Replication protein A (RPA), a heterotrimeric singlestranded DNA binding protein, is required for recombination, and stimulates homologous pairing and DNA strand exchange promoted in vitro by human recombination protein HsRad51. Co-immunoprecipitation revealed that purified RPA interacts physically with HsRad51, as well as with HsDmc1, the homolog that is expressed specifically in meiosis. The interaction with HsRad51 was mediated by the 70 kDa subunit of RPA, and according to experiments with deletion mutants, this interaction required amino acid residues 169–326. In exponentially growing mammalian cells, 22% of nuclei showed foci of RPA protein and 1–2% showed foci of Rad51. After γ-irradiation, the percentage of cells with RPA foci increased to ∼50%, and those with Rad51 foci to 30%. All of the cells with foci of Rad51 had foci of RPA, and in those cells the two proteins co-localized in a high fraction of foci. The interactions of human RPA with Rad51, replication proteins and DNA are suited to the linking of recombination to replication.
Human Rad51 protein (HsRad51) is a homolog of bacterial recombination protein RecA and yeast Rad51 protein (1,2). In addition to a high degree of structural homology to RecA, HsRad51 and yeast Rad51 share significant functional similarity. The proteins form similar nucleoprotein filaments on singlestranded DNA (ssDNA), possess DNA-dependent ATPase activity, and promote homologous pairing and DNA strand exchange between ssDNA and homologous double-stranded DNA (dsDNA) (3–7). The RAD51 gene of Saccharomyces cerevisiae belongs to the RAD52 epistasis group and, together with other members of the group, is implicated in homologous recombination and repair of DNA double-strand breaks (8). Various studies indicate that mammalian Rad51 protein participates in homologous recombination, DNA repair and immunoglobulin switch recombination (9–14).
Replication protein A (RPA) is a heterotrimeric eukaryotic ssDNA binding protein that is required for the replication, repair and recombination of DNA. In addition, RPA has been implicated in the regulation of transcription (for review see 15). The subunits of RPA are 70, 32 and 14 kDa in mass. RPA interacts physically with about a dozen other proteins that are involved in replication, repair, recombination and transcription. These interactions are usually mediated by the 70 kDa subunit (15). Yeast and human RPA interact with Rad52, a protein which is involved in homologous recombination (16–18). In S.cerevisiae, RPA and Rad52 assemble into subnuclear complexes or foci during meiotic recombination (19); mammalian RPA and Rad51 have recently been shown to co-localize on synapsed axes in meiosis (20). Homologous pairing and DNA strand exchange mediated by yeast or human Rad51 in vitro is facilitated by the cognate RPA (4,21–23).
In the course of observations on the polarity of strand exchange promoted by human Rad51 (23), we examined the physical interaction of Rad51 with RPA, as reported here.
Materials and Methods
We used polyclonal rabbit sera raised against HsRad51, RecA or HsRPA; affinity purified rabbit polyclonal anti-HsRad51 antibody and monoclonal antibody 7G9 and 71-9A, which recognize the 32 and 70 kDa subunits of HsRPA, respectively. Secondary antibodies (Sigma) used with the polyclonal sera and monoclonal antibodies were goat anti-rabbit conjugated with alkaline phosphatase and goat anti-mouse conjugated with alkaline phosphate, respectively.
Co-immunoprecipitation of HsRad51 and RPA proteins
Purified HsRad51 protein (at 0.2–0.9 µM final concentration) was incubated with 0.1–0.3 µM RPA protein or a mutant form of the protein in 9 µl of buffer R (25 mM HEPES, pH 7.4, 1 mM DTT, 100 µg BSA/ml) for 30 min at 37°C. Then 1 µl anti-HsRad51 affinity-purified rabbit polyclonal antibody (0.7 mg/ml of stock solution) was added. After 30 min incubation at 37°C, we added 20 µl of a 50% slurry of protein A-agarose beads (Pierce Chemical Company), pre-equilibrated in buffer IP (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.02% NP-40). After 60 min incubation at room temperature (with occasional shaking), the beads were washed four times with 400 µl buffer IP. The washed beads were mixed with 10 µl of loading buffer (100 mM Tris-HCl, pH 7.4, 2% SDS, 20% glycerol, 4 mM EDTA, 200 mM β-mercaptoethanol, 40 mM DTT, 0.08% bromophenol blue), and boiled for 5 min, following which 20 µl of the mixture were loaded on a 12% SDS-polyacrylamide gel. After gel electrophoresis, the presence of RPA protein subunits was analyzed by western blotting, using anti-RPA antibodies for specific detection and secondary anti-IgG antibodies conjugated with alkaline phosphatase for visualization.
In control experiments, HsRad51 protein was replaced by an equimolar amount of RecA protein, and anti-RecA antibodies were used for co-immunoprecipitation instead of anti-HsRad51 antibodies.
Interaction of HsRad51 protein with proteins bound to a nitrocellulose membrane
Purified test proteins, diluted in buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 50 µg BSA/ml) were spotted onto a Biotrace NT nitrocellulose membrane (Gelman Sciences). The membrane was dried for 30 min at room temperature and blocked for 1 h in buffer FW (20 mM Tris-HCl, pH 7.5, 60 mM NaCl, 10 mM MgCl2, 0.1 mM EDTA, 5% glycerol, 0.02% NP-40), containing 5% non-fat dry milk. After washing in buffer FW, the membrane was incubated for 1 h at room temperature with 0.5 mg HsRad51/ml in buffer FW, containing 2% BSA. The membrane was then washed extensively in buffer FW, and HsRad51 protein that was retained on the membrane was detected by the use of anti-HsRad51 antibodies.
A protocol for far-western blotting was described earlier (27–29). Briefly, proteins were electrophoresed through a 12% SDS-polyacrylamide gel and transferred to Immobilon-P membrane (Millipore Corp.). The membrane was immersed in 8 M urea and 1% mercaptoethanol in buffer FW. Proteins adsorbed to the membrane were renatured by incubation in 10 sequential 2-fold dilutions of urea in buffer FW. After blocking in buffer FW containing 5% non-fat dry milk, the membrane was incubated for 1 h at room temperature with 0.5 mg HsRad51 protein/ml in buffer FW containing 2% BSA. The bands which retained HsRad51 were detected using anti-HsRad51 antibodies.
Co-immunoprecipitation of RPA and HsRad51
Using co-immunoprecipitation to test for physical interactions, we mixed HsRad51 and RPA prior to adding purified anti-HsRad51 antibodies and binding immune complexes to beads of protein A-agarose. The material that was retained by thoroughly washed beads was recovered by boiling in SDS and was analyzed by gel electrophoresis and western blotting with anti-RPA antibodies (Fig. 1). All three subunits of RPA were found in the immune complexes retained on beads of protein A-agarose (Fig. 1A, lane 1). Retention of RPA required the presence of both HsRad51 and anti-HsRad51 (Fig. 1A, lanes 2 and 3). Under our conditions, almost all HsRad51 was retained by washed beads (data not shown), and we estimate that ∼10% of RPA was retained.
Polyclonal antibody against HsRad51 cross-reacts with HsDmc1, a RecA homolog that is specific to meiotic cells (data not shown). The substitution of HsDmc1 for HsRad51 also resulted in co-immunoprecipitation of RPA (Fig. 1A, lane 5).
As one test of the specificity of the interactions with RPA, we substituted Escherichia coli RecA protein for HsRad51 and anti-RecA antibody for anti-HsRad51 antibody in the immunoprecipitation. Although antibody to RecA led to almost complete adhesion of RecA to protein A-agarose beads (data not shown), co-immunoprecipitation of RPA by RecA was undetectable (Fig. 1B, lane 6).
The ability of RPA to bind to HsRad51 protein was confirmed by experiments in which RPA was immobilized on a nitrocellulose membrane and subsequently incubated with HsRad51 protein. Retention of the HsRad51 on the washed membrane was analyzed using anti-HsRad51 antibody. As a negative control, we used E.coli ssDNA binding protein (SSB), and as a positive control we used human Rad52 protein, which has been observed to bind to HsRad51 protein (30). By this assay, HsRad51 bound to RPA and to HsRad52, but not to E.coli SSB (Fig. 1E).
Since both HsRad51 and RPA bind to ssDNA, it was necessary to rule out the possibility that the observed HsRad51-RPA association was mediated by contaminant DNA in the enzyme preparations. When we applied the protocol for labeling 34 termini of DNA by terminal transferase (31), we failed to detect any radioactive DNA (<1 pM) in the enzyme preparations that were used in our experiments.
Interaction with HsRad51 is mediated by the 70 kDa subunit of RPA
The interaction of RPA and HsRad51 was further analyzed by the far-western protocol. Equimolar amounts of purified RPA and SSB proteins were electrophoresed through SDS-polyacrylamide gel and transferred to a membrane. Proteins bound to the membrane were renatured in situ by incubation of the membrane in solutions with gradually decreasing concentrations of urea. The membrane was incubated with a solution of HsRad51 protein, and following extensive washing, any retained HsRad51 was detected by use of anti-HsRad51 antibodies. As indicated in Figure 1F, HsRad51 bound to the 70 kDa subunit of RPA protein. Binding was not detected to the other subunits of RPA nor to E.coli SSB. This observation suggests that the interaction of HsRad51 with RPA is mediated by the largest subunit of RPA.
Mapping the interacting domain of the 70 kDa subunit of RPA
To locate the domain of the 70 kDa subunit that interacts with HsRad51, we examined co-immunoprecipitation of HsRad51 with mutant forms of HsRPA (Fig. 1D; 26). We found that both mutant RPA·70ΔN168, which has lost 168 amino acids from the N-terminus of the 70 kDa subunit, and mutant RPA70ΔC327, which consists of 326 N-terminal amino acids of the 70 kDa polypeptide and which lacks the 32 and 14 kDa subunits, co-immunoprecipitated with HsRad51 (Fig. 1C, lanes 4 and 8). On the other hand, mutant RPA·70ΔN236, which has lost 236 amino acid residues from its N-terminus, did not co-immunoprecipitate efficiently with HsRad51 (Fig. 1C, lane 6). These observations indicate that residues between 168 and 327 of the 70 kDa subunit of the RPA are important in the trimeric protein for interaction with HsRad51. The same domain has been implicated in interactions with several other proteins (26).
The ability of mutant RPA70DC327 to interact with HsRad51 indicates that such interaction does not require 32 and 14 kDa subunits. Furthermore, mutant RPA·70DN236, which carries intact 14 and 32 kDa subunits and a deleted form of the 70 kDa subunit, does not interact with HsRad51. Nevertheless, we found that mutant RPA32·14, which consists of only the 14 and 32 kDa subunits, co-immunoprecipitated with HsRad51 (Fig. 1C, lane 9). This additional interaction with the 32 or 14 kDa subunit was apparent only when the 70 kDa subunit was absent from the trimeric complex.
Co-localization of HsRad51 and RPA proteins in mammalian cells
The finding of a direct interaction between HsRad51 and RPA proteins in vitro prompted us to compare the intracellular localization of the two proteins. Consistent with our earlier observations (10), we found that HsRad51 protein concentrated in multiple discrete foci in the nucleoplasm of cultured mouse fibroblasts, although only in 1% of cells. However, after γ-irradiation of cells with doses of 5–10 Gy, the number of cells with focally concentrated HsRad51 protein increased to >30% (Table 1). In the untreated exponentially growing cultures, 22% of nuclei showed focally concentrated RPA protein, whereas the majority of cells displayed a more or less uniform nuclear immunofluorescence of weak to moderate intensity. After γ-irradiation, the percentage of cells with RPA foci increased to >50%. Thus the absolute increases in the fraction of HsRad51-positive cells and RPA-positive cells were similar, and virtually all of the cells with foci of HsRad51 had foci of RPA, both of which were attributable to irradiation (Table 1). Moreover, within 24 h after irradiation, HsRad51 and RPA co-localized in most of the cells that had foci of both proteins (Table 1 and Fig. 2). This association between HsRad51 and RPA occurred in a timedependent manner. The highest number of HsRad51-positive cells as well as the highest number of co-localizations were observed 1 day after DNA damage. After 2 days, the number of HsRad51 positive cells showing co-localization with RPA foci dropped from nearly 100 to 50% (Table 1).
Additional observations indicated that focal co-localization of HsRad51 and RPA was mediated by DNA damage. Focal concentration of HsRad51 was also induced by treatment of cell cultures with the aneuploidogen colcemid, which does not introduce significant DNA damage (T.Haaf, unpublished observations), but these colcemid-induced HsRad51 foci did not co-localize with RPA foci (data not shown).
We conclude that γ-irradiation leads to the concentration of HsRad51 and RPA in common foci.
SSB, the ssDNA-binding protein from E.coli, is an essential protein that plays crucial roles in DNA replication, DNA repair and homologous recombination. This protein removes secondary structure from ssDNA, and by doing so facilitates the activity of a variety of enzymes of DNA metabolism (for review see 32).
RPA, the eukaryotic counterpart of the SSB, is also implicated in various aspects of cellular metabolism. Unlike SSB, which is composed of identical 20 kDa subunits, RPA is a stable heterotrimeric complex of 70, 32 and 14 kDa subunits. Its impact on the replication, repair and recombination of DNA can be explained only partly by its role in removing the secondary structure of ssDNA, since it also interacts with a variety of proteins through which it exerts its vital effects (reviewed in 15).
Using co-immunoprecipitation, we found that RPA interacts with HsRad51 as well as with Dmc1, its meiotic homolog. The interaction of RPA with HsRad51 was also confirmed by a blotting assay. As expected, RPA did not co-immunoprecipitate with RecA, and according to blotting experiments, E.coli SSB protein did not interact with Rad51. At a ratio of two to three molecules of HsRad51 per molecule of RPA heterotrimer, the complete precipitation of HsRad51 was accompanied by the precipitation of only 10% of the input RPA. This apparently weak interaction of RPA with HsRad51 was also reflected in an ELISA which indicated that the interaction of RPA with HsRad52 was stronger than that with Rad51 (data not shown).
In experiments with mutant forms of RPA protein, we found that a region of the 70 kDa subunit of RPA that included residues 169–326 is necessary for the interaction of RPA with Rad51. The same domain of the 70 kDa subunit has been implicated previously in the interactions of RPA with several other proteins (26).
In the G1-phase of the cellular cycle, RPA is dispersed throughout nuclei, as shown by immunostaining. By contrast, in 60–70% cells in S-phase, RPA is rearranged in a punctate pattern, or in a mixed punctate and dispersed pattern. The punctate appearance of RPA in S-phase is presumably due to its localization in replication foci (33–38). In agreement with the observations just cited, we found focally concentrated RPA in 22% of non-synchronized cells. We assume that those foci represent sites where chromosomal replication takes place during S-phase. After γ-irradiation, however, the percentage of cells with RPA foci was increased to 52%. Earlier we had found that after DNA damage HsRad51 protein changes its nuclear distribution and moves to distinct areas, forming foci (10). Recently, we found that the majority of these foci contained a region of ssDNA, which appeared in cells as a result of DNA damage. We postulate that these foci are nucleated by ssDNA where Rad51 forms filaments and mediates recombinational repair (E.Raderschall, E.I.Golub and T.Haaf, submitted for publication). The observations reported here show that these damage-induced Rad51 foci also contain RPA and thus provide further evidence for a role of RPA in recombinational repair. Mammalian RPA and Rad51 have also been shown to co-localize on synapsed axes in meiosis, suggesting an interaction of RPA with Rad51 in meiotic recombination (20).
In the recombination system of S.cerevisiae, RPA has been shown to stimulate homologous pairing and strand exchange promoted by Rad51 (21,22). However, RPA actually inhibits the reaction unless it is added after Rad51 has formed a filament on ssDNA (39,40). Other observations suggested that RPA acts by removing secondary structure from ssDNA, an interpretation supported by the further observation that E.coli SSB could replace RPA in reactions with Rad51 (22). This non-specific aspect of the action of RPA, however, does not exclude other specific interactions. Rad52, which interacts in vivo with Rad51 (41), is able to overcome the inhibition caused by RPA if the Rad51 filament has not already been formed (39,40,42). Rad52 is also able to promote the annealing of complementary strands (43). RPA, which interacts with Rad52 in vivo (16), enables Rad52 to promote the annealing of kb-lengths of ssDNA, and in this function, neither human RPA nor E.coli SSB can replace yeast RPA (44).
In both prokaryotes and eukaryotes, 3′ ends play a favored role in the initiation of recombination (46–49). According to the double-strand break model of recombinational repair, an invading 34 end provides a primer for new DNA synthesis that is important in completing the repair (50). A similar mechanism operates in late phage T4 replication when recombination initiates the formation of replication forks (51), and a similar process is thought to occur in the repair of broken replication forks by recombination (52–54). Moreover, in mammals, Rad51 plays a vital role as evidenced by the embryonic lethality of homozygous Rad51 null mutations (11,55). Human RPA interacts specifically with viral proteins that initiate DNA replication, such as SV-40 large T-antigen, EBNA1 protein from Epstein-Barr virus and E2 protein from bovine papilloma virus (56–58), as well as with cellular DNA polymerase a and the primase subunit (26,56). The 70 kDa subunit of RPA also interacts with mammalian PCNA, the toroidal sliding clamp of replication (59). The present demonstration of a physical interaction between RPA and Rad51, as well as their co-localization in γ-irradiated cells, is consistent with the hypothesis that RPA creates a bridge between recombination and replication in mammalian cells (23).
We are grateful to Jan Zulkeski for data processing and to Zhufang Li for technical assistance. This research was supported by Grant no. R37-GM33504 from the National Institutes of Health to C.M.R., Research Grant HA1374/4-1 from Deutsche Forschungsgemeinschaft to T.H. and Grant no. GM44721 to M.S.W.