Proc. Nadl. Acad. Sci. USA Vol. 91, pp. 9901-9905, October 1994 Biochemistry Dissociation of RecA filaments from duplex DNA by the RuvA and RuvB DNA repair proteins (recombination/branch mlgratlon/beflcase/topolsomerase/DNA supercoiling) DAVID E. ADAMS, IRINA R. TSANEVA, AND STEPHEN C. WEST* Imperial Cancer Research Fund, Clare Hall Laboratories, South Mimms, Herts, EN6 3LD, United Kingdom Communicated by Nicholas R. Cozzarelli, June 15, 1994

ABSTRACT The RuvA and RuvB proteins of Escherichia through regions of tightly bound protein or compacted coUl act late in recombination and DNA repair to catalyze the chromatin (15, 16). A similar problem may occur during the branch migration of Holliday junctions made by RecA. In this late steps of recombination in E. coli in which branch paper, we show that addition of RuvAB to supercoiled DNA migration proteins such as RuvAB or RecG (17) act upon that is bound by RecA leads to the rapid dis tion of the DNA intermediates bound by RecA. Since the RuvAB RecA nucleoprotein filament, as determined by a topological complex enhances the rate of DNA strand exchange when assay that measures DNA underwidin and a restriction added to ongoing RecA reactions (6), it is possible that endonuclease protection assay. Disruption ofthe RecA filament RuvAB actively displaces RecA from DNA. To determine requires RuvA, RuvB, and hydrolysis of ATP. These fndins whether RuvAB can dissociate RecA filaments from duplex suggest several important roles for the RuvAB helicase during DNA we conducted the experiments described in this genetic recombination and DNA repair: (i) dislement of paper. We found that RuvAB promotes the dissociation of RecA filaments from double-stranded DNA, (i) interruption of RecA filaments from duplex DNA and causes transient RecA-mediated strand exchange, (iii) RuvAB-catalyzed unwinding of the DNA double helix. These data are con- branch migration, and (iv) recycling of RecA protein. sistent with a model in which RuvAB displaces RecA from double-stranded DNA to promote branch migration during In Escherichia coli, the RecA protein plays a central role in recombinational repair. recombination by initiating homologous pairing and strand exchange to produce intermediates in which duplex mole- MATERIALS AND METHODS cules are linked by Holliday junctions (1). As a first step, RecA binds to single-stranded orgapped duplex DNA to form Proteins. E. coli RecA, RuvA, and RuvB and wheat germ a nucleoprotein filament in which the DNA is extensively DNA topoisomerase (topo) I were purified as described (18, underwound to 1.5 times the length of B-form DNA. This 19). Restriction enzymes and bovine serum albumin (BSA) underwinding can be detected by topological methods (2, 3) were bought from New England Biolabs and BRL, creatine and may help to promote the formation of multistranded kinase was from Sigma, and proteinase K was from Boeh- in strand ringer Mannheim. Protein concentrations were determined DNA helices that act as intermediates exchange, by the Bradford and Lowry methods using BSA as standard and to create torsional stress that can unwind blocks ofDNA and were confirmed by spectroscopy. Amounts ofprotein are heterology during recombinational repair (4). expressed in moles of monomeric protein. Cells carrying mutations in any of the three ruv genes DNAs. The phagemid pDEA1 (4.44 kb) was derived from (ruvA, ruvB, ruvC) are sensitive to UV light and ionizing the plasmids pBR322, pACYC184, and pGEM-7Zf(+) radiation and exhibit reduced recombination frequencies (5). (Promega). It contains the ColEl and phage fl replication Biochemical studies have shown that the RuvA and RuvB origins and was used for the production of both double- proteins interact with Holliday junctions to promote their stranded and single-stranded DNA. 32P-labeled supercoiled movement along DNA, by a process known as branch DNA (scDNA) was made by growth of pDEA1 in E. coli migration (6-8). The RuvC protein resolves Holliday junc- JM109 cells in the presence of [32P]orthophosphate (370 tions by endonucleolytic cleavage (9, 10). RuvA (22 kDa) and MBq/ml, 10 mCi/ml; Amersham) and purified with a Qiagen RuvB (37 kDa) each play a defined role in branch migration, column. Linear duplex DNA was produced by cleavage at the with RuvA targeting RuvB to the Holliday junction (11, 12). unique Pst I restriction site. DNA concentrations are ex- The selective binding of RuvA to junctions (7, 11), the pressed in moles of nucleotide residues. stimulation of RuvB's ATPase by RuvA/DNA (8), and Topological Assay for Displacement of RecA. To produce observations which show that saturating concentrations of nucleoprotein filaments on duplex DNA, supercoiled 32p- RuvB alone can promote branch migration (13) suggest that labeled pDEA1 DNA (20 KM) was incubated with RecA (20 RuvB is the motor that drives strand exchange during branch puM) in 1 ml of buffer A [27 mM Tris*HCl, pH 7.5/1 mM migration, with RuvA playing an ancillary role (12). Further MgCl2/2 mM dithiothreitol/1.3% (vol/vol) glycerol/0.01% biochemical studies ofRuvAB have revealed a DNA helicase (wt/vol) BSA/27 mM phosphocreatine with creatine kinase activity that unwinds DNA with a 5'-to-3' polarity (14). This at 4 units/ml] containing 2.7 mM ATP. Mixtures were finding may be significant, since DNA helicases could pro- prepared on ice and then transferred to 370C. Following this mote strand exchange through regions of chromatin, blocks initial RecA binding (typically 45 min), MgCl2 was increased of DNA heterology, and DNA lesions during genetic ex- to 12 mM for RuvAB-DNA binding. The higher MgCl2 change and recombinational repair. concentration did not affect the stability of the RecA fila- Several groups have investigated how DNA polymerases ments but inhibited rebinding of RecA to scDNA (20). To do and helicases cope with the problem of translocating Abbreviations: scDNA, supercoiled DNA; BSA, bovine serum al- The publication costs ofthis article were defrayed in part by page charge bumin; ATP[yS], adenosine 5'-[y-thio]triphosphate; topo I, topo- payment. This article must therefore be hereby marked "advertisement" isomerase I. in accordance with 18 U.S.C. §1734 solely to indicate this fact. *To whom reprint requests should be addressed. 9901 Downloaded by guest on September 30, 2021 00A79,02 Biochemistry: Adams et al. Proc. NatL. Acad. Sci. USA 91 (1994) this, aliquots (37.5 ul) were supplemented with 2.5 p4 of0.225 RESULTS M MgCl2 prior to addition of 10 ,ul of a prewarmed mixture RuvA and RuvB Proteins Displace RecA from scDNA. To containing RuvA (10 ,uM) and RuvB (6 ,uM) in buffer B (20 determine whether RuvA and RuvB displace RecA protein mM Tris HCl, pH 7.5/1 mM EDTA/0.5 mM dithiothreitol/ from duplex DNA, we took advantage of the ability of RecA 150 mM NaCl/0.01% BSA/10% glycerol). After 15 min at to bind and underwind negatively supercoiled plasmid DNA 37°C, 20 units of wheat germ topo I (in 2 ,l) was added, an (Fig. 1A). The assay we used was a modification of that amount sufficient to relax protein-free DNA within 30 sec. In described by Iwabuchi et al. (22), in which RecA-mediated most cases, topo was allowed to react for 5 min at 37°C in DNA underwinding produces positive superhelical turns order to ensure complete relaxation. Reactions were stopped which can be relaxed by eukaryotic topo I (indicated sche- by incubation with SDS, EDTA, and proteinase K (0.4%, 40 matically in Fig. 1 B and C). Upon deproteinization, the DNA mM, and 20 uAg/ml, respectively) for 15 min at 37°C. 32p- is seen as a highly negatively supercoiled product, defined as labeled DNA was phenol/chloroform extracted, ethanol pre- form X DNA (Fig. 1D). Addition of RuvAB and consequent cipitated, and analyzed by agarose gel electrophoresis and dissociation of the RecA nucleoprotein filament would be autoradiography. expected to lead to the formation ofrelaxed DNA (rather than Restriction Endonuclease Cleavage Assay. Reaction mixtures form X DNA) after topo I treatment (Fig. 1 F-H). (75 y4) contained 20 ,uM pDEA1 scDNA, 20 ,uM RecA, 1.33 To bind RecA to supercoiled plasmid 32P-labeled DNA, we mM ATP, and 13.3 pM adenosine 5'-[ythio]triphosphate used low concentrations ofMgCI2 followed by a shift up to 12 (ATP[-yS]) in buffer A. After 45 min at 37°C, we added 5 ,u of mM (22). At the higher Mg2+ concentration, the RecA 0.225 mM MgCl2 and 20 1d of a mixture containing 10 uM filament is stable for at least 90 min. Addition of wheat germ RuvA and 6 ,uM RuvB in buffer B. The final MgCl2 concen- topo I led to the formation ofhighly supercoiled form X DNA tration was 12 mM. After 15 min at 37°C, 10 units of Pst I (Fig. 2, lane d), whose agarose gel electrophoretic mobility restriction endonuclease (in 0.5 td) was added to each reaction (after deproteinization) was greater than that of native and incubation was continued for a further 15 min. Reactions scDNA (lane a). Under these conditions, >90%o of the DNA were stopped, deproteinized, and assayed by gel electropho- was converted to form X, indicating the formation of stable resis as described above. The cleavage assay uses low levels nucleoprotein filaments. As expected, in the absence of of ATP[yS] in an otherwise standard RecA-DNA binding RecA, the DNA was completely relaxed by topo I, resulting reaction mixture in order to stabilize RecA filaments formed in a profile oftopoisomers whose gel electrophoretic mobility on double-stranded DNA (21). The extra stability afforded by was slightly positively supercoiled in the gel (lane c). the ATP[IyS] enables bound RecA to shield the DNA from When RuvA and RuvB were added to the nucleoprotein restriction endonuclease cleavage (21) yet did not diminish the filament following the shift to 12 mM MgCl2, the DNA was RuvAB helicase activity (unpublished observations). relaxed by topo treatment, indicating that RecA had been Agarose Gel Electrophoresis. DNA samples were analyzed dissociated (Fig. 2, lane e). The profile of final topoisomer by electrophoresis through 1% (wt/vol) agarose gels in 80 products was the same as that obtained in control reactions mM Tris acetate, pH 7.5/5 mM sodium acetate/1 mM containing RuvAB but without RecA (Fig. 2, lane h). Neither EDTA/0.03% SDS at 1.67 V/cm for 27 hr. Gels were dried RuvA nor RuvB alone was able to dissociate RecA from the and 32P-labeled DNA was visualized by exposure to Kodak DNA (Fig. 2, lanes f and g). In these incomplete reactions, XAR films. Assays were quantitated with a Phosphorlmager form X DNA was observed. Similar results were obtained in (Molecular Dynamics) or by laser densitometry. experiments in which single-stranded DNA fragments were A B C D

a RecA RecA + Topol

sc DNA form X DNA

Topol, RuvAB RuvAB

SDS + ProK RecA RecA

(-) topoisomer E F G H FIG. 1. Diagram of the DNA-unwinding assay used to show RuvAB-mediated dissociation of RecA filaments. Underwinding of negatively supercoiled DNA (A) by RecA results in the introduction ofpositive supercoils (B). Underwinding is initially limited by the positive supercoiling but becomes greater upon topo I-mediated relaxation of the DNA (C). Upon deproteinization, highly underwound DNA adopts a tightly interwound configuration, known as form X DNA (D). The high degree of negative supercoiling in this product differs sharply from that (zero, on average) generated by topo I in the absence of RecA (E). If RuvAB is added to a RecA-DNA binding reaction mixture before (F) or after (G) topo I addition, RuvAB rapidly displaces RecA from the DNA which is relaxed by topo I. Transient unwinding and/or underwinding of the DNA (small bubbles) by RuvAB (F and G) results in partially relaxed negative (-) topoisomer products (H). ProK, proteinase K. Downloaded by guest on September 30, 2021 Biochemistry: Adams et al. Proc. NatL. Acad. Sci. USA 91 (1994) 9903

100- RecA |- | + |+ |+ |+ | -| RuvAl - -- 4t+ -I + l+Il Y75- RuvB 4- +- Topol - -+| + + + I 50- sc 8.88 - oc 4.44 - U U 25- form X 8.88 - - topo- 0 isomers 0 10 20 30 MgCl2 Concentration (mM) FIG. 4. Mg2+ dependence of RecA removal. RecA protein was bound to 32P-labeled scDNA at 1 mM Mg2+. The Mg2+ concentration was then increased to the indicated final concentrations as reactions were taken through the standard DNA-unwinding assay using topo sc 4.44 - I. Reactions contained RecA alone (0); RecA, RuvA, and RuvB (0); RuvA and RuvB only (O); or topo I only (A). The percentage ofform form X4.44 - * X DNA was quantitated by gel electrophoresis followed by phos- a b c d e f g h phorimaging. FIG. 2. Dissociation of RecA filament from scDNA by RuvAB. Mg2+ (Fig. 4). We did not observe substantial disruption of RecA protein was bound to 32P-labeled scDNA by the Mg2+-shift the RecA filament when ATP was replaced with ATP['yS] method as described in Materials and Methods. RuvA and RuvB (Fig. 3, lane i). This result indicates that ATP hydrolysis is were added as indicated prior to treatment with wheat germ topo- required, although we cannot exclude the possibility that the isomerase I. 32P-labeled products were analyzed by agarose gel RecA filament was irreversibly fixed by the ATPyS]. In the electrophoresis followed by autoradiography. The positions of su- presence of 1 mM percoiled (sc) and open circular (oc) pDEA1 plasmid DNA (4.44 kb ATP, dissociation of RecA was only in length) and a small amount ofdimeric DNA (8.88 kb) are indicated. slightly inhibited by the addition of0.5 mM ATP[SJ] (lanej). The relaxed topoisomer products are shown. Control reactions revealed that RuvAB, in the presence of ATP[-v6, bound the supercoiled DNA, resulting in a wide used to load RecA onto the scDNA at 12 mM Mg2+ (data not distribution of topoisomers (lanes l-o). As noted earlier, this shown). observation is consistent with the hypothesis that RuvAB can Unexpectedly, without RecA, the profile of topo I relax- introduce torsional stress into the DNA. ation products obtained in the presence of RuvA and RuvB A time course ofthe standard RecA displacement reaction (Fig. 2, lane h) differed from those obtained with RuvA or is shown in Fig. 5. Within 2 min after RuvA and RuvB RuvB alone (lanes i andj) or with naked DNA (lane c). Most addition, >90%o ofthe substrate DNA was fully relaxed(lanes likely, DNA binding by RuvAB leads to a transient unwind- c-g). We conclude that RuvAB acts rapidly to disrupt RecA- ing and/or underwinding of the substrate, and this small DNA filaments. change was detected by the topoisomerase assay. Requirementfor RuvA and RuvB. To determine the specific Reaction Requirements and ATP Dependence. Dissociation requirement for RuvA and RuvB in mediating RecA disso- of RecA by RuvAB required ATP (Fig. 3) and 7.5-30 mM ciation, the concentration ofone protein was varied while the

+RecA +RecA+RuvAB +RuvAB Topo controls

t i cqn 0.AL CL 2 0~ Q Li U) E E (0) (i + H m m CL m CL' a. EL a. a. L +L + 4CL a.~ .0 F- I +F+ + + + + + + + I + +

sc 8.88 - oc 4.44 S uNno e Id" form X 8.88 - topo- isomers

sc 4.44 - fonr X 4.44 - a a

a b c d e f g h j k m n o p q r s t FIG. 3. Requirements for RuvAB-mediated dissociation of RecA: Effect of ATP, ATP[yS], and low Mg2+. Reactions were performed and the products were analyzed as described in Fig. 2 legend. The final reaction mixtures contained ATP (1 mM) or ATP[S] (0.5 mM) as indicated. In some reactions, the Mg2+ concentration was not increased and remained 1 mM. Topo I was omitted from the sample shown in lane a. Downloaded by guest on September 30, 2021 9904 Biochemistry: Adams et al. Proc. Natl. Acad. Sci. USA 91 (1994)

ITime (min) 01 1 2 L41 71015011 RecA |- +|+ | RuvAB 1-1 . .I I I - oc 4.44 Pst I I-I+I I +I+IM M2iM3 -*"* - .. topo- oc 4.44 isomers - Go linear 4.44 ---_

sc 4.44 - form X 4.44 - sc 4.44 - am_ Relaxed DNA (%) 0 |1 00 00 00|1 00| I 1391|90 form X 4.44 - a b c d e f g a b c d e f g h FIG. 5. Time course of RuvAB-mediated dissociation of RecA FIG. 7. Displacement of RecA from scDNA, as detected by filament. RecA protein was bound to 32P-labeled scDNA by the restriction endonuclease gel assay. RecA was bound to 32P-labeled Mg2+-shift method. RuvA and RuvB were added at 0 min. Thirty plasmid DNA by the Mg2+-shift method prior to addition of RuvAB seconds before the indicated times, 50-g1 aliquots were removed and (as indicated in lane headings). The presence of the RecA filament added to tubes containing 20 units (in 2 p4) of topo I (except lane a, afforded protection ofthe DNA from Pst I, as assayed by agarose gel where topo was omitted). Reactions were stopped and mixtures were electrophoresis and autoradiography. Marker lanes, M1-M3, con- deproteinized after a further 30 sec by addition and incubation with tained supercoiled (sc) substrate (lane f), Pst I-linearized substrate SDS, EDTA, and proteinase K. 32P-labeled products were analyzed (lane g), and form X DNA (lane h). oc, Open circular. by agarose gel electrophoresis followed by autoradiography. The percentage of relaxed DNA was quantitated by laser densitometry we also used a sensitive restriction endonuclease assay which and is indicated below the autoradiograph. measures the protection ofa specific DNA sequence by RecA other was held constant. When RuvA was varied from 0 to 2 (21). Using 10 units of Pst I, which cleaves 32P-labeled ,uM with RuvB held constant at 1.2 AM (at 15 AuM DNA, 15 pDEA1 at a unique site, we observed complete linearization ,pM RecA, and 15 mM MgCl2), the amount of form X DNA ofthe substrate (0.5 yg) in the absence ofother proteins (Fig. decreased sharply between 0.125 AM and 0.5 AM RuvA (Fig. 7, lanes a and b). In the presence of 1 mM ATP and 10 ,IM 6A). In the presence of 7.5 mM MgCl2, the dissociation ATP[yS], 15 uM RecA shielded z40o of the DNA from Pst reaction was less efficient and was proportional to the RuvA I cleavage (lane c). concentration. Similar results were obtained when RuvA was When RuvA and RuvB were added to the RecA-DNA held constant (2 ,uM) and RuvB was varied. In this case, the binding reaction after the formation ofstable RecA filaments, percentage of form X DNA decreased as the RuvB concen- we observed that RecA was displaced as measured by tration increased from 0.075 to 0.6 .uM RuvB (Fig. 6B). At 7.5 cleavage ofthe scDNA by Pst I (Fig. 7, lane d). RuvAB alone mM Mg2+, displacement of RecA by RuvAB was again less did not protect the DNA from cleavage (lane e). These data efficient. These results indicate that the amount of RuvA confirm that RuvAB removes RecA to make the Pst I (0.35 ,tM) and RuvB (0.15 ,uM) required to displace >50% of restriction site accessible for cleavage. the bound RecA is one RuvA tetramer per 85 bp and one RuvB dodecamer per 600 bp. DISCUSSION Restriction Cleavage Assay for Displacement of RecA. The DNA unwinding assay measures global changes in DNA In this work we have shown that the E. coli RuvA and RuvB topology as a result of RecA binding and topo I action. To proteins dissociate RecA protein from supercoiled DNA. confirm that the assay was measuring displacement ofRecA, Displacement of stable RecA filaments required RuvA, RuvB, ATP, and Mg2+. Replacement of ATP with the slowly A B hydrolyzable ATP analog ATP['yS] inhibited RuvAB- 100| 10u mediated RecA displacement, implying a requirement for y0l [RuvB] = 1.2pM 10- A [RuvA] = 2 iM ATP hydrolysis. The results point to a model in which the 1- q 75 e- 75 0 RuvA and RuvB proteins play a dual role late in recombina- tion and repair: (i) to clean up the DNA of proteins used to z z generate the Holliday junction and (ii) to promote branch 50 \ 50 ' migration. Previous studies have shown that RuvA and RuvB play 25 \ 25" independent yet cooperative roles in branch migration. RuvA interacts specifically with Holliday junctions (7, 23) and directs RuvB to the site of a junction (11). Once targeted to 0 0.5 1 1.5 2 0 0.3 0.6 0.9 1.2 DNA, the RuvB ATPase is likely to provide the motor that RuvA Concentration (jiM) RuvB Concentration (jiM) drives branch migration via an active translocation process (12). In recent studies, the RuvB protein was analyzed by FIG. 6. Requirement for RuvA and RuvB in dissociation ofRecA. electron microscopy and found to form a dodecamer on RecA binding and unwinding assays were carried out as described in duplex DNA in which two hexameric encircle the DNA Materials and Methods, except that the concentration of RuvA was rings varied while the RuvB concentration was held constant (A) or the (24). Although it has yet to be established that the double RuvB concentration was varied while RuvA was constant (B). The hexamer represents the functional form of the RuvB motor, final Mg2+ concentrations were 7.5 mM (o) and 15 mM (c). Reactions it is tempting to speculate that dissociation of the RecA were stopped and the 32P-labeled products were analyzed as de- filament results from translocation of RuvB ring structures scribed in Fig. 4 legend. along DNA. Downloaded by guest on September 30, 2021 Biochemistry: Adams et al. Proc. Natl. Acad. Sci. USA 91 (1994) 9905 In the experiments described here, the amounts of RuvA Hildebrandt and Professor Nicholas Cozzarelli (University of Califor- and RuvB required to displace >50o of the RecA filaments nia, Berkeley) for their notes on purification of wheat germ topo I. were significantly lower than the amount of RecA present in the minimal 1. West, S. C. (1992) Annu. Rev. Biochem. 61, 603-640. reaction mixtures. The stoichiometry (one RuvA 2. Ohtani, T., Shibata, T., Iwabuchi, M., Watabe, H., lino, T. & tetramer per 85 bp and one RuvB dodecamer per 600 bp) as Ando, T. (1982) Nature (London) 299, 86-89. determined in Fig. 6 leads us to suggest that DNA binding, in 3. Wu, A. M., DasGupta, C. & Radding, C. M. (1983) Proc. Nati. the absence oftranslocation, is unlikely to be responsible for Acad. Sci. USA 80, 1256-1260. the observed degree of RecA displacement. Indeed, the 4. Jwang, B. R. & Radding, C. M. (1992) Proc. Natl. Acad. Sci. requirements for RuvA and RuvB may be artificially high due USA 89, 75%-7600. 5. Sharples, G. J., Benson, F. E., Illing, G. T. & Lloyd, R. G. to the lack ofa Hollidayjunction which would target RuvAB (1990) Mol. Gen. Genet. 221, 219-226. to the DNA. We currently favor an active RuvAB translo- 6. Tsaneva, I. R., Mailer, B. & West, S. C. (1992) Cell 69, cation model in which RuvB rings unidirectionally translo- 1171-1180. cate along duplex DNA, displacing RecA protein as the DNA 7. Parsons, C. A., Tsaneva, I., Lloyd, R. G. & West, S. C. (1992) passes through the center of the RuvB ring structure. The Proc. Natl. Acad. Sci. USA 89, 5452-5456. physical act of threading double-stranded DNA through the 8. Shiba, T., Iwasaki, H., Nakata, A. & Shinagawa, H. (1991) RuvB with Proc. Natl. Acad. Sci. USA 88, 8445-8449. cavity ofa ring, possibly coupled DNA unwinding 9. Dunderdale, H. J., Benson, F. E., Parsons, C. A., Sharples, (14), could provide a processive mechanism of RecA dis- G. J., Lloyd, R. G. & West, S. C. (1991) Nature (London) 354, placement and branch migration. 506-510. Other possible mechanisms of RuvAB-mediated dissocia- 10. Iwasaki, H., Takahagi, M., Shiba, T., Nakata, A. & Shinagawa, tion of the RecA filament, such as a direct physical interac- H. (1991) EMBO J. 10, 4381-4389. tion between RuvAB and RecA, are not ruled out by the 11. Parsons, C. A. & West, S. C. (1993)J. Mol. Biol. 232, 397-405. 12. West, S. C. (1994) Cell 76, 9-15. current set of experiments. Precedent for such a direct 13. Mailler, B., Tsaneva, I. R. & West, S. C. (1993) J. Biol. Chem. physical interaction between a RecA-like DNA recombinase 268, 17179-17184. and a DNA helicase has been observed between bacterio- 14. Tsaneva, I. R., Miller, B. & West, S. C. (1993) Proc. Natd. phage T4 UvsX protein and Dda helicase (25). Acad. Sci. USA 90, 1315-1319. The active displacement of RecA from duplex DNA by 15. Bonne-Andrea, C., Wong, M. L. & Alberts, B. M. (1990) RuvAB is of particular interest the of the Nature (London) 343, 719-725. given high stability 16. Yancey-Wrona, J. E. & Matson, S. W. (1992) Nucleic Acids RecA filament in the presence ofan adequate supply ofATP. Res. 20, 6713-6721. Although it is known that RecA protein undergoes a transi- 17. Lloyd, R. G. & Sharples, G. J. (1993) EMBO J. 12, 17-22. tion from high- to low-affinity DNA-binding states upon ATP 18. Tsaneva, I. R., Ming, G. T., Lloyd, R. G. & West, S. C. (1992) hydrolysis (26), there is little evidence to suggest that ATP Mol. Gen. Genet. 235, 1-10. hydrolysis and dissociation are coupled. Indeed, experiments 19. Dynan, W. S., Jendrisak, J. J. & Hager, D. A. (1981) J. Biol. by Pugh and Cox (27) showed that RecA protein remains Chem. 256, 5860-5865. 20. Pugh, B. F. & Cox, M. M. (1987) J. Biol. Chem. 262, 1326- associated with the heteroduplex DNA product of strand 1336. exchange well after the reaction was complete. Dissociation 21. Lindsley, J. E. & Cox, M. M. (1989)J. Mol. Biol. 205,695-712. and recycling of RecA may therefore be consequences of 22. Iwabuchi, M., Shibata, T., Ohtani, T., Natori, M. & Ando, T. subsequent RuvAB-mediated processing events that serve to (1983) J. Biol. Chem. 258, 12394-12404. dissociate the RecA filament from DNA. Removal of spent 23. Iwasaki, H., Takahagi, M., Nakata, A. & Shinagawa, H. (1992) RecA protein and catalysis of branch migration would then Genes Dev. 6, 2214-2220. leave the DNA accessible for Holliday-junction resolution by 24. Stasiak, A., Tsaneva, I. R., West, S. C., Benson, C. J. B., Yu, RuvC. data an X. & Egelman, E. H. (1994) Proc. Natl. Acad. Sci. USA 91, Genetic indicate interaction among RuvA, 7618-7622. RuvB, and RuvC in resolution (28) suggesting that the final 25. Kodadek, T. & Alberts, B. M. (1987) Nature (London) 326, stages of Holliday-junction processing-involving RecA re- 312-314. moval, branch migration, and Holliday-junction resolution- 26. Kowalczykowski, S. C. (1991) Annu. Rev. Biophys. Biophys. may well occur via a series of coordinated events in vivo. Chem. 20, 539-575. 27. Pugh, B. F. & Cox, M. M. (1987) J. Biol. Chem. 262, 1337- We thank our colleagues for suggestions, Alison Mitchell for pro- 1343. viding RuvA protein, John Nicholson for photography, and Emily 28. Lloyd, R. G. (1991) J. Bacteriol. 173, 5414-5418. Downloaded by guest on September 30, 2021