During meiosis, accurate chromosome segregation relies on homology to mediate chromosome pairing, synapsis, and crossover recombination. Crossovers are dependent upon formation and repair of double-strand breaks (DSBs) by homologous recombination (HR). In males of many species, sex chromosomes are largely hemizygous, yet DSBs are induced along nonhomologous regions. Here we analyzed the genetic requirements for meiotic DSB repair on the completely hemizygous X chromosome of Caenorhabditis elegans males. Our data reveal that the kinetics of DSB formation, chromosome pairing, and synapsis are tightly linked in the male germ line. Moreover, DSB induction on the X is concomitant with a brief period of pseudosynapsis that may allow X sister chromatids to masquerade as homologs. Consistent with this, neither meiotic kleisins nor the SMC-5/6 complex are essential for DSB repair on the X. Furthermore, early processing of X DSBs is dependent on the CtIP/Sae2 homolog COM-1, suggesting that as with paired chromosomes, HR is the preferred pathway. In contrast, the X chromosome is refractory to feedback mechanisms that ensure crossover formation on autosomes. Surprisingly, neither RAD-54 nor BRC-2 are essential for DSB repair on the X, suggesting that unlike autosomes, the X is competent for repair in the absence of HR. When both RAD-54 and the structure-specific nuclease XPF-1 are abrogated, X DSBs persist, suggesting that single-strand annealing is engaged in the absence of HR. Our findings indicate that alteration in sister chromatid interactions and flexibility in DSB repair pathway choice accommodate hemizygosity on sex chromosomes.
MEIOSIS is essential for the formation of haploid gametes for sexual reproduction. Accurate chromosome segregation during meiosis relies on homology between maternal and paternal homologous chromosomes to drive pairing, synapsis, and crossover (CO) recombination. The presence of differentiated sex chromosomes in the heterogametic sex requires modification of these homology-dependent processes. This is significant as the largely nonhomologous X and Y chromosomes in human males display a higher rate of nondisjunction than autosomes, resulting in infertility and developmental disorders such as Klinefelter and Turner syndromes (Shi et al. 2001; Lange et al. 2009). It is currently unknown how meiosis is altered to accommodate hemizygous sex chromosomes.
Of critical importance to meiotic chromosome segregation is the formation of double-strand breaks (DSBs) catalyzed by the conserved topoisomerase Spo11 and repair by homologous recombination (HR) to generate crossovers (Keeney et al. 1997; Dernburg et al. 1998). Spo11-induced DSBs are processed by CtIP/Com1/Sae2, the MRN (Mrell-Rad50-Nbs1) complex and multiple nucleases to facilitate repair by HR (Terasawa et al. 2008; Manfrini et al. 2010; Zakharyevich et al. 2010; Garcia et al. 2011). Recent studies in Caenorhabditis elegans have revealed that COM-1 and MRE-11 promote HR by inhibiting direct religation of DSBs by nonhomologous end joining (NHEJ) (Lemmens et al. 2013; Yin and Smolikove 2013). The importance of inhibiting error-prone pathways during meiosis is also exemplified by studies in Drosophila, where multiple barriers to NHEJ have been uncovered by mutational analyses (Joyce et al. 2012). In mouse oocytes, an error-prone single-strand annealing (SSA) pathway can process extra-chromosomal DNA. SSA requires only small stretches of homology within a single DNA duplex and results in deletions (Fiorenza et al. 2001); however, the prevalence of this pathway in meiosis has not been investigated.
In addition to a bias toward repair through HR, the choice of repair template is also tightly regulated during meiosis. Unlike somatic cells where the sister chromatid is the preferred template, meiotic DSBs are preferentially repaired using the homolog. This homolog bias is imposed through regulation of the meiotic recombination machinery as well as meiosis-specific chromosomal axis components and cohesins (Couteau et al. 2004; Niu et al. 2005; Niu et al. 2009; Kim et al. 2010; Ho and Burgess 2011; Hong et al. 2013; Shin et al. 2013; Yan and McKee 2013). Thus, meiotic prophase events have evolved to drive homologous chromosome interactions to ensure formation of crossovers for proper segregation at the meiosis I division. Nonetheless, studies in Saccharomyces cerevisiae and C. elegans have revealed that the use of the sister chromatid as a template for DSB repair in meiosis occurs and is important for maintaining genome stability (Bickel et al. 2010; Goldfarb and Lichten 2010).
Sex chromosomes of the heterogametic sex are largely nonhomologous and therefore require adaptation of these homology-dependent meiotic processes. Interestingly, even the homologous pseudoautosomal region(s) (PARs) of mammalian sex chromosomes require modification of both chromosome structure and the meiotic recombination machinery to promote crossovers (Kauppi et al. 2011). Regardless of the extent of homology, meiotic DSBs are induced along the length of sex chromosomes in male C. elegans, mice, and humans (Ashley et al. 1995; Moens et al. 1997; Sciurano et al. 2006; Jaramillo-Lambert and Engebrecht 2010; Checchi and Engebrecht 2011). The hemizygous regions of sex chromosomes subsequently undergo meiotic sex chromosome inactivation (MSCI), which is characterized by elaboration of a specialized heterochromatin domain and transcriptional silencing (Turner 2007). One proposed function of MSCI is to prevent recombination between nonhomologous regions of sex chromosomes (McKee and Handel 1993). While there is no direct evidence in support of this hypothesis, it is likely that the repressive chromatin architecture elaborated during MSCI influences how induced DSBs are repaired, as repair is affected by chromatin environment (Van Attikum and Gasser 2009).
To determine how meiotic DSBs are repaired on hemizygous regions of sex chromosomes, we analyzed repair of SPO-11-induced DSBs on the lone X chromosome of C. elegans males. In males, the X chromosome completely lacks a homologous partner, yet meiotic DSBs are induced and repaired efficiently (Jaramillo-Lambert and Engebrecht 2010; Checchi and Engebrecht 2011). Additionally, as with mammals, the X chromosome of C. elegans males accumulates repressive chromatin marks and is transcriptionally silenced during meiotic prophase (Kelly et al. 2002; Bean et al. 2004; Checchi and Engebrecht 2011). Taking advantage of the spatiotemporal organization of the C. elegans germ line, high-resolution microscopy, and available mutants, we analyzed the requirement of DSB repair pathways, singly and in combination, as well as chromosomal structural components on repair of meiotic DSBs on the single X chromosome of males. We find that there is a brief period of X chromosome pseudosynapsis (i.e., apparent synapsis that does not necessarily constitute full alignment or contain all components of the synaptonemal complex) that we propose allows X sister chromatids to behave as if they are homologs. Consistent with this, neither meiotic kleisins nor the SMC-5/6 complex, implicated in intersister repair, are required for DSB repair on the X chromosome. Further, unlike on autosomes, in the absence of HR an XPF-1-dependent pathway can repair DSBs on the X, suggesting that error-prone SSA can be engaged to repair DSBs on hemizygous sex chromosomes. We propose that both alteration in sister chromatid interactions and flexibility in repair pathway choice on sex chromosomes accommodate hemizygosity during meiosis.
Materials and Methods
Maintenance and genetic analyses of worms were performed using standard procedures (Brenner 1974). C. elegans var. Bristol (N2) was used as the wild-type strain. The following mutations were used in this study: LGI, atm-1(gk186), rad-54 (ok615); LGII, smc-5(ok2421), smc-6(ok3294), xpf-1(ok3039); LGIII, met-2(n4256), unc-32(e189)com-1(t1489), cku-80(ok861), lig-4(ok716), brc-2(tm1086); LGIV, fem-3(e1996), him-8(me4), him-3(e1147), zim-2(tm574), rec-8(ok978); LGV, coh-3(gk112)coh-4(tm1857), fog-2(q71); and LGX, lon-2(e678). COSA-1::GFP worms were obtained from Anne Villeneuve (Yokoo et al. 2012); the LGV lacO integrant was obtained from Paul Sternberg (Gonzalez-Serricchio and Sternberg 2006); the LacO on X strains was derived from lines from Christian Frokjaer-Jensen and Erik Jorgensen; strains harboring the transgene RPA-1::YFP (opls263) were obtained from Michael Hengartner (Stergiou et al. 2011) and RPA-1::GFP from Julian Blow (Sonneville et al. 2012).
fem-3(e1996) X0 females were generated as described in (Jaramillo-Lambert and Engebrecht 2010). Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health National Center for Research Resources (NIH NCRR). All strains were propagated at 20°.
Purification of LacI-His6-GFP
The plasmid used for expressing LacI-His6-GFP was a generous gift from Anna Hines (Darby and Hine 2005). Following induction with 0.5 mM IPTG, E. coli were lysed and protein purified using nickel beads (Ni-NTA agarose; Qiagen, Valencia, CA).
Immunostaining of germ lines was performed as described (Jaramillo-Lambert et al. 2007). The following primary antibodies were purchased and used at the indicated dilutions: mouse anti-histone H3 dimethyl K4 (1:250) (Millipore, Temecula, CA); rabbit anti-COH-3 (1:10,000), rabbit anti-HIM-8 (1:10,000), rabbit anti-RAD-51 (1:10,000), and rabbit anti-GFP (1:500) (Novus Biologicals, Littleton, CO). Guinea pig anti-HTP-3 (1:500), guinea pig anti-HIM-8 (1:500), and rabbit anti-REC-8 (1:1000) were generous gifts from Abby Dernburg and Josef Loidl, respectively. Rabbit anti-SYP-1 (1:200) and rat anti-RAD-51 (1:100) were generously gifted by Anne Villeneuve. The following secondary antibodies from Life Technologies were all used at 1:500 dilutions: Alexa Fluor 555 goat antirat IgG, Alexa Fluor 488 goat antirat IgG, Alexa Fluor 546 goat antimouse IgG, Alexa Fluor 488 goat antimouse IgG, Alexa Fluor 555 goat antirabbit IgG, Alexa Fluor 488 goat antirabbit IgG, and Alexa Fluor 488 goat anti-guinea pig IgG. Alexa Fluor 647 donkey anti-guinea pig, Alexa Fluor 647 donkey antimouse IgG, and Alexa Fluor 647 donkey antirabbit IgG were used at 1:200 dilutions. DAPI (2 µg/ml; Sigma) was used to counterstain DNA.
LacI-His6-GFP was added directly to dissected gonads following standard fixation at a concentration of 5–25 ng/μl and incubated overnight at 4°. RPA-1::GFP and RPA-1::YFP fluorescence was directly visualized by dissecting gonads and fixing in 2.5% paraformaldehyde for 1 min. Slides were subsequently incubated on dry ice for at least 5 min followed by freeze crack and immersion in 95% ethanol for 1 min.
Collection of images was performed using an API Delta Vision deconvolution microscope. Images were deconvolved using Applied Precision SoftWoRx image analysis software and subsequently processed and analyzed using Fiji (ImageJA) (Wayne Rasband, NIH). All images shown are projections through data stacks.
Structured illumination microscopy (SIM) analysis was performed using a Nikon N-SIM super-resolution microscope and NIS-Elements 2 image processing software. Images were further processed using ImageJ.
Fluorescent in situ hybridization
Fluorescent in situ hybridization (FISH) was performed as described in Phillips and Dernburg (2006). A fluorescent Cy5-tagged lacO oligonucleotide (5′/5Cy5/CCACATGTGGAATTGTGAGCGGATAACAATTTGTGG-3′) was generated and used to label LacO insertions. Nuclei scored as containing one spot had either a single focus or two adjacent foci whose signals overlapped; nuclei scored as containing two spots had clearly separated signals. Significance was analyzed using Fisher’s exact test.
RNA-mediated interference analysis
RNA-mediated interference (RNAi) experiments were performed at 20°, using the feeding method (Timmons et al. 2001). L4 larvae were fed RNAi-inducing HT115(DE3) bacteria strains or the same bacteria transformed with the empty feeding vector, L4440. atm-1 was obtained from a genomic RNAi feeding library (Kamath and Ahringer 2003) and smc-6 was cloned from N2 genomic DNA into L4440. Cultures were plated onto NGM plates containing 25 μg/ml carbenicillin and 1 mM IPTG and were used within 2 weeks of plating.
Analysis of fragmented chromosomes and progeny inviability
Number of DAPI-stained bodies present in the −1 to −3 diakinesis nuclei of fem-3(e1996) X0 and XX worms with indicated RNAi treatment were scored by standard fluorescence microscopy; a minimum of 70 nuclei was examined for each genotype. Significance was analyzed using Fisher’s exact test. Percentages of male and inviable progeny were determined from a minimum of eight independent crosses of indicated males to fog-2(q71) females. No significant differences were observed using the Student’s t-test.
Kinetics of chromosome pairing and synapsis are tightly linked with RAD-51 assembly in the male germ line
Meiotic DSBs catalyzed by the conserved topoisomerase Spo11 are essential for crossover formation and successful meiosis. Although Spo11-dependent DSBs are a prerequisite for chromosome pairing and synapsis in many organisms such as yeast and mammals (Giroux et al. 1989; Romanienko and Camerini-Otero 2000), these events can be uncoupled by mutational analysis in other species including Drosophila and C. elegans, as spo-11 mutants are competent for chromosome pairing and synaptonemal complex (SC) assembly (Dernburg et al. 1998; Liu et al. 2002). In C. elegans, kinetics of meiotic DSB repair are monitored by the appearance and disappearance of immunostained RAD-51 recombinase within the spatiotemporal gradient of the germ line (Colaiacovo et al. 2003) (Figure 1A and Supporting Information, Figure S1). While in C. elegans hermaphrodites, DSBs peak in abundance in nuclei with fully assembled SCs in early (EP) to mid pachytene (MP) (Colaiacovo et al. 2003), RAD-51 foci are more abundant in transition zone (TZ; leptotene/zygotene) in the male germ line (Jaramillo-Lambert and Engebrecht 2010). Interestingly, the male germ line contains more TZ nuclei than in hermaphrodites [53.6 ± 2.0 TZ nuclei/bisected gonad (n = 11) vs. 36.1 ± 1.6 TZ nuclei/bisected gonad (n = 9); P < 0.001], suggesting that the early peak of RAD-51 foci may reflect alteration of the kinetics of chromosome pairing in the male germ line. To investigate the spatiotemporal relationship between chromosome pairing and the appearance of meiotic DSBs in the C. elegans male germ line, we monitored RAD-51 assembly in germ lines containing a LacO insertion on chromosome V (LacO-V), which allowed us to track homolog interactions throughout all stages of meiotic progression using purified LacI-His6-GFP (Figure 1).
The C. elegans male germ line has five pairs of autosomes and one sex chromosome, the X. Although the X lacks a homolog and cannot form a chiasma, all chromosomes accumulate SPO-11-dependent RAD-51 foci in the TZ (Jaramillo-Lambert and Engebrecht 2010). In male germ lines, chromosome Vs were paired by early-to-mid TZ, coincident with the timing of RAD-51 assembly (Figure 1, B and C). Analysis of RAD-51 foci throughout the TZ revealed that only 7% of nuclei contained RAD-51 prior to chromosome V pairing (1 focus/nucleus), while in TZ nuclei with paired chromosome Vs, 86% contained one or more RAD-51 foci [average (av) = 4.93 foci/nucleus], indicating a strong bias toward RAD-51 assembly on paired homologs (Figure 1D).
In hermaphrodites, SC assembly takes place primarily in late TZ/EP nuclei (Colaiacovo et al. 2003). To assess the kinetics of SC assembly relative to homolog pairing in the male germ line, we next monitored the loading of the central element component SYP-1 (MacQueen et al. 2002) in LacO-V worms. In male germ cells, SYP-1 stretches first appeared in early TZ nuclei (Figure 2, A and B). Paired chromosome V homologs were detected almost exclusively in nuclei containing SYP-1 on all chromosomes, indicating that completion of SC assembly occurs coordinately on all chromosomes in males. Indeed, all homolog pairs contained a fully assembled SC by mid TZ (Figure 2, A, C, and D).
As homolog pairing and SC assembly is coincident with RAD-51 foci formation in males, we examined the status of the X chromosome in TZ nuclei. Surprisingly, we did not observe a chromosome lacking SYP-1 in several mid-to-late TZ stage nuclei (Figure 2, C and D, arrowhead), suggesting that SYP-1 was loaded on the X chromosome. To investigate this, we used SIM to assess loading of both SYP-1 and the axial component HTP-3, which accumulates on chromosomes in early TZ nuclei (Figure 1, A and B). In every male germ line examined (n = 18), we detected one to three mid-to-late TZ nuclei where we could unambiguously identify the X (by absence of H3K4me2) that contained a SYP-1 track (Figure 2, E–H). By EP, however, a single DAPI-stained body, presumably the X, was lacking SYP-1 (Figure 2A), suggesting that SYP-1 loads on the X only within a brief window coincident with onset of meiotic DSB formation.
As the X lacks a homolog, SYP-1 could be loaded onto regions of the X that have folded back on itself, assembled on a single axis, or loaded between the sister chromatids. The six C. elegans chromosomes are very similar in overall nucleotide length (C. elegans Sequencing Consortium 1998), although the X chromosome of males becomes highly condensed as meiosis proceeds (Bean et al. 2004; Checchi and Engebrecht 2011). We observed two classes of nuclei (n = 30) where SYP-1 tracks were present on the X: nuclei where the X appeared shorter and more compact than the autosomes (36.66%; Figure 2, E and F), consistent with fold-back synapsis, and nuclei where the X and autosomes were comparable in length (63.33%; Figure 2, G and H), suggesting that SYP-1 is assembled on a single axis and/or loaded between the sisters. Further, in some nuclei with an extended X, two tracks of DAPI and HTP-3 could be distinguished (HTP-3; red arrowheads) similar to what was observed on autosomes (HTP-3; green arrowheads) (Figure 2H), suggesting that in these nuclei, SYP-1 is loaded between the sister chromatids.
Meiotic cohesin is altered on the X chromosome of males
A prediction based upon SYP-1 loading between the X sister chromatids is that sister chromatid cohesion (SCC) will also be altered. To investigate this, we first monitored the localization of the meiotic kleisins REC-8 and COH-3, which are essential for SCC during meiosis (Severson et al. 2009; Tzur et al. 2012). In hermaphrodites, REC-8 is nucleoplasmic in mitotic germ cells and localizes to chromatin in early meiotic prophase (Pasierbek et al. 2001; Severson et al. 2009). Similarly, we found that in males, REC-8 was visible in all mitotic germ cell nuclei and became concentrated on chromosomes upon entry to the TZ (Figure 3, A and B). COH-3 was not observed in mitotic germ cells but was recruited to chromatin in a similar pattern to REC-8 upon meiotic entry (Figure 3, C and D).
Whereas REC-8 and COH-3 were detected on autosomes beginning in early meiotic prophase, both REC-8 and COH-3 had an altered staining pattern on the X chromosome. REC-8 was detected on the X in TZ nuclei, yet COH-3 was only observed on autosomes at this stage (Figure 3, B and D). Interestingly, by early pachytene, the X was either devoid or contained significantly reduced levels of both REC-8 and COH-3 as compared to autosomes (Figure 3, B and D). By MP to late pachytene (LP), however, stretches of REC-8 and COH-3 became apparent on the X chromosome. As REC-8 and COH-3 staining was also present in spermatocyte nuclei where the X is very highly condensed, it is unlikely that the compacted nature of the X precludes antibody access at earlier stages. Furthermore, the same staining pattern was observed in met-2 males (data not shown), which fail to load the repressive histone mark H3K9me2 on the X chromosome of males (Bessler et al. 2010; Checchi and Engebrecht 2011). Additionally, in males lacking ZIM-2, a protein required for chromosome V homolog pairing (Phillips and Dernburg 2006), robust staining was observed on the asynapsed chromosome Vs throughout pachytene, indicating that decreased REC-8/COH-3 staining on the X is not a consequence of having unpaired vs. paired homologs (Figure 3E). These results indicate that loading and accumulation of meiotic cohesin is altered on the X chromosome of males.
REC-8 and COH-3/4 are not required for RAD-51 removal on the X chromosome
The reduction/lack of REC-8 and COH-3 on the X chromosome of males in early prophase suggests that meiotic cohesin is not a requirement for DSB repair on the X. To test this, we monitored RAD-51 kinetics in mutants lacking REC-8 and COH-3/4. In wild-type male germ lines, RAD-51 foci are largely disassembled by mid pachytene (Figure S1), reflecting repair. In contrast, rec-8; coh-3/4 male germ lines had abundant RAD-51 foci that persisted throughout meiotic prophase (Figure 4, A and B), suggesting that as in hermaphrodites (Tzur et al. 2012), DSB repair is impaired in males lacking meiotic kleisins. To determine whether RAD-51 disassembly is compromised on the X chromosome in rec-8; coh-3/4 male germ lines, we compared the number of RAD-51 foci on the X chromosome in wild-type vs. rec-8; coh-3/4 males (Figure 4B). We found no significant difference (P = 0.081) in the number of X chromosome-specific RAD-51 foci in late pachytene nuclei, suggesting that DSBs on the X can be repaired in the absence of meiotic kleisins.
We next examined the association of sister chromatids by FISH in males harboring a single copy lacO array on the X chromosome. As expected, both wild-type and rec-8; coh-3/4 males contained predominantly a single FISH focus on the X in proliferative zone nuclei, indicative of cohesion between sisters presumably due to the presence of mitotic kleisins. In TZ, 5.0% of wild-type and 10.0% of rec-8; coh-3/4 nuclei had two distinct foci (P = 0.76), suggesting that sister chromatid association is largely independent of meiotic kleisins at this stage of meiotic prophase. However, as pachytene progressed, rec-8; coh-3/4 germ lines displayed increasing percentages of nuclei with two distinct foci compared to wild type [EP: 14 vs. 3.2% (P = 0.0022); MP: 18.1 vs. 5.6% (P = 0.0072); LP: 31.4 vs. 8.4% (P < 0.0001)] (Figure 4C). These results suggest that meiotic kleisins contribute to sister interactions during pachytene but are not required for meiotic DSB repair on the X chromosome of males.
SMC-5/6 facilitates DSB repair on autosomes but is dispensable for DSB repair on the X chromosome of males
Structural maintenance of chromosome proteins SMC-5 and SMC-6 (SMC-5/6) have been proposed to mediate sister chromatid recombination in the C. elegans hermaphrodite germ line (Bickel et al. 2010). smc-5/6 hermaphrodites give rise to viable offspring, presumably because interhomolog (IH) repair is intact; however, a subset of RAD-51 foci persist through late pachytene (Bickel et al. 2010). To determine whether SMC-5/6 plays a role in recombinational repair on the X chromosome of males, we monitored RAD-51 kinetics in smc-5/6 male germ lines and saw elevated numbers of RAD-51 foci throughout pachytene (Figure 5C and Figure S2A). Interestingly, RAD-51 did not accumulate to the same extent in smc-5/6 males compared to hermaphrodites, suggesting that there are sex-specific constraints underlying meiotic DSB repair (Bickel et al. 2010) (Figure 5C and Figure S2A). Despite these differences, however, SMC-5/6 is important for male gamete quality, as there is a 10% decrease in offspring survival among the progeny of females fertilized by smc-5 male sperm (Figure S3A).
As IH repair cannot be engaged on the partnerless X chromosome of males, we asked whether RAD-51 foci persisted on the X chromosome in the absence of SMC-5/6. Surprisingly, we found no significant difference in the number of RAD-51 foci on the X chromosome of smc-5/6 mutant males as compared to wild type (P = 0.4269; Figure 5D and Figure S2B), suggesting that SMC-5/6 is not required for DSB repair on the X chromosome.
Similar to the X chromosome of males, asynapsed autosomes also lack the ability to repair DSBs via IH repair (Bickel et al. 2010). We therefore tested whether absence of SMC-5/6 impaired RAD-51 removal in zim-2 males. Indeed, smc-6; zim-2 males contained elevated RAD-51 foci that persisted on the unpaired chromosome Vs throughout late pachytene/spermatocyte (LP/Sc) stage, longer than in either single mutant on its own (Figure 5B and Figure S2, C–K). These results indicate that while SMC-5/6 is required for DSB repair on autosomes when a homolog is unavailable in the male germ line, the constitutively unpaired X chromosome can repair meiotic DSBs in the absence of SMC-5/6.
Absence of SMC-5/6 increases chromosome fragmentation of asynapsed X chromosomes of hermaphrodites but not the X chromosome of males
Another consequence of unrepaired DSBs is chromosome fragmentation, which manifests in loss of genetic material, aneuploid gametes, and inviable offspring (Aguilera and Garcia-Muse 2013). In smc-5 hermaphrodites, chromosome fragments are apparent in oocyte nuclei that form crossovers, indicating a propensity toward chromosome loss even in the presence of successful IH repair (Bickel et al. 2010). We therefore asked whether absence of SMC-5 resulted in X chromosome loss in males, which would be predicted to increase the percentage of “nullo” sex chromosome (0) gametes. As sex is determined by the ratio of X chromosomes to autosomes in C. elegans (Zanetti and Puoti 2013), an increase in nullo sex chromosome gametes would in turn increase the percentage of male offspring when crossed with female (X containing) gametes. Despite a significant reduction in progeny viability, however, we observed no increase in the percentage of male offspring when fog-2 females, which do not produce self sperm (Schedl and Kimble 1988), were mated to smc-5 males, consistent with SMC-5/6 being dispensable for break repair on the X chromosome of males (Figure S3B).
To further investigate whether SMC-5/6 plays a role in DSB repair on the X chromosome, we depleted smc-6 by RNAi and assessed chromosome fragmentation. We first tested whether X chromosome asynapsis caused by the him-8 mutation (Phillips and Dernburg 2006) enhanced the smc-6(RNAi) phenotype. In smc-6(RNAi); him-8 XX germ lines 56% of nuclei contained fragmented chromosomes (n = 36) vs. 29% in smc-6(RNAi) XX alone (n = 31) and 4% in him-8 XX single mutants (n = 47) (P < 0.0001; Figure 6E). These data corroborate previous observations that SMC-5/6 activity is specific to intersister recombination, and that depletion of this complex is highly deleterious in situations where a homolog is unavailable as a repair template (Bickel et al. 2010).
In the hermaphrodite germ line, chromosome fragmentation is readily detected in diakinesis (DI) nuclei as chromosomes at this stage are condensed, and in wild-type hermaphrodites all six bivalents are physically separated from each other throughout the oocyte (Schvarzstein et al. 2010). However, nuclei in the equivalent stage of male meiosis (spermatocytes/sperm) are very highly compacted, which precludes identification of fragmented chromosomes. To circumvent this, we assessed fragmentation in heterogametic fem-3 X0 females, which have a feminized germ line but possess a single X chromosome with properties similar to the X chromosome of males (Hodgkin 1986; Jaramillo-Lambert and Engebrecht 2010; Checchi and Engebrecht 2011). We depleted smc-6 by RNAi in fem-3 X0 females, and in both control (empty vector) and smc-6(RNAi); fem-3 X0 germ lines, we were able to identify the X by its smaller size and compact morphology (Figure 6, A–D). Whereas six DAPI bodies (five bivalents and the univalent X) were observed in all control fem-3 X0 DI stage nuclei, we observed at least seven DAPI bodies in 16% of smc-6(RNAi); fem-3 X0 germ lines scored (Fisher’s exact test P = 0.0002). In contrast, fragmented nuclei were observed in 47% of smc-6(RNAi); fem-3 XX germ lines, which was significantly higher than either fem-3 XX controls or smc-6(RNAi); fem-3 X0 germ lines (P < 0.0001; Figure 6E). These data suggest that the excess DAPI bodies observed in smc-6(RNAi); fem-3 X0 nuclei were most likely fragmented autosomes and not a consequence of unrepaired DSBs on the X. Taken together, the lack of impairment of DSB repair on the X chromosome of C. elegans males in the absence of both REC-8/COH-3/4 and SMC-5/6 in combination with the cytology suggests that X sister chromatids are configured to mimic homologs early during meiotic prophase.
COM-1 is required for DSB processing on all chromosomes in the male germ line
In wild-type males, a COSA-1 focus, which marks the site COs (Yokoo et al. 2012) is detected on all chromosomes except the X, indicating this is the only chromosome that does not form a CO (Figure S4). To determine whether absence of a crossover on the X reflects altered DSB repair pathway selection, we examined RAD-51 localization in com-1 males. C. elegans hermaphrodite germ cells are characterized by a robust HR bias wherein the CtIP homolog COM-1 blocks error-prone NHEJ and promotes IH repair to ensure CO formation; in the absence of COM-1, very few RAD-51 foci are observed as DSBs are shunted through NHEJ (Lemmens et al. 2013). As in hermaphrodites, com-1 male germ lines contained substantially fewer RAD-51 foci on all chromosomes, including the X, as compared to wild-type nuclei (Figure 7), which presumably reflects repair of induced DSBs through NHEJ (Lemmens et al. 2013). These data indicate that despite the absence of a homolog, a robust COM-1-mediated HR bias remains intact on the hemizygous X chromosome.
The X chromosome of males is the only chromosome not subject to ATM-1-dependent feedback mechanisms at the level of DSB formation
Recent studies have uncovered feedback mechanisms proposed to monitor homolog pairing and DSB formation and to ensure timely CO formation (Barchi et al. 2008; Carballo et al. 2013; Kauppi et al. 2013; Rosu et al. 2013; Stamper et al. 2013). We speculated that in the heterogametic sex, such feedback mechanisms are altered to handle the challenges of a constitutively unpaired chromosome. In yeast and mice, the ataxia telangiectasia mutated (ATM) kinase plays a role in the establishment and maintenance of feedback mechanisms that ensure DSB homeostasis; in the absence of ATM more DSBs are formed (Carballo et al. 2013; Kauppi et al. 2013). To address the role of feedback mechanisms in the regulation of the hemizygous X, we first monitored RAD-51 kinetics in C. elegans mutants deficient for ATM-1. We found that both hermaphrodite and male atm-1 mutants contained more RAD-51 foci throughout meiotic prophase than wild type (Figure 7, A and C–E), suggesting that ATM-1 is important for inhibiting DSB formation once a CO precursor has formed on every chromosome.
Elevated RAD-51 could be a consequence of a greater number of DSBs formed, an extended period of break formation, and/or slower turnover. rad-54 mutants have been used to distinguish between increased number of DSBs vs. a defect in repair as RAD-54 is essential for RAD-51-mediated strand exchange during HR and is required for RAD-51 disassembly (Solinger et al. 2002; Mets and Meyer 2009). However, recent work indicates that absence of RAD-54 both impedes RAD-51 turnover and also extends the time for which DSBs are formed (Rosu et al. 2013; Stamper et al. 2013). To elucidate the underlying cause of elevated RAD-51 foci in atm-1 mutants, we analyzed RAD-51 foci in rad-54; atm-1(RNAi) male germ lines. We found an even greater number of foci in rad-54; atm-1(RNAi) than either atm-1 or rad-54 mutants alone (Figure 7 and Figure S5), suggesting that ATM-1 regulates the number of DSBs formed independent of the extension of time when breaks are formed in rad-54 mutants.
In mice, ATM is also required for multiple aspects of sex chromosome regulation including the X-Y crossover (Barchi et al. 2008). To test whether the X chromosome of males is subject to ATM-dependent DSB regulation, we next assessed RAD-51 levels on the X in the absence of ATM-1. In atm-1 male germ lines, the X was the only chromosome that did not contain increased RAD-51 foci (Figure 7D), suggesting that the complete absence of a pairing partner relieves the C. elegans X chromosome of males from these constraints. Together these results suggest that although early processing of breaks is similar between autosomes and the X, later steps are altered to accommodate the absence of a homolog.
HR is essential for RAD-51 disassembly on all chromosomes except the X chromosome of males
To examine subsequent steps in DSB processing, we monitored RAD-51 kinetics in the absence of RAD-54. In rad-54 males, RAD-51 was detected on all chromosomes including the X in early meiotic prophase (mid TZ stage), indicating that the X chromosome is competent for DSB formation in the absence of RAD-54 (Figure 8, A and B). By pachytene, RAD-51 foci were abundant on the autosomes, but the X chromosome was largely devoid of foci in rad-54 males (Figure 8B). We did observe nuclei where one to two RAD-51 foci were retained on the X, suggesting that RAD-54 does facilitate break repair on all chromosomes. However, RAD-51 was disassembled on at least a subset of X DSBs (Figure 8C), as a significantly smaller percentage of the total RAD-51 foci localized to the X during pachytene [2% in rad-54 vs. 14% (one-sixth) in wild-type males; Figure 8D]. Further, a single pair of autosomes contained upwards of five times as many RAD-51 foci as compared to the X (Figure 8E). These data suggest that unlike the autosomes, the X chromosome of males is at least partially competent for meiotic DSB repair in the absence of HR. To determine whether this phenotype also occurred in hermaphrodites carrying unpaired X chromosomes, we assessed RAD-51 kinetics in rad-54; him-8 XX germ lines. We found that RAD-51 readily accumulated on all rad-54; him-8 chromosomes including the unpaired Xs in the hermaphrodite germ line (Figure S6A). These data indicate that only DSBs on the X chromosome of males can be repaired when HR is blocked.
XPF-1 contributes to repair of meiotic DSBs on the X in the absence of HR
Our data reveal that DSB repair on the X chromosome of C. elegans males as monitored by RAD-51 disassembly can still occur in the absence of HR, indicating that additional pathways are available to repair DSBs on the X. To determine whether error-prone repair pathways contribute to DSB repair on the X chromosome of males, we monitored RAD-51 accumulation in mutants defective for NHEJ (lig-4; Clejan et al. 2006) or SSA (xpf-1; Saito et al. 2009) in the presence and absence of HR (rad-54). The number and kinetics of RAD-51 foci in lig-4 or xpf-1 single or double mutant germ lines were similar to wild type, both genome-wide and on the X (Figure S7). The total number of RAD-51 foci was also not significantly different in any of the double or triple mutant combinations compared to the elevated levels observed in rad-54 male germ lines (Figure 9A). Further, inactivation of NHEJ in the absence of HR did not alter the number of RAD-51 foci observed on the X (rad-54; lig-4; Figure 9, B and C). In contrast, there was a significant increase in RAD-51 foci on the X chromosome in the rad-54; xpf-1 double mutant (Figure 9, B and C). As an additive effect was not observed in either rad-54; xpf-1; lig-4 or rad-54; xpf-1; cku-80 triple mutants, these data indicate that following RAD-51 loading, SSA but not NHEJ contributes to X-specific break repair in the absence of HR. Taken together, these results suggest that an XPF-1-dependent SSA pathway can be engaged to repair breaks on the hemizygous X chromosome when HR is unavailable.
The tumor suppressor BRCA2 functions in HR by facilitating the loading of RAD-51 (Holloman 2011). In C. elegans brc-2 mutants, HR is blocked and consequently the single-strand binding protein RPA-1 accumulates at breaks (Martin et al. 2005). Additionally, BRC-2 has been suggested to play a role in SSA (Martin et al. 2005). We therefore monitored the accumulation of RPA-1::GFP and RPA-1::YFP in the absence of BRC-2. Surprisingly, RPA-1::GFP and RPA-1::YFP foci accumulated in late pachytene nuclei in brc-2 mutant males on all chromosomes except the X (Figure 10), suggesting that similar to RAD-54, BRC-2 is not absolutely essential for DSB repair on the X chromosome. Further, this is specific for the X chromosome of males as the asynapsed X chromosomes in brc-2; him-8 hermaphrodites accumulated RPA-1::GFP (Figure S6B). To determine whether XPF-1 contributes to DSB repair on the X chromosome in the absence of BRC-2, we monitored RPA-1::GFP foci in xpf-1; brc-2 double mutants and found that disassembly of RPA-1::GFP was significantly delayed on the X (Figure 10). Taken together, these results suggest that in the absence of either RAD-54 or BRC-2, XPF-1 mediates DSB repair on the hemizygous X chromosome.
Here we uncover one strategy for how meiosis is altered to accommodate hemizygous sex chromosomes. In C. elegans males, both the chromosome scaffold and the stringency of repair pathway choice have been modified to ensure that the completely hemizygous X chromosome is transmitted intact through meiosis.
Sister vs. homolog interactions and meiotic DSB formation
Analysis of the temporal relationship between chromosome pairing, synapsis, and DSB formation revealed that these events are tightly linked in the male germ line. DSBs, as monitored by the loading of the recombinase RAD-51, are invariably observed on paired and synapsed chromosomes (Figure 1 and Figure 2). Although these events can be uncoupled by mutational analyses in C. elegans (Dernburg et al. 1998) (Figure 8 and data not shown), the strict temporal relationship suggests that these events are normally interdependent. The interdependence of these events in C. elegans is also suggested by analyses of both the axial protein HTP-3, which links DSB formation with homolog pairing and synapsis (Goodyer et al. 2008) and CRA-1, whose inactivation uncovers a requirement for synapsis in DSB repair (Smolikov et al. 2008).
Our studies of the relationship between homolog pairing, synapsis, and RAD-51 assembly in the male germ line revealed a transient window whereby SYP-1 is loaded on the X chromosome concomitant with initiation of meiotic DSBs. Previous studies have indicated that loading of central element components is not dependent on homology per se and that SYP-1 can be loaded between nonhomologous chromosomes (Couteau and Zetka 2005; Martinez-Perez and Villeneuve 2005), on individual unpaired chromosomes (Smolikov et al. 2008; Zhang et al. 2012) or between regions of the same chromosome that have folded back on itself (Harper et al. 2011). While male germ line cytology is less favorable than in hermaphrodites (Jaramillo-Lambert et al. 2010; Checchi and Engebrecht 2011; Woglar et al. 2013) and SYP-1 is only detected on the X in TZ nuclei, where chromosomes are not well separated and therefore difficult to distinguish as individual entities, our SIM data provide evidence that SYP-1 can be loaded between the X sister chromatids (Figure 2, E–H). We propose that the X sister chromatids behave as homologs during the window when DSBs are induced to promote HR using the “sister” as a repair template (Figure 11). This configuration of the X chromosome may counteract the homolog bias mediated by the chromosome axis (Couteau et al. 2004). One prediction of such a model is that meiotic SCC is altered to enable sisters to interact as homologs. Consistent with this, the meiotic kleisins REC-8 and COH-3 have altered patterns of localization on the X in TZ when SYP-1 is loaded (Figure 3). Further, in contrast to autosome pairs, RAD-51 is disassembled on the X with apparently normal kinetics when all three meiotic kleisins are inactivated (Figure 4), suggesting that the majority of DSBs on the X can be repaired in the absence of meiotic kleisins. Finally, during the period when SYP-1 is loaded, sister chromatid association is not dependent on meiotic kleisins (Figure 4). This model also provides an explanation for why the SMC-5/6 complex, which has been suggested to mediate intersister recombination in the hermaphrodite germ line (Bickel et al. 2010), is not required for DSB repair on the X as monitored by RAD-51 disassembly as well as transmission through meiosis and chromosome fragmentation (Figure 5, Figure 6, and Figure S3). This is in contrast to delayed disassembly of RAD-51 on unpaired autosomes in smc-6 mutants (Figure 5 and Figure S2), indicating that the X chromosome is unique in its lack of requirement for the SMC-5/6 complex.
SC polymerization is dynamic, processive, and has a propensity to assemble on any available substrate (Voelkel-Meiman et al. 2012). What then removes SYP-1 specifically from the X chromosome at the onset of pachytene, well before it is disassembled from autosomes? SYP-1 localization parallels REC-8 localization, suggesting that the removal of SYP-1 follows the removal of REC-8 at the TZ/EP transition. However, it is unclear what removes REC-8 at the onset of pachytene only from the X (Figure 3). One possibility is that both SYP-1 and REC-8 are disassembled as the X chromosome condenses beginning in early pachytene. X chromatin condensation correlates with enrichment of repressive histone modifications (Kelly et al. 2002); however, even in the absence of the repressive histone mark H3K9me2, the X can be distinguished by its compact nature (Checchi and Engebrecht 2011), suggesting that multiple complexes are required for condensation and perhaps by extension SYP-1 and REC-8 removal from the X.
Just as the chromosomal basis of sex has evolved multiple times, it is also likely that the process by which meiosis has been altered to accommodate hemizygosity of sex chromosomes has evolved independently numerous times. Interestingly, in mouse spermatocytes, the SYP-1 equivalent SYCP1 is observed in patches along the non-PAR regions of the X and Y chromosomes (Page et al. 2006a), suggesting that central region components of the SC are loaded on regions of mammalian sex chromosomes that are neither synapsed nor homologous to each other. Further, in marsupials, the X and Y are completely nonhomologous yet are connected in meiotic prophase by a dense plate composed of the axial element SCP3 and the central region component SCP1 (Page et al. 2006b). Incorporation of SCP1 is homology independent and is proposed to be important for segregation of the sex chromosomes. Thus, association of central region components of the SC on nonhomologous regions of sex chromosomes may be a relatively general feature of hemizygous sex chromosomes during meiosis.
Regulation of X DSB processing and repair indicate a HR bias not directed toward crossing over
During meiosis, SPO-11-dependent DSBs are processed by a large set of proteins to promote repair by HR. Of critical importance is the CtIP/Sae2 homolog COM-1, which inhibits direct religation of DSBs by CKU-70/80, prior to RAD-51 filament formation (Lemmens et al. 2013). DSBs on all chromosomes of males, including the X, are subject to COM-1-dependent regulation as the numbers of both genome-wide and X-specific RAD-51 foci were significantly reduced in com-1 mutants (Figure 7). This suggests that initial processing of DSBs is similar on autosomes and the X, and in both cases results in channeling repair through HR. Following nucleolytic degradation of the 5′ end, the resulting 3′ single-stranded DNA is immediately loaded by RPA-1, which is subsequently displaced by RAD-51 via BRC-2 (Martin et al. 2005). Interestingly, we observed similar number and size of RAD-51 foci on the X and each autosome pair (Figure 8D), suggesting that early processing events leading up to strand exchange are similar between the X and autosomes. However, COSA-1, a protein that marks sites of interhomolog crossovers (Yokoo et al. 2012), is not loaded on the X chromosome of males (Figure S4), indicating that DSBs are not designated to become COSA-1 marked crossovers. As crossover sites define bivalent asymmetry for reductional segregation (Nabeshima et al. 2005), and X sisters remain associated through meiosis I (Albertson and Thomson 1993; Shakes et al. 2009), it is likely that a crossover on the X would interfere with segregation at the first meiotic division. Taken together, these observations suggest that a step subsequent to strand exchange excludes X recombination sites from being designated as crossovers.
A number of recent studies have provided evidence that robust feedback mechanisms are in place during meiotic prophase to ensure regulated crossover formation (Carballo et al. 2013; Kauppi et al. 2013; Rosu et al. 2013; Stamper et al. 2013). In both yeast and mice, the ATM checkpoint protein plays a critical role in feedback regulation by inhibiting break formation; consequently, in the absence of ATM more DSBs are formed. Similarly, we found that in the C. elegans germ line, atm-1 mutants have elevated RAD-51 foci throughout meiotic prophase in both hermaphrodites and males (Figure 7 and Figure S5). Work from the Dernburg and Villeneuve labs has revealed that failure to form a crossover intermediate, as in rad-54 mutants, results in extension of the period of time in which breaks are formed (Rosu et al. 2013; Stamper et al. 2013). As rad-54; atm-1(RNAi) mutants have more RAD-51 foci than either of the single mutants (Figure 7 and Figure S5), we interpret this to mean that ATM-1 inhibits break formation independent of the crossover precursor signal that extends the time when breaks can be formed. Interestingly, we found that the X chromosome of males is not subject to this ATM-1 feedback regulation, suggesting that during processing, X breaks become differentiated from breaks on autosomes. These results are also consistent with our previous analyses of X DSBs not being detected by checkpoint machinery (Jaramillo-Lambert and Engebrecht 2010; Checchi and Engebrecht 2011). The interplay between checkpoint machinery, DSB processing and chromatin environment on the X is currently being investigated.
Prevalence of error-prone repair pathways in meiosis
Although we propose that noncrossover HR is the predominant mode of DSB repair on the X chromosome, inactivation of HR by mutation of either RAD-54 or BRC-2 does not block repair on the X to the same extent as on autosomes (Figure 8, Figure 9, and Figure 10). Repair of X DSBs is significantly impaired in the absence of either RAD-54 or BRC-2 and the structure-specific nuclease XPF-1. While recent studies have revealed a role for XPF-1 in crossover resolution, it is clear that XPF-1 plays additional roles in the germ line (Agostinho et al. 2013; O’Neil et al. 2013; Saito et al. 2013) and has been shown to function in SSA in somatic cells (Al-Minawi et al. 2008; Pontier and Tijsterman 2009). Genetic analysis in C. elegans revealed that in addition to the critical role BRC-2 plays in promoting RAD-51 filament formation, BRC-2 also functions independently of RAD-51 in DSB repair (Martin et al. 2005). As C. elegans does not have a Rad52 ortholog, it was suggested that BRC-2 mediates SSA. However, our results suggest that BRC-2 and XPF-1 define two genetically separable pathways for RAD-51-independent repair, and that DSB repair on the X can proceed in the absence of BRC-2 function. Thus, when HR is blocked subsequent to DSB resection as in brc-2 or rad-54 mutants, we propose that XPF-1-dependent SSA is the predominant pathway for DSB repair on the X chromosome of males (Figure 11).
Error-prone pathways such as NHEJ and SSA are believed to be the last resort for repair of DSBs in meiosis due to their propensity to cause mutations or chromosome abnormalities that could result in inviable progeny or be passed on to the next generation and lead to sterility. In the hermaphrodite germ line, NHEJ is engaged only at the very end of prophase after an extended period of time where HR is the predominant mode of repair (Colaiacovo et al. 2003; Hayashi et al. 2007). Release from the HR mode in hermaphrodites correlates with MAP kinase signaling (Hayashi et al. 2007). Males do not have MAP kinase signaling at this juncture of meiotic prophase (Lee et al. 2007), suggesting males regulate these processes differently. This is also supported by our observations that the overall kinetics of break repair are different in hermaphrodites and males. Nonetheless, our genetic and cytological analyses suggest that XPF-1 is only engaged to repair X DSBs and does not play a general role in break repair in the male germ line. One potential consequence of XPF-1-dependent repair would be a higher mutation rate on the X as compared to the autosomes, in lines maintained by outcrossing. Whether this is observed and a consequence of the hermaphroditic life style of C. elegans or a more general property of sex chromosomes that contributes to their rapid evolution awaits future studies.
We thank A. Dernburg, J. Loidl, and A. Villeneuve for generously providing antibodies and the Caenorhabditis Genetic Center for strains. We also thank A. Severson for advice on using LacI-His6-GFP, D. Cortez for advice on FISH, A. Villeneuve for helpful discussion, and E. Tapley and N. Andrews for critical reading of the manuscript. This work was supported by National Institutes of Health (NIH) GM103860 and the Agricultural Experimental Station California-Davis*MCB-7237-H (J.E.), NIH T32 CA10849 (P.M.C.), and NIH T32 GM0070377 (K.S.L.).
Communicating editor: M. C. Zetka
- Received February 18, 2014.
- Accepted March 13, 2014.
- Copyright © 2014 by the Genetics Society of America