Role of Walker Motif A of RuvB Protein in Promoting Branch Migration of Holliday Junctions

Escherichia coli RuvB protein, an ATP-dependent hexameric DNA helicase, acts together with RuvA protein to promote branch migration of Holliday junctions during homologous recombination and recombinational repair. To elucidate the role of the Walker motif A of RuvB (GXGKT; Xindicates a nonconserved residue) in ATP hydrolysis and branch migration activities, we constructed four ruvB mutant genes by site-directed mutagenesis, altering the highly conserved Lys68 and Thr69. K68R, K68A, and T69A mutants except T69S failed to complement UV-sensitive phenotype of theruvB strain. These three mutant proteins, when overexpressed, made the wild-type strain UV-sensitive to varying degrees. K68R, K68A, and T69A were defective in ATP hydrolysis and branch migration activities in vitro. In the presence of Mg2+, K68R showed markedly reduced affinity for ATP, while K68A and T69A showed only mild reduction. K68A and T69A could form hexamers in the presence of Mg2+ and ATP, while K68R failed to form hexamers and existed instead as a higher oligomer, probably a dodecamer. In contrast to wild-type RuvB, K68R, K68A, and T69A by themselves were defective in DNA binding. However, RuvA could facilitate binding of K68A and T69A to DNA, whereas it could not promote binding of K68R to DNA. All of the three mutant RuvBs could physically interact with RuvA. These results indicate the direct involvement in ATP binding and ATP hydrolysis of the invariant Lys68 and Thr69 residues of Walker motif A of RuvB and suggest that these residues play key roles in interrelating these activities with the conformational change of RuvB, which is required for the branch migration activity.

Homologous DNA recombination involves multistep reactions that require many gene products. Much of our knowledge of the molecular mechanisms involved in recombination has been derived from studies of Escherichia coli (1,2). The recombination intermediates called Holliday structures, in which two homologous duplex DNA molecules are held together by a single-stranded crossover (3), are formed by the functions of RecA and several accessory proteins. The Holliday intermediates are processed in a concerted and interactive manner by RuvA, RuvB, and RuvC proteins to give mature products (4 -6).
RuvA, a Holliday junction-specific binding protein, is a tetramer in solution, forms a stable complex with RuvB, and facilitates the binding of RuvB to the junction DNA (7,8). The crystal structure of E. coli RuvA has been determined at the atomic level and it reveals that the four subunits are arranged in a planar flower petal-like structure in the crystal (9,10). More recently, crystal structures of the complexes of E. coli RuvA and a synthetic Holliday junction (11) and Mycobacterium leprae RuvA and a Holliday junction (12) have been reported. In the former structure, the four-way junction DNA was bound by a RuvA tetramer on one face, while in the latter, the junction DNA was sandwiched between two tetramers. It has not been determined which form of the RuvA-junction DNA complex represents the active form in vivo. The RuvA-RuvB complex catalyzes branch migration of Holliday junctions using energy derived from ATP hydrolysis (13,14). RuvB is a helicase that catalyzes unwinding of DNA in a 5Ј to 3Ј direction with respect to single-stranded DNA (ssDNA) 1 in an ATP hydrolysisdependent manner (15). RuvB forms a hexameric ring structure in the presence of Mg 2ϩ and ATP and binds DNA through the central holes of the ring (16,17). However, under the conditions of low Mg 2ϩ and/or low RuvB concentration, RuvB forms a stable dimer (18), suggesting that the dimer is the basic unit of RuvB in forming the hexamer. Thus, RuvB changes its tertiary and quaternary structures by allosteric interactions with various effectors such as Mg 2ϩ , ATP, RuvA, and DNA.
DNA helicases catalyze the unwinding of double-stranded DNA (dsDNA) to produce ssDNA using energy derived from nucleotide 5Ј-triphosphate hydrolysis (19). These enzymes play essential roles in a variety of processes in DNA metabolism, such as replication, recombination, repair, and transcription (20). A large number of helicases have been identified, and comparison of their amino acid sequences has revealed the presence of seven conserved sequence motifs in the majority of them (21), suggesting that many helicases share structural similarities.
Motifs I and II of the helicases correspond to the Walker A and B motifs (22), respectively, found in numerous NTP-binding proteins. The crystal structures of adenylate kinase (23), Ras (24), RecA (25), and F 1 -ATPase (26), for example, show that motif I binds the diphosphate or triphosphate moiety of nucleotides and that motif II is involved in binding nucleotides via Mg 2ϩ . Like all helicases so far characterized, the RuvB family of helicases shares the consensus amino acid sequences for Walker motif A (GXGKT) and B (DEXH) (where X indicates a nonconserved amino acid residue), while the other five common motifs are not clearly identifiable in RuvB. However, from a study of structure-based sequence alignments, the protein architectures of RuvB family helicases are suggested to be closely related to those of the clamp-loader proteins, such as the ␥ subunit of E. coli DNA polymerase III and human RF-C proteins, all of which are involved in the loading of clamp proteins to achieve high processivities of replicative DNA polymerases (27). This suggests the possibility that RuvB, although it is an intrinsic DNA helicase, has a type of protein folding and mechanism of DNA unwinding distinct from those of other typical DEXX DNA/RNA helicase family proteins. Therefore, we use Walker motifs A and B, rather than helicase motifs I and II, throughout this report to designate for the RuvB motifs.
In this study, we constructed four ruvB mutant genes with substitutions in the highly conserved amino acids Lys 68 and Thr 69 in Walker motif A by site-directed mutagenesis and demonstrated that these point mutations affected not only the RuvB activities of ATP hydrolysis and ATP binding, but also those of DNA binding, hexamer formation, and promotion of branch migration. The results suggest that amino acid residues directly involved in binding and hydrolysis of ATP play additional roles in interrelating other functions of RuvB.
Media and Growth Conditions-Bacteria were routinely grown at 37°C in Luria broth medium (32). When needed, ampicillin and chloramphenicol were added to Luria broth at final concentrations of 100 and 15 g/ml, respectively.
Site-directed Mutagenesis-To remove the BglII cleavage site from pAF101, the plasmid was blunt-ended using mung bean nuclease after digestion with BglII and self-ligated. The resultant plasmid was designated pAF102. A synthetic 27-mer (5Ј-AGGTAACATATGATTGAAGC-AGACCGT-3Ј) and 30-mer (5Ј-TGATGGGGATCCGACTTACGGCATT-TCTGG-3Ј) were used for PCR to generate a new NdeI site at the ATG initiation codon and a BamHI site downstream of the TAA termination codon of ruvB in pHS102 (33), respectively. The PCR products containing the ruvB gene were digested with NdeI and BamHI and cloned into the NdeI-BamHI site of pAF102. The KpnI-BglII region of ruvB, which was produced by PCR, was replaced by the KpnI-BglII region of ruvB in pHS102 to produce pTY311 (Fig. 1). The 1.3-kb XbaI-BamHI fragment from pTY311, containing the entire ruvB gene plus the flanking 5Ј sequence, was subcloned into M13mp18 to yield M13mp18RV. Sitedirected mutagenesis by PCR using the four appropriate synthetic 24-mer oligonucleotides and M13mp18RV was carried out to alter codon 68 of ruvB from AAA (Lys) to AGA (Arg) or GCA (Ala) and to alter codon 69 from ACT (Thr) to TCT (Ser) or GCT (Ala). pTY317, pTY318, pTY319, and pTY320 were subsequently constructed by replacing the 0.8-kb KpnI-ClaI fragment of pTY311 with the 0.8-kb KpnI-ClaI fragment from M13mp18RV(K68R), M13mp18RV(K68A), M13mp18RV-(T69S), and M13mp18RV(T69A). To construct low copy number plasmids containing the wild-type ruvB and four ruvB mutant genes, the 1.3-kb XbaI-BamHI fragments of pTY311, pTY317, pTY318, pTY319, and pTY320 were cloned into the XbaI-BamHI site of pSCH19. The DNA sequences of the mutant ruvB genes were confirmed by sequencing the appropriate DNA regions (Applied Biosystems 373S DNA sequencer).
UV Sensitivity Test-Sensitivity to UV irradiation of exponentially growing cells carrying the indicated plasmids was measured as described previously (28).
Purification of Proteins-RuvA proteins were purified as described previously (10). The wild-type and mutant RuvB proteins were overproduced in E. coli HRS2500 using the T7 expression system (29). E. coli HRS2500 strains carrying the plasmids containing wild-type or one of the mutant ruvB genes were grown at 37°C to an A 600 of about 0.4 in 1.5 liter of Luria broth medium containing ampicillin. Isopropyl-␤-Dthiogalactopyranoside was added to a final concentration of 1 mM, and the cultures were incubated for 4 h. The RuvB proteins were purified by the procedures described previously until the step of DEAE-Sepharose column chromatography (34). The fractions containing RuvB proteins were pooled, and ammonium sulfate was added to a final concentration of 70% saturation. The mixtures were stirred for 1 h and centrifuged at 27,000 ϫ g for 15 min. The pellets were suspended in 5 ml of R-buffer (20 mM Tris-HCl, pH 7.5, 7 mM mercaptoethanol, 10% glycerol) containing 1 M NaCl, and the suspensions were applied to a Sephacryl S-400 column (XK 26/70, Amersham Pharmacia Biotech). The RuvB peak fractions were pooled and dialyzed against R-buffer. The dialyzed samples were applied to a 5-ml Hi-trapS column (Amersham Pharmacia Biotech), and the pass-through fractions were applied to a 5-ml Hi-trapQ column (Amersham Pharmacia Biotech). The proteins were eluted with a 100-ml linear gradient from 0 to 1 M NaCl in buffer R, and the peak fractions were pooled and dialyzed against storage buffer (30 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol (DTT), 50 mM KCl, 50% glycerol). The RuvA protein concentration was determined by the Bradford method (Bio-Rad protein assay kit) using bovine serum albumin (BSA) as a standard. The RuvB protein concentration was determined using ⑀280 ϭ 16,900 M Ϫ1 cm Ϫ1 , which was obtained by the method described in Gill and von Hippel (45). This value showed excellent agreement with that obtained by Marrione and Cox (35). Unless stated otherwise, protein concentrations are expressed as moles of monomers.
ATPase ATP Filter Binding Assay-The amount of ATP bound to RuvB was measured essentially as described previously (34). The standard reactions were performed in 20 mM Tris-HCl, pH 7.5, 1 mM DTT, 50 mM NaCl, and various concentrations of ATP and [␣-32 P]ATP as noted, in the presence or absence of Mg 2ϩ (10 mM). Reaction mixtures containing 2 M RuvB were incubated on ice for 15 min and then passed through nitrocellulose filters (0.45 m, Whatman). The filters were washed with the assay buffer and then dried, and the radioactivity of [␣-32 P]ATP bound to RuvB on the filters was measured in a liquid scintillation counter (Beckman). The data obtained were fitted to the equation: is the concentration of ATP, and K d is the dissociation constant.
Gel Retardation Assay-DNA binding activity of RuvB was assayed in a solution (20 l) containing 20 mM triethanolamine HCl, pH 7.5, 15 mM MgCl 2 , 1 mM ATP␥S, 1 mM DTT, 0.01% BSA, and 0.5 g of form I pUC19 DNA. In the case of the reactions to measure the RuvA-assisted DNA binding of RuvB, MgCl 2 was reduced to 5 mM to eliminate the unassisted formation of RuvB-DNA complexes, as described previously (8). The reactions were initiated by the addition of RuvB protein at the indicated concentrations. After incubation for 20 min at 37°C, glutaraldehyde was added to a final concentration of 0.25%. The solutions were further incubated for 30 min, and the reactions were stopped by the addition of 5 l of loading buffer (0.5 M Tris-HCl, pH 7.5, 10 mM EDTA, 40% sucrose, 0.1% bromphenol blue). The formation of the protein-DNA complexes was analyzed by electrophoresis on 0.8% agarose gels, and DNA was stained with ethidium bromide.
Gel Filtration Assay-Gel filtration chromatography was carried out at 15°C using a SMART system (Amersham Pharmacia Biotech). RuvA and RuvB protein samples were applied to a Superdex-200 column equilibrated with buffer A containing 20 mM Tris-HCl, pH 7.5, 1 mM DTT, 50 mM NaCl, and 5% (v/v) glycerol. When required, ATP and Mg 2ϩ were added to buffer A at final concentrations of 0.25 and 15 mM, respectively. To measure the oligomer formation of RuvB, wild-type and mutant RuvBs (35 M) were incubated for 3 h in buffer A containing Mg 2ϩ or Mg 2ϩ and ATP at 4°C, and 10-l aliquots were applied to the column. For the RuvA-RuvB complex formation, RuvA and RuvB were mixed in buffer A to a final concentration of 15 and 10 M, respectively. After the mixtures were incubated at 4°C for 3 h and then at 15°C for 30 min, they were applied to the column. Protein peaks were detected by measuring the absorbance at 280 or 225 nm using a Peak monitor. The following proteins were used as molecular mass standards: blue dextran (void volume), ferritin (440 kDa), aldolase (154 kDa), BSA (67 kDa), ovalbumin (43 kDa).

Construction of the ruvB Mutant Genes with Substitutions of the Walker Motif
A-Walker motif A (GXGKT/S) is highly conserved in many ATPases, including DNA and RNA helicases (21), and it is conserved in all bacterial RuvB homologs so far identified (28). To address the role of the Walker motif A of RuvB in the promotion of branch migration of Holliday junctions, we constructed four mutant ruvB genes (K68R, K68A, T69S, and T69A) with an alterations at the Lys 68 or Thr 69 residues in the Walker motif A of RuvB by site-directed mutagenesis ( Fig. 1) as described under "Experimental Procedures." Complementation and Dominance Test of the Mutant ruvB Genes-Multicopy plasmids carrying the wild-type and mutant alleles were transformed into a ruvB deletion strain, HRS2301, to assay the recombination repair activity and into a ruvB ϩ strain, AB1157, to examine the effects of the mutant ruvB genes on the function of the wild-type gene. pTY319, encoding the T69S mutant allele, fully complemented the DNA repair deficiency of HRS2301 ( Fig. 2A), and all the properties of RuvB (T69S) protein that we studied in vitro were identical with those of wild-type RuvB protein (data not shown), indicating that Thr 69 of Walker motif A is functionally exchangeable with serine. pTY317, pTY318, and pTY320, encoding the K68R, K68A, and T69A alleles, respectively, failed to complement the DNA repair deficiency (Fig. 2A). Furthermore, these three mu-tant genes when overexpressed from the multicopy plasmids made the wild-type strain highly UV-sensitive (Fig. 2B). The mutant ruvB genes K68A and T69A in a low copy number vector, pSCH19, made the wild-type strain as sensitive to UV as they did when in a multicopy vector (Fig. 2C). However, the K68R allele carried in the low copy number vector made the wild-type strain less sensitive to UV than it did when it was carried on the multicopy vector (Fig. 2C). The wild-type ruvB gene did not affect the UV sensitivity of the wild-type strain when carried either on the multicopy or the low copy number vector (Fig. 2, B and C).
The amounts of RuvB proteins expressed from the plasmids were estimated by Western blot analysis using a polyclonal anti-RuvB antibody (data not shown). In both the wild-type and ⌬ruvB strains, all mutant and wild-type RuvB proteins were synthesized at almost the same levels. The amounts of mutant and wild-type RuvB proteins expressed from the multicopy and low copy number plasmids were about 50-and 5-fold higher than the levels expressed from the chromosomal ruvB gene, respectively. RuvB protein is induced about 5-fold from the chromosomal gene by UV. These results suggest that the failure to complement the UV repair deficiency of the ⌬ruvB strain by K68R, K68A, and T69A alleles was not due to the lack of expression or instability of the gene products and that the dominant negative phenotype of the mutant genes for UV repair in vivo was due to the inhibition of the wild-type RuvB function by these mutant proteins coexpressed in the same cells.
The Lack of ATPase and Branch Migration Activities of the RuvB Mutant Proteins-To examine the biochemical properties of the mutant RuvB proteins, the wild-type and mutant ruvB genes were overexpressed in the HRS2500 (⌬ruvABC::cat) strain using the T7 expression system. The proteins were purified to homogeneity as judged by SDS-PAGE (Fig. 3). To assess the contribution of the Walker motif A to the ATPase activity, we performed time course analysis of ATP hydrolysis using the purified mutant proteins. Reactions were carried out in the absence of RuvA and DNA. K68R, K68A, and T69A had no detectable ATPase activity (Fig. 4A). Since the ATPase activity of RuvB is synergistically stimulated by RuvA and DNA, we examined the effects of these cofactors on the ATPase activities of the mutant proteins. The ATPase activity of wildtype RuvB was stimulated by RuvA and DNA individually and synergistically (Fig. 4, B-D), as reported previously (16,36). However, these cofactors did not enhance the ATPase activities of the mutant proteins (Fig. 4, B-D). These results suggest that the mutations altering the conserved Lys 68 and Thr 69 in Walker motif A inactivated the intrinsic ATPase activity of RuvB.
Dissociation of synthetic four-way junctions by the RuvAB complex has been used as a model system to study branch migration activity (14,37). During a 30-min incubation in the presence of RuvA and ATP, wild-type RuvB dissociated 60% of the synthetic Holliday junctions, while K68R, K68A, and T69A showed no detectable branch migration activity (Fig. 5). These results show that the conserved residues in the Walker motif A are critically important for the branch migration activity, which reflects the importance of ATP hydrolysis for this activity.
ATP Binding Activity of RuvB Mutant Proteins-We next examined the ATP binding properties of the mutant RuvB proteins in the presence and absence of Mg 2ϩ using a filter Binding assay (Fig. 6). Samples containing RuvB and ATP were incubated on ice to prevent ATP hydrolysis by wild-type RuvB and then passed through nitrocellulose filters. As shown in Fig. 6, the mutant RuvBs showed different degrees of defects in ATP binding activity. The double-reciprocal plots of the data obtained in the presence of Mg 2ϩ showed that Mg 2ϩ reduced the K d value of wild-type RuvB about 2.5-fold, indicating that Mg 2ϩ enhances the affinity of the RuvB for ATP. K d values of K68R, K68A, and T69A were 66-, 8-, and 12-fold higher than that of wild-type RuvB, respectively (Fig. 6A). The differences in the K d values between wild-type RuvB and the mutant proteins were less pronounced in the absence of Mg 2ϩ than in its presence (Fig. 6B). In the absence of Mg 2ϩ , the K d values of K68R and K68A were 5-and 3-fold higher than that of wildtype RuvB, respectively, and T69A had the same affinity as the wild-type protein. It is intriguing that Mg 2ϩ increased the affinity of wild-type RuvB for ATP, while it decreased the affinities of K68R and T69A and did not significantly change the affinity of K68A.
DNA Binding Activity of RuvB Mutant Proteins-We examined whether mutations in Walker motif A affected the DNA binding activity of RuvB, which should be important for the RuvAB-catalyzed branch migration. The proteins were incubated with form I pUC19 DNA in the presence of 15 mM Mg 2ϩ and 1 mM ATP␥S, conditions that favor the binding of wild-type RuvB to DNA (8). The resultant protein-DNA complexes were fixed with glutaraldehyde to stabilize the weak interaction between DNA and the proteins, and the reaction products were analyzed by agarose gel electrophoresis (Fig. 7). Gel retardation was observed when DNA was incubated with wild-type RuvB. The degree of mobility shift increased with the increasing RuvB concentration (Fig. 7A). In contrast, we could not detect retardation with K68R, K68A, or T69A under the same conditions (Fig. 7A), indicating that these mutants were defective in the ability to bind to dsDNA.
Because previous studies have shown that RuvA facilitates the loading of RuvB onto DNA (8), we examined whether RuvA could load the mutant RuvB proteins onto DNA. RuvA promoted the binding of K68A and T69A to dsDNA, as it did the binding of wild-type RuvB, as shown by the presence of supershift bands of the RuvAB-dsDNA complex (Fig. 7B). In contrast, RuvA could not load K68R onto dsDNA under the same conditions (Fig. 7B). Therefore, although the three mutant proteins are defective in loading onto DNA by themselves, K68A and T69A, but not K68R, can be loaded onto DNA with the help of RuvA.
Complex Formation between the Mutant RuvB Proteins and the RuvA Protein-To directly examine the ability of the mutant RuvB proteins to form complexes with RuvA, the proteins were mixed, incubated, and applied to a Superdex 200 gel filtration column in the absence of ATP and Mg 2ϩ . As shown in Fig. 8A, RuvA was eluted at a position indicating a mass of 105 kDa, corresponding to the tetramer, and wild-type RuvB was eluted at a position indicating a mass of 135 kDa, corresponding to the dimer. The Stokes radius of RuvB calculated from gel filtration was larger than that of a sphercial protein with the molecular mass of RuvB (18). The mixture of RuvA and RuvB eluted at a peak position indicating a mass of 250 kDa. This peak contained both RuvA and RuvB, as shown by SDS-PAGE analysis, indicating formation of a RuvAB complex (Fig. 8A) (10,16). Similarly, the formation of complexes between mutant RuvB proteins and RuvA was analyzed. In the absence of RuvA, K68R, K68A, and T69A were eluted at positions corresponding to 240, 140, and 155 kDa, respectively (data not shown). As shown in Fig. 8B, the mixtures of RuvA and mutant RuvB proteins eluted with molecular masses ranging from 250 to 330 kDa. SDS-PAGE analysis revealed that these peak fractions contained both RuvA and RuvB (data not shown). These results show that all of the mutant RuvB proteins retain the ability to form complexes with RuvA. Oligomeric Structures of Mutant RuvB Proteins-To examine whether the mutations had any effect on the ability of RuvB to form hexameric rings, we investigated the oligomeric states by gel filtration chromatography. Since RuvB hexamer formation is dependent on high protein concentration and cofactors, RuvB proteins at 35 M were applied to a Superdex 200 column in the presence of Mg 2ϩ and ATP (Fig. 9A). Wild-type RuvB was eluted at a position indicating a molecular mass of 230 kDa, corresponding to the RuvB hexamer, in agreement with a previous study (16). K68A and T69A eluted at positions corresponding to molecular masses of 260 and 230 kDa, respectively, indicating that these mutant proteins also formed hexameric ring structures under these conditions. K68A eluted with a broader peak than wild-type RuvB and T69A, suggesting that it contained higher oligomeric species in addition to hexamers. K68R eluted at a position indicating a molecular mass of 430 kDa, corresponding to the dodecamer. We also analyzed the oligomeric states of these mutant proteins in the presence of Mg 2ϩ or EDTA. In the absence of Mg 2ϩ and ATP and the presence of EDTA, K68A and T69A were eluted at a position corresponding to the dimer of RuvB, as was wild-type RuvB (18). However, K68R was eluted at a position corresponding to a molecular mass of 380 kDa (data not shown). In the presence of Mg 2ϩ , wild-type RuvB and all the mutant RuvB proteins were eluted at a position indicating a molecular mass of 430 kDa (Fig. 9B), which corresponds to the size of the RuvB dodecamer, consistent with the findings of a previous study with wild-type RuvB (16). In summary, all the mutant proteins except K68R changed oligomeric states, as did the wild-type protein, in response to the presence of the cofactors ATP and Mg 2ϩ in solution. In contrast, K68R was refractory to the effect of the allosteric cofactor ATP. DISCUSSION In this work, we studied the roles of the conserved residues in the Walker motif A of RuvB in the ATPase and branch migration activities by constructing mutants. One of these, T69S, could fully complement the UV-sensitive phenotype of the ruvB strain, while others, K68R, K68A, and T69A, failed to complement. Overexpression of K68R, K68A, and T69A made the wild-type strain UV sensitive to similar degrees. However, at a lower level of expression, the K68R made the wild-type strain less UV-sensitive than the other two mutants. This may be due to the defective binding of K68R to DNA even in the presence of RuvA, which may make the mutant RuvB-RuvA complex less competitive than the wild type RuvB-RuvA complex for the junction loading.
The three mutant proteins showed markedly reduced ATPase activity in vitro. RuvA and supercoiled DNA stimulated the ATPase activity of wild-type RuvB, but had no stim- ulating effect on the low, intrinsic ATPase activities of these three mutant proteins. The k cat values of K68R, K68A, T69A, and wild-type RuvB for ATPase in the presence of RuvA and DNA were 0.11, 0.16, 0.6, and 30.2 min Ϫ1 , respectively. Thus, the mutations in Walker motif A of RuvB virtually eliminated the fundamental ability to hydrolyze ATP (Fig. 4), as has been shown for many other ATPases (38 -40).
We demonstrated that the mutations in Walker motif A affected the affinity of RuvB for ATP (Fig. 6). In the absence of Mg 2ϩ , K68R and K68A had reduced ability to bind ATP. Mg 2ϩ stimulated the ATP binding ability of wild-type RuvB (2.5-fold decrease in K d ), while it greatly reduced the ATP binding ability of K68R, compared with that of wild-type RuvB (66-fold increase in K d ). Meanwhile, the K d of K68A was unaffected by the addition of Mg 2ϩ . These results suggest that Lys 68 in RuvB plays a key role in the interaction with ATP. X-ray crystallographic analysis of the RecA-ADP complex and F 1 -ATPase indicate that the corresponding lysine residues in the P-loop directly interact with the ␤ and ␥ phosphates of ATP (25,26). The reduced ATP binding of K68A may be due to the loss of positive charge required for the direct interaction with the ␤ and ␥ phosphates of ATP resulting from the substitution of lysine by alanine. K68R not only exhibited reduced ATP binding activity, similarly to K68A, in the absence of Mg 2ϩ , but it also showed a further reduction in ATP binding activity in the presence of Mg 2ϩ . This was surprising because the positive charge of the side chain was maintained. To account for this result, we speculate that some change in tertiary structure is caused by the substitution of lysine with arginine such that the bulky side chain of arginine sterically hinders the ATP binding, and this hindrance may be further enhanced by a Mg 2ϩ -coordinated conformational change around the ATP-binding site of RuvB. Indeed, gel filtration analysis revealed that K68R formed higher oligomeric states (hexamers and dodecamers) under all conditions we studied, unlike the wild-type RuvB. T69A had the same ability to bind ATP as wild-type RuvB in the absence of Mg 2ϩ , consistent with the x-ray crystallographic studies of the RecA-ADP complex and F 1 -ATPase, which revealed that the threonine residue of Walker motif A (GXGKT) interacted with a magnesium ion that bridges the ␤ and ␥ phosphates of ATP. However, T69A had reduced ability to bind ATP in the presence of Mg 2ϩ (5.5-fold increase in K d ). Gel filtration analysis revealed that T69A was able to undergo conformational change in response to Mg 2ϩ (Fig. 9B), indicating that T69A can bind to Mg 2ϩ even without the Thr 69 residue. These results suggest that coordination of Mg 2ϩ not mediated by Thr 69 induces a conformational change that affects the topology around the P-loop. Thus, these equilibrium ATP binding experiments strongly suggest the roles for the lysine and threonine residues of the Walker A sequence of RuvB in ATP binding, which is in good agreement with the x-ray crystallographic data of the Walker A regions in other proteins such as adenylate kinase (23), Ras (24), RecA (25) and F 1 -ATPase (26).
Protein-DNA cross-linking assays revealed that K68R, K68A, and T69A were defective in DNA binding in the presence of ATP␥S. Similar substitutions in the hexameric helicase T7 gene 4 protein (39) also caused defective DNA binding in the presence of dTMP-PCP. However, RuvA could facilitate the loading of K68A and T69A onto DNA. As an explanation for the DNA binding defects of K68A and T69A by themselves, these mutant proteins may not bind ATP␥S or may be defective in the conformational change induced by the ATP␥S, which is necessary for the DNA binding of wild-type RuvB. RuvA may restore these defects of K68A and T69A. Consistent with this idea, it has been shown that RuvA reduced the K m of the RuvB ATPase and enhanced the intimate and continuous contact of RuvB with DNA (36). Electron microscopic and gel filtration studies have shown that RuvB forms a hexameric ring structure as an active form of RuvB that assembles in solution in the presence of ATP and Mg 2ϩ . In this study, gel filtration analysis revealed that K68A and T69A as well as wild-type RuvB could form hexamers in the presence of Mg 2ϩ and ATP and could form complexes with RuvA. These results suggest that these two mutants are capable of undergoing the nucleotide-induced conformational change in the presence of saturating amounts of ATP.
On the other hand, K68R could not bind DNA even in the presence of RuvA, although K68R could form a complex with RuvA (Fig. 7B). Recently, Mezard et al. (41) showed D113N, with a substitution in Walker motif B, was defective in DNA binding both in the presence and absence of RuvA. D113N was also formed hexamers or dodecamers even in the absence of Mg 2ϩ at higher RuvB concentrations (41). K68R, like D113N, formed higher oligomeric states, such as hexamers and dodecamers, under all conditions we studied. Thus, K68R protein may self-aggregate to form hexameric or dodecameric rings in the absence of DNA, which prevents loading onto DNA. These results suggest that the change in oligomeric states induced by nucleotide binding is important for the assembly of RuvB onto DNA.
Although the ATPase activity of RuvB is required for the branch migration activity, the mechanisms by which RuvB protein transmits the energy derived from ATP hydrolysis into the movement necessary for the branch migration are still unclear. It is likely that a conformational change caused by the energy derived from ATP hydrolysis must be coupled to DNA unwinding and translocation. Indeed, nucleotide-induced conformational changes have been reported for several helicases such as E. coli DnaB, Rep, and bacteriophage T7 gene 4 proteins (42)(43)(44). In this study, we have shown that all of the substitution mutations in the conserved lysine and threonine residues in Walker motif A, except T69S, not only affect the ATP binding affinity and inactivate the ATPase activity, but also affect the DNA binding. In addition, K68R affected the oligomer formation. These results suggest that these residues may play key roles in interconnecting ATP binding and ATPase activities with DNA binding and oligomerization through nucleotide-induced conformational changes, all of which are required for the branch migration activity.