Saccharomyces cerevisiae Dmc1 protein promotes renaturation of single-strand DNA (ssDNA) and assimilation of ssDNA into homologous super-coiled duplex DNA.

Dmc1 and Rad51 are eukaryotic RecA homologues that are involved in meiotic recombination. The expression of Dmc1 is limited to meiosis, whereas Rad51 is expressed in mitosis and meiosis. Dmc1 and Rad51 have unique and overlapping functions during meiotic recombination. Here we report the purification of the Dmc1 protein from the budding yeast Saccharomyces cerevisiae and present basic characterization of its biochemical activity. The protein has a weak DNA-dependent ATPase activity and binds both single-strand DNA (ssDNA) and double-strand DNA. Electrophoretic mobility shift assays suggest that DNA binding by Dmc1 is cooperative. Dmc1 renatures linearized plasmid DNA with first order reaction kinetics and without requiring added nucleotide cofactor. In addition, Dmc1 catalyzes strand assimilation of ssDNA oligonucleotides into homologous supercoiled duplex DNA in a reaction promoted by ATP or the non-hydrolyzable ATP analogue AMP-PNP.

and Dmc1 (7)(8)(9)(10). Rad51 is important for both mitotic and meiotic recombination whereas Dmc1 expression and function are meiosis-specific. Additional relatives of Rad51 have been identified in yeast and vertebrates (reviewed in Ref. 11). Although a complex containing two of the five vertebrate Rad51 paralogues has been shown to function together to promote homologous strand assimilation (12), additional data suggest that Rad51 paralogues act as assembly factors for Rad51 (13)(14)(15). Genetic studies have demonstrated that both Dmc1 and Rad51 can promote recombination in the absence of the other (16,17). Despite this partial redundancy of function, Rad51 and Dmc1 have unique functions during recombination that cooperate during a large fraction of recombination events (16 -19). Evidence for direct interaction between Dmc1 and Rad51 has been reported for the human but not the yeast orthologues (20 -22). The mechanistic details and function of Dmc1-Rad51 cooperation during recombination are unknown.
Like RecA, Rad51 from human and yeast has been shown to catalyze homologous pairing and perform strand exchange activity (11,23,24). Its activity is stimulated by accessory factors, including the single-strand DNA binding protein RPA (replication protein A) and proteins encoded by members of the RAD52 epistasis group (11,25). These factors help load Rad51 onto ssDNA and promote unwinding of dsDNA targets during strand invasion and exchange. Rad51 forms helical nucleoprotein filaments in which the DNA is in an underwound and extended conformation relative to the B form, indicating that the structural framework underlying the strand invasion process is conserved between Escherichia coli and eukaryotes (26,27).
Like yeast and human Rad51, human Dmc1 (HsDmc1) has been reported to have homologous pairing activity (22,28,29), although the activity reported is less robust than that of Rad51. The HsDmc1 protein has been shown to form toroidal octomers that can bind DNA as a stack of rings on the DNA (22,30). Helical filaments have not been detected. Discovery of the toroidal form of Dmc1 raises the possibility that the functional form of Dmc1 differs from that of other members of the RecA family. Direct evidence supporting this possibility has yet to be obtained.
In addition to promoting homology-mediated invasion of duplex, RecA is able to renature complementary ssDNA strands, an activity that may contribute to some types of recombination (31). Renaturation activity has also been reported for HsRad51 (32). Furthermore, yeast Rad52 also promotes renaturation, and this activity may be responsible for Rad51-independent recombination events in mitotic cells (33)(34)(35)(36).
Here, we describe the purification of Dmc1 protein from the budding yeast Saccharomyces cerevisiae and biochemical characterization of its function. Like RecA, Rad51, and HsDmc1, ScDmc1 binds both ssDNA and dsDNA and has DNA-dependent ATPase activity. We show for the first time that Dmc1 promotes renaturation of complementary ssDNA and does so with first order reaction kinetics like RecA. We also show that ScDmc1 can promote homologous ssDNA strand assimilation in duplex DNA.

EXPERIMENTAL PROCEDURES
Complementation and Recombination Frequency Assays-To determine if the Dmc1 protein remains functional in S. cerevisiae when modified by the addition of 6 histidine codons at the amino terminus, an HIS 6 -DMC1 allele was constructed. This allele was inserted at the normal DMC1 locus by one-step transplacement. The KanMX4 gene was located 643 bp 3Ј of the transcriptional stop codon, conferring resistance to the antibiotic G418 (Sigma Chemical Co.). The HIS 6 -DMC1::KanMX4 construct was transformed into the haploid strain DKB1586 (MATa; dmc1⌬::ARG4) to create strain DKB1589. As a control, a DMC1::KanMX4 construct (i.e. an unmodified copy of DMC1) was also transplaced into DKB1586 to create strain DKB1687. Proper integrants were detected by screening transformants for loss of arginine prototrophy. Integration into the correct position was confirmed by Southern blot analysis.
Dmc1 Expression in E. coli-An intronless derivative of DMC1 was modified to encode an additional 6 histidine residues at the amino terminus and placed under the control of a phage T7 promoter (pET3a, Novagen) to create pNRB150. This plasmid was chemically transformed into BLR(DE3)/pLysS E. coli cells (Novagen), which carry a null mutation in the recA gene. Cells were grown at 37°C in Luria-Broth supplemented with 100 g/ml ampicillin and 20 g/ml chloramphenicol until the cells reached A 540 ϭ 0.3. Expression of Dmc1 was induced by the addition of 1.0 mM isopropyl-␤-D-thiogalactopyranoside followed by a 1.5-h incubation at 30°C. Cells were harvested by centrifugation, washed, resuspended in equal volume of buffer A (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM BME), frozen in liquid nitrogen, and stored at Ϫ80°C.
Purification of Dmc1-Frozen cells were rapidly thawed and kept at 4°C for the remainder of the purification. The thawed cells were sonicated four times on a Cole-Parmer ultrasonic homogenizer (Chicago, IL) for 30 s each at 30% duty cycle and output of 3 in the presence of protease inhibitors (5 g/ml antipain, 2 g/ml aprotinin, 100 ng/ml leupeptin, 100 g/ml Pefabloc SC). Insoluble material was removed by ultracentrifugation for 45 min at 25,000 rpm in a Beckman 70 Ti rotor. The cleared lysate was incubated with Talon resin (CLONTECH) that had been equilibrated in buffer A. The resin was washed with TP buffer (20 mM Tris-HCl, pH 7.5, 100 mM sodium phosphate, pH 7.5, 10% glycerol, 2 mM BME). The protein-bound resin was then incubated with TP buffer containing 0.36 unit/l of DNase I (Amersham Pharmacia Biotech), 2 mM MgCl 2 , and protease inhibitors. The resin was washed with TP buffer and then with 50 mM imidazole, pH 7.9, in TP buffer. Dmc1 was eluted with 100 mM imidazole, pH 7.9, in TP buffer.
Protein eluted from the Talon resin was applied to Cibacron Blue (Bio-Rad) equilibrated with TP buffer. The column was washed with 200 mM NaCl in buffer G (20 mM Hepes, pH 7.9, 10% glycerol, 1 mM EDTA, 2 mM BME). Dmc1 was eluted with 1.0 M NaCl in buffer G. The eluted protein was dialyzed against 200 mM NaCl in buffer G. The dialyzed fraction was then applied to Q-Sepharose (Amersham Pharmacia Biotech) equilibrated in buffer G with 200 mM NaCl. Dmc1 was eluted from the column with steps of 0.3, 0.5, 0.7, and 1.0 M NaCl in buffer G.
Protein fractions from the 0.5 and 0.7 M NaCl elution were combined and concentrated by dialysis into storage buffer (20 mM Hepes, pH 7.9, 50% glycerol, 0.7 M NaCl, 1 mM EDTA, 2 mM BME). Aliquots were frozen in liquid N 2 and stored at Ϫ80°C. The yield from 8.0 liters of cells was ϳ800 g of Dmc1 protein.
Protein concentrations were determined by Bradford assay (Bio-Rad) and by spectrophotometric analysis at 280 nm (Dmc1 has an extinction coefficient of 12240 M Ϫ1 cm Ϫ1 at 280 nm).
For ATPase assays, the viral strand of M13 was propagated in E. coli strain JM101 and prepared according to published methods (41). For DNA binding assays, replicative form I ⌽X 174 DNA (New England BioLabs) was linearized with BsrBI followed by treatment with alkaline phosphatase (New England BioLabs). The reaction mixture was extracted with phenol and ethanol-precipitated. The ⌽X 174 viral strand was linearized by hybridizing oligos to create a PstI site and a BsrBI site. The reaction mixture was run on an agarose gel. The band was excised and purified using the Geneclean kit (BIO 101, Inc., Vista, CA). For DNA renaturation assays, pBluescript II SK (ϩ) was linearized with AccI followed by treatment with alkaline phosphatase. The reaction mixture was extracted with phenol and ethanol-precipitated.
All DNA was end-labeled with [␥-32 P]ATP by T4 polynucleotide kinase (New England BioLabs) using a protocol provided by the supplier. Radiolabeled DNA was purified from unincorporated nucleotide using a G25 MicroSpin column (Amersham Pharmacia Biotech), and the specific activity was adjusted to roughly 2.5 ϫ 10 10 cpm/mol by addition of unlabeled DNA. Concentrations of nucleic acids are given in moles of nucleotide for ssDNA or moles of base pairs for dsDNA.
ATPase Assay-Reactions (10 l) contained circular single-stranded M13 (0.3 mM nucleotide) in 20 mM Hepes, pH 7.5, 1 mM DTT, 1 mM magnesium acetate, 100 g/ml bovine serum albumin, and 0.1 mM [␣-32 P]ATP. The turnover rate was determined by calculating the V max based on a 50-fold range of ATP concentration. Reactions were started by the addition of the appropriate amount of Dmc1 and incubated at 37°C. At the indicated time points, 1.0-l aliquots were taken and spotted onto thin layer chromatography plates (CEL300 PEI, Machery-Nagel). The plates were developed in 850 mM potassium phosphate, pH 3.4, dried, analyzed using a Storm PhosphorImager (Molecular Dynamics), and quantified using ImageQuant version 4.0 (Molecular Dynamics).
DNA Binding Assay-Reactions (15 l) contained either 1.3 M 32 Plabeled linear ds⌽X 174 or 2.6 M 32 P-labeled linear ss⌽X 174, 20 mM Hepes, pH 7.5, 1 mM DTT, 1 mM magnesium acetate, and 2 mM ATP. Reactions were started by the addition of the appropriate amount of Dmc1 and incubated for 5 min at 37°C. Reaction were analyzed by electrophoresis on an 0.8% agarose gels for 1 h at 10 V/cm using 1ϫ Tris acetate-EDTA (TAE) running buffer containing Tris-HCl reduced to 10 mM. The gel was dried onto Whatman paper and developed on Kodak BioMax MS film.
Renaturation Assay-32 P-Labeled heat-denatured linear pBluescript II SK (ϩ) at various concentrations was incubated with Dmc1 in 20 mM Hepes, pH 7.5, 10 mM magnesium acetate, 1 mM DTT, and 2 mM nucleotide cofactor. Aliquots of 25 l were removed at appropriate time points and treated with 0.1% SDS at 37°C for 10 min. Following addition of SDS, samples were kept at 4°C while additional time points were taken. The samples were resolved on a 0.8% agarose gel in 1ϫ TAE buffer for 1 h at 10 V/cm. The gels were dried on Whatman paper and analyzed by phosphorimaging.
Strand Assimilation Assay-Dmc1 (0.54 M) was preincubated with either 32 P-labeled oligo 18.1, 306.7, or both (1.0 M each) as in the DNA binding assays. Following preincubation, a mixture of pUC18 (12.3 M) and pNRB252 (6 M) was added along with sufficient magnesium acetate to give a final concentration of 10 mM. Oligos were in 5-fold molar excess to dsDNA target sequence. The reactions were incubated at 37°C and stopped by the addition of SDS to a final concentration of 0.08%. DNA from reaction mixtures was resolved on a 0.8% agarose gel in 1ϫ TAE buffer for 14 -16 h at 2.2 V/cm. Gels were dried onto Whatman paper and analyzed by phosphorimaging.

RESULTS
Complementation of His 6 -Dmc1-To aid in purification, the full-length coding region of DMC1 was tagged with His 6 . To ensure that the addition of 6 histidine residues to the amino terminus did not disrupt Dmc1 function, a strain was constructed that expressed only the His 6 -tagged Dmc1 protein. This strain was assayed for sporulation efficiency, spore viability, and the ability to generate recombinant products. The strain expressing His 6 -Dmc1 was indistinguishable from the wild type control in all three assays (Table I). These data indicate that the His 6 -Dmc1 protein is fully functional in vivo.
Purification of ScDmc1-His 6 -tagged Dmc1 was cloned into the phage T7 expression vector pET3a. The protein was expressed and purified from an E. coli strain that allowed inducible expression of the T7 RNA polymerase and carried a null allele of the recA gene to avoid potential contamination of Dmc1 preparations with RecA. Dmc1 was purified to near homogeneity by three chromatographic steps (see "Experimental Procedures"). A Coomassie Blue-stained SDS-polyacrylamide gel electrophoresis gel of the purified protein shows a single prominent band (Fig. 1A). Although there are species of smaller molecular weight detected in the lysate by Western blot, these did not bind to the Talon resin and the final fraction contains little or none of these species as assayed by Western blot (Fig. 1B).
Dmc1 has DNA-dependent ATPase Activity-Dmc1 contains Walker A and B amino acid sequence motifs, which are present in many proteins that bind ATP (42). Like other members of the RecA family, the ability of Dmc1 to hydrolyze ATP to ADP and P i is stimulated in the presence of either ssDNA or dsDNA (Fig.  2). The presence of dsDNA only stimulates the ATPase activity of Dmc1 to half the activity level stimulated by ssDNA. Omission of DNA results in 13% relative to the maximal ATPase activity. The lower ATPase activity in the presence of dsDNA as compared with ssDNA is unlikely to be due to lack of binding of Dmc1 to DNA, because all the Dmc1 should be bound to the DNA under the reaction conditions employed (see below). The ATP turnover rate (k cat ) was calculated by generation of Lineweaver-Burk plots. In two independent experiments, the k cat was calculated to be 0.6 min Ϫ1 and 0.8 min Ϫ1 , giving an average of 0.7 min Ϫ1 . Although this is lower than the turnover rate for RecA (k cat ϭ 30 min Ϫ1 ), it is similar to the k cat of human and yeast Rad51 (k cat ϭ 0.16 min Ϫ1 and 0.7 min Ϫ1 , respectively) (23, 24) and human Dmc1 (k cat ϭ 1.5 min Ϫ1 and 0.7 min Ϫ1 ) (28,29).
Dmc1 Binds ssDNA and dsDNA-The ATPase assays indicate that Dmc1 interacts with DNA. The ability of Dmc1 to stably bind ssDNA and dsDNA was assayed directly by electrophoretic mobility shift assays (EMSAs). Dmc1 was able to bind both ssDNA and dsDNA forming complexes that were stable enough to survive electrophoresis in agarose gels without cross-linking (Fig. 3).
Incubation of ssDNA (2.6 M nucleotide) in the presence of ATP showed that below 0.126 M Dmc1 there was little shifting of DNA, although the mobility of a minor fraction of DNA was slightly shifted. Increasing the concentration of Dmc1 from 0.126 to 0.28 M resulted in a discrete shift in the mobility of linear ss⌽X 174. The lowest concentration at which the shift was observed corresponds to a molar ratio of 1 monomer per 9 nucleotides. Further increasing the Dmc1 concentration resulted in a gradual increase in the mobility of the ssDNA to a plateau that corresponds to a monomer per nucleotide ratio of 1:2 to 1:3. Binding to ssDNA was not substantially altered when nucleotide cofactor was omitted from reactions (data not shown).
Dmc1 did not result in an electrophoretic mobility shift of dsDNA at concentrations below 0.28 M. A small fraction of the DNA was retarded after incubation with 0.28 M DNA, whereas a discrete shift of all DNA was observed at 0.42 M, which corresponds to a monomer per base pair ratio of 1:6. The same shift in mobility of DNA was observed after binding to Dmc1 at concentrations ranging from 0.42 to 1.26 M. However, a further 2-fold increase in Dmc1 concentration from 1.26 to 2.5 M resulted in a further reduction in mobility of DNA. These data suggest that Dmc1 binds to dsDNA with slightly less affinity than to ssDNA. Consistent with this suggestion, binding to dsDNA was sensitive to slightly lower concentrations of NaCl than binding to ssDNA (data not shown). Unlike binding to ssDNA, binding to dsDNA required the presence of a nucleotide cofactor, such as ATP, dATP, ADP, or a non-hydrolyzable ATP analogue (data not shown).
Dmc1 Catalyzes Renaturation of Complementary Single Strands-To determine if it is able to promote ssDNA renaturation, Dmc1 (0.35 M) was incubated with 32 P-labeled heat- a Sporulation efficiency was measured by phase contrast microscopy following sporulation on SPM ϩ 1/5 COM plates. Only asci containing at least two spores scored.
b Spore viability was assayed by tetrad dissection (n ϭ 200) on YPD agar medium.
c Frequency of interhomologue recombination were measured at the his4X/his4B locus for formation of prototrophs (10).  4A), but Dmc1-dependent renaturation activity is readily detected at 0.5 min (Fig. 4B). Essentially all renatured products were in the form of linear duplex as opposed to high molecular weight networks. Such networks are the predominant product of RecA-promoted renaturation under most conditions (31,43,44). The results suggest that, under the conditions employed, the renatured products form without multiple independent "nucleation" events on a single molecule.
Maximum Dmc1-dependent renaturation activity occurs at ratios of Dmc1 to nucleotides between 1:10 and 1:40. (Fig. 4C). If saturation of ssDNA by Dmc1 occurs around 1 monomer to two to three nucleotides (as suggested by the EMSAs), optimal renaturation under these conditions occurs when less than 20% of the ssDNA is bound to protein. Both lower and higher levels of Dmc1 reduce the level of renaturation.
To determine if the Dmc1-dependent renaturation reaction proceeds with first or second order kinetics, the ratio of Dmc1 to nucleotide was kept at 1:14 and the concentration of heatdenatured linear pBluescript was varied over a 100-fold range (0.011-1.1 nM plasmid molecules). A log-log plot of the half times of the renaturation reactions versus the initial concentration of heat-denatured DNA yields a straight line with a slope of Ϫ0.097 (Fig. 4D). The log-log plot in the absence of nucleotide cofactor or in the presence of other nucleotide cofactors produced similar values (Fig. 4D and data not shown). Because the slope of the plot did not differ significantly from 0, the results indicate that the Dmc1-dependent renaturation is a first order reaction, like RecA-dependent renaturation (44,45). A slope of Ϫ1 is expected if the reaction proceeds via second order kinetics as is the case for non-enzymatic renaturation. Under the conditions of our experiments the Dmc1-dependent renaturation reaction did not require ATP hydrolysis nor was the efficiency substantially dependent on added nucleotide. Although RecA-dependent renaturation is more efficient in the presence of ATP under some conditions, both RecA and Rad51 have also been found to promote renaturation in the absence of hydrolysis (32,45). The first order rate constant for renaturation was determined to be 0.35 min Ϫ1 , a value similar to values reported for RecA (45).
Dmc1 Promotes Homologous Strand Assimilation-RecA can promote assimilation of ssDNA into intact dsDNA (46,47). The product of this reaction survives deproteination and is likely to be a "D-loop" in which the assimilated strand forms Watson Crick base pairs with the complementary sequence on the target (48). However, it is also possible that deproteination of homologously paired complexes traps the single strand in a stable, non-B form structure (49). For this reason we refrain from concluding that the activity we observe with the assay is complete strand exchange.
The ability of Dmc1 to catalyze strand assimilation between ssDNA and dsDNA was assayed using oligos homologous to super-coiled duplex plasmids. Homologous joints formed between the oligo and the plasmid are detected following gel electrophoresis. Because of the relatively small size of the oligo compared with the super-coiled target plasmid, the joint molecules have about the same electrophoretic mobility as supercoiled dsDNA and are detected as a radioactive species at the position of the supercoiled dsDNA following electrophoresis.
Dmc1 is able to catalyze assimilation of ssDNA into dsDNA in a homology-dependent manner. Maximum strand assimilation activity was observed after 10 min (Fig. 5A) and when AMP-PNP was used as a nucleotide cofactor (Fig. 5C). In the presence of two super-coiled plasmids, only one of which contained a sequence homologous to the oligo, joints were only detected between the oligo and the plasmid containing the homologous sequence (Fig. 5B, lanes 2 and 4). Although variable, ϳ20% of maximum levels of product formation were detected in the presence of ATP. Product formation was not detected in the absence of nucleotide cofactor or when ATP␥S was substituted for AMP-PNP (Fig. 5C). This failure of ATP␥S to support strand assimilation did not result from failure of the nucleotide to bind Dmc1, because it inhibited Dmc1's ATPase activity under similar conditions (data not shown).
Optimal strand assimilation activity occurred at a Mg 2ϩ concentration between 10 and 15 mM; there was a 2-fold decrease in product formation at Mg 2ϩ concentrations of 5 and 50 mM. Strand assimilation also required that the dsDNA substrate be super-coiled; linearization of the plasmid at various unique restriction sites abolished the signal. Strand assimilation activity was not detected at concentrations of Dmc1 below 1/3 monomer/nucleotide. Additionally, optimal activity required preincubation of Dmc1 with the ssDNA oligo before the addition of the super-coiled dsDNA target. The optimum stoichiometry and sensitivity to Mg 2ϩ concentration are similar to those reported for strand assimilation by RecA and strand exchange by Rad51 (23,47,50,51). DISCUSSION The results presented here add to the evidence that the function of the meiosis-specific protein Dmc1 is closely related to E. coli RecA and eukaryotic Rad51. Like HsDmc1, ScDmc1 is a weak DNA-dependent ATPase, binds both dsDNA and ssDNA, and promotes ssDNA assimilation into duplex DNA. We show for the first time that Dmc1 promotes efficient renaturation of complementary ssDNA strands.
Results of the EMSAs suggest Dmc1 binds DNA cooperatively. Gel shift assays showed distinct shifts of DNA mobility only when a certain protein level was achieved. These shifted species were unaltered by further increases in Dmc1 concentration. This behavior is not expected if monomers bind DNA strands in a non-cooperative manner. Such situations predict that individual DNA molecules will bind different amounts of proteins resulting in a gradual and non-uniform shift in mobility until saturation is reached. The discrete shifts observed in these experiments suggest that Dmc1 binds both ssDNA and dsDNA in a cooperative manner, as shown previously for RecA and Rad51 (7,(52)(53)(54).
EMSAs suggest that Dmc1 saturates ssDNA and dsDNA at about 1:3 monomer/nucleotide and 1:6 monomer/base pairs, respectively. These numbers are similar to those reported for RecA, which saturates DNA at 1:6 monomers per nucleotide under some conditions and 1:3 under others (Ref. 55, and references therein). The results are also similar to results obtained for HsRad51, ScRad51, and HsDmc1 (7, 24, 26 -28, 56). Further studies of Dmc1-DNA binding using methods that allow accurate quantification of binding in solution are needed to determine the degree of cooperativity and stoichiometry of Dmc1-DNA binding.
Dmc1 promotes efficient renaturation of complementary ssDNA with the following characteristics: Optimal renaturation occurs at Dmc1 levels that cover less than 20% of the ssDNA; deviation from this optimum results in less efficient renaturation; and the reaction occurs with first order kinetics. These properties are shared with RecA-mediated renaturation (45) and are in contrast to spontaneous renaturation as well as the renaturation reaction promoted by the yeast recombination protein Rad52 (35). The first order kinetics suggest that the rate-limiting step in renaturation of complementary ssDNA strands is not the pairing of complementary strands but a later step that involves conversion of a complex containing Dmc1 and two ssDNA strands to stable duplex DNA (43). The ability of Dmc1 to promote annealing may result from an ability to promote unstable protein-DNA aggregates and/or from the ability of Dmc1 oligomers to bind 2 DNA strands simultaneously (43,45).
In contrast to the renaturation activity, which did not require nucleotide cofactor, the strand assimilation by Dmc1 required addition of either ATP or the non-hydrolyzable analogue AMP-PNP. The finding that AMP-PNP promotes strand assimilation by Dmc1 is consistent with results obtained with RecA and Rad51, which indicate that strand assimilation and strand exchange require a nucleotide cofactor but not ATP hydrolysis (46,57,58). Effective nucleotide cofactors are predicted to act as allosteric effectors causing Dmc1 to adopt a conformation that is favorable for strand assimilation.
Although the renaturation activity of Dmc1 is quite efficient in vitro, the precise relevance of this activity to Dmc1's function in vivo remains to be determined. The DNA at the sites of meiotic double-strand breaks is rendered singled-stranded by nucleases. The two ssDNA ends invade an intact duplex chromatid to form double-Holliday junctions (59). Both Dmc1 and Rad51 are required for full levels of joint molecule formation in vivo. It is possible that invasion of one ssDNA end by Rad51 forms a D-loop that is enlarged by repair synthesis on the intact chromatid, forming a region of ssDNA complementary to the ssDNA sequences on the second DNA end. Dmc1 could catalyze annealing between the D-loop and the ssDNA region FIG. 4. Dmc1 renatures complementary ssDNA strands with first order reaction kinetics. Heat-denatured 32 Plabeled linear pBluescript II SK(ϩ) was incubated A, in the absence of or B, in the presence of Dmc1 at 37°C. Aliquots were removed at the indicated time points and deproteinized by the addition of SDS to 0.1%. The reaction products were separated on an agarose gel and developed as described under "Experimental Procedures." C, dependence of renaturation on Dmc1 concentration. Product formation was measured after 10 min. D, kinetics of the Dmc1-dependent renaturation activity in the presence of AMP-PNP (Ⅺ) or in the absence of any nucleotide cofactor (E). The logarithm of the half-time (ln t1 ⁄2 ) of the renaturation activity was plotted as a function of the logarithm of the initial molar concentration of the dsDNA. from the second DNA end. However, we believe that this explanation cannot fully account for Dmc1's in vivo recombination activity. A rad51 mutant shows only a 3-fold reduction in the formation of joint molecules, and this level of recombination depends on DMC1 ϩ (16). In addition, a dmc1 mutant allele that contains a lysine to alanine mutation in the putative ATP binding site of Dmc1 (i.e. the "Walker box" motif) was indistinguishable from a null mutant, in respect to promoting formation of meiotic products. 2 Thus nucleotide binding is likely to be important for Dmc1 function in vivo. Because strand assimilation but not single-strand annealing is effected by the absence of nucleotide, it is unlikely that the Walker box mutant's phenotype results from a defect in annealing activity alone. We propose that, like other members of the RecA family, Dmc1 performs strand invasion to form joint molecules by a mechanism distinct from, but closely related to, the mechanism that promotes renaturation of DNA in vitro.
Even under optimal conditions used here, the level of Dmc1's strand assimilation activity was low; a maximum of 1% of oligo and 4% of target duplex were present in joint molecules. It is important to use caution in ascribing biological significance to the strand assimilation activity of a protein when that activity is low. Nonetheless, we argue that the assimilation activity of Dmc1 is biologically relevant for three reasons. First, as discussed in the introduction, there is good evidence that Dmc1 promotes the formation of homologous joint molecules in vivo and does so in the absence of other proteins related to RecA. Second, as discussed above, both Dmc1 recombination activity in vivo and its strand assimilation activity in vitro appear to depend on nucleotide cofactor. This property of Dmc1 strand assimilation is shared with the efficient strand assimilation reaction promoted by RecA. Finally, as discussed further below, previous studies predict that the efficiency of Dmc1's strand assimilation activity will be strongly dependent on accessory proteins.
Although further optimization of the reaction may increase strand assimilation activity of Dmc1 in vitro, it is likely that the efficiency of the reaction is limited due to the absence of one or more accessory proteins. This possibility is underscored by previous studies on Rad51. In contrast to RecA, which can promote strand assimilation in the absence of other proteins, Rad51's strand assimilation activity requires one of two structurally related DNA unwinding proteins, Rad54 or Tid1/Rdh54 (60 -63). These unwinding proteins have not been shown to possess helicase activity, but have been shown to alter the topology of duplexes, probably by reducing the twist of the dsDNA target. Because Dmc1-catalyzed strand assimilation requires a super-coiled duplex target, local unwinding may be required for Dmc1 to promote strand assimilation in the absence of other proteins. Tid1/Rdh54 was isolated based on its interaction with Dmc1 in a two-hybrid assay and subsequently shown to interact with Rad51 (64). Cytological and genetic observations suggest that Tid1/Rdh54 stimulates Dmc1 activity in vivo (19,65). Thus, Dmc1 almost certainly relies on Rdh54/Tid1 to alter the topology of the target duplex as it promotes homologous strand assimilation in vivo. Dmc1 is also likely to rely on factors that direct its assembly into oligomers at the sites of meiotic double-strand breaks. A particularly interesting possibility is that Dmc1's biochemical activity is influenced by interaction with its partner recombinase Rad51, as suggested by cytological and biochemical experiments (18, 20 -22).