1 Novel features of centriole polarity and cartwheel stacking revealed by cryo - 1 tomography 2 3 4 5 Sergey Nazarov 1 , 2 , 4 , Alexandra Bezler 1 , 4 , Georgios N Hatzopoulos 1 , 4 , Veronika 6 Nemčíková Villímová 1 , Davide Demurtas 2 , Maeva Le Guennec 3 , Paul Guichard 3 and 7 Pierre Gönczy 1 8 9 10 1 Swiss Institute for Experimental Cancer Research ( ISREC ) , School of Life Sciences , 11 Swiss Federal Institute of Technology Lausanne ( EPFL ) , Switzerland 12 2 Interdisciplinary Centre for Electron Microscopy ( CIME ) , Swiss Federal Institute of 13 Technology Lausanne ( EPFL ) , Switzerland 14 3 Department of Cell Biology , University of Geneva , Sciences III , Geneva , Switzerland 15 4 These authors contributed equally to this work . 16 17 18 19 Correspondence : pierre . gonczy @ epfl . ch 20 21 22 Keywords : centriole , cartwheel , microtubules , cryo - electron tomography , 23 Trichonympha , Teranympha , SAS - 6 24 25 26 Running title : cryo - ET of centriolar cartwheel 27 28 2 Summary 29 30 Centrioles are polarized microtubule - based organelles that seed the formation of cilia , 31 and which assemble from a cartwheel containing stacked ring oligomers of SAS - 6 32 proteins . A cryo - tomography map of centrioles from the termite flagellate 33 Trichonympha spp . was obtained previously , but higher resolution analysis is likely to 34 reveal novel features . Using sub - tomogram averaging ( STA ) in T . spp . and 35 Trichonympha agilis , we delineate the architecture of centriolar microtubules , pinhead 36 and A - C - linker . Moreover , we report ~ 25 Å resolution maps of the central cartwheel , 37 revealing notably polarized cartwheel inner densities ( CID ) . Furthermore , STA of 38 centrioles from the distant flagellate Teranympha mirabilis uncovers similar cartwheel 39 architecture and a distinct filamentous CID . Fitting the CrSAS - 6 crystal structure into 40 the flagellate maps and analyzing cartwheels generated in vitro indicates that SAS - 6 41 rings can directly stack onto one another in two alternating configurations : with a slight 42 rotational offset and in register . Overall , improved STA maps in three flagellates 43 enabled us to unravel novel architectural features , including of centriole polarity and 44 cartwheel stacking , thus setting the stage for an accelerated elucidation of underlying 45 assembly mechanisms . 46 47 3 Introduction 48 49 Centrioles are evolutionarily conserved microtubule - based organelles that seed the 50 formation of primary cilia , as well as of motile cilia and flagella . Despite significant 51 progress in recent years , the mechanisms orchestrating centriole assembly remain 52 incompletely understood , in part because the detailed architecture of the organelle has 53 not been fully unraveled . 54 The centriole is a 9 - fold radially symmetric cylindrical organelle typically ~ 500 55 nm in length and ~ 250 nm in diameter , which is polarized along a proximal - distal axis 56 ( reviewed in Azimzadeh & Marshall , 2010 ; Gönczy & Hatzopoulos , 2019 ) . In the 57 proximal region lies a likewise symmetrical cartwheel usually ~ 100 nm in length , which 58 is critical for scaffolding the onset of centriole assembly ( reviewed in Guichard et al , 59 2018 ; Hirono , 2014 ) . In transverse view , the cartwheel is characterized by a central 60 hub from which emanate 9 spokes that extend towards peripherally located 61 microtubule triplets . 62 The SAS - 6 family of proteins is thought to constitute the principal building block 63 of the cartwheel and is essential for its formation across systems ( Culver et al , 2009 ; 64 Dammermann et al , 2004 ; Jerka - Dziadosz et al , 2010 ; Kilburn et al , 2007 ; Kleylein - 65 Sohn et al , 2007 ; Leidel et al , 2005 ; Nakazawa et al , 2007 ; Rodrigues - Martins et al , 66 2007 ; Strnad et al , 2007 ; Yabe et al , 2007 ) . SAS - 6 proteins contain an N - terminal 67 globular head domain , followed by a ~ 45 nm long coiled - coil and a C - terminal region 68 predicted to be unstructured ( Dammermann et al , 2004 ; Kitagawa et al , 2011 ; Leidel 69 et al , 2005 ; Van Breugel et al , 2011 ) . In vitro , SAS - 6 proteins readily homodimerize 70 through their coiled - coil moiety ; such homodimers can undergo higher order 71 oligomerization through an interaction between neighboring head domains ( Kitagawa 72 et al , 2011 ; Nievergelt et al , 2018 ; Van Breugel et al , 2011 ) . Ultimately , this results in 73 the formation of a SAS - 6 ring with a central hub harboring 18 juxtaposed head 74 domains , from which emanate 9 paired coiled - coils that extend peripherally . Such ring 75 oligomers are ~ 23 nm in diameter and bear striking resemblance with a transverse 76 section of the cartwheel observed in cells . Moreover , recombinant Chlamydomonas 77 reinhardtii SAS - 6 ( CrSAS - 6 ) possesses the ability not only to self - assemble into ring 78 oligomers , but also to undergo stacking of such entities , together generating a structure 79 akin to the cartwheel present in the cellular context ( Guichard et al , 2017 ) . 80 Additional features of cartwheel architecture have been unveiled through cryo - 81 electron tomography ( cryo - ET ) of centrioles purified from Trichonympha ( Guichard et 82 al , 2013 , 2012 ) . Three closely related species of these unicellular symbiotic flagellates , 83 T . campanula , T . collaris and T . sphaerica - referred to collectively as T . spp . , populate 84 4 the gut of Zootermopsis damp wood termites ( Tai et al , 2013 ) . T . spp . centrioles are 85 particularly well suited for sub - tomogram averaging ( STA ) of cryo - ET specimens 86 because they harbor an exceptionally long cartwheel - bearing region , reaching several 87 microns ( Gibbons & Grimstone , 1960 ; Guichard et al , 2013 , 2012 ) . STA of purified T . 88 spp . centrioles yielded a ~ 34 Å map ( Fourier Shell Correlation ( FSC ) criterion 0 . 143 ) , 89 which established that the cartwheel comprises stacks of ring - containing elements 90 bearing a central hub from which emanate spokes . Suggestively , a 9 - fold symmetrical 91 SAS - 6 ring generated computationally from the crystal structure of the CrSAS - 6 head 92 domain plus the first 6 heptad repeats of the coiled - coil ( CrSAS - 6 [ 6HR ] ) could be fitted 93 in the hub of this STA map ( Guichard et al , 2012 ) . However , some vertical hub 94 densities remained unaccounted for upon such fitting , raising the possibility that 95 additional components are present . In addition , the T . spp . STA map revealed 9 - fold 96 symmetrical cartwheel inner densities ( CID ) inside the hub proper , with contacts 97 between hub and CID occurring where the fitted CrSAS - 6 [ 6HR ] head domains interact 98 with one another ( Guichard et al , 2013 ) . 99 The T . spp . STA map uncovered a vertical periodicity of ~ 8 . 5 nm between 100 spoke densities emanating from the hub ( Guichard et al , 2013 , 2012 ) . Two such 101 emanating densities merge with one another as they extend towards the periphery , 102 where the vertical spacing between merged entities is hence of ~ 17 nm . There , spokes 103 abut a pinhead structure that bridges the central cartwheel with peripheral microtubule 104 triplets . The STA map also revealed the architecture of the A - C linker , which connects 105 the A - microtubule from a given triplet with the C - microtubule of the adjacent one . 106 Interestingly , both pinhead and A - C linker are polarized along the proximal - distal 107 centriole axis ( Guichard et al , 2013 ) . Given that the centrally located hub and CID were 108 not noted at the time as being polarized , this led to the suggestion that the pinhead 109 and the A - C linker might be critical for imparting polarity to the entire organelle 110 ( Guichard et al , 2013 ) . Further cryo - ET analysis of procentrioles from Chlamydomonas 111 and mammalian cells established that aspects of A - C linker architecture are 112 evolutionarily conserved , including the attachment points on the A - and C - 113 microtubules ; moreover , novel features were revealed , such as a vertical crisscross 114 pattern for the A - C linker in Chlamydomonas ( Greenan et al , 2018 ; Li et al , 2019 ) . 115 Whether the central elements of the cartwheel , including the CID , are likewise 116 conserved beyond T . spp . is unclear . 117 Considering that the earlier work in T . spp . was conducted without direct 118 electron detector and that software improvements have occurred since , we sought to 119 achieve a higher resolution map of the T . spp . cartwheel - bearing region . Moreover , to 120 explore the evolutionarily conservation of cartwheel architecture , we investigated two 121 5 other flagellates living in the gut of termites that might likewise harbor long cartwheels 122 well suited for STA . 123 124 125 Results 126 127 Exceptionally long cartwheel region in Trichonympha agilis 128 We set out to obtain a high - resolution STA map of the native cartwheel in T . spp . 129 Moreover , we likewise aimed at investigating Trichonympha agilis , a symbiotic 130 flagellate that lives in the gut of the Japanese termite Reticulitermes speratus ( Ohkuma 131 and Kudo , 1998 ) . This choice was guided by the fact that transcriptomic and genomic 132 information is being assembled in T . agilis ( Yuichi Hongoh , Tokyo Institute of 133 Technology , Japan , personal communication ) , which will be instrumental to map 134 proteins onto the STA map should sub - nanometer resolution be reached in the future . 135 As shown in Fig . 1A , T . agilis bears a large number of flagella , which stem from 136 similarly numerous centrioles inserted below the plasma membrane ( Kubai , 1973 ) . 137 Many of these flagella are tightly packed in a region called the rostrum located at the 138 cell anterior ( Fig . 1A , arrow ) . To determine the length of the cartwheel - bearing region 139 of T . agilis centrioles , cells were resin - embedded and analyzed by transmission 140 electron microscopy ( TEM ) , focusing on the rostral region . Longitudinal views 141 established that the cartwheel - bearing region is ~ 2 . 3 µ m in length on average ( Fig . 1B , 142 pink line ; SD = 0 . 36 µ m , N = 8 ) . This is less than the ~ 4 µ m observed in T . spp . ( Guichard 143 et al , 2013 ) , yet over 20 times the size of the ~ 100 nm cartwheel in centrioles of most 144 systems , including Chlamydomonas reinhardtii and Homo sapiens ( Guichard et al , 145 2017 , 2010 ; O’Toole and Dutcher , 2014 ) . In addition , we found that the centriole distal 146 region devoid of cartwheel is ~ 0 . 4 µ m in T . agilis ( Fig . 1B , white line ; SD = 0 . 07 µ m , 147 N = 6 ) , similar to its dimensions in other systems ( Guichard et al , 2013 ; Le Guennec et 148 al , 2020 ) . 149 We also analyzed transverse sections of resin - embedded T . agilis centrioles 150 using TEM . As shown in Fig . 1C , we found the characteristic features of the cartwheel - 151 bearing region , including a central hub from which emanate 9 spokes that extend 152 towards peripheral microtubule triplets . In addition , we noted the presence of the 153 pinhead and the A - C linker , as well as of the triplet base connecting these two elements 154 ( Gibbons and Grimstone , 1960 ; Vorobjev and Chentsov , 1980 ) , which is more 155 apparent in the circularized and symmetrized image ( Fig . 1C ) . 156 6 Overall , given the presence of a long cartwheel - bearing region , we conclude 157 that T . agilis also provides a suitable system to investigate the architecture of the 158 proximal part of the centriole using cryo - ET and STA . 159 160 Novel features revealed by improved STA of Trichonympha centrioles 161 Using a direct electron detector , we acquired tilt series of purified T . spp . centrioles , 162 focusing on the proximal cartwheel - bearing region ( Fig . S1A , S1B ) , followed by 163 tomogram reconstruction and STA ( Fig . 2A - E ; for all datasets see Fig . S1C - F for raw 164 tomograms , as well as Fig . S1G - J and Table S1 for processing pipeline ) . For the 165 central cartwheel , we achieved a local resolution ranging from ~ 16 Å to ~ 40 Å ( Fig . 166 S2A ) , with a global resolution of ~ 24 Å ( FSC criterion 0 . 143 ; Fig . S2B ; see Fig . S3 - S5 167 for resolution of all other STA maps , which have been deposited in EMDB ) . 168 Using line scans on 2D projections of the STA map , we determined the T . spp . 169 hub diameter to be 23 nm ( Fig . 2A , 2B ) , in line with previous work ( Guichard et al , 170 2013 , 2012 ) . Importantly , the improved resolution achieved here enabled us to uncover 171 novel features in the central cartwheel of the T . spp . centriole . Of particular interest , 172 we discovered that the position of the CID is polarized along the proximal - distal 173 centriole axis with respect to the hub and the spokes that emanate from it ( Fig . 2C , 174 2D ) . This is apparent from vertical intensity profiles of longitudinal views , which show 175 that the CID is positioned distal to the center of the hub density , which itself appears 176 to be elongated in the vertical direction ( Fig . 2C ) . Occasionally , two units can be 177 discerned within one hub density ( Fig . 2C , double peaks in light blue intensity profile 178 and corresponding dashed lines ) , a point that will be considered further below . Such 179 double units were not recognized previously , presumably owing to the lower resolution 180 map ( Guichard et al , 2013 , 2012 ) . Moreover , we found densities that vertically bridge 181 successive hub elements ( Fig . 2D , arrows ) . The polarized location of the CID unveiled 182 here is also apparent with respect to where spoke densities emerge from the hub ( Fig . 183 2E , arrow ) . In addition , we identified discontinuous densities in the center of the CID 184 ( Fig . 2B , 2D ) . 185 We likewise analyzed the central cartwheel in T . agilis observing variations 186 along the centriole axis in 2D longitudinal views ( Fig . S1D ) . Focused 3D classification 187 of sub - volumes indeed uncovered two classes , corresponding to 55 % and 45 % of 188 sub - volumes , which can occur within the same centriole ( Fig . 2F - J ; Fig . S6A - F ) . We 189 found that the central cartwheel STA map of both T . agilis classes exhibits many 190 similarities with that of T . spp . Thus , the CID is present , and spoke densities emanate 191 from a hub ~ 23 nm in diameter ( Fig . 2F , 2G ; Fig . S6A , S6B ) . Moreover , vertical 192 densities bridging successive hub elements are also present in both T . agilis classes 193 7 ( Fig . 2I , arrows ; Fig . S6D , arrows ) . Furthermore , we found that the CID is also 194 polarized along the proximal - distal centriole axis , being distal with respect to the hub 195 in both T . agilis classes , as evidenced from vertical intensity profiles ( Fig . 2H - I ; Fig . 196 S6C - D ) , as well as from the location of the CID relative to where spoke densities 197 emerge from the hub ( Fig . 2J and Fig . S6D , arrowheads ) . In the T . agilis 55 % class , 198 like in T . spp . , hub densities are elongated in the vertical direction and can be 199 sometimes discerned as two units ( Fig . 2H , double peaks in light pink intensity profile 200 and corresponding dashed lines ) . The presence of such double hub units next to the 201 CID is more apparent in the 45 % class , where they also exhibit a slight offset relative 202 to the vertical axis ( Fig . S6C , white dashed line ) , a point considered further below . In 203 addition , we found in this class that double hub units alternate with single hub units 204 that do not have a CID in their vicinity ( Fig . S6C , dashed arrow , Fig . S6D ) , an absence 205 noticeable also in raw tomograms ( Fig . S1D ) and verified in 3D top views ( Fig . S6A 206 and S6B , full circle ) . Moreover , we found that spoke densities emanating from single 207 hub units are thinner than those stemming from double hub units ( Fig . S6D ) . The 208 plausible origin of alternating double and single hub units in the 45 % T . agilis class 209 will be considered below . We noted also that the 45 % sub - volumes exhibit slight 210 variations in the spacing between double and single hub units ( Fig . S6E , 25 % and 20 % 211 sub - classes ) . 212 Taken together , our findings establish that T . spp . and T . agilis central 213 cartwheel architecture shares many features , including a polarized CID position . 214 215 Comparative STA of peripheral centriole elements in Trichonympha 216 We also investigated peripheral elements in the proximal region of T . spp . and T . agilis 217 centrioles . To this end , we extracted peripheral sub - volumes from the tomograms and 218 generated for each species three maps using STA centered either on the microtubule 219 triplet , the pinhead or the A - C linker ( Fig . 3A , 3E ) . 220 For the microtubule triplet , the resulting analysis revealed a characteristic 221 centriolar architecture . Thus , the A - microtubule bears 13 protofilaments , the B - 222 microtubule 10 protofilaments proper , with 3 extra ones shared with the A - microtubule , 223 whereas the C - microtubule exhibits a similar organization as the B - microtubule ( Fig . 224 3B , 3F ) . Moreover , we detected prominent densities corresponding to microtubule 225 inner proteins ( MIPs ) . In both species , we found a MIP located along protofilament A9 , 226 close to protofilament A10 ( Fig . 3B , 3F , empty arrowhead ) . A MIP was discovered at 227 this location in ciliary axonemes ( Nicastro et al , 2006 ) , and was also observed in 228 centrioles of T . spp . , Chlamydomonas and mammalian cells ( Greenan et al , 2020 , 229 2018 ; Guichard et al , 2013 ; Li et al , 2019 ) . In addition , we observed a MIP along 230 8 protofilament A5 in T . spp . ( Fig . 3B , chevron ) , which is visible only at low threshold in 231 T . agilis and positioned like MIP1 in Chlamydomonas centrioles and axonemes ( Li et 232 al , 2019 ; Nicastro et al , 2006 ) . In both Trichonympha species , we also detected 233 additional densities or microtubule associated proteins ( MAPs ) on the microtubule 234 exterior between the A - B and B - C inner junctions ( Fig . 3B , 3F , arrowhead ) , as 235 described also in Chlamydomonas and mammalian centrioles ( Greenan et al , 2018 ; Li 236 et al , 2019 ) . 237 For both T . spp . and T . agilis , we next conducted STA centered on the pinhead 238 or the A - C linker to uncover features in these peripheral elements . We thus found that 239 the pinhead connects to the A3 protofilament in both species ( Fig . 3C , 3G ) , in line with 240 previous observations in T . spp . and in other systems ( Greenan et al , 2018 ; Guichard 241 et al , 2013 ; Li et al , 2019 ) . Moreover , we found that the pinhead is polarized in a similar 242 manner in T . spp . and T . agilis ( Fig . 3C , 3G ) , with the two pinfeet moieties PinF1 and 243 PinF2 pointing proximally from the A3 protofilament , as reported previously for T . spp . 244 ( Guichard et al , 2013 , 2020 ) . We found also that the spacing between PinF1 and PinF2 245 elements is ~ 8 . 6 nm and ~ 7 . 9 nm in T . spp . , whereas it is ~ 8 . 4 nm and ~ 8 . 4 nm in T . 246 agilis ( Fig . 3C , 3G ) , compatible with power spectra of 2D class averages considering 247 the standard deviation of the measurements ( Fig . S2I , S3L ) . 248 Similarities between the two species are also apparent in the A - C linker that 249 bridges neighboring MT triplets . Thus , we found that the T . spp . A - C linker connects 250 protofilament A8 / A9 from one triplet with protofilaments C9 / C10 of the adjacent triplet 251 ( Fig . 3D ) , furthering the mapping of these connections compared to previous work 252 ( Guichard et al , 2013 ) . As shown in Fig . 3H , we found that the A - C linker in T . agilis 253 has a similar architecture . To generate an overview of the entire proximal region of the 254 T . agilis centriole , comprising hub , spoke , pinhead and A - microtubule , we conducted 255 STA centered on the spokes for both 55 % and 45 % classes ( Fig . S7A - H ) . Such maps 256 have a lower resolution due to the larger box size , the binning and the absence of 257 radial symmetrization , but nevertheless uncover a concerted proximal - distal polarity of 258 several elements . First , the CID is positioned distally within double hub units . Second , 259 spoke densities exhibit a slightly asymmetry in their tilt angle , with the proximal spoke 260 being more tilted . Moreover , these two polarized features present centrally are 261 connected with the polarized pinhead and A - C linker present peripherally ( Fig . S7C - D , 262 S7G - H ) . 263 Overall , we conclude that peripheral components of the cartwheel are also 264 generally conserved between the two Trichonympha species and exhibit concerted 265 polarity with central elements along the proximal - distal centriole axis . 266 267 9 Diversity of cartwheel architecture in Teranympha cartwheel 268 The gut of R . speratus termites contains another symbiotic flagellate , namely 269 Teranympha mirabilis ( Koidzumi , 1921 ; Noda et al , 2018 ) , which we found also to 270 harbor numerous centrioles and associated flagella ( Fig . S8A ) . TEM analysis of resin - 271 embedded specimens established that the cartwheel - bearing region in such centrioles 272 is ~ 1 . 1 µ m on average ( Fig . S8B , green line ; N = 9 , SD = 0 . 06 µ m ) , whereas the distal 273 region averages ~ 0 . 5 µ m ( Fig . S8B , black line ; N = 9 , SD = 0 . 08 µ m ) . Transverse 274 sections of the cartwheel - bearing region revealed the characteristic hub and spokes 275 connected through a pinhead to peripheral microtubule triplets , which are joined by an 276 A - C linker , whereas the triplet base is poorly visible ( Fig . S8C ) . 277 We conducted cryo - ET of purified T . mirabilis centrioles , acquiring tilt series of 278 the entire cartwheel - bearing region followed by tomogram reconstruction and STA . We 279 again generated separate maps for the central cartwheel , the microtubule triplet , the 280 pinhead and the A - C linker . Focused 3D classification of the central cartwheel region 281 yielded two classes representing 64 % and 36 % of sub - volumes , which can occur 282 within the same centriole ( Fig . 4A - G ; Fig . S8E ) . Similarly to the two Trichonympha 283 species , a hub ~ 23 nm in diameter is present in both T . mirabilis classes , with spoke 284 densities emanating from it ( Fig . 4A - C ) . Interestingly , we found that the architecture of 285 the central cartwheel in T . mirabilis differs slightly from that in Trichonympha . Indeed , 286 in both T . mirabilis classes , we uncovered a filamentous structure ~ 7 nm in diameter 287 present along the entire cartwheel - bearing region inside the hub , which we dubbed 288 filamentous cartwheel inner density ( fCID ) ( Fig . 4A - G ) . The fCID is also apparent in 289 transverse views of resin - embedded centrioles and symmetrized tomogram slices ( Fig . 290 S8C , S8D ) . Moreover , the fCID is consistently detected in raw tomograms ( Fig . S1E ) , 291 as well as in non - symmetrized STA comprising larger volumes centered on the spokes 292 ( Fig . S7I - P ) . 293 As shown in Figure 4D - G , the two central cartwheel T . mirabilis classes differ 294 in hub architecture , as revealed by vertical line profile intensity measurements . In the 295 64 % class , a periodicity of ~ 8 . 4 nm is apparent between hub densities , each consisting 296 seemingly of a single vertically elongated unit ( Fig . 4D , 4F ) . By contrast , in the 36 % 297 class , each hub density comprises a double unit ( Fig . 4E , 4G ) . Such double hub units 298 exhibit a peak - to - peak distance of ~ 3 . 2 nm and are separated from the adjacent double 299 hub unit by ~ 5 . 2 nm ( Fig . 4E ) . The sum of the two distances , namely 8 . 4 nm is 300 equivalent to that observed in the 64 % class . Moreover , the periodicity at the level of 301 emerging spoke densities is likewise equivalent in the two classes ( Fig . 4F , 4G ) . 302 Furthermore , as in the 45 % T . agilis class , we observed that every other double hub 303 10 unit in the 36 % T . mirabilis class exhibits a slight offset relative to the vertical axis ( Fig . 304 4E , white dashed lines ) . 305 Analysis of the peripheral STA microtubule triplet map of the T . mirabilis 306 centriole uncovered the canonical protofilament configuration for A - , B - and C - 307 microtubules ( Fig . 4H , 4I ) . As in Trichonympha , we detected additional densities on 308 the external side of the microtubules between A - B and B - C inner junctions ( Fig . 4I , 309 arrowheads ) . The previously described C - stretch that extends from protofilament C1 310 in T . spp . ( Guichard et al , 2013 ) is also observed in T . mirabilis ( Fig . 4I , chevron ) , while 311 prominent MIPs are not detected at the selected threshold . 312 The map generated by centering on the pinhead revealed a connection with 313 the A3 protofilament , with PinF1 and PinF2 being separated by 8 . 4 nm ( Fig . 4J ) , as 314 suggested also by power spectra of 2D class averages ( Fig . S4L ) . Moreover , we found 315 that the T . mirabilis A - C linker architecture is similar to that in the two Trichonympha 316 species , with anchoring to microtubules at protofilaments A9 and C9 / C10 ( Fig . 4K ) . 317 Overall , these findings indicate that T . mirabilis peripheral elements share 318 many conserved features with those in Trichonympha , as does the central cartwheel , 319 apart from the discontinuous CID being replaced by the seemingly continuous fCID in 320 T . mirabilis . 321 322 Comparative analysis of SAS - 6 ring stacking mode 323 We set out to investigate how SAS - 6 rings may fit into the central cartwheel STA maps 324 of the three species . Genomic or transcriptomic information is not available for T . spp . 325 and T . mirabilis at present . Thus , in the absence of molecular information for SAS - 6 in 326 T . spp . and T . mirabilis , as well as of structural information for T . agilis SAS - 6 , and 327 given that SAS - 6 crystal structures from a wide range of organisms exhibit extensive 328 similarities , we employed CrSAS - 6 as a model instead . We used homodimers of 329 CrSAS - 6 [ 6HR ] , the longest CrSAS - 6 fragment with a determined crystal structure , 330 which can self - assemble into 9 - fold symmetrical rings ~ 23 nm in diameter and 4 . 2 nm 331 in height ( Kitagawa et al , 2011 ) . 332 Given the presence of two units per hub density in the T . mirabilis 36 % class , 333 and also probably in both Trichonympha species , we computationally assembled 334 CrSAS - 6 [ 6HR ] into single rings , as well as into directly superimposed double rings in 335 register . We then used rigid body fitting to score the goodness of fit of both types of 336 ring assemblies with the STA maps ( Table S2 ) . Starting with the T . mirabilis 36 % 337 class , we found that whereas a single CrSAS - 6 [ 6HR ] ring fits in this map , additional 338 hub densities remain unaccounted for in this case ( Fig . 5A , arrowheads ) . By contrast , 339 the double ring configuration fills the hub density to a larger extent ( Fig . 5B ) , yielding 340 11 a better overlap score ( Table S2 ) . However , coiled - coil moieties extend slightly outside 341 the spoke densities in this case , potentially because of coiled - coil bending in vivo or 342 species - specific features . Moreover , rigid body fitting of CrSAS - 6 [ 6HR ] homodimers 343 confirmed that two directly superimposed dimers provide a better fit for the T . mirabilis 344 36 % class , a fit that also revealed a slight offset between them ( Fig . S9A ) . Next , we 345 fitted computationally assembled single and double CrSAS - 6 [ 6HR ] rings in the T . 346 mirabilis 64 % class map , finding again that head domains of double rings can be 347 readily accommodated in the hub density , although one coiled - coil moiety clearly 348 extends outside spoke densities in this case ( Fig . S9B ) Taken together , these 349 observations are compatible with the possibility that SAS - 6 rings can directly stack on 350 top of one another in vivo . 351 To explore whether direct SAS - 6 ring stacking might occur also in 352 Trichonympha , we likewise performed fitting of computationally assembled CrSAS - 353 6 [ 6HR ] single and double rings . We found again that in all cases directly superimposed 354 double rings can readily be accommodated in the hub densities with improved 355 goodness of fit compared to single rings ( Fig . 5C , 5D ; Fig . S9C , S9D ; Table S2 ) . The 356 T . agilis 45 % class , which comprises alternating double and single hub elements , 357 presents a particularly interesting case . As anticipated , double rings could be placed 358 in the elongated hub density comprising two units , but a single ring could be 359 accommodated in the thinner individual hub unit ( Fig . 5C , 5D ; Table S2 ) . 360 Prompted by the slight offset observed between two SAS - 6 ring oligomers upon 361 fitting CrSAS - 6 [ 6HR ] homodimers in the T . mirabilis 36 % class ( see Fig . S9A ) , as well 362 as the structural complementarity between superimposed SAS - 6 rings following such 363 an offset , we computationally assembled a double ring with a ~ 6 . 5° rotational offset 364 imposed between ring pairs , which allowed the two rings to come closer to one another 365 vertically by 0 . 4 nm ( Fig . S10B , compare with Fig . S10A ; Materials and Methods ) . 366 Manual fitting in the T . mirabilis 36 % class map uncovered that such an offset double 367 ring can be readily accommodated in the hub ( Fig . S10C ) . A similar conclusion was 368 drawn from fitting the offset double ring into the double hub density of the T . agilis 45 369 % class ( Fig . S10D ) . Such a slight offset between superimposed SAS - 6 rings might 370 explain the hub offset observed in the longitudinal view of the central cartwheel STA 371 in both T . mirabilis 36 % class and T . agilis 45 % class ( see Fig . 4E and Fig . S6C , 372 white dashed lines ) . Furthermore , we reasoned that such an offset double ring might 373 likewise affect the connected spoke densities . If spoke densities correspond in reality 374 to two individual spokes slightly offset from one another , as expected from an offset 375 double ring configuration , but which cannot be identified as individual units due to the 376 resolution limit , then the shape of the spoke densities should be elliptical when viewed 377 12 end on . To investigate this possibility , we unwrapped the central cartwheel maps to 378 obtain a complete view at the level of the spokes ( Fig . S10E ) . Importantly , this 379 uncovered the expected elliptical shape , revealing spoke offset in the T . mirabilis 36 380 % class ( Fig . S10F ) , as well as in both T . agilis classes ( Fig . S10G ) and in T . spp . ( Fig . 381 S10H ) . By contrast , no spoke offset was apparent in the T . mirabilis 64 % class ( Fig . 382 S10F ) . 383 Overall , these findings lead us to propose that pairs of SAS - 6 rings directly 384 stack in the cellular context and can do so with an offset ( see Discussion ) . 385 We investigated further the possibility that SAS - 6 rings directly stack onto one 386 another using a cell free assay with a recombinant CrSAS - 6 protein containing the 387 globular head domain and the entire coiled - coil ( referred to as CrSAS - 6 [ NL ] ) . It has 388 been shown previously that CrSAS - 6 [ NL ] bearing 6xHis - and S - tag moieties can self - 389 organize into stacks thought to exhibit ~ 8 . 5 nm periodicity between spokes based on 390 the analysis strictly of top views ( Guichard et al , 2017 ) . We investigated this question 391 anew , now analyzing stacking periodicity of untagged purified CrSAS - 6 [ NL ] protein . 392 Moreover , we conducted STA of side views of stacked assemblies , thus allowing us to 393 measure the underlying periodicities with a global resolution of ~ 25 Å ( Fig . 5E - I , Fig . 394 S5 ) . Importantly , we found that such in vitro assembled CrSAS - 6 [ NL ] stacks exhibit a 395 periodicity of ~ 4 . 5 nm that can each accommodate one computationally assembled 396 single ring of CrSAS - 6 [ 6HR ] ( Fig . 5G - I ) . Moreover , we found that the spokes of 397 successive in vitro assembled rings are almost in register ( Fig . S10I ) . 398 Taken together , these observations lead us to propose that direct stacking of 399 SAS - 6 rings may be an evolutionarily conserved feature of cartwheel assembly at the 400 onset of centriole biogenesis . 401 402 403 Discussion 404 405 The centriole is fundamental for numerous aspects of cell physiology and shows 406 remarkable conservation across the major eukaryotic groups of life . Understanding the 407 mechanisms that govern centriole assembly requires not only the identification of 408 essential building blocks and their regulators , but also a detailed knowledge of 409 organelle architecture . Our findings uncover conserved features in the cartwheel - 410 bearing portion of the centriole . We find that the CID located within the hub is polarized 411 along the proximal - distal centriole axis in both Trichonympha species and that a 412 potentially related fCID is present in T . mirabilis . Moreover , we establish that hub 413 densities can be fitted by directly superimposing 9 - fold symmetrical SAS - 6 double rings 414 13 that exhibit a slight rotational offset between the two rings . Furthermore , we find that 415 such an offset double ring is stacked in register with the next offset double ring . Overall , 416 our work provides novel insights into the architecture of the centriolar cartwheel at the 417 root of centriole biogenesis . 418 419 Conserved features in cartwheel architecture 420 Centriole ultrastructure began to be explored in the 1950s when TEM of resin - 421 embedded fixed specimens revealed its signature 9 - fold radial symmetry and the chiral 422 arrangement of microtubule triplets ( Fawcett & Porter , 1954 ) . More recent cryo - ET 423 analysis uncovered the native architecture of the cartwheel - bearing portion of the 424 centriole ( Guichard et al , 2013 ; Li et al , 2019 ) . Here , we set out to improve the 425 resolution previously obtained with T . spp . and to investigate whether features 426 discovered in this species are present in other flagellates harboring unusually long 427 cartwheels . 428 In the absence of fossil record or complete sequence information , it is not 429 possible to assess sequence divergence between centriolar proteins in the species 430 analyzed here , nor to date with precision the evolutionary times separating them . 431 However , an approximation is provided in the case of T . spp . and T . agilis by the 432 phylogenetic divergence between their respective hosts Zootermopsis spp . and 433 Reticulitermes speratus , which are estimated to have shared their last common 434 ancestor ~ 140 million years ago ( Bucek et al , 2019 ) . Given that Trichonympha are 435 obligate symbiotic organisms and considering that the two host species inhabit 436 different continents ( Park et al , 2006 ; Thorne et al , 1993 ) , it is reasonable to postulate 437 that T . spp . and T . agilis diverged at least this long ago . Teranympha mirabilis belongs 438 to a different genus , and based on small subunit rRNA sequences it is much more 439 distant from the Trichonympha species than they are from one another ; however , the 440 split between the two genera is thus far undated ( Carpenter and Keeling , 2007 ; Noda 441 et al , 2018 , 2009 ; Ohkuma et al , 2009 ) . Regardless of the exact evolutionary times 442 separating these flagellates , our findings , together with those of a companion 443 manuscript ( Klena et al ) , establish that the CID is conserved between distant 444 eukaryotic groups , suggesting shared evolutionary history and / or function ( Fig . 6A ) . 445 The CID exhibits a 9 - fold radial symmetry and connects with the hub approximately 446 where neighboring SAS - 6 homodimers interact with one another ( Fig . 6B ) , a 447 suggestive location raising the possibility that the CID imparts or maintains the 9 - fold 448 symmetrical SAS - 6 ring structure ( Guichard et al , 2013 ) . We discovered here that the 449 CID is polarized along the proximal - distal centriole axis , so that may also play a role in 450 imparting or maintaining organelle polarity ( Fig . 6C ) . Furthermore , the CID exhibits an 451 14 ~ 8 . 4 nm periodicity along the proximal - distal centriole axis , with no apparent continuity 452 between two superimposed CID elements . This is in contrast to the fCID , which runs 453 throughout the center of the cartwheel - bearing portion of the T . mirabilis centriole . 454 Although the fCID seems disconnected from the hub , small elements linking the two 455 can be discerned in the T . mirabilis 36 % class ( Fig . 6D , see Fig . S7N ) . Moreover , the 456 fCID might be linked with the hub through other protein segments that are small , 457 flexible or exhibit different periodicities , rendering them difficult to detect in the current 458 STA map . It will be interesting also to address whether the fCID exhibits helical 459 features . Elements related to the fCID may be present in systems other than T . 460 mirabilis . Thus , electron - dense material is discernable in the geometrical center of the 461 hub in T . spp . and less so in T . agilis ( see Fig . 2D and 2I ) , although it is discontinuous 462 and smaller in diameter than the fCID . Moreover , a density that may be related to either 463 CID or fCID is apparent inside the hub in other species , including Chlamydomonas , 464 Chytrid and several insects ( Geimer & Melkonian , 2004 ; Guichard et al , 2017 ; O’Toole 465 & Dutcher , 2014 ; Olson & Fuller , 1968 ; McNitt 1974 ; Gottardo et al , 2015 ; Uzbekov et 466 al , 2018 , Klena et al ) ( Fig . 6A ) . Determining the molecular nature of CID and fCID in 467 diverse systems is expected to help assess whether they share an evolutionary history 468 and may serve a similar function . 469 We found conserved features also in the peripheral elements of the cartwheel - 470 bearing region of the centriole . Thus , the pinhead is present in all three organisms 471 analyzed here , as is the A - C linker , although the exact arrangement of moieties within 472 these elements differs slightly . Together with the peripheral STA map in the proximal 473 region of the Chlamydomonas centriole ( Li et al , 2019 ) , these findings indicate that 474 peripheral elements have been significantly conserved across evolution . Slight 475 differences between systems are observed notably in the microtubule triplets and 476 prominent MIPs . At the resolution achieved here , we thus identified a prominent MIP 477 in both Trichonympha species next to protofilament A9 , close to protofilament A10 , 478 potentially corresponding to the A - microtubule seam ( Ichikawa et al , 2017 , 2019 ) . A 479 prominent MIP has also been observed at this position in centrioles of mammalian 480 CHO cells and Drosophila S2 cells ( Greenan et al , 2020 , 2018 ; Li et al , 2019 ) . The 481 molecular identity of several Chlamydomonas MIPs present in ciliary axonemal 482 microtubules was identified recently ( Ma et al , 2019 ) , but corresponding MIPs could 483 not be detected here using the selected threshold , perhaps because not all 484 periodicities were explored . Nevertheless , we identified one additional prominent MIP 485 next to protofilament A5 in T . spp . , which appears to be conserved in metazoan 486 centrioles ( Greenan et al , 2018 ) , and was identified as RIB72 and FAP252 in ciliary 487 axonemal microtubules ( Ma et al , 2019 ) . Given the recent spectacular progress in 488 15 assigning the molecular nature of many ciliary MIPs using high resolution cryo - EM ( Ma 489 et al , 2019 , Song et al , 2020 ) , it is likely that better resolution STA maps combined with 490 a search for different periodicities will also enable the efficient identification of centriolar 491 MIPs . 492 493 Root of proximal - distal polarity 494 The centriole is polarized along its proximal - distal axis , as evidenced for instance by 495 the presence of the cartwheel and microtubule triplets in the proximal region versus 496 microtubule doublets in the distal region . Likewise , distal and sub - distal appendages 497 are restricted to the distal region , by definition . Ultimately , proximal - distal polarity of 498 the centriole also translates into that of the axoneme that emanates from it . Prior work 499 in T . spp . raised the possibility that centriole polarity might stem from the pinhead or 500 the A - C linker , because both elements exhibit a clear proximal - distal polarity ( Guichard 501 et al , 2013 ) . Whereas it remains plausible that polarity is imparted by peripheral 502 elements in some settings , in light of the present findings an alternative view can be 503 envisaged . Indeed , the offset between double SAS - 6 rings generates an inherently 504 asymmetric structure , which correlates with the polarized location of the CID , as well 505 as with the asymmetric tilt angle of spoke densities that is also observed in distantly 506 related eukaryotic groups ( Klena et al ) . Given that central elements are present before 507 peripheral ones during organelle assembly in the canonical centriole duplication cycle , 508 it is tempting to speculate that the asymmetric properties of offset double rings and / or 509 the CID are key for imparting organelle polarity . The situation might differ during de 510 novo centriole assembly , which in human cells does not require interactions between 511 HsSAS - 6 head domains ( Wang et al , 2015 ) . In this case , peripheral elements are 512 thought to be more critical , and this may also be the case when it comes to imparting 513 organelle polarity . 514 515 Working model of SAS - 6 ring stacking 516 What are the mechanisms of SAS - 6 ring stacking ? The previous lower 517 resolution T . spp . map ( Guichard et al , 2013 ) , together with an initial analysis of in vitro 518 generated stacks of CrSAS - 6 [ NL ] ( Guichard et al , 2017 ) , led to the suggestion that the 519 periodicity between subsequent SAS - 6 ring oligomers was ~ 8 . 5 nm . The higher 520 resolution analysis performed here in three species leads us to propose instead that 521 SAS - 6 rings stack directly on top of one another , thus doubling the number of SAS - 6 522 molecules per vertical unit length . Although we cannot formally exclude that a single 523 SAS - 6 ring is present and that additional densities correspond to other unassigned 524 protein ( s ) , several pieces of evidence support the direct SAS - 6 stacking scenario . First , 525 16 two SAS - 6 rings fit without clashes within vertically elongated hub densities in 526 Trichonympha and T . mirabilis . Second , the T . agilis 45 % class contains thin hub 527 densities that can accommodate only one SAS - 6 ring , reinforcing the notion that the 528 thicker ones harbor two . Third , stacks reconstituted in a cell free assay from purified 529 CrSAS - 6 exhibit direct ring superimposition . 530 Our findings taken together lead us to propose the following working model of 531 SAS - 6 stacking in the cellular context ( Fig . 6E , 6F ) . This model entails two modes of 532 ring stacking . The first corresponds to the offset double ring configuration , with a tight 533 packing of SAS - 6 rings , which are ~ 3 . 2 nm apart for instance in the T . mirabilis 36 % 534 class . In line with our findings , directly superimposed hub densities with remarkably 535 similar spacing are observed also in Paramecium and C . reinhardtii ( Klena et al ) . In 536 the second stacking mode , two such offset double rings stack in register , being ~ 5 . 2 537 nm apart from each other in the T . mirabilis 36 % class . A similar alternating pattern is 538 observed in T . spp . and T . agilis , albeit with slightly different distances , potentially 539 owing to limitations in measurement precision . Overall , we propose that these two 540 alternating modes of SAS - 6 ring stacking represent a fundamental feature of cartwheel 541 assembly . 542 The periodicity of the basic central cartwheel unit in this working model is ~ 16 . 8 nm , 543 which corresponds to the largest conserved periodicity detected by FFT . This value is 544 consistent with the distance expected from an alternation of the two modes of SAS - 6 545 stacking ( 2x ~ 3 . 2 nm for the rings with an offset + 2x ~ 5 . 2 nm for the rings in register 546 in the case of the T . mirabilis 36 % class = ~ 16 . 8 nm ) ( Fig . 6F ) . Each SAS - 6 hub 547 element possesses one spoke emerging from it , but owing to insufficient resolution , 548 two closely superimposed spokes appear as one joint density in all cases , except for 549 the single hub element in the T . agilis 45 % class . Interestingly , we note also that hub 550 elements with accompanying spoke densities are further apart in the T . mirabilis 64 % 551 class , indicative of a potential variation in stacking mode . Perhaps the densities 552 bridging successive hub elements in this case correspond to another protein than SAS - 553 6 . One intriguing candidate to consider in this context is SAS - 6 - like , an evolutionary 554 ancient SAS - 6 related protein lacking the coiled - coil ( de Leon et al , 2013 ) . 555 556 Concluding remarks 557 In summary , conducting cryo - ET in three flagellates enabled us to uncover 558 important conservation of centriolar architecture , as well as some species specificity . 559 In the future , higher resolution analyses , including without symmetrizing , is expected 560 to reveal additional polarity and stacking features , as well as to enable one to visualize 561 17 protein domains and thus further unravel the mechanisms governing centriole 562 assembly . 563 564 565 Materials and methods 566 567 Transmission electron microscopy of Trichonympha and Teranympha cells 568 T . agilis and T . mirabilis were extracted from the hindgut of Reticulitermes speratus 569 termites ( a kind gift of Yuichi Hongoh , Tokyo University of Technology , Japan ) , as 570 described ( Guichard et al , 2015 ) , placed in a drop of 10 mM K - PIPES , and gravity - 571 sedimented for 3 - 5 min onto coverslips coated with poly - D lysine ( Sigma - Aldrich , 572 catalog number P1024 ) , until cells became in contact with the glass . Cells were fixed 573 overnight in 2 % paraformaldehyde 1 % glutaraldehyde in phosphate buffer 0 . 1 M , pH 574 7 . 4 ( with 0 . 2 % Tween - 20 for some samples ) , washed in cacodylate buffer ( 0 . 1 M , pH 575 7 . 4 ) at 4°C and post - fixed with 0 . 5 % tannic acid in cacodylate buffer ( 0 . 1 M , pH 7 . 4 ) 576 for 40 min at room temperature . After two washes in distilled water , cells were post - 577 fixed in 1 % osmium tetroxide in cacodylate buffer . Thereafter , samples were washed 578 in distilled water and dehydrated in a graded ethanol series ( 1x 50 % , 1x 70 % , 2x 96 579 % , 2x 100 % ) . Finally , samples were embedded in epon resin and polymerized at 65°C 580 overnight . 50 nm sections were cut and then stained in 1 % uranyl acetate in distilled 581 water for 10 min , followed by Reynold’s stain ( 1 . 33 g lead citrate , 1 . 76 g sodium citrate , 582 160 mM NaOH in 50 ml distilled water ( Reynolds , 1963 ) ) for 5 min . Images were 583 acquired using Tecnai Spirit at 80 kV or Tecnai F20 at 200 kV microscopes ( Thermo 584 Fischer Scientific ) . Pear - shaped T . agilis cells with long flagella covering the anterior 585 of the cell can be readily distinguished from elongated T . mirabilis cells with shorter 586 flagella arranged in rows spiraling around the cell . These distinct morphologies are 587 preserved in resin . Purified centrioles from the two species were distinguished in cryo - 588 ET based on the presence or absence of the fCID . Images of centriole cross - sections 589 from TEM and cryo - ET were circularized and symmetrized using the CentrioleJ plugin 590 ( Guichard et al , 2013 ) . 591 592 Purification of Trichonympha and Teranympha centrioles 593 Trichonympha and Teranympha cells were extracted from the hindgut of termites in 10 594 mM K - PIPES in the presence of cOmplete protease inhibitor cocktail ( 1 : 1000 ; Roche ) 595 as described ( Guichard et al , 2015 ) , with minor species - specific modifications . For the 596 purification of centrioles from mixed T . agilis and T . mirabilis populations from R . 597 speratus termites , cells were pelleted at 1000 g for 20 s and the supernatant discarded . 598 18 For enrichment of T . mirabilis centrioles , cells were sedimented on ice 3x10 min in 1 599 ml 10 mM K - PIPES ; the more fragile T . agilis cells spontaneously lysed during this 600 step and thus depleted from the resulting pellet . T . spp . ( T . collaris , T . campanula and 601 T . sphaerica ) were extracted from Zootermopsis angusticollis termites ( a kind gift from 602 Filip Husnik and Patrick Keeling , University of British Columbia , Vancouver , Canada ) , 603 which harbor the same set of Trichonympha species as Zootermopsis nevadensis 604 ( Boscaro et al , 2017 ; Tai et al , 2013 ) . After extraction , T . spp . cells were sedimented 605 on ice 3x10 min in 1 ml 10 mM K - PIPES . In all cases , centrioles were released during 606 a 20 - 40 min incubation on ice in 1 ml K - PIPES 10 mM 0 . 5 % NP - 40 plus protease 607 inhibitors . Centrioles and associated flagella were either pelleted at 500 g for 3 min or 608 the resulting supernatant spun at 1000 g for 5 min at 4° C ; both types of preparations 609 were utilized for cryo - ET . In all cases , centrioles were washed once in 1 ml 10 mM K - 610 PIPES plus protease inhibitors and pelleted at 1000 g for 5 min . The pellet was stored 611 on ice before preparing grids for cryo - ET . 612 613 Cryo - ET data acquisition 614 Grids were prepared as described previously ( Guichard et al , 2015 ) . Briefly , 4 μl of the 615 purified centriole solution were mixed with 10 nm gold beads using a 20 μl tip cut at 616 the extremity and then applied to the copper side of a glow discharged lacey holey 617 carbon grid ( carbon Cu 300 mesh , Ted Pella ) . The grid was then manually blotted from 618 the carbon side with a Whatman grade 1 filter paper and plunged into liquid ethane at 619 - 180° C with a manual plunge freezer . 620 Tilt series of centrioles that were approximately parallel to the tilt axis were 621 collected with the Tomo 4 . 0 software ( ThermoFisher Scientific ) on a Tecnai F20 TEM 622 operated at 200 kV ( ThermoFisher Scientific ) ; data were recorded with a Falcon III DD 623 camera ( ThermoFisher Scientific ) in linear mode at 29’000X magnification , 624 corresponding to a pixel size of 3 . 49 Å . The tilt series were recorded from −60° to 60° 625 using both continuous and bi - directional tilt schemes , with an increment of 2° at 2 . 5 – 626 5 µm underfocus . To allow determination of the proximal - distal axis , the blunt proximal 627 end was included in image acquisition . A total of 54 tilt series were collected ( Table 628 S1 ) . 629 Tilt series alignments using gold fiducials were performed with IMOD v4 . 9 630 ( Kremer et al , 1996 ) . The contrast transfer function ( CTF ) was estimated with 631 CTFFIND4 v1 . 8 ( Rohou and Grigorieff , 2015 ) using relion _ prepare _ subtomograms . py 632 ( Bharat and Scheres , 2016 ) . Variable SkipCTFCorrection was set to ‘True ’ to generate 633 a wedge - shaped tilt - and dose - weighted 3D CTF model . Aligned tilt series were 634 19 corrected by phase - flipping with ctfphaseflip and used for weighted back - projection 635 ( WBP ) tomogram reconstruction with IMOD . 636 637 Sub - tomogram processing 638 Areas corresponding to the central cartwheel region ( CW ) and to the peripheral 639 microtubule triplet region ( MTT ) were identified visually in individual tomograms and 640 their longitudinal axes modeled with closed contours in 3dmod . Proximal - distal polarity 641 of the centriole was assessed by the fact that the proximal end was blunt , in contrast 642 to the distal end , where the flagellum was usually present . Moreover , the clockwise 643 chirality of the microtubule triplets when viewed from the distal end provided an 644 independent means to determine proximal - distal polarity , which was registered and 645 maintained in the resulting sub - volumes . Also , reconstructions from sub - volumes re - 646 centered on the spokes with CW and polar CID density on one side , and spokes and 647 polar pinhead on the other , allowed unambiguous polarity assignment ( see Fig . S7 ) . 648 Individual model points were interpolated along the contours using the 649 addModPts from the PEET package ( Heumann et al , 2011 ) , with a step size of 3 x 85 650 Å = 252 Å corresponding to a 73 pixels shift ( see Fig . S1B ) . The interpolation step size 651 was selected to correspond to approximately three hub element , as smaller step sizes 652 could not separate cartwheel variations visible in the raw tomograms ( see Fig . S1D ) . 653 Larger step sizes were also explored to generate non - overlapping sub - volumes along 654 the proximal - distal axis of the cartwheel hub , shifted relative to each other by more 655 than half of the sub - volume size . The resulting STAs exhibited similar structural 656 features , yet were noisier due to the reduced number of sub - volumes . 657 Interpolation with a step size of 252 Å resulted in 1385 initial CW sub - volumes 658 for T . spp . , 1154 for T . agilis and 1802 for T . mirabilis . A similar approach and step size 659 was selected for extraction of MTT , resulting in 2792 initial MTT sub - volumes for T . 660 spp . , 3850 for T . agilis and 7928 for T . mirabilis ( see Fig . S1G - I ; Table S1 ) . All further 661 processing was performed with RELION v2 . 1 or v3 . 1 ( Kimanius et al , 2016 ) . Briefly , 662 2D projections of all sub - volumes along the Z axis were generated and subjected to 663 reference - free 2D classification . 2D class averages with clear CW or MTT densities 664 were selected for further data processing . At this step , the hub of both T . agilis and T . 665 mirabilis showed variations in vertical spoke and hub element spacing . Power spectra 666 of 2D class averages were used to measure the vertical spacing of the CW elements 667 ( see Fig . S2C , S3C , S3F , S4C , S4F ) or MTT ( see Fig . S2F , S3I , S7I ) . Modeling with 668 3dmod was also used to investigate the proximal - distal twist of microtubule blades . No 669 twist angle of the microtubules was observed along the proximal - distal axis of the 670 20 cartwheel - bearing region in either species , as previously reported for T . spp . ( Gibbons 671 and Grimstone , 1960 ) . 672 Selected 2D projections of CW sub - volumes were re - extracted , now as 3D sub - 673 volumes with a box size of 200 pixels ( corresponding to 80 nm , with the exception of 674 T . spp . , which corresponds to 88 nm ) and re - scaled to a pixel size of 4 Å . Initial 3D CW 675 and MTT references were built de novo with the Stochastic Gradient Descent ( SGD ) 676 algorithm from the re - extracted 3D sub - volumes . The resulting CW initial reference 677 was re - aligned with relion _ align _ symmetry to orient the visible nine - fold rotational ( C9 ) 678 axis along Z axis , and C9 symmetry applied . Similarly , the MTT initial reference 679 proximal - distal axis was re - aligned with the Z axis . Consensus 3D refinement of the 680 CW initial reference with C9 symmetry and missing wedge correction revealed a well 681 resolved CW density . Consensus 3D refinement was followed by focused 3D 682 refinement with a soft mask that included only the CW hub with the CID , as well as 683 part of the emanating spoke density , whose inclusion inside the mask was important 684 to avoid misalignment along the proximal - distal axis . Moreover , one round of focused 685 3D classification in T . agilis and T . mirabilis with a soft tight mask that included only 686 three central stacked rings revealed vertical spacing variations in 3D . 687 The T . agilis CW hub classified into two groups representing 55 % and 45 % of 688 refined sub - volumes . The 55 % class showed uniform 85 Å vertical spacing measured 689 from the 3D reconstruction in real space and from the power spectrum . Further 3D 690 classification did not yield any observable improvement in the quality of the maps as 691 judged by features and resolution . In contrast , further focused 3D classification without 692 alignment of the 45 % class revealed two sub - groups corresponding to 20 % and 25 % 693 of all refined sub - volumes , with differences in distances between alternating thin and 694 thick rings . The T . mirabilis CW hub classified into two main groups representing 64 % 695 and 36 % of refined sub - volumes . The 64 % class exhibited 85 Å vertical spacing 696 measured from the 3D reconstruction in real space and from power spectra and a 697 smooth CW hub surface . In contrast , the 36 % class revealed 85 Å vertical spacing 698 with two smaller units per hub with 42 Å spacing measured from the 3D reconstruction 699 in real space and from power spectrum . Further 3D classification did not yield any 700 observable improvement in the quality of the maps as judged by features and 701 resolution . For all 3D classes , final focused 3D refinements with local alignments were 702 performed . Analogous analyses of T . spp . revealed a single stable class with uniform 703 85 Å vertical spacing measured from the 3D reconstruction and from the power 704 spectrum . 705 To establish the polar relationship between the CID , the corresponding CW and 706 the peripheral elements ( A - microtubule , pinhead , A - C linker ) , the symmetry was 707 21 relaxed from C9 to C1 with relion _ particle _ symmetry _ expand . The resulting symmetry - 708 expanded sub - volumes were re - extracted with binning 3 and a box size of 100 pixels , 709 using the ‘re - center refined coordinates’ option approximately on the spokes center to 710 include CW density on one side and spokes and pinhead on the other . Focused 3D 711 classification of re - extracted sectors without alignment and without symmetry iterated 712 with focused 3D refinement revealed reconstructions with well - resolved connections 713 between central and peripheral elements . No structural features with a periodicity 714 different than a multiple of 85 Å were detected at this step . 715 A similar processing pipeline was applied to peripheral elements ( microtubules , 716 pinhead , A - C linker ) . Selected 2D projections of MTT sub - volumes were re - extracted 717 as binned 2 times 3D sub - volumes with a box size of 128 pixels corresponding to 89 718 nm to include microtubule triplet and emanating pinhead and A - C linker densities ( see 719 Fig . S1B ) . Initial 3D MTT references were built de novo with the SGD algorithm from 720 the re - extracted 3D sub - volumes . The resulting MTT initial reference proximal - distal 721 axis was re - aligned with the Z axis . Re - extracted MTT sub - volumes were subjected to 722 consensus 3D refinement followed by rounds of focused 3D classification and 3D 723 refinement . Refined sub - volumes were re - extracted either at the center of emanating 724 pinhead or A - C linker densities and refined separately . Local resolution distributions 725 were determined with ResMap ( Kucukelbir et al , 2014 ) and directional 3D FSC were 726 measured with 3DFSC web server ( Zi Tan et al , 2017 ) . 727 728 Rigid body fitting and computational assembly of SAS - 6 rings 729 Rigid - body fitting of CrSAS - 6 [ 6HR ] ( PDB - 3Q0X ) into T . spp . , T . agilis and T . mirabilis 730 3D STA maps was performed with ADP _ EM plugin ( Garzón et al , 2007 ) , followed by 731 symmetrical fitting in UCSF Chimera ( Pettersen et al , 2004 ) . 732 Single CrSAS - 6 [ 6HR ] rings were computationally assembled as described 733 ( Kitagawa et al , 2011 ) , while double rings were assembled by superposing two such 734 CrSAS - 6 [ 6HR ] rings using the crystallized stacks of LmSAS - 6 rings as guide ( Van 735 Breugel et al , 2011 ) . Offset double rings were assembled manually by rotating the 736 proximal ring clockwise , with the angle being estimated based on angles between fitted 737 homodimers . The vertical distance between rings in the double offset ring configuration 738 was reduced by 0 . 4 nm , and clashes were only allowed between flexible loops that are 739 not resolved in the crystal structure . CrSAS - 6 [ 6HR ] single ring , double in register rings 740 or double offset rings were fit in the map using fitmap function of ChimeraX ( Goddard 741 et al , 2018 ) . Coordinate models were initially converted to molecular maps at 742 corresponding global resolutions and then fitted into the STA maps . Calculated values 743 22 for correlation and overlap between maps are reported in Table S2 . UCSF Chimera or 744 ChimeraX were used for visualization and segmentation of 3D models . 745 746 In vitro assembly of CrSAS - 6 stacked cartwheels 747 A 6xHis - and S - tag - containing CrSAS - 6 construct harboring the N - terminal globular 748 head domain plus the entire coiled - coil ( referred to a CrSAS - 6 [ NL ] , comprising amino 749 acids 1 - 503 ) was expressed and purified as described ( Guichard et al , 2017 ) , except 750 that the tag moieties were removed by TEV protease cleavage . The resulting purified 751 CrSAS - 6 [ NL ] protein was dialyzed overnight from Tris pH 7 . 5 150 mM NaCl into 10 752 mM K - PIPES pH 7 . 2 using a mini dialysis unit at 4°C ( slide - A - lyzer , 3 . 5 K , 10 – 100 ml , 753 Pierce , catalogue number 69’550 ) . Thereafter , 5 µ l of dialyzed material was pipetted 754 onto a glow - discharged Lacey holey carbon grid ( Carbon Cu 300 mesh , Ted Pella ) , 755 blotted for 3 s ( blot force - 15 , no wait time ) and plunge - frozen by vitrification in liquid 756 nitrogen - cooled liquid ethane , using a Vitrobot MKIV ( ThermoFisher Scientific ) . 757 Tilt series were collected on a Titan Krios TEM ( ThermoFisher Scientific ) 758 operated at 300 kV and equipped with a Gatan Quantum LS energy filter ( zero loss slit 759 width 20 eV ; Gatan Inc . ) on a K2 Summit direct electron detector ( Gatan Inc . ) in 760 electron counting mode ( 110 e - / Å 2 total dose ) , at a calibrated pixel size of 2 . 71 Å . The 761 tilt series were recorded from −60° to 60° using a dose - symmetric tilt scheme ( Hagen 762 et al , 2017 ) with increments of 2° at 2 . 5 - 3 . 5 µm defocus . A total of 31 tilt series were 763 collected automatically using SerialEM ( Mastronarde , 2005 ) . 764 Tilt series alignment using gold fiducials was performed with IMOD v4 . 9 . The 765 CTF was estimated with CTFFIND4 v1 . 8 using relion _ prepare _ subtomograms . py . 766 Variable SkipCTFCorrection was set to ‘True’ to generate a wedge - shaped tilt - and 767 dose - weighted 3D CTF model . Aligned tilt series were corrected by phase - flipping with 768 ctfphaseflip and used for WBP tomograms reconstruction with IMOD . Further STA was 769 performed as described above for Trichonympha centrioles . Briefly , side views of 770 CrSAS - 6 [ NL ] stacks were identified visually in tomograms and their longitudinal axes 771 modeled with closed contours in 3dmod . Next , individual model points were added 772 along the contours using addModPts with a step size of 2 CrSAS - 6 [ NL ] rings ( 2 x 42 Å 773 = 84 Å ) to prevent misalignment of potentially merged spokes , resulting in 926 initial 774 sub - volumes ( see Fig . S1J ; Table S1 ) . Further processing was performed with RELION 775 v2 . 1 or v3 . 1 as described above . 776 777 778 Declarations 779 23 780 Ethics approval and consent to participate 781 Not applicable . 782 783 Consent for publication 784 Not applicable . 785 786 Availability of data and material 787 Sub - tomogram averages have been deposited at Electron Microscopy Data Bank 788 ( EMD - 10916 , EMD - 10918 , EMD - 10922 , EMD - 10923 , EMD - 10927 , EMD - 10928 , EMD - 789 10931 , EMD - 10932 , EMD - 10934 , EMD - 10935 , EMD - 10937 , EMD - 10938 , EMD - 10939 , 790 EMD - 10941 , EMD - 10942 ; Table S1 ) . 791 792 Competing interests 793 The authors declare that they have no competing interests . 794 795 Funding 796 This work was funded by grant from the European Research Council to PG ( AdG 797 340227 ) and the Swiss National Science Foundation ( SNSF ) to PaG 798 ( PP00P3 _ 157517 ) . 799 800 Authors ' contributions 801 SN , AB , GNH , VNV and DD collected and analyzed data . SN , AB , GNH , PG 802 contributed to experimental design and analysis , as well as writing the manuscript . 803 PaG and MLG contributed to data analysis . All authors read and approved the final 804 manuscript . 805 806 Acknowledgements 807 We are grateful to Yuichi Hongoh , Patrick Keeling and Filip Husnik for providing 808 termites . We thank Graham Knott and Stéphanie Rosset ( BioEM platform of the School 809 of Life Sciences , EPFL ) for assistance with TEM , as well as Kenneth Goddie , Lubomir 810 Kovacik and Henning Stahlberg ( Center of Nanoimaging , Biozentrum , Basel , 811 Switzerland ) with data collection on the Titan Krios . Niccolò Banterle , Graham Knott 812 and Fabian Schneider are acknowledged for their critical reading of the manuscript . 813 816 24 Figure legends 817 818 Figure 1 : Exceptionally long centriolar cartwheel in T . agilis 819 ( A ) Differential interference contrast micrograph of live T . agilis cell . The arrow points 820 to the cell anterior , where the rostrum is located ; arrowheads point to some of the 821 flagella . 822 ( B ) Transmission electron micrographs of T . agilis centriole embedded in resin - 823 longitudinal view ; the hub ( arrowhead ) is visible in the cartwheel - bearing region ( pink 824 line ) , but not in the distal region ( white line ) . 825 ( C ) Transmission electron micrographs of T . agilis centriole embedded in resin in 826 transverse view from distal end ( left ) and corresponding image circularized and 827 symmetrized with the CentrioleJ plugin ( middle ) , with schematic indicating principal 828 architectural elements ( right ) . 829 830 Figure 2 : Conserved architecture and polarity in T . spp . and T . agilis cartwheel 831 ( A , F ) Transverse 2D slices through STA of T . spp . ( A ) and T . agilis ( F ) central 832 cartwheel at indicated height , from proximal ( 0 nm ) to distal ( 8 . 7 nm ( A ) and 12 nm 833 ( F ) ) . The dashed box in the 5 . 2 nm ( A ) and 8 . 0 nm ( F ) slices indicate corresponding 834 regions shown in ( C , H ) . Schematic on top illustrates the area used to generate the 3D 835 maps of the central cartwheel . 836 ( B , G ) Transverse views of central cartwheel STA 3D map ; 9 spoke densities emanate 837 from the hub and the CID is present within the hub . The diameter of the hub is 23 . 0 ± 838 0 nm in T . spp . and 22 . 7 ± 0 . 2 in T . agilis ( both N = 3 ; here and thereafter in the figure 839 legends , N corresponds to 2D measurements from STA averages ) . An electron dense 840 structure is present inside the CID in T . spp . ( B , grey ) , which is also visible to some 841 extent before symmetrizing . Note that the coloring of elements in these and other figure 842 panels is merely illustrative and not meant to reflect underlying molecular boundaries . 843 ( C , H ) 2D longitudinal view of central cartwheel STA delineated by a dashed box in ( A , 844 F ) for T . spp . ( C ) and T . agilis ( H ) . Arrowheads denote position of line scans along the 845 vertical axis at the level of the CID ( C , purple ; H , dark pink ) , and hub ( C , light blue ; H , 846 light pink ) , with corresponding pixel intensities in arbitrary units ( AU ) . Plot profiles are 847 shifted horizontally relative to each other for better visibility . Some maxima are 848 highlighted with dashed lines ; the average distance between two CID elements is 8 . 5 849 ± 0 . 2 nm ( N = 4 ) in T . spp . ( C ) and 8 . 3 ± 0 . 5 ( N = 9 ) in T . agilis ( H ) . Note that hub densities 850 are elongated in the vertical direction , resulting in broad peak profiles where two 851 25 maxima can sometimes be discerned ( dashed light blue lines in ( C ) ; dashed light pink 852 lines in ( H ) ) . 853 ( D , I ) Longitudinal views of T . spp . ( D ) and T . agilis ( I ) central cartwheel STA . The 854 average distance between emanating spokes is 8 . 0 ± 1 . 5 nm ( N = 2 ) in T . spp . ( D ) and 855 8 . 2 ± 0 . 5 nm ( N = 9 ) in T . agilis ( I ) . Note that these values are measurements on STA 856 and are within the error of the CID periodicities reported in ( C , H ) ; the same applies for 857 other measurements hereafter . Discontinuous densities in the center of the CID ( grey ) 858 are visible in ( D , I ) ; the T . agilis volume in ( I ) is shown at a lower contour level than in 859 ( G ) . Boxed area is shown magnified in ( E , J ) . Proximal is down in this and all other 860 figure panels showing STA longitudinal views . 861 ( E , J ) The CID axis is located distal relative to axis of spoke densities , corresponding 862 to an average shift of 0 . 9 ± 0 . 7 ( N = 3 ) nm in T . spp . ( E ) and 0 . 8 ± 0 . 3 nm ( N = 9 ) in T . 863 agilis ( J ) . 864 865 Figure 3 : Architecture of peripheral elements in T . spp . and T . agilis . 866 ( A , E ) ( Top ) 2D slices through STA transverse view of microtubule triplet of T . spp . ( A ) 867 and T . agilis ( E ) , with insets showing position of pinhead ( dashed green box ) and A - C 868 linker ( dashed red box ) . ( Bottom ) Longitudinal 2D slice of STA centered on the pinhead 869 ( left ) or A - C linker ( right ) . Schematic on top illustrates the different areas used to 870 generate maps of the microtubule triplets ( B , F ) , pinhead ( C , G ) and A - C linker ( D , H ) . 871 ( B , F ) Transverse view of T . spp . ( B ) and T . agilis ( F ) microtubule triplet STA . 872 Microtubule protofilament numbers are indicated , as are the pinhead and A - C linker 873 ( only the C - link is visible ; the A - link lies on the edge of the volume and is thus less well 874 resolved in this STA - for better view see STA centered on A - C linker in D , H ) . 875 Prominent microtubule inner densities within the A - microtubule are highlighted ( empty 876 arrowhead next to A9 , chevron next to A5 ( B ) ) , as are additional external densities at 877 the A - B and B - C inner junctions ( black arrowheads ) . Double arrowheads point to 878 viewing direction in indicated panels . 879 ( C , G ) Longitudinal view of STA centered on the pinhead in T . spp . ( C ) and T . agilis 880 ( G ) , from the viewing point indicated in ( B , F ) . The pinfeet ( PinF1 and PinF2 ) and 881 pinbody ( PinB ) are indicated , as are microtubule protofilaments A3 and A4 . The 882 average distance between pinfeet elements is 8 . 6 ± 0 . 4 nm and 7 . 9 ± 0 . 4 nm in T . spp . 883 ( N = 3 each ; C ) and 8 . 4 ± 0 nm for each in T . agilis ( N = 3 each ; G ) . Corresponding 884 transverse views are shown below , illustrating the connection of PinF2 with 885 protofilament A3 . 886 26 ( D , H ) Longitudinal view of STA centered on the A - C linker in T . spp . ( D ) and T . agilis 887 ( H ) , from the viewing point indicated in ( B , F ) . Microtubule protofilaments A8 / 9 and 888 C9 / C10 of two adjacent triplets are indicated , as are the A - and C - links . The average 889 distance between A - and C - links is 8 . 4 ± 0 . 4 nm and 8 . 4 ± 0 . 3 nm in T . spp . ( both N = 6 ; 890 D ) as well as 8 . 4 ± 0 . 3 nm and 8 . 4 ± 0 . 9 nm in T . agilis ( N = 6 and N = 5 , respectively ; 891 H ) . Corresponding transverse views are shown below , chevrons point to connecting 892 points ; the connection of the A - link with A9 is only partially visible in the transverse 893 view at this height , as indicated by the dashed chevron ( H ) . 894 895 Figure 4 : Novel architectural features in T . mirabilis cartwheel . 896 ( A ) Transverse 2D slices through STA comprising 64 % and 36 % of sub - volumes of 897 T . mirabilis central cartwheel at indicated height , from proximal ( 0 nm ) to distal ( 8 . 0 898 nm ) . The dashed boxes in the 4 . 0 nm slices indicate region shown in D and E . Note 899 filamentous cartwheel inner densities ( fCID ) in the center of the hub . Schematic on top 900 illustrates the different areas used to generate 3D maps of the central cartwheel ( A - 901 G ) , MT triplets ( I ) , pinhead ( J ) and A - C linker ( K ) . 902 ( B , C ) Transverse views of T . mirabilis central cartwheel 3D map of 64 % ( B ) and 36 903 % ( C ) classes , with corresponding hub diameters of 23 . 5 ± 0 . 2 nm and 23 . 4 ± 0 . 5 nm 904 ( both N = 3 ) . In both classes the fCID is visible and 9 spokes emanate from the hub . 905 ( D , E ) 2D longitudinal view of T . mirabilis central region of the cartwheel STA 64 % ( D ) 906 and 36 % ( E ) classes in the region delineated by dashed boxes in ( A ) . Arrowheads 907 denote position of line scans along vertical axis at the hub level , with corresponding 908 normalized pixel intensities ( in arbitrary units ) . Some maxima are highlighted by 909 dashed lines . The average distance between hub units in the 64 % class is 8 . 4 ± 0 . 6 910 nm ( N = 8 ; D ) ; in the 36 % class , the distance between hub densities alternates between 911 3 . 2 ± 0 . 3 nm and 5 . 2 ± 0 . 3 nm and ( both N = 9 ; E ) . Dashed white line in ( E ) indicates 912 the offset between two superimposed hub units at the level of spoke densities , which 913 occurs every other hub unit pair . 914 ( F , G ) Longitudinal views of T . mirabilis cartwheel STA 64 % ( F ) and 36 % ( G ) classes . 915 The average distance between spokes is 8 . 4 ± 0 . 6 nm ( N = 8 ) in the 64 % class and 8 . 2 916 ± 1 . 5 nm ( N = 7 ) in the 36 % class . Note in both cases the continuous ~ 7 nm fCID inside 917 the hub . Note also densities bridging successive hubs vertically ( arrows ) . Asterisks 918 denote positions of individual units apparent within hub densities . Note that the fCID 919 was not always positioned in the geometrical center of the hub , suggestive of inherent 920 flexibility . 921 27 ( H ) ( top ) 2D slices through STA transverse view of microtubule triplet in T . mirabilis , 922 with insets showing pinhead ( dashed green box ) and A - C linker ( dashed red box ) . 923 ( bottom ) Longitudinal 2D slice of STA centered on the pinhead ( left ) or A - C linker 924 ( right ) . 925 ( I ) Transverse view of microtubule triplet STA in T . mirabilis . Microtubule protofilaments 926 are indicated , as are the positions of the pinhead and the C - link ( the A - link lies on the 927 edge of the volume and is thus less well resolved in this STA - for better views see 928 STA centered on A - C linker in K ) . At this contour level , MIPs are not visible . 929 Arrowheads indicate external densities at the A - B and B - C inner junctions ; chevron 930 indicates C - stretch . Double arrowheads point to viewing directions shown in ( J , K ) . 931 ( J ) Longitudinal view of T . mirabilis STA centered on the pinhead , from the viewing 932 point indicated in ( I ) . Location of pinhead consisting of pinfeet ( PinF1 and PinF2 ) and 933 pinbody ( PinB ) are indicated , as are microtubule protofilaments A3 and A4 . The 934 average distance between pinfeet elements is 8 . 4 ± 0 . 5 nm ( N = 5 ) . Corresponding 935 transverse view is shown below . 936 ( K ) Longitudinal view of T . mirabilis STA centered on the A - C linker , from the viewing 937 point indicated in ( I ) . Microtubule protofilaments A8 / A9 and C9 / C10 of two adjacent 938 triplets are indicated , as are the A - and C - links . The average distance between A - and 939 C - links is 8 . 4 ± 0 . 9 nm ( N = 6 ) and 8 . 4 ± 0 . 3 nm , respectively ( N = 5 ) . Corresponding 940 transverse views are shown below , chevrons point to connections . 941 942 Figure 5 : Direct stacking of SAS - 6 rings 943 ( A - D ) Computationally assembled CrSAS - 6 [ 6HR ] single ( A , C ) or double ring in 944 register ( ribbon diagram shown in different shades for clarity ) ( B , D ) fitted into the 3D 945 maps of the 36 % T . mirabilis class ( A , B ) and the 45 % T . agilis class ( C , D ) . Dashed 946 box shows longitudinal section through hub element ( left ) , double chevron viewing 947 point of longitudinal external views ( right ) . Extra unaccounted densities in single ring 948 fitting are indicated by arrowheads ( A , C ) . Note that each elongated hub density can 949 accommodate two tightly stacked CrSAS - 6 [ 6HR ] rings ( B , D ) , and that only one ring 950 can fit into the thin hub element in the 45 % T . agilis class ( C , D ) . Indicated distances 951 stem from measurements on the fitted models . 952 ( E ) Purified CrSAS - 6 [ NL ] protein self - organized into stacks and analyzed by cryo - ET ; 953 one longitudinal and one transverse view are highlighted by a dashed rectangle and a 954 dashed circle , respectively . 955 ( F , G ) 2D views through STA of in vitro self - assembled CrSAS - 6 [ NL ] proteins . 956 Transverse sections at the indicated heights through one assembly unit ( F ) , as well as 957 28 longitudinal view ( G ) of the area delineated by a dashed box in ( F ) ; white arrowhead 958 denotes position of line scan along the vertical axis at the level of the hub , with 959 corresponding pixel intensities in arbitrary units ( AU ) . The average distance between 960 two hub elements is 4 . 5 ± 0 . 3 nm ( N = 7 ; dashed lines ) . 961 ( H ) Longitudinal view of in vitro self - assembled CrSAS - 6 [ NL ] proteins STA 3D map . 962 Note densities bridging successive hubs vertically ( arrows ) . 963 ( I ) Two CrSAS - 6 [ HR ] single rings fitted into the 3D map of in vitro self - assembled 964 CrSAS - 6 [ NL ] proteins . Dashed box indicates longitudinal section through hub element 965 ( left ) , double chevron indicates viewing point of longitudinal external view ( right ) . Note 966 that the 3D map can accommodate rings stacked 4 . 4 nm apart ( measured on fitted 967 ring models ) , and vertical densities linking hubs are partially occupied . 968 969 Figure 6 : Working model of centriolar cartwheel architecture 970 ( A ) Occurrence of CID or fCID in species analyzed in this study ( T . spp . , T . agilis and 971 T . mirabilis , blue background ) , and in select other organisms mapped on a taxonomic 972 tree ( Tree generated using phyloTv2 based on NCBI taxonomy - 973 https : / / phylot . biobyte . de / ) . See main text for references . 974 ( B ) Schematic of cartwheel hub in T . spp . and T . agilis with 9 - fold symmetrical CID 975 connecting to the hub , equidistant between two spokes . 976 ( C ) The CID is polarized distally relative to the hub element comprising two tightly 977 superimposed ring elements in T . spp . and T . agilis . 978 ( D ) Schematic of cartwheel hub in T . mirabilis with continuous fCID along the proximal - 979 distal axis , with potential thin connections to the hub , equidistant between two spokes . 980 ( E ) Working model of cartwheel stacking , with alternation of two modes of SAS - 6 ring 981 stacking : pairs of rings are tightly superimposed with a rotational offset ( dashed line ) 982 and such offset double rings are then stacked in register ( solid line ) ; the offset direction 983 is consistent in T . spp . , T . agilis and T . mirabilis , yet may be different in other species , 984 as suggested in Paramecium ( Klena et al ) . Surface representation of SAS - 6 rings ; 985 homodimers highlighted by black contour . 986 ( F ) Consensus model of periodicities in centriolar cartwheel along the proximal - distal 987 axis . Surface representation of offset double rings tightly superimposed with ~ 3 . 2 nm 988 periodicity , in register with the next such offset double ring located ~ 5 . 2 nm away , thus 989 amounting to an overall periodicity of ~ 8 . 4 nm . Note that these values are based on 990 the T . mirabilis 36 % class , which has the most clearly resolved periodicities , yet does 991 not have a CID as shown in this composite model , with related values in other species ; 992 see text for details . The CID exhibits a related ~ 8 . 4 nm mean periodicity in the two 993 Trichonympha species . Two spoke periodicities are observed : close to the hub , pairs 994 29 of spokes are ~ 8 . 4 nm apart , whereas spoke pairs merge towards the periphery , where 995 their periodicity is hence ~ 16 . 8 nm , matching that of the pinhead and connected 996 microtubules . Note that spoke angles are merely illustrative and limited by the 997 resolution of the obtained STA maps . Note also that alternative modes of spoke 998 merging may be revealed by analyzing much larger sub - volumes and be present in 999 other species ( Klena et al ) . 1000 1001 1002 1003 Additional Data 1004 1005 Supplemental Table 1 - 2 1006 Supplemental Table Legends 1007 Table S1 1008 Statistics of Cryo - EM data collection and reconstruction and EMDB accession , 1009 see also Fig . S1 . 1010 1011 Table S2 1012 Correlation and overlap score for fitting with computationally assembled 1013 CrSAS - 6 [ 6HR ] single , double , and double rings with offset . 1014 Results of the fit in map function performed with ChimeraX by maximizing correlation 1015 as shown in Fig . 5 and Fig . S9 , Fig . S10 and described in Materials and Methods . 1016 Reported are the percentage of atoms inside the density map , the correlation , and the 1017 overlap between model and map . For each STA map , we report the gain in overlap 1018 upon fitting the double and offset double versus single ring model ; the maximum 1019 possible gain is + 100 % . 1020 1021 Supplemental Figures S1 - S10 1022 Supplemental Figure Legends 1023 Figure S1 : Cryo - EM acquisition , raw data and processing 1024 ( A ) Low magnification view of cryo - preserved purified T . spp . centriole and axoneme 1025 on carbon grid . Note that the proximal end of the centriole is blunt , and that the 1026 diameter of the centriole containing microtubule triplets is larger than that of the 1027 axoneme containing microtubule doublets , enabling one to determine the length of the 1028 30 centriole , as indicated . The black square illustrates the proximal location and the 1429 1029 x 1429 nm size of the field of view typically used for tilt series acquisition ; note that this 1030 can be less than the length of the centriole depending on the species . Dashed white 1031 rectangle : area corresponding to the tomogram shown in panel B . 1032 ( B ) 2D longitudinal view ( sum of 10 - 20 slices ) of the central cartwheel tomogram , with 1033 band - pass filter and Gaussian blur corresponding to the area with a dashed rectangle 1034 in ( A ) . Sub - volumes extracted along the longitudinal axis are ( 80 nm ) 3 for the central 1035 cartwheel ( pink squares represent one plane of the sub - volume ) and ( 89 nm ) 3 for the 1036 peripheral elements comprising microtubule triplet , pinhead and A - C linker ( blue 1037 squares represent one plane of the sub - volume ) , with a step size of 25 nm in both 1038 cases . 1039 ( C - F ) 2D longitudinal views from raw reconstructed tomograms of the central cartwheel 1040 illustrating typical sub - volumes ( sum of 10 - 20 slices of the central cartwheel tomogram , 1041 with band - pass filter and Gaussian blur ) from T . spp . ( C ) , T . agilis ( D ) , T . mirabilis ( E ) 1042 and CrSAS - 6 [ NL ] stacked assemblies ( F ) . Filled arrowhead indicate hub elements 1043 constituted of double units that can sometimes be resolved upon STA , but are not 1044 visible in this raw data ( C - E ) , empty arrowhead indicate hub elements in the T . agilis 1045 45 % class lacking the CID ( D ) . Note that classes cannot be unambiguously 1046 distinguished in the T . mirabilis raw tomograms , empty arrowheads in this case 1047 indicate hub elements with apparent variations in spacing ( E ) . Chevrons indicate 1048 resolved CrSAS - 6 [ NL ] single unit hub elements ( F ) . 1049 ( G - J ) Processing scheme with number of tomograms , initial sub - volumes extracted for 1050 cartwheel and peripheral elements ( microtubule triplet , pinhead and A - C linker ) , 1051 classification / refinement steps , and final sub - volumes used for STA in T . spp . ( G ) , T . 1052 agilis ( H ) , T . mirabilis ( I ) and CrSAS - 6 [ NL ] stacked assemblies ( J ) . See also Table S1 . 1053 1054 Figure S2 : Resolution maps of T . spp . 1055 ( A , D , G , J ) Local resolution distribution in T . spp . STA estimated with ResMap . Shown 1056 are transverse views of central cartwheel ( A ) , microtubule triplet ( D ) , pinhead ( G ) and 1057 A - C linker ( J ) . 1058 ( B , E , H , K ) Histogram and directional Fourier shell correlation ( FSC ) plots for T . spp . 1059 STA of central cartwheel ( B ) , microtubule triplet ( E ) , pinhead ( H ) and A - C linker ( K ) 1060 with global resolution at FSC 0 . 143 criterion indicated . 1061 ( C , F , I , L ) Power spectra of 2D class average with periodicities highlighted by arrows 1062 for T . spp . STA of central cartwheel ( C ) , microtubule triplet ( F ) , pinhead ( I ) and A - C 1063 linker ( L ) . The precision for the measurements in this and remaining power spectra is 1064 1 pixel , corresponding to a minimum of 0 . 4 nm ( see Table S1 ) . 1065 31 1066 Figure S3 : Resolution maps of T . agilis 1067 ( A , D , G , J , M ) Local resolution distribution in T . agilis STA estimated with ResMap . 1068 Shown are transverse views of central cartwheel 55 % class ( A ) , 45 % class ( D ) , 1069 microtubule triplet ( G ) , pinhead ( J ) and A - C linker ( M ) . 1070 ( B , E , H , K , N ) Histogram and directional FSC plots for T . agilis STA of central 1071 cartwheel 55 % class ( B ) , 45 % class ( E ) , microtubule triplet ( H ) , pinhead ( K ) and A - C 1072 linker ( N ) with global resolution at FSC 0 . 143 criterion indicated . 1073 ( C , F , I , L , O ) Power spectra of 2D class average with periodicities highlighted by 1074 arrows for T . agilis STA of central cartwheel 55 % class ( C ) , 45 % class ( F ) , microtubule 1075 triplet ( I ) , pinhead ( L ) and A - C linker ( O ) . 1076 1077 Figure S4 : Resolution maps of T . mirabilis 1078 ( A , D , G , J , M ) Local resolution distribution in T . mirabilis STA estimated with ResMap . 1079 Shown are transverse views of central cartwheel 64 % class ( A ) , 36 % class ( D ) , 1080 microtubule triplet ( G ) , pinhead ( J ) and A - C linker ( M ) . 1081 ( B , E , H , K , N ) Histogram and directional FSC plots for T . mirabilis STA of central 1082 cartwheel 64 % class ( B ) , 36 % class ( E ) , microtubule triplet ( H ) , pinhead ( K ) and A - C 1083 linker ( N ) with global resolution at FSC 0 . 143 criterion indicated . 1084 ( C , F , I , L , O ) Power spectra of 2D class average with periodicities highlighted by 1085 arrows for T . mirabilis STA of central cartwheel 64 % class ( C ) , 36 % class ( F ) , 1086 microtubule triplet ( I ) , pinhead ( L ) and A - C linker ( O ) . 1087 1088 Figure S5 : Resolution maps of CrSAS - 6 [ NL ] stacked assemblies 1089 ( A ) Local resolution distribution in CrSAS - 6 [ NL ] stacked assemblies ( generated from 1090 side - views ) STA estimated with ResMap shown in transverse views . 1091 ( B ) Histogram and directional FSC plots for CrSAS - 6 [ NL ] stacked assemblies STA with 1092 global resolution at FSC 0 . 143 criterion indicated . 1093 ( C ) Power spectra of 2D class average with periodicity highlighted by arrows for 1094 CrSAS - 6 [ NL ] stacked assemblies . 1095 1096 Figure S6 : Variation in hub architecture is some T . agilis sub - volumes 1097 ( A ) Transverse 2D slices through central cartwheel STA of T . agilis 45 % class , which 1098 comprises 25 % and 20 % sub - classes ( see E ) , at indicated height , from proximal ( 0 1099 nm ) to distal ( 18 . 0 nm ) . The pink circles mark spokes with CID ( 12 . 0 nm , dashed line ) 1100 or without CID ( 2 . 0 nm , solid line ) , as represented in ( B ) ; dashed box in the 4 . 0 nm 1101 32 slice indicates longitudinal section shown in ( C ) . Schematic on top illustrates the area 1102 used to generate the 3D maps of the central cartwheel . 1103 ( B ) Transverse views of central cartwheel STA 3D map of T . agilis 45 % class . The 1104 hub diameter is 22 . 7 nm ± 0 . 2 nm ( N = 3 ) with 9 emanating spoke densities , either at a 1105 level where the CID is present ( dashed circle ) or not ( solid circle ) , as indicated in ( A ) . 1106 ( C ) 2D longitudinal view of central cartwheel STA of T . agilis 45 % class delineated by 1107 a dashed box in ( A ) . Arrowheads denote position of line scans along the vertical axis 1108 at the level of the CID ( dark pink ) and the hub ( light pink ) , with corresponding 1109 normalized pixel intensities in arbitrary units ( AU ) . The plot profiles are shifted 1110 horizontally relative to each other for better visibility . The distance between hub 1111 densities alternates between 3 . 2 nm ( N = 1 ) and 6 . 0 ± 0 . 3 nm ( N = 2 ) ; maxima are 1112 indicated by dashed pink lines . Dashed white line indicates offset of two superimposed 1113 hub units . The average distance between two CID elements is 15 . 2 nm ( N = 1 ; dashed 1114 red line ) . The middle hub density that comprises only one unit and lacks a neighboring 1115 CID is indicated by a dashed arrow . 1116 ( D ) Longitudinal view of central cartwheel STA of T . agilis 45 % class at lower contour 1117 level than in ( B ) . Note densities bridging successive hubs vertically ( arrows ) , as well 1118 as proximal location of CID relative to the spoke density axis , resulting in a vertical 1119 offset of 1 . 2 ± 0 nm ( N = 2 ; arrowhead ) . Note also absence of CID in middle hub element 1120 comprising a single unit . 1121 ( E ) The spacing between the spoke density emanating from a double hub unit and the 1122 spoke density emanating from a single hub unit distal to it is 8 . 0 nm in the 25 % class 1123 and 7 . 1 nm in the 20 % class ( both N = 1 ) . By contrast , the spacing between the spoke 1124 density emanating from a double hub unit and the spoke density emanating from a 1125 single hub unit proximal to it is 8 . 7 nm in the 25 % class and 7 . 6 nm in the 20 % class 1126 ( both N = 1 ) . 1127 ( F ) Distribution of sub - volumes of the 55 % ( black circle ) and 45 % ( white circle ) 1128 classes along five T . agilis centrioles ; areas with neither black nor white circle could 1129 not be clearly assigned to either class . Note individual centrioles constituted of mostly 1130 one of the two classes , and others where the two classes are mixed without an 1131 apparent pattern , indicating that the distribution of 55 % and 45 % classes is not 1132 stereotyped along the centriole . 1133 1134 Figure S7 : Polar centriolar cartwheel 1135 ( A - P ) Non - symmetrized STA comprising larger volume than in Fig . 2 - 4 and centered 1136 on the spokes to jointly show the central cartwheel and peripheral elements in T . agilis 1137 ( A - D : 55 % class ; E - H : 45 % class ) and T . mirabilis ( I - L : 64 % class ; M - P : 36 % class ) . 1138 33 2D slices through STA transverse view ( A , E , I , M ) with corresponding 3D views ( B , F , 1139 J , N ) , as well as 2D longitudinal views ( C , G , K , O ) with corresponding 3D views ( D , H , 1140 L , P ) . The concerted proximal - distal polarity is visible from the central CID ( in A - H ) and 1141 the hub all the way to the pinhead . Note that proximal - distal polarity is visible also in 1142 the asymmetric spoke densities tilt angles in all cases , with a more pronounced tilt on 1143 the proximal side ( D , H , L , P ) . Note also that the unsymmetrized CID shown here 1144 resembles that in the symmetrized maps of Figure 2 , indicating that the CID exhibits 1145 bona fide 9 - fold radial symmetry ( A , E ) . Arrowhead in ( N , P ) points to a connection 1146 between fCID and hub . For representation , a Gaussian filter was applied to maps in 1147 ChimeraX . 1148 1149 Figure S8 : Very long cartwheel in T . mirabilis . 1150 ( A ) Differential interference contrast micrograph of live T . mirabilis cell . The arrow 1151 points to cell anterior , where the rostrum is located ; arrowheads point to some of the 1152 flagella . 1153 ( B ) Transmission electron micrograph of T . mirabilis centriole embedded in resin - 1154 longitudinal view . The hub ( arrowhead ) is visible in the proximal cartwheel - bearing 1155 region ( green line ) , but not in the distal region ( black line ) . 1156 ( C ) Transmission electron micrograph of T . mirabilis centriole embedded in resin in 1157 transverse view ( left ) and corresponding image circularized and symmetrized ( right ) . 1158 Note small density present inside the hub corresponding to the fCID . 1159 ( D ) Transverse view slice through cryo - electron tomogram of T . mirabilis centriole 1160 circularized and symmetrized . 1161 ( E ) Distribution of sub - volumes of the 64 % ( black circle ) and 36 % ( white circle ) 1162 classes along four T . mirabilis centrioles , proximal is left ; areas with neither black nor 1163 white circle could not be clearly assigned to either class . Note that the distribution of 1164 the two classes is not stereotyped along the centriole . 1165 1166 Figure S9 : CrSAS - 6 [ 6HR ] fitting in different cartwheel STA maps 1167 ( A ) Ribbon representation of CrSAS - 6 [ 6HR ] homodimers ( top , scale bar 5nm ) used for 1168 rigid body fitting into the 3D map from T . mirabilis 36 % class shown in side view ( left ) 1169 and end - on view magnified , with dashed line indicating offset between two 1170 superimposed homodimers ( right ) . Individual layers of homodimers are indicated in 1171 distinct colors for clarity . In this and all other panels of this figure , dashed box indicates 1172 longitudinal section through hub element ( left ) , double chevron indicates viewing 1173 direction of longitudinal external views ( right ) . 1174 34 ( B - D ) CrSAS - 6 [ 6HR ] single or double rings in register ( ribbon diagram represented in 1175 different shades for clarity ) fitted into the 3D maps from the T . mirabilis 64 % class ( B ) , 1176 T . spp . ( C ) and the T . agilis 55 % class ( D ) . Single ring fitting is shown in the top , 1177 double ring fitting at the bottom . Note unaccounted densities in the hub upon fitting of 1178 single rings in all cryo - ET maps ( arrowheads , B - D ) , whereas all maps readily 1179 accommodate double CrSAS - 6 [ 6HR ] rings at the hub element , although the coiled - coil 1180 elements extend slightly beyond the spoke densities . Note also that one coiled - coil in 1181 the T . mirabilis 64 % class sticks out of the density entirely ( B , bottom ) . 1182 1183 Figure S10 : Superimposed rings of SAS - 6 may be offset with respect to one 1184 another 1185 ( A , B ) ( left ) Ribbon representation of CrSAS - 6 [ 6HR ] homodimers computationally 1186 assembled into double rings ( rings represented in different shades for clarity ) with 1187 spokes in register ( A ) or with a 6 . 5° offset turning the proximal ring clockwise and 1188 moved 0 . 4 nm closer ( B ) ; transverse ( top ) and longitudinal view ( bottom ) . ( Right ) 1189 Magnified views of two vertically superimposed homodimers ( ribbon diagram ) , with 1190 dashed lines indicating angle between axes of the two coiled - coil axes ( top ) and 1191 surface representation of complementary interface ( bottom ) . Distances were 1192 measured on the ribbon diagram . 1193 ( C , D ) CrSAS - 6 [ 6HR ] offset double rings ( shown in B and represented in different 1194 colors for clarity ) fitted into 3D map from T . mirabilis 36 % class ( C ) and T . agilis 45 % 1195 class ( D ) , with dashed line indicating offset between double rings . Dashed box 1196 indicates longitudinal section through hub element ( left ) , double chevron viewing point 1197 of longitudinal external views ( right ) . 1198 ( E ) Schematic illustrating processing reported in ( F - H ) : after linearization of the 1199 cartwheel STA , a longitudinal section in Z - direction ( blue line ) is shown at the level of 1200 the spokes ( represented in pink ) . 1201 ( F - H ) Longitudinal section as explained in ( E ) for T . mirabilis 36 % and 64 % classes 1202 ( F ) , T . agilis 45 % and 55 % classes ( G ) and T . spp . ( H ) at the level of spoke emergence 1203 from the hub . Note offset of individual spokes highlighted by a dashed line . 1204 ( I ) Longitudinal section as explained in ( E ) for in vitro assembled CrSAS - 6 [ NL ] stacks ; 1205 spokes from consecutive stacks are almost in register . Note that in ( F - H ) the spokes 1206 of double hub elements cannot be fully resolved , while they appear as individual units 1207 in ( I ) . 1208 1209 35 References 1210 Azimzadeh , J . , and Marshall , W . F . ( 2010 ) . Building the centriole . Curr . Biol . 20 , R816 - 1211 25 . 1212 Bharat , T . A . M . , and Scheres , S . H . W . ( 2016 ) . Resolving macromolecular structures 1213 from electron cryo - Tomography data using subtomogram averaging in RELION . Nat . 1214 Protoc . 11 , 2054 – 2065 . 1215 Boscaro , V . , James , E . R . , Fiorito , R . , Hehenberger , E . , Karnkowska , A . , Del Campo , 1216 J . , Kolisko , M . , Irwin , N . A . T . , Mathur , V . , Scheffrahn , R . H . , et al ( 2017 ) . Molecular 1217 characterization and phylogeny of four new species of the genus trichonympha 1218 ( Parabasalia , trichonymphea ) from lower termite hindguts . Int . J . Syst . Evol . Microbiol . 1219 67 , 3570 – 3575 . 1220 van Breugel , M . , Hirono , M . , Andreeva , A . , Yanagisawa , H . A . , Yamaguchi , S . , 1221 Nakazawa , Y . , Morgner , N . , Petrovich , M . , Ebong , I . - O . O . , Robinson , C . V , et al ( 2011 ) . 1222 Structures of SAS - 6 suggest its organization in centrioles . Science 331 , 1196 – 1199 . 1223 Bucek , A . , Šobotník , J . , He , S . , Shi , M . , McMahon , D . P . , Holmes , E . C . , Roisin , Y . , Lo , 1224 N . , and Bourguignon , T . ( 2019 ) . Evolution of Termite Symbiosis Informed by 1225 Transcriptome - Based Phylogenies . Curr . Biol . 29 , 3728 - 3734 . 1226 Carpenter , K . J . , and Keeling , P . J . ( 2007 ) . Morphology and phylogenetic position of 1227 Eucomonympha imla ( Parabasalia : Hypermastigida ) . J . Eukaryot . Microbiol . 54 , 325 – 1228 332 . 1229 Culver , B . P . , Meehl , J . B . , Giddings , T . H . , and Winey , M . ( 2009 ) . The two SAS - 6 1230 homologs in Tetrahymena thermophila have distinct functions in basal body assembly . 1231 Mol . Biol . Cell 20 , 1855 – 1864 . 1232 Dammermann , A . , Müller - Reichert , T . , Pelletier , L . , Habermann , B . , Desai , A . , and 1233 Oegema , K . ( 2004 ) . Centriole assembly requires both centriolar and pericentriolar 1234 material proteins . Dev . Cell 7 , 815 – 829 . 1235 Fawcett , D . W . , and Porter , K . R . ( 1954 ) . A study of the fine structure of ciliated epithelia . 1236 J . Morphol . 94 , 221 – 281 . 1237 Garzón , J . I . , Kovacs , J . , Abagyan , R . , and Chacón , P . ( 2007 ) . ADP _ EM : Fast 1238 exhaustive multi - resolution docking for high - throughput coverage . Bioinformatics 23 , 1239 427 – 433 . 1240 Geimer , S . , and Melkonian , M . ( 2004 ) . The ultrastructure of the Chlamydomonas 1241 reinhardtii basal apparatus : Identification of an early marker of radial asymmetry 1242 36 inherent in the basal body . J . Cell Sci . 117 , 2663 – 2674 . 1243 Gibbons , I . R . , and Grimstone , A . V . ( 1960 ) . On flagellar structure in certain flagellates . 1244 J . Biophys . Biochem . Cytol . 7 , 697 – 716 . 1245 Goddard , T . D . , Huang , C . C . , Meng , E . C . , Pettersen , E . F . , Couch , G . S . , Morris , J . H . , 1246 and Ferrin , T . E . ( 2018 ) . UCSF ChimeraX : Meeting modern challenges in visualization 1247 and analysis . Protein Sci . 27 , 14 – 25 . 1248 Gönczy , P . , and Hatzopoulos , G . N . ( 2019 ) . Centriole assembly at a glance . J . Cell Sci . 1249 132 . 1250 Gottardo M , Callaini G & Riparbelli MG ( 2015 ) The Drosophila centriole - conversion 1251 of doublets into triplets within the stem cell niche . J . Cell Sci . 128 . 2437 - 2442 . 1252 Greenan , G . A . , Keszthelyi , B . , Vale , R . D . , and Agard , D . A . ( 2018 ) . Insights into 1253 centriole geometry revealed by cryotomography of doublet and triplet centrioles . Elife 1254 7 . 1255 Greenan , G . A . , Vale , R . D . , and Agard , D . A . ( 2020 ) . Electron cryotomography of intact 1256 motile cilia defines the basal body to axoneme transition . J . Cell Biol . 219 . 1257 Le Guennec , M . , Klena , N . , Gambarotto , D . , Laporte , M . H . , Tassin , A . M . , van den 1258 Hoek , H . , Erdmann , P . S . , Schaffer , M . , Kovacik , L . , Borgers , S . , et al ( 2020 ) . A helical 1259 inner scaffold provides a structural basis for centriole cohesion . Sci . Adv . 6 . 1260 Guichard , P . , Chrétien , D . , Marco , S . , Tassin , A . M . , Chretien , D . , Marco , S . , and 1261 Tassin , A . M . ( 2010 ) . Procentriole assembly revealed by cryo - electron tomography . 1262 EMBO J . 29 , 1565 – 1572 . 1263 Guichard , P . , Desfosses , A . , Maheshwari , A . , Hachet , V . , Dietrich , C . , Brune , A . , 1264 Ishikawa , T . , Sachse , C . , Gönczy , P . , Gonczy , P . , et al ( 2012 ) . Cartwheel Architecture 1265 of Trichonympha Basal Body . Science . 337 , 553 . 1266 Guichard , P . , Hachet , V . , Majubu , N . , Neves , A . , Demurtas , D . , Olieric , N . , Fluckiger , 1267 I . , Yamada , A . , Kihara , K . , Nishida , Y . , et al ( 2013 ) . Native architecture of the centriole 1268 proximal region reveals features underlying its 9 - fold radial symmetry . Curr . Biol . 23 , 1269 1620 – 1628 . 1270 Guichard P , Hachet V , Majubu N , Neves A , Demurtas D , Olieric N , Fluckiger I , Yamada 1271 A , Kihara K , Nishida Y , Moriya S , Steinmetz MO , Hongoh Y & Gönczy P ( 2020 ) 1272 Erratum : Native Architecture of the Centriole Proximal Region Reveals Features 1273 Underlying Its 9 - Fold Radial Symmetry ( Native Architecture of the Centriole Proximal 1274 Region Reveals Features Underlying Its 9 - Fold Radial Symmetry ( 2013 ) 23 ( 17 ) ( 1620 – 1275 37 1628 ) , ( S0960982213007860 ) ) . Curr . Biol . 1276 Guichard , P . , Hamel , V . , Neves , A . , and Gönczy , P . ( 2015 ) . Isolation , cryotomography , 1277 and three - dimensional reconstruction of centrioles . Methods Cell Biol . 129 , 191 – 209 . 1278 Guichard , P . , Hamel , V . , Le Guennec , M . , Banterle , N . , Iacovache , I . , Nemcíková , V . , 1279 Flückiger , I . , Goldie , K . N . , Stahlberg , H . , Lévy , D . , et al ( 2017 ) . Cell - free reconstitution 1280 reveals centriole cartwheel assembly mechanisms . Nat . Commun . 8 , 14813 . 1281 Guichard , P . , Hamel , V . , and Gönczy , P . ( 2018 ) . The Rise of the Cartwheel : Seeding 1282 the Centriole Organelle . BioEssays 40 . 1283 Hagen , W . J . H . , Wan , W . , and Briggs , J . A . G . ( 2017 ) . Implementation of a cryo - electron 1284 tomography tilt - scheme optimized for high resolution subtomogram averaging . J . 1285 Struct . Biol . 197 , 191 – 198 . 1286 Heumann , J . M . , Hoenger , A . , and Mastronarde , D . N . ( 2011 ) . Clustering and variance 1287 maps for cryo - electron tomography using wedge - masked differences . J . Struct . Biol . 1288 175 , 288 – 299 . 1289 Hirono , M . ( 2014 ) . Cartwheel assembly . Philos . Trans . R . Soc . Lond . B . Biol . Sci . 369 . 1290 Ichikawa , M . , Liu , D . , Kastritis , P . L . , Basu , K . , Hsu , T . C . , Yang , S . , and Bui , K . H . 1291 ( 2017 ) . Subnanometre - resolution structure of the doublet microtubule reveals new 1292 classes of microtubuleassociated proteins . Nat . Commun . 8 . 1293 Jerka - Dziadosz , M . , Gogendeau , D . , Klotz , C . , Cohen , J . , Beisson , J . , and Koll , F . 1294 ( 2010 ) . Basal body duplication in Paramecium : the key role of Bld10 in assembly and 1295 stability of the cartwheel . Cytoskeleton ( Hoboken ) . 67 , 161 – 171 . 1296 Kilburn , C . L . , Pearson , C . G . , Romijn , E . P . , Meehl , J . B . , Giddings , T . H . , Culver , B . P . , 1297 Yates , J . R . , and Winey , M . ( 2007 ) . New Tetrahymena basal body protein components 1298 identify basal body domain structure . J . Cell Biol . 178 , 905 – 912 . 1299 Kimanius , D . , Forsberg , B . O . , Scheres , S . H . W . , and Lindahl , E . ( 2016 ) . Accelerated 1300 cryo - EM structure determination with parallelisation using GPUS in RELION - 2 . Elife 5 . 1301 Kitagawa , D . , Vakonakis , I . , Olieric , N . , Hilbert , M . , Keller , D . , Olieric , V . , Bortfeld , M . , 1302 Erat , M . C . , Flückiger , I . , Gönczy , P . , et al ( 2011 ) . Structural basis of the 9 - fold 1303 symmetry of centrioles . Cell 144 , 364 – 375 . 1304 Klena , N . , Le Guennec M . , Tassin , A . M . , van den Hoek , H . , Erdmann , P . S . , Schaffer , 1305 M . , Geimer , S . , Kovacik , L . , Sadian , Y . , Goldie , K . N . , Stahlberg , H . , Engel , B . D . , 1306 Hamel , V . , Guichard , P . Probing the evolutionary conservation of the centriole 1307 38 cartwheel - containing region by cryo - electron tomography . Co - submitted manusript . 1308 Kleylein - Sohn , J . , Westendorf , J . , Le Clech , M . , Habedanck , R . , Stierhof , Y . - D . , and 1309 Nigg , E . A . ( 2007 ) . Plk4 - induced centriole biogenesis in human cells . Dev . Cell 13 , 190 – 1310 202 . 1311 Koidzumi , M . ( 1921 ) . Studies on the intestinal protozoa found in the termites of japan . 1312 Parasitology 13 , 235 – 309 . 1313 Kremer , J . R . , Mastronarde , D . N . , and McIntosh , J . R . ( 1996 ) . Computer visualization 1314 of three - dimensional image data using IMOD . J . Struct . Biol . 116 , 71 – 76 . 1315 Kubai , D . F . ( 1973 ) . Unorthodox mitosis in Trichonympha agilis : Kinetochore 1316 differentiation and chromosome movement . J . Cell Sci . 13 , 511 – 552 . 1317 Kucukelbir , A . , Sigworth , F . J . , and Tagare , H . D . ( 2014 ) . Quantifying the local resolution 1318 of cryo - EM density maps . Nat . Methods 11 , 63 – 65 . 1319 Leidel , S . , Delattre , M . , Cerutti , L . , Baumer , K . , and Gonczy , P . ( 2005 ) . SAS - 6 defines 1320 a protein family required for centrosome duplication in C . elegans and in human cells . 1321 Nat Cell Biol 7 , 115 – 125 . 1322 de Leon , J . C . , Scheumann , N . , Beatty , W . , Beck , J . R . , Tran , J . Q . , Yau , C . , Bradley , 1323 P . J . , Gull , K . , Wickstead , B . , and Morrissette , N . S . ( 2013 ) . A sas - 6 - like protein 1324 suggests that the Toxoplasma conoid complex evolved from flagellar components . 1325 Eukaryot . Cell 12 , 1009 – 1019 . 1326 Li , S . , Fernandez , J . J . , Marshall , W . F . , and Agard , D . A . ( 2019 ) . Electron cryo - 1327 tomography provides insight into procentriole architecture and assembly mechanism . 1328 Elife 8 . 1329 Ma , M . , Stoyanova , M . , Rademacher , G . , Dutcher , S . K . , Brown , A . , and Zhang , R . 1330 ( 2019 ) . Structure of the Decorated Ciliary Doublet Microtubule . Cell 179 , 909 - 922 . e12 . 1331 Mastronarde , D . N . ( 2005 ) . Automated electron microscope tomography using robust 1332 prediction of specimen movements . J . Struct . Biol . 152 , 36 – 51 . 1333 McNitt R ( 1974 ) Centriole ultrastructure and its possible role in microtubule formation 1334 in an aquatic fungus . Protoplasma . 80 , 91 - 108 . 1335 Nakazawa , Y . , Hiraki , M . , Kamiya , R . , and Hirono , M . ( 2007 ) . SAS - 6 is a cartwheel 1336 protein that establishes the 9 - fold symmetry of the centriole . Curr Biol 17 , 2169 – 2174 . 1337 Nicastro , D . , Schwartz , C . , Pierson , J . , Gaudette , R . , Porter , M . E . , and McIntosh , J . R . 1338 ( 2006 ) . The molecular architecture of axonemes revealed by cryoelectron tomography . 1339 39 Science 313 , 944 – 948 . 1340 Nievergelt , A . P . , Banterle , N . , Andany , S . H . , Gönczy , P . , and Fantner , G . E . ( 2018 ) . 1341 High - speed photothermal off - resonance atomic force microscopy reveals assembly 1342 routes of centriolar scaffold protein SAS - 6 . Nat . Nanotechnol . 13 , 696 – 701 . 1343 Noda , S . , Mantini , C . , Bordereau , C . , Kitade , O . , Dolan , M . F . , Viscogliosi , E . , and 1344 Ohkuma , M . ( 2009 ) . Molecular phylogeny of parabasalids with emphasis on the order 1345 Cristamonadida and its complex morphological evolution . Mol . Phylogenet . Evol . 52 , 1346 217 – 224 . 1347 Noda , S . , Shimizu , D . , Yuki , M . , Kitade , O . , and Ohkuma , M . ( 2018 ) . Host - symbiont 1348 cospeciation of termite - gut cellulolytic protists of the genera teranympha and 1349 eucomonympha and their treponema endosymbionts . Microbes Environ . 33 , 26 – 33 . 1350 O’Toole , E . T . , and Dutcher , S . K . ( 2014 ) . Site - specific basal body duplication in 1351 Chlamydomonas . Cytoskeleton 71 , 108 – 118 . 1352 Ohkuma , M . , and Kudo , T . ( 1998 ) . Phylogenetic analysis of the symbiotic intestinal 1353 microflora of the termite Cryptotermes domesticus . FEMS Microbiol . Lett . 164 , 389 – 1354 395 . 1355 Ohkuma , M . , Noda , S . , Hongoh , Y . , Nalepa , C . A . , and Inoue , T . ( 2009 ) . Inheritance 1356 and diversification of symbiotic trichonymphid flagellates from a common ancestor of 1357 termites and the cockroach Cryptocercus . Proc . R . Soc . B Biol . Sci . 276 , 239 – 245 . 1358 Olson , L . W . , and Fuller , M . S . ( 1968 ) . Ultrastructural evidence for the biflagellate origin 1359 of the uniflagellate fungal zoospore . Arch . Mikrobiol . 62 , 237 – 250 . 1360 Park , Y . C . , Kitade , O . , Schwarz , M . , Kim , J . P . , and Kim , W . ( 2006 ) . Intraspecific 1361 molecular phylogeny , genetic variation and phylogeography of Reticulitermes speratus 1362 ( Isoptera : Rhinotermitidae ) . Mol . Cells 21 , 89 – 103 . 1363 Pettersen , E . F . , Goddard , T . D . , Huang , C . C . , Couch , G . S . , Greenblatt , D . M . , Meng , 1364 E . C . , and Ferrin , T . E . ( 2004 ) . UCSF Chimera - A visualization system for exploratory 1365 research and analysis . J . Comput . Chem . 25 , 1605 – 1612 . 1366 Reynolds , E . S . ( 1963 ) . The use of lead citrate at high pH as an electron - opaque stain 1367 in electron microscopy . J . Cell Biol . 17 , 208 – 212 . 1368 Rodrigues - Martins , A . , Bettencourt - Dias , M . , Riparbelli , M . , Ferreira , C . , Ferreira , I . , 1369 Callaini , G . , and Glover , D . M . ( 2007 ) . DSAS - 6 Organizes a Tube - like Centriole 1370 Precursor , and Its Absence Suggests Modularity in Centriole Assembly . Curr . Biol . 17 , 1371 1465 – 1472 . 1372 40 Rohou , A . , and Grigorieff , N . ( 2015 ) . CTFFIND4 : Fast and accurate defocus estimation 1373 from electron micrographs . J . Struct . Biol . 192 , 216 – 221 . 1374 Song K , Shang Z , Fu X , Lou X , Grigorieff N & Nicastro D ( 2020 ) In situ structure 1375 determination at nanometer resolution using TYGRESS . Nat . Methods . 17 , 201 - 208 . 1376 Strnad , P . , Leidel , S . , Vinogradova , T . , Euteneuer , U . , Khodjakov , A . , and Gönczy , P . 1377 ( 2007 ) . Regulated HsSAS - 6 Levels Ensure Formation of a Single Procentriole per 1378 Centriole during the Centrosome Duplication Cycle . Dev . Cell 13 , 203 – 213 . 1379 Tai , V . , James , E . R . , Perlman , S . J . , and Keeling , P . J . ( 2013 ) . Single - Cell DNA 1380 Barcoding Using Sequences from the Small Subunit rRNA and Internal Transcribed 1381 Spacer Region Identifies New Species of Trichonympha and Trichomitopsis from the 1382 Hindgut of the Termite Zootermopsis angusticollis . PLoS One 8 . 1383 Thorne , B . L . , Haverty , M . I . , Page , M . , and Nutting , W . L . ( 1993 ) . Distribution and 1384 Biogeography of the North American Termite Genus Zootermopsis ( Isoptera : 1385 Termopsidae ) . Ann . Entomol . Soc . Am . 86 , 532 – 544 . 1386 Uzbekov R , Garanina A & Bressac C ( 2018 ) Centrioles without microtubules : A new 1387 morphological type of centriole . Biol . Open . 7 , 1 - 8 . 1388 Vorobjev , I . A . , and Chentsov , Y . S . ( 1980 ) . The ultrastructure of centriole in mammalian 1389 tissue culture cells . Cell Biol . Int . Rep . 4 , 1037 – 1044 . 1390 Wang , W . J . J . W . - J . , Acehan , D . , Kao , C . - H . C . H . H . , Jane , W . N . W . - N . N . , Uryu , K . , and 1391 Tsou , M . - F . B . M . F . B . F . B . ( 2015 ) . De novo centriole formation in human cells is error - 1392 prone and does not require SAS - 6 self - assembly . Elife 4 , 1155 – 1160 . 1393 Yabe , T . , Ge , X . , and Pelegri , F . ( 2007 ) . The zebrafish maternal - effect gene cellular 1394 atoll encodes the centriolar component sas - 6 and defects in its paternal function 1395 promote whole genome duplication . Dev . Biol . 312 , 44 – 60 . 1396 Zi Tan , Y . , Baldwin , P . R . , Davis , J . H . , Williamson , J . R . , Potter , C . S . , Carragher , B . , 1397 and Lyumkis , D . ( 2017 ) . Addressing preferred specimen orientation in single - particle 1398 cryo - EMthrough tilting . Nat . Methods 14 , 793 – 796 . 1399 sample T . spp . T . agilis T . mirabilis C . reinhardtii centrioles in vitro CrSAS - 6 Data collection TEM F20 Krios Pixel size ( Å ) 3 . 49 2 . 71 Tilting scheme Bi - directional Dose - symmetric N tomograms 18 18 17 31 Reconstruction : cartwheel Step size for extraction ( Å ) 3 x 85 = 255 2 x 42 = 84 extracted sub - volumes 1385 1154 1802 926 sub - volumes final reconstruction 511 594 737 334 Resolution CW ( Å ) 24 23 23 15 EMDB accession EMD - 10927 EMD - 10916 EMD - 10918 EMD - 10922 EMD - 10923 EMD - 10928 Reconstruction : microtubule triplet ( MMT ) , pinhead ( PH ) , A - C linker ( AC ) Step size for extraction ( Å ) 3 x 85 = 255 na initial extracted sub - volumes : MMT , PH , AC 2792 3850 7928 na sub - volumes MMT , final 823 849 986 na Resolution MMT ( Å ) 38 40 38 na EMDB accession EMD - 10932 EMD - 10931 EMD - 10934 na sub - volumes PH , final 587 714 955 na Resolution PH ( Å ) 39 39 37 na EMDB accession EMD - 10938 EMD - 10935 EMD - 10941 na sub - volumes AC , final 1818 1512 1520 na Resolution AC ( Å ) 34 37 37 na EMDB accession EMD - 10939 EMD - 10937 EMD - 10942 na Table S1 . Map Ring model Percentage ( % ) of atoms inside contour Correlation Overlap Percentage ( % ) gain in overlap score by fitting double over single ring T . mirabilis 36 % Single 77 0 . 83 23 Double 63 0 . 71 31 + 58 Offset 69 0 . 73 27 + 87 T . mirabilis 64 % Single 93 0 . 78 17 Double 78 0 . 71 26 + 47 Offset 71 0 . 74 34 + 47 T . spp . Single 94 0 . 83 34 Double 64 0 . 65 41 + 23 Offset 63 0 . 69 45 + 33 T . agilis 55 % Single 99 0 . 86 38 Double 82 0 . 73 51 + 34 Offset 86 0 . 78 55 + 44 T . agilis 45 % ( thick ) Single 100 0 . 86 41 Double 87 0 . 79 60 + 46 Offset 78 0 . 83 64 + 56 T . agilis 45 % ( thin ) Single 93 0 . 87 33 Double 46 0 . 63 34 + 4 Offset 46 0 . 63 34 + 4 CrSAS - 6 [ NL ] in vitro Single 91 0 . 90 18 Double 91 0 . 90 35 + 97 Offset 84 0 . 88 33 + 83 Correlation and overlap are calculated with ChimeraX ( Goddard et al . , 2018 ) . Correlation value range : - 1 to 1 Table S2 .