Caenorhabditis elegans HIM-18/SLX-4 Interacts with SLX-1 and XPF-1 and Maintains Genomic Integrity in the Germline by Processing Recombination Intermediates

Homologous recombination (HR) is essential for the repair of blocked or collapsed replication forks and for the production of crossovers between homologs that promote accurate meiotic chromosome segregation. Here, we identify HIM-18, an ortholog of MUS312/Slx4, as a critical player required in vivo for processing late HR intermediates in Caenorhabditis elegans. DNA damage sensitivity and an accumulation of HR intermediates (RAD-51 foci) during premeiotic entry suggest that HIM-18 is required for HR–mediated repair at stalled replication forks. A reduction in crossover recombination frequencies—accompanied by an increase in HR intermediates during meiosis, germ cell apoptosis, unstable bivalent attachments, and subsequent chromosome nondisjunction—support a role for HIM-18 in converting HR intermediates into crossover products. Such a role is suggested by physical interaction of HIM-18 with the nucleases SLX-1 and XPF-1 and by the synthetic lethality of him-18 with him-6, the C. elegans BLM homolog. We propose that HIM-18 facilitates processing of HR intermediates resulting from replication fork collapse and programmed meiotic DSBs in the C. elegans germline.


Introduction
DNA double-strand breaks (DSBs) can arise in various ways, including as a result of the collapse of stalled replication forks, exposure to DNA damaging agents and the formation of programmed meiotic DSBs [1]. The importance of DSB repair is therefore highlighted by its critical roles in replication fork restart, the maintenance of genomic integrity and promoting faithful meiotic chromosome segregation. Homologous recombination (HR) provides an efficient and accurate repair of DSBs, in part through the use of an intact donor template for repair. Current models of HR propose that following DSB formation, the DSB ends are resected in a 59 to 39 orientation generating 39 single-stranded DNA (ssDNA) tails [2,3]. Repair can then proceed through different pathways, one of which involves the association of the Rad51 recombinase protein with the ssDNA tails. This generates a nucleoprotein filament that engages in strand invasion with either an intact sister or homologous chromosome resulting in the formation of a D-loop structure. Subsequent second end capture, DNA synthesis and ligation, results in the formation of two four-way DNA junction intermediates referred to as Holliday junctions (HJ). Double HJ (dHJ) intermediates can be resolved through cleavage by HJ resolvases, which results in either a crossover (CO) or noncrossover (NCO) product, or undergo dissolution mediated by RecQ helicases in combination with topoisomerase activity, resulting in NCOs.
Identifying components involved in the processing of HR intermediates has been critical to understand the molecular mechanisms of HR. HJ resolvases have been identified in poxviruses (A22R), bacteriophages (T4 endonuclease VII and T7 endonuclease I), and E. coli (RuvC) [4]. In S. cerevisiae and S. pombe, Cce1 and Cce1/Ydc2, respectively, were identified as HJ resolvases that act in the mitochondria [5,6]. Recently, H. sapiens GEN1 and its homolog in S. cerevisiae, Yen1, have been reported to resolve HJs in vitro via a canonical RuvC-like symmetrical cleavage that does not require further processing [7]. However, their in vivo function remains unclear. Moreover, HJ processing can also be achieved through asymmetric cleavage, as exemplified by the biochemical activity of the MUS81-EME1 complex in eukaryotic cells [8,9].
The identification of proteins involved in HJ processing has been challenging due to the crossfunctionality observed for nucleases throughout different HR pathways. This crossfunctionality can be partly due to changes in complex composition that may modulate specificity of recognition and processing of a given intermediate. In D. melanogaster, where 90-95% of meiotic COs do not require MUS81 [10], COs are dependent on a protein complex consisting of MEI-9, an ortholog of the mammalian XPF and S. cerevisiae Rad1 nucleotide excision repair (NER) endonucleases, ERCC1, an XPF interaction partner, MUS312 and HDM, a member of a superfamily of proteins with ssDNA binding activity [11][12][13][14][15][16]. However, ERCC1 and HDM are only required for a subset of CO events [12,13]. In S. cerevisiae, SLX4 was first identified in a synthetic-lethal screen for genes required for viability in the absence of SGS1, a member of the RecQ family of DNA helicases implicated in the human Bloom and Werner syndromes [17]. Slx4 binds to two structure-specific endonucleases, Slx1 and Rad1, in a mutually exclusive manner [17,18]. The Slx1-Slx4 complex cleaves multiple branched DNA substrates in vitro, particularly 59-flap, simple-Y, HJ and replication fork structures [19], and functions in rDNA maintenance during S phase [20,21]. Meanwhile, the Slx4-Rad1 complex is required for the single-strand annealing (SSA) HR pathway, in which Mec1/ Tel1 dependent phosphorylation of Slx4 is essential [18,22]. Furthermore, an in vivo nuclease activity of Slx4 without Slx1 or Rad1 is proposed [19,23,24].
Here, through a functional genomics approach in the nematode C. elegans, we identified HIM-18, which shares sequence similarity with S. cerevisiae Slx4 and D. melanogaster MUS312. HIM-18 is localized to mitotic nuclei at the premeiotic tip and to meiotic nuclei from late pachytene through diakinesis in wild type germlines. The accumulation of HR intermediates stemming from replication fork collapse in mitosis and from SPO-11-dependent programmed meiotic DSBs in mid to late pachytene observed in him-18 mutant germlines, coupled with the localization pattern of HIM-18, suggests a role in both mitotic and meiotic DSB repair. This is further supported by the observation of reduced CO frequencies and the premature disassembly of bivalent attachments at prometaphase I resulting in increased chromosome nondisjunction in him-18 mutants. Moreover, HIM-18 interacts with SLX-1 and XPF-1, and him-18 mutants show similar DNA damage sensitivity and synthetic lethality with him-6/BLM as observed in slx4 mutants in yeast. Taken together, our analysis suggests that HIM-18 is required for the maintenance of genomic integrity in germline nuclei. We propose a model in which HIM-18/SLX-4 promotes the processing of late HR intermediates in the germline resulting from replication fork collapse in mitotic nuclei and programmed DSBs in meiotic nuclei.

HIM-18 Is a Conserved Chromosome-Associated Protein Present in Both Mitotic and Late Meiotic Prophase Germline Nuclei
him-18 (open reading frame T04A8. 15) was identified in an RNA interference (RNAi) screen performed as in [25] and designed to detect meiotic candidates from among germlineenriched genes [ [26] and M. Colaiácovo, unpublished results]. The HIM-18 protein harbors two DNA binding motifs (zinc finger and SAP) and three protein binding motifs (coiled-coil, BTB and leucine-zipper) ( Figure 1A). Homology searches revealed HIM-18 orthologs from yeast to humans ( Figure 1B). Specifically, the SAP motif is highly conserved throughout these orthologs, with the exception of MUS312 in D. melanogaster. This conservation is particularly interesting because the SAP motif, present in the S. pombe mitochondrial HJ resolvase Cce1/Ydc2, is implicated in promoting HJ binding and resolution [27,28]. Furthermore, the mammalian orthologs, named BTBD12, also contain BTB and zinc finger motifs. Therefore, this sequence analysis suggests that HIM-18 is conserved among eukaryotic organisms and shares homology with proteins predicted to function in the processing of HR intermediates.
The him-18(tm2181) mutant, obtained from the Japanese National Bioresource Project, carries a 394 bp out-of-frame deletion encompassing parts of exons 5 and 6 ( Figure 1A). This deletion results in a premature stop codon and the loss of the predicted coiled-coil, BTB, SAP and leucine-zipper motifs. The analysis of DAPI-stained germlines of him-18(tm2181)/+ hermaphrodites, and a genetic analysis of both embryonic lethality and the incidence of males among the progeny of these worms, indicated that these heterozygotes were indistinguishable from wild type ( Table 1). This indicates that tm2181 is a recessive allele of him-18. A similar analysis indicated that tm2181 homozygotes were indistinguishable from transheterozygotes for tm2181 and sDf121, a deficiency encompassing the him-18 locus (Table 1). Moreover, an affinity purified HIM-18 antibody raised against the first 166 amino acids present in HIM-18, failed to detect a HIM-18 signal on immunostained whole mounted gonads from him-18(tm2181) mutants ( Figure S1). Taken together, these studies suggest that him-18(tm2181) is a null.
To investigate the localization of HIM-18 in the germline, dissected gonads from wild type hermaphrodites were immunostained with the affinity purified HIM-18 N-terminal antibody ( Figure 1C). HIM-18 is first observed in mitotic nuclei at the distal tip region of the germline (premeiotic tip). However, the HIM-18 signal is abruptly reduced as nuclei enter into meiotic prophase, being barely visible in the transition zone (leptotene/zygotene) or in early to mid pachytene nuclei. HIM-18 signal is then clearly detected once again in late pachytene nuclei, persisting through the end of diakinesis. At a higher resolution, the HIM-18 signal is observed as nuclear foci (mainly around the chromosomes) in the premeiotic tip, late pachytene and diplotene/diakinesis stages ( Figure 1C). Although the HIM-18 signal is reduced from transition zone through mid-pachytene, a low level of HIM-18 foci, both on and around chromosomes, is still apparent at these stages upon longer exposure ( Figure 1C). Finally, HIM-18 is no longer observed localizing onto chromosomes after diakinesis (data not shown).
Homologous Pairing, Axis Morphogenesis, and Synapsis Are Normal in him-18 Mutants Analysis of him-18(tm2181) homozygous mutants and following depletion of him-18 by RNAi revealed phenotypes suggestive of

Author Summary
Homologous recombination (HR) is a process that provides for the accurate and efficient repair of DNA double-strand breaks (DSBs) incurred by cells, thereby maintaining genomic integrity. Proper processing of HR intermediates is critical for biological processes ranging from replication fork restart to the accurate partitioning of chromosomes during meiotic cell divisions. This is further emphasized by the fact that impaired processing of HR intermediates in both mitotic and meiotic cells can result in tumorigenesis and congenital defects. Therefore, the identification of components involved in HR is essential to understand the molecular mechanism of HR. Here, we identify HIM-18/SLX-4 in C. elegans, a protein conserved from yeast to humans that interacts with the nucleases SLX-1 and XPF-1 and is required for DSB repair in the germline. Impaired HIM-18 function results in increased DNA damage sensitivity, the accumulation of recombination intermediates, decreased meiotic crossover frequencies, altered late meiotic chromosome remodeling, the formation of fragile connections between homologs, and an increased chromosome nondisjunction. Finally, HIM-18 is localized to both mitotic and meiotic nuclei in wild-type germlines. We propose that HIM-18 function is required during the processing of late HR intermediates resulting from replication fork collapse and meiotic DSBs. Immunolocalization of HIM-18 in germline nuclei of wild type hermaphrodites. Low magnification images show nuclei progressing from the premeiotic tip through all the stages of meiotic prophase, in a left to right orientation. Arrowheads indicate the beginning of transition zone and entrance into meiosis. A single focal plane from a representative nucleus of each stage is shown at a higher magnification. Detection thresholds for a-HIM-18 signals in transition zone and mid-pachytene panels were lower than for other panels to permit imaging of the lower levels of chromosomeassociated HIM-18 observed in these stages. Bars, 20 mm for whole gonad, 1 mm in insets. doi:10.1371/journal.pgen.1000735.g001 errors in meiotic chromosome segregation such as an increased embryonic lethality (79.9%; n = 2748, and 72.5%; n = 436, respectively) and a high incidence of males (11.9% and 5.5%, respectively) among their surviving progeny (Table 1 and Figure  S2A). Errors in meiotic chromosome segregation can stem from earlier defects in homologous chromosome pairing, axis morphogenesis or synapsis. We examined homologous pairing via fluorescence in situ hybridization (FISH) and immunofluorescence analysis, by monitoring both the establishment and maintenance of pairing, as chromosomes enter into meiotic prophase at the transition zone and progress into pachytene where they are fully synapsed. Specifically, we utilized a FISH probe to the 5S ribosomal DNA region (chromosome V) and observed wild type levels of pairing between homologs in him-18 mutants both in the transition zone and pachytene nuclei (Figure 2A, 2B, and 2E and Figure S3). 98% (n = 107) of the pachytene nuclei examined in him-18 mutants carried fully paired homologs in this analysis, compared to 97% (n = 101) in wild type. We obtained a similar result by immunostaining both wild type and him-18 mutant gonads with a HIM-8 antibody, which specifically localizes to the pairing center end of the X chromosome [29] ( Figure 2C and 2D and data not shown). Therefore, homologous pairing is normal for autosomes and the X chromosome in him-18 mutants. We determined that axis morphogenesis was indistinguishable from wild type, as exemplified by the normal kinetics and pattern of localization of axis-associated proteins such as the meiosis specific cohesin REC-8 and the cohesin SMC-3 ( Figure S4). We then examined the formation of the synaptonemal complex (SC), the proteinaceous scaffold that forms between fully paired and aligned homologous chromosomes during meiosis. Immunostaining for SYP-1, a SC central region protein, revealed a SYP-1 localization between paired homologous chromosomes that is indistinguishable from wild type both at the transition zone and pachytene ( Figure 2 and data not shown). Taken together, these results indicate that events occurring upon entrance into meiosis, such as homologous pairing, axis morphogenesis and synapsis do not require HIM-18.

SPO-11-Dependent and -Independent Recombination Intermediates Accumulate in him-18 Mutants
To investigate whether the defects in meiotic chromosome segregation observed in him-18 mutants reflect defects in DSB repair, we performed a quantitative comparison of the levels of RAD-51 foci in the germlines of both wild type and him-18 mutants ( Figure 3A and Figure S5). Since nuclei are positioned in a temporal-spatial gradient throughout the germline in C. elegans, proceeding in a distal to proximal orientation from mitosis into the various stages of meiotic prophase I, levels of RAD-51 foci were assessed both in mitotic (zones 1 and 2) and meiotic (zones 3-7) nuclei ( Figure 3A and 3B, Figures S5, S6, S7, S8 and [30]). In wild type, only a few mitotic RAD-51 foci were observed at zones 1 and 2 (5.8% and 1.2% of nuclei contained 1 and 2-3 RAD-51 foci, respectively; Figure 3A and Figure S5). In these mitotic nuclei, RAD-51 foci are thought to be mainly derived from singlestranded DNA gaps formed at stalled replication forks or resected DSBs resulting from collapsed replication forks [31]. During meiotic prophase, as a result of SPO-11-dependent programmed meiotic DSB formation, levels of RAD-51 foci start to rise at the transition zone (zone 3), then accumulate maximally at early to mid-pachytene (zones 4 and 5; nearly 90% of nuclei contain an average of 3.3 RAD-51 foci), and are reduced at late pachytene (zones 6 and 7) [30]. In him-18 mutants, levels of RAD-51 foci were higher than those observed in wild type germlines, both in mitotic (23%, 14% and 3% of nuclei contained 1, 2-3 and 4-6 RAD-51 foci, respectively, in zones 1 and 2; P,0.0001 for both zones) and meiotic nuclei (an average of 5 RAD-51 foci/nucleus in zones 4 and 5; P,0.0001 and P = 0.0021, respectively) ( Figure 3A and Figure S5). Moreover, higher levels of RAD-51 foci persisted through late pachytene in him-18 mutants compared to wild type (3.2 RAD-51 foci/nucleus compared to 0.9, P,0.0001, and 0.9 foci/nucleus compared to 0.1, P = 0.0001, in zones 6 and 7, respectively) suggesting either a delay in meiotic DSB repair or an overall increase in the levels of DSBs formed during meiosis. Interestingly, larger RAD-51 foci were observed throughout both mitotic and meiotic nuclei in him-18 mutants compared to wild type ( Figure S8), some of which were still present in pachytene nuclei in him-18; spo-11 double mutants. However, these SPO-11-independent RAD-51 foci failed to result in physical attachments between homologs (chiasmata). Specifically, instead of six bivalents, corresponding to the six pairs of homologous chromosomes held together by chiasmata, observed in DAPI-stained diakinesis oocytes in wild type, an average of 11.9 DAPI-stained bodies were observed in him-18; spo-11 double mutants (n = 36), similar to spo-11 single mutants (11.7 DAPI-stained bodies; n = 44) ( Figure 3B and 3C). We also observed elevated levels of germ cell apoptosis in both him-18 and him-18; spo-11 double mutants, compared to wild type (P,0.0001 and P = 0.0009, respectively) and spo-11 mutants (P,0.0001, respectively; Figure 4A and 4B). Furthermore, the elevated germ cell apoptosis in both him-18 and him-18;spo-11 mutants was suppressed following depletion of cep-1/p53 by RNAi ( Figure S2C and Figure S9), suggesting that the elevated apoptosis stems from the activation of the DNA damage checkpoint in late pachytene as a result of damage incurred both during mitosis and meiosis. Altogether, these observations suggest that SPO-11independent mitotic RAD-51 foci persist into pachytene and that these unresolved recombination intermediates contribute in part to the increase in germ cell apoptosis observed in him-18 mutants. However, the increased RAD-51 foci and germ cell apoptosis observed in mid to late pachytene, also support a role for HIM-18 specifically in HR during meiosis and are not simply a result of DNA damage being carried over into meiosis from defects in mitosis. This is further supported by the observation that events occurring upon entrance into meiosis were indistinguishable from wild type, and that despite homologous pairing and synapsis, repair of SPO-11dependent DSBs was impaired.

Loss of HIM-18 Confers DNA Damage Sensitivity
To gain further support for a role for HIM-18 in DNA damage repair, young adult hermaphrodites were exposed to different types of DNA damage and embryonic viability was monitored as an index of sensitivity (see Experimental Procedures). Embryonic viability after DNA damage treatment was plotted as a percentage of the hatching after DNA damage normalized by that in untreated him-18 mutants. him-18 mutants were hypersensitive to DSBs induced by c-irradiation (IR). Only 60%, 5% and 0% hatching was observed in him-18 mutants exposed to 10, 50 and 100 Gy, respectively, compared to wild type worms where a high level of hatching was observed even following the highest IR exposure level (83% at 100 Gy) ( Figure 4C). Exposure to nitrogen mustard (HN2), which induces DNA interstrand crosslinks (ICLs) that obstruct essential cellular processes such as transcription and replication, resulted in significantly decreased hatching levels in him-18 mutants (5% hatching at 100 mM HN2) compared to wild type (95% hatching at 100 mM HN2) ( Figure 4D). To further examine the role of HIM-18 in responding to lesions that affect replication fork progression, worms were exposed to camptothecin (CPT), which inhibits the detachment of topoisomerase I from DNA, thus preventing DNA re-ligation at a single-strand nick, which in turn results in a single ended DNA double-strand break when collision of a replication fork occurs at the lesion [32]. Treatment with CPT resulted in a decrease in hatching in him-18 mutants (17% hatching at 500 nM CPT) compared to wild type (96% at 500 nM CPT) ( Figure 4E). In addition, both wild type and him-18 mutants were examined following exposure to the ribonucleotide reductase inhibitor hydroxyurea (HU), which results in a checkpoint-dependent cell cycle arrest ( Figure S10C). him-18 mutants showed hypersensitivity to HU, further suggesting that him-18/slx-4 is required to resolve stalled replication forks. Furthermore, the levels of RPA-1 foci were indistinguishable between both wild type and him-18 germlines ( Figure S10A), in contrast to the increase in RAD-51 foci observed at the premeiotic region in him-18 mutants compared to wild type ( Figure 3A and Figure S5). This observation suggests that the frequency of replication stalling is similar between wild type and him-18 mutants, but that there is a defect in the recovery from stalled replication forks in him-18 mutants. Mitotic germ cell nuclei with larger nuclear diameters are observed in him-18 mutants compared to wild type even prior to HU treatment, further suggesting replication stress is occurring in this background ( Figure S10B and Table S1). Following HU treatment, larger nuclear diameters were observed in mitotic germ cell nuclei in both wild type and him-18 mutants, suggesting that the S phase checkpoint is intact in him-18 mutants. Since the checkpoint is apparently intact, but reduced survival was still observed in him-18 mutants, this implies a DNA repair rather than a checkpoint defect. Taken together, a drastic reduction in relative hatching frequencies was observed in him-18 mutants compared to wild type following exposure to all four kinds of genotoxic agents. These results suggest that HIM-18 is required for DSB repair, ICL repair and recovery from replication fork collapse.
him-18 Is Synthetic Lethal with him-6, the C. elegans BLM Homolog Budding yeast SLX4 was first identified in a synthetic-lethal screen for genes that are essential in an sgs1 mutant background [17]. Given that HIM-18 and Slx4 share sequence homology and the analysis of DSB repair progression and DNA damage sensitivity suggest a functional conservation, we investigated whether him-18 mutants show synthetic lethality with loss of him-6, the C. elegans homolog of BLM, the Bloom syndrome helicase gene [33]. The total number of eggs laid either by him-18 or him-6 single mutants were only moderately reduced compared to wild type (66% and 79% of wild type levels were observed, respectively), and among those eggs laid, 20% and 40%, respectively, hatched (Table 1). In contrast, him-18;him-6 double mutants showed a drastic reduction in brood size (only 3% of wild type) and all the eggs laid failed to hatch (Table 1). To further investigate whether the increased embryonic lethality observed in the him-18;him-6 double mutants correlated with defects in DSB repair, we quantified the levels of RAD-51 foci in their germlines ( Figure 3A and Figure S5). Indeed, both mitotic and meiotic RAD-51 foci levels were drastically increased in him-18;him-6 double mutants compared to either single mutant. On average, a 2.7-and 19-fold increase in RAD-51 foci/nucleus was observed in mitotic nuclei at zone 1 in him-18;him-6 double mutants compared with him-18 and him-6 single mutants, respectively (P,0.0001), and a 2.9-and 2.3-fold increase in meiotic mid-pachytene nuclei at zone 5 (P = 0.0001) ( Figure S6). Furthermore, germ cell apoptosis was elevated nearly two-fold in him-18;him-6 mutants compared to either single mutant (P,0.0001, respectively; Figure 4A and 4B). We also detected thin DAPI-stained threads (hereafter referred to as chromatin bridges), frequently stained with RAD-51, connecting nuclei at the premeiotic tip in him-18;him-6 mutants. Specifically, 9 out of 11 gonads contained pairs of nuclei connected by chromatin bridges. Chromatin bridges were observed in 8.6% (n = 19/221) of nuclei in zone 1. Moreover, between 1 and 25 RAD-51 foci were observed on 95% of these chromatin bridges ( Figure S11). These data suggest that accumulation of unresolved toxic recombination intermediates results in synthetic lethality in him-18;him-6 double mutants.

HIM-18 and MUS-81 Have Additive Roles in DNA Repair
Mus81 has been shown to be required for meiosis in fission yeast [8,34,35] and has been implicated in HJ processing in eukaryotic cells [8,9]. To determine whether it plays a role with HIM-18 in DNA repair in the C. elegans germline, we examined phenotypes suggestive of errors in chromosome segregation and the levels of RAD-51 foci in mus-81 and mus-81;him-18 double mutants (Table 1, Figure 3A, and Figures S6, S7, S8). The total number of eggs laid by mus-81 mutants was reduced compared to wild type (47% of wild type levels), and 20.7% of the eggs laid failed to hatch. This embryonic lethality was not accompanied by an increase in the frequency of males among the surviving progeny, suggesting that MUS-81 is not required for the proper disjunction of the X chromosome at meiosis I and in agreement with [36]. In mus-81;him-18 mutants embryonic lethality, but not the incidence of male progeny, was elevated compared to him-18 single mutants (87.8% of the eggs laid failed to hatch; 5.6% males; P,0.0001 and P = 0.2272, respectively, by the Fisher's Exact test). Levels of RAD-51 foci in mus-81 mutants were only increased in nuclei in the premeiotic region compared to wild type (levels of RAD-51 foci were similar during pachytene). However, levels of RAD-51 foci were increased in both premeiotic and pachytene nuclei in mus-81;him-18 double mutants compared to either single mutant ( Figure 3A, Figures S5, S6, S7, S8). Therefore, while MUS-81 may not play a critical role during meiosis in C. elegans, most of the him-18 phenotypes are aggravated by mus-81. These results suggest that MUS-81 may have additive roles with HIM-18 during repair both in mitosis and meiosis.

HIM-18 Interacts with SLX-1 and XPF-1
Slx4 interacts with Slx1, Rad1/XPF, Rtt107/Esc4 and Cdc27 in S. cerevisiae [17,[37][38][39]. In D. melanogaster, MUS312 interacts with MEI-9/XPF [16]. To determine whether HIM-18 interacts with SLX-1, XPF-1 or ERCC-1, which forms a heterodimer with XPF-1 in C. elegans [40], we tested the full length and various regions of HIM-18 for interactions with these proteins using the yeast twohybrid system ( Figure 5). We divided HIM-18 into three parts, namely HIM-18N (amino acids 1 to 166), HIM-18M (amino acids 165 to 437), and HIM-18C (amino acids 420 to 718). HIM-18N contains the zinc finger domain, HIM-18M contains the coiledcoil and BTB domains, and HIM-18C contains the SAP and leucine zipper motifs. DB-HIM-18M showed self-activation precluding further analysis with this construct. Both HIM-18 full length and HIM-18C interact with SLX-1 in either orientation in the yeast two-hybrid system indicating that SLX-1 binds to the C-terminal region of HIM-18. HIM-18 full length also interacts with XPF-1 in either orientation although this interaction is weaker than those observed between HIM-18-SLX-1 or XPF-1-ERCC-1. Similar to D. melanogaster, where an interaction between MUS312 and ERCC1 was not detected [16], we also failed to observe an interaction between HIM-18 and ERCC-1 (data not shown). We obtained similar results using different combinations of yeast strains and plasmids ( Figure S12). Thus, HIM-18 physically interacts with SLX-1 and XPF-1 in C. elegans, similar to the interactions observed involving Slx4 in S. cerevisiae [17], and its orthologs in S. pombe [20] and D. melanogaster [16].
To further refine the region of HIM-18 required for the interaction with SLX-1, we specifically examined two domains contained within the C-terminus defined based on recent studies of the mammalian SLX4/BTBD12 protein [41][42][43]. Specifically, we examined the conserved C-terminal domain (CCD) [41,42] and the helix-turn-helix (HtH) region contained within the CCD [43]. We observed that HIM-18CCD, but not HIM-18HtH, binds to SLX-1 ( Figure 6). These results are in agreement with the analysis of the human SLX4/BTBD12 [41] [42].

HIM-18 and XPF-1 Are Required for Normal Levels of Meiotic Crossovers
The S. cerevisiae Slx4-Slx1 complex can cleave branched DNA substrates such as HJs in vitro and both MUS312 and MEI-9/XPF in D. melanogaster have been implicated in HJ resolution. Since both xpf-1(e1487) mutants and xpf-1(RNAi) worms showed embryonic lethality (20.2%, n = 2348, and 7.8%, n = 2890, respectively), a high incidence of males (4.5% and 3.5%, respectively) ( Table 1 and Figure S2B) and elevated levels of RAD-51 foci at late pachytene (zone 6) (3.1 and 2.160.29 foci/nucleus, respectively; P,0.0001 and P = 0.0028 by the two-tailed Mann-Whitney test; 95% C.I.) compared to wild type (0.9 foci/nucleus) and control (RNAi) (0.960.15 foci/nucleus) ( Figure 3A and Figure S5), it is possible that XPF-1 may also play a role in meiotic recombination in C. elegans. Given that meiotic COs require HJ resolution, we assessed the role of HIM-18 and XPF-1 in meiotic CO formation by comparing CO frequencies along both chromosomes I and X between wild type and mutants for him-18 and its interaction partner xpf-1 ( Figure 7A and Table S2). We were precluded from performing this analysis for slx-1 mutants due to the lack of an available strong loss-of-function mutant (T. Saito and M. Colaiácovo, unpublished results). On chromosome I, a 55.6cM interval corresponding to 96% of this chromosome's whole length (interval A to E) was assayed using 5 snip-SNPs. The CO frequency in this interval was reduced to 69.8% (P = 0.0005) and 79% (P = 0.0302) of wild type, respectively in him-18 and xpf-1 mutants. On chromosome X, a 44cM interval corresponding to 76% of this chromosome's whole length (interval A to E) was assayed using 5 snip-SNPs. The crossover frequency observed in this interval was reduced to 50.5% (P,0.0001) and 76.1% (P = 0.0434) of wild type, respectively in him-18 and xpf-1 mutants. As in wild type, double COs were not detected in either mutant for either chromosome, indicating that CO interference was not impaired. Finally, in wild type C. elegans, CO distribution is biased towards the terminal thirds of autosomes and is more evenly distributed along the X chromosome [44]. Our analysis suggests that these distribution patterns are not altered among the remaining COs observed in him-18 and xpf-1 mutants (the reduction in COs observed for interval D-E on the right end of the X chromosome in him-18 mutants was not significant compared to wild type; P = 0.0553). Taken together, these data suggest that HIM-18 and XPF-1 do not play a role in CO positioning, but are required for normal levels of CO formation in the autosomes and the X chromosome.

HIM-18 Acts Following Holliday Junction Formation during Meiosis
The fact that the frequency but not the position of meiotic crossover events is affected in him-18 and xpf-1 mutants suggests that HIM-18 and XPF-1 may be required very late in the process of crossover formation. To further assess this, we examined the genetic interaction between him-18 and msh-5 ( Figure 3B and 3C and Table 1). MSH-5 and HIM-14/MSH-4 function downstream of the chromosomal association of the RAD-51 strand-exchange protein, but upstream of dHJ resolution during meiosis [30]. We observed approximately 12 DAPI-stained bodies in diakinesis oocytes in both msh-5 and him-18;msh-5 mutants, compared to the nearly 6 DAPI-stained bodies observed in both wild type and him-18 mutants. These results suggest that HIM-18 may act downstream of MSH-5, after dHJ formation. However, until further studies address the nature of the bivalent connections observed in him-18 mutants, we cannot exclude the possibility that HIM-18 may also act on a completely different pathway from MSH-5.
Given that CO recombination results in chiasmata that persist until the metaphase I to anaphase I transition during meiosis, it was intriguing that despite the reduction in meiotic CO  Asymmetric disassembly of the SC, exemplified by SYP-1 in this diagram, starts in late pachytene and progresses through early diplotene where SYP-1 is still observed in regions where homologs remain coaligned, but is no longer present in homologue segments that are disassociating. The region where SYP-1 is still present corresponds to the short arm of the bivalent at early diakinesis, which is later occupied by AIR-2, whereas LAB-1 localizes on the long arm. The short and the long arms are defined by the off-center position of the single crossover event that occurs between each pair of homologs. This late prophase chromosome remodeling process around the off-center crossover reveals the chiasma. To better understand the conformational change undergone by the bivalents from diplotene to diakinesis, the end of each chromosome is labeled a, b, c, and d, respectively. Bivalents align at the metaphase I plate (the long arms are positioned perpendicular to the metaphase plate) and this involves microtubules and the motor activity of chromokinesin at the short arm (not shown). At anaphase I, AIR-2 is proposed to phosphorylate the cohesin REC-8 localized at the short arm (REC-8 localized at the long arm is protected by LAB-1) resulting in loss of sister chromatid cohesion at the short arm. The recombined homologs then segregate away from each other towards opposite poles of the meiosis I spindle (bold arrows). REC-8 (yellow bars), SYP-1 (black bars), LAB-1 (red circles), AIR-2 (green circles), and microtubules (green lines). doi:10.1371/journal.pgen.1000735.g007 recombination frequencies observed in him-18 mutants, we mostly observed 6 pairs of attached bivalents in him-18 diakinesis oocytes ( Figure 3B and 3C). However, careful examination revealed that the morphology of the DAPI-stained bivalents observed in him-18 mutants was distinct from wild type in diakinesis and prometaphase I, although this was more clearly apparent in the latter ( Figure 7B and 7C). Specifically, homologs were loosely attached and 84% (n = 16/19) of the oocytes examined had at least one bivalent held together by a thin DAPI-stained thread at prometaphase I. The ''fragility'' of the connections between homologs is further highlighted by the localization of the LAB-1 and AIR-2 proteins on bivalents at prometaphase I. LAB-1 is a protein recently implicated in the protection of sister chromatid cohesion and is restricted to the longer axes of the bivalents from late prophase I through the metaphase I to anaphase I transition [45]. AIR-2 is the C. elegans Aurora B kinase and is restricted to the mid-bivalent from late diakinesis through metaphase I. However, in him-18 mutants the mid-bivalent is observed separating prematurely at prometaphase I as indicated by the partially separated (V-shaped) AIR-2 ring-like signals in 68% of the oocytes examined (n = 13/19; P,0.0001 by the two-sided Fisher's Exact test, 95% C.I.) compared to 8.3% (n = 2/24) in wild type ( Figure 7B). Taken together, these fragile attachments between homologs suggest that HIM-18 is required for chiasma formation and support the increased chromosome nondisjunction observed in him-18 mutants.

The Formation of Mature Bivalents Is Delayed in him-18 Mutants
Crossovers or crossover precursors trigger chromosome remodeling in late meiotic prophase resulting in mature bivalent formation [46]. Therefore, to investigate whether HIM-18 affects the timing of mature bivalent formation, we assessed the kinetics of events correlated with chromosome remodeling in late prophase such as SC disassembly, the chromosomal dissociation of a CO recombination site marker and Histone H3 serine 10 phosphorylation ( Figure 8). During SC disassembly, which initiates in late pachytene in wild type germlines, SC central region components that were previously localized throughout the full length of chromosomes become progressively restricted to the mid-bivalent by early diakinesis and are mostly gone from chromosomes by late diakinesis (only 10.9% of -2 oocytes carried 3 or more bivalents with residual SYP-1 signal, n = 55) ( Figure 8A and 8G and [46]). In contrast, although early meiotic events proceed with normal kinetics in him-18 mutants, SC disassembly is delayed in this background, as evidenced by the higher levels of -2 oocytes with chromosome-associated SYP-1 (28.8%, n = 52; P,0.0345) ( Figure 8A and 8B). Recently, ZHP-3, a homolog of the budding yeast Zip3 protein, was suggested to mark CO recombination sites starting in late pachytene in C. elegans [47]. Interestingly, despite the reduction in CO frequencies observed in him-18 mutants, we observed approximately six ZHP-3::GFP foci/nucleus in oocytes at early diakinesis (-5 oocytes) in both wild type (n = 25) and him-18 mutants (n = 18) ( Figure 8C and 8D). This suggests that ZHP-3 may mark a CO precursor instead of the mature CO during late pachytene through diakinesis. Moreover, the timing of dissociation of ZHP-3::GFP from chromosomes was delayed in a similar fashion to that of SC disassembly in him-18 mutants ( Figure 8C and 8D). While in wild type, between 5.3 to 3.7 ZHP-3::GFP foci/ nucleus were observed until mid-diakinesis (-4 and -3 oocytes, n = 29 and 32, respectively) and were mostly no longer detected by late diakinesis (0.2 foci/-2 oocyte, n = 32), in him-18 mutants, in average 2.6 ZHP-3::GFP foci (n = 26) were still present in the -2 oocytes (Figure 8C and 8D). Finally, quantitative analysis of Histone H3 phosphorylation (pH 3), a chromosomal substrate of AIR-2 kinase [45,48,49], indicated that the appearance of nuclei with pH 3 positive chromosomes is delayed in him-18 mutants compared to wild type in late diakinesis (30% of -3 oocytes in him-18 mutants carried pH 3 positive chromosomes compared to 72% in wild type) ( Figure 8E and 8F). Interestingly, CO-defective spo-11 mutants in which, similar to him-18 mutants, events occurring upon entrance into meiosis such as chromosome pairing and synapsis are normal [50], showed the most delay in Histone H3 phosphorylation, further implicating mature CO formation as a requirement for proper timing of Histone H3 phosphorylation. Taken together, these data suggest that chromosome remodeling during late prophase is delayed in him-18 mutants due to impaired CO formation.

HIM-18 Is a New Late HR Intermediate Processing Factor Conserved Among Eukaryotes
Several groups reported that the Slx1-Slx4 complex cleaves HJs in vitro, although this cleaving activity is weak and inconsistent with an authentic HJ resolvase activity [7,19,20]. Although we do not know yet whether C. elegans SLX-1-HIM-18 has authentic HJ cleaving activity in vitro, our genetic and cytological data suggest that some HR intermediates are processed in a HIM-18dependent manner. We did not identify any known nuclease motifs in HIM-18. However, we showed that two nucleases, SLX-1 and XPF-1, interact with HIM-18. SLX-1 is conserved from yeast to humans and contains an URI nuclease domain and a PHD finger domain. Although it has been reported that XPF-ERCC1 mainly cuts simple-Y, bubble, stem loop and 39-flap structures in S. cerevisiae and H. sapiens [51], whether the substrate specificity of XPF-1 is altered from those reported DNA structures to HJs due to the interaction with HIM-18 during HR is unknown. Notably, the SAP motif of S. pombe Cce1, a mitochondrial HJ resolvase, is required for stable binding to HJs. An attractive hypothesis that builds on this observation is that HIM-18 may bind to the HJs via its SAP motif and promote the nuclease activity of SLX-1 and XPF-1. In this vein, HIM-18 could serve as a scaffold accommodating different interaction partners (nucleases), thereby facilitating the resolution of HJ intermediates arising in different biological contexts as depicted in our model ( Figure 9). Specifically, we propose that HIM-18 function is required for replication restart after DNA damage and correct CO formation during meiosis to maintain genomic integrity in the germline.

HIM-18 Function in Replication Restart
We showed that HIM-18 is required for the repair of DNA damage arising during DNA replication in the germline. When a replication fork collides either with a spontaneous or an artificial barrier, single-strand gaps (SSG) can be generated at either the lagging or leading strands ( Figure 9A). To complete an error-free DNA replication, SSGs must be repaired by HR, and during mitosis, this involves the use of a sister chromatid, instead of a homologous chromosome, as a template for repair. We propose that HIM-18 is required for processing late HR intermediates after the RAD-51-mediated strand exchange and pairing ( Figure 9A-c and 9A-d). It is conceivable that MUS-81 also functions in this step as it can cleave substrates mimicking dHJs in biochemical assays [52] and has been implicated in the repair of spontaneous DNA damage [36] and DNA ICLs [53]. Therefore, it will be interesting to test if HIM-18 and MUS-81 have overlapping functions in HJ processing during replication collapse. In BLM-deficient cells, sister chromatid exchange (SCE) is increased [54]. As BLM has been implicated in the dissolution of HJs to yield NCOs in multiple organisms [55,56], the increased levels of chromatin bridges with numerous RAD-51 foci observed in him-18; him-6 double mutants may represent an accumulation of repair intermediates during DNA replication stress, which are not processed either via HIM-18 or dissolution by HIM-6. While dHJ unwinding activity has not been reported for HIM-6 in C. elegans, our data is consistent with distinct roles for HIM-6 and HIM-18 in processing dHJ intermediates ( Figure 9A-e'). Apparently, him-18 and xpf-1 mutants are distinct from each other regarding repair of spontaneous DNA damage during mitosis. In contrast to him-18 mutants, xpf-1 mutants do not show an increase in the levels of RAD-51 foci in the premeiotic region of the germline and the RAD-51 staining pattern observed in xpf-1;him-18 double mutants is very similar to that in him-18 single mutants. XPF-1 is required for unwinding or repairing G4 DNA (G-quadruplex) structures during DNA replication in mutants of the C. elegans homolog of the FANCJ DNA helicase, dog-1 [57,58]. In contrast, we did not detect elevated levels of deletions on polyG/C-tracts in dog-1; him-18 double mutants ( Figure S13). Therefore, HIM-18 does not play a role in the unwinding of G4 DNA during replication. Likewise, whereas XPF-1 is required for NER [59], fly MUS312 and yeast Slx4, the HIM-18 orthologs, do not function in NER [16,20]. Taken together, these observations suggest that HIM-18 has a distinct function from that of XPF-1 during mitotic proliferation.

HIM-18 Function in Meiotic CO Formation
COs play a critical role in ensuring accurate meiotic chromosome segregation, as exemplified by the alterations in CO number and/or distribution frequently associated with human aneuploidies [60]. Ensuring the formation of at least one CO per homolog pair (obligate CO) is vital to the transmission of an intact genome during gametogenesis. In wild type C. elegans, one DSB per pair of homologous chromosomes engages in a CO pathway and is marked by ZHP-3. In him-18 mutants, CO frequencies are decreased and offloading of ZHP-3 from nascent CO sites is delayed. This delay could be simply explained by dHJs being dissolved instead of resolved in him-18 mutants and the dissolution process perhaps being time consuming ( Figure 9B). Occasionally, however, dissolution is not completed in him-18 mutants, because fragile connections (possibly hemicatenanes) are still detectable between homologs. This fragile connection persists through diakinesis into prometaphase I possibly leading to nondisjunction (NDJ) at anaphase I. This is further supported by our observation of chromosome bridges and lagging chromosomes at the metaphase I to anaphase I transition in him-18 mutants (n = 9/15 oocytes), in contrast to wild type where these were never detected (n = 0/17) ( Figure S14). However, the presence of oocytes lacking either chromosome bridges or lagging chromosomes in him-18 mutants suggests that at least a portion of the hemicatenanes may be finally dissolved by TOP-3 just prior to anaphase I. Taken together, our data are consistent with a role for HIM-18 in processing dHJs leading to CO formation during meiosis ( Figure 9B).
him-18;him-6 double mutants are synthetic lethal. Interestingly, with regard to the numbers of DAPI-stained bodies in oocytes at diakinesis, him-18 suppresses the him-6 mutation (P,0.0001, by the two tailed Mann-Whitney test; 95% C.I.). To explain this meiotic phenotype we propose that HIM-6 may play a role in stabilizing the meiotic D-loop and dHJ intermediates, perhaps via its helicase activity, and specifically promote a CO outcome. In him-18;him-6 mutants, there is no HIM-6-dependent stabilization of D-loops. Some unstable D-loops may be processed via the SDSA pathway. The remaining D-loops may become unstable dHJs, which are not cleaved by HIM-18 and persist during diakinesis. Therefore, the number of DAPI-stained bodies in him-18;him-6 mutants (6.39) is lower than that observed in him-6 mutants (7.32) and higher than in him-18 mutants (6.04).
A requirement for XPF in meiotic CO formation had only been observed, thus far, in D. melanogaster, where MEI-9 is essential for meiotic COs. Our genetic and cytological analyses suggest that XPF-1 seems to function in CO formation during C. elegans meiosis. Interaction of HIM-18 with XPF-1 may be important for altering the substrate-specificity of XPF-1 towards HJs during meiosis. We observed that the decrease in CO frequency on chromosome X in xpf-1 mutants is milder than that in him-18 mutants (P = 0.0120). These data suggest that HIM-18 may define yet another HJ processing activity distinct from XPF-1. Mus81, which is also known as a meiotic HJ resolvase, is specialized in interference-independent COs in yeast, plants and mammals [61]. However, in C. elegans, virtually all COs are HIM-14/MSH-4 and MSH-5-dependent, and therefore interference-dependent [62,63]. Additionally, mus-81 mutants are largely viable [36] suggesting that MUS-81 may not be required for meiotic COs in C. elegans. Further experiments will address whether SLX-1, which we demonstrated is a HIM-18 interaction partner, functions during meiotic CO formation.

HIM-18 Expression May Undergo Post-Transcriptional Regulation
HIM-18 is expressed in the C. elegans germline, being enriched at the premeiotic tip and in late meiotic prophase (from late pachytene to diakinesis). HIM-18 protein levels, as detected by immunostaining, are low between the transition zone and the midpachytene stage. In contrast, analysis of mRNA expression by in situ hybridization (Y. Kohara, personal communication) suggests a more uniform pattern of expression from transition zone until diakinesis. These data suggest tight regulation of HIM-18 at the protein level, either by translational repression and/or protein degradation between transition zone through the mid-pachytene stage. The translation of a number of mRNAs in the C. elegans germline is repressed by GLD-1, a member of the STAR KHdomain family of RNA binding proteins [64]. However, sequence analysis revealed that HIM-18 harbors a destruction box (D-box; RXXL), an APC/C recognition motif and an Ubc9 recognition motif (yKXE). These motifs are usually required for the ubiquitinand SUMO-dependent proteolytic pathways. In S. cerevisiae Slx4 interacts with the APC/C component Cdc27, although it remains to be determined whether Slx4 is degraded by APC/C. Further analysis will reveal whether HIM-18 may be a target for a proteolytic pathway during transition zone to mid-pachytene.
In summary, we identified HIM-18 as an ortholog of yeast Slx4, fly MUS312 and mammalian BTBD12 proteins, which plays a critical role in the germline during DSB repair upon replication fork collapse in mitosis and SPO-11-dependent programmed meiotic DSB formation. Our results therefore identified HIM-18 as a new HJ processing factor in C. elegans, which is distinct from the Mus81-Eme1 complex.

Analysis of HIM-18 Protein Conservation and Motifs
HIM-18 homology searches were performed using the Ensembl genome browser (http://www.ensembl.org/index.html) and Pfam (http://pfam.sanger.ac.uk/). Although the HIM-18 ortholog in S. cerevisiae was not identified by the Ensembl program, Pfam predicted that the SAP motif of HIM-18 is similar to that in yeast Slx4. The following motif prediction programs were applied to HIM-18 and its orthologs: COIL and P-SORT II for coiled-coil and leucine zipper predictions, Pfam and HHpred for zinc finger and BTB domain predictions [69][70][71][72].

Antibody Preparation, DAPI Analysis, Immunostaining, and FISH
Rabbit anti-HIM-18 antibody was produced using a HIStagged fusion protein expressed from plasmid pDEST17 (Invitrogen) containing coding sequence corresponding to the first 166 amino acids of HIM-18. 6xHis-HIM-18N was expressed in BL21 E. coli cells and purified with the Ni-NTA Purification System (Invitrogen). Animals were immunized and bled by Sigma-Genosys, The Woodlands, TX. The antisera were affinity-purified against the 6xHis-HIM-18N peptide as described in [73].
FISH was performed as in [74] utilizing a probe to the 5S rDNA locus on chromosome V prepared as in [50].

DNA Damage Sensitivity Experiments
Young adult him-18/him-18 animals were picked from the progeny of him-18/qC1 parent animals. To assess ionizing radiation (IR) sensitivity, animals were treated with 0, 10, 50 or 100 Gy of IR from a Cs 137 source at a dose rate of 2.16 Gy/min. For nitrogen mustard (HN2) sensitivity, animals were treated with 0, 50, 100 or 150 mM of HN2 (mechlorethamine hydrochloride; Sigma) in M9 buffer containing E. coli OP50 with slow shaking in the dark for 19 hours. Treatment with camptothecin (CPT; Sigma) was similar, but with doses of 0, 100 or 500 nM. Following treatment with HN2 or CPT, animals were washed twice with M9 containing TritonX100 (100 ml/L) and plated to allow recovery for 3 hours. For hydroxyurea (HU) sensitivity, animals were placed on seeded MYOB plates containing 0 or 40 mM HU for 24 hours. Apoptotic cells surrounded by CED-1::GFP signal were observed in the late pachytene region of the germline with a Leica DM5000 B fluorescence microscope. Between 21 and 95 gonads were scored for each genotype. Statistical comparisons between genotypes were performed using the two-tailed Mann-Whitney test, 95% C.I.