Homologous pairing promoted by the human Rad52 protein.

The Rad52 protein, which is unique to eukaryotes, plays important roles in the Rad51-dependent and the Rad51-independent pathways of DNA recombination. In the present study, we have biochemically characterized the homologous pairing activity of the HsRad52 protein (Homo sapiens Rad52) and found that the presynaptic complex formation with ssDNA is essential in its catalysis of homologous pairing. We have identified an N-terminal fragment (amino acid residues 1-237, HsRad52(1-237)) that is defective in binding to the human Rad51 protein, which catalyzed homologous pairing as efficiently as the wild type HsRad52. Electron microscopic visualization revealed that HsRad52 and HsRad52(1-237) both formed nucleoprotein filaments with single-stranded DNA. These lines of evidence suggest the role of HsRad52 in the homologous pairing step of the Rad51-independent recombination pathway. Our results reveal the striking similarity between HsRad52 and the Escherichia coli RecT protein, which functions in a RecA-independent recombination pathway.

The ability of cells to repair double strand breaks that occur on chromosomal DNA is critical to maintain the integrity of their genomic DNA. Double strand breaks can result from several events: ionizing radiation, DNA-damaging agents, replication errors, and specific enzymes that act in meiosis, mating-type switching, and V(D)J recombination (1)(2)(3)(4)(5). To repair such lesions, cells have developed homologous recombination (HR) 1 and nonhomologous DNA end joining. HR is an important process for the preservation of the DNA sequence in chromosomal DNA. The homologous pairing step of the HR pathway, which is the process of searching for sequence homology in either homologous chromosomes or sister chromatids, is an essential step. Therefore, the identification of the enzymes that promote homologous pairing has been one of the central drives toward an understanding of the HR pathway (6,7).
In Escherichia coli, two homologous pairing enzymes, the RecA and RecT proteins, have been reported (8 -10). RecA and RecT, which have no amino acid sequence homology, catalyze homologous pairing in separate recombinational repair pathways: the RecBCD and RecF pathways for RecA and the RecE pathway for RecT. (11)(12)(13). In eukaryotes, two RecA homologues, Rad51 and Dmc1, have been identified, and both proteins promote homologous pairing (14 -19). No proteins with significant sequence homology to RecT have been found in eukaryotes, although recombination pathways that are independent of Rad51 have been identified (20,21). An electron microscopic visualization showed that RecT forms a heptameric or octameric ring structure (22). Interestingly, the ScRad52 (Saccharomyces cerevisiae Rad52) and HsRad52 (Homo sapiens Rad52) proteins also form ring structures similar to that of RecT (23,24). RAD52 genes have been identified in several eukaryotes, including yeast, chicken, mouse, and human (25)(26)(27). The role of Rad52 in the Rad51-dependent recombination pathway has been extensively studied both in vivo (28,29) and in vitro. The ScRad52 and HsRad52 proteins both bind single-stranded DNA (ssDNA) and promote annealing of complementary ssDNA molecules (23, 30 -35), suggesting a role of Rad52 in the single-stranded annealing (36). In addition, ScRad52 and HsRad52 both reportedly enhance the joint molecule formation between circular ssDNA and linear double-stranded DNA (dsDNA) that is promoted by the ScRad51 (S. cerevisiae Rad51) and HsRad51 (H. sapiens Rad51) proteins (32,(37)(38)(39). Therefore, Rad52 has been proposed to be a mediator protein that binds ssDNA and facilitates the loading of Rad51 onto the recombination site (36,40,41). In addition to its role in the Rad51-dependent recombination pathway, the role of Rad52 in the Rad51-independent recombination pathways has also been studied in vivo (42)(43)(44)(45). However, its precise molecular mechanism is not well understood. Interestingly, HsRad52 itself promotes homologous pairing between ssDNA fragments and superhelical dsDNA substrates (46). This activity, along with the single-stranded annealing, may provide the framework for the role of Rad52 in such a pathway.
In the present study, the homologous pairing activity of HsRad52 has been biochemically characterized. HsRad52 showed maximal homologous pairing activity at a concentration that just saturated the ssDNA. The HsRad52-ssDNA complex formation is a critical step in the homologous pairing reaction, because an incubation of HsRad52 with dsDNA before the addition of ssDNA inhibited the reaction. The conserved N-terminal domain of HsRad52, which was revealed by a limited proteolysis, formed a nucleoprotein filament with ssDNA and catalyzed homologous pairing with an efficiency similar to that of the wild type HsRad52 protein.

EXPERIMENTAL PROCEDURES
Overexpression and Purification of HsRad52 and HsRad52 1-237 -HsRad52 and HsRad52 1-237 were both purified in a three-step procedure involving Ni-NTA-agarose purification, the removal of the hexahistidine tag, and heparin-Sepharose purification, to facilitate comparison of the activities. MonoS column chromatography was used to concentrate the HsRad52 protein. Both proteins were cloned into the T7 polymerase expression vector pET-15b (Novagen). The proteins were overexpressed in the E. coli strain JM109 (DE3) along with E. coli tRNA Arg3 and tRNA Arg4 , which recognize the CGG and AGA/AGG codons, respectively. Expression levels of both proteins were extremely low without the inclusion of the plasmid containing the tRNA genes. For an average purification, HsRad52 and HsRad52 1-237 were purified from 5-and 1-liter cultures, respectively. All cultures were incubated at 30°C. At the logarithmic phase of growth (A 600 ϭ 0.6), protein expression was induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside (final concentration). The cells were harvested after 4 -6 h of culture and were lysed by sonication in buffer A (pH 7.8) containing 50 mM Tris-HCl, 0.3 M KCl, 2 mM 2-mercaptoethanol, 10 mM imidazole, 10% glycerol, and protease inhibitors (Complete EDTA-free; Roche Molecular Biochemicals) on ice. All manipulations after harvesting were done at 4°C. The lysates were centrifuged at 27,700 ϫ g for 20 min, and the supernatants were gently mixed by the batch method with 4 ml of Ni-NTA agarose beads (Qiagen) for 1 h. The HsRad52-coupled Ni-NTAagarose beads were packed into Econo-columns (Bio-Rad) and were washed with 30 column volumes of buffer B (pH 7.8), which contained 50 mM Tris-HCl, 0.3 M KCl, 2 mM 2-mercaptoethanol, 50 mM imidazole, and 10% glycerol at a flow rate of about 0.3 ml/min. The HsRad52 proteins were eluted in a 30-column volume linear gradient from 50 to 300 mM imidazole. Peak fractions, which eluted at about 0.15 M imid-azole, were collected, and 2 units of thrombin protease (Amersham Pharmacia Biotech)/mg of HsRad52 protein were added to remove the hexahistidine tag. Fractions were immediately dialyzed against buffer C (pH 7.5), which contained 50 mM Tris-HCl, 0.2 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, and 10% glycerol, for more than 12 h. After the removal of the hexahistidine tag, the HsRad52 proteins were loaded onto a 4-ml Heparin-Sepharose column (Amersham Pharmacia Biotech). The column was washed with 20 column volumes of buffer C, and the proteins were eluted with a 20-column volume linear gradient from 0.2 to 1 M KCl. HsRad52 and HsRad52 1-237 both eluted in sharp peaks at about 0.3 M KCl. The peak fractions of HsRad52 1-237 , which had a concentration of 1-2 mg/ml, were dialyzed against buffer D (pH 7.0), which contained 20 mM Hepes-KOH, 0.2 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, and 50% glycerol, and were stored at Ϫ20°C. For HsRad52, the heparin-Sepharose peak fractions were dialyzed against buffer E (pH 7.5), which contained 20 mM potassium phosphate, 0.2 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, and 10% glycerol, and were applied to a 1-ml MonoS column (Amersham Pharmacia Biotech) for concentration. The MonoS column was washed with buffer E and was eluted in a 20-ml linear gradient from 0.2 to 1 M KCl. The HsRad52 protein sharply eluted at about 0.3 M KCl. Peak fractions were collected, dialyzed against buffer D, and stored at Ϫ20°C. For subsequent experiments, HsRad52 and HsRad52 1-237 were dialyzed against buffer F (pH 7.0), which contained 20 mM Hepes-KOH, 0.2 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, and 10% glycerol. Protein concentrations were determined using the Bio-Rad protein assay kit with bovine serum albumin (Pierce) as the standard.
Overexpression and Purification of HsRad51-The HsRad51 expression vector was constructed as described (46). Hexahistidine-tagged HsRad51 was overexpressed in E. coli strain JM109 (DE3) along with FIG. 1. Purified HsRad52. A, SDS-PAGE of the column fractions containing HsRad52. The peak fraction of Ni-NTAagarose (lane 2, 2 g), the fraction after the removal of the hexahistidine tag (lane 3, 2 g), and the peak fraction of MonoS (lane 4, 2 g) were analyzed on a 15-25% gradient SDS-polyacrylamide gel and stained with Coomassie Brilliant Blue. B, limited proteolysis of HsRad52. The protein was treated with 30 g/ml proteinase K at 25°C for 4 -6 h, and the proteolytic fragments were fractionated on a 12% SDS-PAGE. The amino acid residues of the two major proteolytic fragments, which are indicated on the right of the panel, were revealed by the combination of N-terminal amino acid sequencing and mass spectrometry. C, SDS-PAGE of the purified HsRad52 1-237 protein. HsRad52 1-237 was purified using the purification scheme for HsRad52. D, gel filtration analysis of HsRad52 and HsRad52 1-237 . HsRad52 was excluded from the Superdex 200 HR 10/30 gel filtration column. In contrast, HsRad52 1-237 eluted from the column with an estimated molecular mass of 300 kDa. the tRNA described above. The culture conditions for HsRad51 were identical to those for HsRad52. The HsRad51 protein was purified from a 10-liter LB culture. Harvested cells were lysed by sonication in buffer G (pH 8.0), which contained 50 mM Tris-HCl, 0.5 M NaCl, 5 mM 2-mercaptoethanol, 10 mM imidazole, 10% glycerol, and protease inhibitors (Complete EDTA-free; Roche Molecular Biochemicals). Lysates were mixed gently by the batch method with Ni-NTA-agarose beads at 4°C for 1 h. The HsRad51-coupled Ni-NTA-agarose beads (4 ml) were then packed into an Econo-column (Bio-Rad) and were washed with 30 column volumes of buffer H (pH 8.0), which contained 50 mM Tris-HCl, 0.5 M NaCl, 5 mM 2-mercaptoethanol, 60 mM imidazole, and 10% glycerol, at a flow rate of about 0.3 ml/min. Hexahistidine-tagged HsRad51 was eluted in a 30-column volume linear gradient from 60 to 400 mM imidazole. HsRad51, which eluted in a broad peak, was collected, and was treated with 1 unit of thrombin protease (Amersham Pharmacia Biotech)/mg of HsRad51. The HsRad51 protein was immediately dialyzed against buffer I (pH 8.0), which contained 50 mM Tris-HCl, 0.2 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, and 10% glycerol, at 4°C. The HsRad51 protein precipitated overnight but redissolved after changing the dialysis buffer the following day. After more than 24 h of dialysis, the HsRad51 protein was collected and filtered to remove residual precipitates. About 20 mg of HsRad51, which was more than 99% pure as judged by SDS-PAGE and Coomassie Brilliant Blue staining, were then applied to a 1-ml MonoQ column (Amersham Pharmacia Biotech) for concentration. The MonoQ column was washed with 20 ml of buffer I and was eluted with a 20-ml linear gradient from 0.2 to 0.6 M KCl. The HsRad51 protein sharply eluted between 0.3-0.4 M KCl, and peak fractions had concentrations of about 2 mg/ml. These fractions were dialyzed against buffer J (pH 8.0), which contained 20 mM Tris-HCl, 0.1 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, and 10% glycerol, and were used for subsequent studies. The protein concentrations were determined using the Bio-Rad protein assay kit with bovine serum albumin (Pierce) as the standard.
DNA Substrates-To prevent the dsDNA substrates used in the D loop formation from undergoing irreversible denaturation, alkaline treatment of the cells harboring the plasmid DNA was avoided. Instead, the cells were gently lysed using sarkosyll, as described (47). pGsat4 was created by inserting a 198-base pair fragment of the human ␣-satellite sequence into the pGEM-T Easy vector (Promega). The resulting 3.2-kilobase plasmid DNA and the ⌽X174 form I DNA were both prepared following the method described above.
For the ssDNA substrates, the following high pressure liquid chromatography-purified oligonucleotides were purchased from Roche Molecular Biochemicals: SAT-1 (50-mer, 5Ј-ATT TCA TGC TAG ACA GAA GAA TTC TCA GTA ACT TCT TTG TGC TGT GTG TA-3Ј) and ⌽X-1 (50-mer, 5Ј-ATT TTG TTC ATG GTA GAG ATT CTC TTG TTG ACA TTT TAA AAG AGC GTG GA-3Ј). The 5Ј ends of the oligonucleotides were labeled with T4 polynucleotide kinase (New England Biolabs) in the presence of [␥-32 P]ATP at 37°C for 90 min. Labeled oligonucleotides were purified with Chromaspin-10 columns (CLONTECH). The TE buffer in the spin columns was exchanged with H 2 O prior to purification of the oligonucleotides. This allowed accurate absorbance readings of the purified oligonucleotides at 260 nm for concentration determination.
Limited Proteolysis of HsRad52-HsRad52 (5 l of 1-3 mg/ml) was mixed with 5 l of 30 g/ml proteinase K, and the mixture was incubated at 25°C for 4 -6 h. The proteinase K solution was diluted in 10 mM Tris-HCl (pH 7.5). The reaction mixture was fractionated by 12% SDS-PAGE, and the protein bands were transferred to a polyvinylidene difluoride membrane using a semi-wet blotting apparatus (Bio-Rad). Proteolytic bands were visualized by staining with a 100-fold diluted Coomassie Brilliant Blue staining solution, followed by a brief destaining of the membrane with methanol. Portions of the polyvinylidene difluoride membrane containing the bands were excised for N-terminal amino acid sequencing. For mass determination of the proteolytic fragments, the reaction mixture containing the proteolyzed HsRad52 was treated with 1 l of 2 N HCl to stop further proteolysis by proteinase K. The mixture was diluted 100-fold with H 2 O and was subjected to mass spectrometry.
Assay for D-loop Formation-To maintain the activity of the HsRad52 proteins, the desired concentrations of the  1 and 2). In the presence of HsRad52 (1 M), D-loops were observed in homologous combinations that have migration distances corresponding to the length of the dsDNA (lanes 3 and 4). HsRad52 was unable to catalyze any D-loops between heterologous combinations of the DNA substrates (lanes 5 and 6). tion, the reaction was stopped by adding 1 l of 5% SDS, followed by immediately adding 1 l of 6 mg/ml proteinase K (Roche Molecular Biochemicals). The reaction mixtures were further incubated at 37°C for 15 min. After adding the 6-fold loading dye (15% Ficoll, 0.1% bromphenol blue, 0.1% xylene cyanole), the reaction mixtures were subjected to 1% agarose gel electrophoresis (SeaKem GTG-agarose) in 0.5ϫ TBE buffer. Electrophoresis was run at 3.3 V/cm at room temperature. The recombination products were visualized by autoradiography of the dried gel. Products and reactants were quantified using a Fuji BAS2500 image analyzer. Because the ssDNA is in excess, the yield of products was expressed as a percentage of the pGsat4 dsDNA incorporated into D-loops.
Assay for ssDNA and dsDNA Binding-The reaction mixture was essentially identical to that used in the assay for the D-loop formation to facilitate comparisons of the results. For the ssDNA binding, one volume of the 32 P-labeled SAT-1 ssDNA (10 M) was mixed with nine volumes of the nonlabeled SAT-1 ssDNA (10 M) to dilute the 32 P label 10-fold. After a 5-min incubation of HsRad52 and SAT-1 ssDNA (1 M), 1 l of a 2% glutaraldehyde solution was added to fix the HsRad52-ssDNA complex. The incubation was continued for 20 min. Afterward, the 6-fold loading dye was added, and the reaction mixtures were subjected to 1% agarose gel electrophoresis in 0.5ϫ TBE buffer. Electrophoresis was run at 3.3 V/cm at room temperature. The resulting complexes were visualized by autoradiography of the dried gel. The amounts of free ssDNA were quantified using a Fuji BAS2500 image analyzer. For the dsDNA binding, HsRad52 was incubated with pGsat4 dsDNA (15 M) for 5 min and subjected to 1% agarose gel electrophoresis without fixation of the complex. The complexes were visualized by ethidium bromide staining (0.5 g/ml) of the gel. The amounts of free dsDNA were quantified using a MacBAS software.
Assay for HsRad52-HsRad51 Interactions-The ability of HsRad52 to bind to Affi-Gel 15-conjugated HsRad51 was observed. An Affi-Gel 15 slurry (250 l; Bio-Rad) was washed two times with 0.5 ml of H 2 O, followed by three washes with 0.5 ml of buffer K (pH 8.0), which contained 20 mM Hepes-KOH, 0.1 M KCl, 0.5 mM EDTA, 2 mM 2-mercaptoethanol, 10% glycerol, and 0.05% Triton X-100. The washed beads were mixed with 0.5 ml of 2.8 mg/ml HsRad51 and were incubated at 4°C for 5 h. Afterward, the supernatant was removed, and the conjugated beads were washed six times with 0.5 ml of buffer L. The final concentration of HsRad51 conjugated to Affi-Gel 15, determined by the amount of unbound HsRad51, was 3.3 mg/ml. The Affi-Gel 15-HsRad51 matrix was adjusted to a 33% slurry with binding buffer and was stored at 4°C.
For the binding assay, 10 g of Rad51 (30 l of the slurry) were mixed with 20 g of HsRad52 and HsRad52 1-237 . The proteins were mixed at room temperature for 90 min. The unbound HsRad52 proteins were then removed, and the Affi-Gel-HsRad51 beads were washed six times with 0.5 ml of buffer L. SDS-PAGE sample buffer (10 l of a 2-fold stock) was mixed directly with the washed beads. After heating the mixture at 98°C for 2 min, the HsRad51 and the bound HsRad52 proteins were fractionated by 15-25% gradient SDS-PAGE. Bands were visualized by Coomassie Brilliant Blue staining.
Size Exclusion Chromatography of HsRad52-A Superdex 200 HR 10/30 gel filtration column (Amersham Pharmacia Biotech) was used to determine the estimated molecular masses of the HsRad52 proteins. All gel filtrations were done in buffer F. For each injection, 0.1-0.2 ml of 1-2 mg/ml HsRad52 was used. To create a standard molecular mass curve, ferritin, aldolase, albumin, ovalbumin, and ribonuclease A (Amersham Pharmacia Biotech) were used.
Electron Microscopic Analysis of HsRad52-HsRad52 and HsRad52 1-237 and their complexes with ⌽X174 circular ssDNA were negatively stained on a copper-plated carbon grid with 2% uranyl acetate. The proteins and complexes were observed with a JEOL JEM 2000FX electron microscope.

RESULTS AND DISCUSSION
Homologous Pairing Activity by HsRad52-We previously reported that HsRad52 catalyzes homologous pairing independently from HsRad51, as observed in the D-loop formation assay, which is a standard homologous pairing assay for RecA (9). However, in those previous experiments, we used dsDNA substrates prepared by a method involving alkaline treatment (46), which may irreversibly denature the double helix of the dsDNA. To exclude the possibility of observing an annealing reaction between a denatured dsDNA and its homologous ssDNA, we prepared the dsDNA by a conventional method that does not involve alkaline treatment (47) and successfully reproduced the formation of D-loops by purified HsRad52 (Figs.  1A and 2). The use of this DNA substrate now allowed us to characterize the homologous pairing activity of HsRad52. In the reaction, a 50-mer oligonucleotide was preincubated with HsRad52, and then the dsDNA was added to initiate the reaction. The migration distance of the reaction product on the agarose gel was identical to that of the D-loop product formed by RecA (Fig.  2B, lanes 3 and 4). When the superhelical tension in the Dloops was released by cutting the dsDNA with restriction enzymes at sites outside the homologous region, the D-loops dissociated by spontaneous branch migration, as in the case of authentic D-loops formed by RecA or a nonenzymatic method (48,49) (Fig. 2B, lanes 5-7). These two results indicate that the D-loop formed by HsRad52 has the same physical characteristics as that formed by RecA.
The HsRad52-catalyzed D-loop formation was a homologydependent reaction. When ⌽X174 dsDNA, which was also prepared by the conventional method, and its homologous ssDNA were used as D-loop substrates, HsRad52 formed D-loops with an efficiency similar to that of the pGsat4 DNA substrates (Fig.  2C, lanes 3 and 4). In contrast, no D-loops were formed between heterologous combinations of the DNA substrates (Fig. 2C,  lanes 5 and 6).
The HsRad52 protein converted more than 10% of the dsDNA into D-loops when HsRad52 was preincubated with ssDNA followed by the addition of dsDNA (Fig. 3, A and C). However, when the dsDNA was preincubated with HsRad52, followed by the addition of ssDNA, no D-loops were observed even after 32 min of incubation (Fig. 3, B and C). Therefore, the HsRad52-dsDNA complex was incapable of forming the D-loop. The D-loop formation catalyzed by the HsRad52-ssDNA complex was dependent on the protein concentrations (Fig. 4, A and  B). The D-loop formation increased with higher concentrations of HsRad52 and reached maximal activity at about a 1:1 stoichiometric ratio of monomeric HsRad52 and nucleotides of the ssDNA substrate. This maximal D-loop formation activity occurred at concentrations of HsRad52 that just saturated the ssDNA (Fig. 4, A-D). Concentrations exceeding this saturation point were inhibitory toward D-loop formation, suggesting that the complex between the excess HsRad52 and the dsDNA inhibited the reaction as in the case of RecT (Fig. 3B and Ref. 50). In fact, HsRad52 bound to superhelical dsDNA at concentrations that inhibited the D-loop formation (Fig. 4, E and F).
These results indicate that the ssDNA, and not the dsDNA, is contained in the active presynaptic complex for homologous pairing. The established homologous pairing proteins, RecA, Rad51, and RecT, also catalyze the homologous pairing reaction by initially forming a complex with ssDNA (50 -53). The presynaptic complex formation with ssDNA appears to be a universally conserved mechanism among homologous pairing proteins.
In eukaryotes, Rad51-dependent and Rad51-independent homologous recombination pathways are both found (54). In the Rad51-dependent pathway, Rad51 is believed to be the key protein that catalyzes homologous pairing. However, it has been reported that the homologous pairing activity of Rad51 is significantly weaker than that of the E. coli RecA protein (55,56). Our results show that HsRad52 directly promotes homologous pairing, which suggests that HsRad52 may cooperate with HsRad51 to promote the homologous pairing reaction in the Rad51-dependent pathway. In the Rad51-independent pathway, HsRad52 may be a key homologous pairing protein. This is consistent with the hypothesis that Rad52 promotes homologous pairing in pathways that are independent of HsRad51, as suggested by the yeast genetic studies (21, 43, 45).
A Conserved N-terminal Domain of HsRad52 Catalyzes Homologous Pairing-The N-terminal region (amino acid residues 1-177) of HsRad52 is highly conserved from the yeast to human Rad52 proteins, and genetic studies have shown that this domain is responsible for the central functions of the yeast Rad52 protein (57,58). Our protease mapping of the domains of HsRad52 revealed that the polypeptide containing amino acid residues 1-237 (HsRad52 1-237 ) was resistant to limited proteolysis by proteinase K (Fig. 1B). Hence, we overexpressed HsRad52 1-237 in E. coli and purified the protein to homogeneity (Fig. 1C) to test its ability to catalyze D-loop formation using the assay established with the wild type HsRad52. Interestingly, HsRad52 1-237 retained the D-loop formation activity with almost the same efficiency as that of the wild type HsRad52. The maximal D-loop formation activity was observed at a 1.25:1 stoichiometric ratio (monomeric HsRad52 1-237 : nucleotides of ssDNA) (Fig. 5, A and B), where HsRad52 1-237 just saturated the ssDNA substrate (Fig. 5, C and D). Similar to the wild type HsRad52, HsRad52 1-237 exhibited a decreased D-loop formation at higher protein concentrations (Fig. 5, A and B), probably because of the dsDNA binding by the excess HsRad52 1-237 (Fig. 5, E and F). Our results strongly suggest that the N-terminal domain is responsible for the homologous pairing activity of HsRad52.
Electron Microscopic Visualization of HsRad52 and HsRad52  and Their Complexes with ssDNA-The HsRad52 and HsRad52 1-237 proteins were analyzed by gel filtration chromatography (Fig. 1D). HsRad52 was excluded from the Superdex 200 HR 10/30 gel filtration column (Amersham Pharmacia Biotech), and the oligomerization state of the protein could not be characterized. However, HsRad52 1-237 eluted with an apparent molecular mass of about 300 kDa, which is consistent with the previous reports that this protein forms a multimeric structure (Fig. 1D). Despite this difference, when both proteins were visualized by electron microscopy, we observed ring structures that were similar to each other (Fig. 6, A and C; Refs. 24 and 59). These results suggest that wild type HsRad52 rings exhibit higher order associations, whereas HsRad52 1-237 rings do not (Fig. 1D). This is consistent with the previous report that the C-terminal region of HsRad52 is involved in higher order associations ( Fig. 1D; ref. 59). This higher order association of HsRad52 is apparently not essential in the homologous pairing, because HsRad52 1-237 has similar homologous pairing efficiency as the wild type (Figs. 4A and 5A). Surprisingly, when HsRad52 or HsRad52 1-237 was complexed with circular ⌽X174 ssDNA and visualized by electron microscopy, we observed nucleoprotein filaments that were about 10 nm thick ( Fig. 6B and 6D). These results indicate that the N-terminal region (HsRad52 1-237 ) is sufficient for the nucleoprotein filament formation, which is important in the homologous pairing, like the case of RecA.
Similarity between the HsRad52 and RecT Proteins-Although Rad52 homologues are not found outside of eukaryotes, our biochemical characterization of HsRad52 revealed significant similarities between HsRad52 and the E. coli RecT protein. Both homologous pairing proteins do not have an ATPbinding motif and do not require ATP for the reaction (Figs. 3A and 4A for HsRad52; Ref. 50 for RecT). These proteins can catalyze the D-loop formation without the assistance of other recombination factors. The HsRad52 protein can catalyze homologous pairing at protein concentrations below those saturating the ssDNA substrate, but the presence of any excess HsRad52 above the saturation point results in the drastic inhibition of homologous pairing. This drastic inhibition has also been reported with RecT (50). Without DNA, RecT and HsRad52 both reportedly form ring structures that contain seven or eight protein monomers, as observed by electron microscopy (22,24). When these two proteins bind to ssDNA, they form filamentous complexes. For HsRad52, the nucleoprotein filaments are probably composed of stacked rings or rings packed in an edge-to-edge manner (Fig. 6, B and D). These filament structures are probably important for the homologous pairing activity of these two proteins. Finally, HsRad52 and FIG. 7. HsRad51 binding of HsRad52 and HsRad52 1-237 . A, HsRad52 or HsRad52 1-237 (20 g) was mixed with HsRad51 (10 g) that was covalently conjugated to an Affi-Gel 15 matrix. After 90 min of incubation at 24°C, the Affi-Gel 15 matrix was washed with binding buffer and was directly mixed with 2-fold SDS-PAGE sample buffer (10 l). This mixture containing the Affi-Gel 15 matrix was boiled at 98°C for 2 min and was fractionated on a 15-25% gradient SDS-PAGE. Lanes 1 and 2 are one-tenth (2 g) of the input proteins, lanes 3 and 4 are the negative controls using the Affi-Gel 15 matrices with active ester sites blocked by ethanolamine, and lanes 5 and 6 are the Affi-Gel 15 matrices conjugated with HsRad51. The bands were visualized by Coomassie Brilliant Blue staining. B, graphical representation of the binding intensity to HsRad51 by HsRad52 and HsRad52 1-237 . The binding intensities of HsRad52 and HsRad52 1-237 were adjusted so that the average binding intensity of HsRad52 was 100%.
RecT have been suggested to function in homologous recombination pathways that are independent of HsRad51 and RecA, respectively (60). These similarities suggest that HsRad52 may be a functional homologue of RecT.
The C-terminal Region of HsRad52 Is Important in the Interaction with HsRad51-The yeast two-hybrid analysis suggested that the region containing amino acid residues 291-330 is involved in the interaction with HsRad51 (61). Because this segment is missing in HsRad52 1-237 , its binding to HsRad51 was predicted to be impaired. To test the interaction between HsRad52 1-237 and HsRad51, an HsRad51-conjugated affinity column was prepared. The HsRad52 1-237 protein that coprecipitated with the HsRad51-conjugated Affi-Gel 15 beads (Bio-Rad) was detected by 15-25% gradient SDS-PAGE with Coomassie Brilliant Blue staining (Fig. 7A). As compared with the wild type HsRad52, HsRad52 1-237 lost more than 90% of its ability to interact with HsRad51 (Fig. 7B). These results indicate that the C-terminal region of HsRad52 directly interacts with HsRad51.
Multiple Homologous Pairing Activities in Eukaryotes-Several enzymes in eukaryotes reportedly have homologous pairing activity. They include Rad51, Dmc1, Xrcc3-Rad51C, and HsRad52 (14 -19, 62). In E. coli, multiple recombination pathways have been found (63). These pathways play important roles in the repair of chromosomal breaks and the generation of genetic diversity (64). Because the architecture of the chromosome is much more complex in eukaryotes than in prokaryotes, the number of recombination pathways and the enzymes involved in those pathways are expected to be larger in eukaryotes. In this context, the multiple homologous pairing activities can be explained by the involvement of different subsets of recombination enzymes in different recombination pathways. For those pathways that are Rad51-independent, Rad52, Xrcc3-Rad51C, or other proteins may be the central catalysts in the homologous pairing step. It is also possible that several homologous pairing proteins act in the recombination process together. The C-terminal region of HsRad52 is important for the physical interaction with HsRad51. This interaction may be important for the cooperative homologous pairing reaction with HsRad51. In fact, HsRad52 and ScRad52 have both been shown to catalyze homologous pairing, in cooperation with HsRad51 and ScRad51, respectively, in the Rad51-dependent pathway (32,(37)(38)(39)46). The C-terminal region of HsRad52 may also interact with other human homologous pairing proteins, such as Xrcc3 and Rad51C. Further investigation is required to prove this hypothesis.