1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Running title: Electron Cryo-Tomography of Procentriole Electron Cryo-Tomography Provides Insight into Procentriole Architecture and Assembly Mechanism Sam Li1*, Jose-Jesus Fernandez2, Wallace F. Marshall1 and David A. Agard1,3* 1. Department of Biochemistry and Biophysics, University of California, San Francisco, CA 94158, USA. 2. Centro Nacional de Biotecnologia – CSIC, Campus Universidad Autonoma, Cantoblanco, 28049 Madrid, Spain 3. Howard Hughes Medical Institute, University of California, San Francisco, CA 94158, USA. 21 22 23 24 25 26 27 28 29 30 31 32 33 *Correspondence: Sam Li Phone: 1-415-502-2930 Fax: 1-415-476-1902 Email: samli@msg.ucsf.edu David A. Agard Phone: 1-415-476-2521 Fax: 1-415-476-1902 Email: agard@msg.ucsf.edu 34 35 36 37 38 39 1 40 41 42 43 Abstract 44 biogenesis and homeostasis is tightly regulated. Using electron cryo-tomography 45 (cryoET) we present the structure of procentrioles from Chlamydomonas reinhardtii. We 46 identified a set of non-tubulin components attached to the triplet microtubule (MT), many 47 are at the junctions of tubules likely to reinforce the triplet. We describe structure of the 48 A-C linker that bridges neighboring triplets. The structure infers that POC1 is likely an 49 integral component of A-C linker. Its conserved WD40 β-propeller domain provides 50 attachment sites for other A-C linker components. The twist of A-C linker results in an 51 iris diaphragm-like motion of the triplets in the longitudinal direction of procentriole. 52 Finally, we identified two assembly intermediates at the growing ends of procentriole 53 allowing us to propose a model for the procentriole assembly. Our results provide a 54 comprehensive structural framework for understanding the molecular mechanisms 55 underpinning procentriole biogenesis and assembly. Centriole is an essential structure with multiple functions in cellular processes. Centriole 56 57 Keywords: centriole, procentriole, basal body, probasal body, microtubule triplet, 58 electron cryo-tomography, subtomogram averaging, structure heterogeneity 59 2 60 61 62 63 64 Introduction 65 characteristic nine-fold symmetry. As an evolutionarily conserved organelle, the 66 centriole, also known as the basal body, fulfills many cellular functions. In cycling cells, 67 a pair of centrioles recruits pericentriolar material (PCM). Together they form the 68 centrosome, the primary microtubule organizing center (MTOC) in animal cells. Related 69 to its function as an MTOC, the centrosome is essential for mitotic spindle formation, 70 spindle orientation and faithful mitotic chromosome segregation and for intracellular 71 transport of cargoes. In non-dividing cells, the centriole functions as basal body to 72 template cilium formation. Based on their structure and functions, the cilia can be further 73 classified into primary or motile cilia. The primary cilium functions as a cell “antenna” 74 that senses diverse signals on both sides of cell membrane. Motile cilia are responsible 75 for cellular swimming and fluid flow. Given the array of diverse functions carried out by 76 centrioles and basal bodies, it is not surprising that mutations affecting centrioles and 77 basal bodies cause diverse human diseases, ranging from tumors to different forms of 78 ciliopathies (Nigg and Raff, 2009; Reiter and Leroux, 2017) The centriole is a barrel-shaped structure composed of a set of MT triplets arranged in a 79 80 Centriole biogenesis is tightly controlled. Recent studies have illuminated the 81 mechanisms that regulate centriole assembly. Genetics and cell biological studies in 82 various model organisms show that centriole biogenesis occurs by a cascade of molecular 83 events performed by evolutionarily conserved components, reviewed by (Banterle and 84 Gonczy, 2017). In vertebrate cells, the sequential recruitment of centriole components 3 85 during assembly is accompanied by a series of morphological changes at the newly 86 emerging centrioles. First, the nine-fold symmetric structure called the cartwheel forms at 87 the proximal side of the mature centriole (mother centriole). This is followed by assembly 88 of MT triplets at the tip of the nine radial spokes in the cartwheel. Together, they form 89 the procentriole, a precursor of the daughter centriole. The procentriole continue to 90 elongate and develop as the new daughter centriole prior to mitosis. During mitosis, the 91 mother and daughter centrioles are “disengaged” but remain loosely linked. In the next 92 cell cycle, the daughter centriole will completely separate from the mother. Meanwhile, it 93 acquires the appendages and the PCM and becomes fully competent as an MTOC. These 94 mark the completion of the centriole duplication cycle (Kong et al., 2014). Similar to the 95 centriole duplication in metazoans such as in mammals, the centriole duplication process 96 in the unicellular organism Chlamydomonas reinhardtii is tightly controlled. The process 97 has been described in a series of seminal studies (Cavalier-Smith, 1974; Geimer and 98 Melkonian, 2004; O'Toole and Dutcher, 2014). Compared to the vertebrates, despite 99 many morphological and ultrastructural differences in the duplication steps, a number of 100 key components in the Chlamydomonas centriole assembly have been found conserved in 101 other organisms (Dutcher et al., 2002; Dutcher and Trabuco, 1998; Hiraki et al., 2007; 102 Keller et al., 2009; Matsuura et al., 2004; Nakazawa et al., 2007). In addition, proteomics 103 and bioinformatics studies in several model organisms have identified a list of major 104 structural components of centrioles (Andersen et al., 2003; Keller et al., 2005; Kilburn et 105 al., 2007; Li et al., 2004; Muller et al., 2010). Together, these studies concur that the 106 centriole is assembled by a series of evolutionarily conserved protein building blocks. 4 107 The process is tightly controlled spatially and temporally by a set of regulatory proteins 108 (Carvalho-Santos et al., 2010; Hodges et al., 2010). 109 110 Meanwhile, structural approaches, including super-resolution light microscopy, X-ray 111 crystallography and electron cryo-microscopy, have been applied to put the building 112 blocks into the context of the centriole’s 3D structure. Several crystal structures are now 113 available describing components of the centriole, including Plk4, Spd2, Sas6, Cep135, 114 STIL and CPAP. In addition, there have been cryoET studies on assembly of centriole in 115 several organisms (Greenan et al., 2018; Guichard et al., 2010; Guichard et al., 2012; Li 116 et al., 2012). In particular, the events of cartwheel assembly has been studied extensively 117 (Guichard et al., 2012; Guichard et al., 2017; Hilbert et al., 2016; Kitagawa et al., 2011; 118 van Breugel et al., 2011), leading to a molecular mechanism that at least in part 119 establishes the 9-fold symmetry, reviewed in (Guichard et al., 2018). 120 121 Despite the structural and functional study of many centriole components in the past 122 years, a complete picture of the centriole architecture and its assembly mechanism is 123 lacking. Using cryoET and subtomogram averaging, we describe the triplet structure of 124 the Chlamydomonas reinhardtii procentriole. We identify 11 non-tubulin components in 125 the structure that are associated to the triplet tubules in an asymmetric manner. We 126 further present the structure of the A-C linker that laterally bridges neighboring triplets. 127 Finally, using extensive classification and averaging in image processing, we identified 128 two partially assembled triplets at the growing ends to the procentrioles that shed light on 129 the mechanism of triplet and procentriole assembly. Overall, our work presented here 5 130 builds a framework for understanding the mechanism of centriole biogenesis in molecular 131 details. 6 132 Results 133 134 Overall Architecture of the Procentriole 135 To study the structure of both the centriole and procentriole, the nuclear-flagellar- 136 apparatus (NFAp) from the unicellular green algae Chlamydomonas reinhardtii were 137 isolated and visualized by cryoET. The collected tilt series and the reconstructed 138 tomograms show good conservation of the stereotypical structures of NFAp (Video 1). 139 These include the proximal and distal striated fibers that connect two mature centrioles, a 140 set of rootlet microtubules (rMT), in the distal end of centriole there is transition zone 141 where it connects to flagellum (Figure 1A). At the centriole proximal end, a cartwheel 142 structure, including the central hub and the radial spoke, are visible. In many tomograms, 143 two well-preserved procentrioles are attached to their respective mother centrioles via the 144 rMT (Figure 1B, Video 1). Here, we focused our study on both the proximal ~100 nm 145 region of mother centrioles and on the attached procentrioles. Using subtomogram 146 averaging, we obtained an averaged structure of MT triplet at 23.0 Å resolution (Figure 147 1C, Figure1-figure supplement 1). In this structure, all MT protofilaments (PF) can be 148 resolved, as well as the 4 nm repeat of tubulin along the PF. This confirms previous 149 observations from the central core region and from centrioles in other organisms that the 150 A-tubule is composed of 13 PFs as an elliptical ring deviated from the canonical MT 151 structure. Both B- and C-tubules are partial rings with 10 PFs. In additional, a number of 152 non-MT structures are readily visible. These include the pinhead, a structure that 153 connects the A-tubule to the cartwheel, and the A-C linker that bridges A-tubule to the C- 154 tubule of its neighboring triplet. To visualize the structure of the entire procentriole, we 7 155 took reverse steps of subtomogram averaging and put the averaged triplet into the context 156 of entire procentriole tomogram. The result is a model representing a cross section of the 157 procentriole (Figure 1D). In the model, without imposing symmetry, the pinhead is about 158 70 nm away from the central hub. The A-C linker is clearly visible and makes multiple 159 connections to both the neighboring A- and C-tubules, although with blurred details 160 compared to the MTs in the triplet, indicating structural heterogeneity and flexibility in 161 this region. 162 163 To assess whether there are structural difference between the proximal region of mother 164 centrioles (10854 subtomograms) and adjacent procentrioles (2083 subtomograms), we 165 averaged these two datasets independently. This results in two triplet averages at 23.0 Å 166 and 30.1 Å resolution respectively (Figure1-figure supplement 2). Even though the MT 167 backbones are nearly identical, there are notable differences between these two averages. 168 A number of microtubule inner proteins (MIPs) that are attached to the centriole B- 169 tubule, in particular MIPs present in the inner junction of A-B and B-C tubule, are absent 170 in the procentriole triplet. Conversely, some luminal densities in the procentriole are 171 missing in the centriole. These differences may reflect difference in kinetics or transient 172 association of proteins at various stages of assembly, or incompletion of procentriole 173 assembly (see below for further detailed analysis on the B- and C-tubules in the 174 procentrioles). 175 176 Non-tubulin Components Associated with the Procentriole Triplet 8 177 Unlike many biological complexes for which the individual components might have rigid 178 and well-defined structures, large organelle-scale assemblages are often intrinsically 179 flexible and heterogeneous, creating challenges for their structure study. To improve 180 resolution, we applied a “focused refinement” strategy widely used in single particle 181 cryoEM (Scheres, 2016), to align the A-, B- and C-tubules of the triplet separately. These 182 resulted in three averaged tubule structures, with improved resolution at 21.4 Å, 22.3 Å 183 and 22.3 Å, respectively (Figures 2A-C, Figure1-figure supplement 1). Based on the 184 improved structures, we built a MT model for each tubule, allowing us to identify a total 185 of 11 MIPs associated to the MT wall (Figures 2A-C, Figure 2-figure supplements 1-11). 186 Among them, two components MIP1 and MIP2 share similar binding pattern as the MIPs 187 previously seen in the core region (Li et al., 2012). MIP1 is a cone-shaped structure 188 projecting from PF A5 into the lumen of the A-tubule and having an 8 nm longitudinal 189 periodicity. Each MIP1 has two legs that recognize the luminal side of α/β tubulin dimer 190 (arrowheads in Figure 2-figure supplement 1). By contrast, MIP2 is more filamentous, 191 forming a trellis-like meshwork, laterally spanning PFs A11 to A13. It follows the pitch 192 of the 3-start helical MT lattice running along the A-tubule luminal wall (Figure 2-figure 193 supplement 2), presumably to reinforce the “ribbon” structure that is shared by the A- and 194 B-tubules. Besides MIP1 and MIP2, the overall pattern of the MIPs bound within the 195 procentriole is markedly different from the core region. The procentriole MIP binding 196 patterns are shown in detail in Figure 2-figure supplements 1-11 and their characteristics 197 are summarized in Table 2. Of interest, 7 of the 11 MIPs (MIPs 3~6, MIPs 9~11) are 198 localized at or in the vicinity of the inner junctions of the A-B or B-C tubules. Among 199 them, the MIP9 at the inner B-C tubule junction, is of particular interest. Even though it 9 200 forms a filament exhibiting 4 nm periodicity, the structure deviates from a canonical MT 201 protofilament (Figure 2-figure supplement 12). Therefore, we assigned it as a non-tubulin 202 protein MIP9. Together, these inner junctions MIPs crosslink multiple PFs forming 203 intricate networks. They are likely important in strengthening the overall triplet structure. 204 205 The cartwheel is a unique procentriolar structure, emerging early in centriole assembly, 206 that is essential in establishing both the 9-fold symmetry and for scaffolding procentriole 207 construction. The overall structure of the cartwheel can be divided into 3 parts, the central 208 hub, the radial spoke and the pinhead (Figure 1B) (Hirono, 2014). The pinhead connects 209 the cartwheel to the MT triplets. Using local classification to focus on the pinhead, we 210 identified 4763 subtomograms having relatively intact pinhead structures, resulting in an 211 averaged pinhead structure at 23.1 Å resolution (Figures 2D, Figure 2-figure supplement 212 14). It is a hook-shaped structure anchored to PF A3 of the A-tubule. Consistent with a 213 previous study (Guichard et al., 2013), the pinhead can be partitioned into three parts, 214 namely pinbody (PinB), pinfoot 1 (PinF1) and pinfoot 2 (PinF2) where both pinfeet 215 attach to the MT wall. Each PinF1/PinF2 unit forms a ring structure bound to α/β tubulin 216 dimer on PF A3, therefore, the entire pinhead repeats every 8 nm. This is in agreement 217 with a recent triplet structure from the proximal region of mammalian centriole (Greenan 218 et al., 2018). The PinB turns 90 degree from the pinfeet, forming an inverted L-shaped 219 structure (Figure 2D). There is a longitudinal gap between neighboring PinB every 8 nm 220 without connection. The distal tip of PinB is flexible exhibiting less well-defined 221 structure in the average (arrowheads in Figure 2D and Figure 2-figure supplement 13). 222 Interestingly, among the 13 PFs that form the elliptical-shaped A-tubule, the PFs A2 and 10 223 A3 display the highest curvature of the MT wall with a large lateral gap between these 224 two PFs where the MIP4 laterally crosslinks. The high local curvature and the gap may 225 create a unique site for anchoring the pinhead to the A-tubule. 226 227 Structure of the A-C linker that Connects Neighboring Triplets 228 Procentriole is a cylindrical structure composed of 9 MT triplet blades. Each triplet is 229 laterally connected to its neighboring triplets by a structure called the A-C linker. In the 230 averaged triplet structure (Figure 1C), the A-C linker is weaker than the MT triplet, 231 indicating either compositional and/or conformational heterogeneity. By extensive 232 subtomogram classification and alignment using the A-tubule as a reference (Figure 3- 233 figure supplement 1 and detailed in Materials and Methods), we obtained 4 classes whose 234 averaged structures showing a more complete and detailed structure of the A-C linker 235 most closely associated with the A-tubule (Figure 3-figure supplements 2, 5). Similarly, 236 in a “reciprocal classification” by using the C-tubule as the reference point, we identified 237 6 classes of subtomograms whose averaged structure shows enhanced detail for the 238 portion of the A-C linker associated with the C-tubule (Figure 3-figure supplements 3-5). 239 The differences in the classes largely reflect the underlying flexibility of the A-C linker 240 about its midpoint. We combined the above two averages by docking them into a lower 241 resolution class-averages, resulting in a structure of the complete A-C linker for each 242 identified class (Figure 3-figure supplement 6). 243 244 Overall, the A-C linker forms a crisscross-shaped structure (Figure 3A, Video 2) that can 245 be divided into five parts: a central trunk region, from which two arms (Arm A, B) and 11 246 two legs (Leg A, B), extend out to the C- and A-tubules, respectively. Arm A forms a 247 longitudinal helical filament with 8 nm periodicity and binds to PF C8 of the C-tubule 248 (Figure 3B). The Arm B inserts into the trough between PFs C9 and C10 of the C-tubule, 249 likely binding to both PFs. Leg A and B fork out from the central trunk towards the A- 250 tubule. The Leg A has a thin rod shape that tilts ~30° towards the proximal end of 251 procentriole and extends about 10 nm to reach and contact PF A6 of A-tubule (Figure 252 3C). Leg B tilts ~30° towards the distal end of procentriole. It spans about 22 nm to 253 connect to both PFs A9 and A10 of the A-tubule. It continues to reach as far as to the 254 outer junction of A- and B-tubules (Figures 3A,B). Lastly, the central trunk can be further 255 divided into three longitudinal filaments bundled into a spiral: f1, f2, f3 (Figures 3A,C). 256 257 Strikingly there is a doughnut-shaped density with the Leg B. The averaged diameter of 258 the ring is ~4 nm and they stack longitudinally with 8 nm periodicity (Figures 3A,B). 259 Previous studies found the conserved centriole components, POC1, to localize to the 260 centriole proximal ends (Keller et al., 2009; Pearson et al., 2009b). Knockdown or 261 deletion of POC1 in several organisms resulted in defective centrioles or basal bodies 262 (Khire et al., 2016; Pearson et al., 2009a; Venoux et al., 2013). Human mutations of 263 either paralog, POC1A or POC1B, cause ciliopathy-like pathologies (Beck et al., 2014; 264 Roosing et al., 2014; Sarig et al., 2012; Shaheen et al., 2012). In Tetrahymena, POC1 is 265 important for maintaining A-C linker integrity. poc1 null mutants display basal body 266 defects ranging from missing or disintegrated triplets to disconnected neighboring triplets 267 and aberrant A-C linkers (Meehl et al., 2016; Pearson et al., 2009a). Interestingly, one of 268 predicted conserved signatures of POC1 is multiple WD40 repeats at its N-terminus 12 269 (Woodland and Fry, 2008), for example, in Chlamydomonas, POC1 is predicted to have 7 270 tandem WD40 repeats at its N-terminus (Keller et al., 2009). A crystal structure of a 7- 271 repeat WD40 β-propeller domain fits remarkably well into the ring density observed on 272 the Leg B (Figures 3D,E). Due to relatively modest resolution of our average map, the 273 precise orientation of the WD40 domain could not be defined. Nevertheless, in the 274 docked model, the WD40 β-propeller makes multiple contacts with the rest of Leg B, 275 consistent with the function of WD40 domain as one of the most abundant protein- 276 protein interaction domains that scaffolds multi-protein complexes (Stirnimann et al., 277 2010). Notably, human ciliopathy mutations in POC1A or POC1B map to the WD40 278 repeats. Thus, combined with these data, our structure suggests that POC1 is likely an 279 integral component of the procentriole A-C linker that bridges neighboring triplets. Its N- 280 terminal WD40 β-propeller domain provides multiple sites for interacting with other A-C 281 linker components. However, in addition to procentriole, POC1 has also been found at 282 other locations in mother centriole and in flagellum (Keller et al., 2009; Pearson et al., 283 2009b). A precise localization of POC1 gene product on the centriole has to wait for 284 future higher resolution structure by cryoET and other studies. 285 286 The Twist of A-C linker Results in Iris Diaphragm Motion of the Procentriole 287 As discussed above, analysis of the A-C linker heterogeneity led to the identification of 4 288 and 6 conformational classes in two classification schemes where either the A- or C- 289 tubule was used as a reference point (Figure 3-figure supplements 1,3). Overlaying 290 projections from these classes, in both schemes, reveals a large and continuous swinging 291 motion of the remaining portion of the linker relative to the reference point (Videos 3,4). 13 292 The swing angle θ (Figure 4A, Table 3) changes up to 10 or 16 degrees respectively in 293 two schemes. Since the proximal region of the mother centriole has a defined length, 294 calculating the weighted average of the swing angle at each longitudinal position along 295 the triplet (Figures 4B,C) shows that two neighboring triplets, by using the A-C linker as 296 a pivot point, twist progressively along the procentriole in a left-handed manner with the 297 thumb pointing towards the MT plus end, as illustrated in Figure 4A. 298 299 To further identify the origin of this twist motion, we overlaid the complete A-C linker 300 for each class obtained by following steps illustrated in Figure 3-figure supplement 6. 301 The two sets of overlaid structures are colored as a “heat map” with increasing 302 temperature from the proximal to the distal position along the longitudinal axis of the 303 centriole. The overlays clearly show the central trunk region of the A-C linker is at the 304 pivot point of this twist motion (Figures 4D,E). Interestingly, the trunk is composed of a 305 bundle of 3 inter-connected filaments, f1, f2 and f3 (Figures 3A-C). Together they form a 306 spiral that likely drives, at least in part, the twist motion of neighboring triplets. 307 308 Finally, to directly visualize the twist of neighboring triplets in the context of the entire 309 procentriole, we placed the averaged triplet structure back into one of the procentriole 310 tomograms by reversing the steps of subtomogram averaging. The result is a longitudinal 311 segment of the procentriole. Proceeding from the proximal towards the distal end (Video 312 5), the 9 triplets rotate concomitantly while the A-C linkers are at the pivot points of the 313 twist. At the distal end, the triplets become more tangential to the circumference of 314 procentriole cylinder. As a consequence, the luminal diameter gradually increases as if 14 315 opening up an iris diaphragm. Similar diaphragm-like motion of the centriole has been 316 observed previously in several organisms, including Chlamydomonas, Tetrahymena and 317 mammalian cells (Anderson, 1972; Li et al., 2012; Meehl et al., 2016; Paintrand et al., 318 1992). Interestingly, a Tetrahymena POC1 null mutation results in a reduced triplet twist 319 angle (Meehl et al., 2016), likely due to defective A-C linker structure. It is likely that 320 this longitudinal twist of triplets is a common structure feature of procentrioles and 321 centrioles. Taken together, our structural analysis of the A-C linker has revealed a 322 characteristic iris diaphragm motion of the procentriole that is accommodated by the twist 323 of a spiral in the central trunk of the A-C linker. 324 325 Intermediate State of the B-tubule during Triplet Assembly 326 Many of our tomograms contain procentrioles attached to mother centrioles that are in the 327 process of assembly (Figure 1B, Video 1). To identify any assembly intermediates and to 328 study the assembly process, we set out to analyze the triplets from these procentrioles. 329 Among 110 tomograms collected, the length of the procentriole triplet varied 330 substantially, with the A-tubule always being the longest tubule, followed by the B- then 331 the C-tubules. Most of the procentriole MTs are straight with a slightly flared end at the 332 MT tip (Figure 5A), consistent with the morphology of slow growing MT and elongating 333 triplets (Guichard et al., 2010; Hoog et al., 2011). The B- and C-tubules are attached to 334 the middle of A- or B-tubules at different heights, respectively, suggesting that the B- and 335 C-tubules initiate assembly by using the A- and B-tubule wall as a template rather than a 336 minus-end template such as the -tubulin ring complex. Once initiated, the B- and C- 337 tubules then extend longitudinally towards both the MT minus and the plus ends. 15 338 Histograms of the measured tubule length in 110 procentrioles show wide distributions 339 for all 3 tubules (Figure 5B), demonstrated by the large standard deviations. This length 340 variation also implies substantial structural heterogeneity in the procentriole triplets 341 where the B- and C-tubules are likely partially assembled. We set out to identify any 342 assembly intermediates using classification. 343 344 We first focused on classification of the B-tubules. Based on 2083 subtomogram volumes 345 from 110 procentrioles, we obtained 8 classes (Figure 5-figure supplement 1). 346 Surprisingly, one of the classes, composed of 157 subtomograms, shows partially 347 assembled B-tubules. The average of this class shows three PFs B1, B2 and B3 at outer 348 A-B junction having more prominent density than the other PFs in the B-tubule. We 349 mapped these 157 subtomograms to find their location in the procentriole. Based on their 350 longitudinal positions, we divided them into 5 classes. These are 1) at the proximal end of 351 the triplet (28.7%), 2) at the distal end (25.5%), 3) a nascent B-tubule where the entire B- 352 tubule is partially assembled (33.8%), 4) extremely short and noisy B-tubule less than 10 353 nm long (7.0%), 5) partial B-tubule structure found in the midst of complete tubule 354 (5.1%), likely these are defective structures or errors in classification. The first three 355 classes comprise 88% of the 157 subtomograms, and the averages are shown in Figure 356 5C. Interestingly, all three averages show an incomplete B-tubules with strong densities 357 for PFs B1~B3. As a check, we directly visualized the identified partial doublets in their 358 procentriole tomograms (Figure 5D), confirming that they are indeed partially assembled 359 doublets at the longitudinal extremity of the triplet. In summary, based on the 360 classification and the mapping analysis, we identified an intermediate in the B-tubule 16 361 assembly at both the polar ends of triplet. It shows the B-tubule initiates lateral expansion 362 from PF B1 at the outer A-B junction towards the luminal side. 363 364 Intermediate State of the C-tubule during Triplet Assembly 365 We applied a similar analysis to the C-tubule and after two rounds of classification, we 366 identified 208 subtomograms from 110 procentrioles that have partially complete C- 367 tubules (Figure 6-figure supplement 1). In contrast to the incomplete doublet, the average 368 of these 208 subtomograms shows PFs at both the inner and the outer B-C junctions with 369 a gap in between. We further mapped these incomplete triplets in their corresponding 370 tomograms. Similar to the B-tubule result, based on the location these triplets can be 371 divided into five classes. Three major classes, including the proximal end, the distal end 372 and the nascent C-tubule where the entire length of the triplet has partially assembled C- 373 tubule, comprise 83% of the 208 subtomograms. Averages from these three major classes 374 shows a consensus structure of incomplete triplet (Figure 6A), where the PFs C8~C10 375 along with the MIP9 at the inner B-C junction are clearly visible while the PFs C1~C3 376 are visible at the outer junction. Interestingly, in all three averages, even though the C- 377 tubule is incomplete, the A-C linker is visible albeit with weaker density likely due to its 378 flexible nature or incomplete assembly. This indicates that the linkage between 379 neighboring triplets has been established at this early stage of the triplet assembly. We 380 further confirmed these averages by directly visualizing corresponding structure in the 381 procentriole tomograms (Figure 6B). Taken together, our analysis has identified C-tubule 382 assembly intermediates at either end of the triplet longitudinal extremity, suggesting that 383 the C-tubule assembles laterally from both the inner and the outer junction in a bi- 17 384 directional fashion. This allows prompt establishing of the A-C linker during the triplet 385 assembly, reinforcing the procentriole as a barrel-shaped structure. 18 386 387 Discussions 388 389 Here, by using cryoET and image analysis, we present procentriole structures from 390 Chlamydomonas reinhardtii. These new structures and findings have implications 391 concerning the mechanism of centriole biogenesis and assembly. 392 393 Comparing the Chlamydomonas Procentriolar Triplet to Other Triplet Structures 394 As important cell organelles, there have been extensive electron microscopic studies on 395 both the procentriole and the centriole. Most were focused on the morphology of these 396 organelles in the context of their life cycles. In recent years, cryoET has been applied to 397 study molecular architecture of centriole and basal body in several organisms, including 398 Chlamydomonas, Drosophila, Trichonympha and mammalian cells (Greenan et al., 2018; 399 Guichard et al., 2013; Li et al., 2012). These have provided insights into mechanism of 400 triplet/doublet assembly and centriole biogenesis. 401 402 In Chlamydomonas, there are marked differences between the procentriole triplet 403 structure reported here and that in the core region (Li et al., 2012). First, a Y-shaped 404 structure observed in the luminal side of core region is absent in procentriole, instead, the 405 space is taken in part by the pinhead structure. Second, the A-C linker structures in these 406 two regions are different. Whereas the procentriole has a crisscross shaped A-C linker 407 structure that links PFs A6, A9/A10 and C8, C9/C10 from its neighboring triplets, in the 408 core region, the A-C linker bridges PF A6 to various positions on neighboring C-tubule 19 409 depending on the longitudinal position (Li et al., 2012). Third, the PF C1 in the 410 procentriole is composed of tubulin having a 4 nm periodicity, while the C1 in distal half 411 of the core region exhibits an 8 nm repeat, indicating a non-tubulin protein. Lastly, even 412 though a number of MIPs are shared in both regions, for example a cone-shaped structure 413 MIP1 at PF A5 and a number of MIPs spanning the luminal side of PFs A1-A3, the 414 overall patterns of MIPs decorated on the triplet are different in these two regions. This is 415 particularly noticeable in the inner junctions of A-B and B-C tubules. All these 416 differences suggest that two sets of proteins are enlisted sequentially during the centriole 417 assembly, controlled by as yet unknown mode of spatial and temporal regulation. For 418 example, both pinhead and the Y-shaped structure are near the inner junction of A-B 419 tubules. As part of the cartwheel structure, the pinhead function as a scaffold for initial 420 establishment of nine triplets in the procentriole. Once the procentriole is fully 421 assembled, the Y-shaped structure will follow to further strengthen the central core 422 region. Structural difference between these two regions has also been observed recently 423 in mammalian centriole (Greenan et al., 2018), but with details that are distinct from 424 those with Chlamydomonas. However, this does suggest that a multiple-step hierarchical 425 spatial and temporal control is likely a common theme of centriole assembly in many 426 eukaryotes. 427 428 The procentriole triplet structure presented in this work also exhibits several notable 429 differences from the triplets from other organisms. For example, in Trichonympha, the 430 pinhead exhibits 17 nm periodicity, while in Chlamydomonas, it is 8 nm. Interestingly, in 431 both organisms the A-C linker has 8.5 nm or 8 nm periodicity, respectively. This 20 432 structural difference might reflect the difference in mechanism of length control as the 433 procentriole in Trichonympha has an exceptionally long cartwheel. 434 435 The Roles of A-C Linker in the Procentriole Assembly 436 In addition to serving as a structural linkage between neighboring triplets, our structure 437 has revealed several important roles for the A-C linker during procentriole assembly. 438 In a Sas6 null mutation in Chlamydomonas, a circular centriole can still form without the 439 cartwheel, albeit at low efficiency (Nakazawa et al., 2007). Among these cartwheel-less 440 centrioles, the majority remains nine-fold symmetric, but other symmetries have been 441 observed as well, suggesting that, at least in some organism, the cartwheel does not fully 442 account for setting up the nine-fold symmetry in the centriole. Interestingly, in vitro 443 assembly of CrSas6 results in cartwheel-like structures with the majority having eight and 444 nine-fold symmetry (Guichard et al., 2017; Hilbert et al., 2016). Likewise, mammalian 445 cells can form centrioles de novo without the central hub formed by Sas6, although the 446 assembly is less efficient and more error-prone than the wild type (Wang et al., 2015). 447 These observations have led to speculation that other factor(s) might be needed for 448 efficient and robust centriole assembly (Gönczy, 2012; Hirono, 2014). Our finding that 449 the A-C tubule linkage is established at an early stage of procentriole assembly while the 450 triplets are still elongating provides direct evidence supporting a model that the A-C 451 linker helps establish the 9-fold symmetry. In our model, A-C linker can accomplish this 452 simply by restricting the angle between neighboring MT triplets to be approximately 140 453 degree as needed for forming a nonagon. Thus in wild type cells, the combination of both 454 Sas6 oligomerization and a stabilizing linkage between neighboring triplets that has a 21 455 defined angle will provide a robust mechanism for establishing the nine-fold symmetry. 456 In addition to restricting the angle between neighboring triplets, the A-C linker also plays 457 a structural role that connects nascent incomplete triplets therefore bolstering the nascent 458 procentriole structure during its assembly. 459 460 Instead of being a static structure, the A-C linker displays substantial twist motion along 461 the triplet wall. This can accommodate an apparent left-handed iris diaphragm movement 462 that progresses along the longitudinal direction. The twist observed both in the 463 procentriole and in the core region results in change of angle for each triplet relative to 464 the cylindrical circumference, from the most acute angle at the very proximal side 465 towards more tangential in the distal region. As a result, the inner diameter of the 466 centriole gradually increases. This change may facilitate multiple processes: First, the 467 acute angle of the triplet at the proximal end places the A-tubule including the pinhead 468 close to the cartwheel, likely reflecting its attachment as the earliest stages of biogenesis. 469 Second, the gradual twisting will increase the exposed surface area, facilitating 470 recruitment of other centriole components, such as the subdistal and the distal 471 appendages. Third, the tangential angle of the triplet at the distal end of basal body must 472 match the angle of the axoneme doublets, where the basal body is ready to serve as a 473 template for the flagellum assembly. Lastly, this iris diaphragm motion of the MT triplet 474 might also have impact on controlling the procentriole length by twisting the pinheads 475 away from the hub while the luminal diameter becomes wider. This suggests that the 476 radial spokes will be under strain of pulling force that continually increases along the 477 procentriole. This pinhead movement can be directly visualized in Video 5. As part of the 22 478 radial spoke, the C-terminus of Sas6 and Cep135/Bld10 are predicted to have highly 479 coiled-coil structures with intrinsically elastic properties. It is appealing to envision that 480 this strain and elasticity will reach to a balance point once the MT triplets elongate to 481 certain length, suggesting an intrinsic length control mechanism due to the iris diaphragm 482 motion of the triplet. 483 484 A Model for the Procentriole Triplet Assembly 485 Previous EM studies on mammalian centriole and basal body showed the triplet assembly 486 was a non-uniform process where the A-tubule assembled first, followed by sequential B- 487 and C-tubule assembly. During the tubule elongation, their lengths varied substantially 488 (Anderson and Brenner, 1971; Guichard et al., 2010). Furthermore, by analyzing the 489 nascent procentrioles, Anderson and Brenner observed the B- and C-tubule first emerged 490 as sheets laterally attached to the A- and B-tubule respectively. Consistent with these 491 findings, by analyzing a population of procentrioles, we found the average length of three 492 tubules in the triplet varies substantially. In addition, we have identified two triplet 493 intermediates, showing that the B-tubule initiates assembly uni-directionally from the 494 outer A-B junction. By contrast, the C-tubule assembles laterally in a bi-directional 495 fashion, from both the inner and the outer B-C junctions. These lateral expansions are 496 concomitant with the longitudinal elongation of the triplets at their extremities. Based on 497 the analysis, we propose a possible model for the triplet assembly during the procentriole 498 biogenesis (Figure 7). In our model, the A-tubule first emerges and anchors to the 499 pinhead of cartwheel. It continues to grow. Then the B-tubule initially assembles at the 500 outer A-B junction by making an unusual tubulin-tubulin lateral interface and branching 23 501 out at the PF A10. This branch laterally expands from the outer junction to reach the 502 luminal inner junction where the PF B10 and other components including the MIPs 5~6 503 seal and complete the B-tubule. Unlike the B-tubule, the C-tubule initiates assembly on 504 both junctions. PFs C1 and C10 branch out from the B-tubule at the inner and the outer 505 junctions, respectively. These two branches expand laterally in a bi-directional fashion. 506 At this point, while the C-tubule is still incomplete, the newly assembled PFs C8~C10 are 507 readily available to provide a site for the A-C linker that connects the C-tubule to the A- 508 tubule of its (N-1) neighbor triplet. Equivalently, its A-tubule will connect to the C-tubule 509 of its (N+1) neighbor triplet. As a result, all nine triplets form an inter-connected barrel 510 even though some triplets are still incomplete. In the last step, the gap is filled by the PFs 511 in the C-tubule. This completes the assembly of all three tubules at this longitudinal 512 point, while the lateral expansion and the longitudinal elongation will continue in both 513 extremities of the triplet. 514 515 Unlike the rapid kinetics of the MT self-assembly, the dynamics of growing MT triplets 516 in both procentriole and centriole are slow but stable compared to the MT (Kochanski 517 and Borisy, 1990). Although we cannot exclude the possibility that there is a fraction of 518 partial B- or C-tubules that is due to their lability during the purification process, based 519 on the observed tubule length variation, the morphology of the MT ends, the stability of 520 centriole MT compared to the canonical MT and the consistent pattern of the partial B- 521 and C-tubule observed at the growing ends (Figures 5C-D, 6A-B), the identified 522 intermediate structures likely represent two snapshots of an underpinning mechanism for 523 the MT triplet assembly, which is tightly controlled by a set of factors. These include 24 524 structural components such as MIPs identified in this work or other microtubule 525 associated proteins (MAPs) (Sharma et al., 2016; Zheng et al., 2016). In addition, other 526 regulatory components such as, acetyltransferase or glutamylase will alter the 527 polymerization properties and stability of the MT triplet as well (Yu et al., 2015). 528 Interestingly, many MIPs identified in this work and in mammalian centrioles (Greenan 529 et al., 2018) are localized at the junction of A-B or B-C tubules in the triplet. For 530 example, MIPs 5~6 are found at the inner A-B junction, while MIPs 9~11 are at the inner 531 B-C tubules junction. This asymmetric localization of the MIPs will likely modulate the 532 assembly kinetics of the MT PFs. Furthermore, since the association of MIPs to triplet 533 might be concomitant with the MT elongation, the MIPs’ own assembly kinetics on MT 534 will in turn have directly impact on the assembly of the triplet. In the future, with 535 enhanced dataset size and improved algorithm, it is likely that more assembly 536 intermediates will be identified in Chlamydomonas and in other organisms. This will 537 provide insight into the centriole assembly mechanism with improved spatial and 538 temporal resolution. Our study presented here has paved the way in this direction. 539 540 Finally, like many large and functional biological assemblages, the procentriole structure 541 presented in this work is an example demonstrating the extensive diversity and 542 heterogeneity of a biological system in situ, including composition heterogeneity such as 543 the incomplete B- and C-tubules in triplet and conformation heterogeneity such as the 544 continuous twisting of the A-C linker. CryoET is well poised as a unique and 545 unparalleled tool to reveal this phenomenon and to provide high-resolution insights for 546 many biological mechanisms waiting to be explored. 547 25 548 549 550 Acknowledgement 551 collection, Matt Harrington for supporting the computational infrastructure, Hiroaki 552 Ishikawa for discussion on the biology of Chlamydomonas, centrioles and basal bodies, 553 Tom Goddart for help with UCSF Chimera program. We are grateful to many of our 554 colleagues for critical reading of the manuscript and for the encouragements. This work is 555 supported in part by NIH grants GM031627 (D.A.A.), GM118099 (D.A.A.), PO1 556 GM105537 (D.A.A.), GM113602 (W.F.M.) and by HHMI (D.A.A.) and by the Spanish 557 AEI/FEDER (SAF2017-84565-R) (J.-J.F.) and by Fundacion Ramon Areces (J.-J.F.). We thank Michael Braunfeld and Shawn Zheng for advice on tomography data 558 559 The EM structures have been deposited in the EMDB with the following accession 560 numbers: EMD-9167, EMD-9168, EMD-9169, EMD-9170, EMD-9171, EMD-9172, 561 EMD-9173, EMD-9174. 562 26 563 564 Declaration of Interests 565 The authors declare no competing interests. 566 567 568 569 27 570 Figure Legends 571 572 Figure 1. Tomographic Reconstruction and Subtomogram Averaging of the MT 573 Triplet in Centriole and Procentriole 574 (A) A slice from the reconstructed tomogram showing a pair of mother centrioles, also 575 known as basal bodies. The central hub as part of the cartwheel is marked by a yellow 576 asterisk. The centriole is partitioned longitudinally into three regions. The procentriole 577 region spans 100 nm at the proximal end of the mother centriole. The central core region 578 spans 250 nm. The distal region spans 150 nm where the triplets become doublets before 579 reaching the transition zone (TZ). PSF: proximal striated fibers; DSF: distal striated 580 fibers; rMT: rootlet MT. 581 (B) A slice from the reconstructed tomogram of procentriole attached to the mother 582 centriole via the rootlet MT. The central hub, radial spokes and pinheads are clearly 583 visible, as well as protofilaments in some MT triplets. RS: radial spoke; PH: pinhead; 584 rMT: rootlet MT. 585 (C) Subtomogram average of the MT triplet from both the proximal region of mother 586 centrioles and the procentrioles. 587 (D) A model of procentriole viewed from its distal end. The model is generated by 588 docking the subtomograms average in (C) into the tomogram volume in (B). The A- B- 589 and C-tubules and the A-C linker are labeled. The central hub and the radial spokes are 590 depicted schematically. PH: pinhead. 591 592 Figure 2. Identifying Non-tubulin Components in Procentriole MT Triplet 593 (A) The A-tubule structure. A model of 13-pf MT (red) are docked into the subtomogram 594 average of the A-tubule. 4 MIPs identified on the A-tubule are highlighted in colors. 595 (B) The B-tubule structure. A model of 10-pf MT (blue) are docked into the 596 subtomogram average of the B-tubule. 4 MIPs identified on the B-tubule are highlighted 597 in colors. 598 (C) The C-tubule structure. A model of 10-pf MT (green) are docked into the 599 subtomogram average of the C-tubule. 3 MIPs found at the B-C inner junction are 600 highlighted in colors. 28 601 (D) Subtomogram averaging of the pinhead structure. The pinhead is highlighted in pink 602 color. Two dashed arrows indicate viewing directions for two images in the side view on 603 the right. The arrows indicate PinB, PinF1 and PinF2, respectively. The arrowheads mark 604 the tip of PinB. 605 606 Figure 3. The Structure of A-C Linker 607 (A) A top view of the A-C linker structure with its linked A- and C-tubules. The A-C 608 linker is highlighted in red. Three filamental structures f1, f2 and f3 making up the 609 central trunk running in longitudinal direction are marked with *. 610 (B) The A-C linker viewed from outside of procentriole. The dashed square highlights the 611 area that will be viewed in (D). 612 (C) The A-C linker viewed from the luminal side of procentriole. Three filaments f1, f2 613 and f3 that make up the central trunk are marked with vertical dashed lines. They are 614 interconnected and form a spiral. 615 (D and E) Close up views of the Leg B in the A-C linker show a possible location of 616 POC1. (D) is in side view from outside of the procentriole. (E) is in top view from the 617 distal end of the procentriole. An atomic model of WD40 β-propeller domain (PDB ID 618 1S4U) is fitted into the doughnut-shaped density in Leg B. The red * indicate connecting 619 points of this WD40 domain to the rest of Leg B structure and to the f3 in central trunk. 620 621 Figure 4. The Twist of A-C Linker Drives Procentriole Triplets in Iris Diaphragm 622 Motion 623 (A) Schematic diagram of two triplets connected by the A-C linker. θ is defined as the 624 angle between the Arm A and the Leg B. The results of mapping and angle measurement 625 as shown in (B-C, Table 3) are consistent with a twisting motion of two triplets as 626 indicated by two curved arrows when moving longitudinally from proximal to distal 627 direction. The pivot point of this twist is at the A-C linker as indicated by an arrow. 628 (B and C) Weighted averages of twist angle along the triplet wall in two classification 629 schemes shown in Figure 3-figure supplements 1,3. Six points along 90 nm longitudinal 630 length of centriole are sampled starting from the proximal end. The weighted average of 631 twist angle T at point i is defined as: Ti = ∑ θj*Nij / ∑ Nij 29 632 Nij: number of subtomogram belong to class j at point i. 633 θj: twist angle as defined in (A) for class j. 634 (D and E) Overlay of two sets A-C linker structure based on two classification schemes 635 (Figure 3-figure supplements 1,3). Based on its longitudinal position, each structure is 636 colored following a “heat map” scheme. Arrows point to the pivot point, the central 637 trunk. 638 639 Figure 5. Variation of the Tubule Length in Procentriole Triplets and Detecting the 640 B-tubule Intermediate. 641 (A) Two examples of procentriole in tomogram. In each, two triplets marked by yellow 642 rectangle box are selected to depict tubule length variation as shown on the right in the 643 longitudinal view. The distal end of each tubule is indicated by an arrowhead. Black 644 arrows indicate examples of the slightly flared morphology at the growing end of the 645 tubule. Scale bar: 100 nm. 646 (B) Histogram showing length distribution for A-, B- and C-tubules. For each tubule, the 647 average length, its standard deviation and the number of measurement are indicated. 648 (C) The averages of incomplete B-tubule from the proximal, distal and nascent doublet, 649 respectively. Each image is a z-projection of doublet of 9.6 nm long. The numbers of 650 subtomogram in each class and their percentages are indicated. The arrowhead indicates 651 PF B1. 652 (D) Examples of incomplete B-tubule in procentriole tomogram from three classes. The 653 incomplete B-tubules are marked by yellow arrowheads. 654 655 Figure 6. Detecting the C-tubule Intermediate 656 (A) The averages of subset of subtomogram with incomplete triplet at proximal, distal 657 end and from nascent triplet. The images are z-projection of the average where the triplet 658 length is 9.6 nm. The numbers of subtomogram in each class and their percentages are 659 indicated. The arrowhead points to PF C1. The arrow points to the A-C linker. 660 (B) Examples of incomplete C-tubule in procentriole tomogram in three classes. The 661 incomplete C-tubules are marked by yellow arrowheads. The A-C linkers are marked by 662 yellow arrows. 30 663 664 Figure 7. A Model for the MT Triplet Assembly in the Procentriole 665 In the model, the triplet assembly can be divided into 5 sequential steps: 1) A-tubule 666 emerges and anchors to the pinhead in the cartwheel. It continues to elongate 667 longitudinally. 2) B-tubule branches out at PF A10 in outer A-B junction. 3) B-tubule 668 expands laterally from outer A-B junction toward the luminal inner A-B junction where it 669 completes the B-tubule. 4) C-tubule branches out at both inner and outer B-C junctions. 670 Meanwhile, the A-C linker is established. 5) The bi-directional lateral expansion of C- 671 tubule completes the triplet assembly at this longitudinal position. 672 673 674 31 675 Supplemental Figure Legends 676 Figure 1-figure supplement 1 Assessing the Resolutions of Subtomogram Average by 677 Fourier Shell Correlation (FSC) Method. 678 FSC curves are for the MT triplet (black, 23.0 Å), the A-tubule (red, 21.4 Å), the B-tubule 679 (blue, 22.2 Å) and the C-tubule (green, 22.2 Å). All resolution assessments are carried out 680 in a “gold-standard” Scheme using FSC 0.143 criterion. 681 682 Figure 1-figure supplement 2 Comparing Subtomogram Averages from a Subset of 683 Data. 684 (A) The triplet in blue on left is an average based on 10854 subtomograms from the 685 proximal end of mother centrioles. 686 (B) The triplet in light green on right is an average based on 2083 subtomograms from 687 the 110 procentrioles that remain attached to the mother centrioles. 688 The differences between these two structures are indicated by red arrows and by red 689 asterisks, respectively. 690 691 Figure 2-figure supplements 1-11 The non-tubulin Procentriole Components Associated 692 to the MT Triplet. 693 The figures show the binding pattern of 11 MIPs and their periodicity along the MT wall. 694 In each figure, image on the left is a longitudinal cross section of the average of 695 corresponding tubule. The middle image is a surface-rendered map with the structure of 696 corresponding MIP highlighted in color. Image on the right is a cartoon model of the 697 tubule. The dashed line indicates the location where the cross section goes through. The 698 red arrow indicates viewing direction of the structure shown in the middle image. They 699 are also summarized in Table 2. 700 701 Figure 2-figure supplements 12. MIP9 forms a non-tubulin filament at the inner 702 junction of B- and C-tubule. 703 The MIP9 density is highlighted in blue. A model of MT protofilament in green is fit into 704 the MIP9 density. On the right panel, only the MIP9 filament is shown in two orthogonal 32 705 views, the rest triplet structure is removed for clarity. The MIP9 density differs 706 substantially from a canonical MY protofilament. 707 708 Figure 2-figure supplements 13. The anisotropic resolution in the pinhead structure. 709 The distal tip of PinB is flexible exhibiting the lowest resolution in the average 710 (arrowhead). The local resolution is estimated by blocres in Bsoft. 711 712 Figure 2-figure supplements 14. FSC curve for the averaged pinhead structure after 713 classification (resolution: 23.1 Å). 714 715 Figure 3-figure supplement 1. Schematic diagram shows the classification process for 716 identifying intact A-C linker associated to the A-tubule. 717 The number of subtomogram in each class and the 4 classes identified with intact A-C 718 linker are indicated. Total 3992 subtomograms from 4 classes are identified with relative 719 intact A-C linker. Among them, 2917 (73.1%) are from the proximal end of mother 720 centrioles. 1075 (26.9%) are from the procentrioles. 721 722 Figure 3-figure supplement 2. Subtomogram average of the A-tubule with associated A- 723 C linker. 724 This is based on 3992 subtomograms (Class 1-4) identified in classification scheme in 725 Figure 3-figure supplement 1. 726 727 Figure 3-figure supplement 3. Schematic diagram showing classification process for 728 identifying intact A-C linker associated to the C-tubule. 729 The number of subtomogram in each class and the 6 classes identified with relative intact 730 A-C linker are indicated. The Class 1 is not used for further refinement and average due 731 to its low quality. Total 3478 subtomograms from 6 classes are identified with relative 732 intact A-C linker. Among them, 2611 (75.1%) are from the proximal end of mother 733 centrioles. 867 (24.9%) are from the procentrioles. 734 33 735 Figure 3-figure supplement 4. Subtomogram average of the C-tubule with associated A- 736 C linker. 737 This is based on 3245 subtomograms (Class 2-6) identified in classification scheme in 738 Figure 3-figure supplement 3. 739 740 Figure 3-figure supplement 5. FSC curves for the averaged A-tubule and the C-tubule 741 with more intact A-C linker after classification. 742 The A-tubule with associated A-C linker (red, 23.1 Å), the C-tubule with associated A-C 743 linker (green, 23.1 Å). 744 745 Figure 3-figure supplement 6. The scheme for Reconstruction of Full A-C Linker 746 Structure. 747 By using this scheme, based on the 4 and 6 classes identified (illustrated in Figure 3- 748 figure supplements 1,3), total 10 structures with full A-C linker are reconstructed 749 showing different degrees of the twist motion. The Class 2 (568 subtomograms) 750 identified in the second classification scheme (Figure 3-figure supplement 3) is used as 751 an example shown in Figure 3. 752 753 Figure 5-figure supplement 1. Classification and Identification of Partially Assembled 754 B-tubule. 755 Total 157 subtomograms with incomplete doublet are identified as the result of 756 classification. Images displayed are the central z-sections from the average of each class. 757 The number of subtomogram in each class is indicated. 758 759 Figure 6-figure supplement 1. Classification and Identification of Partially Assembled 760 C-tubule. 761 Total 208 subtomograms with incomplete triplet are identified as the result of 762 classification. Images displayed are the central z-sections from the average of each class. 763 The number of subtomogram in each class is indicated. 764 765 766 34 767 768 769 770 Tables Table 1. Summary of Statistics of Subtomogram Averages Structure Number of subtomogram Resolution (Å) Description EMDB ID # 1 12937 23.0 Triplet structure 9167 2 12517 21.4 A-tubule 9168 3 12179 22.3 B-tubule 9169 4 12075 22.3 C-tubule 9170 5 4763 23.1 Pinhead structure with its associated partial A-tubule 9171 6 3992 23.1 A-tubule with more complete A-C linker after classification 9172 7 3245 23.1 C-tubule with more complete A-C linker after classification 9173 Full A-C linker structure with Aand C-tubule it links to 9174 8* 771 772 773 774 775 776 Composite map derived from EMD-9172 and EMD-9173 All resolutions are assessed by using the “gold standard” scheme with FSC 0.143 criterion. * This is a composite map by merging EMD-9172 and EMD-9173 onto the map of one of classes, class#2 (See Figure 3-figure supplement 6) 35 777 778 Table 2. Estimated Molecular Weight of MIPs associated to Triple Name 779 780 781 782 783 784 Estimated MW Periodicity (KD) Location MIP1 45 8 nm cone-like structure associated to the lumen side of A5 MIP2 n/a 4 nm trellis-like structure spanning laterally from A11 to A13 MIP3 24 8 nm laterally link A13, A1 and A2 in the lumen of Atubule MIP4 28 8 nm laterally link A2 and A3 in the lumen of A-tubule MIP5 † 74 8 nm inner junction of A, B-tubule, laterally link A1, A2 and B10 MIP6 27 8 nm inner junction of A, B-tubule, laterally link A1, A13, B9, B10 MIP7 n/a 8 nm fin-like filamental structures running longitudinally along B5, B4 B3 in the lumen of Btubule MIP8 15 8 nm outside of B-tubule but in the lumen of C-tubule, laterally link B6 and B7 MIP9 92 8 nm inner junction of B, C-tubule, link B8, B9 and C10 MIP10 19 8 nm inner junction of B, C-tubule, link B7 and C10 MIP11 6 8 nm laterally crosslink C9 and C10 in the lumen of Ctubule Protein density = 0.849 Dalton/Å3 Orange: MIPs associated to the A-tubule of triplet Blue: MIPs associated to the B-tubule of triplet Green: MIPs associated to the C-tubule of triplet 36 785 786 787 Table 3. A. Measurement of Twist Angle at A-C Linker Class θ Angle (degree) 788 789 790 791 792 793 794 795 I 93 II 90 III 87 IV 83 The assignment for each class is following convention defined in Figures 4D and it is based on the resulting 4 classes in the first classification scheme, illustrated in Figure 3figure supplement 1. In the scheme, the A-tubule is used as a reference point. The angle θ is defined as the angle between the Arm A and the Leg B as illustrated in Figure 4A. B. Measurement of Twist Angle at A-C Linker Class θ Angle (degree) 796 797 798 799 800 801 802 803 I 96 II 93 III 90 IV 89 V 84 VI 80 The assignment for each class is following convention defined in Figures 4E and it is based on the resulting 6 classes in the second classification scheme, illustrated in Figure 3figure supplement 3. In the scheme, the C-tubule is used as a reference point. The angle θ is defined as the angle between the Arm A and the Leg B as illustrated in Figure 4A. 37 804 Video 1 An Aligned Tomography Tilt Series of NFAp Showing a Pair of Mother 805 Centriole and the Attached Procentrioles. 806 807 Video 2 Surface Rendered Structure of the A-C Linker. 808 809 Video 3 Swing Motion of the A-C Linker and its Linked C-tubule. The A-tubule on the 810 left is static and is used as a reference point to show the motion. 811 812 Video 4 Swing Motion of the A-C Linker and its Linked A-tubule. The C-tubule on the 813 right is static and is used as a reference point to show the motion. 814 815 Video 5 An Iris-Diaphragm Motion of the Procentriole Triplets.. A longitudinal segment 816 of the procentriole is generated by putting the averaged triplet structure back into one of 817 the procentriole tomograms. The movie is displayed as moving through the longitudinal 818 direction from the proximal to the distal end then backwards. The triplets twist 819 progressively along the procentriole in a left-handed chirality with the thumb pointing 820 towards the distal end. Two yellow arrows mark the two A-C linkers. They appear at the 821 distal end of the longitudinal segment. 822 823 38 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 References: Andersen, J.S., C.J. Wilkinson, T. Mayor, P. Mortensen, E.A. Nigg, and M. Mann. 2003. Proteomic characterization of the human centrosome by protein correlation profiling. Nature. 426:570-574. Anderson, R.G. 1972. The three-dimensional structure of the basal body from the rhesus monkey oviduct. J. Cell Biol. 54:246-265. Anderson, R.G., and R.M. Brenner. 1971. The formation of basal bodies (centrioles) in the Rhesus monkey oviduct. J. Cell Biol. 50:10-34. Banterle, N., and P. Gonczy. 2017. Centriole Biogenesis: From Identifying the Characters to Understanding the Plot. Annu. Rev. Cell. Dev. Biol. 33:23-49. Beck, B.B., J.B. Phillips, M.P. Bartram, J. Wegner, M. Thoenes, A. Pannes, J. Sampson, R. Heller, H. Gobel, F. Koerber, A. Neugebauer, A. Hedergott, G. Nurnberg, P. Nurnberg, H. Thiele, J. Altmuller, M.R. Toliat, S. Staubach, K.M. Boycott, E.M. Valente, A.R. Janecke, T. Eisenberger, C. Bergmann, L. Tebbe, Y. Wang, Y. Wu, A.M. Fry, M. Westerfield, U. Wolfrum, and H.J. Bolz. 2014. Mutation of POC1B in a severe syndromic retinal ciliopathy. Hum. Mutat. 35:1153-1162. Carvalho-Santos, Z., P. Machado, P. Branco, F. Tavares-Cadete, A. Rodrigues-Martins, J.B. Pereira-Leal, and M. Bettencourt-Dias. 2010. Stepwise evolution of the centriole-assembly pathway. J. Cell Sci. 123:1414-1426. Cavalier-Smith, T. 1974. Basal body and flagellar development during the vegetative cell cycle and the sexual cycle of Chlamydomonas reinhardii. J. Cell Sci. 16:529-556. Dutcher, S.K., N.S. Morrissette, A.M. Preble, C. Rackley, and J. Stanga. 2002. Epsilontubulin is an essential component of the centriole. Mol. Biol. Cell. 13:3859-3869. Dutcher, S.K., and E.C. Trabuco. 1998. The UNI3 gene is required for assembly of basal bodies of Chlamydomonas and encodes delta-tubulin, a new member of the tubulin superfamily. Mol. Biol. Cell. 9:1293-1308. Geimer, S., and M. Melkonian. 2004. The ultrastructure of the Chlamydomonas reinhardtii basal apparatus: identification of an early marker of radial asymmetry inherent in the basal body. J. Cell Sci. 117:2663-2674. Gönczy, P. 2012. Towards a molecular architecture of centriole assembly. Nature reviews Molecular cell biology. Greenan, G.A., B. Keszthelyi, R.D. Vale, and D.A. Agard. 2018. Insights into centriole geometry revealed by cryoTomography of doublet and triplet centrioles. Elife. 7. Guichard, P., D. Chrétien, S. Marco, and A.-M. Tassin. 2010. Procentriole assembly revealed by cryo-electron tomography. The EMBO journal. 29:1565-1572. Guichard, P., A. Desfosses, A. Maheshwari, V. Hachet, C. Dietrich, A. Brune, T. Ishikawa, C. Sachse, and P. Gonczy. 2012. Cartwheel architecture of Trichonympha basal body. Science. 337:553. Guichard, P., V. Hachet, N. Majubu, A. Neves, D. Demurtas, N. Olieric, I. Fluckiger, A. Yamada, K. Kihara, Y. Nishida, S. Moriya, M.O. Steinmetz, Y. Hongoh, and P. Gonczy. 2013. Native architecture of the centriole proximal region reveals features underlying its 9-fold radial symmetry. Curr. Biol. 23:1620-1628. Guichard, P., V. Hamel, and P. Gonczy. 2018. The Rise of the Cartwheel: Seeding the Centriole Organelle. Bioessays. 40:e1700241. 39 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 Guichard, P., V. Hamel, M. Le Guennec, N. Banterle, I. Iacovache, V. Nemcikova, I. Fluckiger, K.N. Goldie, H. Stahlberg, D. Levy, B. Zuber, and P. Gonczy. 2017. Cell-free reconstitution reveals centriole cartwheel assembly mechanisms. Nat Commun. 8:14813. Hilbert, M., A. Noga, D. Frey, V. Hamel, P. Guichard, S.H. Kraatz, M. Pfreundschuh, S. Hosner, I. Fluckiger, R. Jaussi, M.M. Wieser, K.M. Thieltges, X. Deupi, D.J. Muller, R.A. Kammerer, P. Gonczy, M. Hirono, and M.O. Steinmetz. 2016. SAS6 engineering reveals interdependence between cartwheel and microtubules in determining centriole architecture. Nat. Cell Biol. 18:393-403. Hiraki, M., Y. Nakazawa, R. Kamiya, and M. Hirono. 2007. Bld10p constitutes the cartwheel-spoke tip and stabilizes the 9-fold symmetry of the centriole. Current biology : CB. 17:1778-1783. Hirono, M. 2014. Cartwheel assembly. Philos Trans R Soc Lond B Biol Sci. 369. Hodges, M.E., N. Scheumann, B. Wickstead, J.A. Langdale, and K. Gull. 2010. Reconstructing the evolutionary history of the centriole from protein components. J. Cell Sci. 123:1407-1413. Hoog, J.L., S.M. Huisman, Z. Sebo-Lemke, L. Sandblad, J.R. McIntosh, C. Antony, and D. Brunner. 2011. Electron tomography reveals a flared morphology on growing microtubule ends. J. Cell Sci. 124:693-698. Keller, L.C., S. Geimer, E. Romijn, J. Yates, I. Zamora, and W.F. Marshall. 2009. Molecular architecture of the centriole proteome: the conserved WD40 domain protein POC1 is required for centriole duplication and length control. Mol. Biol. Cell. 20:1150-1166. Keller, L.C., E.P. Romijn, I. Zamora, J.R. Yates, and W.F. Marshall. 2005. Proteomic analysis of isolated chlamydomonas centrioles reveals orthologs of ciliary-disease genes. Current biology : CB. 15:1090-1098. Khire, A., K.H. Jo, D. Kong, T. Akhshi, S. Blachon, A.R. Cekic, S. Hynek, A. Ha, J. Loncarek, V. Mennella, and T. Avidor-Reiss. 2016. Centriole Remodeling during Spermiogenesis in Drosophila. Curr. Biol. 26:3183-3189. Kilburn, C.L., C.G. Pearson, E.P. Romijn, J.B. Meehl, T.H. Giddings, B.P. Culver, J.R. Yates, and M. Winey. 2007. New Tetrahymena basal body protein components identify basal body domain structure. J. Cell Biol. 178:905-912. Kitagawa, D., I. Vakonakis, N. Olieric, M. Hilbert, D. Keller, V. Olieric, M. Bortfeld, M.C. Erat, I. Fluckiger, P. Gonczy, and M.O. Steinmetz. 2011. Structural basis of the 9-fold symmetry of centrioles. Cell. 144:364-375. Kochanski, R.S., and G.G. Borisy. 1990. Mode of centriole duplication and distribution. J. Cell Biol. 110:1599-1605. Kong, D., V. Farmer, A. Shukla, J. James, R. Gruskin, S. Kiriyama, and J. Loncarek. 2014. Centriole maturation requires regulated Plk1 activity during two consecutive cell cycles. J. Cell Biol. 206:855-865. Li, J.B., J.M. Gerdes, C.J. Haycraft, Y. Fan, T.M. Teslovich, H. May-Simera, H. Li, O.E. Blacque, L. Li, C.C. Leitch, R.A. Lewis, J.S. Green, P.S. Parfrey, M.R. Leroux, W.S. Davidson, P.L. Beales, L.M. Guay-Woodford, B.K. Yoder, G.D. Stormo, N. Katsanis, and S.K. Dutcher. 2004. Comparative genomics identifies a flagellar and basal body proteome that includes the BBS5 human disease gene. Cell. 117:541-552. 40 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 Li, S., J.J. Fernandez, W.F. Marshall, and D.A. Agard. 2012. Three-dimensional structure of basal body triplet revealed by electron cryo-tomography. EMBO J. 31:552-562. Matsuura, K., P.A. Lefebvre, R. Kamiya, and M. Hirono. 2004. Bld10p, a novel protein essential for basal body assembly in Chlamydomonas: localization to the cartwheel, the first ninefold symmetrical structure appearing during assembly. 165:663-671. Meehl, J.B., B.A. Bayless, T.H. Giddings, Jr., C.G. Pearson, and M. Winey. 2016. Tetrahymena Poc1 ensures proper intertriplet microtubule linkages to maintain basal body integrity. Mol. Biol. Cell. 27:2394-2403. Muller, H., D. Schmidt, S. Steinbrink, E. Mirgorodskaya, V. Lehmann, K. Habermann, F. Dreher, N. Gustavsson, T. Kessler, H. Lehrach, R. Herwig, J. Gobom, A. Ploubidou, M. Boutros, and B.M. Lange. 2010. Proteomic and functional analysis of the mitotic Drosophila centrosome. EMBO J. 29:3344-3357. Nakazawa, Y., M. Hiraki, R. Kamiya, and M. Hirono. 2007. SAS-6 is a Cartwheel Protein that Establishes the 9-Fold Symmetry of the Centriole. Curr. Biol. 17:2169-2174. Nigg, E.A., and J.W. Raff. 2009. Centrioles, centrosomes, and cilia in health and disease. Cell. 139:663-678. O'Toole, E.T., and S.K. Dutcher. 2014. Site-specific basal body duplication in Chlamydomonas. Cytoskeleton (Hoboken). 71:108-118. Paintrand, M., M. Moudjou, H. Delacroix, and M. Bornens. 1992. Centrosome organization and centriole architecture: their sensitivity to divalent cations. J. Struct. Biol. 108:107-128. Pearson, C.G., T.H. Giddings, and M. Winey. 2009a. Basal body components exhibit differential protein dynamics during nascent basal body assembly. Mol. Biol. Cell. 20:904-914. Pearson, C.G., D.P.S. Osborn, T.H. Giddings, P.L. Beales, and M. Winey. 2009b. Basal body stability and ciliogenesis requires the conserved component Poc1. J. Cell Biol. 187:905-920. Reiter, J.F., and M.R. Leroux. 2017. Genes and molecular pathways underpinning ciliopathies. Nat. Rev. Mol. Cell Biol. 18:533-547. Roosing, S., I.J. Lamers, E. de Vrieze, L.I. van den Born, S. Lambertus, H.H. Arts, P.B.S. Group, T.A. Peters, C.B. Hoyng, H. Kremer, L. Hetterschijt, S.J. Letteboer, E. van Wijk, R. Roepman, A.I. den Hollander, and F.P. Cremers. 2014. Disruption of the basal body protein POC1B results in autosomal-recessive cone-rod dystrophy. Am. J. Hum. Genet. 95:131-142. Sarig, O., S. Nahum, D. Rapaport, A. Ishida-Yamamoto, D. Fuchs-Telem, L. Qiaoli, K. Cohen-Katsenelson, R. Spiegel, J. Nousbeck, S. Israeli, Z.U. Borochowitz, G. Padalon-Brauch, J. Uitto, M. Horowitz, S. Shalev, and E. Sprecher. 2012. Short stature, onychodysplasia, facial dysmorphism, and hypotrichosis syndrome is caused by a POC1A mutation. Am. J. Hum. Genet. 91:337-342. Scheres, S.H. 2016. Processing of Structurally Heterogeneous Cryo-EM Data in RELION. Methods Enzymol. 579:125-157. Shaheen, R., E. Faqeih, H.E. Shamseldin, R.R. Noche, A. Sunker, M.J. Alshammari, T. Al-Sheddi, N. Adly, M.S. Al-Dosari, S.G. Megason, M. Al-Husain, F. AlMohanna, and F.S. Alkuraya. 2012. POC1A truncation mutation causes a 41 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 ciliopathy in humans characterized by primordial dwarfism. Am. J. Hum. Genet. 91:330-336. Sharma, A., A. Aher, N.J. Dynes, D. Frey, E.A. Katrukha, R. Jaussi, I. Grigoriev, M. Croisier, R.A. Kammerer, A. Akhmanova, P. Gonczy, and M.O. Steinmetz. 2016. Centriolar CPAP/SAS-4 Imparts Slow Processive Microtubule Growth. Dev. Cell. 37:362-376. Stirnimann, C.U., E. Petsalaki, R.B. Russell, and C.W. Muller. 2010. WD40 proteins propel cellular networks. Trends Biochem. Sci. 35:565-574. van Breugel, M., M. Hirono, A. Andreeva, H.A. Yanagisawa, S. Yamaguchi, Y. Nakazawa, N. Morgner, M. Petrovich, I.O. Ebong, C.V. Robinson, C.M. Johnson, D. Veprintsev, and B. Zuber. 2011. Structures of SAS-6 suggest its organization in centrioles. Science. 331:1196-1199. Venoux, M., X. Tait, R.S. Hames, K.R. Straatman, H.R. Woodland, and A.M. Fry. 2013. Poc1A and Poc1B act together in human cells to ensure centriole integrity. J. Cell Sci. 126:163-175. Wang, W.J., D. Acehan, C.H. Kao, W.N. Jane, K. Uryu, and M.F. Tsou. 2015. De novo centriole formation in human cells is error-prone and does not require SAS-6 selfassembly. Elife. 4. Woodland, H.R., and A.M. Fry. 2008. Pix proteins and the evolution of centrioles. PLoS One. 3:e3778. Yu, I., C.P. Garnham, and A. Roll-Mecak. 2015. Writing and Reading the Tubulin Code. J. Biol. Chem. 290:17163-17172. Zheng, X., A. Ramani, K. Soni, M. Gottardo, S. Zheng, L. Ming Gooi, W. Li, S. Feng, A. Mariappan, A. Wason, P. Widlund, A. Pozniakovsky, I. Poser, H. Deng, G. Ou, M. Riparbelli, C. Giuliano, A.A. Hyman, M. Sattler, J. Gopalakrishnan, and H. Li. 2016. Molecular basis for CPAP-tubulin interaction in controlling centriolar and ciliary length. Nat Commun. 7:11874. 42 989 990 Materials and Methods 991 Sample Preparation 992 Nucleo-flagellar apparatus (NFAp) from Chlamydomonas reinhardtii was purified based 993 on previous published method (Wright et al., 1985). Briefly, cells of a cell-wall defect 994 strain (cc849, cw10 mt-, Chlamydomonas Resource Center) were grown in TAP medium. 995 Cultures were bubbled continuously with filtered air and illuminated with continuous 996 white light. After reaching log phase, the cells were harvested by centrifugation at 250g 997 for 5 minutes. The cells were resuspended in MT Buffer (30mM Tris-acetate, 5mM 998 MgSO4, 5mM EGTA, 25mM KCl, 1mM DTT, pH=7.3). The suspension was overlaid on 999 a 30% Percoll (GE Health) in MT Buffer. The sample was centrifuged for 5 minutes at 1000 300g. The supernatant was discarded. The pellet was resuspended in MT buffer, placed 1001 on ice, equal volume of Lysis Buffer (MT Buffer, 1% Nonidet P-40 (Roche), EDTA-free 1002 protease inhibitor cocktails (Roche) was added, mixed vigorously. The cells were lysed 1003 rapidly and completely in 10 minutes. 1004 The lysate was overlaid on top of discontinuous gradient of 50%, 40% and 30% Percoll 1005 in MT Buffer, spun at 10,000g. Fractions in 30% Percoll was recovered and diluted 10- 1006 fold with MT buffer. The sample was pelleted at 800g followed by resuspension in a 1007 small volume of MT Buffer. 1008 1009 Cryo-ET Data Collection and Reconstruction 1010 The purified NFAp was mixed with BSA-coated 10 nm colloid gold (Ted Pella). 4 µl 1011 sample was applied onto Quantifoil grid Rh/Cu 200 R2/2 (Quantifoil, Inc), blotted for 2 1012 second in 100% humidity and flash-frozen in liquid ethane using a Vitrobot (FEI, Inc). 43 1013 1014 Tomography tilt series were collected on a field emission gun (FEG) microscope (Polara, 1015 FEI, Inc) operated at 300kV. The microscope was equipped with a post-column energy 1016 filter Bio-Quantum GIF (Gatan, Inc). The slit width was set at 25 eV. UCSF Tomography 1017 software (Zheng et al., 2007) was used for collecting single-axis tilt series. For screening 1018 condition purpose, initial tomographic tilt series were collected on a CCD camera 1019 (UltraCam, Gatan Inc). But all tilt series used for subtomogram averaging and for 1020 classification were collected on a K2 direct electron detector in the counted mode (Gatan, 1021 Inc), in a nominal magnification of 50,000, the effective pixel size is 4.82 Å. The 1022 specimen was tilted from -60° to +60° in 1° increment. The exposure time at each tilt was 1023 1.0 second with 0.2 second per frame. This resulted in 5 frames at each tilt. The dose rate 1024 was set at 8 electron/pixel/second. The total accumulative dose for each tilt series was 1025 about 60 electron/Å2. The defocus was set in the range of 3~6 μm. 1026 1027 For tomogram reconstruction, a movie of 5 frames at each tilt was corrected of motion 1028 using MotionCorr (Li et al., 2013). Tomographic tilt series were aligned in IMOD 1029 (Kremer et al., 1996) by using 10 nm colloid gold beads as fiducials. The contrast transfer 1030 function for each tilt series was determined and corrected by TOMOCTF (Fernández et 1031 al., 2006). 3D tomograms were calculated by TOMO3D (Agulleiro and Fernandez, 1032 2015). 1033 1034 Subtomogram Averaging and Classification 44 1035 From 193 reconstructed tomograms containing NFAps, 201 centrioles and 110 1036 procentrioles were identified and selected for further processing. They were first 1037 annotated to find the proximal end of MT triplet. Based on these, the longitudinal spans 1038 of the MT triplet were defined. The subtomograms that contained segment of MT triplet 1039 were boxed out. Initially the triplet segment length was set to 38 nm with pixel size in 1040 2xbinned format. This was followed by ab initio alignment without a reference. The 1041 alignment of subtomograms were carried out by the programs MLTOMO implemented in 1042 the software package Xmipp (Scheres et al., 2009). After generating the initial reference, 1043 the subtomograms were re-extracted from the tomogram volumes in unbinned format. 1044 The MT triplet length in each subtomogram was gradually decreased. The final triplet 1045 length in the averaged structures for triplet, A-, B- and C-tubules was 29 nm, containing 1046 more than 3 repeats as the largest periodicity detected in the procentriole triplet is 8 nm. 1047 Since MT triplet is a continuous filament, the neighboring subtomograms should have 1048 similar geometric parameters. We have developed a program called RANSAC to impose 1049 constraint on the neighboring subtomograms from the same triplet. It was used to detect 1050 any alignment outliers and to impose constraint to correct misaligned subtomograms by 1051 regression. The program is available upon request. The program MLTOMO was also 1052 used in the classification of various structures such as the pinhead, the A-C linker, the A- 1053 tubule, B-tubule and C-tubule to detect heterogeneity. The processes are illustrated 1054 schematically in the supplementary figures. 1055 1056 For each tubule structure from 110 procentrioles, we first carried out focused 1057 classification on 2083 subtomograms on the A-, B- and C-tubule. This allowed us to 45 1058 identify 420, 758 and 862 defective or incomplete structure in each tubule respectively. 1059 These subtomograms were excluded from further focused alignment. The results are 1060 more homogeneous datasets for each tubule that combine procentriole and the proximal 1061 region of the mother centriole. The improvements are evident in the increased resolutions 1062 shown in Figure 1-figure supplement 1 and in Table 1 for all 3 tubules. These averaged 1063 tubule structures were used for further segmentation and estimation of the molecular 1064 weight for each MIP associated to the tubule. 1065 1066 For all the structures reported, resolution assessments were carried out in a “gold- 1067 standard” scheme (Chen et al., 2013). A post-processing program in the software package 1068 Relion (Scheres, 2012) was used to calculate the FSC between two halves of the maps. 1069 The statistics of the structures are summarized in Table 1 and the FSC plots are shown in 1070 the supplementary figures. The local resolution of the pinhead structure, as shown in 1071 Figure 2-figure supplement 13, is estimated by blocres in Bsoft (Heymann and Belnap, 1072 2007). 1073 1074 UCSF Chimera was used for displaying the surface-rendered averaged structures, the 1075 model building and the volume segmentation (Goddard et al., 2005). 1076 1077 To segment out the non-tubulin components, the MIPS, bound to the triplet and to 1078 estimate their molecular weights, an atomic model of MT protofilament was docked into 1079 the subtomogram averages of the A-, B- and C-tubules, respectively. These resulted in 1080 the atomic models for all 3 tubules (Figure 2). The differences between the subtomogram 46 1081 average map and the model density were attributed to the MIPs. The MIP density was 1082 segmented in UCSF Chimera. Based on the volume of segmented MIPs and assuming the 1083 protein density is 0.849 Dalton/Å3 (Fischer et al., 2004), the approximate molecular 1084 weight of each MIP was estimated (Table 2). 1085 1086 To model Poc1 in the A-C linker structure, a crystal structure of WD40 β-propeller 1087 domain with 7 repeats (PDB ID 1S4U) was first manually docked into the doughnut- 1088 shaped density in Leg B of the A-C linker. The docking was further optimized by using 1089 the “Fit in Map” function in UCSF Chimera. A local cross correlation coefficient 1090 between the model and the density map is 0.80, indicating a good fit. 1091 1092 Subtomogram Classification and Alignment on the A-C Linker 1093 For classification on the A-C linker structure, based on the initial alignment result of the 1094 triplet and by using the A- or C-tubule as reference point, respectively, we searched for 1095 subtomograms that exhibit intact A-C linker. In two parallel schemes, we identified 3992 1096 subtomograms from 4 classes and 3245 subtomogram from 5 classes (Class I with 233 1097 subtomograms was excluded due to its small number and low SNR) that show relatively 1098 intact A-C linker attached to the A- or the C-tubule, respectively (Figure 3-figure 1099 supplements 1,3). These two datasets are subjected to further refinement. The results are 1100 two structures, both at 23.1 Å resolution, showing more complete and detailed structure 1101 of the A-C linker (Figure 3-figure supplement 2,4,5). 1102 47 1103 To generate a complete structure of the A-C linker, since we have identified multiple 1104 classes of subtomogram showing intact A-C linker (Figure 3-figure supplements 1,3), 1105 though each in relative low resolution, we docked the above two averages (Figure 3- 1106 figure supplements 2,4) into these low-resolution class-averages. The result is a structure 1107 having complete A-C linker for each identified class. An example process is illustrated in 1108 Figure 3-figure supplement 6. 1109 1110 Analysis on the Periodicity of the Pinhead Structure 1111 For the pinhead structure, we carried out extensive studies on its longitudinal periodicity. 1112 There are several factors that can potentially affect the periodicity in the average. These 1113 are, 1) the length of the triplet in subtomogram, 2) the percentage of overlap between 1114 neighboring triplet segments, 3) the structural component included in the soft-edge mask 1115 that is used in subtomogram alignment, 4) a reference is provided or not during the initial 1116 alignment, 5) the structure difference between the triplet in the procentriole and the triplet 1117 from the proximal region of the mother centriole. All these factors can potentially 1118 introduce errors in the subtomogram alignment. We took these factors into account and 1119 did extensive tests. All results consistently show 8 nm periodicity in the pinhead 1120 structure. In the final map of the pinhead, the longitudinal length is 19 nm, including 1121 more than two repeats of the pinhead. 1122 48 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 References for Materials and Methods Agulleiro, J.I., and J.J. Fernandez. 2015. Tomo3D 2.0--exploitation of advanced vector extensions (AVX) for 3D reconstruction. J. Struct. Biol. 189:147-152. Chen, S., G. McMullan, A.R. Faruqi, G.N. Murshudov, J.M. Short, S.H. Scheres, and R. Henderson. 2013. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy. 135:24-35. Fernández, J.J., S. Li, and R.A. Crowther. 2006. CTF determination and correction in electron cryotomography. Ultramicroscopy. 106:587-596. Fischer, H., I. Polikarpov, and A.F. Craievich. 2004. Average protein density is a molecular-weight-dependent function. Protein science : a publication of the Protein Society. 13:2825-2828. Goddard, T.D., C.C. Huang, and T.E. Ferrin. 2005. Software extensions to UCSF chimera for interactive visualization of large molecular assemblies. Structure (London, England : 1993). 13:473-482. Heymann, J.B., and D.M. Belnap. 2007. Bsoft: image processing and molecular modeling for electron microscopy. J. Struct. Biol. 157:3-18. Kremer, J.R., D.N. Mastronarde, and J.R. McIntosh. 1996. Computer visualization of three-dimensional image data using IMOD. J. Struct. Biol. 116:71-76. Li, X., P. Mooney, S. Zheng, C.R. Booth, M.B. Braunfeld, S. Gubbens, D.A. Agard, and Y. Cheng. 2013. Electron counting and beam-induced motion correction enable near-atomic-resolution single-particle cryo-EM. Nat. Methods. Scheres, S.H. 2012. RELION: implementation of a Bayesian approach to cryo-EM structure determination. J. Struct. Biol. 180:519-530. Scheres, S.H.W., R. Melero, M. Valle, and J.-M. Carazo. 2009. Averaging of electron subtomograms and random conical tilt reconstructions through likelihood optimization. Structure (London, England : 1993). 17:1563-1572. Wright, R.L., J. Salisbury, and J.W. Jarvik. 1985. A nucleus-basal body connector in Chlamydomonas reinhardtii that may function in basal body localization or segregation. J. Cell Biol. 101:1903-1912. Zheng, S.Q., B. Keszthelyi, E. Branlund, J.M. Lyle, M.B. Braunfeld, J.W. Sedat, and D.A. Agard. 2007. UCSF tomography: an integrated software suite for real-time electron microscopic tomographic data collection, alignment, and reconstruction. J. Struct. Biol. 157:138-147. 49