Centriole elimination during C. elegans oogenesis initiates with loss of the central tube protein SAS-1

Centrioles are lost during oogenesis in most metazoans, ensuring that the zygote is endowed with the correct number of two centrioles, which are paternally contributed. How centriole architecture is dismantled during oogenesis is not understood. Here, we analyze with unprecedent detail the ultrastructural and molecular changes during oogenesis centriole elimination in C. elegans. Centriole elimination begins with loss of the so-called central tube and organelle widening, followed by microtubule disassembly. The resulting cluster of centriolar proteins then disappears gradually, usually moving in a microtubule- and dynein-dependent manner to the plasma membrane. Moreover, we find that neither Polo-like kinases nor the PCM, which modulate oogenesis centriole elimination in Drosophila, do so in C. elegans. Furthermore, we demonstrate that the central tube protein SAS-1 normally departs first from the organelle, which loses integrity earlier in sas-1 mutants. Overall, our work provides novel mechanistic insights regarding the fundamental process of oogenesis centriole elimination.


SUMMARY 21
Centrioles are lost during oogenesis in most metazoans, ensuring that the zygote is endowed 22 with the correct number of two centrioles, which are paternally contributed. How centriole 23 architecture is dismantled during oogenesis is not understood. Here, we analyze with 24 unprecedent detail the ultrastructural and molecular changes during oogenesis centriole 25 elimination in C. elegans. Centriole elimination begins with loss of the so-called central tube 26 and organelle widening, followed by microtubule disassembly. The resulting cluster of 27 centriolar proteins then disappears gradually, usually moving in a microtubule-and dynein-28 dependent manner to the plasma membrane. Moreover, we find that neither Polo-like 29 kinases nor the PCM, which modulate oogenesis centriole elimination in Drosophila, do so in 30 C. elegans. Furthermore, we demonstrate that the central tube protein SAS-1 normally 31 departs first from the organelle, which loses integrity earlier in sas-1 mutants. Overall, our 32 work provides novel mechanistic insights regarding the fundamental process of oogenesis 33 centriole elimination. control are tuned to serve specific physiological contexts is poorly understood. 48 Centriole number control must be modified at fertilization to ensure that the two 49 gametes together contribute a total of two centrioles to the zygote. This requirement is 50 achieved in most metazoan organisms by inactivating or eliminating maternally-derived 51 centrioles and providing two centrioles from the sperm (reviewed in (Delattre and Gönczy,52 2004; Manandhar et al., 2005)). There are two principal removal modes of maternally-derived 53 centrioles. In the first mode, maternal centrioles are removed following the meiotic divisions 54 (Crowder et al., 2015). This is the case in starfish, for example, where centrioles are present 55 at spindle poles during the two female meiotic divisions, such that three maternal centrioles 56 are eliminated through polar body extrusion, with the last one being removed in the zygote 57 Overall, our findings uncover that oogenesis centriole elimination in C. elegans 217 comprises a late phase during which the focus containing diminishing levels of centriolar and 218 PCM proteins can detach from the nucleus and move towards the plasma membrane in a 219 microtubule-and dynein-dependent manner. Importantly, however, in approximately 20% of 220 oocytes, the focus of RFP::SAS-7 becomes undetectable in the vicinity of the nucleus, without 221 apparent movement towards the plasma membrane. Moreover, when microtubules or 222 dynein components were depleted, RFP::SAS-7 centriolar foci eventually disappeared (see 223 operate similarly in the worm. There are three Polo-like kinases in C. elegans: PLK-1, which is 231 essential and closest to Polo, exerting analogous functions during centrosome maturation and 232 embryonic cell division (Chase et al., 2000); PLK-2, which is not essential and functions in 233 meiotic chromosome organization (Brandt and Kim, 2021); the more divergent and non-234 essential PLK-3, which has no ascribed function. Previous immunostaining established that 235 PLK-1 and PLK-2 in the gonad localize at centrosomes exclusively in the mitotic zone (Harper 236 et al., 2011). Accordingly, we found that endogenously tagged PLK-1::GFP marks centrosomes 237 exclusively in dividing cells in that region (Fig. 3A). 238 In principle, PLK-1 could be present below detection levels in later stages when 239 elimination occurs, or exerts a function in the mitotic zone that translates into subsequent 240 organelle elimination. To explore these formal possibilities, we set out to investigate the role 241 of PLK-1, as well as PLK-2 and PLK-3 in case redundancies were at play. We used plk-1(RNAi) 242 feeding conditions that result in meiotic arrest (Budirahardja and Gönczy, 2008), as well as 243 the null alleles plk-2(ok1936) and plk-3(gk1103). All analyzed animals expressed RFP::SAS-7 244 and RME-2::GFP to score centriolar focus disappearance relative to oocyte maturation, using 245 the presence of RME-2::GFP on the plasma membrane to this end (Materials and methods). 246 Importantly, we found that the timing of RFP::SAS-7 centriolar focus elimination is 247 comparable in plk-1(RNAi) animals and control worms, with ~70% of -3/-4 oocytes harboring 248 both RFP::SAS-7 foci and RME-2::GFP ( Fig. 3B-3E). In plk-2(ok1936) mutant gonads, oocyte 249 maturation occurs earlier than in the control, as evidenced by the presence of RME-2::GFP 250 positive oocyte as early as positions -12/-11 (Fig. 3F, 3G). RFP::SAS-7 foci are also sometimes 251 lost earlier, but the concordance between RME-2::GFP rise and RFP::SAS-7 decay is preserved, 252 demonstrating that the timing of elimination is not altered (Fig. 3G). When PLK-1 is depleted 253 by RNAi in plk-2(ok1936) animals, oocyte maturation is even more precocious, but RFP::SAS-254 7 disappearance still occurs after RME-2::GFP increase (Fig. 3H, 3I). Finally, plk-1(RNAi) plk-255 2(ok1936) plk-3(gk1103) triply inactivated animals do not exhibit additional phenotypes 256 regarding oocyte maturation or centriole elimination timing (Fig. 3J, 3K). 257 Because RNAi-mediated depletion can sometimes be incomplete, we combined 258 plk-2(ok1936) with the thermosensitive allele plk-1(or683ts) at the restrictive temperature to 259 achieve the strongest possible depletion condition compatible with life. This yielded highly 260 disorganized gonads, yet with seemingly normal centriole elimination timing, as SPD-2 and 261 SPD-5 foci persist until the RME-2::GFP signal appears, as in control worms (Fig. S3A, S3B). In 262 order to further test whether centriole elimination in plk-2(ok1936) plk-1(or683ts) animals 263 correlates with oocyte maturation, we immunostained gonads with IFA-1 pan-centriolar 264 antibodies and counterstained them with a DNA dye to assess chromosome condensation as 265 a proxy for meiotic progression (see Fig. 1B) (Phillips et al., 2009). We found in both control 266 and mutant worms that centriolar foci are present in the vicinity of nuclei with chromosomes 267 characteristic of the late diplotene stage (Fig. S3C, S3D). Overall, we conclude that Polo-like 268 kinases do not modulate centriole elimination timing in C. elegans. 269 We next tested whether the PCM may be required for centriole stability 270 independently of PLK-1, PLK-2 and PLK-3. PCMD-1 is essential for generating the PCM core 271 Moreover, we found that GFP::SAS-7 decay in such animals coincides with RME-2::GFP 277 enrichment, further demonstrating that elimination timing is not impacted in the absence of 278 SPD-5 (Fig. 4D). We conclude that PCM loss does not lead to precocious loss of foci with 279 centriolar and PCM core proteins during C. elegans oogenesis. 280 281

Central tube loss and centriole widening mark the onset of centriole elimination 282
We performed CLEM to characterize the centriole elimination process at the ultrastructural 283 level to gain insights into the mechanisms leading to organelle removal. We dissected, fixed 284 and imaged two gonads expressing GFP::SAS-7, followed by resin embedding and 50 nm serial 285 section EM analysis. A total of 69 foci bearing GFP::SAS-7 were analyzed in this manner, 286 representing successive stages of oogenesis and, therefore, centriole elimination. Each 287 nucleus with an accompanying GFP::SAS-7 focus was first visualized by brightfield and then 288 identified in the corresponding serial sections, thus providing spatial coordinates to spot the intensity of the GFP::SAS-7 focus begins to decline (see Fig. 1), ultrastructural analysis 296 revealed that the central tube is no longer detectable (Fig. 5F). In addition, this is 297 accompanied by an increase of organelle diameter (Fig. 5I, 5J). Loss of centriole integrity was 298 observed thereafter, with microtubule singlets no longer being recognized (Fig. 5G), which 299 was followed by the complete loss of a detectable centriole (Fig. 5H). Interestingly, this is the 300 case despite GFP::SAS-7 remaining present in a focus at this stage, which will be cleared 301 eventually as reported above. 302 Taken together, these results reveal that centriole elimination during C. elegans 303 oogenesis is characterized by an initial loss of organelle integrity, followed by loss of an 304 amorphous cluster of centrosomal proteins. Moreover, these findings raise the possibility that 305 the first observable ultrastructural alteration, namely central tube removal, may cause 306 subsequent loss of centriole integrity. germline. To investigate this possibility, we used U-Ex-STED to analyze the distribution of 316 endogenously tagged SAS-1::3xFLAG during the course of the changes uncovered by EM, in 317 relationship to SAS-7 and SAS-4 distribution. Since the central tube is the first ultrastructural 318 element to vanish, we expected fluorescence intensity of the SAS-1 focus to diminish before 319 that of SAS-7 and SAS-4, and found this to be the case indeed ( Fig. 6A-6C). We used U-Ex-STED 320 to refine the timing of SAS-1 loss relative to that of microtubules, finding that GBP::RFP::SAS-1 fluorescence at centrioles decays faster than that of microtubules, consistent with the early 322 loss of the central tube observed by EM (Fig. 6D, 6G). Using an analogous experimental 323 strategy, we set out to investigate the integrity of other ultrastructural elements in 324 relationship to microtubules. Importantly, whereas SPD-5 rings are dismantled concomitantly 325 with microtubule singlets (Fig. S4A, S4B), consistent with the early loss of SPD-5 uncovered in 326 live specimens (see Fig. 1I, 1J), we found that centriolar microtubules are lost before SPD-2 or 327 SAS-4 ( Fig. 6E-6G). This finding likely explains why foci containing centriolar proteins with 328 decaying signal intensity remain present in maturing oocytes despite the complete loss of 329 ultrastructural integrity observed by EM (Fig. 5H). Moreover, like in the EM analysis, U-Ex-330 STED uncovered a widening of centrioles as they progress through meiosis, as evidenced by 331 measurements of SPD-2, microtubules and SAS-4 ring diameters (Fig. 5H). 332 Together, these data demonstrate that removal of the microtubule-binding protein 333 SAS-1 is an early event in the sequence of events leading to organelle demise. 334

SAS-1 is required for centriole structural integrity and stability during oogenesis 336
If SAS-1 departure from centrioles not only marks the onset of oogenesis centriole 337 elimination, but is also important for this process to occur, then impairing sas-1 function 338 should exacerbate loss of centriole integrity. To test this prediction, we performed U-Ex-STED 339 on gonads from the strong reduction-of-function allele sas-1(t1521ts) at the restrictive 340 temperature. Strikingly, we found that centrioles are as wide in early prophase as they are in 341 diplotene in this mutant background, whereas widening only occurs during diplotene in the 342 control ( Fig. 7A-7C). Moreover, whereas SAS-4 is normally present in approximately equal 343 amounts next to each of the nine centriolar microtubules in early prophase, this is not the 344 case in sas-1(t1521) mutant animals, where intensities are more variable (Fig. 7D). Together, 345 these two data sets suggest that centriole integrity is already affected in early prophase when 346 sas-1 function is compromised. Furthermore, U-Ex-STED analysis revealed that centriolar 347 microtubules and SAS-4 signals decay faster during meiosis progression in sas-1(t1521ts) 348 mutant animals than in the control (Fig. 7E, 7F). This is accompanied by premature loss of 349 organelle integrity, as evidenced by the fraction of centrioles in diplotene with less than nine 350 foci of centriolar SAS-4 and completely disorganized centrioles (Fig. 6G). Overall, these data 351 demonstrate that SAS-1 is required for centriole stability during oogenesis, and suggest that 352 14 354 355 DISCUSSION 356 We established with precision the course of events leading to centriole elimination during 357 C. elegans oogenesis using multi-scale microscopy and molecular genetic approaches. We 358 discovered that oogenesis centriole elimination is characterized by an initial loss of the central 359 tube and organelle integrity, followed by the gradual disappearance of an amorphous cluster 360 of centrosomal proteins. The latter often occurs after detachment from the nucleus and 361 movement to the plasma membrane in a microtubule-and dynein-dependent manner, 362 although such movement is not essential for disappearance . Finally, we demonstrate that the 363 central tube protein SAS-1 is the first component to leave centrioles and propose that this 364 event triggers centriole elimination. 365 366

Clearance of amorphous cluster of centrosomal proteins 367
We found that, following the loss of centrioles recognizable by EM, an amorphous cluster the absence of these components upon pcmd-1 inactivation, centrioles tend to detach from 378 nuclei already in early prophase, localizing to the rachis and the loop region. Moreover, in the 379 control condition, the SPD-5 focus is reduced to ~30% of initial levels already in -7 oocytes, 380 shortly before cluster detachment initiates, compatible with a causative link. Alternatively, a 381 more global remodeling of the nuclear envelope might promote detachment of different 382 molecular complexes in maturing oocytes since P-granules, which do not contain PCM 383 proteins, also detach from the nuclear envelope around that time (Spike et al., 2008). 384 Cluster movement to the plasma membrane occurs at an average velocity of 385 ~0.65µm/min, which is very slow considering that it relies on microtubules and dynein, and 386 partly on kinesin. Dynein and kinesin motor proteins in C. elegans exhibit velocities along 387 microtubules that are two orders of magnitude faster (Gönczy et al., 1999;Pierce et al., 1999). 388 Interestingly, centrioles also translocate in a microtubule-and dynein-dependent manner 389 from the nuclear vicinity to the tip of the dendrite in the PQR sensory neuron in C. elegans, a 390 image. Note absence of SPD-5 in pcmd-1(t3412ts) mutant animals as evidenced by the 552 presence of green foci. Note also that some nuclei lack centrioles already in early pachytene, 553 indicating that centrioles detach precociously, accumulating in the rachis and the loop region. 554 In the insets, dashed lines delineate the rachis and loop borders, dashed circles nuclei, 555 whereas arrowheads point to mislocalized centriolar foci.  for RFP or GFP and co-stained with an antibody against a-tubulin. The LUT "Fire" reports 635 fluorescence intensity (from bright to dim: white, yellow, red, purple, blue and black). The 636 expansion factor is the same for each sample so that relative sizes within each series can be 637 compared (i.e. within D, E and F). Note that we used GBP::RFP::SAS-1 instead of SAS-1::3xFLAG 638 in these experiments as RFP antibodies gave stronger signals than FLAG antibodies, allowing 639 better detection following U-Ex-STED. D, grey boxes were used because of the impossibility 640 to identify centrioles remnant based on the tubulin staining at the oocytes stage.

C. elegans strains and RNAi 676
The C. elegans lines generated for this study (see Table1) are available from the lead contact 677 upon request. Strains were maintained following standard methods on nematode growth 678 medium (NGM) plates seeded with Escherichia coli OP50 as food source (Brenner, 1974). 679 Strains were kept at 20°C or 24°C, except for thermosensitive strains, which were kept at 16°C 680 until the L4 stage, when they were shifted to 24°C or 25°C for 20-24 h prior to imaging. The 681 list of strains used in this study is given in Table S1. Synchronized populations were obtained 682 by allowing 20-50 gravid adults to lay eggs for 1 h at room temperature (RT) and imaging 683 gonads of the resulting adults after ~65h of growth at 20°C. RNAi by feeding was performed 684 with clones from either Ahringer or Vidal library to deplete plk-1(Vidal), tba-2 (Vidal), dhc-1 685 (Arhinger), dlc-1 (Vidal), act-1 (Vidal) and unc-116 (Arhinger), feeding L3/young L4s at 24°C 686 and imaging 20-24h thereafter. in water), and gonads were immediately distributed over the coverslip using a pipette tip.
After expansion, gels were cut into pieces fitting into a 5 cm Petri dish. Prior to staining, gels 729 were blocked for 1h at RT in blocking buffer (10mM HEPES (pH=7.4), 3% BSA, 0.1% Tween 20, 730 sodium azide (0.05%)), followed by incubation overnight at RT with primary antibodies diluted 731 in blocking buffer. Gels were washed three times in blocking buffer for 10 min each, before 732 incubation with secondary antibodies diluted in blocking buffer supplemented with 0.7 ug/L 733 Hoechst for 3 h at 37˚C in the dark. Gels were washed three times in blocking buffer for 10 734 min each before transfer into a 10cm Petri dish for re-expansion by washing 6 times 20 min 735 in distilled water. For imaging, gels were cut and mounted on a 60x24 mm coverslip coated 736 with poly-D-lysine (Sigma, # P1024) diluted in water (2 mg/ml) and supported on both  for each region normalized with that of the most-distal region 1. 790

Timelapse imaging and movie analysis 791
For long-term live imaging, worms were loaded and immobilized in the microfluidic as 792 described (Berger et al., 2018). Briefly, synchronized day 1 adults were collected and washed 793 3 times in fresh S-basal buffer (5.85 g NaCl, 1 g K2HPO4, 6 g KH2PO, 1 ml cholesterol (5 mg/ml 794 in ethanol), H2O to 1L) and left to sediment. In parallel, a 50 ml Falcon tube of NA22 grown 795 overnight at 37°C was centrifuged (at 4000rcf for 20 min) to obtain a bacteria pellet. This We focused on oocytes in positions -6 to -4 at the onset of the experiment, so that the entire 808 movement of foci could be monitored. Cropped images were extracted and aligned based on 809 the center of the nucleus, which was manually selected at each timepoint for a given oocyte.  Levene's test for equality or inequality of variance was performed by using: