Function of Yeast Rad52 Protein as a Mediator between Replication Protein A and the Rad51 Recombinase*

The RAD51 and RAD52 genes of Saccharomyces cerevisiae are key members of theRAD52 epistasis group required for genetic recombination and the repair of DNA double-stranded breaks. The RAD51encoded product mediates the DNA strand exchange reaction. Efficient strand exchange is contingent upon the addition of the heterotrimeric single-stranded DNA binding factor replication protein A (RPA) after Rad51 has nucleated onto the single-stranded DNA. However, if the single-stranded DNA is incubated with Rad51 and RPA simultaneously to mimic what may be expected to occur in vivo, the efficiency of strand exchange decreases dramatically, revealing an inhibitory effect of RPA that is distinct from its stimulatory function. Interestingly, the inclusion of Rad52 protein, which has been purified in this study from yeast cells, restores the efficiency of strand exchange. Thus, Rad52 functions as a co-factor for the Rad51 recombinase, acting specifically to overcome the apparent competition by RPA for binding to single-stranded DNA.

Saccharomyces cerevisiae genes of the RAD52 epistasis group, including RAD50, RAD51, RAD52, RAD54, RAD55, RAD57, RAD59, MRE11, and XRS2, function in genetic recombination and the recombinational repair of DNA doublestranded breaks induced by ionizing radiation. Because meiotic recombination is required for the proper disjunction of chromosomal homologs during meiosis I, mutants of the RAD52 group often also exhibit severe meiotic abnormalities, including a failure to sporulate and low spore viability (1).
The RAD51 encoded product is structurally related to the Escherichia coli recombination protein RecA (2). Like RecA, Rad51 protein mediates the homologous DNA pairing and strand exchange reaction (3). The efficiency of the Rad51-mediated strand exchange reaction is markedly stimulated by the heterotrimeric ssDNA 1 binding factor RPA (3)(4)(5). However, the manner in which RPA is incorporated is critical for obtaining maximal stimulation, such that if the ssDNA substrate is incubated with Rad51 and RPA simultaneously, the extent of ensuing strand exchange is only a fraction of what is seen when the ssDNA is first incubated with Rad51 prior to the incorporation of RPA (6). These results suggest that, although it is an important accessory factor for the Rad51 recombinase activity, RPA can also compete with Rad51 for binding sites on the ssDNA and thus reduce the efficiency of strand exchange (6). That RPA competes with Rad51 protein for sites on ssDNA has also been inferred from the observation that an excess of RPA inhibits the ssDNA-dependent ATPase activity of Rad51 protein (5).
Here Rad52 protein is expressed in yeast and purified to near homogeneity. It is demonstrated that inclusion of Rad52 protein in the strand exchange reaction alleviates the inhibition by RPA, providing evidence for a co-factor function of Rad52 protein in the Rad51-catalyzed DNA strand exchange reaction.

MATERIALS AND METHODS
Polyclonal Antibodies-The portion of Rad52 protein encompassing amino acid residues 168 -456 was expressed as a fusion protein with the E. coli transcriptional terminator rho (). The insoluble -Rad52 fusion protein was purified by preparative SDS-polyacrylamide gel electrophoresis and used for polyclonal antiserum production in rabbits. Antibodies were affinity-purified from the antiserum as described (7). Antibodies against Rad51 protein were prepared as described (3), and the preparation of anti-Rad14 antibodies (8) was kindly provided by Dr. Sami Guzder.
Immunoprecipitation-For immunoprecipitation, extract was prepared in cell breakage buffer (50 mM Tris-HCl, pH 7.5, 10% sucrose, 2 mM EDTA, 300 mM KCl, 2 mM DTT) with protease inhibitors at 2 ml of buffer/gram of cells using a French press (6). The extract was clarified by ultracentrifugation (100,000 ϫ g, 90 min) and dialyzed against buffer I (25 mM Tris-HCl, pH 7.5, 10% glycerol, 0.2 mM EDTA, and 1 mM DTT) containing 150 mM KCl and protease inhibitors. After centrifugation (100,000 ϫ g, 90 min), the clarified dialysate (0.5 ml) was mixed at 4°C for 4 h with 10 l of protein A beads bearing affinity-purified anti-Rad14 antibodies, anti-Rad51 antibodies, and anti-Rad52 antibodies, all at 2 mg antibodies/ml matrix. The beads were washed once each with 300 l of buffer I containing 150 mM KCl, buffer I containing 250 mM KCl, and buffer I alone and then incubated with 30 l of 3% SDS at 37°C for 10 min to elute bound proteins. An aliquot of the eluates (5 l) was subjected to immunoblot analysis to determine their content of the Rad51 and Rad52 proteins.
Rad52 Protein Purification-RAD52 gene from nine nucleotides upstream of the first translation initiating ATG codon until 600 nucleotides downstream of the TGA translation stop codon was placed under the control of the phosphoglycerate kinase (PGK) promoter, yielding plasmid pR52.1 (2, PGK-RAD52). This plasmid was introduced into yeast strain LP2749-9B harboring the RAD51 overexpressing plasmid pR51.1 (3). For the purification of Rad52 protein, extract was prepared from 400 g of frozen yeast paste (7) and subjected to the purification scheme described in Fig. 2B. The yield of Rad52 protein was about 100 g, and the full purification details will be described elsewhere. 2 Other Proteins-Rad51 protein was purified from yeast strain LP2749-9B harboring plasmid pR51.3 (2, PGK-RAD51) as described (9). RPA was purified from strain LP2749-9B as described (6).
Preparation of Affi-Gel Rad51-Rad51 protein (2 mg) and bovine serum albumin (BSA, 3 mg) in 1 ml of coupling buffer (0.1 M potassium * This work was supported by Public Health Service Grant RO1ES07061 from the National Institute for Environmental Health Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  MOPS, pH 7.5) were mixed with 0.5 ml of Affi-Gel 15 beads (Bio-Rad) at 4°C for 5 h. The coupling efficiencies were determined by SDS-polyacrylamide gel electrophoresis, and the final concentrations of the immobilized ligands were 3.2 mg/ml of Rad51 and 5 mg/ml of BSA. The Affi-Gel matrices were stored at Ϫ20°C in 20 mM Tris-HCl, pH 7.5, 0.2 mM EDTA, 1 mM DTT, and 50% glycerol.
Binding of Rad52 Protein to Affi-Gel Rad51-Purified Rad52 protein (1 g) was diluted to 200 l with buffer T (10 mM Tris-HCl, pH 7.5, 10% glycerol, 0.2 mM EDTA, 0.5 mM DTT, and 0.01% Nonidet P-40) containing 150 mM KCl and 100 g/ml BSA, and 150 l of this solution was mixed with 7.5 l of Affi-Gel 15 beads containing covalently conjugated BSA or Rad51 at 25°C for 45 min. The beads were then spun down in a microcentrifuge and washed at 4°C once with 150 l each of buffer T containing 150 mM KCl, buffer T containing 300 mM KCl, and buffer T alone. The beads were treated with 30 l 3% SDS at 37°C for 10 min to elute proteins, and 3 l of the eluates along with 15 l of the starting material, 15 l of the supernatant containing unbound Rad52, and 15 l the two KCl washes were subjected to immunoblot analysis to determine their content of Rad52 protein.
Strand Exchange Reactions-In the experiment shown in Fig. 3B, the reaction (final volume, 50 l) was assembled by mixing 25 g of Rad51 protein (11.6 M) added in 4 l of storage buffer with 2.4 l of buffer K containing 250 mM KCl (used for Rad52 storage) and 520 ng of X174 viral (ϩ) strand (32 M nucleotides) added last in 4 l of TE (10 mM Tris-HCl, pH 7.5, 0.2 mM EDTA) in 40 l of buffer R (35 mM potassium MOPS, pH 7.2, 25 mM KCl, 3 mM ATP, 3 mM MgCl 2 , 1 mM DTT). After a 5-min incubation at 37°C, 8 g of RPA (1.35 M) in 2 l of storage buffer was added, followed by a 5-min incubation at 37°C, then 520 ng of X174 dsDNA (16 M base pairs) was added in 4 l of TE, and 4 l of 50 mM spermidine were incorporated. The complete reaction mixture was incubated at 37°C, and 5.5-l portions were withdrawn at the indicated times and processed for agarose gel electrophoresis as described (4). The reaction that had no RPA (Fig. 3B) was assembled and incubated in exactly the same way, except that 2 l of storage buffer that did not contain RPA was added after preincubation of Rad51 protein with the viral (ϩ) strand.
Effect of Rad52 Protein on Strand Exchange-In the experiment in

RESULTS AND DISCUSSION
As seen in immunoblot analyses, Rad52 protein in yeast extract consists of a number of closely spaced bands with sizes ranging from 61 to 63 kDa ( Figs. 1 and 2). The abundance of these Rad52 protein species increases with elevated expression of RAD52, as when extract from yeast cells harboring the overproducing plasmid pR52.1 (2, PGK-RAD52) was analyzed (see Fig. 2A). The gel mobility of these Rad52 species was not altered by treatment with calf intestinal alkaline phosphatase, suggesting that the difference in gel mobility of these Rad52 species is not due to phosphorylation. The RAD52 protein coding frame within the first 40 amino acid residues contains five potential translation initiating ATG codons; it is possible that the multiple Rad52 species are the result of alternate ATG codons being used in the translation of the RAD52 encoded message.
Immunoprecipitation experiments using protein A-agarose beads bearing covalently conjugated anti-Rad51 and anti-Rad52 antibodies were carried out to determine whether Rad52 protein in cell extract is physically associated with Rad51 protein. As shown in Fig. 1A, anti-Rad51 immunobeads precipitated, in addition to Rad51, also Rad52 protein, and likewise, anti-Rad52 immunobeads co-precipitated Rad51 protein. The amount of Rad52 protein that co-precipitated with Rad51 protein was very similar to that directly precipitated by its cognate antibodies, suggesting that a substantial portion of the cellular Rad52 protein exists as a complex with Rad51 protein. Interestingly, the amount of Rad51 protein that co-precipitated with Rad52 protein was less than 10% of that precipitated by anti-Rad51 immunobeads, suggesting that Rad51 is present in considerable excess over Rad52 protein in yeast cells. Consistent with this deduction, the amount of Rad51 protein that coprecipitated with Rad52 increased with overexpression of the Rad52 protein, as when extract from LP2749-9B harboring pR52.1 (2, PGK-RAD52) was used for immunoprecipitation. As a result of a 20-fold overproduction of Rad52 protein, approximately 40% of the Rad51 protein in cell extract became associated with the Rad52 protein (Fig. 1A). The association of Rad52 with Rad51 in cell extract was likely due to a direct interaction between the two proteins, because purified Rad52 bound to purified Rad51 immobilized on Affi-gel 15 beads (Fig. 1B). Interestingly, the level of Rad52 protein in strain LP2749-9B harboring pR52.1 was enhanced about 2.5-fold upon the introduction of the RAD51 overexpressing plasmid pR51.1 (2, ADCI-RAD51; Ref. 3). Because Rad51 and Rad52 proteins interact (Refs. 2 and 10 and see Fig. 1), it appears that interaction of Rad52 with Rad51 results in stabilization of the former in yeast cells. For the purification of Rad52 protein, extract from strain LP2749-9B co-harboring pR51.1 and pR52.1 was subjected to the chromatographic procedure outlined in Fig.  2B. Rad52 protein from the last step of purification in Mono Q was nearly homogeneous (Fig. 2C) and was used in the studies below.
In the earliest or the presynaptic phase of the in vitro homologous pairing and strand exchange reaction (Refs. 11 and 12 and Fig. 3A), Rad51 protein polymerizes on the ssDNA substrate to form a nucleoprotein filament, within the confines of which the ensuing pairing and strand exchange occur (4). Previous work has indicated that the most efficient pairing and strand exchange is effected by incubating ssDNA during the presynaptic phase, first with Rad51 protein alone for a few minutes before incorporating RPA (3,4). In this standard strand exchange reaction, about 60% (11% joint molecules and 49% nicked circular duplex) and 87% (12% joint molecules and 75% nicked circular duplex) of the input linear dsDNA had been converted into strand exchange products after 36 and 72 min of incubation, respectively. As reported previously (3,5) and reiterated in Fig. 3 (B and D), RPA plays an important accessory role in the strand exchange reaction, because in its absence, the efficiency of pairing and strand exchange decreased to a point where only 6% of the input linear dsDNA had been converted to joint molecules and Յ1% to the full strand exchange product, nicked circular duplex, by the reaction end point of 72 min.
When RPA was added together with Rad51 protein to the ssDNA during presynapsis, the level of ensuing pairing and strand exchange was only a fraction of what could be achieved in the standard reaction. Specifically, the level of reaction products after 60 min of incubation was only 9% (4% nicked circular duplex) of the input linear duplex (Fig. 3, C and D), as compared with about 80% conversion into products after the same incubation time in the standard reaction (Fig. 3, B and  D). These results indicated that RPA can also exert a negative effect on DNA strand exchange and suggested that other RAD52 group proteins may function to neutralize the inhibition by RPA. I therefore examined whether purified Rad52 protein was capable of alleviating the inhibition by RPA. Because the results from the immunoprecipitation experiments summarized earlier (Fig. 1A) revealed that Rad52 protein is of a much lower cellular abundance than Rad51 protein, initially, a molar amount of Rad52 protein about one-tenth that of Rad51 protein was used in the strand exchange reaction. Interestingly, the inclusion of Rad52 protein (1.25 M) with Rad51 (11.6 M) and RPA (1.35 M) resulted in marked stimulation (Fig. 4,  A and B), restoring strand exchange to a level comparable with that seen in the standard reaction (Fig. 3, B and D). For instance, at the mid-point of the reaction (36 min of incubation), the amount of strand exchange products was 56% (44% nicked circular duplex) of the input linear dsDNA when Rad52 was present (Fig. 4, A and B), whereas only 7% of the input dsDNA (3% nicked circular duplex) had been converted to products in its absence (Fig. 3, C and D).
As shown in Fig. 4A (lanes 1-3), in the absence of RPA, the level of pairing and strand exchange obtained with Rad51 and Rad52 proteins was essentially the same as the extremely low level seen with Rad51 alone (Fig. 3C), which suggested that Rad52 does not function to replace RPA. Strand exchange enhancement was also examined as a function of Rad52 protein concentration, and it was found, in reactions containing 11.6 M Rad51 and 1.35 M RPA, that the maximal level of stimulation occurred at about 1.0 M Rad52 protein (Fig. 4C).
The possibility that Rad52 might have pairing and strand exchange activity was also examined. However, in the absence of Rad51 protein, there was no evidence of homologous pairing, even at a Rad52 protein concentration (5 M) much higher than those used in the experiments shown in Fig. 4 and regardless of whether or not RPA was added to the reaction during or after incubation of Rad52 with the ssDNA (data not shown).
In its role as a co-factor for the Rad51 recombinase, Rad52 protein resembles the heterodimer of the Rad55 and Rad57 proteins, which is also capable of overcoming the inhibition by RPA (6). Interestingly, the recombination defects in rad55 and rad57 mutants can be suppressed partially by introducing a multicopy plasmid that contains the RAD52 gene (13), suggesting that a common biochemical function exists in the Rad55-Rad57 heterodimer and Rad52 protein. Our results now provide evidence that this common function may be an ability of these protein factors to facilitate Rad51-ssDNA nucleoprotein assembly in the presence of RPA. The function of Rad52 protein and the Rad55-Rad57 heterodimer in the Rad51-mediated strand exchange reaction is reminiscent of that described for the bacteriophage T4 UvsY protein (Refs. 14 and 15 and references therein) and the combination of the E. coli RecO and RecR proteins (16), which act to allow their cognate recombinases UvsX protein and RecA protein to efficiently utilize ssDNA coated with T4 gene 32 protein and E. coli SSB protein for pairing and strand exchange. Both Rad52 protein and the Rad55-Rad57 heterodimer bind ssDNA (6,17) and interact physically with Rad51 protein (2, 6, 10, 13, 18), suggesting a mechanism by which ssDNA-bound Rad52 or Rad55-Rad57 heterodimer actively recruits Rad51 to the ssDNA substrate.