Structure-Function Analysis of the Three Domains of RuvB DNA Motor Protein*[boxs]

RuvB protein forms two hexameric rings that bind to the RuvA tetramer at DNA Holliday junctions. The RuvAB complex utilizes the energy of ATP hydrolysis to promote branch migration of Holliday junctions. The crystal structure of RuvB from Thermus thermophilus (Tth) HB8 showed that each RuvB monomer has three domains (N, M, and C). This study is a structure-function analysis of the three domains of RuvB. The results show that domain N is involved in RuvA-RuvB and RuvB-RuvB subunit interactions, domains N and M are required for ATP hydrolysis and ATP binding-induced hexamer formation, and domain C plays an essential role in DNA binding. The side chain of Arg-318 is essential for DNA binding and may directly interact with DNA. The data also provide evidence that coordinated ATP-dependent interactions between domains N, M, and C play an essential role during formation of the RuvAB Holliday junction ternary complex.

Homologous recombination enhances genetic diversity, contributes to genomic stability, and plays an important role in DNA repair. In prokaryotes, RuvA, RuvB, and RuvC proteins process a central intermediate of homologous recombination, the Holliday junction, in a late stage of homologous recombination (1)(2)(3). RuvA and RuvB form a complex that promotes branch migration of the Holliday junction. The RuvA tetramer, a Holliday junction-specific binding protein, facilitates the binding of RuvB to the Holliday junction. Two RuvA tetramers bind to opposite sides of the Holliday junction, and RuvB hexamers are tethered to and flank the RuvA tetramers (4 -6). The RuvB hexamer forms a ring that encircles double-stranded DNA (dsDNA) 1 and promotes DNA translocation in an ATP hydrolysis-dependent manner (2,3). RuvC is a structure-specific endonuclease that resolves the Holliday junction (7,8). The nicked ends of the two DNA molecules are sealed by DNA ligase to complete the recombination reaction.
RuvB has intrinsic ATPase activity (9,10) that is synergistically enhanced by RuvA and DNA (11). RuvB binds to DNA in the presence of ATP␥S in vitro (12). The structural analysis of RuvB suggests that it is a member of the AAA ϩ (ATPase associated with various cellular activities) ATPase superfamily rather than a member of the hexameric helicase family as previously suggested (13,14). The crystal structures of RuvB from Thermus thermophilus HB8 (Tth) and Thermotoga maritima were recently solved (15,16). These structures and other data suggest that the structure and function of the RuvAB complex are highly conserved among bacteria. The RuvB monomer has distinct amino-terminal (N), central (M), and carboxyl-terminal (C) domains (see Fig. 1A). The N and M domains are structurally similar to equivalent domains in other AAA ϩ family ATPases, except that a unique ␤-hairpin protrudes from the RuvB domain N. This ␤-hairpin physically interacts with RuvA and is required for formation of the RuvAB complex (17)(18)(19). Domain C of RuvB contains a winged helix motif, which is topologically similar to the DNA-binding domains of metallothionein repressor SmtB and histone H5 (15,16). Structural and mutational analyses of RuvB suggest that it contains several functional motifs related to ATPase activity, including Walker motifs, sensor motifs, and an arginine finger (14 -16, 20). RuvB is a multifunctional protein that interacts with RuvA, ATP, DNA, and Mg 2ϩ . These interactions and RuvB ATPase are required for RuvAB-dependent branch migration of Holliday junctions.
This study analyzes the function of seven RuvB mutants, five truncated RuvB mutants lacking one or more RuvB domains and two RuvB point mutants with amino acid substitution mutations of Arg-318, a C-domain residue that plays a role in DNA binding (15). The results provide insight into the specific functional roles of each of the three RuvB domains. Furthermore, the data presented here suggest that coordinated ATPdependent functions of the N, M, and C domains of RuvB are required for DNA binding, ATP hydrolysis, and RuvAB-catalyzed branch migration of Holliday junctions.

EXPERIMENTAL PROCEDURES
Escherichia coli Strains and Plasmids-HRS2301 (⌬ruvB::Cm r ) (21), derived from AB1157 (22), was used as a ⌬ruvB strain for the UV light sensitivity test. BL21 (DE3) (23) was used for overexpression of the GST fusion proteins. HRS4000 (⌬ruvABC::Km r ) (24), derived from BL21 (DE3), was used for overexpression of the recombinant RuvA and RuvB proteins. pAF101 was used for overexpression of wild-type and mutant * This work was supported by Grants-in-aid for Scientific Research on Priority Areas from Ministry of Education, Culture, Sports, Science, and Technology of Japan (to H. S.) and grants from CREST of JST (Japan Science and Technology Agency) (to Y. H. and H. S.). 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 U.S.C. Section 1734 solely to indicate this fact.
UV Light Sensitivity Assay-pRB100 derivatives expressing the truncated RuvB proteins were introduced into HRS2301 or wild-type strain. Transformants were grown to stationary phase, and 10-fold serial dilutions of the cultures were spotted onto LB agar plates con-taining ampicillin at 100 g/ml. The agar plates were irradiated by UV light at 45 J/m 2 and incubated overnight at 37 or 26°C.
Protein Purification-The RuvA, wild-type RuvB, and RuvB R318A proteins were prepared as described (18,(25)(26)(27)(28). HRS4000 was used as a host for overproduction of the GST fusion RuvB proteins. HRS4000 carrying wild type or the truncated ruvB plasmids were grown at 37°C in 2 liters of LB medium containing 100 mg/ml ampicillin. When reached A 600 ϭ 0.3, isopropyl-1-thio-␤-D-galactopyranoside was added to a concentration of 2 mM for induction of the GST fusion RuvB protein synthesis, and culture was grown for another 6 h at 26°C. The cells were collected by centrifugation and resuspended in R buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol, 2 mM 2-mercaptoethanol) containing 0.1 M NaCl. After sonication and centrifugation, the supernatant was applied to a glutathione-Sepharose column (Amersham Biosciences) and eluted with elution buffer (50 mM Tris-HCl, pH 8.0, 10 mM reduced glutathione). GST was removed from the fusion protein using precision protease (Amersham Biosciences). The samples were dialyzed against storage buffer (10 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 50% glycerol) and stored at Ϫ30°C. Protein concentrations were determined using a Bio-Rad protein assay kit.
Branch Migration and ATPase Assay-The branch migration assay was carried out as described previously (18,20,29). The substrate used for the branch migration assay was a 32 P-labeled synthetic Holliday junction DNA (27). The reaction mixture (20 l) was incubated for 30 min at 26 or 37°C, and the reaction was stopped by the addition of 5 l of stop buffer. The products were analyzed by PAGE in a 9% gel and visualized by autoradiography. The ATPase assay was performed as described previously (27).
DNA Binding Assay-The DNA binding assay was performed essentially as described (30). The standard reaction mixture (20 l) contained 20 mM Tris acetate (pH 8.0), 10 mM Mg(OAc) 2 , 0.01% bovine serum albumin, 1 mM ATP␥S, 10 nM 70-mer duplex DNA labeled with 32 P, and wild type or truncated RuvB at the indicated concentration. Samples were incubated at 26 or 37°C for 30 min and analyzed by PAGE in a 6% gel in TAM buffer (40 mM Tris acetate (pH 7.8), 0.5 mM Mg(OAc) 2 ), followed by visualization by autoradiography.
Formation of RuvAB Holliday Junction Complex-The assay was carried out as described previously (18,31). The reaction mixture (20 l) was incubated for 20 min at 26°C. Protein-DNA complexes were fixed by incubation with 0.2% glutaraldehyde at 26°C for 30 min. The reaction products were analyzed by PAGE in a 6% gel and visualized by autoradiography.
Gel Filtration Analysis of RuvB Oligomer and RuvAB Complex Formation-Gel filtration experiments to assess RuvAB protein complex formation were carried out as described previously (27). Samples of RuvA (15 M), RuvB (10 M), and a mixture of RuvA and RuvB were applied to a Superdex-200 column equilibrated with buffer A (20 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 50 mM NaCl, and 5% glycerol). RuvB samples contained intact RuvB, RuvB-N, or RuvB-NM. The truncated RuvBs containing RuvB-M, RuvB-MC, and RuvB-C were applied to a Superdex-75 column. Protein peaks were monitored by absorbance at the wavelengths 225 and 280 nm. The assay to examine RuvB hexamer formation was carried out as described previously (27). The wild type and mutant RuvBs (35 M) were incubated for 3 h at 4°C in buffer A containing 10 mM MgCl 2 and 0.25 mM ATP and 10-l aliquots were applied to the column equilibrated with buffer A containing 10 mM MgCl 2 and 0.25 mM ATP.

RESULTS
Structure-Function Analysis of RuvB Using Domain Truncations-RuvB has an N, M, and a C domain (15) (Fig. 1A) and each of these domains may have distinct or overlapping func-tional and/or structural roles in RuvB. The role(s) of each RuvB domain was examined using ruvB mutants that lack one or more of these domains (Fig. 1B). These mutants are called RuvB-N, -NM, -M, -MC, and -C, where the letter indicates the intact domain(s) of the truncated variant. RuvB mutants were tested for complementation of the DNA repair deficiency of E. coli strain HRS2300 (⌬ruvB) and for dominant negative character in wild-type AB1157 at 37 or 26°C. Repair deficiency was assayed by treating wild type and mutant cells with or without UV irradiation. Truncated mutants were expressed in vivo from high copy number expression plasmids constructed with vector pAF101 (see "Experimental Procedures"). Fig. 1, C and D, shows the characteristics of each RuvB truncation mutant. None of the RuvB truncation mutants complemented the UV light sensitivity phenotype of ⌬ruvB at 37 or 26°C (Fig. 1C), but the positive control (wild-type RuvB) did complement the DNA repair defect. In addition, RuvB-N and RuvB-NM demonstrated dominant negative effects on wildtype cells, and the dominant negative phenotype was more severe at 26 than at 37°C (Fig. 1D). The dominant negative phenotype was stronger in cells overexpressing RuvB-NM than in cells overexpressing RuvB-N. RuvB-M, RuvB-MC, and RuvB-C had no dominant negative effects on wild-type cells at 37 or 26°C.
Domain Requirements for RuvB ATPase-To examine the biochemical properties of the mutant RuvB proteins, truncated RuvB mutants were purified by affinity chromatography as GST fusion proteins. After affinity chromatography, the recombinant RuvB moiety was separated from the GST moiety using precision protease cleavage of the fusion protein linker. When overexpressed in E. coli at 37°C, RuvB-NM and RuvB-N were largely insoluble. However, the solubility of these proteins was significantly higher when they were overexpressed at 26 instead of 37°C (data not shown), suggesting that the mutant proteins may fold incorrectly at the higher temperature. This result is consistent with the observation that the dominant negative effects of RuvB-NM and RuvB-N are more severe at 26 than at 37°C. RuvB ATPase activity is required for RuvAB-catalyzed branch migration at Holliday junctions. RuvB ATPase is stimulated by the addition of RuvA or DNA and simultaneous addition of RuvA and DNA synergistically activates the RuvB ATPase (9,34,35). Fig. 2A shows the effect of domain truncation on RuvB ATPase activity at 37°C. Unlike wild-type RuvB, RuvB-NM ATPase is stimulated equally by RuvA or RuvA-DNA and is not stimulated by DNA alone. RuvB-NM also has a temperature-sensitive ATPase that is much weaker than the ATPase of wild-type RuvB at 37°C, but equivalent to the basal wild-type ATPase at 26°C (Fig. 2B). This result may reflect the higher solubility of RuvB-NM at 26 than at 37°C. RuvB-N, -M, -MC, and -C had no detectable ATPase activity at 26°C (Fig.  2C), suggesting that domains N and M are each required but neither is sufficient for RuvB ATPase.
Domain Requirements for RuvB Branch Migration-The capacity of RuvB truncation mutants to carry out branch migration in vitro was assessed using a synthetic Holliday junction as a substrate. Wild-type RuvB dissociated the Holliday junction efficiently at 26°C in the presence of ATP and RuvA; in contrast, none of the RuvB truncation mutants were capable of promoting branch migration or dissociating the synthetic Holliday junction under identical conditions (Fig. 3A). Furthermore, RuvB-N and RuvB-NM inhibited wild-type RuvAB-catalyzed branch migration (Fig. 3B, lanes h-t). This result may reflect the same mechanism that leads to the dominant negative phenotype of RuvB-N and RuvB-NM with respect to repair of UV-induced DNA damage in vivo.
Interaction between Truncated RuvB Mutants and RuvA-The capacity of RuvB truncation mutants to interact with RuvA was examined using gel filtration analysis in the presence or absence of ATP and Mg 2ϩ . In the absence of ATP and Mg 2ϩ , RuvA, wild-type RuvB and the RuvAB complex eluted at positions of 120, 100, and 280 kDa (Fig. 4A), respectively (27). RuvB-N eluted at 70 kDa and the RuvB-N complex with RuvA eluted at 190 kDa (Fig. 4B). RuvB-NM eluted at 90 kDa and the RuvB-NM complex with RuvA eluted at 250 kDa (Fig. 4C). These results indicate that RuvB-N and RuvB-NM form complexes with RuvA. RuvB-M, RuvB-MC, and RuvB-C eluted at 6.7, 6.9, and 5.8 kDa, which may correspond to their predicted molecular sizes of 8, 17, and 9 kDa, suggesting that these proteins probably exist in solution as monomers. The elution of RuvB-MC at 6.9 kDa seems somewhat anomalous; however, the sedimentation coefficiency of RuvB-MC was estimated to be 2.24 S, which predicts a native molecular size of 12 kDa (data not shown). RuvB-MC may take a structure deviated from a sphere in solution. None of the three RuvB truncation mutants, RuvB-M, RuvB-MC, and RuvB-C, showed any shift in the elution positions by the addition of RuvA (Fig. 4, D-F), indicating that they do not form complexes with RuvA.
Domain N Is Required for RuvB-RuvB Intersubunit Interaction-The results of gel filtration analysis of RuvB, RuvB-N, and RuvB-NM indicate that they form oligomers larger than a monomer. Previous sedimentation analysis indicated that RuvB exists as a dimer in the absence of ATP and Mg 2ϩ (35). Glycerol gradient sedimentation analysis was used here to calculate the native molecular size of RuvB truncation variants by the method of Siegel and Monty (33) ( Table I). These data indicate that RuvB, RuvB-N, and RuvB-NM form dimers with the calculated native molecular sizes of 74, 39, and 59 kDa, respectively. These results suggest that domain N includes an interface required for RuvB dimer formation in the absence of ATP and Mg 2ϩ .
Intersubunit interactions of truncated RuvB mutants were also examined in the presence of ATP and Mg 2ϩ . Previous studies have shown that intact RuvB forms a hexamer under these conditions (27 ,36). A small fraction of RuvB-N migrated as an oligomer larger than a dimer, which had an approximate molecular size of 166 kDa. In contrast, the main elution position for RuvB-NM corresponded to an oligomer larger than a dimer, which had an approximate molecular size of 230 kDa. These data suggest that domains N and M are both required for ATP/Mg 2ϩ -induced oligomerization (possibly hexamers) of RuvB.

Effect of Domain Truncation on Formation of a RuvAB Holliday Junction
Complex-RuvAB plays an essential role in driving branch migration of Holliday junctions during homologous recombination and repair. This requires formation of a tripartite RuvAB-Holliday junction (HJ) complex (2,3). The capacity of RuvB-N and RuvB-NM to participate in ternary complex formation in vitro was examined here using a gel shift assay. Fig. 5A shows that RuvB forms a RuvAB-Holliday junction complex, which is detected as a species with slower electrophoretic mobility than the RuvA-Holliday junction complex (Fig. 5A, lanes b-e). In contrast, RuvB-N and RuvB-NM do not form a co-complex with RuvA and a Holliday junction (Fig. 5A,  lanes f-k).
Effect of Domain Truncation on RuvB DNA Binding Activity-The above studies suggest that RuvB-N and RuvB-NM are defective in DNA binding and that domain C plays an important role in promoting interaction with Holliday junctions. Thus, the DNA binding activity of RuvB-N, RuvB-NM, and RuvB-C was examined using a gel shift assay. Intact RuvB binds to and retards the mobility of dsDNA, but the truncated RuvB mutants did not change the electrophoretic mobility of dsDNA (Fig. 5B), suggesting that all three domains are required for binding of RuvB to duplex DNA in vitro.
Importance of Arg-318 for DNA Repair Activity of RuvB-Previous studies indicate that winged helix motifs may include positively charged amino acids that are critical for the DNA binding activity associated with this motif. In earlier studies on RuvB, the RuvB R318C mutant, which has a mutation in the winged helix motif in domain C, was identified as a dominant negative RuvB mutant that inhibits RuvB-catalyzed DNA repair in vivo (14). To further examine the role of Arg-318 in DNA binding, two additional RuvB Arg-318 mutants, R318K and R318A, were constructed by site-directed mutagenesis (Fig.  6A). The ruvB R318K mutant on multicopy plasmid retains DNA repair activity, whereas R318A lacks DNA repair activity and has a dominant negative phenotype with respect to DNA repair in vivo (Fig. 6B). These results suggest that only conservative substitution of Arg-318, such as R318K, support the DNA repair activity of RuvB. Thus, the positively charged character of Arg-318 is required for wild-type RuvB function and activity.
Biochemical Properties of RuvB R318A-RuvB R318A and R318K were purified and their biochemical properties were characterized in vitro. As suggested by in vivo complementation assay, RuvB R318K possessed branch migration and DNA-RuvA-dependent ATPase activities, although the RuvB activities in RuvB-R318K activities were slightly lower than those of wild-type RuvB. At least 4-fold more RuvB R318K protein is required to unwind a comparable amount of Holliday junction substrate or to hydrolyze a comparable amount of ATP (supplemental Fig. 1). On the other hand, RuvB R318A had no detectable level of branch migration activity and inhibited wild-type RuvB-catalyzed branch migration (Fig. 7A, data not  shown). Fig. 7B shows that RuvB R318A has basal ATPase activity in the absence of cofactors, and the basal ATPase activity is stimulated by RuvA but not by DNA. Moreover, the DNA-RuvA-dependent ATPase activity of RuvB R318A is significantly lower than the activity of wild-type RuvB (Fig. 7B). These results suggest that R318A is defective in binding duplex DNA, but it does interact with RuvA. This is consistent with the observation that RuvB R318A does not retard the electrophoretic mobility of duplex DNA, and it does not participate in ternary complex formation with Holliday junction DNA (Fig. 7,  C and D). These results suggested that the positively charged side chain of Arg-318 plays a critical role in DNA binding, and the DNA binding activity of RuvB is essential for the formation of a stable RuvAB complex on Holliday junctions. Thus, the carboxyl-terminally located winged helix motif in RuvB may play an important role in RuvB functions in vitro and in vivo.

DISCUSSION
This study analyzes the functional roles of the N, M, and C domains of RuvB. The results provide evidence for cooperative ATP-dependent interactions between all three RuvB domains during Holliday junction binding and branch migration. This study also supports the following conclusions. 1) The N and M domains of RuvB are required for ATPase activity, 2) RuvB domains N and M play essential roles in formation of RuvB oligomers and assembly of RuvB hexamers, 3) domain N is required for the interaction between RuvB and RuvA, 4) neither the N, NM, nor the C domain is sufficient for DNA binding activity, and 5) RuvB Arg-318, located in the winged helix motif in the C domain, is required for DNA-dependent ATPase activity and dsDNA binding (Fig. 7, B and C).
Domains N and M Constitute a Catalytic Domain of the ATPase Activity-The crystal structure of Tth RuvB shows that the architecture of the N and M domains resembles protein members of the AAA ϩ ATPase family. The fact that the ATPase activity of RuvB requires both N and M domains (Fig. 2) is explained by the fact that ATP binds to the interface between domains N and M in the Tth crystal structure. The RuvB N and M domains also possess Walker A, Walker B, sensor 1, and sensor 2 motifs (Fig. 1B), all of which are essential for ATPase activity in members of the AAA ϩ ATPase family. The fact that RuvB domains N and M are also required for oligomerization of RuvB suggests that ATP-dependent conformational changes FIG. 6. DNA repair activity of RuvB Arg-318 mutants. A, schematic representation of mutations made in the winged helix motif of RuvB. The boxes represent amino acid sequences that are highly conserved among the bacterial RuvBs. The highly conserved arginine residue at position 318 was changed to Lys or Ala (highlighted). B, complementation and negative dominance assays were performed at 37°C as described in the legend to Fig. 1. may play a role in RuvB intersubunit interactions (also see below).
Domain N Includes Interfaces for RuvB-RuvB and RuvA-RuvB Interactions-Previous studies suggested that the ␤-hairpin in domain N is required for the interaction between RuvB and RuvA (17)(18)(19). This is consistent with the data presented here, indicating that the N and the NM domains of RuvB are proficient in binding RuvA, but the M, MC, and C domains of RuvB are deficient in interacting with RuvA. These results suggest that domain N is required for the interaction with RuvA and contributes to the interface between RuvA and RuvB. Gel filtration and sedimentation analysis also showed that the N or the NM domains form RuvB dimers in solution, suggesting domain N is interface of RuvB-RuvB interaction. The crystal structure of Tth RuvB provides evidence that RuvB dimer is formed through the interaction between the ␤-hairpins of the two RuvB molecules (15). However, the ␤-hairpin motif is required for the interaction of RuvB with RuvA. The RuvB dimer formed by the interaction between the ␤-hairpins should be mechanistically different from the polar hexamer formation through a head to tail association of monomer RuvB. Therefore, assembly of RuvB hexamer might require another RuvB-RuvB interface. Indeed, in the presence of ATP/Mg 2ϩ , RuvB-NM oligomerizes, but RuvB-N does not. Furthermore, RuvB-N binds ATP with lower affinity than wild-type RuvB or RuvB-NM (data not shown). Because RuvB requires ATP for the oligomer formation, ATP-induced conformational change of RuvB may play a key role in RuvB-RuvB interaction required for the ATP-dependent hexamers assembly. In summary, domain N plays a role in RuvB-RuvA and RuvB-RuvB interactions, and domain M is required for ATP-dependent hexamer assembly.
Domain C and Arg-318 Are Required for DNA Binding-RuvB-NM possesses intrinsic ATPase activity in the absence of RuvA and DNA; however, unlike the ATPase of RuvB, the ATPase of RuvB-NM is not stimulated by DNA ( Fig. 2A). Moreover, RuvB-NM is defective in DNA binding and branch migration (Figs. 3A and 5B). These results suggest that domain C is required for DNA binding. The domain C includes a winged helix motif, which is associated with DNA binding in SmtB, Histone H5 (15,16), and other DNA-binding proteins. X-ray crystallographic and electron microscopic analyses showed that Tth RuvB Arg-302 (which corresponds to E. coli RuvB Arg-318) faces the central opening of the RuvB hexameric ring, which accommodates duplex DNA (15,19). Results presented here show that ruvB R318K but not ruvB R318A fully complements the DNA repair deficiency of a ⌬ruvB mutant in vivo, and that ruvB R318A has a dominant negative phenotype that inhibits wild-type RuvB activities in vivo (Fig. 6B). Biochemical characterization of RuvB R318A revealed that it was defective in DNA-dependent ATPase activity and dsDNA binding activity (Fig. 7, B and C). These result strongly suggest that Arg-318 is required for and may play a direct role in binding of RuvB to DNA. However, it should be emphasized that the C domain of RuvB has no DNA binding activity by itself. Moreover, although RuvB-NM and RuvB R318A interact with RuvA and form RuvB oligomers in an ATP-dependent manner, they are deficient in forming the RuvAB Holliday junction ternary com-plex. Thus, this study supports the conclusion that domains N, M, and C of RuvB act in a coordinated manner to promote the essential homologous recombination function of RuvB, ATP-dependent binding and branch migration at Holliday junctions.