever mothers come to maternal clinic were interviewed and investigated. Maternal age, level of education and family income was collected from direct questionaries' and from hospital log book; maternal health card was used to collect maternal anthropometric measurement during pregnancy. Additionally maternal height and MUAC was also measured during the interview period according to standardized techniques \[[@CIT0020]\]. A non-stretchable tape made of fiberglass was used to measure maternal MUAC. All of the maternal variables were categorize into several category ([Table 1](#T0001){ref-type="table"}) to aid in data processing and evaluation of specific effect on infant birth weight. The entire maternal characteristics were used as independent variable. Infant birth weight was used as dependent variable. Birth weight of every infant was collected from maternal and infant health cards and from visited hospital\'s log book. Birth weight was categorized into three categories ([Table 1](#T0001){ref-type="table"}) as \< 2500 gm, 2500-2999 gm and = 3000 gm. According to WHO cutoff points \< 2500 gm of birth weight was considered as Low Birth Weight (LBW) \[[@CIT0021]\]. Details data on the pattern of breast feeding practice was also collected during the interview session. The women were asked about their current and previous infant feeding practices, including the time of initiation and duration of breastfeeding, use of formula, bottle feeding and introduction of complementary food. Use of any pre-lacteal feeding was also documented and categorize into another group. Exclusive breast feeding (EBF) up to six months was categorized into one category and EBF for less than six months was categorizing into other category. The other categories of breast feeding practice were as breast feeding on demand and continuing breast feeding for 24 months. Three infant variables were used as Stunting, Wasting and Stunting and Wasting. Infant anthropometric measurements were done during the period of interview. Infant age was collected from infant health card or from questionnaire. A research team consisting of students and teacher were directly involved in data collection in the maternal clinic. Data were collected with the help of health assistant working in the maternal clinic, Infant weight and height was measured using electronic scale and wooden measuring board following standard measuring procedure \[[@CIT0020]\]. Before measurement, the scales were calibrated and validated by the research team. Waterlow classification was used to classify infant into normal, wasting, stunting and wasting and stunting. In Waterlow classification% height for age and% weight for height was calculated using standard reference data for height of normal infant of the same age and weight of normal infant of the same height \[[@CIT0020]\].
######
Relative distribution of infant according to their birth weight and categories of maternal variables.
Variable Total Birth Weight (gm)
-------------------------------------------------- -------------- ------------------- --------------- ---------------
**Maternal Age (years)**
Up to 20 110(21.57) 34(30.9) 48(43.6) 28(25.5)
21-25 282(55.29) 80(28.4) 112(39.7) 90(31.9)
26-30 55(10.78) 13(23.6) 20(36.4) 22(40)
≥ 31 63(12.35) 23(36.5) 23(36.5) 17(27)
**Maternal Education**
Illiterate 58(11.37) 32(55.2) 15(25.9) 11(18.9)
Primary 75(14.71) 35(46.7) 32(42.6) 08(10.7)
class 6- S.S.C[\*](#TF0001){ref-type="table-fn"} 272(53.33) 69(25.3) 11542.3) 88(32.4)
H.S.C[\*](#TF0001){ref-type="table-fn"} 52(10.20) 11(21.2) 26(50) 15(28.8)
Graduation & above 53(10.39) 03(5.7) 1528.3) 35(66)
**Family Income (Tk)**
Up to 3999 81(15.88) 35(43.2) 28(34.6) 18(22.2)
4000-7999 135(26.47) 42(31.1) 62(45.9) 31(23.0)
8000-11999 119(23.33) 36(30.1) 46(38.7) 37(31.1)
12000-15999 87(17.05/ 25(28.7) 32(36.8) 30(34.5)
≥16000 88(17.25) 12(13.6) 3539.8) 41(46.6)
**Maternal Height (cm)**
Up to 144.9 74(14.51) 32(43.2) 22(29.8) 20(27)
145-149.9 89(17.45) 33(37.1) 28(31.5) 28(31.4)
150-154.9 101(19.8) 27(26.7) 46(45.5) 28(27.8)
155-159.9 118(23.14) 32(27.1) 52(45.7) 32(27.7)
160-164.9 61(11.96) 13(21.3) 25(41) 25(37.7)
≥165 67(13.14) 13(19.4) 28(41.8) 26(38.8)
**Maternal Weight (kg)**
Up to 45.9 57(11.18) 30(52.6) 10(17.5) 17(29.8)
46-49.9 75(14.71) 38(50.7) 22(29.3) 15(20)
50-54.9 120(23.53) 30(25) 53(44.2) 37(30.8)
55-59.9 106(20.78) 30(28.3) 49(46.2) 27(25.5)
≥60 152(29.8) 22(14.5) 69(45.4) 61(40.1)
**Maternal MUAC (cm)**
≤ 22.0 75(14.71) 41(54.7) 25(33.3) 09(12.0)
22.1-24 118(23.14) 42(35.6) 51(43.2) 25(21.2)
24.1-26 133(26.08) 33(24.8) 55(41.4) 45(33.8)
26.1-28 83(16.27) 19(22.9) 31(38.6) 33(39.7)
28.1-29 50(9.8) 10(20.0) 18(36.0) 22(44.0)
\>29 51(10) 05(9.8) 23(45.0) 23(45.1)
**Total** ***(n%)*** **510(100)** **150(29.4)** **203(39.8)** **157(30.8)**
LBW = Low Birth Weight, S.S.C. = Secondary School Certificate, H.S.C. = Higher Secondary School Certificate.
Statistical analysis {#S20005}
--------------------
Maternal variables were also categorized into different groups. Descriptive statistics were used to analyze the relative distribution of birth weight in different categories of maternal variable. Each of the maternal independent variable was dichotomized into two groups as up to 20 years and more than 20 years for age; uneducated and educated and so on ([Table 2](#T0002){ref-type="table"}) and coded as 0 and 1. Maternal MUAC was dichotomized into = 22 cm and \> 22 cm and maternal MUAC = 22 cm was considered as poor maternal nutrition \[[@CIT0022]\]. Dependent variable infant birth weight was also dichotomized into two groups as low birth weight and normal birth weight and coded as o and 1 respectively. Chi square test was used to assess the relationship between dependent and independent variable. Odd Ratio and Relative Risk with 95% confident interval (CI) were calculated using crosstab, to measure the magnitude of odd and risk of LBW and infant growth retardation in terms of wasting and stunting. To interpret the relationship between pattern of exclusive breastfeeding and infant nutritional status, two (02) important maternal age groups, 26-30 years and (=31 years were selected. Infant nutritional status was also dichotomized into two groups as wasting and normal; stunting and normal and wasting and stunting and normal and codded as o and 1 respectively. Breastfeeding pattern was also dichotomized into two groups as exclusive breast feeding and non-exclusive breast feeding. Association between pattern of breast feeding (independent variable) and infant nutritional status (dependent variable) was also assessed by calculating Chi square and OR with 95% CI using crosstab. All analysis was done using SPSS software, version 15.0 (SPSS Inc, Chicago, IL, USA). P value of \<0.05 was considered to be significant.
######
Independent effects of demographic, socio-economic and anthropometric risk factors for risk for low birth weight (LBW) analyzed by binary logistic regression.
Socio-demographic Factors Pearson Chi-Square *P* Value OR[\*](#TF0002){ref-type="table-fn"} (CI 95%)
------------------------------------------------------------------------------------ -------------------- ----------- -----------------------------------------------
**Maternal Age:** (≤20 years vs. \>20)[†](#TF0003){ref-type="table-fn"} 0.151 0.697 1.095 (0.693- 1.732)
**Maternal Education:** (Illiterate vs. Literate)[†](#TF0003){ref-type="table-fn"} 20.917 0.000 3.484 (1.993-6.089)
**Family Income (Tk):** (≤ 3999 vs. ≥4000)[†](#TF0003){ref-type="table-fn"} 8.830 0.003 2.078 (1.274-3.387)
**Maternal Height:** (\<150 cm vs. ≥ 150)[†](#TF0003){ref-type="table-fn"} 12.639 0.000 2.044 (1.374- 3.043)
**Maternal Weight:** (\<50 kg vs. ≥ 50 kg)[†](#TF0003){ref-type="table-fn"} 41.910 0.000 3.835 (2.520- 5.837)
**Maternal MUAC:** (≤ 22 cm vs. ≥ 22cm)[†](#TF0003){ref-type="table-fn"} 24.817 0.000 3.385 (2.058- 5.567)
OR: Odd Ratio.
Maternal age \> 20 years, Literate, Family income ≥ 4000 Tk, Maternal height ≥ 150 cm, Maternal weight ≥ 50 kg and maternal MUAC ≥ 22 cm was considered as reference categories.
Ethical approval {#S20006}
----------------
This research was approved by the ethics committee of the faculty of Applied Science and Technology of Islamic University, Kushtia, Bangladesh. The objective of this study and study protocol were first described to the subjects and oral consent to participate in the study was taken. All of the maternal clinics that took part in this study also gave their approval.
Results {#S0007}
=======
Maternal characteristics and infant birth weight {#S20008}
------------------------------------------------
According to maternal age category, highest percentage (36.5%) of low birth weight infant was found in the age group of = 31 years; however the lowest percentage (23.6%) was found in the 26-30 age group ([Table 1](#T0001){ref-type="table"}, [Figure 1](#F0001){ref-type="fig"}, [Figure 2](#F0002){ref-type="fig"}). 40% infant were in the age group of 26-30 year with birth weight greater than 3 kg, which was the highest among all of the maternal age and birth weight category. However only 11.37% mother-infant pair were in the illiterate mother group, highest percentage (55.2%) of LBW infant were in this group, following 46.7% in the group where mother has completed primary education. In contrast, mothers, who were graduated, were shown to have the highest percentage (66) of infants with normal weight and only 5.7% of infant with LBW. According to family income category the lowest family income category (≤ 3999) were with highest percentage (43.2%) of low birth weight infant, whereas the lowest percentage (13.6%) of the LBW were in the highest family income category (≥ 16000). Percentages (46.6%) of normal weight infant were also high in the highest family income group. According to maternal height and weight category the highest percentage of LBW infant (43.2 and 52.6% respectively) were found in the lowest maternal height (\<145 cm) and weight (\< 46 kg) category respectively. Mother with mid upper arm circumference \< 22 cm were tend to give highest percentage (54.7%) of LBW infant. As a total, 29.4% of infant were with LBW.
{#F0001}
{#F0002}
Except age (p = 0.697), all of the maternal independent variable were significantly (p ≤ 0.003) associated with LBW ([Table 2](#T0002){ref-type="table"}). Mother with age ≤ 20 years were associated with similar Odd of having LBW (OR: 1.095, CI: 0.693- 1.732) as compared with mother with age greater than 20 years. Illiterate mother were shown to have more than 3 times more odd (OR: 3.484, CI: 1.993-6.089; P = 0.000) of having low birth weight infant than those of educate mother. Mothers from low income family (\< 3999/= BDT) were also tend to have more (OR: 2.078, CI: 1.274-3.387; p = 0.003) LBW babies than mothers from higher income family (\>16000/= BDT). Similarly maternal height, weight and MUAC also showed a significant effect (p = 0.000) on LBW. Mother weighted less than 50 kg were associated with 383% (CI: 2.520- 5.837) more odd of having LBW than mother with higher weight. Mother with MUAC \< 22 cm also shown similar result (OR: 3.385, CI: 2.058- 5.567; p = 0.000).
Breast feeding practice and infant nutritional status {#S20009}
-----------------------------------------------------
Among the studied mother-infant pair 61.1% mother were shown to initiate breastfeeding within 1 hour of child birth, while the rest of the mother (38.1%) did not do that ([Table 3](#T0003){ref-type="table"}). The study showed that 45% mother breastfeed their child exclusively for the first six month, whereas 90% mother gives breast milk to their child on demand. 97.8% of studied mothers were shown to have a wish to continue the breastfeeding up to 2 years of infant\'s age. Only 21% of studied mothers were shown to give pre-lacteal feed. Honey was the most common among the pre-lacteal feeds. Analysis of the association of exclusive breast feeding practice with infant nutritional status were shown that among 25 exclusively breastfed infant only 2 were found with wasting; while in the case of 33 non-exclusively breastfed infant the number was 6 ([Table 4](#T0004){ref-type="table"}). Using Waterlow classification, weight and height of each infant was compared with the reference infant\'s weight and height with same age. The highest percentages (26.7%) of nutrition related complication was growth retardation which was among non-exclusively breastfed infant and the second highest was both wasting (20%). Among exclusively breast fed infant only 16% were stunted, whereas only 8% and 4% were wasted and both stunted and wasted respectively. In the case of greater than 30 years of maternal age group, 52.6% infants were stunted among nonexclusively breastfed which was the highest percentage both in exclusively and non-exclusively breastfed infant. The percentages of all form of malnutrition related complication were higher among non-exclusively breast fed infant than exclusively breastfed infant. The crosstab analysis were shown that all of the association in both maternal age groups was non-significant (p \> 0.05), except only the association of wasting with breastfeeding practice (p\< 0.05). However the odd and risk ([Table 5](#T0005){ref-type="table"}) of all types of malnutrition related complication were higher among non-exclusively breastfed infant, but was not found statistically significant (1 was included in the 95% CI).
######
Pattern of Breastfeeding practice
Breast Feeding Practice Total Number (%) Positive *n (%)* Negative *n (%)*
----------------------------------------------------------------------------- ------------------ ------------------ ------------------
Initiated breastfeeding within one hour of child birth. 510 (100) 312(61.1) 198(38.9)
Give pre-lacteal feeds. 510 (100) 107(21) 403(79)
Breastfed on child demand. 510 (100) 464(90.9) 46(9.1)
Children that Breastfed Exclusively (EBF) for the first six months of life. 510 (100) 230(45) 280(55)
Continuing and will continue breastfeeding for 2 years. 510 (100) 499(97.8) 11(2.2)
######
Prevalence of wasting Stunting and wasting and wasting according to pattern of exclusive breastfeeding practice
Breast Feeding Status Maternal age Group 26-30 years
--------------------------------- ---------- -------------------------------- ----------- ----------- ----------- ----------- ------------
Exclusive Breast Feed Infant 25 (100) 02 (8) 23 (92) 01 (4) 24 (96) 04 (16) 21 (84)
Non-Exclusive Breastfeed Infant 30 (100) 06 (20) 24 (80) 03 (10) 27 (90) 08 (26.7) 22 (73.3)
Total 55 (100) 8 (14.5) 47 (85.5) 04 (7.3) 51 (92.7) 12 (21.8) 43 (78.20)
**Maternal Age Group \> 30**
Exclusive Breast Feed Infant 25 (100) 04 (24) 21 (76) 02 (8) 23 (92) 11 (44) 14 (56)
Non-Exclusive Breastfeed Infant 38 (100) 15 (39.5) 23 (60.5) 07 (18.4) 31 (81.6) 20 (52.6) 18 (47.4)
Total 63 (100) 19 (31.1) 44 (69.9) 09 (14.3) 54 (85.7) 31 (49.2) 32 (50.8)
######
Independent Effects of exclusive breast feeding practice for risk for infant growth retardation analyzed by binary logistic regression
Effect Pearson Chi-Square *P* Value OR[\*](#TF0004){ref-type="table-fn"} (CI 95%)
-------------------------------------------------------------------- -------------------- ----------- -----------------------------------------------
**Maternal Age Group ≤ 30 years[†](#TF0005){ref-type="table-fn"}**
Wasting (Non-EBF vs. EBF) 1.580 0.209 2.875 (0.526- 15.729)
Stunting and Wasting (Non-EBF vs. EBF) 0.728 0.617 2.667 (0.260- 27.381)
Stunting (Non-EBF vs. EBF) 0.910 0.340 1.909 (0.499- 7.298)
**Maternal Age Group \>30 years[†](#TF0005){ref-type="table-fn"}**
Wasting (Non-EBF vs. EBF) 3.945 0.047 3.424 (0.979- 11.969)
Stunting and Wasting (Non-EBF vs. EBF) 1.337 0.298 2.597 (0.493- 13.677)
Stunting (Non-EBF vs. EBF) 0.450 0.503 1.414 (0.513- 3.900)
OR: Odd Ratio.
Exclusive Breast Feeding (EBF) was considered as reference category
Discussion {#S0010}
==========
The study was conducted on sample of 510 mothers and their infants selecting randomly to assess the effects of maternal status and breastfeeding practices on infant development. Then we selected some factors randomly to identify the effect of maternal status on infant development. Two types of characteristics of mothers were studied against different ranges of birth weight of infant to identify the effect of maternal status on infant and the effect of the pattern of the exclusive breast feeding practices on infant nutritional status were studied.
In Asia the prevalence of LBW is highest and in South-East Asia one third of the newborn are with LBW \[[@CIT0023]\]. The current study showed the prevalence of LBW as 29.4%, however according to study of Bangladesh Bureau of Statistics the prevalence was 36% in Bangladesh \[[@CIT0024]\]. A study reported the rate of LBW as 31.2%, studied on 1000 mothers in Dhaka Medical College \[[@CIT0021]\]. A study conducted in Bangladesh reported that, 15.9% mothers were found to be under 20 years and 8.9% were above 30 years of age \[[@CIT0010]\]. The current study also revealed that the percentage of mother-infant pair were highest in the 21-25 age groups. About 11.37% of mothers had no formal educational background, 14.71% had primary level, 53.33% had six to S.S.C level, 10.2% had higher secondary and 10.39% completed graduation. There is some dissimilarity in educational status in comparison to the study conducted by Karim and Taylor (1997) \[[@CIT0010]\].
It was revealed from the logistic regression analysis that maternal age is not an important predictor of LBW. However the percentages of LBW infant in the maternal age group of up to 20 years and more than 30 years were little bit higher than the age groups of 21-30 years though was not significant. Thus maternal age ranging from 21-30 years was found to be most suitable age group for giving birth with normal birth weight. The findings of the present study were in agreement with many similar studies both in developed and developing countries \[[@CIT0002], [@CIT0009], [@CIT0010], [@CIT0025]\]. When maternal age groups were dichotomized into to two groups for binary logistic regression analysis for the effect of maternal age \< 20 years on LBW did not show any significant effect (p = 0.697). Mother\'s educational status has great influence on birth weight of the infant. The present study showed a decreasing trend of the percentage (55.2% - 03%) of LBW with increasing maternal educational status. Maternal education was shown to be an important predictor of LBW. Literate mother were shown to give less number of babies with LBW. Many other studies have shown similar findings \[[@CIT0009], [@CIT0010], [@CIT0026]\]. It is possible that the effect of education on birth weight is to some extent a reflection of other influences like maternal nutrition, which can also be enhanced by better income opportunities for educated couples. The highest percent (26.47%) of mother-infant pair were in the family income group of 4000-7999 BDT. Family income was associated with LBW in a similar trend of decreasing percentage of LBW with increasing family income. In the present study, it was revealed that the highest incidence of LBW (43.2%) was found among the families having income less than 4000 BDT and the lowest LBW (13.6%) among the families with income equal to or greater than TK. 16000 BDT. Family income less than 4000 BDT showed a higher Odd (Odd: 2.078, CI: 1.274-3.387) of having LBW than more family income and the association was significant (p = 0.003). Others studies reported similar findings in Bangladesh \[[@CIT0021], [@CIT0027]\] and other countries \[[@CIT0028], [@CIT0029]\]. Mother in a family with higher income usually gets proper nutrition and better care than a mother in low income family.
The magnitude of effect of maternal weight and maternal MUAC were more intense than the maternal height but statistically significant. Regression analysis showed that women with height and weight \< 150 cm and \< 50 kg respectively were more likely to have 2 and 3 times more LBW respectively than women with more height and weight. A similar study using cut-off points for low maternal weight at \< 45 kg was shown similar findings of having (OR: 3.51, 95%CI: 1.74-7.15) LBW; however the same study shown a non-significant association between maternal height and LBW using a cut-off point of \< 145cm \[[@CIT0022]\]. Some other study reported significant association between maternal height and LBW \[[@CIT0030], [@CIT0031]\], while some others study report were non-significant \[[@CIT0032]\]. However the other study findings were controversial, the current study confirms maternal weight and height as strong predictor of LBW. Mother with MUAC \< 22 cm were more than 3 times likely to have LBW babies as compared to mother with MUAC SUPER_OR_EQUAL 22cm and the association between these variables was found statistically significant (p = 0.000). Ojha and Malla in 2007, Husaini et al. in 1995 and Neumann et al. in 1995 also reported positive association between maternal MUAC and LBW, though different cutoff-points were used \[[@CIT0022], [@CIT0033], [@CIT0034]\]. MUAC is usually reflected by the nutritional status. Mother with MUAC \> 22 cm were tends to give more babies with normal birth weight. Mother with MUAC \< 22 cm was negatively associated with birth weight of infants.
Since the time immemorial mammalians babies were feeding with their mother\'s breast milk. Breast milk is considered to be the only and ideal food for the infants for the first six months of life and exclusive breast feeding for the first six months of life is an important strategy for the proper growth and development of infants. In spite of an increased focus on the necessity to ensure exclusive breast feeding for the first six months of life, the percentage of exclusive breast feeding is not appreciable both in developed and developing country. However the practice of feeding colostrum has improved in the past decade (87%) and 43% of children were exclusively breastfeed for the first six months of life \[[@CIT0035]\], whereas 45% of mother was found to breast feed exclusively in our current study. More than 97% of studied mother continued and wish to continued breastfeeding for up to 2 years, which is certainly a positive sign for breast feeding practice. The use of honey and other sweet item as pre-lacteal feeding were more common as a cultural believe that the first milk is dirty and as a believer to give honey. The potential contamination for pre-lacteal feeding are more and there are chance of developing allergic manifestation in the future life \[[@CIT0036], [@CIT0037]\]. However several studies showed higher percentage (79-93%) of giving pre-lacteal feeding \[[@CIT0038], [@CIT0039]\], current study showed to have only 21% to do this practice. According to WHO, it is essential to initiate breastfeeding within half an hour after birth \[[@CIT0040]\]. Early initiation of breastfeeding helps to ensure proper nutrition and first defense for the infant and help to secrete milk \[[@CIT0041]\] and initiate a positive bonding between mother and infant \[[@CIT0042]\]. However only 22% of mothers were found to initiate breastfeeding within 1 hour as reported by Sayed Mahmood et al., 61.1% mother were found to do this in the current study, \[[@CIT0043]\]. To identify the effects of exclusive breastfeeding practices on infant development, two important maternal age groups were selected. However beneficial effect of exclusive breastfeeding has been supported by several studies \[[@CIT0044]--[@CIT0046]\], the current study did not show any effect of exclusive breastfeeding on child growth, compared with non-exclusive breastfeeding using crosstab analysis. The odd of having under nutrition in non-exclusively breast feed infant was higher than in exclusively-breast feed infant, however the association between breast feeding practice and nutritional status of infant was found non-significant. This might be due to small sample size (55 and 63).
Conclusion {#S0011}
==========
Low birth weight is a common problem in developing world where malnutrition effect particularly on maternal nutrition. This study identified that lower level of maternal education and anthropometry as well as family income increases the risk of LBW. This indicates increased level of maternal education ensures maternal awareness and higher income help to ensure proper food, which is essential for proper maternal nutrition and perinatal infant development. Practices of exclusive breastfeeding were found less than half of the studied mother. It is important to ensure exclusive breastfeeding practice for all of the mothers. Though in the current study, Exclusive breastfeeding up to 6 months of infant\'s age were shown to have no effects on infant\'s nutritional status, exclusive breast feeding is important for infant nutrition. Practice of EBF was not found satisfactory in the studied mother infant pairs. Measure should need to ensure exclusive breastfeeding up to six months and early initiation of breastfeeding within one hour.
The study was supported by the Dept. of Applied Nutrition and Food Technology, Islamic University, Kushtia, Bangladesh. The authors are thankful to the authorities of Kushtia General Hospital and Maternity Health Clinic. The authors are also thankful to the participant mothers and infants for their cooperation and patients.
Competing interests {#S0012}
===================
The authors declared that they have no conflicts of interest.
Authors' contributions {#S0013}
======================
Mohidul Islam: Principally responsible for data collection and gathering and partly in writing of the paper. Shahinur Rahman: Contributed in writing and revising of the paper. Kamruzzaman: Responsible for writing and final revising of the article and statistical analyses and interpretation of result was done by this author. Mominul Islam: Contributed in writing and revising of the paper. Abdus Samad: Supervised the whole and took part in the conception of methodology. All the authors have rad and approved the final version of the manuscript.
| {
"pile_set_name": "PubMed Central"
} |
*Bioscience Reports* (2019); <https://doi.org/10.1042/BSR20182466>
The authors are retracting their accepted manuscript "MicroRNA-367 promotes progression of hepatocellular carcinoma through PTEN/PI3K/AKT signaling pathway" at the request of the corresponding author and Hunan Provincial Tumor Hospital, in conjunction with the policies of their Institutional Review Board (IRB), as their study had not been granted at Hunan Provincial Tumor Hospital. They had also noticed an error in terms in the institutional review board within their Declarations.
All the named authors of the original accepted manuscript have agreed to this retraction, and apologise for any inconvenience caused to the readers and to the Editor of *Bioscience Reports*.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
The term small for gestational age (SGA) is used to describe a neonate who is smaller than average after adjusting for sex and gestational age (GA). SGA infants usually have a birth weight below the 10th percentile for GA \[[@b1-apem-2019-24-4-226]\]. Some infants born SGA, particularly due to intrauterine growth restriction (IUGR), suffer acute and chronic consequences. In the neonatal period, some suffer from perinatal events, such as hypoglycemia, gastro-esophageal reflux, and hypothermia \[[@b2-apem-2019-24-4-226]\]. Such individuals can be and remain short in stature or suffer premature adrenarche and polycystic ovarian syndrome \[[@b3-apem-2019-24-4-226],[@b4-apem-2019-24-4-226]\]. Furthermore, SGA individuals demonstrate significantly increased risks for obesity, hypertension, dyslipidemia, insulin resistance, type 2 diabetes mellitus, and cardiovascular diseases (CVD) from childhood to adulthood \[[@b2-apem-2019-24-4-226]\]. For proper assessment and management of SGA individuals, up-to-date, ethnicity-specific birth length or birth weight references according to GA are needed \[[@b5-apem-2019-24-4-226]\]. Many countries, including the United States, have reported new birth weight or birth length references by sex and GA, as the old references no longer fit the contemporary population \[[@b6-apem-2019-24-4-226]-[@b11-apem-2019-24-4-226]\]. The new references and cutoffs can be used to identify children at risk for short adult stature and future metabolic disease. Previously, we reported Korean birth weight references according to GA and sex \[[@b12-apem-2019-24-4-226]\]. However, birth length references based on the Korean population remained absent at the time of this study. Thus, the objectives of this study were as follows: (1) to create sex-specific birth length reference data for Koreans; (2) to create a sex-specific birth length cutoff to define SGA; and (3) to compare Korean-specific birth length reference data and cutoffs to those of other countries.
Materials and methods
=====================
1. Study subjects
-----------------
This study was performed using data from the 4th Korean National Health and Nutrition Examination Survey (KNHANES-IV; 2007--2009), a cross-sectional and nationally representative survey with a multistage and stratified sampling design conducted by the Division of Chronic Disease Surveillance of the Korea Centers for Disease Control and Prevention. Written informed consent was secured from the respective caretakers of all participants before the study began, and the KNHANES-IV was conducted following ethical approval by the Institutional Review Board of Korea Centers for Disease Control and Prevention (No: 2007-02-04-P, 2008-04EXP-01-C, 2009-01CON-03-2C). In the KNHANESIV, the birth weight and birth length of children 1--3 years of age (n=895, representing 42,756 subjects) were available. Those without birth length were excluded (n=52). Thus, the final analytical sample consisted of 843 subjects (male, 459; female, 384).
2. Reference birth weights and lengths for 36-37 week and 38- to 41-week gestational ages
-----------------------------------------------------------------------------------------
In the KNHANES-IV, GA was identified using the closedended question \"Was your baby born between the expected birth date±2 weeks?\" If the answer was yes, the baby was defined as full-term (38--41 weeks). When the birth date was between 2--4 weeks earlier than expected, the baby was designated as GA 36--37 weeks. For reference, a GA of 38 weeks represented 38 weeks + 0--6 days. Multiple births (n=10) were excluded, as the negative impact on intrauterine growth is well known. Thus, we used data of singleton newborns. We further excluded \"extreme outliers,\" which were defined as value \> 2 times the interquartile range (25th to 75th percentiles) above the third quartile for each GA \[[@b13-apem-2019-24-4-226]\]. There were 6 extreme outliers based on the birth length and 3 based on birth weight. Those with a GA below 36 (n=6) or an unspecified GA (n=18) were also excluded. Thus, data of 67 (39 males) singleton newborns with GA of 36--37 weeks and data of 736 (399 males) with GA of 38--41 weeks were used for reference. The 3rd, 10th, 25th, 50th, 75th, 90th, and 97th percentiles for male and female birth lengths were calculated.
3. Small for gestational age definition
---------------------------------------
We used 2 identifiers to define SGA: a birth weight below the 10th percentile (SGAW) and a birth length below the 10th percentile (SGAL) \[[@b1-apem-2019-24-4-226],[@b6-apem-2019-24-4-226]\].
4. Statistical analysis
-----------------------
All statistical analyses were performed using SPSS ver. 17.0 (SPSS Inc., Chicago, IL, USA). The *P*-values of \<0.05 were considered significant. The data are presented as the mean±standard deviation concerning the percentile values for birth weight and length according to GA of 36--37 weeks and of 38--41 weeks.
Results
=======
1. Subject characteristics
--------------------------
The characteristics of the study subjects are presented in [Table 1](#t1-apem-2019-24-4-226){ref-type="table"}. The 843 subjects in this study were born between 2005 and 2009 in Korea. The mean birth length was 50.7±2.9 cm, and the mean birth weight was 3,283±438 g. Female infants had lower birth lengths and birth weights than male infants (50.2±3.2 cm vs. 51.1±2.7 cm, 3,225±464 g vs. 3,327±413 g, respectively; *P*\<0.001). There was no difference in birth weight concerning vaginal delivery and caesarean section. However, the birth lengths of infants born by caesarean section were shorter those born by vaginal delivery (50.4±3.4 cm vs. 50.9±2.6 cm, respectively; *P*=0.046). Full-term neonates comprised 91.3% of the participants (n=770), and preterm neonates younger than 37 weeks made up 8.7% (n=73).
2. Percentile distributions of birth length and birth weight by sex
-------------------------------------------------------------------
The mean and percentiles of birth length and weight according to GA of males are presented in [Table 2](#t2-apem-2019-24-4-226){ref-type="table"}, and the corresponding female data are presented in [Table 3](#t3-apem-2019-24-4-226){ref-type="table"}. In creating the percentiles, outliers and twins were excluded; therefore, data from 736 (399 males) singleton infants born between 38--41 weeks and from 67 infants born between 36--37 weeks were used. At full-term, the 10th percentile birth length was 48 cm in both males and females, the 10th percentile birth weight was 2,900 g for males and 2,828 g for females, and the 3rd percentile birth length was 47 cm for males and 46 cm for females.
3. Prevalence of SGA
--------------------
The prevalence (%) of SGA in full-term neonates when using the 10th percentile as the cutoff point for either birth length or weight is shown in [Fig. 1](#f1-apem-2019-24-4-226){ref-type="fig"}. In full-term neonates, a total of 745 subjects (including outliers) were divided based on birth length and weight. Of the subjects, 86.3% (n=643) possessed a greater than 10th percentile value for both birth length and weight (group I). The others fell into 3 groups: 3.9% (n=29) were of normal length and low weight (group II); 7.0% (n=52) were short and of normal weight (group III); and 2.8% (n=21) were short and of low weight (group IV). Considering SGA based only on birth weight, short newborn infants of normal weight (group III) were neglected from all SGA classifications, consisting of 51% of all individuals.
Discussion
==========
In this study, we introduced references for Korean birth length and weight based on the nation-wide survey data from the KNHANES-IV. This new Korean birth length data may be useful in assessing long-term health risks, such as short stature in childhood and metabolic risk in adulthood for children born with SGA. Furthermore, we showed that approximately half of infants classified as SGA, who are short but have a normal birth weight, might be misclassified as appropriate gestational age (AGA); this supports the need for updates and Korean-specific reference curves for birth length.
Contemporary ethnicity-specific birth weight references are needed to identify SGA neonates who might suffer from acute and chronic consequences \[[@b5-apem-2019-24-4-226]\]. Some SGA neonates are born following IUGR, and the suboptimal fetal growth occurring in IUGR fetuses is an important cause of perinatal mortality and morbidity \[[@b2-apem-2019-24-4-226]\]. Furthermore, SGA children have demonstrated significantly increased risks for obesity, hypertension, dyslipidemia, insulin resistance, type 2 diabetes mellitus, and cerebrovascular disease from childhood to adulthood \[[@b2-apem-2019-24-4-226],[@b14-apem-2019-24-4-226]\]. Furthermore, approximately 13% of children born with SGA remain short-statured and constitute a significant proportion of adults with short stature \[[@b15-apem-2019-24-4-226]\]. In a Swedish birth cohort, SGA infants demonstrated a 7-fold higher risk of short stature in adulthood compared to non-SGA infants \[[@b4-apem-2019-24-4-226]\]. They also revealed that the majority (61.7%) of the short newborn population may be misclassified as AGA if classified only in terms of birth weight. Furthermore, a slightly higher relative risk of short stature was obtained using birth length rather than birth weight to define SGA \[[@b16-apem-2019-24-4-226],[@b17-apem-2019-24-4-226]\]. Additionally, the postnatal growth patterns were similar in SGAL and SGAW. In our study, when SGA was defined only by birth weight, 71.2% of short newborns were classified as AGA.
Growth hormone (GH) treatment for SGA children is standard practice. Many studies have shown that adult height is increased by GH therapy in these children \[[@b18-apem-2019-24-4-226],[@b19-apem-2019-24-4-226]\]. Short-term GH treatment reduces body fat while promoting lean body mass and may reduce cardio-metabolic risk in SGA children \[[@b20-apem-2019-24-4-226]\]. Furthermore, a study from Sweden reported that GH therapy showed a significant and cost-effective improvement in health and quality of life \[[@b21-apem-2019-24-4-226]\]. Thus, the U.S. Food and Drug Administration and the European Medicines Agency have approved GH treatment for children with short stature who are born SGA. In Korea, medical insurance benefits have been applied to GH treatment for SGA children since 2014. Therefore, it is important to define SGA not only to reduce perinatal morbidity, but also to manage short stature in children and adolescents.
Moreover, it is well known that there are racial and ethnic differences in birth weight and birth length \[[@b22-apem-2019-24-4-226]-[@b24-apem-2019-24-4-226]\]. Furthermore, significant interactions have been found between race and maternal variables, especially after beginning antenatal care. Thus, ethnicity- and sex-specific references for birth weight and birth length are important. As the KNHANES-IV data did not show the specific GA, we first compared the birth weight reference of this study with that in our previous work concerning Korean birth weight. The values for the 10th, 50th, and 90th percentiles of this study for full-term (GA of 38--41 weeks) were between a GA of 39 weeks and a GA of 40 weeks in previous studies concerning both sexes \[[@b25-apem-2019-24-4-226]\].
Compared to the new USA intrauterine growth data of a GA of 39 weeks and a GA of 40 weeks \[[@b6-apem-2019-24-4-226]\], our references for full-term birth length values were similar to the USA data for a GA of 40 weeks. The 3rd, 10th, 25th, 50th, 75th, 90th, and 97th percentiles for birth length at a GA of 40 weeks were 46.9, 48.4, 49.9, 51.6, 53.2, 54.7, and 56.1 cm, respectively, in male neonates and 46.1, 47.6, 49.1, 50.8, 52.4, 53.8, and 55.1 cm in female neonates. Compared to the new Japanese neonatal anthropometric charts for GA \[[@b11-apem-2019-24-4-226]\], our reference values for the full term were higher than the former. The 3rd, 10th, 50th, 90th, and 97th percentiles for the birth length of a GA of 40 weeks were 45.9, 47.1, 49.4, 51.7, and 52.7 cm, respectively, in both sexes. These variations can mainly be attributed to ethnic differences, which were previously discussed in a study of the Korean reference for birth weight \[[@b12-apem-2019-24-4-226]\]. The other reason for these differences is that the new Japanese charts were based on data from subjects born only via vaginal delivery. Compared to other countries\' GA of 40 weeks, the Korean birth length was similar to that of Turkey and Israel and higher than that of Indonesia \[[@b26-apem-2019-24-4-226]-[@b28-apem-2019-24-4-226]\].
There were some limitations in the present study. First, the most important limitation of the provided reference is that, like other population-based studies, it is cross-sectional in nature rather than comprised of longitudinal measurements \[[@b5-apem-2019-24-4-226],[@b29-apem-2019-24-4-226]\]. Second, there was a small number of subjects in the KNHANESIV, and the specific GA per subject was lacking, which invalidated the study\'s ability to create a reference concerning length for GA. Third, it was not easy to distinguish the infant of a multicultural family, but their portion is not higher than the present, so the effect on the result was considered insignificant. Thus, further research, including greater numbers of subjects with specified GAs, is needed concerning Koreans.
In conclusion, we calculated the means and percentiles for birth length in Korean neonates based on data from the KNHANES-IV. We created new reference curves using the 10th percentile as the cutoff point for birth length and weight in Korean neonates. We also created a new cutoff for SGA by birth length. As some SGA neonates experienced short-term and long-term health risks, these data and curves provide useful information for not only Korean endocrinologists, but also for other specialists, including neonatologists, for planning research or targeting prevention of metabolic diseases, such as obesity, diabetes, and CVD.
**Conflicts of interest**
No potential conflict of interest relevant to this article was reported.
**Ethical statement**
The Research Ethics Committee of the Korea Center for Disease Control approved the study protocol (No:2007-02-04-P, 2008-04EXP-01-C, 2009-01CON-03-2C), and all participants or their parents signed a written informed consent form.
{#f1-apem-2019-24-4-226}
######
The characteristics of the study subjects
Variable No. (%) Birth weight (g) *P*-value Birth length (cm) *P*-value
-------------------- ------------ ------------------ ----------- ------------------- -----------
Total 843 3,283±438 50.7±2.9
Sex
Male 459 (54.4) 3,327±413 \<0.001 51.1±2.7 \<0.001
Female 384 (45.6) 3,225±464 50.2±3.2
Multiplicity
Singleton 833 (98.8) 3,289±436 0.001 50.8±2.9 0.002
Twin+ 10 (1.2) 2,855±349 47.9±2.1
Delivery type
Vaginal 522 (61.9) 3,295±403 0.333 50.9±2.6 0.046
Caesarean section 321 (38.1) 3,264±491 50.4±3.4
Values are presented as mean±standard deviation unless otherwise indicated.
######
Mean and percentiles of the birth length and weight of male Korean neonates (n=440)
GA^[\*](#tfn1-apem-2019-24-4-226){ref-type="table-fn"}^ week No. Mean±SD Percentiles
------------- -------------------------------------------------------------- ----------- ----------- ------------- ------- ------- ------- ------- ------- -------
Length (cm) 38--41 399 51.2±2.4 47.0 48.0 50.0 51.0 52.2 54.0 57.0
36-37 39 50.4±2.5 44.2 47.0 49.0 50.5 52.0 53.0 56.4
\<35 2 46.5±2.1 \- \- \- \- \- \- \-
Weight (g) 38--41 399 3,358±367 2,693 2,900 3,120 3,360 3,600 3,800 4,020
36-37 39 3,225±438 2,435 2,780 2,900 3,200 3,460 3,780 4,275
\<35 2 2,420±113 \- \- \- \- \- \- \-
SD, standard deviation.
GA: gestational age; 36 weeks represents 36 weeks + 0--6 days.
######
Mean and percentiles of the birth length and weight of female Korean neonates (n=369)
GA^[\*](#tfn2-apem-2019-24-4-226){ref-type="table-fn"}^ week No. Mean±SD Percentiles
------------- -------------------------------------------------------------- ------------- ----------- ------------- ------- ------- ------- ------- ------- -------
Length (cm) 38--41 337 50.5±2.8 46.0 48.0 49.0 50.0 52.0 54.0 56.4
36--37 28 49.0±2.6 43.0 45.0 47.4 49.0 50.6 52.0 53.7
\<35 4 40.6±7.5 \- \- \- \- \- \- \-
Weight (g) 38--41 337 3,293±374 2,612 2,828 3,040 3,280 3,535 3,752 4,064
36--37 28 2,899±468 1,930 2,140 2,548 2,960 3,230 3,506 3,620
\<35 4 1,673±1.007 \- \- \- \- \- \- \-
SD, standard deviation.
GA: gestatioal age; 36 weeks represents 36 weeks + 0--6 days.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
The endoplasmic reticulum (ER) is the site of synthesis, maturation, folding, and oligomerization of newly synthesized proteins, such as secretory and membrane proteins. Moreover, the ER plays a key role in protein homeostasis. The folding process is regulated by a group of proteins, referred to as ER chaperones \[[@B1]\]. One of them is oxygen regulated protein 150/glucose regulated protein 170 (ORP150/GRP170) alternatively known as HSP12A \[[@B2]\]. It was originally identified in cultured astrocytes exposed to hypoxia. The ORP150 is a part of the ER machinery that assists in the assemble and folding of secretory and membrane proteins \[[@B2], [@B3]\]. Its expression is upregulated in hypoxia, serum starvation, glucose deprivation, ischemia, and treatment with tunicamycin or 2-deoxyglucose. The ORP150 plays a cytoprotective role for the maintenance of cellular viability \[[@B3]\].
Many stress conditions, such as hypoxia or nutrient deprivation, slow down the folding process and cause accumulation of unfolded/misfolded proteins in the cell \[[@B4]--[@B6]\]. Hypoxia is a factor inducing ER stress, although the underlying mechanism is unknown. Yoshida \[[@B7]\] suggested that during this process glycolytic enzymes are induced to sustain ATP production and then cells consume glucose. Perhaps a decrease in glucose concentration induced by hypoxia inhibits N-glycosylation, leading to ER stress \[[@B7]\].
The accumulation of unfolded proteins activates the so-called "unfolded protein response" (UPR), which involves at least four types of reaction: general translational attenuation (phosphorylation of the translation initiation factor 2); increased folding capacity (upregulation of ER chaperones); enhanced ER-associated degradation of misfolded and unfolded proteins; and, if the stress is severe and unsolvable, apoptosis \[[@B4]--[@B6]\].
Misfolded and unfolded proteins are returned from the ER to the cytosol and degraded in the proteasome \[[@B8]\]. The ubiquitin proteasome pathway is responsible for cell quality control by eliminating defective proteins from the cytosol and endoplasmic reticulum. Proteins are degraded by the ubiquitin proteasome pathway via two distinct steps: the covalent attachment of multiple monomers of ubiquitin molecules to a protein substrate and degradation of the tagged protein by the 26S proteasome \[[@B8], [@B9]\].
Bortezomib ((\[(1*R*)-3-methyl-1-({(2*S*)-3-phenyl-2-\[(pyrazin-2-ylcarbonyl) amino\]propanoyl}amino)-butyl\]boronic acid), velcade, formerly known as PS-341) is a competitive inhibitor of 20S proteasome activity in whole cells. It is used in the treatment of multiple myeloma and some forms of non-Hodgkin\'s lymphoma \[[@B8], [@B9]\]. Bortezomib inhibition of the proteolytic activity of the 20S proteasome has been shown to induce proapoptotic ER stress and inhibit proteasome degradation of I*κ*B, an inhibitor of nuclear factor-*κ*B (NF-*κ*B) in the cancer cell \[[@B8]--[@B10]\].
The inhibition of the proteasome results in many toxic effects, including the accumulation of unfolded and damaged proteins \[[@B11]--[@B13]\]. In response to proteasome inhibition, the cell induces specific protective mechanisms, including the unfolded protein response \[[@B11]\], autophagy \[[@B14], [@B15]\], and, if the damage is severe, apoptosis \[[@B11], [@B16]\].
It is known that apoptosis plays the major role in control of cancer development. An organelle that can induce apoptosis when stressed is the ER. In fact, cells encounter multiple apoptotic stimuli during cancer progression, including nutrient deprivation or hypoxia. Accordingly, it was suggested that the well-documented antiapoptotic potential of chaperons may play a critical role in suppression of apoptosis in cancer cells \[[@B1], [@B8]\]. Recently, the attention of the scientists has shifted towards a novel role of cell senescence in control of cancer development, and with this shift our view on the role of chaperons in cancer has also evolved towards appreciation of the major role of chaperons in regulation of the senescence program \[[@B17]\].
Senescence can be generally characterised as a cellular stress response. This is a signal transduction process that leads to an irreversible growth arrest of cells in the G1 cell cycle phase \[[@B18]\]. Our knowledge about cell senescence occurring *in vivo*and, most importantly, as a desired result for cancer treatment, is very limited and a rather new field of research. Senescence was originally applied to the irreversible growth arrest of cells after prolonged proliferation under *in vitro*cell culture conditions. Now it has been extended to the irreversible proliferation arrest of cells caused by various stresses, including oxidative damage, telomere dysfunction, DNA damage, and oncogene induced senescence as well \[[@B19]--[@B24]\]. Tumour cells are exposed to many different---external as well as internal---sources of stress; therefore the induction of senescence constitutes an important block to tumour progression \[[@B18]\]. Senescence is a potent anticarcinogenic program and the process of neoplastic transformation involves series of events that allow cells to bypass senescence by inactivation of senescence associated pathways. Still, many tumour cells have retained the capacity to senesce in response to external stress stimuli. Most conventional anticancer therapies activate DNA damage signaling pathways, which aim to induce primarily apoptotic cell death but often treated cells do not die by apoptosis but rather undergo growth arrest or senescence. It is not fully understood currently which specific signals cause cells to undergo either senescence or initiate apoptosis \[[@B18]\]. Senescence runs parallel with an accumulation of damaged proteins at the molecular-cellular level. The attenuation of molecular chaperone inducibility and the simultaneous accumulation of damaged proteins raise the possibility that preservation of protein homeostasis is a major determinant in the occurrence and duration of cellular senescence \[[@B25]\].
The cellular and molecular effects of the proteasome inhibitor---bortezomib---on human skin fibroblasts are as yet poorly characterised. Its participation in senescence process remains unclear. We decided to study the effect of bortezomib on apoptosis and senescence of human skin fibroblasts incubated in hypoxia in comparison to normoxic conditions. We investigated the effect of bortezomib on the activity of senescence marker SA-*β*-galactosidase and induction of ORP150 chaperon in normal fibroblasts incubated in hypoxic and normoxic conditions and its correlation with apoptosis of these cells.
2. Materials and Methods {#sec2}
========================
2.1. Reagents {#sec2.1}
-------------
Dulbecco\'s modified Eagle\'s medium (DMEM), containing glucose at 4.5 mg/mL (25 mM) with Glutamax, penicillin, streptomycin, and trypsin-EDTA were provided by Invitrogen (San Diego, USA), passive lysis buffer by Promega (Madison, USA), FBS Gold by Gibco (USA), BCA Protein Assay Kit by Thermo Scientific (USA), PE Annexin V Apoptosis Detection Kit I by BD Pharmingen (USA), Senescence Detection Kit by bioVision (USA), Sigma-Fast BCIP/NBT reagent, and alkaline phosphatase-labelled anti-mouse immunoglobulin G (IgG) by Sigma (St. Louis, USA). The monoclonal antihuman ORP150 antibody was purchased from IBL (Gunma, Japan) and anti-HIF-1*α* antibody from BD Transduction Laboratory (USA).
2.2. Cell Cultures {#sec2.2}
------------------
Normal human skin fibroblasts cell line (CRL1474) were obtained from American Type Culture Collection (ATCC). Cells were maintained in high glucose DMEM supplemented with 10% heat-inactivated foetal bovine serum GOLD (FBS GOLD), 2 mM L-glutamine, penicillin (100 U/mL), and streptomycin (100 *μ*g/mL). Cells were cultured in Falcon flasks (BD) in a 5% CO~2~ incubator (Galaxy S+; New Brunswick), at 37°C. Subconfluent cultures were detached with 0.05% trypsin 0.02% EDTA in calcium-free phosphate-buffered saline (PBS) and counted in cells counter Scepter (Millipore).
2.3. Cell Viability {#sec2.3}
-------------------
Cell viability was measured according to the method of Carmichael et al. \[[@B26]\] using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT). Briefly, cells were seeded in 24-well plate at a density of 5 × 10^4^ per well. Confluent cells, cultured for 12 h, 24 h, and 48 h in normoxic conditions with different concentrations of bortezomib, were washed three times with PBS and then incubated with 1 mL of MTT solution (0.25 mg/mL in PBS) for 4 h at 37°C in 5% CO~2~ in an incubator. The medium was removed and 1 mL of 0.1 mol/L HCl in absolute isopropanol was added. Absorbance of converted dye in living cells was measured at wavelength of 570 nm. The viability of fibroblasts cultured in hypoxic conditions was calculated as percentage of control cells, incubated in normoxia. All the experiments were done in triplicate in at least three cultures.
2.4. Induction of Hypoxia in Cell Cultures {#sec2.4}
------------------------------------------
The cells (2.5 × 10^5^ in 2 mL of medium) were seeded in six-well plates and incubated until they achieved confluence (about 48 h). The high glucose DMEM was removed and replaced with 2 mL of the same fresh medium with 25 nmol/L, 50 nmol/L or without bortezomib. Control cell cultures were kept in normoxic conditions whereas the test cells were incubated in hypoxic conditions. Hypoxia was evoked by 12 h, 24 h, and 48 h incubation of cells in atmosphere containing a reduced to 1% oxygen concentration in hypoxia chamber (Galaxy 170R; New Brunswick an Eppendorf Company).
After incubation the culture media were removed, the cell layers were washed with PBS and submitted to the action of lysis buffer for determination of ORP150/GRP170 expression and protein assay. It allowed the separation of cells and extracellular matrix from the bottom of culture vessels and their suspension in the buffer. The cells incubated on six-well plates were detached with trypsin and analyzed by flow cytometry method.
2.5. Detection of Apoptosis {#sec2.5}
---------------------------
The cells (2.5 × 10^5^ in 2 mL of medium) were seeded in six-well plates and incubated until they achieved confluence. The fibroblasts were incubated in the high glucose DMEM in hypoxic and normoxic conditions with 25 or 50 nmol/L of bortezomib. The incubation was continued for 12 h, 24 h and 48 h. The cells were trypsinised and resuspended in DMEM and then in binding buffer. Cells were stained with FITC Annexin V and PI for 15 min at room temperature in the dark following the manufacturer\'s instructions (FITC Annnexin V apoptosis detection Kit I). Flow cytometry was performed using a FACSCanto II cytometer (Becton Dickinson). Data were analysed with FACSDiva software and dead cells were excluded based on forward- and side-scatter parameters.
2.6. Detection of SA-*β*-Galactosidase Expression {#sec2.6}
-------------------------------------------------
For cytochemical of SA-*β*-galactosidase staining 2.5 × 10^5^ cells in 2 ml of growth medium were seeded in Petri dishes (9.5 cm^2^) and incubated for 24 h in the high glucose DMEM. In this conditions they reached 70--80% confluence. Cells were incubated in the high glucose DMEM in hypoxic and normoxic conditions with 25 nmol/L, 50 nmol/L, or without bortezomib. The incubation was continued for 12 h, 24 h, and 48 h. After this time the medium was removed and the cell layers ware washed with 2 mL of PBS. SA-*β*-galactosidase-positive cells were detected using Senescence Detection Kit (bioVision). Briefly, the cells were fixed with 1 mL of Fixative Solution for 15 minutes at room temperature and washed with 2 mL of PBS. After washing Staining Solutions Mix was added \[940 *μ*L of Staining Solution, 10 *μ*L of Staining Supplement, and 10 *μ*L of 20 mg/mL 5-bromo-4-chloro-3-indolyl-*β*-D-galactopyranoside (X-gal)\] at pH 6.0. This pH ensures that nonsenescence cells remain unstained. The cells were incubated in these conditions for 12 h at 37°C. After incubation the cells were washed with PBS and stained cultures were viewed under an inverted microscope (Olympus CKX 41) at 200 magnification. Representative fields were photographed. The percentage of SA-*β*-galactosidase positive cells was determined by counting the number of blue cells under bright field illumination, and then the total number of cells in the same field under phase contrast.
2.7. Sodium Dodecyl Sulphate/Polyacrylamide Gel Electrophoresis (SDS/PAGE) {#sec2.7}
--------------------------------------------------------------------------
The cell cultures were washed with cold PBS and solubilised in 200 *μ*L of passive lysis buffer per well. The lysates were centrifuged at 10,000 ×g, at 4°C, for 10 min. The samples of lysates, containing 20 *μ*g of protein, were subjected to SDS-PAGE, as described by Laemmli \[[@B27]\]. The electrophoresis was run for 40--45 minutes. In each experiment 7.5% polyacrylamide gel and constant current (25 mA) were used.
2.8. Immunoblotting {#sec2.8}
-------------------
The proteins were transferred to nitrocellulose membranes and then pretreated for 2 h with Tris-buffered saline (TBS) containing 0.05% Tween 20 (TBS-T) and 5% nonfat dry milk, at room temperature. Membranes were probed with a mixture containing anti-ORP150 antibody (1 : 100) or anti-HIF-1*α* (1 : 500) in 5% dried milk in TBS-T, at 4°C, for 16 h. Then the alkaline phosphatase conjugated antibody against mouse IgG (whole molecule) was added at concentration 1 : 2500 in TBS-T for 1 h with slow shaking. The nitrocellulose was washed with TBS-T (five times for 5 min) and exposed to Sigma-Fast BCIP/NBT reagent.
2.9. Protein Assay {#sec2.9}
------------------
Protein concentration in lysates of cell layers was determined by the method of Smith et al. \[[@B28]\] using BCA Protein Assay Kit (Thermo Scientific, USA). Bovine serum albumin was used as a standard.
2.10. Statistical Analysis {#sec2.10}
--------------------------
The mean values from three independent experiments ± standard deviations (SD) were calculated. Statistical analysis was performed using Student\'s *t*-test.
3. Results {#sec3}
==========
3.1. The Effect of Bortezomib on Viability of Fibroblasts {#sec3.1}
---------------------------------------------------------
The antiproliferative effects of bortezomib were assessed by MTT method in fibroblasts cultured with increasing concentrations of bortezomib for periods of 12 h, 24 h, and 48 h. [Figure 1](#fig1){ref-type="fig"} shows that bortezomib, in the concentration from 25 nmol/L to 1000 nmol/L, caused a time-dependent and dose-dependent reduction in cell viability of the tested cells. An evident inhibition in cell viability was observed as early as 48 h. In cells treated with 1000 nmol/L of bortezomib, the effect on cell viability was markedly more pronounced than in others ([Figure 1](#fig1){ref-type="fig"}). Two concentrations of bortezomib (25 nmol/L and 50 nmol/L) were chosen for further study. Both were up to value of the half maximal inhibitory concentration (IC~50~) for bortezomib.
3.2. Detection of HIF-1*α* in Fibroblasts Submitted to Hypoxia {#sec3.2}
--------------------------------------------------------------
We also characterised the expression of HIF-1*α*, a biochemical marker of hypoxia. [Figure 2](#fig2){ref-type="fig"} shows that cells grown in normoxic conditions (for 12 h---lane 1, for 24 h---lane 3 and for 48 h---lane 5) did not demonstrate the expression of HIF-1*α*. In contrast, those cells incubated in hypoxic conditions demonstrated an intense expression of HIF-1*α* after 12 incubation hours ([Figure 2](#fig2){ref-type="fig"}, lane 2). The expression of HIF-1*α* after 24 h and 48 h (lane 4 and lane 6) was lower in comparison to 12 h (lane 2).
3.3. The Effect of Bortezomib on Apoptosis {#sec3.3}
------------------------------------------
[Figure 3](#fig3){ref-type="fig"} shows the percent of apoptotic fibroblasts in cultures incubated for 12 h, 24 h, and 48 h in normoxic and hypoxic conditions with 25 or 50 nmol/L of bortezomib. We did not observe any effect of bortezomib on apoptosis of fibroblasts in normoxic conditions. The percent of apoptotic cells did not change significantly, independently on incubation time and examined concentration of bortezomib (25 nmol/L or 50 nmol/L). Only in the case of cultures incubated for 12 h in hypoxia with 25 nmol/L of bortezomib we observed 2-fold increase in apoptosis and 3-fold rise with 50 nmol/L of bortezomib in comparison to control cells, incubated without bortezomib. In contrast to 12 h we did not observe any effect of bortezomib on apoptosis of fibroblasts incubated for 24 h and 48 h with or without bortezomib in hypoxic conditions. The percent of apoptotic cells did not change significantly after 24 h or 48 h with 25 nmol/L or 50 nmol/L of bortezomib in comparison to control ([Figure 3](#fig3){ref-type="fig"}).
3.4. The Effect of Bortezomib on SA-*β*-Galactosidase Expression {#sec3.4}
----------------------------------------------------------------
The SA-*β*-galactosidase expression was determined with the use of chromogenic X-gal substrate. Normal cells showing traits of division do not undergo reaction with this substrate.
However, the cells with evoked cell division blockade become intensely blue stained. Higher SA-*β*-galactosidase expression in cytoplasm caused the intensification of this reaction.
[Figure 4](#fig4){ref-type="fig"} shows the percent of SA-*β*-galactosidase positive fibroblasts incubated for 12 h, 24 h, and 48 h in normoxic and hypoxic conditions with 25 or 50 nmol/L of bortezomib. Bortezomib, in examined concentrations---25 nmol/L and 50 nmol/L, caused a time-dependent increase in senescence of normal fibroblasts ([Figure 4](#fig4){ref-type="fig"}). SA-*β*-galactosidase expression was 1.5- to 2.0-fold higher in the cells incubated for 12 h in normoxic conditions with 25 and 50 nmol/L of bortezomib, than in control cells. The prolongation of incubation time up to 24 h resulted in an increase of SA-*β*-galactosidase expression---up to 40% and after 48 h---up to 50%. In hypoxia conditions we observed an increase in senescence of fibroblasts incubated with 25 and 50 nmol/L of bortezomib. The percent of senescence positive cells after 12 h was similar in cultures incubated in hypoxia and normoxia ([Figure 4](#fig4){ref-type="fig"}). It is worthy of note that bortezomib exerted very strong effect on senescence of fibroblasts incubated in hypoxic conditions. The SA-*β*-galactosidase expression was raised after 24 h over 40% and after 48 h-over 70%.
[Figure 5](#fig5){ref-type="fig"} displays representative photographs of positive for SA-*β*-galactosidase staining fibroblasts incubated for 12 h, 24 h, and 48 h in normoxic and hypoxic conditions with 25 nmol/L or 50 nmol/L of bortezomib. Morphological changes of cells play an important role in senescence. After 12 h of cells incubation with 25 nmol/L or 50 nmol/L of bortezomib in hypoxic and normoxic conditions irregular and enlarged cell body shape was observed in comparison with control cells incubated in normoxia and hypoxia without bortezomib. The prolongation of incubation time up to 24 h resulted in a higher number of irregular body shape cells and development of granulation in cytosol, independently of the concentration of bortezomib in comparison to control cells incubated in hypoxia and normoxia without bortezomib. This effect was more evident after 48 h. Additionally, we observed the development of vacuolization in the cytosol of cells incubated in hypoxia conditions only with 50 nmol/L of bortezomib, in comparison to these incubated in hypoxic conditions without bortezomib.
3.5. The Effect of Bortezomib on the Expression of ORP150/GRP170 {#sec3.5}
----------------------------------------------------------------
[Figure 6](#fig6){ref-type="fig"} shows the expression of ORP150 and its glycosylated form GRP170 in fibroblasts incubated in normoxic and hypoxic conditions with 25 nmol/L, 50 nmol/L of bortezomib. In hypoxic conditions we observed induction of ORP150 expression. The expression of GRP170 in fibroblasts was observed in cultures incubated in normoxic conditions for 12 h, 24 h, and 48 h with or without bortezomib ([Figure 6](#fig6){ref-type="fig"}; lanes 1--3). It is of interest that fibroblasts incubated in normoxia did not express ORP150 ([Figure 6](#fig6){ref-type="fig"}; lanes 1--3). In hypoxic conditions we observed the induction of ORP150 expression in fibroblasts incubated for 12 h, with or without bortezomib ([Figure 6](#fig6){ref-type="fig"}; lanes: 4--6). Prolongation of incubation up to 24 h and 48 h resulted in intensification of ORP150 expression, especially after 48 h.
4. Discussion {#sec4}
=============
The 26S proteasome inhibitor, bortezomib, selectively induces apoptosis in some cancer cells. However, the nature of its selectivity remains unknown. The work presented here provides novel information on cellular effects of bortezomib in normal fibroblasts. We observed that bortezomib slightly reduces cell viability of the tested cultures in a concentration-dependent and time-dependent manner. We have found, that in normoxic conditions the percent of apoptotic cells did not change significantly, independently on incubation time and examined concentration of bortezomib (25 nmol/L or 50 nmol/L). In hypoxic conditions we did not observe any effect of bortezomib on apoptosis of fibroblasts incubated for 24 h and 48 h in comparison to control. Only in the case of fibroblasts incubated for 12 h in hypoxia significant increase in apoptosis, dependent on concentration of bortezomib, was observed. Moreover, we demonstrated that bortezomib, in examined concentrations---25 nmol/L or 50 nmol/L, causes a time-dependent increase in senescence of normal fibroblasts, especially of these incubated in hypoxic conditions. Our findings demonstrated that in contrast to normal fibroblasts, bortezomib treatment evoked strong effect on apoptosis in cancer cells. We observed a time-dependent increase up to 70% in apoptosis of MDA-MB-231 cells, especially these incubated in hypoxic conditions (non published data). Proteasome inhibition have widespread ability to disrupt tumor cell homeostasis and to induce apoptosis *in vitro* in a broad spectrum of tumor cell lines without inducing toxicity in normal cells \[[@B29]\]. There was no effect of bortezomib action on senescence of breast cancer cells (data not shown). In contrast to them, bortezomib in examined concentration caused a time-dependent and dose-dependent increase in senescence of normal fibroblasts. It is worthy of note, that bortezomib exerted very strong effect on senescence of fibroblasts incubated in hypoxic conditions.
The maintenance of protein homeostasis in cell requires the activities of chaperones and the ubiquitin-proteasome system, which together serve to inactivate and degrade misfolded proteins. When proteins are not folded properly, they are directed to 26S proteasomal degradation \[[@B10]\]. If misfolded or unfolded proteins are not degraded by the proteasome, they form aggregates and lead to the ER stress. The ER stress triggers UPR to reduce the accumulation of unfolded proteins and restore the ER function. When protein aggregation or ER stress persists, UPR signaling switches from the prosurvival to proapoptotic \[[@B2]\]. Consequently, the 26S proteasome complex also plays an important role in regulating the ER stress and cell survival \[[@B10]\]. Therefore, inhibition of the proteasomal function in cancer cells would promote apoptosis and have an antitumour function \[[@B9]\]. In fact, the inhibition of the proteolytic activity of the 26S proteasome has been shown to induce proapoptotic ER stress in multiple myeloma \[[@B30]\], pancreatic \[[@B31]\], head and neck cancer \[[@B32]\], and nonsmall cell lung carcinoma \[[@B33]\].
Proteasomal activity is essential for eliminating excess proteins and, by counteracting protein production, establishing steady protein levels \[[@B34]\]. The ubiquitin proteasome pathway represents the major pathway for intracellular protein degradation. The 26S proteasome is responsible for the degradation of approximately 80% of cellular proteins, including misfolded and mutated proteins as well as those involved in the regulation of development, differentiation, cell proliferation, signal transduction, apoptosis, and antigen presentation \[[@B8]\]. Prolonged proteasome inhibition induces stress responses that initiate apoptosis via intrinsic pathway \[[@B34]\]. This is exploited clinically in the treatment of multiple myeloma with the proteasome inhibitor bortezomib. Inhibition of proteasome activity by bortezomib is associated with an accumulation and transcriptional induction of BH3-only proteins such as PUMA, BIM, NOXA, or BIK. BH3-only proteins antagonise antiapoptotic BCL-2 family members such as BCL-2, BCL-xL, or Mcl-1 and can activate the proapoptotic members BAX and BAK \[[@B34]\]. Activated BAX and BAK form pores in the outer mitochondrial membrane, resulting in cytochrome c and Smac/Diablo release from the intermembrane space into the cytosol. This results in caspase-9 activation, inhibition of inhibitor of apoptosis (IAP) proteins, and subsequent apoptosis execution by effector caspases \[[@B35]\]. Induction of NOXA has been reported to be a key mechanism in bortezomib-mediated apoptosis, which is independent on P53 status but dependent on c-Myc \[[@B36]--[@B39]\]. Bortezomib-mediated apoptosis is accompanied by induction of c-Jun-NH~2~ terminal kinase, generation of reactive oxygen species, release of cytochrome c, second mitochondria-derived activator of caspases and apoptosis-inducing factor (AIF), and activation of the intrinsic caspase-8 pathway and extrinsic caspase-9 pathway \[[@B8]\].
In agreement with the cytoprotective role of molecular chaperones it has been shown that they can prevent stress-induced apoptotic death \[[@B25], [@B40]\]. Overexpression of Hsp70 chaperons (ORP150 belongs to this family) prevents cytochrome c release from mitochondria, blocks apoptosome formation by binding to the apoptotic protease-activating factor (Apaf-1), inhibits the release of AIF from mitochondria, and prevents the loss of mitochondrial transmembrane potential. The AIF released from mitochondria binds to Hsp70. This interaction makes impossible the nuclear import of AIF \[[@B40]\].
Little is known about the function of chaperones in nondividing cells. Senescence is thought to play an important role in tumour suppression. Cellular senescence is more than just replicative senescence. It is a common program that is activated by normal cells in response to various types of stress. It has been named "stress-induced premature senescence" \[[@B23]\]. This type of senescence can act as a tumour suppressor in order to prevent damaged cells to multiply as well as a secondary outcome for cancer cells due to therapeutic treatments \[[@B18]\]. Tumour suppressors prevent cells from transforming into cancer by forming molecular barriers for genomic instability and infinite proliferation. They normally induce either apoptosis or a permanent cell cycle arrest-senescence. Unfortunately, it is not known at present which factors determine apoptosis or senescence \[[@B18]\].
Chaperones induction in cancer cells can lead to cancer progression and may be a major cause of chemotherapeutics resistance. It is one of the mechanisms protecting cancer cells against entering the apoptotic pathway. Hence, chaperone inhibition may be a promising tool to decrease cytoprotection and to initiate apoptosis of cancer cells \[[@B18]\]. Senescence and apoptosis normally counteract tumour development and cancer cells must therefore overcome these important tumour suppressor mechanisms to disrupt this barrier. Senescence emerges as an important tumour suppressor mechanism *in vivo*. Understanding and application of cellular senescence for cancer therapy has recently become a field of extensive research \[[@B18]\]. Gabai et al. have shown that knockdown of Hsp72 in certain cancer cells, but not in untransformed breast epithelial cells, triggers senescence via p53-dependent and p53-independent mechanisms. They demonstrate that the p53-dependent pathway controlled by Hsp72 depends on the oncogenic form of phosphatidylinositol 3-kinase (PI3K). Moreover, in cancer cell lines, activation of the p53 pathway caused by depletion of Hsp72 was dependent on oncogenes that activate the PI3K pathway. On the other hand, the p53-independent senescence pathway controlled by Hsp72 was associated with the Ras oncogene. In this pathway, extracellular signal-regulated kinases (ERKs) were critical for senescence, and Hsp72 controlled the ERK-activating kinase cascade at the level of Raf-1. Therefore, Hsp72 is intimately involved in suppression of at least two separate senescence signaling pathways that are regulated by distinct oncogenes in transformed cells, which explains why cancer cells become "addicted" to this heat shock protein \[[@B41]\].
The expression of ORP150 in cultured human cells is essential for their survival under hypoxia \[[@B42]\]. Such data suggest that ORP150 contributes in the cellular response to environmental stress. Our experiments have shown that ORP150 expression was induced in fibroblasts in hypoxia conditions only, suggesting that this protein may play an important role in the cytoprotective response to environmental stress.
At present, the exact signaling mechanism underlying bortezomib-induced apoptosis and senescence is poorly understood. Our study has shown that bortezomib causes a time-dependent increase in senescence of normal fibroblasts, especially of these incubated in hypoxic conditions. Moreover, we demonstrated that oxygen regulated protein 150 (ORP150) expression was induced in fibroblasts in hypoxia conditions only, suggesting that this protein may play an important role in the cytoprotective response to environmental stress. Further investigation of the protective responses will elucidate the mechanism by which fibroblasts are resistant on proteasome inhibition-mediated apoptosis and are directed for senescence.
Conflict of Interests
=====================
The authors declare that there is no conflict of interests regarding the publication of this paper.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
{#fig5}
{#fig6}
[^1]: Academic Editor: Michael Kalafatis
| {
"pile_set_name": "PubMed Central"
} |
The melting temperatures of lower mantle minerals are fundamental properties for understanding the evolution of the Earth from an early magma-ocean state to now, and for understanding the likelihood of melting in the current deep Earth (i.e., in Ultra Low Velocity Zones, ULVZs). Orthorhombic MgSiO~3~ perovskite (now bridgmanite[@b1]) is the end-member of the most voluminous phase in the lower mantle and so its melting temperature is a key cornerstone to any comprehensive understanding of lower mantle melting. Despite this, considerable doubt remains as to its high-pressure melting temperature.
At low pressures (\~25 GPa) the melting temperature of bridgmanite is reasonably well constrained from multi-anvil experiments to be between about 2800 and 2900 K (Ito and Katsura)[@b2]. This agrees well with more recent experiments of Liebske and Frost[@b3]. At higher pressure, however, there is little agreement between different studies. First of all, there are currently no static experimental measurements on the high-pressure melting T of perovskite above 100 GPa. The early diamond anvil cell experiments of Heinz and Jeanloz[@b4] and Knittle and Jeanloz[@b5] produced extremely flat melting curves and, therefore, very low high-pressure melting temperatures (\<4000 K at 100 GPa). These low melting temperatures have not been reproduced by other experiments or by theoretical estimates, and are likely to underestimate the high pressure melting temperature of bridgmanite. Other than these, the highest pressure static experiments only reach about 60 GPa. The higher pressure melting curve is constrained by a single shock-wave measurement of Akins *et al*.[@b6], which produced a melting temperature of 5500 K at 110 GPa.
The lack of high-pressure data has driven many attempts to estimate the melting temperature of bridgmanite by extrapolating the lower pressure data with simple melting equations, or by using theoretical methods such as molecular dynamics. However, these produce widely varying melting curves. For instance, simple extrapolations of the low-pressure data with various different melting equations predict melting temperatures ranging from 6800 K to \~8000 K at 120 GPa (Zerr and Boehler[@b7]). Lower melting temperatures of around 6000 K at 120 GPa were found using molecular dynamics simulations (Belonoshko[@b8]) and a temperature near 5200 K was found by Stixrude and Karki[@b9]. In this study they used the low-pressure melting temperature of Ito and Katsura[@b2] to anchor the melting curve and then used the Clapeyron slope from enthalpies and volumes from *ab initio* molecular dynamics (AIMD) simulations to extrapolate the melting curve to higher pressures. An even lower melting temperature of 5000 K at 120 GPa was suggested by Mosenfelder *et al*.[@b10] using equations of state constrained from shock-experiments.
It is fair to say, therefore, that current estimates for the melting temperature of the MgSiO~3~ end-member of bridgmanite at 120 GPa range from somewhere between 5000 K and 8000 K.
In this study we estimate the melting temperature of bridgmanite at 25 GPa and 120 GPa using the so-called two-phase model (2P phase) and *ab initio* molecular dynamics. This method avoids overheating issues typical of melting a fully solid system and has been used in a number of studies to produce melting temperatures for a range of materials[@b11][@b12][@b13][@b14][@b15][@b16][@b17][@b18][@b19][@b20][@b21][@b22][@b23][@b24][@b25][@b26][@b27]. As described in Methods section, a system of coexisting solid and liquid is set up and allowed to evolve through time. If the system melts, we assume it is above its melting temperature, and if it solidifies then it is below the melting temperature. If solid and liquid coexist, we assume it is on the melting temperature. This is an elegant and simple method used many times in the past (Fe, Al[@b28][@b29][@b30][@b31][@b32][@b33][@b34][@b35][@b36][@b37][@b38][@b39]). However, as discussed below, diffusion is relatively slow in silicate melts even near the melting temperature, and the system takes a long time to melt and appears to coexist at hundreds of degrees above its melting temperature. This is particularly a problem when using *ab initio* forces since the calculations are restricted to 100 ps or so, and we would predict melting temperatures that are too high. We show, however, that we can use the results from much cheaper empirical potential simulations to extrapolate the *ab initio* simulations to much longer timescales and produce very accurate melting temperatures.
Results
=======
In order to account for the very long times it can take the two-phase system to melt, while at the same time keeping the accuracy of the *ab initio* forces and energies, we characterised the melting kinetics (time to melting) using a classical empirical potential model and extremely long simulation times. To do this we used the Matsui pair-potential[@b40]. This potential model was designed to describe the MgSiO~3~ perovskite solid crystal over a small range of pressures and temperatures, and produces melting temperatures that are somewhat higher than via first principles (FP) methods. However, as we show later, diffusion rates and coordination numbers are very similar to those obtained *ab initio*, and so we use the classical potential to correct the *ab initio* simulations to infinite time.
To minimise any differences between the classical and DFT results, simulations were run in an NVT ensemble at the *ab initio* volume of the liquid at the target pressure (found via a set of fully liquid NPT simulations at 25 GPa and 120 GPa), using the coexistence approach for the model and recording the time it takes the system to melt as a function of temperature. The detailed procedure can be found in the [Supplementary Information](#S1){ref-type="supplementary-material"}.
[Figure 1](#f1){ref-type="fig"} shows the time, *τ*, it took the two-phase system to melt at different temperatures using the classical potential model. The results are shown for two pressures (25 GPa and 120 GPa). At very high temperatures the two-phase system melts very quickly, while *τ* increases dramatically as the temperature of the system approaches the solid-liquid transition point. The error bars are the uncertainties obtained from 10 trials per temperature, and show that the time to melting for the same temperature can vary quite significantly for different starting conditions (i.e., initial atomic velocities). This is particularly pronounced near the melting temperature.
The T-*τ* profile cannot be fit via a single activation energy or rate constant, and suggests that the perovskite melting process is dominated by different kinetics at than at temperatures closer to the melting temperature. An explanation for this behaviour comes from an interpretation of the melting process of solids by Samanta *et al*.[@b41]. Metadynamics calculations showed that the melting of a solid proceeds through multiple barrier-crossing events and competitive pathways when the temperature is relatively close to the melting point. On the other hand, when the system is heated to temperatures much higher than T~*M*~, the solid melts via a single step process with a small activation energy. We have, therefore, adopted a similar multi-barrier formulation to interpolate (*R*^2^ = 0.99) our melting time T-*τ* curve with a '2-rate exponential decay' model:where *T*~*M*~ is the melting temperature at the desired pressure, *κ*~*s*~ and *κ*~*f*~ (in ps^−1^) represent the kinetic constants for the slow and fast processes respectively and the pre-exponential factors *A* and *B* (in Kelvin) define the percentage of the fast process that spans between the fastest point to the melting temperature *T*~*M*~ as follows: .
Using this kinetic model, we tentatively suggest three different behaviours of the two-phase system of MgSiO~3~ around the melting point: 1) close to the melting point the slow kinetics is dominated by very low diffusion coefficients associated to Mg, Si and O in the melt; 2) at higher temperatures faster melting rates are associated to a very low transition barrier due to the solid becoming vibrationally unstable; 3) for temperatures in between the kinetics derives from a balance between the stability of the solid and the diffusivity of the liquid phase. Regardless of the exact mechanisms, this provides a convenient way of parameterising T vs *τ*, and moreover, for obtaining the true melting temperature for a set of simulations well above the actual melting point.
By fitting the two-rate equation we find a melting point of T~*M*~ = 3016 ± 35 K (as standard error of the mean SEM) and T~*M*~ = 5463 ± 15 K at 25 and 120 GPa, respectively. The fit to this is shown in [Fig. 1](#f1){ref-type="fig"}. We also find that the kinetic constants are weakly dependent on the pressure and that the fast rate is one order of magnitude smaller than the slower one. Moreover, it has to be noticed that the statistical value of %~*fast*~ calculated at both pressures (see caption [Fig. 1](#f1){ref-type="fig"}) showed that the fast-kinetics dominated the two-phase to liquid phase transition. As a consequence, the most important points for the fitting were those with *τ* calculated at higher temperatures. When applied to the AIMD data, this could be especially convenient in terms of computational cost, since the cpu time for *ab initio* calculations dramatically increases where the slow kinetics dominates.
[Figure 2](#f2){ref-type="fig"} shows the AIMD results on the same set of two-phase models used with the empirical potentials above. Given the vastly increased computational cost, only the higher temperatures melted within a manageable simulation time (a few 10 s of ps). Adopting the same kinetic model as above, and using the same kinetic constants and prefactor B, this strategy allows us to fit the results with one adjustable parameter, the prefactor A (given in caption of [Fig. 2](#f2){ref-type="fig"}). This results in a melting temperature of T~*M*~ = 2848 K at 25 GPa. At higher pressure the predicted melting point was identified as T~*M*~ = 5054 K, with a standard error below 100 K. To further demonstrate the need for this approach, in [Fig. 2](#f2){ref-type="fig"} we have also included the fitting curves obtained without the constraints from the CMD simulations. At P = 25 GPa the 2-phase decay model is preserved and a T~*M*~ = 3031 K is obtained which is similar to that when using the rate constants from the classical MD simulations. At P = 120 GPa, however, the model shows 1-phase decay and we obtain a melting temperature of T~*M*~ = 5775 K, \~700 K higher than when using the rate constants from the CMD simulations. Moreover, in both cases the standard error of the melting temperatures is very large. In other words, we would need T vs *τ* results to much lower temperature and much longer simulations to use the *ab initio* calculations alone.
Discussion
==========
The melting temperatures obtained at 25 GPa and 120 GPa are plotted in [Fig. 3](#f3){ref-type="fig"} together with available experimental data and some other theoretical results. As mentioned above, the classical model tends to overestimate the melting temperature, both with respect to the *ab initio* results and to the low-pressure experimental data of Ito *et al*.[@b2]. In contrast, the *ab initio* melting temperature at 25 GPa (2848 K) agrees almost perfectly with that obtained experimentally in the multi-anvil cell[@b2]. At higher pressures we predict a melting temperature 500 K below that obtained from shock-experiments[@b6]. Our results, also, predict a similar melting temperature (and well within the published uncertainty) to that[@b9] extrapolated with a Clapeyron slope based on enthalpies and melting volumes obtained from *ab initio* simulations but fixing the low-pressure results to the value found by Ito *et al*.[@b2].
Since we used three kinetic parameters for T vs *τ* obtained from the empirical potentials to obtain the melting temperature of the *ab initio* results, it is worth comparing some properties of the liquid from the two types of simulations. We concentrate on the liquid properties since the Matsui pair-potential[@b40] was parameterised for the solid and has already been shown to be a reasonable model for MgSiO~3~ bridgmanite.
Firstly, we suggest above that the kinetics of melting near the melting temperature may be dominated by atomic self diffusivity in the liquid. Diffusion coefficients for the atom types present in the system (Mg, Si and O) were calculated from the mean square displacement (MSD), as described in Allen and Tildesley[@b42]. The *ab initio* and classical potential results are compared to each other in [Fig. 4](#f4){ref-type="fig"}. As no experimental data are found in the literature, we also compare our classical and *ab initio* values with previous theoretical results. We find a good agreement between our AIMD and CMD data, and with previous classical (see Spera *et al*.[@b43]), *ab*-*initio*[@b44] and NPT Car-Parrinello DFT calculations[@b45]. We did not observe any dramatic changes in diffusivity between the two approaches either at 25 GPa or at 120 GPa. At higher pressure ([Fig. 4](#f4){ref-type="fig"}, lower panel), our calculations confirmed that the changes of volume and density could be divided into two regions: T \< 5000 K, where a slow diffusion (around 10^−9^ m^2^ s^−1^) denoted a very strong amorphisation of the molten bridgmanite reaching very low diffusivity at T = 3000 K (liquid to glass/solid transition); and T ≥ 5000 K, where the diffusivity characterized a fully liquid system.
Secondly, we compare the first-neighbour coordinations of oxygen around silicon atoms between AIMD and the CMD at the same volume. The AIMD results are in agreement with previous works[@b9], showing that the coordination number CN~*Si*--*O*~ in the melt varies from 5 at 25 GPa to 6 at 120 GPa with almost no effect due to the temperature. A further increase up to 7 was observed at pressures as high as 500 GPa[@b46]. Our analysis of the CMD liquid system revealed the same trend found at *ab initio* level, where the CN~*Si*--*O*~ was smoothly increasing with compression, but showed a slightly higher averaged value of CN~*Si*--*O*~ \~ 6.4 at 120 GPa.
Finally, we consider whether the results are sensitive to the particular pair-potential used to constrain the rate parameters. If we use Oganov[@b47] instead of Matsui[@b40] as the classical potential, we find a very good agreement between the predicted AIMD melting temperatures at P = 25 GPa (T~*M*~ (Matsui's fitting) = 2848 ± 69 K and T~*M*~ (Oganov's fitting) = 2872 ± 59 K), despite the fact that the melting temperatures of these two potential models themselves are quite different. At P = 120 GPa the results also agree well and the melting temperatures agree within error (T~*M*~ (Matsui's fitting) = 5054 ± 45 K and T~*M*~ (Oganov's fitting) = 4878 ± 190 K). The whole set of data can be found in the [Supplementary Information](#S1){ref-type="supplementary-material"}.
In conclusion we studied the melting process of MgSiO~3~ perovskite at pressures typical of the lower mantle by means of empirical potential and first principles molecular dynamics using a 2-phase (solid-liquid) model. Due to the huge *ab initio* computational cost, we provide a way to predict *T*~*M*~ (*AIMD*) using less expensive classical molecular dynamics simulations.
This was done by fitting a T-*τ* curve calculated from the CMD simulations with a '2-phase exponential decay' kinetic model. We then used a similar model on the AIMD melting results obtained far above the melting temperature to obtain the true *ab initio* melting temperature. The low pressure results agree extremely well with the most accurate experimental results. Our predicted melting temperatures are *T*~*M*~ \~ 2850 *K* at 25 GPa and *T*~*M*~ \~ 5050 *K* at 120 GPa.
We believe that the approach presented in this work is of general applicability. It does not depend on the material under study and shows low sensitivity to the potential in use to model the system. Therefore, it may well represent a powerful and computational affordable approach for studying the melting of a wide range of solids under critical conditions, where the classical potentials usually fail.
Methods
=======
We use both *ab initio* molecular dynamics and classical potential molecular dynamics on a system of 900 atoms (3 × 3 × 5 supercell), starting with a 3:2 ratio between liquid and solid. The two-phase system is prepared from well-equilibrated systems of fully solid and liquid at the required pressure and temperature.
For the AIMD simulations we use the DFT code VASP[@b48][@b49] in the Local Density Approximation (LDA) for the exchange-correlation energy[@b50]. Due to a fairly large dimension of the chosen supercell, the electronic energy is calculated at the Γ point of the Brillouin zone. To reduce the calculation time, we use an energy cutoff of 400 eV and for the finite basis set consequently apply a Pulay correction to the target pressure. In addition a further 2 GPa are subtracted to the pressure for the error associated to the exchange-correlation functional in use[@b51][@b52].
For the classical MD we performed the simulations by means of the dl_poly[@b53] code, with the MAM0K empirical potential[@b40] to approximate the potential energy via Coulombic, van der Waals and Gilbert-type repulsion energy parameters. We also considered a different pair-potential (Oganov *et al*.[@b47]) to measure the robustness of our computational approach.
In both cases, the simulations use a timestep of 1 fs and a period for the thermostat and barostat of 1 ps with a PMASS of 10^−3^--10^−4^. The canonical ensemble (NVT) is propagated in time using a Nosé-Hoover thermostat[@b54], while the CMD pressure (NPT ensemble) is constrained via a Melchionna modification of the general Nosé-Hoover algorithm[@b55]. Isothermal-isobaric simulations are performed at FP level using a Parrinello-Rahman scheme with implementation of the Nosé-Poincare approach for isothermal sampling[@b56].
The 2P system is prepared from well-equilibrated systems of fully solid (FS) and liquid (FL) at the required pressure and temperature. The FL system was melted at very high temperature and then gradually cooled until the desired T and P. This procedure was followed for both classical and *ab initio* MD runs.
Additional Information
======================
**How to cite this article**: Di Paola, C. and Brodholt, J. P. Modeling the melting of multicomponent systems: the case of MgSiO~3~ perovskite under lower mantle conditions. *Sci. Rep.* **6**, 29830; doi: 10.1038/srep29830 (2016).
Supplementary Material {#S1}
======================
###### Supplementary Information
The authors acknowledge NERC for grant NE/I010734/1. This work used the ARCHER UK National Supercomputing Service (<http://www.archer.ac.uk>), together with local HPC resources provided by UCL (Iridis and Legion clusters).
**Author Contributions** C.D.P. designed the theoretical model and performed the calculations. C.D.P and J.P.B. interpreted the results and prepared the manuscript. J.P.B. supervised the project.
{#f1}
{ref-type="fig"} where N = 6 and N = 7. Upper panel: P = 25 GPa, T~*M*~ = 2848 ± 69 K, A = 1809 K and fixed B = 515 K, *κ*~*f*~ = 0.2783 *ps*^−1^ and *κ*~*s*~ = 0.01182 *ps*^−1^, %~*f*~ = 77.84% and R^2^ = 0.978. Unconstrained fitting (violet) curve also included with T~*M*~ = 3031 ± 768 K, A = 2061 K, B = 642 K, *κ*~*f*~ = 0.6564 *ps*^−1^ and *κ*~*s*~ = 0.04536 *ps*^−1^, %~*f*~ = 76.24% and R^2^ = 0.998. Lower panel: P = 120 GPa, T~*M*~ = 5054 ± 45 K, A = 1381 K, fixed B = 709 K, *κ*~*f*~ = 0.1914 *ps*^−1^ and *κ*~*s*~ = 0.01030 *ps*^−1^, %~*f*~ = 66.07% and R^2^ = 0.979. Unconstrained fitting curve for P = 120 GPa has the following form: *T* = *T*~*M*~ + *Ce*^−*κτ*^ and parameters: T~*M*~ = 5775 ± 3327 K, C = 1431 K, *κ* = 0.2272 *ps*^−1^ and R^2^ = 0.987. The red dot-dash lines, defining the area included inside the 95% of statistical confidence (±2*σ*), are only shown for constrained fittings due to the very wide uncertainty found in the unconstrained interpolations.](srep29830-f2){#f2}
![Comparison between present AIMD (full circle) for Matsui's (red symbols) and Oganov's (orange symbols) fittings at pressures of 25 GPa and 120 GPa and previous experimental and theoretical results.\
Present work error bars are smaller than the symbol itself, apart for AIMD point (Oganov's fitting) at P = 120 GPa. We have also included the value (full magenta diamond symbol) and the error bar (magenta vertical line) around the melting temperature at 120 GPa found by Stixrude *et al*.[@b9] for a better comparison with our melting temperature predicted at the same pressure.](srep29830-f3){#f3}
![Self diffusion coefficients (DC) of magnesium (full circle), silicon (full square) and oxygen (full triangle up) calculated over mean square displacement at CMD (green) and AIMD (magenta) level.\
Upper panel: comparison of atomic DC present work at 25 GPa with first-principles simulations[@b44] (black symbols) at 25 GPa (3000 K) and 24 GPa (6000 K) and with classical molecular dynamics data (red symbols) at 26.9 GPa[@b57] using BKS force field. Lower panel: comparison of atomic DC present work with Car-Parrinello NPT-MD calculations (black full triangle down symbols) at 120 GPa[@b45].](srep29830-f4){#f4}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
It has long been known that cancer cells hijack cellular programs that regulate survival, growth and proliferation, leading to tumor formation and progression. The best-known causes of malignant transformation are the genetic and epigenetic changes that induce stem-cell-like properties, such as unlimited cell division and blocked differentiation. Traditionally, the proposed role of the cellular metabolism of cancer cells was to primarily support and sustain malignant growth. However, it is clear today that cellular metabolism actively regulates the malignant phenotype. For example, loss of the p53 tumor suppressor may contribute to malignant transformation independently of its well-described functions in cell cycle regulation, DNA repair and senescence (see [Box 1](#DMM034272B1){ref-type="boxed-text"} for a glossary of terms). Instead, through the induction of glycolysis and anabolic pathways ([Box 1](#DMM034272B1){ref-type="boxed-text"}), p53 dysfunction leads to an early-onset metabolic malignant transformation ([@DMM034272C63]). Another example of a key role of a mutation-driven metabolic reprogramming leading to malignant transformation are oncometabolites. For example, a consequence of loss-of-function mutations in succinate dehydrogenase (SDH; [Box 1](#DMM034272B1){ref-type="boxed-text"}) is that cancer cells can massively accumulate succinate, an intermediate metabolite in the tricarboxylic acid (TCA) cycle. Intriguingly, succinate, now in the role of an oncometabolite, can induce epigenetic alterations through the inhibition of α-ketoglutarate-dependent dioxygenases ([@DMM034272C117]), ultimately leading to a malignant phenotype ([@DMM034272C115]). Some features of metabolic reprogramming in cancer have, however, been puzzling, such as the Warburg effect, in which tumor cells increase glucose consumption and lactate excretion ([@DMM034272C111]). This phenotype is energetically inefficient compared to mitochondrial respiration, and could theoretically limit tumor growth in glucose-depleted tissues. However, anaerobic glycolysis can be beneficial if the malignant cell requires a high metabolic flux to synthesize building blocks such as nucleotides. Moreover, this phenotype can induce a unique metabolic milieu with low glucose and high lactate ([@DMM034272C22]; [@DMM034272C53]; [@DMM034272C100]). Intriguingly, evidence from murine *in vitro* and *in vivo* models suggests that glucose deprivation and lactate accumulation in the tumor microenvironment can have detrimental effects on the immune cells that were poised to infiltrate and destroy tumors ([@DMM034272C20]; [@DMM034272C22]).
Box 1. Glossary**^13^C-labeling:** method to interrogate intracellular metabolic pathways. Detection of labeled metabolites is performed using nuclear magnetic resonance spectroscopy.**Anabolic pathways:** synthesis of macro-molecules out of smaller biochemical components.**CD4+ T cells:** T cells expressing CD4. Often referred to as 'helper' T cells; can differentiate to inflammatory ('effector') and anti-inflammatory ('regulatory') subtypes.**CD4+ Treg cells:** CD4+ T cells with regulatory properties. Usually described by high CD25 and FOXP3 expression. Critical for maintenance of self-tolerance.**CD8+ T cells:** T cells expressing CD8. Often referred to as 'cytotoxic' T cells; capable of direct engagement with infected cells or tumor cells.**Chimeric antigen receptor (CAR)-transduced T cells:** engineered effector T cells, recognizing specific antigens expressed by tumor cells, such as CD22 in B-cell leukemia.**Costimulatory receptors:** in addition to T-cell receptor (TCR) stimulation, ligation of costimulatory receptors such as CD28, CD137 and ICOS increases or modulates T-cell activation.**Germinal center:** area in lymphoid follicles where B cells become activated, proliferate intensively after antigen contact, switch immunoglobulin class and increase affinity for the antigen.**Granzyme-B and perforin:** cytolytic molecules stored in the granules of cytotoxic T cells and natural killer (NK) cells.**Immune checkpoint inhibitors:** monoclonal antibodies that block immune inhibitory pathways such as CTLA-4, PD-1 and PD-L1, and induce immune-cell activation.**Interferon-ɣ (IFN-ɣ):** inflammatory cytokine, mainly produced by T cells and NK cells, with anti-tumoral, anti-viral and immunostimulatory properties.**L-kynurenine:** product of L-tryptophan degradation through tryptophan dioxygenase and indoleamine 2,3-dioxygenase.**Lymphoid/lymphatic organs:** spleen, bone marrow, thymus, appendix, lymph nodes, lymph vessels and tonsils. Critical for formation, maturation, differentiation and activation of immune cells.**Myeloid-derived suppressor cells (MDSCs):** heterogeneous population of immature myeloid cells consisting of precursors for granulocytes, macrophages or dendritic cells. Associated with resolution of inflammation and tumor progression.**Pentose phosphate pathway (PPP):** series of metabolic steps leading to degradation of glucose to pentoses via the formation of NADPH and carbon dioxide.**Plasma cells:** differentiated B cells capable of antibody production and secretion.**Programmed death 1 (PD-1) receptor:** surface protein on activated T cells repressing an immune response. Activated through PD-1 ligands (PD-L1, PD-L2), which are expressed in various tissues, including tumors.**Retinoic acid receptor-related orphan receptor gamma (RORɣt):** ligand-dependent transcription factor expressed only in cells of the lymphoid compartment, typically in CD4+ T cells secreting IL-17 (Th17 cells).**Senescence:** age-related alterations in all stages of immune-cell development.**Succinate dehydrogenase (SDH):** also known as respiratory complex II; catalyzes the oxidation of succinate to fumarate with the reduction of ubiquinone to ubiquinol.**Toll-like receptor (TLR) ligands:** ligands to the pattern recognition receptors and activator of innate immune cells; e.g. microbial cell wall components (e.g. lipopolysaccharide) and viral molecules.**Tumor-draining lymph nodes:** closest lymph nodes to the tumor. Typically a primary site of tumor dissemination.
Cancers are highly diverse and, in addition to the genetic and functional heterogeneity of malignant cells, a broad spectrum of immune populations can be found in human tumor tissue. Among adaptive immune cells, the tumor-infiltrating T cells are the best documented. T cells are highly heterogeneous, and various phenotypic sub-populations \[CD4+ and CD8+ T cells ([Box 1](#DMM034272B1){ref-type="boxed-text"})\] and functional (effector, memory) and differentiation \[CD4+ Th1, CD4+ Treg ([Box 1](#DMM034272B1){ref-type="boxed-text"})\] states have been identified. T cells can affect tumor growth either through direct engagement or through stimulation of other cells in the tumor microenvironment. This feature has been exploited in clinical approaches that aim to increase their anti-tumor potential, such as through blockade of the T-cell-inhibitory PD-1 receptor ([Box 1](#DMM034272B1){ref-type="boxed-text"}), or through employment of *ex vivo* engineered chimeric antigen receptor (CAR)-transduced T cells ([Box 1](#DMM034272B1){ref-type="boxed-text"}). The tumor infiltration with B cells is less well documented, but both their pro- and anti-tumorigenic functions ([@DMM034272C106]) are intriguing and require extensive elucidation in future studies.
The interaction of adaptive immune cells with cells of innate immunity is critical for an effective and well-regulated response, and innate immune cells are often found in tumors. Indeed, the first immune cells to be described in human tumors were macrophages ([@DMM034272C62]). Outside of the context of cancer, these innate immune cells are responsible for fast clearance of pathogen-infected cells during infections. Upon stimulation with interferon-γ (IFN-γ) and toll-like receptor (TLR) ligands ([Box 1](#DMM034272B1){ref-type="boxed-text"}), macrophages polarize to a pro-inflammatory (M1) phenotype, with an anti-tumor potential ([@DMM034272C119]). Additionally, macrophages can also polarize toward an anti-inflammatory phenotype with pro-tumoral characteristics through alternative activation (M2) when stimulated with IL-4 and IL-10. M1 and M2 macrophages participate in inflammatory responses and modulate tissue homeostasis and repair through their distinct functional specialties ([@DMM034272C71]). Hence, macrophages must be highly plastic to adapt their functions in response to infection and tissue damage. Emerging evidence reveals that macrophages engage distinct metabolic demands during M1 and M2 activation. For example, M1 macrophages enhance their anabolic metabolism, including anaerobic glycolysis, pentose phosphate pathway ([Box 1](#DMM034272B1){ref-type="boxed-text"}) activation and fatty acid synthesis. In contrast, M2 macrophages prefer catabolic metabolism and heavily utilize oxidative phosphorylation (OXPHOS) to support their metabolic demands ([@DMM034272C48]). These changes provide metabolic checkpoints to fine-tune macrophage behavior and contribute to their altered functions in diseases, especially in the tumor microenvironment. Also part of the innate immunity, natural killer (NK) cells are critical for direct engagement and killing of cells identified as non-self. Accordingly, they have the potential to destroy cancer cells ([@DMM034272C70]). Compared to T cells and macrophages, NK cell metabolism is less well documented. Accordingly, if and how the tumor microenvironment affects NK cell metabolism is mostly unknown. We hypothesize, however, that several of the key mechanisms of metabolic immune cell suppression in tumors are shared between the various immune cell types that infiltrate tumors.
Tumors modulate local concentrations of nutrients that are critical for immune cell function {#s2}
============================================================================================
Glucose metabolism {#s2a}
------------------
To perform effector functions, including killing cancer cells and excreting cytokines, effector immune cells, such as activated cytotoxic T cells, undergo extensive metabolic reprogramming. Aerobic glycolysis, a process where glucose is metabolized to pyruvate and ultimately to lactate in a series of enzymatic steps that yield ATP and substrates for other metabolic pathways, was first described in malignant cells. Interestingly, non-malignant proliferating cells perform aerobic glycolysis, which is also considered necessary for optimal T-cell function ([@DMM034272C20]). However, T cells are metabolically flexible and aerobic glycolysis may not be critical for their activation and survival ([@DMM034272C21]; [@DMM034272C92]). In contrast, glycolysis is essential for T-cell proliferation ([@DMM034272C92]). Under normoglycemic conditions, T cells upregulate the key glucose transporter Glut1, followed by increased glucose uptake and glycolysis upon antigenic stimulation ([@DMM034272C37]; [@DMM034272C67]). In parallel, pyruvate, one of the terminal products of glycolysis, is mainly reduced to lactate, rather than being oxidized in mitochondrial respiration (see poster). However, T cells may not be able to sustain a permanent state of anaerobic glycolysis that is necessary for their effector function. During acute infections, the population of effector T (Teff) cells performing aerobic glycolysis contracts and memory T cells, which are less dependent on glycolysis and rather engage in mitochondrial respiration, arise ([@DMM034272C14]; [@DMM034272C68]).
In contrast, cancer cells are able to maintain and eventually increase high glucose uptake and glycolysis, leading to a decrease of intratumoral glucose levels ([@DMM034272C16]; [@DMM034272C110]; [@DMM034272C8]). In turn, glucose deprivation can directly impede production of IFN-γ, a key T-cell effector molecule in tumor-infiltrating CD8+ T cells ([@DMM034272C22]). It has been proposed that these effects are mediated through glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a key glycolytic enzyme that can also affect post-transcriptional modification of mRNA. When T cells are able to perform high rates of glycolysis, GAPDH is committed to its metabolic role. Under low glycolytic flux, however, GAPDH prevents translation of IFN-γ ([@DMM034272C21]). It has also been documented that glucose deprivation suppresses T cell receptor (TCR)-dependent activation of Ca^2+^ and nuclear factor of activated T cells (NFAT) signaling, leading to T-cell hypo-responsiveness. This effect is mediated by the absence of the glycolytic product phosphoenolpyruvate, which sustains Ca^2+^ and NFAT by blocking sarco/endoplasmic reticulum Ca^2+^-ATPase ([@DMM034272C49]). In line with this, our group observed a negative correlation between accelerated tumoral glucose metabolism and T-cell infiltration in renal cell carcinoma ([@DMM034272C99]), with similar observations made in oral squamous cell carcinoma ([@DMM034272C85]). Interestingly, emerging clinical data point towards tumor glucose metabolism as a mechanism of resistance to T-cell-mediated tumor rejection. As recently shown by Cascone et al., overexpression of glycolysis-related genes in cancer cells impairs the anti-tumor activity of T cells and, inversely, the inhibition of cancer glycolysis enhanced T-cell-mediated tumor rejection ([@DMM034272C19]).
CD4+ regulatory T (Treg) cells can suppress inflammation and are often associated with tumor progression. Tumor-infiltrating Treg cells may inhibit local anti-tumor immunity. Interestingly, as described in murine systems, Treg cells express low levels of Glut1, do not rely on glucose uptake and glycolysis, and, similarly to CD8+ memory T cells, perform OXPHOS and lipid oxidation ([@DMM034272C74]). Forkhead box protein P3 (FOXP3), the lineage-defining transcription factor of murine Treg cells, was proposed to be a key regulator of this phenotype ([@DMM034272C42]). Mechanistically, FOXP3 induces the expression of genes involved in lipid and peptide hormone metabolism. Additionally, it downregulates genes involved in glucose uptake and glycolysis. Importantly, forced expression of FOXP3 inhibits the PI3K-Akt-mTORC1 signaling pathway, which is involved in the induction of the glycolytic machinery ([@DMM034272C42]). Paradoxically, the abundance of glucose may be important for Treg induction, as glycolysis in conventional CD4+ T cells is essential for the initiation of the regulatory phenotype in Treg cells through the translocation of the glycolytic enzyme enolase-1 to the nucleus, where it binds to *FOXP3* regulatory regions, such as its promoter and its CNS2 (conserved noncoding sequence 2) ([@DMM034272C28]). These studies suggest that glucose is necessary for the emergence of Treg cells, e.g. in a lymphatic organ ([Box 1](#DMM034272B1){ref-type="boxed-text"}), and, upon a subsequent tumor infiltration, glucose may favor Treg survival and function. However, future studies are required to address this hypothesis in cancer patients.
Similar to T cells, B cells are highly metabolically active. During B-cell development, each stage has a different dependency on glucose metabolism and pre-B-cells are less glucose-dependent than immature B cells ([@DMM034272C59]). After stimulation, naïve B cells proliferate and increase glucose uptake and lactate production, similarly to what occurs upon T-cell activation ([@DMM034272C39]). In line, extensive B-cell metabolic reprogramming is required for antibody production ([@DMM034272C17]). Recently, Jellusova et al. described the requirement of high glycolytic activity of germinal center ([Box 1](#DMM034272B1){ref-type="boxed-text"}) B cells to support their growth and proliferation in a hypoxic microenvironment. In addition to glycolysis, an increase of mitochondrial content has been observed in germinal center B cells ([@DMM034272C54]). We speculate that glucose deprivation in combination with hypoxia in the tumor microenvironment might favor certain B-cell phenotypes, creating an immunosuppressive milieu.
Resembling the features of Teff cell metabolism, NK cells increase aerobic glycolysis upon activation ([@DMM034272C40]). With high IL-15 stimulation, NK cells elevate the activity of mammalian target of rapamycin (mTOR) to boost bioenergetic metabolism, increase glucose uptake, and upregulate the expression of transferrin receptor CD71 and amino acid transporter CD98 ([@DMM034272C69]). Accordingly, impairment of glucose metabolism and disruption of mTOR signaling leads to a diminished cytotoxic activity in NK cells ([@DMM034272C29]). Assmann et al. recently showed that sterol regulatory element-binding protein (Srebp) transcription factors play an essential role in the cytokine-induced metabolic reprogramming of NK cells. Srebp was required for the increase in both glycolysis and OXPHOS. Moreover, Srebp inhibition prevented this phenotype and decreased NK cell cytotoxicity ([@DMM034272C7]). It remains unclear, however, whether metabolic alterations found in tumors may affect the metabolic activity and the Srebp-mediated NK cell function.
Amino acid metabolism {#s2b}
---------------------
In addition to glucose, glutamine has been described as a crucial nutrient for the effector function of T cells. Glutamine is the most abundant amino acid in circulation and its uptake is critical for various T-cell metabolic processes, including the TCA cycle, nucleotide synthesis and detoxification of reactive oxygen species (ROS) ([@DMM034272C55]). As shown by Nakaya et al., T-cell glutamine uptake depends on the neutral amino acid transporter type 2 (ASCT2). ASCT2 deficiency blocks the induction of T helper 1 (Th1) and Th17 cells ([@DMM034272C80]). In line with this, it has been reported that glutamine deprivation supports differentiation into Tregs, despite *in vitro* conditions favoring a Th1 differentiation. Moreover, addition of α-ketoglutarate reversed this effect and rescued Th1 differentiation under glutamine deprivation through the induction of Tbet, a Teff cell transcription factor, which correlated with increased mTORC1 signaling ([@DMM034272C58]). As reported by Lee et al., 6-diazo-5-oxo-L-norleucine, a naturally occurring glutamine antagonist, inhibited glutamine metabolism in activated T cells and was able to inhibit immune-mediated rejection of allografts in fully mismatched skin and heart allograft transplantation models ([@DMM034272C61]). Similarly, glutamine is essential for B-cell proliferation and differentiation into plasma cells ([Box 1](#DMM034272B1){ref-type="boxed-text"}) ([@DMM034272C24])*.* Not much is known about glutamine concentrations in tumors, but many cancer types harbor mutated Myc, leading to high glutamine uptake ([@DMM034272C38]). Myc transcriptionally induces mitochondrial glutaminolysis and leads to glutamine addiction of cancer cells ([@DMM034272C114]). Thus, glutamine may be limited in the tumor environment and glutamine deprivation can play a role in tumor-induced immunosuppression.
Upon activation, T cells heavily consume arginine and tryptophan ([@DMM034272C98]). L-arginine enhances the generation of central memory-like T cells by inducing a metabolic switch from glycolysis to OXPHOS, with enhanced anti-tumor activity in an OVA-antigen-expressing B16 melanoma mouse model ([@DMM034272C41]). The authors suggest that L-arginine has a direct effect on specific T-cell nuclear proteins (BAZ1B, PSIP1 and TSN) by changing their structure, leading to increased pro-survival signaling and enhanced anti-tumor function in T cells ([@DMM034272C41]).
However, cancer cells often overexpress the amino-acid-catabolic enzyme indolamine-2,3-dioxygenase (IDO), which can lead to extracellular depletion of tryptophan (see poster, cancer cell). Constitutive expression of IDO depends on cyclooxygenase-2 and prostaglandin E2 via PKC and PI3K signaling ([@DMM034272C45]). Similar to glucose, deprivation of tryptophan can impair T-cell function. Specifically, tryptophan depletion activates general control nonderepressible 2 (GCN2), a stress-response kinase that is activated by elevations in uncharged transfer RNA (tRNA), leading to inhibition of T-cell function ([@DMM034272C78]), impaired Th17 differentiation and promotion of Treg development ([@DMM034272C104]). Similar to IDO, tryptophan 2,3-dioxygenase (TDO) is expressed by cancer cells in various human tumors ([@DMM034272C91]) and its activity, presumably through tryptophan depletion or kynurenine ([Box 1](#DMM034272B1){ref-type="boxed-text"}) production, induces immune dysfunction ([@DMM034272C95]). As shown by Pilotte et al., TDO inhibition through a novel synthetic inhibitor was able to restore the ability of mice to reject TDO-expressing tumors in a preclinical mastocytoma model ([@DMM034272C91]).
Additionally, degradation of arginine by tumors or myeloid-derived suppressor cells ([Box 1](#DMM034272B1){ref-type="boxed-text"}) through arginase-1 upregulation leads to reduced expression of the CD3ζ chain, cell cycle arrest and an impaired antigen-specific T-cell response ([@DMM034272C93]). In line with these observations, tumor cells overexpressing IDO were not rejected by tumor-specific T cells in a P815 mastocytoma murine model ([@DMM034272C108]), further confirming the immunoregulatory role of amino acid catabolism in cancer.
Oxygen {#s2c}
------
Tumors are often hypoxic, as malignant growth can exceed the capacity of healthy progenitor cells to form new blood vessels. In turn, hypoxia can function as a metabolic adjunct to further promote a malignant phenotype. Indeed, hypoxia can boost glucose uptake and glycolysis through induction of various glycolytic genes ([@DMM034272C57]; [@DMM034272C96]), and elevated glycolysis is associated with sustained malignant growth ([@DMM034272C53]). The effects of hypoxia on immune cell activation are not sufficiently explored. On the one hand, hypoxic conditions lead to less efficient TCR- and CD28-mediated T-cell activation ([@DMM034272C81]). Moreover, hypoxia-inducible factor (HIF)-1α-deficient CD4+ and CD8+ T cells from Lck-Cre/HIF-1-floxed mice show an increased ability to proliferate and to produce IFN-γ ([@DMM034272C65]). On the other hand, it has been shown that HIF-1α does not affect T-cell proliferation, but favors the differentiation of Th17 cells via direct transcriptional induction of the RAR-related orphan receptor gamma (RORγt; [Box 1](#DMM034272B1){ref-type="boxed-text"}) ([@DMM034272C27]). Interestingly, hypoxia-induced HIF-1α is able to increase the expression of costimulatory receptors ([Box 1](#DMM034272B1){ref-type="boxed-text"}), such as CD137, on tumor-infiltrating T cells ([@DMM034272C89]). Hypoxia may therefore be selectively required for effective immunotherapies that aim to stimulate the anti-tumor activity of T cells.
Oxygen is necessary for OXPHOS and the generation of ROS. Both processes are part of mitochondrial respiration and ROS are necessary for proper T-cell effector function and antigen-specific proliferation ([@DMM034272C97]). The effects of ROS on CD8+ T-cell function are mediated by lymphocyte expansion molecule (LEM), which regulates the expression of OXPHOS proteins (such as NADH ubiquinone oxidoreductase chain 1) and, accordingly, ROS production. LEM is necessary for cytotoxic T-cell expansion and memory T-cell development ([@DMM034272C83]). Therefore, basal ROS levels are required for proper T-cell signaling. Under hypoxia, ROS levels may be insufficient. Conversely, high ROS levels can be toxic and the ROS generated in the tumor microenvironment can impair immune cells by downregulation of the CD3ζ chain or impairment of Ca^2+^ mobilization ([@DMM034272C5]; [@DMM034272C60]). ROS induce oxidation of lipids such as 4-hydroxynonenal (4-HNE), inducing apoptosis as well as defects in NFκB signaling in T cells ([@DMM034272C64]). Additionally, 4-HNE activates X-box binding protein 1 in tumor-associated dendritic cells (DCs), leading to an accumulation of lipid bodies driving ovarian cancer progression via the suppression of anti-tumoral immune responses ([@DMM034272C26]). In line with this, high intracellular levels of ROS derived from mitochondria may impair anti-tumoral T-cell function ([@DMM034272C102]).
Macrophages are sensitive to changes in oxygen availability and it has been reported that the anti-inflammatory M2-like tumor-associated macrophages (TAMs) accumulate in hypoxic tumor regions, whereas the pro-inflammatory M1-like TAMs reside in normoxic regions. Mechanistically, intratumoral hypoxia-induced semaphorin 3A ([Box 1](#DMM034272B1){ref-type="boxed-text"}) attracts TAMs to hypoxic regions by triggering vascular endothelial growth factor (VEGF) receptor 1 phosphorylation ([@DMM034272C18]). In addition, hypoxic TAMs upregulate the expression of REDD1 (regulated in development and DNA damage responses 1), which inhibits mTOR. This leads to decreased glycolysis in TAMs and correlates with further hypoxia by abnormal blood vessel formation and promotion of metastases ([@DMM034272C112]). Moreover, hypoxia stabilizes HIF-1α in TAMs, leading to high production of chemokines and chemokine receptors such as C-X-C motif chemokine ligand 12 (CXCL12) and receptor 4 (CXCR4) ([@DMM034272C94]), as well as VEGF ([@DMM034272C36]). Hypoxic TAMs also secrete proteolytic enzymes, such as matrix metalloproteinases (MMP)-1 ([@DMM034272C79]) and -7 ([@DMM034272C15]). Production of metalloproteinases by TAMs is likely to affect the interaction of both endothelial and tumor cells with the extracellular matrix, contributing to cell proliferation and tumor dissemination.
Waste products of tumor metabolism affect immunity {#s3}
==================================================
Glucose metabolism and acidity {#s3a}
------------------------------
Although the concentration of essential nutrients may be lower in the tumor microenvironment compared to normal tissues, several waste products of tumor cell metabolism accumulate and can affect immune cell function. The most prominent metabolite in the tumor microenvironment is lactate. After the reduction of pyruvate to lactate, the monocarboxylate transporters (MCT)-1 and -4 co-transport lactate and protons out of the cell, leading to an accumulation of lactate and to a decreased pH in the extracellular space (see poster, cancer cell). Accordingly, intratumoral lactate can reach levels of up to 40 mM ([@DMM034272C11]), and high intratumoral lactate concentrations correlate with a more aggressive tumor biology and decreased patient survival in some cancers, such as in head-and-neck tumors and melanoma ([@DMM034272C13]; [@DMM034272C11]).
To date, several studies demonstrated strong effects of lactate and lactic acid on immune cell populations *in vitro* and *in vivo*. For example, lactate/lactic acid promoted IL-17 production by CD4+ Th17 cells ([@DMM034272C43]), while inhibiting proliferation and activation of cytotoxic CD8+ T cells ([@DMM034272C35]). This effect is induced by the prevention of TCR-triggered phosphorylation of JNK, c-Jun and p38, as well as the expression of NFAT ([@DMM034272C72]; [@DMM034272C11]). In line with this, our group showed that lactate dehydrogenase A (LDHA)-mediated production of lactate in tumor cells and subsequent acidification constrains IFN-γ production in tumor-infiltrating T cells, resulting in a loss of immune surveillance and promoting tumor growth in a mouse melanoma model ([@DMM034272C11]). Importantly, ^13^C-labeled ([Box 1](#DMM034272B1){ref-type="boxed-text"})-lactate uptake experiments demonstrated that protons are required for the effects of lactate/lactic acid on immune cells, since the addition of protons could increase lactate uptake into the immune cells ([@DMM034272C11]; [@DMM034272C35]). Innate immune cells are also sensitive to lactate. High amounts of lactate in the tumor microenvironment stimulate TAM polarization into the M2-like phenotype by stabilizing HIF-1α ([@DMM034272C23]). In addition, high levels of lactic acid in tumors downregulate NK cell activation, resulting in diminished IFN-γ production and tumor immune escape ([@DMM034272C11]).
Although the negative impact of lactate on immune cells is often in concert with a decreased pH in the tumor microenvironment, acidity has distinct effects on a variety of immune populations. Tumors can be highly acidic, and it has been extensively reported that low pH supports cancer growth and spreading ([@DMM034272C56]). Proton secretion from tumor cells to induce extracellular acidity can be carried out by several transporters, such as the Na^+^/H^+^ exchanger, the above-mentioned lactate/H^+^ co-transporting MCTs and the H^+^ ATPase. In addition, intratumoral hypoxia can induce carbonic anhydrase, which can form protons by catalyzing the hydration of CO~2~ ([@DMM034272C52]). In line with this, increased CO~2~ production through the pentose phosphate pathway in cancer cells has been linked to an acidification of the tumor microenvironment ([@DMM034272C56]). As broadly reviewed by [@DMM034272C51], low pH negatively impacts the effector functions of both innate and adaptive immune cells. This was first described by Fischer et al., who demonstrated that low extracellular pH (pHe: 5.8) leads to decreased cytokine production and to a loss of cytotoxic effector functions without affecting cell viability ([@DMM034272C33],[@DMM034272C34]; [@DMM034272C77]). The effect of low pH on cytokine production by T cells correlated with impaired signaling pathways involving STAT5, ERK, AKT, p38 and NFAT ([@DMM034272C51]). Importantly, buffering of low pH with bicarbonate therapy increased T-cell infiltration and impaired tumor growth. Furthermore, a combination of bicarbonate with anti-CTLA4 and anti-PD-1 ([Box 1](#DMM034272B1){ref-type="boxed-text"}) treatments improved antitumor responses in B16 melanoma and Panc02 pancreatic cancer mouse models and increased the survival of mice in a pmel-B16 model of adoptive T-cell therapy ([@DMM034272C90]). Despite these results, a specific acidity-sensing machinery in T cells and other immune cells still needs to be identified.
Amino acid metabolism {#s3b}
---------------------
In addition to tryptophan depletion, high activity of IDO leads to an accumulation of tryptophan catabolism byproducts, most prominently kynurenines. Similarly to lactate, kynurenines can suppress the proliferation and the effector function of CD8+ T cells through the aryl hydrocarbon receptor (AHR) ([@DMM034272C84]; [@DMM034272C32]). Interestingly, the interaction of kynurenines with AHR favors Treg induction in a TGFβ-dependent manner ([@DMM034272C73]). The combination of tryptophan starvation and accumulation of tryptophan catabolites downregulates TCRζ and induces a regulatory phenotype in naïve T cells.
Similarly, overexpression of glutaminase, observed in many cancers, might not only decrease glutamine levels, but could lead to high intratumoral glutamate levels. Indeed, Briggs et al. showed that breast cancer cells secrete glutamate, leading to paracrine induction of HIF-1α via inhibition of the glutamate/cystine antiporter Xc^−^ and HIF-PH2 inactivation ([@DMM034272C12]). In addition, macrophages and DCs, which are often found in tumors and tumor-draining lymph nodes ([Box 1](#DMM034272B1){ref-type="boxed-text"}), can release glutamate in concentrations up to 30 µM ([@DMM034272C87]). Even though the minimal effective concentration necessary to affect immune cells is unknown, T cells constitutively express the glutamate transporter mGlu5R. mGlu5R stimulates adenylate cyclase and prevents TCR-mediated T-cell activation and IL-6 production ([@DMM034272C86], [@DMM034272C87]), possibly through the activation of protein kinase A, which can inhibit several pathways, including ERK, JNK and NFκB signaling ([@DMM034272C88]). However, upon stimulation, T cells express mGlu1R, which signals through the MEK-ERK1/2 pathway, and this counteracts T-cell inhibition through mGlu5R ([@DMM034272C86]). mGlu1R stimulation by glutamate enhances the secretion of several cytokines, including IL-2 and IFN-γ, and increases proliferation ([@DMM034272C87]). A high concentration of extracellular glutamate can also affect other transporters, such as Xc^−^. Our group and others have shown that T cells express Xc^−^ and that T-cell stimulation leads to an increased uptake of cystine. Cystine uptake, followed by its intracellular degradation to cysteine and subsequent glutathione synthesis, is critical for the ROS detoxification machinery, and inhibition of cystine uptake impairs T-cell activation ([@DMM034272C101]). High levels of extracellular glutamate might therefore impair the export of glutamate and the import of cystine, possibly leading to ROS dysregulation and T-cell dysfunction (see poster, Increased metabolic products). Glutamate receptors have also been found on other immune cells, including B cells and DCs ([@DMM034272C88]), and future studies of the intratumoral glutamine/glutamate homeostasis may therefore uncover intriguing new mechanisms of tumor-induced immune dysregulation.
Nucleotide metabolism {#s3c}
---------------------
In addition to the direct effects of hypoxia on intratumoral immune cells, hypoxia induces increased adenine nucleotide breakdown through the 5′-nucleotidase pathway, leading to adenosine accumulation in tumors ([@DMM034272C10]). Specifically, ATP is rapidly degraded to adenosine by the ectonucleotidases CD39 ([@DMM034272C31]) and CD73 ([@DMM034272C105]) expressed on tumor cells, which convert ATP to AMP and AMP to adenosine, respectively. Adenosine can in turn inhibit the activation and cytotoxic capacity of T and NK cells ([@DMM034272C50]; [@DMM034272C44]). The accumulated extracellular adenosine then binds to A2AR and A2BR adenosine receptors expressed by T cells and NK cells, inducing intracellular cAMP accumulation and signaling ([@DMM034272C82]). Even at low levels, adenosine strongly inhibits both TCR-induced proliferation of T cells and IL-2 receptor expression ([@DMM034272C50]). Intriguingly, A2AR and A2BR blockade can enhance NK cell function by increasing granzyme-B ([Box 1](#DMM034272B1){ref-type="boxed-text"}) expression, and promotes the anti-metastatic effects of NK cells by secreting perforin ([Box 1](#DMM034272B1){ref-type="boxed-text"}) ([@DMM034272C9]; [@DMM034272C75]). These data suggest that the adenosine pathway also contributes to NK cell dysfunction in the tumor microenvironment.
Similar to its effects on T and NK cells, adenosine enhances activation of immunoregulatory M2 macrophages via A2AR and A2BR, inhibits TNFα, IL-6 and IL-12 release, and augments IL-10 as well as VEGF production ([@DMM034272C25]). Lastly, many tumors show a deficiency in S-methyl-5′-thioadenosine phosphorylase (MTAP). MTAP is responsible for the breakdown of S-methyl-5′-thioadenosine (MTA) and, because of MTAP deficiency, MTA levels can increase in the tumor microenvironment. MTA inhibits antigen-specific T-cell proliferation, activation and cytokine production by interfering with asymmetric protein methylation events upon T-cell stimulation and through decreased Akt phosphorylation ([@DMM034272C46]). Subsequently, MTA-treated T cells do not upregulate the expression of molecules such as CD25 and CD69, and maintain a naïve phenotype. Functionally, highly activated cytotoxic T cells are still not able to lyse target cells and to produce IFN-γ, resulting in a loss of their anti-tumoral capacity. Thus, MTA might hamper T-cell signaling, rendering anti-tumoral T cells unresponsive.
Importantly, the relevance of adenosine-mediated immune dysfunction has been tested in several studies. For example, dual targeting of A2AR and CD73 showed a significant combination benefit in controlling tumor growth and lung metastases in mice ([@DMM034272C118]). These promising results led to the initiation of several clinical trials with small-molecule inhibitors targeting A2AR ([@DMM034272C109]). However, more data need to be collected to prove the feasibility of such approaches in cancer patients.
Salts and other factors as overlooked players in the tumor microenvironment {#s4}
===========================================================================
In addition to glucose, amino acids and other well-researched metabolic substrates, nutrients such as fatty acids and complex lipids might play an immunometabolic role in the tumor microenvironment ([@DMM034272C1]; [@DMM034272C74]). Interestingly, vitamins and trace elements are also used by immune cells ([@DMM034272C76]; [@DMM034272C113]). Therefore, disturbances in the pathways that involve various nutrients and soluble factors may play a role in tumor-induced immune-cell modulation. Not much is known about these pathways, even in a non-cancer setting, and future research to address this area might provide new interesting targets for fine tuning of immune therapy. Interestingly, high salt \[sodium chloride (NaCl)\] concentrations have recently been linked to enhanced growth, increased glucose consumption and lactate production in cancer cells ([@DMM034272C3]; [@DMM034272C2]). In addition, epidemiological studies demonstrated that high salt is correlated with increased incidence of gastric cancer ([@DMM034272C107]). Studying breast cancer cells, Amara et al. observed a proliferative effect of salt on cancer cells that appeared to be mediated through a G0/G1 phase release following phosphorylation of salt-inducible kinase 3 (SIK3) through mTORC2 ([@DMM034272C4]). However, other studies reported that high salt concentrations are anti-proliferative for cancer cells ([@DMM034272C6]). Although there is very limited information about electrolyte concentrations in the tumor microenvironment, given the high proliferative rate, extensive metabolic reprogramming, overexpression of several ion transporters and dysregulated vascularization of tumors, it would not be surprising to find that salt homeostasis is disturbed in malignant tissues.
Interestingly, increased salt concentrations may also affect immune cells. In this context, high NaCl conditions triggered a switch to an inflammatory Th17 phenotype by inducible salt-sensing kinase SGK1 ([@DMM034272C116]). In contrast, NaCl inhibited the suppressive function of Treg cells ([@DMM034272C47]). In a recent cancer vaccine study, an increased salt concentration in the vaccine formulation dramatically improved vaccination-induced tumor rejection through CD8+ T cells in a mouse E.G7-OVA lymphoma model ([@DMM034272C66]). In contrast to the pro-inflammatory effects of salt that might support anti-tumor immunity, it has been reported that salt can induce the pro-tumor M2 phenotype of macrophages ([@DMM034272C2]). Potassium (K^+^) is highly abundant intracellularly and is released upon cell death. In a recent paper, high extracellular K^+^ concentrations were detected in necrotic areas of mouse and human tumors, reaching concentrations 5- to 10-times higher than normal serum levels ([@DMM034272C30]). As intact ion transport is essential for T-cell function, high intratumoral K^+^ levels could lead to an impairment of TCR-driven Akt-mTOR phosphorylation. Eil et al. showed that increased K^+^ disturbs Akt-mTOR signaling through the serine/threonine phosphatase PP2A. Importantly, T-cell function could be restored via overexpression of K^+^ channels in T cells, leading to a prolonged survival of tumor-bearing mice ([@DMM034272C30]). Thus, the toolbox that tumors might use to evade or modulate anti-tumor immune responses is extensive and future research will determine whether these pathways might provide novel targets for cancer therapy.
Outlook {#s5}
=======
The field of cancer immunometabolism gained significant attention in recent years. Taking advantage of the extraordinary amount of data collected in the field of cancer metabolism, experts in immunometabolism were able to apply pre-established techniques and modify pre-existing hypotheses. However, the vast majority of the collected data was obtained in *in vitro* experiments or *in vivo* animal studies. Despite the challenges in experiments involving human samples, such as heterogeneity both within the same tumor and between patients, it is critical to assess the metabolic interaction of immune and cancer cells in human tumors. Nevertheless, several investigators aim to transfer the knowledge from preclinical models to a clinical setting. For example, a recent Phase 1/2 study that assessed the effect of IDO1 inhibition in combination with a DC vaccine showed a good therapy tolerance and suggested a possible chemo-sensitization effect in patients with advanced cancers ([@DMM034272C103]). IDO1 inhibition in combination with checkpoint inhibition (e.g. PD-1 blockade; Box 1) has also been tested in several Phase 1 and 2 studies, with encouraging results. However, recent results from a Phase 3 study that combined the IDO1 inhibitor epacadostat with pembrolizumab, an anti-PD1 antibody, showed that adding epacadostat had no benefit (<https://clinicaltrials.gov/ct2/show/NCT02752074>).
As the number of mechanisms and thus possible targets is steadily increasing, a key question arises -- can modulation of one metabolic pathway, such as through IDO1 blockade, influence the outcome of immune-cancer interaction to induce tumor regression? We speculate that future studies will aim to address the metabolic complexity of the tumor microenvironment rather than target a specific protein or gene. It is currently unclear how the modulation of such a complex system to treat cancer can be achieved. However, exciting diagnostic (e.g. broad gene transcription assessment) or interventional (high-throughput platforms testing compound libraries) technologies are currently emerging and being implemented into laboratory and clinical practice, and might assist in reaching this goal.
This article is part of a special subject collection 'Cancer Metabolism: models, mechanisms and targets', which was launched in a dedicated issue guest edited by Almut Schulze and Mariia Yuneva. See related articles in this collection at <http://dmm.biologists.org/collection/cancermetabolism>.
**Competing interests**
The authors declare that they do not have any competing or financial interests.
**Funding**
This work was supported by Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (SNSF) project grant (31003A_163204) and Cancer Research Institute-CLIP (both P.-C.H.), Deutsche Forschungsgemeinschaft (KFO262) (M.K.) and Else Kröner-Fresenius-Stiftung (P.J.S.).
| {
"pile_set_name": "PubMed Central"
} |
{.719}
{.720}
{.721}
{.722}
{.723}
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Ischemia is a condition of poor oxygen supply to the tissues; prolonged ischemia and subsequent reperfusion causes severe irreversible damage to the myocardium \[[@b1-medscimonit-22-1250],[@b2-medscimonit-22-1250]\]. Consequences of myocardial ischemia reperfusion (MI/R) injury include thrombolysis, angioplasty, coronary by-pass, and heart transplantation \[[@b3-medscimonit-22-1250]--[@b6-medscimonit-22-1250]\]. Mediation of reperfusion injury is multifactorial and includes increased production of oxygen free radicals, Ca^2+^ levels, loss of membrane phospholipids, and endothelial dysfunction \[[@b7-medscimonit-22-1250],[@b8-medscimonit-22-1250]\]. These factors lead to changes in myocardial functional status. Oxidative stress is one of the key mediators of MI/R injury. Increased reactive oxygen and nitrogen species target cellular protein and lipid moieties. These reactive species arise from the arachidonic acid pathway, mitochondrial electron transport chain, and through neutrophil activation \[[@b9-medscimonit-22-1250]\]. In recent years, antioxidants have gained much importance in preventing I/R injury because these compounds acts as redox balancers. Antioxidant molecules directly scavenge the free radicals and also induce antioxidant enzyme activities to mediate protective effects.
Plumbagin (5-hydroxy-2-methyl-1, 4-naphthoquinone) is a potent antioxidant. Plants rich is this compound were used in ancient times for protection against heart and liver diseases and for their neuroprotective properties \[[@b10-medscimonit-22-1250]\]. Plumbagin had been reported to mediate anti-inflammatory, analgesic, and anti-arthritic activities \[[@b11-medscimonit-22-1250],[@b12-medscimonit-22-1250]\]. Various studies have reported a potential role in anti-cancer effects, including cancers in the breasts \[[@b13-medscimonit-22-1250]\] and lungs \[[@b14-medscimonit-22-1250]\], as well as leukemia \[[@b15-medscimonit-22-1250]\], melanoma \[[@b16-medscimonit-22-1250]\], prostate cancer \[[@b17-medscimonit-22-1250]\], and osteosarcoma \[[@b18-medscimonit-22-1250]\]. Plumbagin modulates redox status through targeting oxidative stress and redox-sensitive transcription factor (NF-κB) \[[@b19-medscimonit-22-1250],[@b20-medscimonit-22-1250]\]. Further, inflammation-associated cytokine expression was significantly reduced by plumbagin and it effectively reduces inflammation and prevents endotoxemia \[[@b21-medscimonit-22-1250]\]. The potential role of plumbagin in prevention of oxidative stress-associated diseases and other important biological functions \[[@b22-medscimonit-22-1250]\], as well as antifungal \[[@b23-medscimonit-22-1250]\] and anti-atherosclerotic \[[@b24-medscimonit-22-1250]\] action have been reported. In the present study, we aimed at understanding the protective role of plumbagin against MI/R injury and its mechanism.
Material and Methods
====================
Animals and myocardial ischemia-reperfusion (I/R) treatment
-----------------------------------------------------------
Male C57BL6/J mice 8--12 weeks of age were used for the present study. All experimental procedures were approved and followed the guidelines of the Institute for Animal Care and Use Committee at the Chinese Ministry of Education and Chinese Ministry of Health, Qilu Hospital, China. The animals were maintained in separate cages with controlled conditions of temperature (22±1^o^C) and relative humidity (70--72%) with alternate dark and light cycles. The animals were acclimatized to their environment for 1 week and fed with standard rat pellets and water ad libitum. Surgical ligation of the left coronary artery (LCA) was performed as described previously \[[@b25-medscimonit-22-1250]\]. The rats were randomly divided into 4 groups with 10 animals in each (n=10): Group 1 (sham); Group 2 (plumbagin); Group 3 (MI/R injury); and Group 4 (plumbagin+ MI/R group) subjected to 45 min of myocardial ischemia followed by 4 h of reperfusion. Animals with MI/R injury received plumbagin (5 mg/kg) with i.p. injection 1 h before the reperfusion. A preliminary study was carried out with different plumbagin concentrations (2.5, 5, and 10 mg/kg) (data not shown); however, 5 mg/kg showed better cytoprotection. Therefore, further studies were performed with this dose. After the treatment, the hearts were removed and the LV was frozen in liquid nitrogen and stored at −80°C.
Oxidative stress parameters
---------------------------
### Lipid Peroxidation
The lipid peroxidation content was determined as described by Ohkawa et al. \[[@b26-medscimonit-22-1250]\]. Thiobarbituric acid reactive substance (TBARS) was measured spectrophotometrically at 532 nm.
### Reactive oxygen species
The tissue samples were incubated in 50 μl of a 30-μM c-H~2~DCFDA stock solution for 30 min. The reaction mixture was centrifuged and fluorescent intensity was measured using a 485/520-nm filter set. The results are expressed as percentage of ROS generation \[[@b27-medscimonit-22-1250]\].
Antioxidant status
------------------
### Non-enzymic antioxidant -- Glutathione
The total GSH content was determined using Cayman's GSH assay kit. The principle involves the reaction between the sulfhydryl group of GSH and DTNB (5, 5′-dithiobis-2-nitrobenzoic acid) in the presence of glutathione reductase. The formation of 5-thio-2-nitrobenzoic acid (TNB) in reaction with Ellman's reagent was measured at 412 nm.
#### Enzymic antioxidant activity -- GST
The principle involves measurement of the conjugation product of 1-chloro-2, 4-dinitro benzene (CDNB) with reduced glutathione, which is measured 340 nm. One unit of GST activity is the amount of enzyme producing 1 mmol of CDNB-GSH conjugate/min \[[@b28-medscimonit-22-1250]\].
#### GPx
The principle involves reduction of oxidized glutathione (GSSG) formed during GPx reaction, which is reduced by nicotinamide adenine dinucleotide phosphate (NADPH) and glutathione reductase. Thus, the rate of NADPH consumption is proportional to GSSG formation. The kinetic change was measured at 340 nm (37°C) for 3 min. GPx activity was expressed as mmol of NADPH oxidized/minute/mg protein (U/mg protein) \[[@b29-medscimonit-22-1250]\].
#### Catalase
The CAT activity was measured as described by Clairborne (1985) \[[@b30-medscimonit-22-1250]\]. The principle involves the measurement of CAT activity by the rate of H~2~O~2~ degradation, which was measured at 230 nm. The results are expressed as H~2~O~2~ consumed/min/mg protein.
#### SOD
The SOD activity was estimated as described by Kakkar et al. (1985) \[[@b31-medscimonit-22-1250]\]. Superoxide dismutase (SOD) activity is based on the inhibition of the formation of (NADH-PMS-NBT) complex. 1 U of SOD activity is calculated as a 50% reduction in NBT/1 min. The results are expressed as U/mg of protein.
### ELISA: MCP-1, TNF-α, IL-6, and IL-8 levels
The serum cytokines were measured using Cayman's EIA kit (Cayman Chemicals, Ann Arbor, MI). The protocol was performed as described in the manufacturer's instructions. The interleukin levels are expressed as pg/ml.
Western blot analysis
---------------------
The tissues (20 mg) were placed in pre-chilled glass petri dishes and minced on ice using sharp scissors followed by addition of 200 μL of cytoplasm isolation buffer (10 mM HEPES-KOH, pH 7.9, at 4°C, 1.5 mM MgCl~2~, 10 mM KCl, 0.5 mM DTT, and 0.2 mM PMSF) incubated on ice for 15 min and centrifuged at 13 000 rpm for 20 min. The supernatant containing the cytoplasm was discarded. Nuclear extract was isolated from the remaining pellets. We added 50 μL of nuclear fraction isolation buffer (20 mM HEPES-KOH, pH 7.9 at 4°C, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl~2~, and 0.2 mM EDTA) incubated for 40 min in ice and vortexed at 10-min intervals. The mixture was centrifuged at 13 000 rpm for 5 min. The supernatant containing the nuclear extract was aliquoted in separate tubes and stored at −80°C for further analysis. Approximately 50 μg of protein was loaded on 12% SDS-PAGE gels and resolved at 100 V for 2 h. The proteins were transferred into PVDF membranes. Membranes were blocked with skimmed milk for 1 h. Later, the blots were washed and incubated with primary antibodies directed against NF-κB-p65 and COX-2 (1: 1000, Calbiochem, La Jolla, CA) and Nrf-2, HO-1, NQO1, and GST (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA) proteins overnight at 4°C. After washing in TBST, we added secondary antibody (horseradish peroxidase-coupled rabbit IgG) and incubated for 1 h. The immunoreactive proteins were detected with the ECL-Western blot system (Amersham Pharmacia, Braunschweig, Germany) and subsequent autoradiography for 2 min.
Statistical analysis
--------------------
Data are expressed as mean ± standard deviation. One-way analysis of variance \[ANOVA\] followed by Tukey's multiple comparison test and the *t* test were used for statistical analysis. p\<0.05 was considered significant. All experiments were performed 3 times to ensure reproducibility.
Results
=======
Plumbagin prevents oxidative stress during myocardial I/R injury
----------------------------------------------------------------
Reactive oxygen species generation and lipid peroxide content was significantly higher (p\<0.05) during myocardial I/R injury in Wistar rats compared to control. Plumbagin treatment reduced the oxidative stress by decreasing ROS and lipid peroxide content (p\<0.05) compared to myocardial I/R injury rats ([Figure 1](#f1-medscimonit-22-1250){ref-type="fig"}).
Plumbagin improved antioxidant defense system: Upregulation of antioxidant enzyme levels
----------------------------------------------------------------------------------------
Myocardial I/R injury was significantly less (p\<0.05) in the enzymic and non-enzymic antioxidant defense system of GSH, SOD, CAT, GPX, and GST activities compared to that of sham rats. Rats treated with plumbagin followed by MI/R injury had better (p\<0.05) antioxidant status than myocardial I/R injury rats ([Figure 2](#f2-medscimonit-22-1250){ref-type="fig"}).
Plumbagin induces Nrf-2 activation and exerts anti-inflammation
---------------------------------------------------------------
Myocardial I/R injury showed improved oxidative stress and inflammation by upregulation (p\<0.05) of NF-κB and downregulation of Nrf-2 and their target gene expression compared to that of controls. Plumbagin treatment followed by MI/R injury induced Nrf-2 activation with concomitant increases (p\<0.05) in HO-1, NQO1, and GST protein expression. Further, plumbagin reduced inflammatory markers such as NF-κB, COX-2, and iNOS protein expression (p\<0.05) compared to that of MI/R injury rats ([Figure 3](#f3-medscimonit-22-1250){ref-type="fig"}).
Plumbagin induced anti-inflammatory effect by decreasing cytokine levels
------------------------------------------------------------------------
[Figure 4](#f4-medscimonit-22-1250){ref-type="fig"} shows significant upregulation (p\<0.05) of pro-inflammatory cytokine expression (MCP-1, IL-6, IL-8, and TNF-α) in MI/R injury compared to sham rats. Cytokine expressions were downregulated (p\<0.05) by plumbagin treatment in MI/R injury rats, showing its anti-inflammatory action against myocardial injury ([Figure 4](#f4-medscimonit-22-1250){ref-type="fig"}).
Discussion
==========
In the present study, we determined that plumbagin provides significant cytoprotection against MI/R injury. Myocardial injury showed a significant increase in reactive oxygen species and modulated oxidative stress and inflammation through Nrf-2 and NF-κB pathways. Plumbagin treatment effectively protected against myocardial I/R injury by upregulating Nrf-2 and downregulating NF-κB expressions.
Myocardial I/R injury impairs the blood supply to the myocardium, leading to irreversible cardiac damage, which is mainly mediated through cell death mechanisms, including apoptosis and necrosis \[[@b32-medscimonit-22-1250]\]. However, oxidative stress and inflammation are early events, ultimately leading to cell death. Ischemic myocardium converts the reintroduced molecular oxygen during reperfusion and increases the reactive oxygen species \[[@b33-medscimonit-22-1250]\]. Treatment with antioxidants against myocardial I/R injury might regulate the overall protection mechanism against initiation and progression of myocardial I/R-induced effects. In the present study, plumbagin significantly prevented myocardial oxidative stress by decreasing generation of reactive oxygen species and lipid peroxide content. Plumbagin has 5 functional hydroxyl and 2 methyl groups, it shows potential antioxidant effects and anti-mutagenic effects against gamma radiation-induced oxidative stress \[[@b34-medscimonit-22-1250]\]. Plumbagin modulated LPS-induced redox status and prevented endotoxic shock and oxidative stress in macrophages \[[@b21-medscimonit-22-1250]\]. Thus, the antioxidant potential of plumbagin might be involved in the protective effect against myocardial I/R injury through reducing oxidative stress.
The central mechanism in regulating oxidative stress and inflammatory mechanism is redox balance \[[@b35-medscimonit-22-1250]\]. Cells are equipped with enzymic and non-enzymic antioxidant defence mechanisms to scavenge the free radicals and check the balance, thereby preventing the activation of inflammatory mediators. NF-κB is activated under redox imbalance and regulates the transcription of inflammation-associated genes, including COX-2 \[[@b36-medscimonit-22-1250],[@b37-medscimonit-22-1250]\]. In the present study, we found that myocardial I/R injury activated NF-κB and expression of its downstream targets, COX-2 and iNOS \[[@b37-medscimonit-22-1250]\]. Treatment with plumbagin suppressed the oxidative stress levels and prevented expression of inflammatory proteins. Further, pro-inflammatory cytokine expressions such as TNF-α, MCP-1, IL-6, and IL-8 levels were significantly reduced by plumbagin treatment compared to that of MI/R injury rats. Similar results were reported earlier in studies in which LPS-induced redox status and inflammation was suppressed by plumbagin treatment by downregulating NF-κB, MAPK, and pro-inflammatory cytokines TNF-α and IL-6 \[[@b21-medscimonit-22-1250]\]. The anti-inflammatory effect of plumbagin was shown to be mediated through inhibition of NF-κB, COX-2, and iNOS and pro-inflammatory cytokine expression in rat paw edema models \[[@b12-medscimonit-22-1250]\]. Thus, plumbagin might regulate inflammatory protein expressions of NF-κB and their downstream targets (COX-2, iNOS), as well as pro-inflammatory cytokine expressions, by regulating ROS levels.
One of the main regulators of redox status in the cells is Nrf-2, a redox-sensitive transcription factor. Nrf-2 maintains basal and induced levels of antioxidant status and protects against cellular oxidants. Under basal or un-stressed conditions, Nrf-2 is bound to Kelch-like ECH-associated protein1 (Keap1), which is an endogenous inhibitor that prevents Nrf-2 activation. During stress, Nrf-2 -Keap-1 cysteine interaction is modified, leading to Nrf-2 activation. The Nrf-2 then translocates into the nucleus and binds with other transcription factors and induces antioxidant gene expression through ARE (Antioxidant Response Element) binding \[[@b38-medscimonit-22-1250],[@b39-medscimonit-22-1250]\]. Nrf-2 regulates expression of about 200 genes related to the cellular antioxidant defense system, some of which include (NAD(P)H quinone oxyreductase, glutathione, and heme oxygenase-1 (HO-1). Disruption of Nrf-2 levels results in excessive oxidative stress and inflammation, leading to the onset of various diseases \[[@b39-medscimonit-22-1250]\]. In the present study, myocardial I/R injury induced increased oxidative stress and inflammation, which might be related to downregulation in Nrf-2 levels and antioxidant enzyme activities. Plumbagin improved the antioxidant defense mechanism through activation and expression of Nrf-2. Further, Nrf-2 downstream targets (NQO1, GST, and HO-1) expression were significantly upregulated by plumbagin compared to rats with MI/R injury. Previous studies on plumbagin showed that cerebral ischemia was effectively ameliorated by translocation and expression of Nrf-2 and HO-1 proteins \[[@b40-medscimonit-22-1250]\]. Thus, plumbagin reduces oxidative stress and inflammation in MI/R injury rats through upregulating Nrf-2 expression and improving antioxidant status.
Conclusions
===========
The present study shows novel evidence of the cardioprotective role of plumbagin against myocardial I/R injury by modulating important mechanisms of oxidative stress and inflammation through Nrf-2 expression.
**Conflict of interest**
The authors declare no conflict of interest.
**Source of support:** Departmental sources
{#f1-medscimonit-22-1250}
{#f2-medscimonit-22-1250}
{#f3-medscimonit-22-1250}
{#f4-medscimonit-22-1250}
[^1]: Study Design
[^2]: Data Collection
[^3]: Statistical Analysis
[^4]: Data Interpretation
[^5]: Manuscript Preparation
[^6]: Literature Search
[^7]: Funds Collection
| {
"pile_set_name": "PubMed Central"
} |
######
Key Messages
CPRD data have been extensively used for observational research. For example, the data were used to show that there was no association between MMR vaccine and autism, and to show an association between oral corticosteroid use and increased risk of fractures.The CPRD has a large UK dataset bringing together longitudinal primary care medical records from participating practices. Over half of CPRD patients are eligible for linkage to additional datasets, including hospital data, national cancer registration data and national mortality records.Quality of some data is driven by the Quality and Outcomes Framework in the UK, and data are also monitored by CPRD internal processes. Analyses described in this paper show that active (alive, currently registered) CPRD patients are representative of the UK population in terms of age and sex.CPRD data originate from routine clinical practice, and their use for epidemiological studies typically requires extensive data processing and an understanding of the way the data are originally recorded and stored.
Data resource basics
====================
UK primary care data for research
---------------------------------
Over 98% of the UK population are registered with a primary care general practitioner (GP)[@dyv098-B1] and under the National Health Service (NHS), visits to the GP are free of charge. The GP is the gatekeeper of care in the UK National Health Service. GPs act as the first point of contact for any non-emergency health-related issues, which may then be managed within primary care and/or referred to secondary care as necessary. Secondary care teams also feed back information to GPs about their patients, including key diagnoses. Patient data are routinely recorded onto computers by practice staff, against a unique patient NHS number. These facets of UK primary care provide good capture of health information in a longitudinal electronic health record.
The Clinical Practice Research Datalink (CPRD)
----------------------------------------------
The CPRD harnesses general practice data and produces a primary care dataset, which is one of the largest databases of longitudinal medical records from primary care in the world ([Table 1](#dyv098-T1){ref-type="table"}). Established in London in 1987, the small Value Added Medical Products (VAMP) dataset grew to become the General Practice Research Database (GPRD) in 1993,[@dyv098-B2]^,^[@dyv098-B3] before expanding to become the CPRD in 2012. The CPRD collates routinely collected anonymised electronic health record data from general practices who have agreed at a practice level to provide data on a monthly basis. All patients registered with the participating practices are included in the dataset, unless they have individually requested to opt out of data sharing, by asking their GP to amend their registration details on the system to disable the extraction of their data. Table 1.Key details about the Clinical Practice Research DatalinkCounties participatingUK: England, Wales, Scotland and Nortdern IrelandWho is included?Patients registered at general practices that contribute data to CPRD, who have not dissented from secondary use of GP patient-identifiable dataWhat is recorded?Demographics, diagnoses, symptoms, signs, prescriptions, referrals, immunisations, behavioural factors, testsPeriod of data collection1987 to presentAverage duration of follow-up 5.1 yearsFunding sourceCPRD has received funding from the MHRA, Wellcome Trust, Medical Research Council, NIHR Health Technology Assessment programme, Innovative Medicines Initiative, UK Department of Health, Technology Strategy Board, Seventh Framework Programme EU, and various universities, contract research organizations and pharmaceutical companies
Data linkage
------------
A subset of English practices (currently 75%, representing 58% of all UK CPRD practices) have consented to participate in the CPRD linkage scheme and have provided patient-level information. Patient-level data from consenting practices are linked via a trusted third party (the Health and Social Care Information Centre[@dyv098-B4]) to other existing data sources. Established linkages include Hospital Episode Statistics[@dyv098-B5] (hospitalisation data), Office for National Statistics[@dyv098-B6] (mortality data including causes of death), Index of Multiple Deprivation and Townsend scores (deprivation data)and disease registries including the National Cancer Intelligence Network,[@dyv098-B7] and the Myocardial Ischaemia National Audit Project[@dyv098-B8] (details in [Supplementary Table 1](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1), available as [Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) at *IJE* online). Other linkages are planned (see CPRD website[@dyv098-B9]) and researchers can make requests for bespoke linkage for individual studies.
Uses for observational research and interventional research
-----------------------------------------------------------
Subject to the appropriate data governance and approvals, the CPRD can supply primary care and linked patient data to researchers in the UK and internationally. Through the CPRD, researchers can approach practices and patients to take part in biosample collection studies or trials. The feasibility of this work has been tested: patients from the CPRD have been recruited to a pharmacogenetic study of statin-induced myopathy,[@dyv098-B10]^,^[@dyv098-B11] practices have been recruited to cluster randomised trials[@dyv098-B12]^,^[@dyv098-B13] and patients have been recruited to pragmatic point-of-care randomised trials.[@dyv098-B14] The electronic health record data can be used alongside the study data to provide a full clinical picture for the recruited patients.
Ethics
------
The CPRD has broad National Research Ethics Service Committee (NRES) ethics approval for purely observational research using the primary care data and established data linkages. Other uses of CPRD data may require separate ethical approval. This is likely if there is any specific patient involvement in the study; for example, if the researcher wishes to ask patients to complete a questionnaire for Patient Reported Outcomes, or to conduct an interventional trial among CPRD patients.
Data governance, practice and patient confidentiality
-----------------------------------------------------
The CPRD strives to operate within UK and European laws to protect confidentiality. Governance requirements to protect patient confidentiality where patient consent has not been obtained are respected by ensuring that patient identifiers are held separately from the clinical data and that there is separation between researchers with access to identifiable information from the primary study and those using CPRD data.
Funding sources
---------------
The CPRD is a joint venture from the Medicines and Healthcare Regulatory Agency (MHRA) and the National Institute for Health Research (NIHR).The CPRD is owned by the UK Department of Health and operates within the MHRA. The CPRD has received funding for studies from the MHRA, Wellcome Trust, Medical Research Council, NIHR Health Technology Assessment programme, Innovative Medicines Initiative, UK Department of Health, Technology Strategy Board, Seventh Framework Programme EU and various universities, contract research organizations and pharmaceutical companies.
Data resource area and population coverage
------------------------------------------
[Figure 1](#dyv098-F1){ref-type="fig"} describes the population coverage of CPRD primary care data across England, Wales, Scotland and Northern Ireland. At the mid-year date of 2 July 2013, the dataset held information on 11.3 million patients who were deemed acceptable for research based on data quality checks (Appendix 1, available as [Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) at *IJE* online, and described below). The population of active patients (alive and currently registered) on 2 July 2013 was 4.4 million, representing 6.9% of the total UK population (based on the UK 2013 mid-year population of 64.1 million). The remaining 6.9 million records represent inactive patients who have died or are no longer registered with a participating practice. Patient numbers by age, sex, deprivation, ethnicity and region are described in [Table 2](#dyv098-T2){ref-type="table"}. Figure 1.Distribution of 674 CPRD practices by region in England, and in Wales, Scotland and Northern Ireland.Note: practices mapped are those contributing up to standard data to the dataset on 2 July 2013, based on the January 2014 dataset build Table 2.Demographic characteristics of acceptable CPRD patients (January 2014 dataset build), and the subset of those active on 2 July 2013All patientsActiveNo. patients112992214425016Men, *n* (%)5478715 (48.5)2183161 (49.3)Women, *n* (%)5820506 (51.5)2241855 (50.7)Age in 2013, *n* (%) (years) \<18-- 742765 (20.2) 18-64-- 4402926 (61.8) 65+-- 1728514 (18.1)Region, *n* (%) North East184753 (1.6)67639 (1.5) North West1257846 (11.1)523356 (11.8) Yorkshire & The Humber441933 (3.9)48480 (1.1) East Midlands446799 (4)29954 (0.7) West Midlands943011 (8.4)394115 (8.9) East of England1117235 (9.9)306538 (6.9) South West943295 (8.4)377821 (8.5) South Central1236351 (10.9)544979 (12.3) London1532066 (13.6)600824 (13.6) South East Coast1130468 (10)474593 (10.7) Northern Ireland275640 (2.4)153576 (3.5) Scotland960121 (8.5)499969 (11.3) Wales829703 (7.3)403172 (9.1)Duration of follow-up (median years, IQR)[^a^](#dyv098-TF2){ref-type="table-fn"}5.1 (1.8-11.1)9.4 (3.4-13.9)[^1][^2]
Frequency of data collection
----------------------------
Data collection happens as part of normal clinical care of patients in participating practices on a daily basis. The frequency of data recording is determined by patient need and varies by age, sex and underlying morbidity. Patients are included in the primary care dataset from their first until their last contact with the participating practice. Data are collected by practices and usually uploaded to the CPRD secure servers on a monthly basis. The date of last data collection corresponds to the date of the last data upload from each practice. Monthly builds of the primary care dataset are generated and made available for researchers to use.
Measures
========
Practice and patient data
-------------------------
The database structure broadly separates information into clinical, referral, immunisation, test and therapy data (see [Supplementary Table 1](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1), available as [Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) at *IJE* online). Data are recorded against practice and patient pseudo-identifiers. At the practice level, geographical region is recorded by the CPRD as one of 10 regions in England, with Wales, Scotland and Northern Ireland as separate regions ([Figure 1](#dyv098-F1){ref-type="fig"}); a practice-level deprivation score is also calculated based on practice lower super output area.
All general practice encounters are recorded electronically and practitioners are encouraged to make these records available for research. Data are collected on demographic information, prescription details, clinical events (symptoms, diagnoses), preventive care provided, tests, immunisations, specialist referrals, hospital admissions and their major outcomes, and details relating to death (details are shown in [Supplementary Table 2](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1), available as [Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) at *IJE* online).
All entries to a patient record are considered as 'consultations', not all of which will involve a face-to-face encounter. Within a consultation multiple 'events' may be recorded, each with an associated date ([Figure 2](#dyv098-F2){ref-type="fig"}). Figure 2.Example of dataset structure.Note: patients consult with practice staff, where clinical, therapy, referral, test and immunisation information is coded in the medical record.
Data are largely recorded by general practice staff using version 2 Read codes, a hierarchical clinical classification system containing over 96, 000 codes.[@dyv098-B15] For example, during a consultation, a GP, nurse, other healthcare professional, practice manager or administrator may enter a number of Read codes to describe a patient's condition (e.g. lifestyle measures such as smoking status, symptoms, past medical history, diagnoses, tests performed such as blood pressure measurement, and therapies offered). Numerical data on additional clinical measures (e.g. height, weight, blood pressure, alcohol intake) can also be recorded during consultations. Prescriptions issued by the GP are automatically recorded with a product name and British National Formulary code, alongside the dosage instructions and quantity. Results of laboratory tests ordered by the GP are commonly added to the patient record via electronic links to laboratories. Data fed back to the GP from other sources may also be entered into the patient record by practice staff; this might include information from secondary care such as key diagnoses, discharge data from hospitals, or follow-up information from specialist clinics.
The GP is also able to make additional uncoded notes and observations about patients as free text. This often contains identifiable information and is not part of the standard database available to researchers.
Data resource use
=================
Data from the CPRD (or formerly the GPRD or VAMP) have been used in the UK and internationally[@dyv098-B16] to produce close to 2000 research reports, with over 1000 published in peer-reviewed journals, across all major therapeutic areas. A bibliography is maintained by the CPRD and is available online.[@dyv098-B17] These publications cover a range of health-related research topics including pharmacoepidemiology, comparative effectiveness research, health services research, assessments of temporal trends in disease incidence, health economics, prognosis research, classical risk factor epidemiology and more recently randomised controlled trials.[@dyv098-B12]^,^[@dyv098-B18] Publications to date include studies showing the absence of an association between measles, mumps and rubella (MMR) vaccine and autism,[@dyv098-B19] cardiovascular risk after acute infection,[@dyv098-B20] the lower risk of dementia associated with statin use,[@dyv098-B21] the risk of myocardial infarction in patients with psoriasis,[@dyv098-B22] the use of oral corticosteroids and fracture risk[@dyv098-B23] and the association between body mass index and cancer.[@dyv098-B24]
Strengths and weaknesses
========================
Strengths
---------
The strengths of the CPRD data as a research resource lie in the breadth of coverage, size, long-term follow-up, representativeness and data quality.
### Breadth of data
The CPRD primary care dataset is one of few large, ongoing databases that include data on morbidity and lifestyle variables and with a linkage to secondary care and mortality data.
### Size and long- term follow-up
A key strength of this database is its size; the CPRD holds data from 674 practices and includes over 79 million person-years of follow-up (on 2 July 2013, January 2014 dataset). This allows epidemiological associations to be investigated in more detail and estimated with a higher level of statistical precision than is possible with smaller data sources, which is of particular importance for the study of rare exposures and diseases.[@dyv098-B25]^,^[@dyv098-B26] For individual patients, there is a median prospective follow-up of 9.4 years for active patients \[interquartile range (IQR) 3.4--13.9\] and 5.1 years (IQR 1.8--11.1 years) ([Table 2](#dyv098-T2){ref-type="table"}) overall, enabling research into diseases with long latency and the study of long-term outcomes.[@dyv098-B27]
### Representativeness
When compared with the UK census in 2011,[@dyv098-B30] CPRD patients are broadly representative of the UK population in terms of age and sex ([Figure 3](#dyv098-F3){ref-type="fig"}). Patients are also comparable to the UK census in terms of ethnicity,[@dyv098-B31] and comparable to the Health Survey for England for body mass index distribution in most patient subgroups.[@dyv098-B32] However, the CPRD may not be representative of all practices in the UK based on geography and size.[@dyv098-B33] Figure 3.Age distribution of the CPRD primary care data on 27 March 2011 compared with UK Census data 2011, in men (top panel) and women (lower panel). These data are based on a one-million patient sample of CPRD. All patients are acceptable.
### Data quality
Aspects of data quality in English general practice are enhanced by the Quality and Outcomes Framework,[@dyv098-B34] an incentive payment programme for GPs, which encourages recording of key data items (for example smoking status and the delivery of services to key patient groups). The Quality and Outcomes Framework was introduced in 2004, and completeness in recording of many variables showed subsequent improvement ([Figure 4](#dyv098-F4){ref-type="fig"}, and [Supplementary Figure 1](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1), available as [Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) at *IJE* online). Figure 4.Recording of key lifestyle and demographic variables by calendar year (A: ever recorded in patient follow-up; B: recorded in the past 3 years of patient follow-up). These data are based on a one-million patient sample of primary care data from the CPRD. All patients are acceptable.
Validation of the CPRD has shown high positive predictive value of some diagnoses and, where evaluated, comparisons of incidence with other UK data sources are also broadly similar.[@dyv098-B35] However, reporting of validation studies was often too poor to permit a clear interpretation, and the majority of studies focused on positive predictive value rather than sensitivity or specificity.[@dyv098-B39]
The quality of primary care data is variable because data are entered by GPs during routine consultations, not for the purpose of research. Researchers must therefore undertake comprehensive data quality checks before undertaking a study. The CPRD provides two sets of data quality criteria: acceptability for patients and up to standard (UTS) time for practices. These criteria do not ensure data quality, but the CPRD recommends that these measures are used as a first step to selecting research-quality patients and periods of quality data recording. The acceptable patient metric is based on registration status, recording of events in the patient record, and valid age and gender. The UTS date is a practice-based quality metric based on the continuity of recording and the number of recorded deaths. The UTS date is calculated for each participating practice, corresponding to the latest date at which practices meet these minimum quality criteria (Appendix 1, available as [Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) at *IJE* online). The figures given in this paper reflect data for patients labelled as acceptable and who have at least 1 day of follow-up that is 'up to standard'. Research into data quality has shown that, despite these criteria, there were large variations in inter-practice recording of data.[@dyv098-B40]
Weaknesses
----------
### Missing data
The variability in completeness of data across patients and across time requires careful consideration; restriction to those with complete data may result in biased analyses, and imputation may not be a straightforward approach because the patterns of missingness are complex. For example, body mass index may be recorded more frequently in patients with a health issue, and blood pressure more frequently in women of reproductive age and those with existing cardiovascular disease. Complex algorithms are often required to deal with missingness, to resolve discrepancies in measures between consultations and to decide whether historical measurements, for example of body mass index, blood pressure or smoking status, are still appropriate to a patient's disease risk much later in follow-up.[@dyv098-B32]
An additional complexity of primary care data is that the absence of a Read code for disease must be interpreted as an absence of the disease itself, so whereas positive predictive value tends to be high,[@dyv098-B39] sensitivity may be lower. This potential misclassification arises partly due to patients failing to present to the GP with disease, and also from variations between GPs in coding diagnoses in the patient electronic record; if GPs enter information as free text, researchers will miss valuable information. The extent of misclassification may vary between diseases.[@dyv098-B39]
### Definitions
There are not generally standardised definitions for diagnoses and other details, so Read code lists and algorithms need to be developed for each study to identify exposures and outcomes of interest. This may lead to inconsistent definitions (and therefore results) between studies using the same data.
### Information from secondary care
General practices receive information about patient contacts with secondary care but this information must be manually entered into the patient record. Therefore, details about hospital admissions (dates, diagnoses, tests performed, length of stay) may be incomplete.
### Data not captured
Some aspects of health may be recorded very infrequently or not at all, for example level of social support, number of people in a household, over-the-counter medication use, prescriptions in secondary care, prescriptions filled, and adherence to treatments. There are also certain patient groups that are missing from primary care records, such as prisoners, private patients, some residential homes and the homeless.
Data Resource access
====================
Access to patient level data is provided by the CPRD for health research purposes and is dependent on approval of a study protocol by the MHRA Independent Scientific Advisory Committee (ISAC).
Researchers intending to use the data should be aware that the CPRD data files contain millions of rows of data, requiring extensive data management and an in-depth understanding of the way the data are input and stored.
The CPRD provide data dictionaries and coding dictionaries to researchers, and guidance on creating code lists is available to help identify codes of interest.[@dyv098-B41] Read code repositories for electronic health record research are also now available.[@dyv098-B42]^,^[@dyv098-B43]
Details about ISAC applications and data costs are available on the CPRD website, and any other queries can be directed to the CPRD Knowledge Centre \[kc\@cprd.com\].[@dyv098-B9]
Supplementary Data
==================
[Supplementary data](http://ije.oxfordjournals.org/lookup/suppl/doi:10.1093/ije/dyv098/-/DC1) are available at *IJE* online.
Funding
=======
LS is supported by a Wellcome Trust Senior Research Fellowship in Clinical Science grant number 098504/Z/12/Z.
***Conflict of interest:****AG is employed by the CPRD and TvS is an ex-employee of the CPRD. No other conflicts declared.*
Supplementary Material
======================
###### Supplementary Data
[^1]: Active patients are alive and currently registered on 2 July 2013.
[^2]: ^a^Includes only up to standard follow-up.
| {
"pile_set_name": "PubMed Central"
} |
The authors confirm that all data underlying the findings are fully available without restriction. The toolbox and the scripts to generate all simulated data are available at this address: <http://dx.doi.org/10.6084/m9.figshare.1005245>; the cardiovascular data are included in the toolbox package; the epilepsy data are available at this address: <http://math.bu.edu/people/kolaczyk/datasets.html>.
Introduction {#s1}
============
Since its first introduction by Schreiber [@pone.0109462-Schreiber1] transfer entropy (TE) has been recognized as a powerful tool to detect the transfer of information between joint processes. The most appealing features of TE are that it has a solid foundation in information theory and it naturally detects directional and dynamical information. Moreover, the formulation of TE does not assume any particular model as underlying the interaction between the considered processes, thus making it sensitive to all types of dynamical interactions. The popularity of this tool has grown even more with the recent elucidation of its close connection with the ubiquitous concept of Granger causality [@pone.0109462-Barnett1], which has led to formally bridge information-theoretic and predictive approaches to the evaluation of directional interactions between processes. Given all these advantages, TE has been increasingly used to assess the transfer of information in physiological systems with several applications in neurophysiology [@pone.0109462-Wibral1]--[@pone.0109462-Gourvitch1]. It is worth noting that speaking of the transfer of information as measured by TE we refer to the "predictive information transfer" intended as the amount of information added by the past (and present) states of a source process to the present state of a target process.
The estimation of TE from time series data which constitute realizations of the investigated physiological processes is complicated by a number of practical issues that need to be addressed and that are contributing to the development of several recipes to compute this measure.
In this study we discuss three different approaches (binning, nearest neighbor, linear) to evaluate the probability distribution function which constitutes the basis for TE in multivariate systems. In turn, each approach has to be paired with the choice of the time series\' past values which contribute information to the knowledge of the present state of a given target time series. The first choice is the classical uniform embedding (UE) that considers a fixed amount of past terms for each series; the second approach is quite recent and employs a non-uniform embedding (NUE) [@pone.0109462-Faes1], [@pone.0109462-Kugiumtzis1] iteratively selecting the most informative terms through an optimization criterion.
These recipes, some of them already established, some novel, are accordingly revisited or explained. Then, in order to contribute to the foundation of a common framework for the application of TE, we describe their implementation in a modular MATLAB toolbox. Several examples are presented allowing a critical comparison of UE and NUE approaches for all the three entropy estimators.
The paper is organized as follows. We first provide an overview of TE. We then distinguish between UE and NUE approaches to the representation of the history of the observed processes. We describe in detail the methods used to estimate the probabilities involved in the evaluation of the TE and their implementation in the toolbox. The approaches are then validated on synthetic time series and then tested on real data: the electroencephalogram of an epileptic patient and cardiovascular measurements in healthy subjects.
Materials and Methods {#s2}
=====================
Transfer entropy {#s2a}
----------------
Let us consider a composite system described by a set of *M* interacting dynamical subsystems and suppose that, within the composite system, we are interested in evaluating the information flow from the source system to the destination system , collecting the remaining systems in the vector . We develop our framework under the assumption of stationarity, which allows to perform estimations replacing ensemble averages with time averages (for non-stationary formulations see, e.g., [@pone.0109462-Ledberg1] and references therein). Accordingly, we denote *X*, *Y* and **Z** as the stationary stochastic processes describing the state visited by the systems , and over time, and , and as the stochastic variables obtained by sampling the processes at the present time *n*. Moreover, we denote , , and as the vector variables representing the whole past of the processes *X*, *Y* and **Z**. In some cases it can be desirable to take into account also the instantaneous influences of the candidate drivers. In this case, the vectors and defined above should contain also the present terms and . Then, the multivariate transfer entropy from *X* to *Y* conditioned to **Z** is defined as:where the sum extends over all the phase space points forming the trajectory of the composite system. (**a**) is then the probability associated with the vector variable **a** while is the probability of observing knowing the values of . The conditional probabilities used in (1) can be interpreted as transition probabilities, quantifying to which extent the transition of the target system towards its present state is affected by the past states visited by the source system . Specifically, the TE quantifies the information provided by the past of the process *X* about the present of the process *Y* that is not already provided by the past of *Y* or any other process included in **Z**.
The formulation presented in (1) is an extension of the original TE measure proposed for pairwise systems [@pone.0109462-Schreiber1] to the case of multiple interacting processes. The conditional TE formulation, also denoted as partial TE [@pone.0109462-Vakorin1], [@pone.0109462-Kugiumtzis1], rules out the information shared between *X* and *Y* that is mediated by their common interaction with **Z**. Note that the TE can be seen as a difference of two conditional entropies (CE), or equivalently as a sum of four Shannon entropies:
TE has a great potential in detecting information transfer because it does not assume any particular model that can describe the interactions governing the system dynamics, it is able to discover purely non-linear interactions and to deal with a range of interaction delays [@pone.0109462-Vicente1]. Recent research has proven that TE is equivalent to Granger Causality (GC) for data that can be assumed to be drawn from a Gaussian distribution, a case in which the data covariance is fully described by a linear parametric model [@pone.0109462-Barnett1], [@pone.0109462-HlavckovSchindler1]. This establishes a convenient joint framework for both measures. Here we evaluate GC in the TE framework and compare this model-based approach with two model-free approaches.
Reconstruction of the system\'s past states and TE evaluation {#s2b}
-------------------------------------------------------------
We will discuss here the crucial issue of how to approximate the infinite-dimensional variables representing the past of the processes. This problem can be seen in terms of performing suitable conditioned embedding of the considered set of time series [@pone.0109462-Vlachos1].
The main idea is to reconstruct the past of the whole system represented by the processes *X*, *Y*, with reference to the present of the destination process *Y*, in order to obtain a vector containing the most significant past variables to explain the present of the destination. Once *V* is computed it is easy to evaluate TE as the difference of two CEs or through the four entropies using the whole *V* or convenient subsets of it according to [equation (2](#pone.0109462.e024){ref-type="disp-formula"}).
### Uniform embedding {#s2b1}
The large majority of approaches applied so far to estimate TE implicitly follow uniform conditioned embedding schemes where the components to be included in the embedding vectors are selected a priori and separately for each time series. For instance the vector is approximated using the embedding vector , where *d* and *m* are respectively the embedding dimension and embedding delay (the same for and , approximated by and ). In this way it is possible to distinguish between a first phase during which the past states are collected and a second phase during which the estimate of the entropy, and consequently of the CE, is evaluated by means of the chosen estimator, according to the following pseudo-code:
1. build the vector ;
2. use *V* and to evaluate the last two entropies of (2) and, consequently, the lowest CE term (CE2);
3. use to evaluate the first two entropies of (2) and, consequently, the highest CE term (CE1);
4. compute TE as equal to the difference CE1--CE2.
The obvious arbitrariness and redundancy associated with this strategy are likely to cause problems such as overfitting and detection of false influences [@pone.0109462-Vlachos1]. Moreover one should assess which TE values are significant. The significance tests associated with TE estimation based on UE are different for model-based and model free estimators, and are described in the respective following subsections.
### Non-uniform embedding {#s2b2}
Non-uniform embedding constitutes the methodological advance, with respect to the state of art, that we implement as a convenient alternative to UE. This approach is based on the progressive selection, from a set of candidate variables including the past of *X*, *Y*, and considered up to a maximum lag (*candidate set*), of the lagged variables which are most informative for the target variable . At each step, selection is performed maximizing the amount of information that can be explained about *Y* by observing the variables considered with their specific lag up to the current step. This results in a criterion for maximum relevance and minimum redundancy for candidate selection, so that the resulting embedding vector includes only the components of , and , which contribute most to the description of . Starting from the full candidate set, the procedure which prunes the less informative terms is described below:
1. 1\. Get the matrix with all the candidate terms MC , with , , representing the maximum lag considered for the past variables of the observed processes; these matrices will contain also the terms and in case one wants to take into account instantaneous effects.
2. 2\. Run the procedure to select the most informative past variables and the optimal embedding vector:
1. \(a\) Initialize an empty embedding vector
2. \(b\) Perform a while loop on *k*, where *k* can assume values from 1 to the number of initial available candidates, *numC*, in the MC matrix. At the *k-*th iteration, after having chosen candidates collected in the vector :
for number of current candidate terms
3. - • add the *i-*th term of MC, , to a copy of to form the temporary storage variable
- • compute the mutual information between and , estimating the probability density function according to the chosen estimator
4. \(c\) Among the tested , select the term which maximizes the mutual information
5. \(d\) **if** fulfills a test for *candidate significance*, as described below, include it in the embedding vector, , delete it from MC and set .
6. \(e\) **else** end the procedure setting and returning
3. 3\. Use and the full embedding vector to evaluate the third and fourth entropy values of (2) and, consequently, the lowest CE term (CE2)
4. 4\. Take the subset of without the past states belonging to the source process, to evaluate the first and the second term of (2) and, consequently, the highest CE term (CE1)
5. 5\. compute TE subtracting CE2 from CE1.
As described above, candidate selection is performed maximizing the mutual information between the target variable and the vector of the candidates already selected, incremented by the candidate under examination. As we will see in the following sections, the practical implementation of this general criterion consists of an optimization process (i.e., minimization of the conditional entropy or maximization of the conditional mutual information, depending on the estimator chosen). The performances of the processes mentioned above in the reconstruction of the optimal embedding for an assigned target process are also discussed in [@pone.0109462-Kugiumtzis1].
The complexity of the algorithm concerns mainly step 2, in particular step 2(b), involving a *for* loop nested inside a *while* loop: in the worst case the body of the *for* loop is executed times resulting in a complexity .
At step 2(d), the test for candidate significance is performed at the *k*-th step comparing the conditional mutual information between the target variable and the selected candidate given the candidates previously selected up to the -th step, , with its null distribution empirically built by means of a proper randomization procedure applied to the points of . The test for candidate significance is fulfilled if the original measure is above the percentile (where is the desired significance level) of its null distribution. In order to maximize detection accuracy, the adopted randomization procedure is varied for each estimator, and is thus described in the relevant section.
Summarizing, the non-uniform embedding is a feature selection technique selecting, among the available variables describing the past of the observed processes, those who are the most significant - in the sense of predictive information - for the target variable. Moreover, given the fact that the variables are included into the embedding vector only if associated with a statistically significant contribution to the description of the target, the statistical significance of the TE estimated with the NUE approach results simply from the selection of at least one lagged component of the source process. In other words, if at least one component from X is selected by NUE, the estimated TE is strictly positive and can be assumed as statistically significant. If this is not the case, the estimated TE results exactly zero and is assumed as non-significant. This latter case occurs also when the first candidate () does not reach the desired level of significance, meaning that none of the candidates provides statistically significant information about the target variable. In such a case, that is encountered for instance when the target process is a white noise, the code returns an empty embedding vector and assigns a value of zero to the TE.
Entropy estimators {#s2c}
------------------
Estimation of the TE, performed according to either UE or NUE presented above, results from the application of estimators of entropy and CE to the various terms in (2). The toolbox contains three of such estimators. The first is the linear estimator (LIN) that assumes that data are drawn from a Gaussian distribution. Under this assumption, the two CE terms defining the TE can be quantified by means of linear regressions involving the relevant variables taken from the embedding vector [@pone.0109462-Barnett1]. The second estimator is the classical binning estimator (BIN), which consists of coarse-graining the observed dynamics using Q quantization levels, and then computing entropies by approximating probability distributions with the frequencies of occurrence of the quantized values [@pone.0109462-HlavkovSchindler1]. The third estimator is based on -nearest neighbor techniques (NN) which exploit the statistics of distances between neighboring data points in the embedding space to estimate entropy terms; we adopted the bias-reduction method of estimating entropies through neighbor search in the space of higher dimension and range searches in the subspaces of lower dimension [@pone.0109462-Kraskov1].
A problem that can arise dealing with UE and NUE procedures when we use entropy estimators that does not assume any probability distribution concerns the curse of dimensionality. Indeed the more candidates we work with, the more the data points will be spread in the phase space, the more the probability density function will assume a constant value. Consequently the NUE should be the most apt method to avoid the curse of dimensionality because it reduces the dimension of the phase space. We will prove this statement in the [Results](#s3){ref-type="sec"} section when it will be clear from the comparison between UE and NUE for the BIN and NN estimators in multidimensional spaces. We are now going to introduce each estimator in detail.
### Linear estimator (LIN) {#s2c1}
The linear estimator method works under the assumption that the overall process has a joint Gaussian distribution. This assumption allows to work with well-known expressions for the probability density functions. Under this assumption, the two CE terms defining the TE in (2) are expressed by means of linear regressions involving the past states of the systems collected in the vector variables [@pone.0109462-Barrett1]. When the UE is implemented, is approximated with the vector of length , , and the same for and which are approximated by and (here ). When the NUE is implemented, the embedding vectors will contain only the components resulting from the selection procedure. Then, an unrestricted regression of on the full vector , and a restricted regression of on the reduced vector , are performed as follows: where and are vectors of linear regression coefficients. The terms and are scalar white noise residuals with variance and . Under the joint Gaussian assumption, it has been demonstrated [@pone.0109462-Barnett1] that the entropy of conditioned to the unrestricted or restricted regression vectors is, respectively, and , from which follows immediately that:
In this study, the unrestricted and restricted regression models in (3) and (4) were estimated by the least-squares method. In the UE implementation, the order *p* of the regressions was selected by the Bayesian information criterion [@pone.0109462-Schwarz1]; in the NUE implementation, the order resulted implicitly from the selection procedure. In NUE, maximization of the mutual information between the component selected at the step *k* and the target variable (step 2d) was obtained in terms of minimization of the CE , where denotes the variance of the residuals of the linear regression of on . Here, the randomization procedure applied to test candidate significance consisted time-shifting the points of by a randomly selected lag (of at least 20 lags, set to avoid autocorrelation effects) [@pone.0109462-Quiroga1].
The statistical significance of the TE estimated through the UE approach is assessed by a parametric F-test for the null hypothesis that the *p* coefficients of which weigh the past components of the driving process, collected in , are all zero [@pone.0109462-Brandt1]. In this case, the test statistic is , where and are the residual sum of squares of the restricted and the unrestricted model, and *N* is the time series length. The TE is considered statistically significant if *F* is larger than the value of the Fisher distribution with degrees of freedom at the significance level .
### Binning estimator (BIN) {#s2c2}
Here we describe the estimator based on fixed state space partitioning. This approach consists of an uniform quantization of the time series followed by estimation of the entropy approximating probabilities with the frequency of visitation of the quantized states [@pone.0109462-HlavkovSchindler1]. This is the classical approach adopted in the first definition of TE [@pone.0109462-Schreiber1]. A time series *y*, realization of the generic process *Y*, is first normalized to have zero mean and unit variance, and then coarse grained spreading its dynamics over quantization levels of amplitude , where and represent minimum and maximum values of the normalized series. Quantization assigns to each sample the number of the level to which it belongs, so that the quantized time series takes values within the alphabet . Uniform quantization of embedding vectors of dimension *d* results in an uniform partition of the *d*-dimensional state space into disjoint hypercubes of size *r*, such that all vectors *V* falling within the same hypercube are associated with the same quantized vector , and are thus indistinguishable within the tolerance *r*. The entropy is then estimated as:where the sum is extended over all vectors found in the available realization of the quantized series, and the probabilities are estimated for each hypercube simply as the fraction of quantized vectors falling into the hypercube (i.e., the frequency of occurrence of within ). According to this approach, the estimate of TE based on binning results from the application of (6) to the four embedding vectors defined in (2) and determined either by UE or by NUE.
In the NUE implementation, maximization of the mutual information between the component selected at the step *k* and the target variable (step 2d) was obtained in terms of minimization of the CE , with the two entropy terms estimated through the application of (6). As for the LIN estimator, the randomization procedure applied to test candidate significance consisted in time-shifting the points of by a randomly selected lag [@pone.0109462-Quiroga1].
The statistical significance of the TE estimated through the BIN UE approach exploited the method of surrogate data implemented by the time-shift procedure proposed in [@pone.0109462-Vlachos1], [@pone.0109462-Quiroga1], [@pone.0109462-Faes2]. Specifically, the estimated TE is tested against its null distribution formed by the values of TE computed on replications of the original series, where in each replication the source series is time-shifted by a randomly selected lag, set to exclude autocorrelation effects.
### Nearest Neighbor estimator (NN) {#s2c3}
Since its first introduction in 1967 [@pone.0109462-Cover1], the nearest neighbor method has been shown to be a powerful non-parametric technique for classification, density estimation, and regression estimation. This method can be used to estimate the entropy of a *d*-dimensional random variable *X*, , starting from a random sample of *N* realizations of *X*. Following the reasoning in [@pone.0109462-Kraskov1], if we consider the probability distribution for the distance between and its *k*-th nearest neighbor, the probability is equal to the chance that there is one point lying within a distance from , that there are other points at smaller distances from it, and that points have larger distances from . Let be the mass of the -sphere centered at , , where is the density of the variable . Then, the expectation value of iswhere is evaluated through the trinomial formula and is the digamma function. The expectation is taken here over the positions of all other points, with kept fixed. An estimator for is then obtained by assuming that is constant in the entire -sphere. The latter giveswhere *d* is the dimension of *x* and is the volume of the *d*-dimensional unit sphere. For the maximum norm one has simply , while for the Euclidean norm. From (7) and (8) we can evaluate and finally:
The NN estimator faces the issue of the bias in the estimation of multiple entropies for vector variables of different dimensions by computing entropy sums through a neighbor search in the space of higher dimension, and range searches in the projected sub-spaces of lower dimensions [@pone.0109462-Kraskov1]. This approach can be fruitfully exploited for the estimation of the TE, as previously done, e.g., in [@pone.0109462-Wibral1], [@pone.0109462-Vicente1]. To do this, we first rewrite the expression for TE in (2) in terms of the components of the embedding vector spanning a space of dimension :
The term is estimated through neighbor search in the dimensional space, while the three other terms are estimated through range searches in the spaces of dimension , and . Accordingly, adaptation of (9) to the four terms in (10) yields the equation for TE based on the nearest neighbor estimator:where , and are the number of points whose distance from , and *V*, respectively, is strictly less than the distance from to its *k*-th neighbor, and denotes average over all n.
In the NUE implementation of the NN estimator, maximization of the mutual information between the component selected at the step *k* and the target variable (step 2d) was obtained in terms of maximization of the conditional mutual information , which was computed as described above by estimating the four relevant entropies through a neighbor search in the complete space, and range searches in the projected sub-spaces of lower dimensions. Moreover, the randomization procedure applied to test candidate significance consisted in shuffling randomly and independently both the points of and those of . These techniques have been recently shown to be optimal for the selection of candidates in a non-uniform embedding approach using nearest neighbor entropy estimators [@pone.0109462-Kugiumtzis1]. As for the BIN UE method, the statistical significance of the TE estimated through the NN UE approach exploited the method of surrogate data implemented by the time-shift procedure proposed in [@pone.0109462-Vlachos1], [@pone.0109462-Quiroga1], [@pone.0109462-Faes2].
Toolbox structure {#s2d}
-----------------
This section describes how the three TE estimators presented above are implemented in the toolbox, exploiting either the UE or the NUE approach for system state reconstruction.
The same main structure, consisting of the following steps, is common to all methods:
1. normalize the data and perform quantization when needed;
2. evaluate the probability density function (PDF);
3. evaluate CE2 (the second conditional entropy in (2)). This term, accounting for the present state of the target series conditioned to the past of the remaining series including the driver, is evaluated first since it is needed to obtain the complete set of conditional terms including all the series;
4. evaluate CE1 (the first conditional entropy in (2)): this term accounts for the present state of the target series conditioned to a vector including the past of the target series and of the all other series except the driver; such a vector is obtained subtracting the candidates belonging to the driver series from the set of candidates evaluated in the previous step.
Keeping this general scheme in mind, specific steps will be performed for any method of choice. For instance, when using the NUE with the BIN estimator, the steps to be performed are:
1. data quantization;
2. estimation of the PDF, as described in *Binning estimator* section;
3. evaluation of the first and second transfer entropy terms according to *Non-uniform embedding* section.
Given the modularity of the structure shown previously it has been possible to build a user friendly toolbox that allows one to compare all the methods at the same time. The toolbox is available at this link <http://dx.doi.org/10.6084/m9.figshare.1005245>. The package also contains two existing MATLAB toolboxes which are used in some of the calculations: ARFIT [@pone.0109462-Schneider1], a collection of modules for modeling and analyzing multivariate time series with autoregressive models, used for choosing the model order in LIN UE, and OPENTSTOOL [@pone.0109462-Merkwirth1], a software package for signal processing with emphasis on nonlinear time-series analysis, and used in searching for neighbor in NN. In order to optimize the toolbox for speed, the routine *evaluateEntropy*, that estimates the entropy among variables according to , has been converted in a.cpp executable substantially reducing the computation time.
In the following we provide guidelines for the use of the toolbox. Let\'s start from a hypothetical *main* function and let\'s explore how a user should set the parameters to chose which methods to use and, possibly, how to build a new method to be inserted within the toolbox.
In the *exampleMain* file, included in the folder /*MuTE/exampleToolbox*, some commented lines remind the method order that has to be kept in mind when setting the parameters, and the parameters available for each method. A first part then follows, devoted to setting the name of the folder that contains data as, for instance, *mat* files. Each file should contain a matrix with the time series as the rows. Then the folders in which all the output files will be stored are defined. In the second part the function *parametersAndMethods* is called.
The function *parametersAndMethods* requires the following inputs, as reported in [table 1](#pone-0109462-t001){ref-type="table"}.
10.1371/journal.pone.0109462.t001
###### How to set the input parameters: an example.
{#pone-0109462-t001-1}
Name Parameter Description
--------------------- ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
dataDir folder containing data to be analysed
numProcessors number of processors used for the parallel session
dataType filename extension
iresDirGenTS folder in which results will be stored
dataFileName data filename
channels vector containing the series id, among the available series, chosen for the analysis
samplingRate variable used to resample data
endPoint value to cut the series length if necessary
autoPairwiseTarDriv vector containing a 1 or a 0 for each chosen method, reflecting whether TE has to be computed among all the pairs or not. In this latter case, the desired drivers and targets will be specified by *idTargets* and *idDrivers*. By default the instantaneous effects of the drivers are not considered. This can be changed in *parametersAndMethods*, by setting *params_nameMethod.idDrivers = \[tarDrivRows(2,:); tarDrivRows(2,:)\]*.
- the number of data realizations;
- the sampling rate;
- the subset of interest;
- a value to cut the series length if necessary
- a vector specifying whether each method will take into account all the pairwise combinations of chosen variables. By default the instantaneous effects won\'t be considered;
- a vector specifying whether the user will set by hand all the pairwise combinations of the chosen variables. This vector will also be used for visualizing the output. It is worth noting that in this case the user should provide as input also the sequence of the destination series and the driver series;
- the folder in which results can be stored, previously defined;
- the folder in which data are stored, previously defined;
- the folder in which results can be eventually copied;
- the number of processors if the code can be run in parallel on several nodes;
- the name of the method chosen and all the relevant parameters as shown in the comments. Here attention should be paid in setting four parameters if the instantaneous effects have to be considered. First of all the function choosing the candidate terms should be set and consequently the variable *usePresent*: *generateConditionalTerm*, and *usePresent* if the instantaneous effects do not have to be taken into account, *generateCondTermLagZero* and *usePresent* otherwise. Then, if one is interested in the action of more than one driver on a target series, for each driver it can be specified whether its instantaneous effect should be considered by writing twice in a row the number of the driver series. One can also choose which variables belonging to the **Z** set can be considered with their instantaneous effects, filling the vector *idOtherLagZero*, [table 2](#pone-0109462-t002){ref-type="table"}, third column.
10.1371/journal.pone.0109462.t002
###### Example of the parameters required to define the methods for an experiment on 5 variables.
{#pone-0109462-t002-2}
Without Instantaneous Effects With Instantaneous Effects
---------------- ------------------------------- ----------------------------
usePresent 0 1
idTargets
idDrivers
idOtherLagZero
In the second column the instantaneous effects are neglected both for targets and conditioning. In the third column we set instantaneous effects for some drivers and the respective targets. For example, when the target is 1, instantaneous effects are taken into account for driver 2 (first two rows, right column, parameter *idDrivers*) and conditioning variable 3 (first row, right column, parameter *idOtherLagZero*).
For an example of how these parameters should be set, let\'s consider 5 variables; a conditioned analysis and a vector *idTargets* would result in the situation shown in [table 2](#pone-0109462-t002){ref-type="table"}, second column, in which no instantaneous effect are set and the variable *idDrivers* contains on the columns the id of the driver series only once and the variable *idOtherLagZero* is the null vector. An example considering instantaneous effects is reported in [table 2](#pone-0109462-t002){ref-type="table"}, third column, when looking at how drivers 1 and 4 influence the target 2 and how drivers 5 and 2 influence the target 3, with series 5 and 2 as conditoning variables.
The input parameters, including the methods of choice, specified in the function *createNameMethodParams* are stored in a structure called *params* by the function *parametersAndMethods*. This function then computes TE according to the chosen methods, via the function *callingMethods*, and stores the significant results through the function *storingOutput*. In case of multiple realizations/data sets to analyze, the computation can be performed in parallel on separate pools.
The description of the toolbox structure should take into account the structure of the function *callingMethods* that receives in input the data matrix with the time series points in row and the structure *params*. The function reads the names of the methods stored in the *params* structure and computes the TE with all the chosen methods (in parallel if the hardware architecture allows it). This function will return a cell array containing the output of each method.
The open structure of the toolbox allows users to integrate in it their own method. The *main* function should in this case be modified with some comments showing which parameters should be passed as an input to *parametersAndMethods*, and in which order. Each new method should then be implemented following the steps described above using all the necessary parameters conveniently grouped via the function *createNameNewMethodParams*. The new method will be called by setting the appropriate name in *callingMethods*.
The execution time for a single run of the system 14 ranged from 0.4 s for the LIN UE to 90.4 s for NN NUE on a Dell Mini Tower Computer, OptiPlex 990 with four Intel Core i5-2400 CPU at 3.10 GHz, 16 GB of RAM.
One of the purposes of this toolbox is to provide a common framework for all the researchers interested in the application of Transfer Entropy to their data. As part of this effort, MuTE will soon be interfaced with the toolbox TRENTOOL [@pone.0109462-Wibral2]. Readers and users are invited to check periodically the webpages of both toolboxes, that will announce when this interface has been set up.
Simulated data {#s2e}
--------------
The first set of simulated data, implemented to validate the simplest approach to TE, BIN UE, consists of two coupled chaotic maps:where is the coupling coefficient according to which is influencing , , , , is the coefficient that regulates the noise and is a Gaussian noise [@pone.0109462-Marinazzo1]. The function generating these data is *multichaoticmap* available in the folder */MuTE/commonFunctions*.
In the second experiment we simulated five time series in two cases: linear time series, for which we can assume a normal distribution of the variables, and non-linear ones, both generated by an autoregressive (AR) model, [equations (13](#pone.0109462.e205){ref-type="disp-formula"}), (14) [@pone.0109462-Baccal1]. The following equations are for the linear Gaussian autoregressive model: where are drawn from Gaussian noise with zero mean and unit variance. The following are the equations for the non-linear model:
where are drawn from Gaussian noise with zero mean and unit variance. A schematic representation of the simulated couplings, valid for both systems, is reproduced in [figure 1](#pone-0109462-g001){ref-type="fig"}. The function generating these data is *generateTS* available in the folder */MuTE/commonFunctions*.
{#pone-0109462-g001}
Electroencephalogram in epilepsy {#s2f}
--------------------------------
The second experiment is performed on intracranial electroencephalography (EEG) measurements recorded from a patient with refractory epilepsy. The dataset consists of time series from 76 contacts. The first sixty-four of these contacts were placed on a 8×8 grid at the cortical level, while the other 12 were along two six-electrode strips that were implanted in deeper brain structures. Eight sets of measurements were taken on this patient, corresponding to eight different epileptic seizures. An epileptologist, examining the data for each seizure, identified two key periods relating to the seizure i.e., a pre-ictal period, just before the clinical onset, and an ictal one, corresponding to the seizure spread and to the clinical symptoms. Each epoch contained 10 seconds of data recorded at 400 Hz. The data are available at <http://math.bu.edu/people/kolaczyk/datasets.html> and described in [@pone.0109462-Kramer1]. In order to reduce overfitting, in this application data were downsampled to 100 Hz.
Cardiovascular and Cardiorespiratory time series {#s2g}
------------------------------------------------
We considered cardiorespiratory time series measured from 15 young healthy subjects ( years old) undergoing a standard head-up tilt testing protocol [@pone.0109462-Faes3]. The acquired signals were the surface electrocardiogram (ECG), the finger arterial blood pressure, and the respiratory nasal flow, measured at 1 kHz sampling rate for 15 minutes in the resting supine position, and 15 further minutes in the position after passive head-up tilting of the bed table. From these signals, the beat-to-beat variability series of heart period (RR interval), *RR(n)*, systolic arterial pressure (SAP), *Sap(n)*, and respiratory activity, *Resp(n)*, were offline measured respectively as the temporal interval occurring between the *n*-th and the -th R waves of the ECG, as the local maximum of the systolic arterial pressure signal inside the *n*-th heartbeat, and as the nasal flow taken at the onset of the *n*-th heartbeat. The time series are available in the folder /*MuTE/cardiovascular_data*. This measurement convention allows instantaneous effects from *Sap(n)* to *RR(n)*, as well as from *Resp(n)* to *Sap(n)* and to *RR(n)*, which were implemented using the relevant feature of the toolbox. The subsequent data analysis was performed on stationary windows of 300 beats taken in supine and upright body positions; inside these windows, the series were normalized to zero mean and unit variance, obtaining the dimensionless series *resp(n)*, *sap(n)*, *rr(n)*.
Results {#s3}
=======
Simulated data {#s3a}
--------------
The aim of testing the BIN UE approach on the coupled maps of [eq. 12](#pone.0109462.e196){ref-type="disp-formula"} was to show a simple case of applicability for this method, which constitutes the most basic approach to the model-free evaluation of TE. We generated 100 realizations of [eq. 12](#pone.0109462.e196){ref-type="disp-formula"}, each of 512 points, and performed the analysis setting 1 as maximum lag for the candidates, 100 surrogates, and 6 quantization levels. As we can see in [figure 2](#pone-0109462-g002){ref-type="fig"}, the method detected correctly the information transfer returning 100 significant realizations for the link and an average TE much higher than the average TE for the link by means of the detection of only 2 significant realizations over 100. We tested also the other methods, which gave similar positive results as the BIN UE, thus demonstrating the applicability of the toolbox for simulations of bivariate systems with short memory.
{#pone-0109462-g002}
Then we moved to a more challenging situation in terms of number of interacting systems and lag of the interaction effects, considering the time series simulated with [equations 13](#pone.0109462.e205){ref-type="disp-formula"} and [14](#pone.0109462.e207){ref-type="disp-formula"}, which involve five systems and contain influences up to 3 points in the past. The experiments were run on 100 realizations of [eqs. 13](#pone.0109462.e205){ref-type="disp-formula"} and [14](#pone.0109462.e207){ref-type="disp-formula"}, of length equal to 512 points. We investigated the TE between each pair of variables conditioned to the other three. The setup of the experiment was the following: for all estimators, used either in the UE or in the NUE framework, the maximum lag for the candidates was set as 5, the number of surrogates was fixed to 100 and . We set 6 quantization levels for BIN and 10 nearest neighbor for NN estimator.
In order to check whether the methods were able to detect the right information transfers, taking into account [figure 1](#pone-0109462-g001){ref-type="fig"}, we expect the estimators to find a TE greater than zero with the highest significance at the following matrix elements: (1,2), (1,3), (1,4), (4,5), (5,4). [Figures 3](#pone-0109462-g003){ref-type="fig"} and [4](#pone-0109462-g004){ref-type="fig"} report the analysis results obtained respectively for the linear system and the non-linear system. Looking at [Figure 3](#pone-0109462-g003){ref-type="fig"} one can notice that LIN UE has very good performances: this reflects the fact that this approach is, in this case of a linear AR system, "by construction", the most likely to correctly detect information transfer. Its NUE version can detect the same links between the variables, though with a slightly higher number of false positives. The LIN estimator, therefore, is able to reveal the correct information flows for this simulation. On the contrary, BIN UE suffers from the curse of dimensionality mentioned in *Entropy estimators* section: evaluating the influences up to the first 5 past points for all the series implies that the uniform embedding procedure projects the data into a phase space of dimensions, where *M* is the number of time series, resulting in a phase space with 25 dimensions, with the points spread enough to lose relevant information about the transfer entropies in the system. As a consequence, no significant link is retrieved with this approach. NN UE retrieves all the true links, but also detects a number of false interactions. Its better performance compared with BIN UE reflects the ability of the nearest neighbor approach to achieve bias compensation in the estimation of entropies of variables of different dimension. Still, the performance of NN UE is not optimal due to the curse of dimensionality. On the other hand, BIN and NN used in the NUE framework are able to recover all the correct links, with only a few false positives. Moving to [Figure 4](#pone-0109462-g004){ref-type="fig"} depicting TE analysis for the non-linear systems, one can notice that the LIN estimator cannot detect all the correct information flows, returning in addition some false positives. Again, BIN UE cannot detect any link because of the curse of dimensionality; conversely BIN NUE, in which the dimensionality of the space is considerably reduced, has high specificity and sensitivity. NN NUE can achieve almost the same performance as BIN NUE but its specificity is lower, especially along the direction . NN UE this time is not able to detect all the correct information transfers ( remains undetected) and reveals some false positives (, ).
{#pone-0109462-g003}
{#pone-0109462-g004}
To better clarify whether and how much the methods are able to distinguish between the true information transfer links and the false ones, in [Figures 5](#pone-0109462-g005){ref-type="fig"} and [6](#pone-0109462-g006){ref-type="fig"} we plotted the average TE with respect to the number of significant realizations found by the methods. Each retrieved link is a point in this bidimensional space. The true links should be in the upper right corner of the plot corresponding to high TE and high number of significant realizations, and they should be apart from the false links, whose natural location would be around the origin of the plot (low TE and low number of significant links). Looking at [figure 5](#pone-0109462-g005){ref-type="fig"} one can notice that for all the methods, except BIN UE and partly NN UE, the two groups of links are well separated and the false links with an averaged TE greater than zero in [figure 3](#pone-0109462-g003){ref-type="fig"} can be neglected. The opposite reasoning holds for BIN UE that is not able to distinguish between false and true links. For the non-linear system ([figure 6](#pone-0109462-g006){ref-type="fig"}) only BIN NUE can separate well true positive from false positive links.
{#pone-0109462-g005}
{#pone-0109462-g006}
To understand how stable the performance of the methods is, in terms of sensitivity and the specificity, with respect to the length of the analyzed data set, we computed the analysis varying the series length from 128 to 1024 points. [Figures 7](#pone-0109462-g007){ref-type="fig"} and [8](#pone-0109462-g008){ref-type="fig"} depict, respectively for the systems 13 and 14, the Receiver Operating Characteristic (ROC) curves obtained for all methods as a function of the series length. Evaluating the amount of TP (true positives), TN (true negatives), FP (false positives) and FN (false negatives) after grouping together all coupled directions (positives) and all uncoupled directions (negatives), we computed sensitivity as and specificity as . In the case of the linear system ([Figure 7](#pone-0109462-g007){ref-type="fig"}), all methods except the BIN UE provide good performance, with the LIN estimator providing the best sensitivity and specificity. All methods provided robust results with respect to the series length, with only a limited decay in the performance observed for 128 points. In the case of the non-linear system ([Figure 8](#pone-0109462-g008){ref-type="fig"}), the performance was optimal for BIN NUE and NN NUE (with a slightly lower specificity), while the methods implementing either the LIN estimator or the UE approach were considerably less sensitive.
{#pone-0109462-g007}
{#pone-0109462-g008}
Electroencephalogram in epilepsy {#s3b}
--------------------------------
In such high dimensional and redundant data, a non-uniform embedding approach is intuitively the most appropriate to identify the patterns of information transfer specific to the onset and spread of the epileptic seizure. The aim of the experiment was to use the NUE approach in order to characterize the dynamical interactions in the epileptic brain by looking at the information transfer between the variables during the pre-ictal and ictal phases. The embedding size in the embedding matrix (EM) was set to eight. The results are reported in [figure 9](#pone-0109462-g009){ref-type="fig"}. The regions corresponding to one of the depth strips (contacts 70 to 76) and the lower left corner of the grid (contacts 1--4, 9--11 and 17) were resected during anterior temporal lobectomy as they were identified by the epileptologists as the seizure onset zone. The Binning approach to NUE seems to be the one which best identifies these areas as those most influential at the start of the seizure and in the early phases of the spread, signature of a putative seizure onset zone. The Binning approach is more selective with respect to the target variables for each driver and less sensitive to the confounding effect of volume conduction resulting in the diagonal patterns observed with the other methods and probably due to conduction effects on the grid.
{#pone-0109462-g009}
Cardiovascular data {#s3c}
-------------------
The analysis of the information transfer for cardiovascular and cardiorespiratory time series was focused on the directions of interaction that are more studied from a physiological point of view: the link from SAP to RR which is related to the so-called cardiac baroreflex, and the links originating from Resp and directed either to RR or to SAP, related respectively to cardiopulmonary or vasculo-pulmonary regulation mechanisms [@pone.0109462-Faes3]. The particular protocol considered allows to establish a sort of verifiable ground truth. Indeed, in the studied protocol, the transition from supine to upright is known to evoke an activation of the sympathetic nervous system and a concurrent deactivation of the parasympathetic nervous system [@pone.0109462-Cohen1]. Accordingly, the two main physiological regulation mechanisms that are expected to be solicited by this transition are: (i) a substantial increase of the baroreflex regulation (direction ), reflecting the necessity of the cardiovascular system to react with changes in the heart rate to the higher fluctuations in the arterial pressure induced by the sympathetic activation; and (ii) a substantial decrease of cardiopulmonary regulation (direction ), reflecting the dampening of respiratory sinus arrhythmia consequent to the parasymphatetic deactivation [@pone.0109462-Nollo1]. On the contrary, no known alterations of the vasculo-pulmonary regulation (direction ) are expected when moving from supine to upright [@pone.0109462-Faes3]. In our analysis all these trends are well reflected in terms of information transfer when the multivariate TE is estimated using the BIN NUE and NN NUE methods. [Figure 10](#pone-0109462-g010){ref-type="fig"} reports the distribution of the multivariate TE computed along these directions using all methods, with subjects studied in the supine and upright body positions. We observe in [Figure 10](#pone-0109462-g010){ref-type="fig"} that BIN NUE and NN NUE reveal, moving from supine to upright, a substantial increase of the TE from Sap to RR, a substantial decrease of the TE from Resp to RR, and an unchanged TE from resp to SAP. These trends were also observed, though with less evident differences, computing the TE according to the LIN estimator. These results suggest the appropriateness of model free TE estimators based on NUE for detecting the information transfer in physiological time series. On the contrary, the BIN UE estimator shows different trends of difficult physiological interpretation, thus suggesting also in experimental data that the estimated TE may be unreliable due to the curse of dimensionality.
{#pone-0109462-g010}
Conclusions {#s4}
===========
In this work we have considered three entropy estimators able to reveal the information transferred among variables represented by time series. We implemented the estimators in two different ways according to UE and NUE approaches, resulting in six methods, two of which are novel, BIN NUE and NN NUE. We compared all the methods validating them on simulated data first and then on real data. We checked whether and how the methods were affected by the number of variables and by the time lag at which the series influenced each other. From the results obtained we can conclude that the new methods introduced, not assuming any model to explain the data and exploiting the NUE strategy for component selection, can detect the correct information flows and are less affected by the number of involved processes and by their interaction lags. The NUE approaches are indeed prone to work in high dimensional spaces as well as in low dimensional spaces because of their ability to reduce the effective dimension of the phase space, choosing only the right variables at the specific time lag that are better able to explain the destination series. On the contrary, BIN UE and NN UE suffer from the curse of dimensionality when several time series and longer interaction delays are present. Finally, looking at LIN UE and LIN NUE performances we can conclude that, even though the equivalence between Granger causality and TE establishes a convenient joint framework for these two measures, there are some drawbacks in having a predefined model to explain the data when these are non-linear. The better performances obtained by the new methods appear when looking at the ROC curves: BIN NUE and NN NUE have high sensitivity and specificity both for linear and non-linear systems.
All the methods have been implemented in an organic toolbox in MATLAB, allowing straightforward comparisons between the methods, and flexible enough to allow other users to implement their own methods.
[^1]: **Competing Interests:**The authors declare that Dr. Montalto is currently an Associate Editor for PLOS One, and this does not alter adherence to PLOS ONE Editorial policies and criteria.
[^2]: Conceived and designed the experiments: AM LF DM. Performed the experiments: AM. Analyzed the data: AM. Wrote the paper: AM LF DM.
| {
"pile_set_name": "PubMed Central"
} |
**To the Editor**I read the article of Yatsu et al. with great interest ([@B1]). Hypoalbuminemia is key to understanding cardiac cachexia and malnutrition, which are closely associated with a poor outcome in heart failure patients. However, I have two concerns regarding their article.
First, they used a serum albumin level of 3.4 g/dL of as the cut-off value for indicating hypoalbuminemia. The optimal cut-off might be calculated by a receiver operating characteristics analysis, despite the fact that the continuous serum albumin level was not found to be a significant predictor of the outcome in their study.
The second concern involves the observational duration. The authors showed survival durations of approximately six years; however, the number of patients at risk was very small, particularly after several years. The two curves cross also after several years. It might be better to confirm whether or not their data meet the proportional hazard before performing log-rank tests and Cox proportional hazard ratio analyses. Furthermore, generally speaking, acute diseases rarely affect long-term outcome. The short-term outcome might therefore be of more interest in patients with acute decompensated heart failure. Hypoalbuminemia with an optimal cut-off value might be useful for stratifying the short-term outcome in such populations.
**The author states that he has no Conflict of Interest (COI).**
[^1]: Correspondence to Dr. Teruhiko Imamura, <[email protected]>
| {
"pile_set_name": "PubMed Central"
} |
Subsets and Splits