The BRCA1-BARD1 complex associates with the synaptonemal complex and pro-crossover factors and influences RAD-51 dynamics during Caenorhabditis elegans meiosis

During meiosis, the maternal and paternal homologous chromosomes must align along their entire length and recombine to achieve faithful segregation in the gametes. Meiotic recombination is accomplished through the formation of DNA double-strand breaks, a subset of which can mature into crossovers to link the parental homologous chromosomes and promote their segregation. Breast and ovarian cancer susceptibility protein BRCA1 and its heterodimeric partner BARD1 play a pivotal role in DNA repair in mitotic cells; however, their functions in gametogenesis are less well understood. Here we show that localization of BRC-1 and BRD-1 (Caenorhabditis elegans orthologues of BRCA1 and BARD1) is dynamic during meiotic prophase I; they ultimately becoming concentrated at regions surrounding the presumptive crossover sites, co-localizing with the pro-crossover factors COSA-1, MSH-5 and ZHP-3. The synaptonemal complex is essential for BRC-1 loading onto chromosomes but recombination is not. BRC-1 forms an in vivo complex with the synaptonemal complex component SYP-3 and the crossover-promoting factor MSH-5. Furthermore, BRC-1 is essential for efficient stage-specific recruitment of the RAD-51 recombinase to DNA damage sites when synapsis is impaired and upon induction of exogenous DNA double-strand breaks. Taken together, our data provide new insights into the localization and meiotic function of the BRC-1–BRD-1 complex and highlight their essential role in DNA double-strand break repair during gametogenesis. Author summary Sexually reproducing species rely on meiosis to transmit their genetic information across generations. Parental chromosomes (homologues) undergo many distinctive processes in their complex journey from attachment to segregation. The physiological induction of DNA double strand breaks is crucial for promoting correct chromosome segregation: they are needed to activate the DNA repair machinery responsible for creating physical connections, or crossovers (COs), between the homologues. In turn, crossovers promote the accurate segregation of the chromosomes in daughter cells. The BRCA1–BARD1 complex has a pivotal role during DNA repair in somatic cells and is exclusively located on unaligned chromosomal regions during mammalian meiosis. We show that in Caenorhabditis elegans, BRCA1 and BARD1 localize to chromosomes at all stages of meiotic prophase I and are enriched at presumptive crossover sites. We found that BRCA1 promotes DNA loading of the repair factor RAD-51 in specific mutant backgrounds and upon exogenous damage induction. Our data provide evidence for a direct physical association between BRCA1 and pro-crossover factors (including the synaptonemal complex) and identify an important role for BRCA1 in stimulating meiotic DNA repair. Further studies are necessary to identify the substrates acted upon by BRCA1–BARD1 complex to maintain genome stability in the gametes.


24
During meiosis, the maternal and paternal homologous chromosomes must align along their 25 entire length and recombine to achieve faithful segregation in the gametes. Meiotic 26 recombination is accomplished through the formation of DNA double-strand breaks, a subset 27 of which can mature into crossovers to link the parental homologous chromosomes and 28 promote their segregation. Breast and ovarian cancer susceptibility protein BRCA1 and its 29 heterodimeric partner BARD1 play a pivotal role in DNA repair in mitotic cells; however, 30 their functions in gametogenesis are less well understood. Here we show that localization of 31 BRC-1 and BRD-1 (Caenorhabditis elegans orthologues of BRCA1 and BARD1) is dynamic 32 during meiotic prophase I; they ultimately becoming concentrated at regions surrounding the 33 presumptive crossover sites, co-localizing with the pro-crossover factors COSA-1,  and ZHP-3. The synaptonemal complex is essential for BRC-1 loading onto chromosomes 35 but recombination is not. BRC-1 forms an in vivo complex with the synaptonemal complex 36 component SYP-3 and the crossover-promoting factor MSH-5. Furthermore, BRC-1 is 37 essential for efficient stage-specific recruitment of the RAD-51 recombinase to DNA damage 38 sites when synapsis is impaired and upon induction of exogenous DNA double-strand breaks. 39 Taken together, our data provide new insights into the localization and meiotic function of 40 the BRC-1-BRD-1 complex and highlight their essential role in DNA double-strand break 41 repair during gametogenesis. 42 43 Introduction synthesis-dependent strand annealing (13). In C. elegans, only one CO is formed between 88 each homologous pair (14), and this depends on the function of the MSH-4/MSH-5 89 heterodimer (orthologues of the yeast and mammalian MutScomplex components, 90 MSH4/MSH5) (15-18), the cyclin COSA-1 (orthologue of mammalian CNTD1) (19,20) and 91 the E3 SUMO-ligase ZHP-3 (orthologue of yeast Zip3) (21-23). CO formation is abolished in 92 absence of DSBs (e.g. in spo-11 mutants) or synapsis; however, unlike in other model 93 systems, lack of DNA breaks does not prevent SC formation in C. elegans (7,11). Meiotic 94 DSB repair also relies on RAD-51-mediated repair in C. elegans (24,25): the RAD-51 95 recombinase localizes to discrete chromatin-associated foci starting in the transition zone and 96 peaking in early pachytene; RAD-51 disengages from DNA in mid-pachytene (7). Markers of 97 aberrant RAD-51 loading, such as increased foci number and/or extended accumulation, are 98 bona fide indicators of defective DSB processing and recombination. CO induction triggers 99 reorganization of the SC components into distinct domains on bivalents (pairs of homologous 100 chromosomes held together by a chiasma): the central elements are confined to the short arm 101 (containing the CO) and the axial elements to the long arm (26)(27)(28)(29)(30). This reorganization is 102 particularly evident during diplotene, at which stage bivalents progressively condense and 103 appear as six DAPI-stained bodies in diakinesis nuclei, which are a read-out for the 104 successful execution of prophase I events (aberrant structures include achiasmatic 105 chromosomes (univalents) or fused/fragmented chromatin masses (11, 16, 31)). 106

107
The breast and ovarian cancer susceptibility protein BRCA1 and its obligate heterodimeric 108 partner BARD1 form an E3 ubiquitin ligase module (the BCD complex), the functions of 109 which have been extensively studied in mitotic cells (32). The BRCA1-BARD1 heterodimer 110 promotes homologous recombination (HR) during the S-G2 stages, by both favouring 111 extended DNA break resection and preventing the non-homologous end joining (NHEJ)-112 . MSH-5 and ZHP-3 are detected at early meiotic stages, with the former 188 accumulating in many foci (these are probably recombination intermediates with both CO 189 and non-CO (NCO) outcomes) and the latter localizing along the SC (20-22). COSA-1 is 190 prominently detected at mid-late pachytene transition as six foci (one CO for each 191 homologue pair), which also contain . Since we observed BRC-1 and 192 BRD-1 recruitment to the short arm of bivalents (chromosome subdomains caused by the 193 formation of CO intermediates (26, 27, 29)), we wondered whether local enrichment of the 194 BCD complex coincides with the regions labelled with pro-CO factors. Comparison of the 195 localization dynamics of GFP::COSA-1 and BRC-1::HA showed that BRC-1 starts to 196 become concentrated concomitantly with enhanced COSA-1 loading and defines a discrete 197 area which later also contains SYP-1 (Fig 3A). We obtained the same localization pattern by 198 monitoring BRD-1 loading ( Fig S2). Furthermore, staining with anti-ZHP-3 antibody (21) 199 also revealed full co-localization of ZHP-3 with BRC-1 ( Fig 3A). To evaluate BRC-1 co-200 localization with MSH-5, we first added a 5′ GFP tag to the endogenous msh-5 locus with 201 CRISPR/Cas9. The tagged line was fully functional, with no defects in chiasmata formation 202 (not shown), suggesting that GFP::MSH-5 is competent to promote CO formation. Similar to 203 ZHP-3 and COSA-1, BRC-1::HA was enriched at defined regions containing a single 204 GFP::MSH-5 focus, which also labels the CO site ( Fig 3B). We performed structured 205 illumination microscopy to further analyse BRC-1 association with the CO site. For this, we 206 added a 5′ OLLAS tag to the endogenous cosa-1 locus (65, 66). This fully functional line was 207 crossed into brc-1::HA worms and co-stained for OLLAS (COSA-1), BRC-1 and  This further confirmed BRC-1 enrichment around COSA-1-labelled CO sites; however, in 209 these nuclei BRC-1 decorates the region of the SC embracing the putative recombination site; 210 thus, it appears to surround, rather than overlapping with, COSA-1 (Fig 3C). 211 212 To assess whether BRC-1-BRD-1 redistribution depends on CO establishment, we generated 213 a brc-1::HA; spo-11 mutant strain to monitor BRC-1::HA loading in absence of meiotic 214 DSBs, which are essential for inducing CO formation. A previous report showed that in spo-215 11 mutants COSA-1 occasionally forms very few foci (possibly arising from mitotic or 216 spontaneous DSBs) and ZHP-3 remains localized along the SC without forming retraction 217 "comets" due to a lack of chiasmata (20). In spo-11 mutants, BRC-1 remained co-localized 218 with ZHP-3 along the SC, without redistributing to chromosome subdomains. This confirms 219 that BRC-1 redistribution depends on chiasmata formation ( Fig 3D). 220 221 Exogenous DSB induction is sufficient to temporarily restore COSA-1 loading and therefore 222 chiasmata formation in spo-11 mutants (11,20,64). Thus, we investigated whether -223 irradiation could rescue the failure in BRC-1 redistribution. We exposed brc-1::HA; spo-11 224 mutant worms to 20 Gy and analysed BRC-1 and ZHP-3 loading at 8 hours post irradiation: 225 at this time point, all late pachytene nuclei in spo-11 mutants display six COSA-1 foci, 226 suggesting that CO induction is fully rescued (20). In the irradiated samples, ZHP-3 was 227 retracted towards the CO site and, consistent with this, BRC-1 also became concentrated 228 around the CO site ( Fig 3E). Based on these data, we conclude that BRC-1 and BRD-1 229 localize to the short arms of bivalents and that their reorganization in mid-pachytene nuclei is 230 dependent on CO establishment. 231

Synapsis and recombination have different effects on BRC-1 and BRD-1 loading 247
Given that CO establishment triggers BRC-1-BRD-1 redistribution (Fig 3C, D), we sought to 248 analyse their localization in mutants that have impairment at different steps of CO formation. 249 As already mentioned, an absence of DSBs leads to a lack of recombination, which prevents 250 BRC-1 and BRD-1 retraction to the short arms of bivalents. We therefore asked whether 251 impaired DNA repair by HR, but not by DSB induction, influences BRC-1 and BRD-1 252 localization. To address this, we crossed brc-1::HA into the msh-5 mutant, which cannot 253 convert recombination intermediates into mature CO products (7, 16). In msh-5 mutants, 254 BRC-1 accumulated along the SC but retraction was not observed (Fig 4A), similar to the 255 localization pattern observed in spo-11 (Fig 3). Then, we analysed BRC-1::HA staining in 256 rad-51 mutants, which have normal SC assembly but no homologous DNA repair due to lack 257 of RAD-51-dependent strand displacement and invasion of the homologous chromosome (24, 258 25). Interestingly, BRC-1 had a rather punctate staining pattern, perhaps through labelling 259 recombination-independent DNA joined molecules ( Fig 4A). Despite this, a strong 260 association with SYP-1 in chromosome subdomains was observed in nuclei exiting the 261 pachytene stage (we also observed this in msh-5 mutants). We observed a similar pattern of 262 BRD-1 localization in com-1 mutants ( Fig S3): here, interfering with DSB resection impairs 263 RAD-51 loading and therefore abolishes CO formation (68). These results suggest that a lack 264 of COs per se impairs redistribution of the BCD complex in late pachytene cells without 265 perturbing loading along the SC. However, in mutants such as rad-51 that are defective in the 266 early steps of recombination, BRC-1-BRD-1 association with the SC is also dramatically 267 reduced. Next, we sought to analyse whether BRC-1 and BRD-1 loading is regulated by 268 synapsis. We analysed BRC-1::HA loading in the complete and partial absence of SC, as well 269 as in mutants in which synapsis occurs between non-homologous chromosomes. The central 270 portion of the SC is formed by several proteins (SYP-1-4) which are loaded in an 271 interdependent manner; thus, all are necessary to establish synapsis (7,8,58,59). In the syp-2 272 synapsis-null mutant (7), BRC-1::HA had a rather punctate staining pattern throughout 273 meiotic prophase I. Strikingly, unlike in the wild type, where BRC-1 starts to spread along 274 the SC immediately after the disappearance of RAD-51, in syp-2 mutants BRC-1 foci 275 remained in close proximity to and co-localized with RAD-51 foci in mid and late pachytene 276 nuclei ( Fig 4B). In C. elegans, a family of zinc-finger nuclear proteins connects 277 chromosome-specific ends (i.e. pairing centres) to the nuclear envelope to promote 278 chromosome pairing and synapsis (69, 70). ZIM-2 and HIM-8 bind to the ends of 279 chromosomes V and X, respectively. Therefore, chromosome V is asynapsed in zim-2 280 mutants and chromosome X is asynapsed in him-8 mutants. We asked whether a partial 281 deficiency in synapsis establishment (affecting only one chromosome pair) also changes 282 BRC-1 loading dynamics. Analysis of BRC-1::HA expression in him-8 and zim-2 mutants 283 revealed a lack of BRC-1 on unsynapsed chromosomes pairs, despite normal loading along 284 the SC and retraction towards the CO site in the remaining bivalents ( Fig 4C,D), suggesting 285 that local synapsis defects do not impair BRC-1 loading. Lastly, we analysed BRD-1 loading 286 in two mutants with deregulated SC assembly. HTP-1 is a HORMA-domain-containing 287 protein essential to prevent SC assembly between non-homologous chromosomes and 288 PROM-1 is an F-box protein involved in promoting meiotic entry and homologous pairing. 289 Both htp-1 and prom-1 mutants display extensive SYP-1 loading between non-homologous 290 chromosomes as well as asynapsed chromosome regions; consequentially, chiasmata 291 formation is severely impaired (26, 71). Remarkably, the degree of BRD-1 co-localization 292 with SYP-1 was extremely reduced in both htp-1 and prom-1 mutants, with most BRD-1 293 detected as bright agglomerates within the nucleus (Fig S4). Thus, we conclude that BRC-1 294 and BRD-1 redistribution during meiotic progression requires CO establishment and is tightly 295 regulated by SCs. 296

BRC-1 promotes RAD-51 recruitment in the absence of synapsis 298
BRC-1 is dispensable for establishing synapsis and chiasmata; however, brc-1 mutant 299 germlines have a higher number of and more persistent RAD-51-labelled recombination 300 intermediates compared with the wild type ( Fig S5) (46, 47). Impaired BRC-1 localization, 301 and probably also impaired function, in CO-defective mutants leads to the formation of 302 abnormal chromosome structures in diakinesis nuclei, possibly due to deficient IS repair (47). 303 DSB repair during meiosis is channelled into both CO and NCO pathways. Since it has been 304 suggested that BRC-1 might preferentially function in NCOs (47), we investigated whether 305 other factors involved in resolving the recombination intermediates required for both CO and 306 NCO repair might also be affected. In somatic cells, the RTR complex mediates efficient 307 resolution of recombination intermediates by promoting the dissolution of double Holliday 308 junctions to yield non-CO products (72-74). RMI1 is an essential component of the RTR 309 complex and a scaffolding component for other complex members, BLM and TOP3A, which 310 promotes their dissolution activity (74). The C. elegans RMI1orthologue, RMH-1, localizes 311 to recombination foci during meiosis: it appears in early pachytene and peaks in mid-312 pachytene, accumulating in many foci and possibly labelling all recombination intermediates. 313 At late pachytene transition, the number of RMH-1 foci is reduced to roughly six per nucleus; 314 these foci co-localize with foci of the pro-CO factors COSA-1, MSH-5 and ZHP-3. Lack of 315 RMH-1 causes a drastic reduction in chiasmata formation due to impaired COSA-1 and 316 MSH-5 loading. However, in CO-deficient backgrounds such as cosa-1, msh-5 and zhp-3 317 mutants, RMH-1 is still recruited in early pachytene but is not retained until late pachytene. 318 Therefore, it has been postulated that RMH-1 functions in both the CO and NCO pathways 319 (75). MSH-5 displays a similar localization, but does not fully co-localize with RMH-1 (20, 320 75). We scored COSA-1, MSH-5 and RMH-1 nuclear localization in brc-1 mutants in nuclei 321 spanning the transition zone to late pachytene stage. Interestingly, GFP::MSH-5 322 accumulation was reduced in early and mid-pachytene, with a similar, but less prominent, 323 trend for GFP::RMH-1 (Fig 5A,B). By late pachytene, both proteins had been recruited into 324 six foci, together with COSA-1, suggesting that the early processing of recombination 325 intermediates might be defective in absence of BRC-1. 326 327 Given that BRC-1 and BRD-1 loading are regulated by synapsis and the establishment of 328 COs, and that a lack of BRC-1 might affect the processing of NCOs rather than COs, we next 329 assessed the effects of BRC-1 depletion in genetic backgrounds defective in chiasmata 330 formation, which hence rely solely on NCOs to repair meiotic DSBs. We first analysed 331 DAPI-stained bodies in diakinesis nuclei from cosa-1 brc-1 and brc-1; syp-2 double mutants 332 to confirm the presence of aberrant chromatin structures (Fig 6A), as previously reported (46, 333 47). As abnormalities in diakinesis nuclei can result from impaired RAD-51-dependent repair 334 of meiotic DSBs (24, 31, 76), we sought to analyse whether lack of brc-1 altered RAD-51 335 dynamics. To this end, we quantified RAD-51 in cosa-1 brc-1 and brc-1; syp-2 mutants. 336 Failure to convert recombination intermediates into mature CO products has been linked to 337 increased RAD-51 levels and its delayed removal during meiotic prophase due to either 338 excessive DSB induction or slower processing of recombination intermediates (5,6,15,16), 339 which are eventually channelled into alternative repair pathways (e.g. IS repair) (7). In fact, 340 both cosa-1 and syp-2 mutants accumulated high levels of RAD-51, which disengaged from 341 chromatin in mid and late pachytene, respectively (Fig 6B, C) (7, 20). Remarkably, removal 342 of BRC-1 from cosa-1 and syp-2 mutants had different effects on RAD-51 dynamics: in both 343 cosa-1 brc-1 and brc-1; syp-2 double mutants, there were far fewer RAD-51 foci in early 344 pachytene compared with both single mutants; however, in cosa-1 brc-1 mutants RAD-51 345 accumulation was dramatically prolonged until diplotene, whereas in the brc-1; syp-2 mutant 346 overall RAD-51 staining was dramatically reduced (Fig 6B, C). Aberrant chromosome 347 structures occurred at a particularly high frequency in brc-1; syp-2 mutants, consistent with 348 the severe reduction in RAD-51 loading in pachytene nuclei (Fig 6A). Thus, in CO-defective 349 mutants, BRC-1 regulation of RAD-51 dynamics is altered by the presence of the SC. 350 351 Efficient RAD-51-mediated repair upon exogenous DSB induction requires functional 352

BRC-1 353
Exposure of brc-1 and brd-1 mutants to IR causes dose-dependent hypersensitivity which 354 eventually culminates in full sterility, possibly due to the formation of highly unstructured 355 chromatin bodies in diakinesis nuclei (46). These structures resemble those formed upon 356 BRC-2/BRCA2 depletion, which in worms is essential for RAD-51 loading (31, 76), and 357 COM-1/Sae2 depletion, which promotes DSB resection (68, 77). Both mutants lack RAD-51 358 recruitment onto DNA during meiotic prophase I. We therefore sought to investigate whether 359 the aberrant chromatin masses observed in irradiated brc-1 mutants were caused by impaired 360 RAD-51 recruitment. To be efficiently loaded to the single-stranded DNA (ssDNA) tails 361 generated after resection, RAD-51 must be exchanged with RPA (RPA-1 in worms), which 362 coats ssDNA tails to stabilize them and prevent DNA from self-winding (78, 79). We 363 generated a brc-1 mutant strain expressing RPA-1::YFP (80) and analysed RAD-51 and 364 RPA-1 loading at two different time points post irradiation. We observed a dramatic 365 reduction in RAD-51 focus formation specifically in mid to late pachytene nuclei of brc-1 366 mutants, along with enhanced RPA-1 levels (Fig 7A). At 24 hours post irradiation, both 367 RAD-51 and RPA-1 were still abundant in [rpa-1::YFP] animals; in contrast, in brc-1; [rpa-368 1::YFP] mutants RPA-1 was still expressed at higher levels than in controls, but RAD-51 was 369 remarkably reduced ( Fig 7B). Prompted by these results, we decided to analyse the loading 370 dynamics of BRC-1::HA and RAD-51 after IR exposure to assess whether exogenous DSB 371 formation affects the mutual spatio-temporal regulation of these proteins. Under 372 physiological growth conditions, BRC-1 and RAD-51 localization did not overlap prior to 373 BRC-1 enrichment in the SC, which occurs after RAD-51 disappearance (Fig S6A, B). At 374 1 hour post irradiation, BRC-1::HA started to form discrete chromatin-associated foci in pre-375 meiotic nuclei, often in close proximity to (but not co-localizing with) RAD-51 foci 376 (Fig S6A,B). Although abundant RAD-51 accumulation was triggered by IR exposure 377 throughout the germline, BRC-1::HA levels were only modestly increased. However, western 378 blot analysis revealed a shift in BRC-1::HA migration after IR which remained unchanged 379 throughout the time course (Fig S6C), suggesting that exogenous DNA damage might elicit 380 post-translational modifications of BRC-1. Western blot analysis also showed a slight 381 increase in BRC-1::HA abundance, confirming our immunofluorescence data (Fig S6A). In 382 meiotic nuclei, BRC-1 was detected along the SC at an earlier time point than in non-383 irradiated animals, but retraction towards the short arms of bivalents appeared delayed 384 ( Fig S6A). Samples analysed 8 hours after IR revealed robust BRC-1 and RAD-51 co-385 localization in nuclei residing in the mitotic tip; however, as at the earlier time point, no clear 386 co-localization was observed in pachytene nuclei (Fig S6A, B). At 24 hours post irradiation, 387 BRC-1::HA foci in the mitotic nuclei had largely disappeared and bright RAD-51 foci were 388 observed only in enlarged, G2-arrested nuclei that were still undergoing repair; in contrast, 389 bright RAD-51 foci co-localizing with BRC-1 were occasionally seen in non-arrested nuclei. BRC-1 and BRD-1 display a highly dynamic localization pattern during meiotic prophase I 412 progression, shifting from a pattern of rather diffuse accumulation at early stages to a robust 413 association with the SC, which culminates in retention of the BCD complex at the region of 414 the bivalent harbouring the chiasma (Figs 1-3). Remarkably, accumulation of BRC-1-BRD-1 415 at specific chromosomal subdomains occurred prior to retraction of the SC central elements 416 to those domains but was concomitant with recombination factor-dependent enrichment of 417 PLK-2 at the SC (Fig 2B) (63, 64), suggesting that the BCD complex is actively targeted to 418 the region surrounding the CO rather than passively recruited following SC remodelling. 419 The fact that recruitment of BRC-1-BRD-1 to the region surrounding the chiasma has similar 420 kinetics to PLK-2 recruitment and precedes SYP-1 redistribution suggests that the BCD 421 complex (i) is brought into place via physical interaction with the CO machinery (Fig 3) Our data favour a model in which the SC is essential for initial recruitment of the BCD 432 complex onto the chromosomes and later accumulates at the CO site due to the local 433 concentration of recombination factors. In fact, BRC-1 recruitment to the SC is not prevented 434 in msh-5 or spo-11 mutants (both of which are defective in CO formation but proficient in 435 synapsis establishment). However, similar to ZHP-3, BRC-1 fails to retract (Figs 3C and 4A). 436 Irradiation of spo-11 mutants restored BRC-1 and ZHP-3 redistribution to the short arms of 437 bivalents (Fig 3D), confirming that CO establishment per se is the key trigger of local BCD 438 complex enrichment. Abrogation of synapsis dramatically changed the BRC-1 expression 439 pattern: it remained punctate throughout meiotic prophase I and displayed extensive and 440 specifically co-localization with RAD-51 in late pachytene cells (Fig 4B). However, in 441 mutants in which only one chromosome pair was asynapsed, such as him-8 and zim-2 442 mutants, BRC-1 was not loaded onto the unsynapsed regions but loading dynamics were 443 normal for the other chromosome regions (Fig 4C, D). It was recently shown that PLK-2 444 plays a pivotal role in modulating the physical state of the SC in response to recombination 445 and that an absence of synapsis impairs PLK-2 redistribution from the nuclear envelope to 446 chromosome subdomains (63, 64, 82), which might explain the different BRC-1 localization 447 patterns in syp-2 mutants. Different BRD-1 localization patterns were observed in htp-1 and 448 prom-1 mutants, but both were characterized by extensive non-homologous synapsis. BRD-1 449 accumulated in bright agglomerates in the nucleus, suggesting that SYP loading per se is not 450 sufficient to recruit BRC-1-BRD-1 onto the SC (Fig S6). 451 452

Crosstalk between the BCD complex and RAD-51 is governed by the SC 453
Blocking BRC-1 function had opposing effects on the progression of recombination 454 intermediates in cosa-1 and syp-2 (CO-defective) mutants. RAD-51 accumulation was 455 exacerbated in cosa-1 single mutants and largely suppressed in syp-2 mutants (Fig 6), leading 456 to the formation of aberrant chromatin masses in diakinesis nuclei both mutant backgrounds, 457 as previously reported (47). Based on genetic data, BRC-1 function was previously 458 postulated to be essential for IS repair of meiotic DSBs (47, 83, 84); our data corroborate this 459 model. In cosa-1 brc-1 double mutants, the presence of an intact SC might still impose a 460 homologue-biased constraint for an inter-homologue, CO-independent pathway that relies on 461 RAD-51-mediated repair but not on BRC-1 function. However, in the absence of synapsis, 462 repair of recombination intermediates is probably channelled entirely through the IS repair 463 pathway because the sister chromatid is the only available repair template: SC depletion 464 triggers association of BRC-1 with RAD-51 in late pachytene cells at presumptive repair 465 sites, thereby promoting HR-mediated repair. We also observed fewer RAD-51 foci in brc-1; 466 syp-2 double mutants during early pachytene, suggesting that BRC-1 is nonetheless required 467 to (directly or indirectly) promote efficient RAD-51 loading, although co-localization with 468 RAD-51 at meiotic onset might be very transient. We observed that a lack of BRC-1 reduces 469 the loading of recombination markers such as MSH-5 and RMH-1 in early pachytene, 470 suggesting that even in the presence of the SC, BRC-1-BRD-1 function is required to 471 efficiently promote the processing of recombination intermediates. Moreover, in brc-1 472 mutants exposed to exogenous DSB induction, RAD-51 is not efficiently retained in mid-to 473 late pachytene cells (Fig 7). This is not due to impaired resection, as shown by the abundant 474 recruitment of RPA-1, which stabilizes ssDNA. However, RAD-51 loading is comparable to 475 controls in later stages, suggesting that stabilization, rather than loading per se, might require 476 the action of the BCD complex. This is in line with the findings reported by Li et al. (see 477 accompanying manuscript). 478 When we scored BRC-1 levels after exposure to IR, we detected a slight increase in 479 abundance but a marked difference in protein migration on western blots (Fig S6), suggesting 480 that exogenous DNA damage promotes post-translational modification of BRC-1. 481 Importantly, despite dramatically enhanced RAD-51 levels upon irradiation, we observed 482 clear co-localization with BRC-1 only in mitotic cells and not during pachytene, once again 483 confirming that these proteins co-localize only when the SC is indeed absent (Fig S6). For cytological analysis of whole-mount gonads, age-matched worms (24 hours post-L4 501 stage) were dissected in 1× PBS on a Superfrost Plus charged slide and fixed with an equal 502 volume of 2% PFA in 1× PBS for 5 min at room temperature. Slides were freeze-cracked in 503 liquid nitrogen and then incubated in methanol -20C for 5 min, followed by three washes in 504 PBST (1× PBS, 0.1% Tween) at room temperature. Slides were blocked for 1 hour at room 505 temperature in PBST containing 1% BSA and then primary antibodies were added in PBST 506 and incubated overnight at 4C. Slides were then washed in PBST at room temperature and 507 secondary antibodies were applied for 2 hours. After three washes in PBST for 10 min each, 508 60 μl of a 2 μg/ml stock solution of DAPI in water was added to each slide and stained for 509 1 min at room temperature. Samples were washed again for at least 20 min in PBST and then 510 mounted with Vectashield. For detection of GFP::MSH-5, worms were dissected and fixed in 511 1× EGG buffer containing 0.1% Tween (instead of PBST). Detection of [RPA-1::YFP] was 512 performed as previously described (85). Primary antibodies used in this study were: mouse 513 monoclonal anti-HA tag (pre-absorbed on N2 worms to reduce non-specific binding; 1:1000 514 dilution; Covance), rabbit anti-HA tag (1:250 dilution; Invitrogen), rabbit anti-BRD-1 (1:500 515 dilution) (50), chicken anti-SYP-1 (1:400 dilution) (51), guinea pig anti-HTP-3 (1:500 516 dilution) (60), mouse monoclonal anti-GFP (1:500 dilution; Roche), guinea pig anti-ZHP-3 517 (1:500 dilution) (21), rabbit anti-OLLAS tag (pre-absorbed on N2 worms to reduce non-518 specific binding; 1:1500 dilution; GenScript), rabbit anti-RAD-51 (1:10,000 dilution; SDIX) 519 and rabbit anti-PLK-2 (1:500 dilution) (86). Appropriate secondary antibodies were 520 conjugated with Alexa Fluor 488 or 594 (1:500 dilution) or with Alexa Fluor 647 (1:250 521 dilution). Images were collected as z-stacks (0.3 μm intervals) using an UPlanSApo 100x NA 522 1.40 objective on a DeltaVision System equipped with a CoolSNAP HQ2 camera. Files were 523 deconvolved with SoftWORx software and processed in Adobe Photoshop, where some false 524 colouring was applied. Samples acquired by super-resolution microscopy ( Fig 3C) were 525 prepared as previously reported (63)  For whole-cell protein extraction, 200 age-matched animals (24 hours post-L4 stage) were 530 picked into 1× Tris-EDTA buffer (10 mM Tris pH 8, 1 mM EDTA) containing 1× protein 531 inhibitor cocktail (Roche), snap-frozen in liquid nitrogen. After thawing, an equal volume of 532 2× Laemmli buffer was added. Samples were boiled for 10 min, clarified and separated on 533 pre-cast 4-20% gradient acrylamide gels (Bio Rad). 534 535 Fractionated protein extracts for western blotting and immunoprecipitation were prepared as 536 previously reported (51). Western blotting used 50 μg protein samples from each fraction, 537 whereas immunoprecipitation assays used at least 1 mg samples of pooled soluble nuclear 538 and chromatin-bound fractions. Proteins were transferred onto nitrocellulose membrane for 539 1 hour at 4C at 100V in 1× Tris-glycine buffer containing 20% methanol. Membranes were 540 blocked for 1 hour in 1× TBS containing 0.1% Tween (TBST) and 5% milk; primary 541 antibodies were added into the same buffer and incubated overnight at 4C. Membranes were 542 then washed in 1× TBST and then incubated with appropriate secondary antibodies in TBST 543 containing 5% milk for 2 hours at room temperature. After washing, membranes were 544 incubated with ECL and developed with a ChemiDoc system (BioRad). To detect 545 phosphorylated CHK-1 S345 , TBST containing 5% BSA (instead of milk) was used for 546 blocking and antibody dilution. The following antibodies were used for western blotting: 547 mouse monoclonal anti-HA tag (1:1000 dilution; Cell Signalling), rabbit anti-HA tag (1:500 548 dilution; Invitrogen), chicken anti-GFP (1:4000 dilution; Abcam), mouse anti-GAPDH 549 (1:5000 dilution; Ambion), rabbit anti-Histone H3 (1:100,000 dilution; Abcam); rabbit anti-550 phospho-CHK-1 S345 (1:1000 dilution; Cell Signalling), HRP-conjugated anti-mouse (1:2500 551 dilution) and anti-rabbit (1:25,000 dilution; both Jackson ImmunoResearch) and HRP-552 conjugated anti-chicken (1:10,000 dilution; Santa Cruz). 553 554 Irradiation 555 Age-matched worms (24 hours post-L4 stage) were exposed to the indicated dose of IR with 556 a Gammacell irradiator containing a 137 Cs source. For viability screening, irradiated worms 557 were allowed to lay eggs for 24 hours and then removed; hatched versus unhatched eggs were 558 scored the following day. For cytological analysis, worms were dissected and immunostained 559 at the indicated times. 560 561 CRISPR-Cas9 Tagging 562 All the details relative to the tagging strategy followed to generate the brc-1::HA, 563