The SMC-5/6 Complex and the HIM-6 (BLM) Helicase Synergistically Promote Meiotic Recombination Intermediate Processing and Chromosome Maturation during Caenorhabditis elegans Meiosis

Meiotic recombination is essential for the repair of programmed double strand breaks (DSBs) to generate crossovers (COs) during meiosis. The efficient processing of meiotic recombination intermediates not only needs various resolvases but also requires proper meiotic chromosome structure. The Smc5/6 complex belongs to the structural maintenance of chromosome (SMC) family and is closely related to cohesin and condensin. Although the Smc5/6 complex has been implicated in the processing of recombination intermediates during meiosis, it is not known how Smc5/6 controls meiotic DSB repair. Here, using Caenorhabditis elegans we show that the SMC-5/6 complex acts synergistically with HIM-6, an ortholog of the human Bloom syndrome helicase (BLM) during meiotic recombination. The concerted action of the SMC-5/6 complex and HIM-6 is important for processing recombination intermediates, CO regulation and bivalent maturation. Careful examination of meiotic chromosomal morphology reveals an accumulation of inter-chromosomal bridges in smc-5; him-6 double mutants, leading to compromised chromosome segregation during meiotic cell divisions. Interestingly, we found that the lethality of smc-5; him-6 can be rescued by loss of the conserved BRCA1 ortholog BRC-1. Furthermore, the combined deletion of smc-5 and him-6 leads to an irregular distribution of condensin and to chromosome decondensation defects reminiscent of condensin depletion. Lethality conferred by condensin depletion can also be rescued by BRC-1 depletion. Our results suggest that SMC-5/6 and HIM-6 can synergistically regulate recombination intermediate metabolism and suppress ectopic recombination by controlling chromosome architecture during meiosis.


Introduction
Homologous recombination (HR) accurately repairs DNA double-strand breaks (DSBs). During recombination, DSBs are resected to create long single-strand DNA tails, which can invade intact homologous donor sequences with the aid of the conserved Rad51 recombinase. An initial recombination intermediate formed after strand invasion is called a displacement loop (Dloop). The invading strand then primes DNA synthesis using the intact homologous chromosomes as a template and second-end DNA capture leads to formation of cruciform recombination intermediates known as Holliday junctions (HJs). HJs can be dissolved by BLM helicase combined with topoisomerase III and RMI1 [1] or resolved by different structure-specific endonucleases leading to crossover (CO) or non-crossover (non-CO) products.
Homologous recombination is essential for meiosis [2]. Meiotic DSBs are generated by the Spo11 enzyme [3]. The choice of template for repairing programmed DSBs during meiosis is distinct from that during mitosis [4]. In mitotic cells, the sister chromatids are preferentially used as repair templates, thereby preventing potentially deleterious effects of recombination, such as loss of heterozygosity. In contrast, during meiosis there is a bias to use the homolog as a repair template, therefore facilitating the formation of COs between homologous chromosomes [5]. COs are not only important for the exchange of genetic information between maternal and paternal chromosomes, but together with sister chromatid cohesion they also provide a transient physical linkage between homologs (chiasmata) that prevents precocious chromosomes segregation before the end of meiosis I.
Although interhomolog recombination is generally thought to prevail during meiosis, intersister recombination also plays an important role in meiotic DSB repair (for review see [6]). In budding yeast, DSBs are efficiently repaired by intersister recombination during meiosis, even when the homolog is present [7]. In C. elegans, interhomolog recombination is favored to repair SPO-11-induced DSBs during the early and mid-pachytene stage [8]. In contrast, intersister recombination contributes to rapid DNA repair at the late pachytene stage in order to preserve genomic integrity before exit from meiotic prophase [9,10].
At the end of the meiotic prophase, all the recombination intermediates generated by interhomolog and intersister recombination must be resolved to facilitate CO formation and accurate chromosome segregation. Numerous studies have been done to identify HJ resolvases contributing to CO formation in different organisms [11][12][13][14]. In budding yeast, at least five endonucleases, belonging to two distinct pathways, are required for CO formation during meiosis [12,15]. The major pathway including Exo1 and the Mlh1-Mlh3 (MutLγ) complex produces the majority of COs while the minor pathway involves the Mus81-Mms4, Yen1, and Slx1-Slx4 complex. Although these enzymes (except for Mlh3) are conserved in C. elegans, different sets of resolvases appear to be used [16][17][18]. Based on genetic and cytological data, C. elegans might have at least two redundant resolvase activities for meiotic recombination intermediate resolution and CO formation. The first resolvase activity depends on XPF-1 and BLM helicase HIM-6. The other resolvase activity includes SLX-1 and MUS-81. Deletion of both resolvase activities results in severely reduced progeny viability and persistence of chromatin bridges between homologs, which represent unresolved recombination intermediates [16]. However, few wild type COs are still present even in the absence of both resolvase activities, and the mus-81 slx-1; xpf-1; him-6 quadruple mutant is not 100% lethal, suggesting the existence of further factors capable of resolving HJs [16].
The Smc5/6 complex belongs to the structure maintenance of chromosome (SMC) protein family, which also includes components of cohesin and condensin and is well conserved among eukaryotes [19][20][21]. Smc5 and Smc6 form a heterodimeric ring-like structure via the interaction through their hinge domains [22]. In budding yeast, at least six additional subunits are associated to Smc5 and Smc6. These subunits are termed non-SMC elements (Nse1-6) [23], whereas in human and worms only four Nse subunits have been identified so far [24]. Notably, Nse1 contains a RING finger domain that resembles those found in ubiquitin ligases and the C-terminal portion of Nse2 contains a conserved RING-like domain and shows SUMO-ligase activity both in vitro and in vivo [25,26]. While the function of cohesin and condensin in sister chromatid cohesion and chromosome condensation has been well characterized, little is known about the exact cellular function of Smc5/6. Previous studies show that the Smc5/6 complex is required for recombinational repair in mitotic cells [27,28]. Recently, the Smc5/6 complex has been found to locate at the pericentromeric heterochromatin in mouse spermatocytes, indicating that it might be involved in preventing aberrant HR in these repetitive regions during meiosis [29,30]. However, this localization is not conserved in human prophase spermatocytes [31]. The Smc5/6 complex is also important for the elimination of meiotic recombination intermediates in budding yeast [32][33][34]. Notably, budding yeast Smc5/6 are essential for cell viability while genes involved in homologous recombination are not, suggesting that the compromised DNA repair in smc5/6 mutants might also originate from a more fundamental defect in chromosome organization. However, the cross-talk between the Smc5/6 complex and other SMC complexes remains elusive. While a regulatory role of the Smc5/6 complex in Mus81-Eme1 dependent HJ resolution has been revealed in fission yeast [35], the exact role of Smc5/6 in higher organisms and the interplay between Smc5/6 complex and other meiotic recombination intermediate resolvases, such as the BLM helicase, is largely unknown.
The hereditary breast/ovarian cancer predisposition gene BRCA1, which is not present in budding and fission yeast, forms a heterodimer with BARD1 and is evolutionarily conserved. The BRCA1/BARD1 complex has been reported to be involved in a variety of processes in somatic cells, including DNA replication, DNA damage response and chromatin remodeling [36]. BRCA1 has also been suggested to have a role in meiotic DSB repair. Previous studies showed that BRCA1 interacts with RAD51 and that it is required for loading of RAD51 to DSB sites [37]. In contrast, a recent study suggested that BRCA1 has a modest impact on RAD51 assembly and CO formation [38]. While deletion of Brca1 exon 11 disrupts spermatogenesis in mice, Brca1 Δ11/Δ11 female mice are fertile [39]. However, Brca1 mutants have a decreased number of MSH4 foci, which is thought to be associated with the stabilization of single-strand invasion intermediates formed at early stages of recombination. This indicates a role of BRCA1 in the regulation of recombination intermediates [38]. Mutants of the C. elegans BRCA1 ortholog brc-1 are also viable and fertile but display a weak meiotic chromosome segregation defect [10]. However, when interhomolog recombination is abrogated by depletion of the synaptonemal complex component SYP-2, mutation of brc-1 results in significantly increased chromosome fragmentation, suggesting that BRC-1 is important for meiotic DSB repair through intersister recombination [10].
In this study, we show that SMC-5/6 acts synergistically with the BLM helicase HIM-6 in meiotic DSB repair in C. elegans. The combined function of SMC-5/6 and HIM-6 is essential for recombination intermediate processing, CO formation and accurate chromosome segregation during meiosis. Loss of BRC-1 rescued the progeny lethality of smc-5; him-6 double mutants. Furthermore, deletion of SMC-5/6 and HIM-6 leads to an irregular distribution of condensin and defective chromosome morphology. Our results suggest that SMC-5/6 and HIM-6 can synergistically regulate recombination intermediate metabolism and prevent ectopic recombination by controlling chromosome architecture during meiosis.
To assess if the synthetic lethality we observed was due to a defect in recombinational repair, we examined the appearance and disappearance of the strand exchange protein RAD-51 on meiotic chromosomes. RAD-51 binds to the resected single strands and the number of RAD-51 foci provides an estimate of the number of recombination intermediates [44]. Increased RAD-51 foci formation is typically observed in the transition zone, a stage where meiotic DSBs are induced (Fig 1B, Zone 3, S1 Fig). As recombination progresses, all DSBs are repaired and RAD-51 foci finally disappear at the late pachytene stage (Fig 1B, zone 6, S1 Fig). Consistent with previous studies, a significantly increased number of RAD-51 foci were observed in early and mid pachytene zones in the smc-5 and him-6 single mutants when compared to the wild type animals [17,41,43]. While the RAD-51 staining was rarely seen in the late pachytene in wild type and him-6 mutants, it persisted in smc-5 mutants as reported previously [43]. Notably, in the smc-5(ok2421); him-6(ok412) double mutants more RAD-51 foci were detected at all meiotic zones compared to either single mutant (Figs 1B and S1). We did not observe a dramatically elevated number of RAD-51 foci in mitotic germ cells (Fig 1B, zone 1 and 2, S1 Fig) of smc-5(ok2421); him-6(ok412), and the elevated number of RAD-51 foci depends on SPO-11 (Figs 1 and S1), confirming that SMC-5/6 and HIM-6 cooperate to process meiotic SPO-11 dependent DSBs. SMC-5/6 and HIM-6 are dispensable for meiotic chromosome axis formation and synapsis Compromised meiotic DSB repair can be due to a defect in the establishment of early meiotic events, such as chromosome axis formation and synapsis [45]. To test whether these events are Synergistic function of SMC-5/6 and the BLM helicase HIM-6. A. Genetic interaction between SMC-5/6 complex and HIM-6 was examined by combining smc-5/6 deletion with different him-6 alleles. Progeny viability in % was determined by counting number of viable eggs/total number of eggs laid. The sample size (n) indicates the number of embryos examined for each genotype. Error bars represent standard deviation of the mean. Asterisks indicate statistically significant reduction in embryonic viability in smc-5(ok2421); him-6(ok412), smc-5(tm2868); him-6(ok412), smc-5(ok2421); him-6(e1423) and smc-6(ok3294); him-6(ok412) double mutants (p < 0.005 by two-tailed Student T-test) when compared with him-6(ok412) and him-6(e1423). B. Distribution of RAD-51 foci in wild type, smc-5(ok2421), him-6(ok412), and smc-5(ok2421); him-6(ok412) animals. defective in smc-5(ok2421); him-6(ok412), we analyzed the localization of HTP-3, a component of the C. elegans axial element, and SYP-1, a component of the synaptonemal complex (SC) central region. Axial element HTP-3 coordinates meiotic DSB formation, homologous pairing and synapsis and localizes along the length of parallel DAPI tracks in pachytene stage [46]. SYP-1 is a component of the central region of the synaptonemal complex [47]. We found that the localization of HTP-3 and SYP-1 occurs normally in smc-5(ok2421) and him-6(ok412) single mutants and smc-5(ok2421); him-6(ok412) double mutants during pachytene by immunostaining (S2 Fig). We therefore conclude that SMC-5/6 and HIM-6 are dispensable for chromosome axis and SC formation.
We next directly investigated whether the CO frequency and distribution are defective in smc-5(ok2421); him-6(ok412) double mutant employing Snip-SNPs (single nucleotide polymorphisms) to differentiate between N2 and polymorphic Hawaii chromosomes [50]. We generated the respective single and double mutants with chromosome V being heterozygous for Hawaii and N2. To investigate the recombination frequency and distribution we used five snip-SNPs, which together cover 92% of chromosome V [51]. It is known that COs are enriched at the arm regions of the autosomes in C. elegans and are suppressed at the center of the chromosomes [52,53]. In the wild type, 44.7% of chromatids have a single CO, and no double COs could be detected. smc-5(ok2421) mutants did not show an altered CO frequency or distribution ( Fig 2B). Nevertheless, consistent with previous studies, him-6(ok412) shows a reduced recombination rate in general, with 33.9% of single CO chromatids [16,41,54,55] ( Fig 2B). However, although the frequency of COs was dramatically decreased in the arm regions of chromosome V in him-6(ok412) mutants, it was 1.8 fold higher in the central region which we defined as between -5 and +5.8 and comprises~one-fourth of the chromosome V. We further confirmed the occurrence of a higher CO frequency (~2-fold) at the central region of chromosome I in him-6(ok412) mutants (S3 Fig), consistent with a previous study [55]. This result is also supported by a recent report showing that the HIM-6 interacting protein RMH-1 (RMI1 homolog) antagonizes CO formation in the center of chromosomes [56]. Interestingly, when him-6 mutation combined with smc-5, CO frequency was enhanced, the strongest enhancement occurring at the center of the chromosome (Fig 2B). In addition, in smc-5 (ok2421); him-6(ok412) a significant frequency of double COs was also detectable (13.8%), indicating that CO interference may be impaired in the absence of both SMC-5/6 complex and HIM-6 ( Fig 2B).
CO formation can also be analyzed cytologically by examining the maturation of bivalents in the diakinesis stage of meiosis I. In wild type, the establishment of COs is accompanied by the differentiation of bivalents into functionally distinct short and long arms demarked by the CO site. In early diakinesis, synaptonemal complex proteins such as SYP-1/2 are concentrated on short arm of the bivalent. In late diakinesis, SYP-1/2 disappears as bivalents mature [57]. While SYP-1 could hardly be detected in the -2 oocytes (the next most proximal oocyte, A. Quantification of ZHP-3::GFP foci in pachytene nuclei of wild type, smc-5(ok2421), him-6(ok412), and smc-5(ok2421); him-6(ok412) mutants. The sample size number (n) indicates the number of germ nuclei examined for each genotype. B. Analysis of CO frequencies and distribution on chromosome V. The genetic map positions of the five SNPs, which together cover 92% of chromosome V, are indicated. n is the number of cross-progeny scored. The frequency of 2 COs, 1 CO or 0 CO per chromosome is indicated in absolute numbers and as percentage (in brackets). The relative recombination frequencies (mutant/ wild type) are indicated by different coloured tags. Red reflects the greatest increase and green reflects the greatest decrease. position relative to spermatheca, Fig 3A) in wild type and him-6 mutants, most bivalents still exhibited SYP-1 staining in -2 oocytes in smc-5 mutants. However, no visible SYP-1 remained in -1 oocytes (the most proximal oocyte) in all single mutants and wild type animals ( Fig 3A). In contrast, in smc-5(ok2421); him-6(ok412) double mutants, SYP-1 staining persisted in all -1 oocytes observed (Fig 3A arrowheads). Furthermore, bivalent maturation can be assessed by staining the axial element component HIM-3 [58]. In nuclei from wild type and him-6(ok412) single mutants, bivalents were highly condensed and appeared as compact DAPI-staining bodies with a cruciform HIM-3 staining pattern indicative of normal chiasma formation (Fig 3B). In contrast, some bivalents in smc-5(ok2421) and smc-6(ok3294) single mutants were not well resolved as previously reported (Fig 3B) [43]. Affected bivalents had an elongated chromosome axis ( Fig 3B, red arrow, for high resolution S4 Fig). Intriguingly, the bivalents in smc-5 (ok2421); him-6(ok412) double mutants were poorly condensed, HIM-3 staining was disorganized and no cruciform structure could be observed, indicating a defect in chiasmata formation at the diakinesis stage ( Fig 3B). Abrogation of SMC-5/6 and HIM-6 leads to formation of chromatin bridges between meiotic chromosomes It is known that proper recombination intermediate processing and CO formation are required for chromosome segregation [16,32,59]. Because smc-5(ok2421); him-6(ok412) double mutants have an increased number of recombination intermediates as evidenced by RAD-51 staining, compared to smc-5(ok2421) and him-6(ok412) single mutants, we sought to examine whether chromosome segregation was impaired by chromatin bridges using live cell imaging of meiotic cell divisions. In wild type and him-6(ok412) or smc-5(ok2421) single mutants, meiotic chromosomes segregated with no detectable chromatin bridges (Fig 4A, S1-S3 Videos). In contrast, in smc-5(ok2421); him-6(ok412) double mutants chromosomes were linked by chromatin bridges during anaphase I and II and could not separate properly (Fig 4A, S4 Video). Measurements of the distance between the metaphase plate and the polar body during the first meiotic division in different mutants revealed that there is no difference between wild type and smc-5 (p = 0.6575) or him-6 (p = 0.5428) single mutants, but the distance was dramatically decreased in smc-5(ok2421); him-6(ok412) double mutants (p = 0.00093, <0.001). In summary, our analysis shows that both homologs (meiosis I) and sister chromatids (meiosis II) are connected by chromatin bridges in smc-5(ok2421); him-6(ok412) double mutants (Fig 4A and 4B, S4 Video). This finding is consistent with recent observation that SMC-5 and RMH-1 cooperate to prevent accumulation of aberrant chromosome connections [56].
We next carefully examined diakinesis chromosomes from different mutants using super resolution microscopy. While wild type had six bivalents per oocyte and no univalent can be detected, most of the oocytes (88%) from him-6(ok412) mutants we examined had some univalents as indicated by more than six DAPI staining bodies (S5 Fig for quantification). In contrast, no univalent could be observed in smc-5(ok2421); him-6(ok412) double mutants, and 53% of oocytes have less than six DAPI stained bodies, bivalents linked by chromosome bridges being scored by us as one (Fig 4C, red arrow for chromatin bridge, S3 Fig for quantification, S5 Video, see linked bivalent at the bottom left). smc-5(ok2421) mutants occasionally contained linked bivalents (S5 Fig for quantification).
The Smc5/6 complex and the BLM helicase have been implicated in DNA repair in mitotically dividing cells [60,61]. Thus, the chromatin bridges could represent unresolved DNA repair intermediates, possibly carried over from mitotic cell divisions. Alternatively, the bridges we observed might originate from SPO-11 induced DSBs, and represent unresolved meiotic recombination intermediates. To distinguish between these possibilities, we generated smc-5(ok2421); him-6 (ok412); spo-11 triple mutants. Loss of SPO-11 abolishes meiotic recombination, which leads to presence of twelve univalents in diakinesis oocytes due to the absence of chiasma between homolog pairs. In smc-5(ok2421); him-6(ok412); spo-11 triple mutants, we rarely observed chromosome bridges between univalents. Although chromosome fragments, which likely originated from defects in mitotic DNA repair, could be detected, the number of DAPI-stained bodies didn't change significantly (11.9±1.37, N = 20) (Fig 4D). We thus conclude that the chromatin bridges are largely derived from SPO-11 induced unresolved meiotic recombination intermediates.
Chromosomes undergo extensive structural reorganization to achieve accurate chromosome segregation during meiosis [67]. Chromosome condensation involves the remodeling of each pair of homologous chromosomes around the site of CO into highly condensed cruciform bivalents [68]. Mutation of hcp-6, a component of condensin II, leads to a dramatically decreased number of cruciform bivalents and defective chromosome segregation during meiotic cell divisions [69,70]. The phenotypes reported in hcp-6 mutants are reminiscent of the phenotypes we observed in in smc-5(ok2421); him-6(ok412) double mutants [70], indicating that the depletion of SMC-5/6 and HIM-6 could similarly lead to a defect in higher-ordered chromosome organization. We further examined the distribution of condensin in wild type and the various single and double mutants by HCP-6 staining. In the wild type and the him-6 (ok412) mutant, HCP-6 is associated with sister chromatids and formed four distinct well organized patches in each bivalent (Fig 6A). In the smc-5(ok2421) single mutant, although HCP-6 still formed patches, an increased number of patches were detected in some bivalents ( Fig 6A). However, in the in smc-5(ok2421); him-6(ok412) double mutant HCP-6 staining is scattered along the bivalent and no distinct well-organized patches can be observed, indicating that the distribution of condensin is severely compromised in the absence of SMC-5/6 and HIM-6 ( Fig  6A). hcp-6 RNAi has previously been shown to lead to a partial HCP-6 depletion in the germ line and, to the formation of chromatin bridges during meiotic divisions [70]. Given that the embryonic lethality of in smc-5(ok2421); him-6(ok412) can be partially rescued by depletion of BRC-1/BRD-1 and that the chromatin bridges in in smc-5(ok2421); him-6(ok412) mutant are brc-1 dependent, we tested whether mutation of brc-1 or brd-1 could also rescue the lethality conferred by hcp-6 depletion. As expected we found that the RNAi depletion of HCP-6 in wild type cause 100% embryonic lethality. However progeny viability is rescued to 65% and 23% when HCP-6 is depleted in brc-1 and brd-1 respectively (Fig 6B), suggesting that BRC-1/BRD-1 may be also required for the formation of toxic chromatin bridges when chromosome condensation is defective during meiosis.

Discussion
In this study we examined the synergistic functions of the BLM helicase and SMC-5/6 during C. elegans meiosis. We show that BLM helicase and SMC-5/6 complex cooperate to process meiotic recombination intermediates and promote the maturation into structured bivalents. We also uncover an unexpected role of HIM-6 in regulating CO distribution. I) HIM-6 and SMC-5/6 are required for normal CO formation and distribution In all organisms studied so far, a significant excess of DSBs are generated relative to the number of CO [49,[71][72][73][74][75]. In C. elegans, although each chromosome undergoes several DSBs, only one DSB results in interhomolog CO between each chromosome pair. The remainder are repaired either by interhomolog non-CO or the intersister recombination pathway [76]. BLM helicase is highly conserved, but the functions of BLM in CO formation differ in various organisms. The budding yeast BLM homologue Sgs1 shows anti-CO activity. Mutation of sgs1 led to 50% increase in CO frequency [77]. In contrast, the fission yeast BLM homologue Rqh1 is able to promote the recombination outcome towards CO formation [78]. Similarly, meiotic recombination is reduced to about half of the wild type frequency in D. melanogaster in the absence of the BLM ortholog mus309 [79]. In C. elegans, the BLM homolog HIM-6 has been found to be required for normal levels of recombination during meiosis [16,41,54,55]. Mutations in him-6 lead to a decrease in progeny viability, reduced number of COs between homolog chromosomes and increased univalent formation. Interestingly, our careful examination of CO frequency and distribution revealed that although the overall CO level is decreased in him-6 mutants, him-6 leads to a strong reduction of COs in the arm regions while the frequency of COs in the central region is significantly increased. Our findings, which are based on the analysis of two chromosomes, are consistent with a previous study showing an altered CO frequency and distribution in him-6 mutants [55]. Our data are also supported by a recent publication showing the same effect when rmh-1 encoding for a regulatory subunit of HIM-6/Blooms is deleted [56]. Thus, HIM-6 might have a pro-CO activity at arm regions but have an anti-CO activity in the center of chromosomes (Fig 7A). How is CO distribution regulated by HIM-6 in C. elegans meiosis? In mice, the BLM helicase colocalizes with the recombinases RAD51 and DMC1 and probably has an early role in the CO/non-CO decision [80]. Additionally, the BLM helicase has been shown to interact physically or functionally with different partners [81]. On one hand, BLM can form a complex with TopIIIα and RMI and is able to dissolve double HJ to generate non-CO [82]. On the other hand, the C. elegans BLM helicase HIM-6 can also act with XPF-1 endonuclease as a HJ resolvase to promote CO formation [16]. Therefore, the proor anti-CO activity of BLM on different regions of chromosome might be conferred by different BLM complexes. Notably, the CO distribution shifting from the arm region to the center of chromosome has also been reported for the C. elegans rec-1, xnd-1 and him-5 mutants and a weaker such phenotype also occurs in slx-1 [53,[83][84][85]. While XND-1 is thought to regulate CO distribution by modulating acetylation levels of histone H2A lysine 5, REC-1 and HIM-5 act redundantly to facilitate the formation of meiotic DSBs and the phosphorylation of REC-1 by cyclin-dependent kinase could be important for the CO distribution [86].
At the moment we can only speculate as to why recombination is increased at the center of the chromosomes in him-6 mutants, why this effect is stronger in smc-5(ok2421); him-6(ok412) Schematic diagram depicting the CO formation in the wild type (i) or in the smc-5/6 mutants (ii). In the wild type, SMC-5/6 complex can promote the repair of SPO-11 induced DSBs by intersister recombination pathways. The CO formation generated by interhomolog recombination can be regulated by HIM-6. However, in the absence of SMC-5/6, more DSBs are channelled to be repaired by interhomolog recombination due to the compromised intersister recombination. In this case, HIM-6 becomes essential to maintain a normal CO landscape. B. Reduced chromosome compaction caused by smc-5/6 and him-6 mutation or by condensin depletion may lead to formation of multi-chromatid joint molecules supressed by brc-1. The dashed lines represent a potential role of SMC-5/6 in preventing the chromosome decondensation probably through cross-talk between SMC-5/6 complex and condensin II complex. double mutants, and why CO recombination is generally increased in this double mutant. Notably, neither smc-5(ok2421); him-6(ok412) double mutants nor the corresponding single mutants show an excessive number of COSA-1 foci, arguing that 'excessive recombination' in the double mutants might not be due to an increased number of CO designated sites, but might reflect COs outcome of ectopic recombination intermediates. Such increased recombination could be due to delayed disassembly of the synaptonemal complex observed in smc-5(ok2421); him-6(ok412) double mutants, as indicated by a prolonged SYP-1 staining on the bivalents in the -1 oocytes at diakinesis. Such delayed desynapsis could promote interhomolog recombination to form excessive COs and could also account for the increased chance for COs occurring at the central chromosomal regions. However, this model cannot account for increased recombination at the center of him-6 chromosomes as we did not observe a desynapsis delay in him-6 single mutants. Furthermore, also not consistent with this model, the frequency and distribution of COs are not altered in smc-5 single mutants, despite desynapsis being delayed.
In budding yeast, the Smc5/6 complex is critical for interhomolog bias by promoting the orderly formation of interhomolog recombination intermediates [32][33][34]. In contrast, we confirmed a previous study showing that the SMC-5/6 complex is not required for CO formation in C. elegans and has a weak or negligible effect on interhomolog recombination [43]. This previous study also revealed that when interhomolog recombination is disrupted by depletion of the chromosome axis component HIM-3 or the synaptonemal complex element SYP-2 in smc-5/6 mutants, chromosome fragments occur. This phenotype indicates an important function of SMC-5/6 in promoting intersister recombination during meiosis in C. elegans [43]. Therefore, in line with previous studies and our genetic and recombination mapping data, we propose that SMC-5/6 and HIM-6 act synergistically to repair meiotic DSBs and regulate CO formation (Fig 7A). While HIM-6 is required for interhomolog recombination, a proportion of meiotic DSBs are repaired by the SMC-5/6 promoted intersister recombination. Compromised intersister recombination repair caused by loss of SMC-5/6 complex probably results in more DSBs to be repaired by interhomolog recombination. In such a scenario increased interhomolog recombination might be quelled by HIM-6, the additional depletion of which would allow for excessive CO formation in addition to an abnormal CO interference. II) SMC-5/6 complex and HIM-6 are required for correct chromosome organization during meiosis The SMC complexes, cohesion, condensin and SMC-5/6 complex, are essential for chromosome organization and dynamics. Numerous studies have revealed that chromosome morphogenesis is important for meiotic recombination regulation and CO formation [67]. Condensin is essential for preventing/resolving chromatin bridges formation probably by maintaining an organized and ordered chromosome structure during meiotic recombination [70]. Depletion of Smc5/6 in human mitotic cells led to a decondensed chromosome conformation accompanied by defective condensin localization, indicating that there might be cross-talk between Smc5/6 and condensin [87]. Related to this, we show that the chromosome organization and condensin distribution are abnormal in smc-5(ok2421) and smc-6(ok3297) mutants in C. elegans. Further depletion of the BLM helicase HIM-6 in smc-5(ok2421) mutant led to much more severe defects in chromosome organization. The phenotype we observe is reminiscent to condensation defects, a disrupted chiasma structure and the formation of chromatin bridges that occurs upon HCP-6 depletion [69,70]. The partial decondensation of pachytene chromosomes in dosage compensation mutants (for instance dpy-28) has been previously linked to increased recombination [88,89]. Thus the excessive recombination occurring in smc-5 (ok2421); him-6(ok412) could be due to the failure to form properly condensed chromosomes. While we have not observed overt chromosome decondensation in smc-5(ok2421); him-6 (ok412) mutants during pachytene, the exact state of those pachytene chromosomes remains to be investigated.
How could him-6 mutations enhance the chromosome disorganization observed in smc-5/6 mutants? It is possible that defective chromosome organization in smc-5(ok2421); him-6 (ok412) double mutants could be partially due to the compromised meiotic DSBs repaired by HR. An extreme case of such a situation occurs in rad-51 mutants, where spo-11 dependent DSBs cannot be repaired by HR, leading to an uncondensed mass of chromatin [90]. It is also possible that the frequency of ectopic recombination could be increased when the chromosome is decondensed (Fig 7B). Toxic recombination intermediates generated by ectopic recombination, such as multi-chromatid joint molecules in budding yeast, are resolved or dissolved by BLM helicase homolog Sgs1 [91,92]. In the absence of BLM helicase, unresolved multi-chromatid intermediates could lead to the formation of aberrant chromosome organization and chromatin bridges. Interestingly, we find that depleting the BRCA1 ortholog BRC-1 could rescue the progeny lethality conferred by smc-5(ok2421); him-6(ok412) mutants and by the RNAi depletion of hcp-6. BRCA1 plays a role in meiotic DNA-damage repair and CO formation during spermatogenesis in mice [39]. BRCA1 deficiency resulted in decreased number of MSH4 foci and delayed appearance of MLH1 foci, indicating that mouse BRCA1 could be involved in regulating efficient formation or stabilization of meiotic recombination intermediates [38]. Consistent with this possibility, we found that depletion of BRC-1 in smc-5(ok2421); him-6 (ok412) double mutants and hcp-6 (RNAi) worms suppressed chromatin bridge formation in C. elegans during meiosis, suggesting that BRC-1 might promote ectopic recombination by driving the formation or stabilization of aberrant joint molecules between non homologous chromosomes and between chromatids, possibly leading to multi-chromatid linkage formation (Fig 7B).
In conclusion, our results suggest that the BLM helicase HIM-6 and the SMC-5/6 complex act synergistically to promote recombination intermediate processing and chromosome maturation during C. elegans meiosis. We also reveal a role of the BLM helicase HIM-6 in regulating CO distribution. Finally, we highlight an important role of chromosome architecture in preventing ectopic meiotic recombination.

C. elegans strains and maintenance
All strains were maintained at 20°C under standard conditions. N2 Bristol was used as the wild type strain. CB4856 Hawaii was used to generate strains for CO recombination frequency analysis. Mutant strains used in this study are listed in S1 Table. All the mutants used in this study were obtained from the Caenorhabditis Genetics Center. Details are described at the National Bioresource Project for the nematode and on www. wormbase.org. All mutants were outcrossed for a minimum of four times to the wild type strain to eliminate background mutations. The TG2512 gtIs2512 [Ppie-1::his-11::GFP unc-119+] strain was generating by biolistic bombardment of pAZ132 of unc-119(ed3) mutants.

RNA interference
RNA interference by feeding was performed using bacteria RNAi feeding strains from the Ahringer library [93]. L4 worms were placed on plates seeded with bacteria expressioning dsRNA. After 24 hours, the adult animals were transferred to new RNAi plates and allowed to lay eggs for approximately 6 hours. The resulting progeny viability for wild type, brc-1 and brd-1 mutants were scored. Assays were performed at 20°C. Bacteria containing empty RNAi vector L4440 or mcm-7 RNAi vector were used for control experiments.

Recordings of meiotic divisions
Embryos were dissected in isotonic growth medium for blastomeres containing 35% bovine FCS. Before use, bovine FCS (heat treated for 30 min at 56°C; Invitrogen) was added. Embryos were mounted on 2% agarose pads. Vaseline patches on the slide were used to reduce the pressure of the coverslip on the embryo. Images were captured every 10 seconds using a widefield DeltaVision microscope. Exposure time was 250 milliseconds and binning used was 2×2. Image analysis and video processing were performed using ImageJ software,

Image acquisition
Microscopy images were acquired with a Delta Vision Image restoration system (Applied Precision). Raw data obtained were analysed and deconvolved using SoftWoRx Suite and soft-WoRx Explorer software (Applied Precision, Issaquah, WA, USA). For SIM microscopy, established protocols were followed [16]. Images were acquired using a UPlanSApochromat 100× 1.4NA, oil immersion objective lens (Olympus, Center Valley, PA) and back-illuminated Cascade II 512×512 electron-multiplying charge-coupled device (EMCCD) camera (Photometrics, Tucson, AZ) on the SIM version 3 system (Applied Precision) equipped with 405-, 488-, and 593-nm solid-state lasers. Samples were illuminated by a coherent scrambled laser light source that had passed through a diffraction grating to generate the structured illumination by interference of light orders in the image plane to create a 3D sinusoidal pattern, with lateral stripes approximately 0.2 μm apart. The pattern was shifted laterally through five phases and through three angular rotations of 60°for each Z-section, separated by 0.125 μm. Exposure times were typically between 100 and 200 ms, and the power of each laser was adjusted to achieve optimal intensities of between 2,000 and 4,000 counts in a raw image of 16-bit dynamic range, at the lowest possible laser power to minimize photo bleaching. Raw images were processed and reconstructed to reveal structures with greater resolution implemented on Soft-WoRx, ver. 6.0 (Applied Precision, Inc.). The channels were then aligned in x, y, and rotationally using predetermined shifts as measured using 100 nm TetraSpeck (Invitrogen) beads with the SoftWoRx alignment tool (Applied Precision, Inc.).

Determining meiotic crossover recombination frequencies
Meiotic CO recombination frequencies were assayed essentially as described, using five snip-SNPs on Chromosome V that differ between N2 Bristol and CB4856 Hawaii [51]. Strains used to determine CO recombination assays were crossed into Hawaii to obtain mutant strains carrying ChrV homozygous for Hawaii DNA. Single and double mutant strains containing smc-5 were balanced with mIn1. GFP positive balanced mutant males with ChrV homozygous Hawaii were then crossed with hermaphrodites of identical genotype in the N2 Bristol background to obtain mutant strains heterozygous for Hawaii. Non-GFP homozygous mutant F1 cross-progeny hermaphrodites were then crossed with males of CB5584, a myo-2::GFP expressing strain, which expresses high levels of green fluorescent protein in pharyngeal muscles, allowed to lay eggs for 24-48 h before removing them for genotype confirmation by PCR and DraI digest. 100-200 individual F1 0 GFP-positive embryos and larvae were lysed and analysed for CO recombination by PCR and DraI digest. , him-6(ok412), smc-5(ok2421); him-6(ok412) and smc-5(ok2421); brc-1; him-6(ok412) mutants. The sample size (n) indicates the total number of -1 diakinesis oocytes examined for each genotype. The proportion of oocytes with six DAPI staining body in him-6 (11.8%, n = 51), smc-5; him-6 double mutants (46.7%, n = 45) and smc-5; brc-1; him-6 (80.6%, n = 31) were significantly different from the wild type (100%, n = 49) (p<0.05). There is no difference between wild type (100%, n = 49) and smc-5 (97.4%, n = 39) (p = 0.258). Statistical significance was determined by two-tailed Z-test for two population proportions. P Values below 0.05 were considered significant. are grateful to Ulrike Gartner for proofreading. We thank the anonymous reviewers for their constructive comments that improved the manuscript and refined the interpretation of our data.