Role of RAD51C and XRCC3 in Genetic Recombination and DNA Repair*

In germ line cells, recombination is required for gene reassortment and proper chromosome segregation at meiosis, whereas in somatic cells it provides an important mechanism for the repair of DNA double-strand breaks. Five proteins (RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3) that share homology with RAD51 recombinase and are known as the RAD51 paralogs are important for recombinational repair, as paralog-defective cell lines exhibit spontaneous chromosomal aberrations, defective DNA repair, and reduced gene targeting. The paralogs form two distinct protein complexes, RAD51B-RAD51C-RAD51D-XRCC2 and RAD51C-XRCC3, but their precise cellular roles remain unknown. Here, we show that, like MLH1, RAD51C localized to mouse meiotic chromosomes at pachytene/diplotene. Using immunoprecipitation and gel filtration analyses, we found that Holliday junction resolvase activity associated tightly and co-eluted with the 80-kDa RAD51C-XRCC3 complex. Taken together, these data indicate that the RAD51C-XRCC3-associated Holliday junction resolvase complex associates with crossovers and may play an essential role in the resolution of recombination intermediates prior to chromosome segregation.

In germ line cells, recombination is required for gene reassortment and proper chromosome segregation at meiosis, whereas in somatic cells it provides an important mechanism for the repair of DNA double-strand breaks. Five proteins (RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3) that share homology with RAD51 recombinase and are known as the RAD51 paralogs are important for recombinational repair, as paralog-defective cell lines exhibit spontaneous chromosomal aberrations, defective DNA repair, and reduced gene targeting. The paralogs form two distinct protein complexes, RAD51B-RAD51C-RAD51D-XRCC2 and RAD51C-XRCC3, but their precise cellular roles remain unknown. Here, we show that, like MLH1, RAD51C localized to mouse meiotic chromosomes at pachytene/diplotene. Using immunoprecipitation and gel filtration analyses, we found that Holliday junction resolvase activity associated tightly and co-eluted with the 80-kDa RAD51C-XRCC3 complex. Taken together, these data indicate that the RAD51C-XRCC3-associated Holliday junction resolvase complex associates with crossovers and may play an essential role in the resolution of recombination intermediates prior to chromosome segregation.
Genetic recombination is required for the maintenance of genome stability, proper chromosome segregation, and the reassortment of genetic traits during meiosis. In humans, homologous recombination requires a number of proteins, including RAD51, RAD52, RAD54, replication protein A, MRE11/RAD50/NBS1, and the RAD51 paralogs (RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3) (1). The RAD51 recombinase plays a critically important genome caretaker role, as indicated by the accumulation of chromosome breaks in chicken DT40 cells carrying a repressible RAD51 transgene (2). The requirement for other homologous recombination proteins is less extreme, but cell lines defective in these proteins generally show a characteristic sensitivity to DNA-damaging agents and exhibit a spontaneous chromosome instability phenotype.
Two primary RAD51 paralog complexes are thought to exist in vivo: one contains RAD51B, RAD51C, RAD51D, and XRCC2 (designated the BCDX2 complex), and the other RAD51C and XRCC3 (3,4). In mice, disruption of any of the five RAD51 paralogs leads to embryonic lethality (5)(6)(7). 6 This indicates that each of the RAD51 paralogs is essential and non-redundant during development. However, hamster and human cell lines that are defective in the RAD51 paralogs have been generated and shown to exhibit a recombination/repair-defective phenotype (9 -12).
Following DNA damage, RAD51 forms nuclear foci that are thought to represent sites where repair reactions take place. The formation or stability of these RAD51 foci is compromised in paralog-defective cells (13,14), leading to the suggestion that the paralogs are required as mediators for the formation of active RAD51 filaments (15,16). In support of this, purified RAD51B-RAD51C stimulates the activity of substoichiometric amounts of RAD51 (17), and the overexpression of RAD51 can partially complement paralog-defective chicken DT40 cells (14).
Recent evidence suggests that some of the paralogs may also act late in recombination. For example, XRCC3-defective cell lines exhibit gene conversion lengths in excess of the normal length (18), and RAD51C-defective cells show an increased frequency of isochromatid breaks (11). These events may result from defects in the resolution of recombination intermediates. In Arabidopsis XRCC3 mutants, meiotic chromosomes pair but then fail to segregate properly at the late stages of meiotic recombination (19). These defects are not observed in RAD51B or XRCC2 mutants, which remain fertile (19,20). Support for a late role for the RAD51C-XRCC3 complex comes from observations showing that extracts made from XRCC3-or RAD51C-defective hamster cells lack normal levels of Holliday junction resolvase activity (21).
These results indicate that the RAD51C-XRCC3 complex may function in a manner that is distinct from other RAD51 paralogs by playing a specific role in the resolution of recombination intermediates prior to chromosome segregation. Consistent with this notion, an xrcc3/rad51d double deletion in DT40, involving proteins from both the BCDX2 and RAD51C-XRCC3 complexes, exhibits a greater DNA repair defect than either of the two single deletions (22). In the work reported here, we further define the roles played by the RAD51C-XRCC3 complex.
Protein Fractionation-Nuclear extracts were prepared from HeLa cells, and a 25-55% (NH 4 ) 2 SO 4 cut was made as described (21). The precipitate (ϳ3.6 g) was recovered by centrifugation and stored as two 1.8-g pellets. Each was resuspended in 8 ml of buffer A (50 mM K 2 HPO 4 /KH 2 PO 4 (pH 6.8), 10% glycerol, 1 mM EDTA, 1 mM dithiothreitol, and 0.01% Nonidet P-40) containing 250 mM KCl and dialyzed against the same buffer before loading onto a 320-ml Sephacryl S-300 gel filtration column (Amersham Biosciences). Proteins were eluted with 320 ml of buffer A containing 250 mM KCl. Fractions (4 ml) were collected and stored at Ϫ80°C. Molecular mass standards (Bio-Rad) were analyzed under the same conditions.
RAD51C Affinity Chromatography-Purified His-RAD51C protein was pre-bound to a nickel-nitrilotriacetic acid column and incubated overnight at 4°C with gentle shaking with HeLa nuclear extract in buffer B (50 mM K 2 HPO 4 /KH 2 PO 4 (pH 7.5), 10% glycerol, 300 mM KCl, 2 mM ␤-mercaptoethanol, and 0.01% Nonidet P-40) containing 5 mM imidazole. The unbound proteins were removed, and the column was washed with the same buffer. RAD51C and bound proteins were subsequently eluted using buffer B supplemented with 10, 20, 40, 80, and 500 mM imidazole.
DNA Substrates-Synthetic Holliday junctions X26 and X0 were made by annealing four oligonucleotides and purified by gel electrophoresis (26). Both junctions were 5Ј-32 P-labeled on oligonucleotide 1. Control double-stranded DNA (ds26) was made by annealing 32 P-labeled strand 1 of Holliday junction X26 with its complement.
Resolution and Branch Migration Assays-Resolution assays were carried out using 32 P-labeled synthetic Holliday junction X0 or X26 in the presence of a 300-fold excess of poly(dI⅐dC) DNA (21). Branch migration reactions by RecQ5␤ were performed as described using 32 P-labeled Holliday junction X26 (25).
Immunoprecipitation-Aliquots (200 l) of gel filtration fractions 19 -23 were incubated for 10 h at 4°C with polyclonal antibody (pAb) 7 against either XRCC3 (SWE30) or XRCC2 (SWE35) that had been pre-bound to protein A-Sepharose beads (15 l; GE Healthcare). The supernatant was removed, and the beads were washed five times with buffer A containing 1 M KCl. The washed beads were analyzed for activity.
Immunofluorescence Staining-Spermatocytes from 6 -8week-old wild-type or Mlh1 knock-out (31) mice were surfacespread and immunostained as described (28). To visualize MLH1 or RAD51C, the spermatocyte suspension was transferred to a sterile Petri dish and incubated with E4 cell culture medium (1 ml) containing 5 M okadaic acid (Sigma) for 2-4 h at 37°C (32). Slides were then washed and further processed for immunostaining. The okadaic acid treatment accelerates meiotic progression and has been used previously to visualize MLH1 at chiasmata during male meiosis (33). The specificity of RAD51C staining was confirmed using four different antibodies (mAb 2H11 and pAbs SWE31, SWE68, and SWE72).

Interaction of the Resolvase Complex with Holliday Junctions-
Previously, HeLa nuclear extracts were fractionated through several columns to enrich Holliday junction resolvase activity (21). The final fraction, designated SP-15, was found to bind 7 The abbreviations used are: pAb, polyclonal antibody; mAb, monoclonal antibody; ChIP, chromatin immunoprecipitation; HJ, Holliday junction; PAR, pseudoautosomal region.

RAD51C-XRCC3 in Recombinational Repair
synthetic Holliday junction DNA to form two distinct protein-Holliday junction (HJ) complexes (*1 and *2), even in the presence of a 50-fold molar excess of competitor duplex DNA (Fig.  1A, lanes d-h). Under similar conditions, SP-15 did not bind duplex DNA substrates (Fig. 1B, lanes c-e). Similar results were obtained with the Escherichia coli Holliday junction-binding protein RuvA (Fig. 1, A, lane c; and B, lanes a and f ). The HJ resolvase activity in SP-15 is known to be RAD51C-dependent (21), and consistent with this result, immunodepletion of RAD51C from SP-15 resulted in a loss of Holliday junctionbinding activity (Fig. 1B, lane k). We therefore concluded that the DNA-binding component in SP-15 is a RAD51C-containing complex.
When experiments were carried out with purified recombinant RAD51C-XRCC3, we observed complexes similar to those formed by SP-15 (Fig. 1A, lanes i-l). However, the total amount of protein-DNA complex formed using purified RAD51C-XRCC3 was small in comparison with the SP-15 fraction. As observed with SP-15, recombinant RAD51C-XRCC3 bound specifically to HJ DNA, and little binding to linear duplex DNA was observed under the same conditions (Fig. 1A, lane b). The small amount of protein-DNA complex formed by the recombinant RAD51C-XRCC3 complex may either be due to the instability of the protein-DNA complex under the conditions of the band shift assays or result from the observed high levels of insoluble protein in the RAD51C-XRCC3 preparation.
We therefore used an alternative approach to demonstrate junction binding by RAD51C-XRCC3 by determining whether the presence of RAD51C-XRCC3 could affect the access of other enzymes to the Holliday junction. We found that RAD51C-XRCC3 blocked the dissociation of a junction mediated by the RecQ5␤ DNA helicase (25). Indeed, RAD51C-XRCC3 was equally as active as E. coli RuvA (Fig. 1, C, compare lanes c-e and i-k; and D). In contrast, the BCDX2 complex failed to block RecQ5␤-mediated dissociation of the DNA substrate ( Fig. 1, C, lanes  f-h; and D). Because a subcomplex of BCDX2 (RAD51D-XRCC2) interacts with BLM (another RecQ family protein) and stimulates its HJ dissociation activity (36), we cannot rule out the possibility that inhibition of RecQ5␤ by RAD51C-XRCC3 occurs by direct interaction of this paralog pair with RecQ5␤. However, direct interactions between RAD51C-XRCC3 and RecQ5␤ have not been observed, 8 leading us to suggest that the inhibitory effects of RAD51C-XRCC3 on RecQ5␤mediated junction dissociation are most likely due to the junction-binding properties of RAD51C-XRCC3.
Direct Association of RAD51C and XRCC3 with HJ Resolvase Activity-The resolvase activity present in HeLa and hamster cell extracts has also been detected in a variety of calf and rabbit organ tissues (thymus, testis, and spleen) (37)(38)(39). An identical activity is found in plant extracts, such as those obtained from wheat germ, giving us a unique opportunity to determine whether exogenous expression of RAD51C (effectively using an in vitro transcription/translation system) in a wheat germ extract stimulates or blocks the endogenous resolvase activity present in the extract. We found that expression of human RAD51C efficiently inhibited HJ resolution catalyzed by the wheat germ HJ resolvase (Fig. 2A, compare lanes b and c). Inhibition was dependent upon a functional RAD51C, as indicated by the inability of RAD51C(K131A), which carries a single amino acid substitution in the ATP-binding domain (21), to act as an inhibitor (Fig. 2A, lane i). Expression of other RAD51 paralogs, such as RAD51B (Fig. 2A, lane e), XRCC3 (lane d), and XRCC2 (data not shown), failed to inhibit resolution, even though all proteins were expressed at comparable levels as detected by [ 35 S]methionine labeling (lower panels).
To determine whether inhibition of the wheat germ resolvase activity was due to the binding of the Holliday junc-8 P. Janscak, personal communication.  JANUARY 19, 2007 • VOLUME 282 • NUMBER 3 tion by RAD51C, we analyzed the effects of expressing RuvA in the same in vitro transcription/translation system. In contrast to RAD51C, RuvA expression did not inhibit the wheat germ resolvase (Fig. 2B, lane d). On the basis of Western analysis, we estimated the amount of RAD51C or RuvA to be no more than 0.5 nM, which is less than the HJ substrate concentration. It is therefore unlikely that human RAD51C inhibits wheat germ resolvase activity by direct junction binding, leading us to suggest that inhibition may be due to the formation of an inactive interspecies complex between human RAD51C and a component of the wheat germ HJ resolvasome.

RAD51C-XRCC3 in Recombinational Repair
Next, we demonstrated the direct interaction between RAD51C and HJ resolvase activity using affinity chromatography (Fig. 2C). When nuclear extracts made from HeLa cells were incubated with a nickel column loaded with recombinant His-tagged RAD51C, we found that only a very small proportion of the applied protein bound to the column, as detected by PAGE of the column eluate (Fig. 2D, lane d). However, we found that all resolvase activity present in the extract bound to the RAD51C column and subsequently eluted in this fraction, as detected by HJ cleavage assays (Fig. 2E, upper panel). Simi-larly, XRCC3 present in the HeLa extract bound to the RAD51C column and eluted with resolvase activity (Fig. 2E, lower panel).
To further analyze the interaction of the resolution activity with RAD51C and XRCC3, we fractionated the nuclear extracts made from HeLa cells by gel filtration after ammonium sulfate precipitation. We found that a peak of HJ resolvase activity eluted with an average molecular mass of ϳ80 -90 kDa (Fig. 3A). The peak of the endogenous RAD51-C-XRCC3 complex, indicated by the elution profile of XRCC3 as detected by Western blotting, was coincident with resolvase activity (Fig. 3A, second panel). RAD51C was present in these peak fractions but was also found in fractions containing larger protein complexes; this is due to the fact that RAD51C is also a component of the BCDX2 complex, which migrates with a molecular mass of 160 kDa. The gel filtration profile of the BCDX2 complex was determined by Western blotting for XRCC2 (Fig. 3A, fourth  panel).
Immunoprecipitation of XRCC3 from resolvase peak fractions 19 -23, containing protein complexes with a mass below 100 kDa, pulled down the RAD51C-XRCC3 complex together with quantitative amounts of resolvase activity even in the presence of 1 M KCl (Fig. 3, B and C, lanes d and h). In contrast, an XRCC2 pulldown from the same fractions failed to bring down RAD51C, XRCC3, or the resolvase activity (Fig. 3C,  lanes f and i), consistent with previous data indicating that XRCC2 is not required for HJ resolution (21). These results confirm the association between RAD51C-XRCC3-containing complexes and HJ resolvase activity.
The average molecular mass of the resolvase complex (80 -90 kDa) containing RAD51C and XRCC3 suggests that the RAD51C/XRCC3 heterodimer itself is responsible for the observed resolution activity. However, preparations of recombinant RAD51C-XRCC3 protein, overexpressed and purified from insect cells, failed to provide a convincing resolution signal in our assays. Further studies will therefore be required to determine whether modification(s) of the RAD51C-XRCC3 complex could result in activation of resolvase activity or to define the involvement of additional protein factors.
Immunofluorescence Detection of RAD51C during Meiosis-Previously, immunofluorescence localization and ChIP techniques were used to show the association of RAD51C with chromatin at sites of double-strand breaks (40). However, these FIGURE 2. Interaction of HJ resolvase with RAD51C. A and B, RAD51C expression in wheat germ extracts blocks HJ resolution by the wheat resolvase. The indicated human and E. coli proteins were expressed in wheat germ transcription/translation extracts, and HJ resolution assays were carried out as described under "Experimental Procedures." The DNA products were deproteinized and analyzed on 10% neutral polyacrylamide gel (upper panels). The amounts of the indicated proteins made by the wheat germ extract were monitored by the addition of [ 35 S]methionine and analyzed by SDS-PAGE (lower panels). C, the schematic diagram indicates the binding of HJ resolvase activity by RAD51C affinity chromatography. Ni-NTA, nickel-nitrilotriacetic acid; On, sample applied to column; FT, flow-through. D, nuclear extracts from HeLa cells were applied to a RAD51C affinity column. Bound proteins were eluted as described under "Experimental Procedures" and analyzed by SDS-PAGE, followed by silver staining. E, eluted fractions were analyzed for their ability to promote the resolution of a 32 P-labeled Holliday junction (upper panel) and for XRCC3 by Western blotting (lower panel).
techniques were not sensitive enough to determine whether RAD51C plays an early or late role in subsequent repair reactions. To overcome this limitation, we analyzed the chromo-somal localization of RAD51C and XRCC3 during meiotic progression in male mice. The advantage offered by meiotic recombination is that the various stages of the recombination process are carefully timed and coordinated, so it is relatively easy to distinguish early from late functions. As controls for early/late functions, we used RAD51, which associates with the chromosomes at the early leptotene/zygotene stages of prophase I (Fig. 4A), and MLH1, which serves as a marker for the later pachytene/diplotene stages of the process (Fig. 4B). At diplotene, the physical connections (chiasmata) between chromosomes can be visualized (41), indicating the formation of crossover products. The mismatch repair protein MLH1 is required for normal levels of crossover formation (31,42) and is often used as a marker for recombinational interactions that are designated to become, or have already been realized as, crossovers (43). As expected, we observed one or two MLH1 foci/chromosome pair at pachytene/diplotene.
When similar experiments were carried out for RAD51C, we found a pattern of staining that was strikingly similar to that of MLH1 (Fig.  4C). At the early stages of meiotic recombination (leptotene/zygotene), we observed only background levels of RAD51C, comparable with those obtained with MLH1. We do not exclude the possibility that our inability to visualize RAD51C foci at this stage could be due to the limitations of immunofluorescence techniques that require the accumulation of significant amounts of protein or the transient nature of such complexes. However, at pachytene, we observed one or two distinct RAD51C foci associated with each synapsed bivalent. The average number of RAD51C foci/ pachytene nucleus was 29.9 (sample size, 10 nuclei; S.D., 2.55). Many of these RAD51C foci persisted through diplotene (average number of RAD51C foci/nucleus, 26; sample size, 10 nuclei). Additionally, a small number of RAD51C foci were observed away from the SCP3-delineated axes or were simply in the proximity of FIGURE 3. Direct association of RAD51C and XRCC3 with HJ resolvase activity. A, nuclear extracts from 50 liters of HeLa cells were fractionated by ammonium sulfate precipitation and gel filtration through a Sephacryl S-300 column. Aliquots (1 l) of each fraction were assayed for resolution using 32 P-labeled synthetic Holliday junction X0 (first panel). The same fractions were analyzed by 9% SDS-PAGE and immunoblotted using anti-XRCC3 mAb 10F1, anti-RAD51C mAb 2H11, and anti-XRCC2 pAb SWE35 (second, third, and fourth panels, respectively). B, gel filtration fractions containing the peak of the resolvase activity were mixed with protein A-Sepharose beads cross-linked with anti-XRCC3 pAb SWE30. Immunoprecipitated beads were assayed for their ability to resolve 32 P-labeled synthetic Holliday junction X26 (upper panel) and for the presence of XRCC3 using anti-XRCC3 mAb 10F1 (lower panel). C, gel filtration fractions 19 -23 (containing the peak of the resolvase activity) were mixed with protein A-Sepharose beads cross-linked with either anti-XRCC3 pAb SWE30 or anti-XRCC2 pAb SWE35 (left panel). Pre-I, pre-immune serum. Immunoprecipitated (IP) beads were assayed as described for B and for the presence of RAD51C, XRCC3, and XRCC3 by Western blotting using mAbs against RAD51C (2H11), XRCC3 (10F1), and XRCC2 (7B7) (right panels). FIGURE 4. Localization of RAD51C during male meiotic prophase I. A, mouse spermatocyte nuclear spreads at successive stages of prophase I were immunostained with anti-RAD51 pAb FBE2 (green) and anti-SCP3 mAb 10G11 (red). B, same as A but stained with anti-MLH1 mAb (green) and anti-SCP3 pAb SWE36 (red). C, same as A but stained with anti-RAD51C pAb SWE68 (green) and anti-SCP3 mAb 10G11 (red). axial elements. Whether these represent active protein complexes or are artifacts of the spreading technique is unknown.
MLH1 disruption results in meiotic arrest with reduced crossovers (31), leading us to analyze RAD51C foci in spermatocytes obtained from Mlh1 Ϫ/Ϫ mice. We found that Mlh1 Ϫ/Ϫ spermatocytes arrested at zygotene/early pachytene and that early pachytene cells from these mice showed fewer than five RAD51C foci/nucleus on average (data not shown). These results are consistent with the notion that the RAD51C-XRCC3 complex is targeted to crossover sites. However, attempts to colocalize RAD51C with MLH1 were unsuccessful, as those nuclei exhibiting RAD51C foci (ϳ7% of total) typically did not show MLH1 foci. We suggest that the association of RAD51C and MLH1 with synapsed bivalents at the late stages of recombination is temporal and that the stable chromosomal association of one factor may precede the other. When similar experiments were carried out using antibodies specific for other paralogs such as RAD51B, XRCC2, and XRCC3, we failed to observe foci on the autosomes at either early or late stages of meiotic recombination (data not shown).
RAD51C and XRCC3 Localization at the Pseudoautosomal Region (PAR)-During mammalian spermatogenesis, the X and Y chromosomes pair and synapse over a short region of homology known as the XY PAR, where an obligatory crossover event takes place (44). Analysis of wild-type meiotic spreads revealed that RAD51C localized at or close to the XY PAR (62% frequency; 100 pachytene nuclei examined) (Fig. 5A). Some RAD51C foci were also observed on the unpaired arms of the X and Y chromosomes and may mark sites of intersister recombination events. Although we found it difficult to visualize autosomal XRCC3 foci in spermatocyte nuclear spreads, we observed XRCC3 foci at the XY PAR. MLH1 also localized to this region. In spermatocyte spreads prepared from the Mlh1 Ϫ/Ϫ mouse, we failed to observe RAD51C and XRCC3 foci at the PAR, supporting the conclusion that crossover formation and the presence of RAD51C-XRCC3 foci are interrelated events.
In humans, the PAR crossover hot spot has been mapped and shown to have a recombination frequency ϳ20-fold higher than the genome average (45). In mice, the XY hot spot has yet to be fully characterized because of the presence of variable numbers of pseudoautosomal repeats in different mouse strains. However, the pseudoautosomal boundary, where PAR diverges into X-and Y-specific sequences, has been defined using flanking markers and occurs between DXYCbl1 and DXCbl1 (34,35). Thus, DXYCbl1 and DXYCbl3 serve as markers for the PAR (Fig. 5B). When ChIP assays were performed on mouse testicular cells, we found that antibodies against RAD51C and XRCC3 immunoprecipitated the DXYCbl1 and DXYCbl3 loci (Fig. 5, C and D). In contrast, only background amounts of the control X-and Y-specific regions (DXCbl1 and Y4.2M, respectively) were pulled down. These results confirm the immunofluorescence data, showing that RAD51C and XRCC3 associate with the PAR adjacent to the pseudoautosomal boundary.

DISCUSSION
The results presented here show that the RAD51C-XRCC3 complex associates with HJ resolution activity, a feature not shared by the other RAD51 paralogs. Although the average molecular mass of the resolvase activity (80 -100 kDa) is close to the combined masses of RAD51C (42,192 Da) and XRCC3 (37,850 Da), efforts to detect resolvase activity from purified recombinant RAD51C-XRCC3 have so far proven negative. However, it is possible that the RAD51C/XRCC3 heterodimer could be activated for resolution by post-translational modification or by interaction with an additional small nuclease subunit. Further experiments are now in progress to determine whether RAD51C-XRCC3 constitutes the HJ resolvase and/or to identify RAD51C-associated factors.
The biochemical studies are supported by immunofluorescence studies showing that RAD51C associates with paired bivalents during the late stages of prophase I at meiosis, when FIGURE 5. Association of RAD51C-XRCC3 with the obligate crossover event in the XY PAR determined by immunofluorescence and ChIP analysis. A, sex chromosomes from wild-type (WT) and Mlh1 Ϫ/Ϫ mice were stained with anti-RAD51C pAb SWE68 (green) and anti-SCP3 mAb 10G11 (red) (upper panels), with anti-XRCC3 pAb SWE30 (green) (middle panels), and with anti-MLH1 mAb (green) and anti-SCP3 pAb SWE36 (red) (lower panels). Yellow arrows point to the XY PAR. B, shown is a schematic representation of the mouse XY pair. DNA markers for the PAR (DXYCbl1 and DXYCbl3) and for the X-specific (DXCbl1) and Y-specific (Y4.2M) sequences are shown. C, RAD51C and XRCC3 localize to PAR as determined using ChIP assays. Two 50 -60-day-old male mice (strain 129OLA) were used for each experiment. ChIP assays were carried out on mouse testicular cells using anti-RAD51C pAbs SWE68 and SWE72 (51C), anti-XRCC3 pAb SWE30 (X3), anti-SCP3 pAb SWE36, and anti-TRF2 pAb SWE38. The immunoprecipitated genomic DNA was then PCR-amplified using pairs of primers as described under "Experimental Procedures." In this experiment, anti-SCP3 antibody (the structural component of axial elements) was used as a positive control to indiscriminately immunoprecipitate all four markers. The telomeric protein TRF2 served as a negative control. D, the results of the ChIP assays shown in C were quantified and are shown as the average of three independent experiments. crossover formation and processing occur. Although RAD51C and XRCC3 were seen only at the late stages of meiotic recombination, an observation that supports the concept that they are late-acting proteins (18,19,21), our data do not rule out the possibility that these proteins also play an role early in recombination. Indeed, RAD51C and XRCC3 may form foci at the earlier stages of recombination, but these may be too small or transient to permit their detection by standard immunofluorescence techniques. However, the MLH1-dependent localization of RAD51C and XRCC3 to the PAR, where an obligate crossover event takes place, appears to provide a compelling case for a defined role late in homologous recombination. Taken together, the data presented in this study show that the RAD51C-XRCC3 complex interacts directly with HJ intermediates and associates with HJ resolution activity. Although it is possible that other components of the human HJ resolvase complex remain to be identified, these observations provide a step toward understanding how recombination intermediates are processed in mammalian cells.
Infertility in mice expressing a hypomorphic allele of Rad51c has been shown to be associated with defects at two stages of meiotic recombination. 6 Spermatocytes undergo developmental arrest during the early stages of meiotic prophase I, suggestive of a role for RAD51C in the early stages of recombination. In contrast, oocytes progress normally to metaphase I but display precocious separation of sister chromatids, aneuploidy, and broken chromosomes at metaphase II. These defects are suggestive of a late role for RAD51C in meiotic recombination and are consistent with observations showing that extracts from the Rad51c null mouse exhibit reduced HJ resolvase activity. These observations are in accord with the primary conclusions of this work.