Branch Migration During Homologous Recombination: Assembly of a Ruvab-Holliday Junction Complex in Vitro

Home , Branch migration, Holliday junction

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

Kevin Hiom and Stephen C. West tion activity, hexamers of RuvB have been found to associ- Imperial Cancer Research Fund ate with tetramers of RuvA (Mitchell and West, 1994), sug- Clare Hall Laboratories gesting that RuvA and RuvB associate in solution prior to South Mimms, Hertfordshire EN6 3LD DNA binding. England In vitro, the RuvA and RuvB proteins exhibit DNA hell- case activity (Tsaneva et al., 1993; Tsaneva and West, 1994). This observation is of particular interest since it Summary suggests that the RuvB ring structure drives branch migration by a reaction that involves transient strand separation. The RuvA and RuvB proteins of E. coli promote the Other DNA helicases, such as the SV40 large T antigen branch migration or movement of Holliday junctions and E. coil DnaB protein, are also known to form hexa- during genetic recombination and DNA repair. Using meric structures on DNA (Mastrangelo et al., 1989; Reha- small synthetic Holliday junctions in which the cross- Kranz and Hurwitz, 1978), indicating that the structural over point is confined near one end of the DNA mole- arrangement observed with RuvB may be common to a cule, we show that RuvAB-mediated branch migration subset of DNA helicases. occurs with a defined polarity. The assembly of RuvA The biochemical properties of RuvA and RuvB are con- and RuvB on the Holliday junction has been investi- sistent with genetic studies that show that ruv mutants are gated by sedimentation analysis and by DNase I foot- defective in a late step in recombination and the recombi- printing. We find that RuvA protein binds and protects national repair of damaged DNA (Benson et al., 1991). The all four strands of DNA at the crossover point, whereas importance of branch migration as a means to generate RuvB protein binds the DNA asymmetrically. The po- heteroduplex DNA, a process central to the exchange of larity of branch migration is defined by the asymmetric genetic information, led us to investigate the mechanism assembly of the RuvAB branch migration complex rel- of protein-driven branch migration. In this paper we report ative to the junction and is consistent with a model in the detection of a stable RuvAB-Holliday junction com- which RuvAB drives branch migration by passing the plex, which we term the RuvAB prebranch migration com- DNA through the hexameric rings of RuvB. plex. Within this complex, RuvA protein contacts all four strands of DNA at the Holliday junction, whereas RuvB Introduction binds to two arms of the flanking DNA. The structural arrangement observed within this prebranch migration com- The DNA damage-inducible RuvA and RuvB proteins of plex indicates that two RuvB ring structures (each ring Escherichia coil are involved in the late stages of genetic encompassing one duplex) interact via RuvA to promote recombination and the recombinational repair of damaged unidirectional branch migration. DNA (Benson et al., 1988, 1991; Lloyd et al., 1984; Sargen- tini and Smith, 1989; Shinagawa et al., 1988). In vitro studies Results have helped define the individual properties and combined functions of these two proteins. The RuvA protein (22 kDa) RuvA Protein Binds the DNA Crossover binds DNA and interacts specifically with Holliday junc- In previous studies, small synthetic Holliday junctions con- tions to form stable RuvA-Holliday junction complexes taining centrally located DNA crossovers were used to (Iwasaki et al., 1992; Parsons et al., 1992). RuvA protein demonstrate branch migration by RuvA and RuvB (Par- is also known to interact with RuvB protein (37 kDa) and sons et al., 1992). Branch migration of these substrates to direct the loading of RuvB onto the Holliday junction occurred in either direction, presumably because the sym- (Parsons and West, 1993). Together RuvA and RuvB pro- metrical nature of the junctions enabled RuvAB to bind in mote the movement of the Holliday junction by a process either orientation. known as branch migration (M011er et al., 1993a; Parsons The specific binding of a DNA crossover by RuvA and et at., 1992; Shiba et al., 1991; Tsaneva et al., 1992b). RuvB can be viewed as the primary step in the assembly Catalysis of branch migration requires ATP hydrolysis, of an active branch migration complex. To investigate the which is provided by the RuvB ATPase activity (Iwasaki structural organization of RuvAB-Holliday junction com- et al., 1989; MOiler et al., 1993b). plexes, we limited the assembly of these complexes to a Electron microscopic studies have shown that RuvB as- single orientation. For this purpose we constructed asym- sembles on duplex DNA as hexameric ring structures in metric junctions by annealing oligonucleotides (88 or 89 which the DNA passes through the hollow core of each ring nt long) to form junctions in which the crossover point was (Stasiak et al., 1994). Though the rings tend to associate in confined near one end of the DNA molecule with flanking pairs on duplex DNA, biochemical studies indicate that duplex arms of 65 bp and 24 bp (Figure 1A). single hexameric rings of RuvB most likely represent the Although RuvA-Holliday junction complexes are stable active form of the protein in the presence of RuvA (Mitchell and can be detected by polyacrylamide gel electrophore- and West, 1994). Under conditions similar to those in sis, previous studies showed that the RuvAB-Holliday which the RuvA and RuvB proteins exhibit branch migra- junction complex could only be detected by this assay Cell 788

A 65nt 24nt I I I -I 5' strand 1 3' strand4

5' X 3' strand 2 3' 5' strand 3 AB ~B lAB AB B

iI i~ + RuvA x 3 DNA 6 + RuvB J =< ,I 0 L ", I£

%1=11!i 6 + RuvA + RuvB ii!iiiiiAil i + ATP i'/ili ii ,~ d_ ,====, , , J 5 10 15 20 5 10 15 20 abcd etgh ijkt rnnop FractionNumber Figure 2. DNaselFootprintingofRuvA-HollidayandRuvAB-Holliday Figure 1. Formationof RuvA-Hollidayand RuvAB-HollidayJunction Junction Complexes Complexes In Vitro Four 5'-32P-labeledjunctions (labeledin strand 1,2, 3, or 4) were incu- (A) Diagramof the asymmetric Hollidayjunction. The crossoverpoint bated with RuvA (0.57 I~M), RuvB (4.8 p.M), or both as described in is fixed as shown. ExperimentalProcedures. Reactionmixtures were then treatedwith (B) Glycerol gradient centrifugationof RuvA and RuvAB complexes DNase I, and the products were analyzedby denaturing polyacryl- with the Holiday junction DNA. Binding reactionswere carried out as amidegel electrophoresis.The 5'-32P-labeledstrand is indicated.The described in ExperimentalProcedures using RuvA (0,57 IIM), RuvB position of the DNA crossover for each junction is marked with an (4.8 p.M), or both, as indicated, and the resulting complexeswere asterisk, and the footprints of RuvA and RuvAB are indicated. No analyzed by centrifugationthrough glycerol gradients. The reaction DNase I protectionby either RuvA or RuvB was observedwith linear shown in graph d also contained 1 mM ATP. The junction was 5'-32P duplex DNA. labeledin strand 2.

after glutaraldehyde fixation in the presence of the nonhy- Holliday junction complexes were formed prior to addition drolyzable analog of ATP, ATPyS (Parsons and West, of DNase I, we observed a footprint that extended approxi- 1993). Because of the inherent instability of the RuvAB- mately 13 nt on either side of the crossover point. This Holliday junction complexes during polyacrylamide gel pattern of protection was observed on all four strands of electrophoresis, in the following analyses we used a com- DNA (Figure 2, lanes c, g, k, and o; data not shown). Under bination of glycerol gradient centrifugation and DNase I the same experimental conditions, we did not observe any footprinting assays to investigate the binding of RuvA and protection of linear duplex DNA, indicating that the foot- RuvB to the Holiday junction. Incubation of 32P-labeled print was due to structure-specific interactions with the junction DNA with RuvA produced DNA-protein com- Holiday crossover (data not shown). The interaction of plexes that sedimented 1.3 times faster than free junction RuvA with the junction is summarized in Figure 3 and DNA through a glycerol gradient (compare graphs a and shows that RuvA bindsthe Holliday junction symmetrically b in Figure 1 B). Gel analyses of fractions from the glycerol about the crossover point, contacting all four flanking du- gradient confirmed that these complexes were identical plex arms. DNase I footprints of symmetrical Hollidayjunc- to RuvA-Hoilidayjunction complexes observed previously tions, in which the crossover point was located centrally, in band shift assays (data not shown). yielded similar results (data not shown). Using DNase I footprinting, we examined the structure of the RuvA-Holliday junction complex by analyzing the Assembly of a RuvAB-Holliday Junction Complex contacts of RuvA with each of the four DNA strands. Four We next examined the interaction of the RuvAB complex identical junctions were prepared, each uniquely 5'-32P with the Holiday junction. When junction DNA was incu- end-labeled in a different DNA strand. The Holliday junc- bated with both RuvA and RuvB and then analyzed by tions were incubated in the absence or presence of RuvA glycerol gradient centrifugation, we observed a complex and treated with DNase I, and the products were analyzed that sedimented 1.5 times faster than free junction DNA by denaturing gel electrophoresis. In the absence of RuvA, (see Figure 1B, graph c). Unlike the band shift assay re- a region of the DNA corresponding to the position of the ported previously, the stability of this complex required crossover was refractory to DNase I digestion, owing to neither the presence of a nonhydrolyzable analog of ATP steric hindrance of the exchanging DNA helices (Figure nor chemical fixation. When reactions were performed in 2, lanes a, e, i, and m) (Murchie et al., 1990). When RuvA- the presence of ATP, complex formation was not seen by Branch Migrationof HollidayJunctions 789

~-- .... RUVB =~4 RuvA ------'~ Figure 3. Summary of DNase I Footprinting Data The footprints of RuvA and Ruv8, determined × in the experimentshown in Figure 2 and other data not shown, are indicated.Footprints were quantified using a phosphorimagerand corre- spond to greaterthan 60% protectionof DNA from digestion by DNase I. The shaded and Branch Migration boxedareas indicate the footprintsof RuvAand RuvB, respectively.The gap in the sequence representsthe crossoverposition. The directionof branch migration,determined in the experimentpresented in Figure5, is also indicated.It was not possibleto map the RuvA footprint, with nucleotideprecision, at the 3' end of strands 2 and 4. This region is indicatedby diffused shading.

glycerol gradient centrifugation, tn this case, we observed analog ATPyS improves the stability of the RuvAB-Holli- RuvAB-mediated branch migration resulting in the dissoci- day junction complex, without supporting branch migra- ation of the DNA substrate (see Figure 1B, graph d). A tion activity (Parsons and West, 1993). To determine number of experiments suggested the presence of both whether the structure of the RuvAB-Hollidayjunction com- RuvA and RuvB within the fast-migrating complex formed plex was altered by the presence of a nucleotide cofactor, in the absence of ATP. First, without addition of RuvA, we we performed further DNase I protection experiments and did not observe complex formation by glycerol gradient compared the RuvAB footprints formed in the absence of centrifugation (data not shown) or by the formation of any nucleotide with those formed in the presence of ATPTS. detectable DNase I footprint resulting from the binding of Using junctions that were 32p labeled in strand 1 or strand DNA by RuvB (see Figure 2, lanes b, f, j, and n). These 2, we observed that the DNase I footprints of the RuvAB results confirm previous observations showing that RuvA complex obtained in the absence or presence of ATP~,S is required for junction binding by RuvB (Parsons and were very similar (Figure 4, lanes c, d, h, and i). One small West, 1993). Second, related experiments in which similar RuvAB-Holliday junction complexes were analyzed by cross reaction with anti-RuvA and anti-RuvB antibodies revealed the presence of both RuvA and RuvB (data not RuvB ATP~S shown). ATP r .... + f The sedimentation coefficient of the RuvAB-Holliday junction complex, in comparison with marker proteins run in parallel, was consistent with a molecular mass of approximately 300 kDa (data not shown). However, quite different results were obtained by gel filtration, which indicated a molecular mass approaching 1000 kDa (data not AB ~ AB shown). This difference most likely arises from the non- globular shape of the RuvAB-Holliday junction complex. When the RuvAB-Holliday junction complex was probed with DNase I, we observed a footprint (see Figure 2, lanes d, h, I, and p) that was considerably larger than that seen with RuvA alone (see Figure 2, lanes c, g, k, and o). Interestingly, the footprint in each strand was extended to one side only of RuvA. The results are summarized in Figure 3. No footprint was observed for RuvAB using linear duplex DNA (data not shown). In the presence of RuvA and RuvB, DNase I protection was observed over 50-54 nucleotides that incorporated the 26-27 nt long footprint exhibited by RuvA. The additional footprint (24- 27 nt) was observed in all four strands that formed the long arms of the Holliday junction DNA substrate. This region of protection was attributed to binding by RuvB. It was not possible to define the intermolecular boundary abcde fghij between RuvA and RuvB molecules, indicating close contacts between the two proteins. We did not observe any Figure 4. Effectof ATPTS and ATP on the RuvAB-HollidayJunction binding by RuvB to the short arms of the junction. Complex DNase I footprintingexperiments were carriedout as describedin the Comparison of RuvAB-Holliday Junction legendto Figure 2 in the presenceor absenceof nucleotidecofactors, Complexes Formed in the Presence as indicated. In these reactions,the junctions were 5'-32P labeled in either strand 1 or strand2. The positionof the DNA crossoverfor each or Absence of ATPyS junction is marked with an asterisk, and the footprints of RuvA and Previous studies showed that the nonhydrolyzable ATP RuvAB are indicated. Cell 790

difference was observed at the 3' side of the RuvAB foot- A print in strand 1, which was more defined in the presence 65nt 24nt of ATP3,S (Figure 4, lane d) than in its absence (lane c). xz~"1 When RuvAB-Holliday junction complexes assembled in g the presence of ATP~,S were analyzed by glycerol gradient centrifugation, we observed that they sedimented one to BranAhP migration / Mg2/ ~eft / RuvAB ~TP/Mg2*Branch migration right two fractions faster than those formed in the absence of cofactor (data not shown). These results may indicate that the presence of ATPTS results in a more compact complex .~~4 ~ without significantly affecting the DNA-binding patterns of RuvA and RuvB. DNase I protection experiments performed following assembly of the RuvAB-Holliday junction complex in the B presence of ATP indicated an increased susceptibility of RuvA - + - + + the DNA to digestion (Figure 4, lanes e and j). Under these RuvB - + + + Markers ATP + + + . . conditions, however, the 5' end of strand 1 and the 3' end of strand 2 remained refractory to cleavage by DNase I.

These data are consistent with RuvAB-mediated branch : junction migration of the junctions, followed by protein dissociation, ~--~*~ partial duplex resulting in the formation of partial duplex molecules with _ .-', single-stranded DNA "tails" that cannot be cleaved by --~ 2 partial duplex DNase I (Parsons et al., 1992; Parsons and West, 1993).

The RuvAB Complex Promotes Unidirectional abcdefg Branch Migration To confirm that the RuvAB complex is capable of branch Figure 5. RuvAand RuvB Promote Unidirectional Branch Migration of the Asymmetric Holliday Junction migration in the presence of ATP and to determine the (A) Schematic diagram indicating the possible consequences of directionality (if any) of the branch migration reaction with branch migration, based on previousexperiments with symmetric Hol- this asymmetric substrate, a junction that was 5'-32P la- liday junctions (Parsons et al., 1992). The 5'-32P label in strand 1 is beled in strand 1 was incubated with RuvA and RuvB in indicated by an asterisk. the presence of ATP. Owing to the asymmetric structure (B) The 5'-~P-labeled junction (as shown in [A]) was incubated with RuvA, RuvB, or both as described in ExperimentalProcedures. RuvA of the junction, the direction of branch migration (i.e., right was present at 0.45 p.M and RuvB at 1.08 p.M (lane c) and 0.54 pM or left, as shown in Figure 5A) could be determined by the (lanes d and e). ATP was included as indicated. The products of the nature of the labeled DNA products. By comparison with reactions were deproteinizedand electrophoresedthrough a 6% neu- DNA markers, produced by annealing oligonucleotides tral polyacrylamidegel, and 32P-labeledDNA was detected by autoradi- used in the preparation of the junction, we found that ography. RuvAB catalyzed branch migration in a rightward direction only (Figure 5, lane e). We did not observe any products mediated branch migration of an asymmetric Holliday that were characteristic of branch migration leftward. As junction is unidirectional and that the polarity of the reac- expected, the branch migration reaction required the tion is defined by the assembly of the RuvAB-Holliday RuvA and RuvB proteins and the presence of ATP (Figure junction complex. DNase I footprinting experiments showed 5, lanes b-d). These results indicate that the asymmetric that RuvA and RuvB proteins bound Holliday junctions as assembly of the RuvAB complex imparts a polarity to the an asymmetric complex with RuvB located on two arms branch migration reaction. that flank the crossover. The interaction of RuvA with Holliday junctions is struc- Discussion ture specific. RuvA binds with high affinity to synthetic Holliday junctions but with much lower affinity to duplex Using an asymmetric synthetic Holliday junction (in which DNA of similar sequence (Iwasaki et al., 1992; Parsons the crossover point is confined near one end of the mole- et al., 1992). in the present experiments, binding of RuvA cule by virtue of the DNA sequences used to construct to the crossover was demonstrated directly by DNase I the junction), we have been able to dissect the structural footprinting (Figure 2). The footprint covered 26-27 nt and organization of the RuvAB-Holiiday junction complex, was symmetric, extending approximately 13 nt on either thus providing insight into the assembly of a branch migra- side of the crossover point on all four DNA strands. Since tion complex and the mechanism of branch migration. RuvA monomers associate to form tetramers in solution Branch migration catalyzed by RuvAB has been demon- (Mitchell and West, 1994; Tsaneva et al., 1992a), this foot- strated previously using either recombination intermedi- print most likely represents the binding of one or more ates generated by RecA (Tsaneva et al., 1992a, 1992b) tetramers to the junction. or synthetic Holliday junctions (Parsons et al., 1992). In The combined interaction of RuvA and RuvB with the both systems, RuvAB-mediated branch migration was bi- Holliday junction produced a DNase I footprint that directional. In this paper we have shown that RuvAB- spanned 50-54 nt along all four DNA strands. This foot- Branch Migrationof HollidayJunctions 791

1994), it appears too small to encompass both duplexes, leading us to propose that each arm of DNA is bound by a separate hexameric RuvB ring (Figure 6). A model for the structure of the RuvAB prebranch migration complex based on the DNase I protection study is presented in Figure 6. In this model, we assume that the functional form of RuvB is a hexameric ring structure (Mitchell and West, 1994; Stasiak et al., 1994) and incorpo- rate the observation that RuvA tetramers associate with direction of branch RuvB hexamers (Mitchell and West, 1994). Since the two migration proteins interact in the absence of DNA, we consider them Figure6. Asymmetric Assembly of a RuvAB-Holliday Junction to be subunits of a RuvAB branch migration enzyme. We Complex suggest that RuvA and RuvB interact together and bind Schematic model for branch migration by RuvA and RuvB showing the relationshipbetween the structure of the RuvAB-Holiidayjunction the junction by virtue of the specificity of RuvA for the complex and the polarity of branch migration. In accord with DNase DNA crossover. This results in the assembly of a RuvAB- I footprintingdata, the RuvA subunit binds the crossover, and RuvB Holliday junction complex in which two RuvB rings bind rings bind two of the four flanking arms. In this scheme, RuvB is as- onto flanking duplex DNA adjacent to RuvA protein (Figure sumed to be a hexamericring structure and RuvAis multimeric(Mitch- 6). Assembly of this complex may require the opening of ell and West, 1994; Stasiaket al., 1994;Tsaneva et al., 1992a). In this model, the asymmetryof the RuvAB complexrelative to the junction a"gate" in the RuvB ring, a reaction that could be facilitated determinesthe polarityof branch migration(left to right, as indicated) by RuvA. A similar reaction occurs during the loading of as the DNA passesthrough the RuvAB complex.With a natural(sym- the ~ subunit of the DNA polymerase III holoenzyme, a metric) Hollidayjunction, complexeswould also be expectedto form hexameric ring structure that requires the efforts of the "y by RuvAB binding the alternate pair of arms. In this case, branch migration would occur in the oppositedirection (i.e., right to left). In complex for its association with DNA (Kuriyan and O'Don- the diagram,the Hollidayjunction is drawn with the arms in a parallel nell, 1993). Alternatively, RuvA may bind the junction and orientation; it could, however,lie antiparallelor square planar. direct subsequent loading of the two RuvB ring structures. Since RuvB protein is a DNA-dependent ATPase (Iwasaki print incorporated the 26-27 nt RuvA footprint and an addi- et al., 1989), each hexameric ring is thought to provide tional 24-27 nt footprint extending along the long duplex the motor for the branch migration reaction. The reaction arms, which was attributed to the region bound by RuvB may involve the rotation of DNA resulting from ATP- (Figure 3). Since this footprint was found in complexes dependent DNA unwinding by RuvAB (Tsaneva et al., formed in the presence or absence of ATP~,S, we believe 1993; Tsaneva and West, 1994). Translocation of the it represents the structure of the RuvAB prebranch migra- RuvAB complex along the DNA (i.e., passage of the DNA tion complex prior to ATP hydrolysis and branch migration. through the RuvAB complex) would then drive the forma- In the absence of RuvA, RuvB protein was unable to tion of extensive lengths of heteroduplex DNA. The asym- bind the junction (Figure 2), and branch migration was metric assembly of the RuvAB complex relative to the not observed (Figure 5B). It is likely that specific protein- crossover, as reported here, would provide the unidirec- protein interactions between RuvA and RuvB are required tionality required for this efficient protein-driven branch for stable assembly of the RuvAB-Holliday junction com- migration reaction. In vivo, RuvAB would be expected to plex. Further evidence for these interactions has been pro- bind a natural (symmetric) Holliday junction in either of vided by the association of RuvA and RuvB to form com- the two possible orientations since RuvA binds in a se- plexes in the absence of DNA, as detected by gel filtration quence-independent manner. In support of this, branch (Mitchell and West, 1994; Shiba et al., 1993). migration of recombination intermediates made by RecA Electron microscopic and three-dimensional image re- or of symmetrical synthetic Holliday junctions has been construction studies have demonstrated that RuvB binds shown to occur in either direction (MLiller et al., 1993a; duplex DNA as hexameric ring structures, which tend to Parsons et al., 1992; Tsaneva et al., 1992b). associate in pairs (Stasiak et al., 1994). The dimensions In previous studies, it was shown that RuvA and RuvB of the double ring suggest that it would encom pass 49 bp of interact with three-armed Y junctions to promote a unidi- B-form DNA (Stasiak et al., 1994). However, biochemical rectional branch migration reaction (Lloyd and Sharpies, studies indicate that the single hexameric ring structure 1993; Tsaneva and West, 1994). We have investigated most likely represents the active form of the RuvB ATPase the interaction of RuvAB with a three-armed structure by (Mitchell and West, 1994), consistent with our DNase I DNase I footprinting (K. H. and S. C. W., unpublished data) protection patterns, which indicate a footprint of approxi- and again find that the RuvAB complex assembles asym- mately 24-27 bp. metrically, with RuvB positioned to one side of RuvA. In DNA binding by RuvB afforded protection to the two long accord with the results described in this paper, the polarity arms of the junction. Since the protection was estimated by of branch migration was directly related to the asymmetry phosphorimaging to be approximately 90%, we propose of the RuvAB footprint. that RuvB binds both flanking duplexes simultaneously (binding to one or the other arm would give only 50% Experimental Procedures protection from DNase I). Given that the central core of Proteins a RuvB ring measures 20-25 ,~, in diameter (Stasiak et al., RuvA and RuvB proteinsof E. coil were preparedas describedpre- Cell 792

viously (Tsaneva et al., 1992a). Protein concentrations were deter- using synthetic Holliday junction DNA (18 ng), and the indicated mined by the Bradford method (Bio-Rad assay kit) using bovine serum amounts of RuvA or RuvB protein or both. Where indicated, 1 mM albumin as a standard and are expressed as moles of monomeric ATP was added. Reactions were incubated at 37°C for 15 rain and protein. loaded directly onto 5 ml of 10%-30% glycerol gradients containing 20 mM Tris-HCI (pH 7.5), 10 mM MgCI2, 1 mM dithiothreitol. Centrifu- DNA Substrates gation was for 16 hr at 45,000 rpm in a Beckman SW50.1 rotor at 4°C. Oligonucleotides were synthesized by phosphoramidate chemistry us- Fractions (approximately 250 Id) were collected, and 32P-labeled DNA ing an Applied Biosystems 394 DNA synthesizer and were purified by was detected by scintillation counting. The protein markers were aldo- reverse phase HPLC. Oligonucleotides requiring further purification lase (156,000), catalase (232,000), ferritin (440,000), and thyroglobulin were subjected to preparative gel electrophoresis. (669,000). Synthetic Holliday junctions were prepared by annealing the appro- priate oligonucleotides essentially as described previously (Parsons Acknowledgments et al., 1990). One strand of each DNA substrate was 5"32P labeled prior to annealing using T4 polynucleotide kinase (Pharmacia) and We thank David Adams, Edward Egelman, Robert Lloyd, Alison Mitch- [7-a2P]ATP (Amersham). The asymmetric Holliday junction was con- ell, Carol Parsons, Andrzej Stasiak, and Irina Tsaneva for interactive structed from the following DNA sequences: oligonucleotide 1,5'-CCA- discussions and communication of results prior to publication. We also GTGATCACATACGCTTTG CTAGGACATCTTGATATCAGCCCACGT- thank Richard Bennett for help with DNase I footprinting experiments. TCACCCG CCTACCAGTGCCACG'I-i'GTATG CCCACGI-IGACC-3'; oli- We are grateful to lain Goldsmith's oligonucleotide synthesis unit for gonucleotide 2, 5'-GGGTCAACGTGGGCATACAACGTGGCACTGGT- providing the DNA and to John Nicholson for photography. This work AGGCGGGTGAACGTGGGCTGATATCAAGATGTCCATCTGTCCGT- was supported by the Imperial Cancer Research Fund. TCATCTATGACGT-3'; eligonucleotide 3, 5'-AACGTCATAGATGAAC- GGACAGATCATGGTGCT1TrAAAGTCTAGAGACTATCGAGCATTA- Received September 12, 1994; revised December 23, 1994. GTACCAGTATCGAATCCGTC'I-rGTCAA-3'; and oligonucleotide 4, 5'-TTTGACAAGAC GGATTCGATACTGGTACTAATGCTCGATAGTCT- References CTAGACTTIAAAAGCACCATGTAGCAAAGCGTATG TGATCACTG-3'. Annealed junctions were purified by gel electrophoresis and their con- Benson, F. E., filing, G. T., Sharpies, G. J., and Lloyd, R. G. (1988). centration determined by calculation of the specific activity using the Nucleotide sequencing of the ruv region of E. coil K-12 reveals a LexA DE81 filter binding method (Sambrook et al., 1989). The partial duplex regulated operon encoding two genes, Nucl. Acids Res. 16, 1541- markers used in the experiment shown in Figure 5B were prepared 1550. by annealing 200 ng of 32P-labeled eligonucleotide 1 with excess oligo- Benson, F. E., Collier, S., and Lloyd, R. G. (1991). Evidence of abortive nucleotide 2 or oligonucleotide 4 as previously described (Parsons recombination in ruv mutants of Escherichia coil K-12. Mol. Gen. Genet. and West, 1990). 225, 266-272. Iwasaki, H., Shiba, T., Makino, K., Nakata, A., and Shinagawa, H. Branch Migration Assay (1989). Overproduction, purification, and ATPase activity of the Esche- Reaction mixtures (20 pl) contained 2.5 ng of 32P-labeled synthetic richia coil RuvB protein involved in DNA repair. J. Bacteriol. 171,5276- Holliday junction DNA in 20 mM Tris-HCI (pH 7.5), 10 mM MgCI2, 1 5280. mM dithiothreitol, 100 i~g/ml bovine serum albumin, 1 mM ATP (unless Iwasaki, H., Takahagi, M., Nakata, A., and Shinagawa, H. (1992). Esch- otherwise stated), and the indicated amounts of RuvA and RuvB pro- erichia coil RuvA and RuvB proteins specifically interact with Holliday teins. Incubation was for 15 rain at 37°C. Reactions were stopped and junctions and promote branch migration. Genes Dev. 6, 2214-2220. deproteinized by the addition of 2 p.I of 10x stop buffer (to a final Kuriyan, J., and O'Dennell, M. (1993). Sliding clamps of DNA polymer- concentration of 20 mM Tris-HCI [pH 7.5], 25 mM EDTA, 0.5% SDS, ases. J. Mol. Biol. 234, 915-925. and 2 mg/ml proteinase K) and incubated for a further 15 rain at 37°C. Lloyd, R. G., and Sharpies, G. J. (1993). Processing of recombination To each reaction, 5 Id of gel loading buffer (25% glycerol containing intermediates by the RecG and RuvAB proteins of Escherichia coil 0.01% bromophenol blue and 0.01% xylene cyanol) was added, and Nucl. Acids Res. 21, 1719-1725, samples were analyzed by electrophoresis through a 6% polyacrylamide gel using a Tris-borate buffer system. ~P-labeled DNA was Lloyd, R. G,, Benson, F. E., and Shurvinton, C. E. (1984). Effect of visualized by autoradiography. ruv mutations on recombination and DNA repair in Escherichia coil K12. Mol. Gem Genet. 194, 303-309. DNase I Footprinting Mastrangelo, I. A., Hough, P, V. C., Wall, J. S., Dobson, M., Dean, Reactions (40 Id) contained 5'-32P-labeled synthetic Holliday junction F. B., and Hurwitz, J. (1989). ATP-dependent assembly of double hex- DNA (5-10 ng) and the indicated amounts of protein in 20 mM Tris- amers of SV40 T antigen at the viral origin of DNA replication. Nature HCI (pH 7.5), 10 mM MgCI2, 5 mM CaCI2, 1 mM dithiothreitol, 100 lig/ 338, 658-662. ml bovine serum albumin, and 200 p.g/ml calf thymus DNA. Binding Mitchell, A. H., and West, S. C. (1994). Hexameric rings of Escherichia reactions were carried out at 37°C for 15 rain and then left for 5 rain coil RuvB protein: cooperative assembly, precessivity, and ATPase at room temperature to equilibrate prior to the addition of DNase I to activity. J. Mol. Biol. 243, 208-215. a final concentration of 15 ng/ml. After incubation for 1 min at room MOiler, B., Tsaneva, I. R., and West, S. C. (1993a). Branch migration temperature, reactions were stopped by addition of 20 p.I of 20 mM of Holliday junctions promoted by the Escherichia coil RuvA and RuvB EDTA. DNA was precipitated with ethanol following addition of tRNA proteins. I. Comparison of the RuvAB- and RuvB-mediated reactions. carrier, washed, and dried. Samples were then quantified for 32p by J. Biol. Chem. 268, 17179-17184. Cerenkov counting and resuspended in denaturing gel loading buffer MOiler, B., Tsaneva, I. R., and West, S. C. (1993b). Branch migration (800/0 formamide, 0.1% bromophenol blue, 0.1% xylene cyanol). The of Hoiliday junctions promoted by the Escherichia coil RuvA and RuvB samples were heated to 95°C for 3 rain, and equal amounts of radioac- proteins. II. Interaction of RuvB with DNA. J, Biol. Chem. 268, 17185- tivity were loaded onto a 12% polyacrylamide gel containing 7 M urea. Following electrophoresis, gels were fixed in 10% methanol, 10% ace- 17189, tic acid before drying on a piece of Whatman 3MM paper. DNA frag- Murchie, A. I. H., Carter, W. A., Portugal, J., and Lilley, D. M. J. (1990). ments were detected by autoradiography and quantified by scanning The tertiary structure of the four-way DNA junction affords protection gels directly with a Molecular Dynamics 425E Phospholmager using against DNase I cleavage. Nucl. Acids Res. 18, 2599-2606. ImageQuant software. The footprint area is defined by greater than Parsons, C..A., and West, S. C. (1990). Specificity of binding to four- 60% protection of DNA from digestion by DNase i. way junctions in DNA by bacteriophage T7 endonuclease I. Nucl. Acids Res. 18, 4377-4384. Glycerol Gradient Centrifugation Binding reactions (70 Id)were carried out in 20 mM Tris-HCI (pH 7.5), Parsons, C. A., and West, S. C. (1993). Formation of a RuvAB-HolUday 10 mM MgCI2, 1 mM dithiothreitol, 100 lig/ml bovine serum albumin. junction complex in vitro. J. Mol. Biol. 232, 397-405. Branch Migration of Holliday Junctions 793

Parsons, C. A., Kemper, B., and West, S. C. (1990). Interaction of a four-way junction in DNA with T4 endonuclease VII. J. Biol. Chem. 265, 9285-9289. Parsons, C. A., Tsaneva, I., Lloyd, R. G., and West, S. C. (1992). Interaction of Escherichia coil RuvA and RuvB proteins with synthetic Holliday junctions. Proc. Natl. Acad. Sci. USA 89, 5452-5456. Reha-Kranz, L. J., and Hurwitz, J. (1978). The dnaB gene product of Escherichia coil: purification, homogeneity and physical properties. J. Biol. Chem. 253, 4043-4050. Sambrook, E. F., Fritsch, E. F., and Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual, Second Edition (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press). Sargentini, N. J., and Smith, K. C. (1989). Role of ruvAB in UV- and y-irradiation and chemical mutagenesis in Escherichia coli. Mutat. Res. 215, 115-129. Shiba, T., Iwasaki, H., Nakata, A., and Shinagawa, H. (1991). SOS- inducible DNA repair proteins, RuvA and RuvB, of Escherichia coil: functional interactions between RuvA and RuvB for ATP hydrolysis and renaturation of the cruciform structure in supercoiled DNA. Proc. Natl. Acad. Sci. USA 88, 8445-8449. Shiba, T., Iwasaki, H., Nakata, A., and Shinagawa, H. (1993). Esche- richia coli RuvA and RuvB proteins involved in recombination repair: physical properties and interactions with DNA. Mol. Gen. Genet. 237, 395-399. Shinagawa, H., Makino, K., Amemura, M., Kimura, S., Iwasaki, H., and Nakata, A. (1988). Structure and regulation of the Escherichia coli ruv operon involved in DNA repair and recombination. J. Bacteriol. 170, 4322-4329. Stasiak, A., Tsaneva, I. R., West, S. C., Benson, C. J. B., Yu, X., and Egelman, E. H. (1994). The Escherichia coil RuvB branch migration protein forms double hexameric rings arcund DNA. Proc. Natl. Acad. Sci. USA 91, 7618-7622. Tsaneva, I. R., and West, $. C. (1994). Targeted versus non-targeted DNA helicase activity of the RuvA and RuvB proteins of Escherichia coli. J. Biol. Chem. 269, 26552-26558. Tsaneva, I. R., Illing, G. T., Lloyd, R. G, and West, S. C. (1992a). Purification and properties of the RuvA and RuvB proteins of Esche- richia coli. Mol. Gen. Genet. 235, 1-10. Tsaneva, I. R., MLiller, B., and West, S. C. (1993). The RuvA and RuvB proteins of Escherichia coil exhibit DNA helicase activity in vitro. Proc. Natl. Acad. Sci. USA 90, 1315-1319. Tsaneva, I. R., Meller, B., and West, S. C. (1993). The RuvA and RuvB proteins of Escherichia coli exhibit D NA helicase activity in vitro. Proc. Natl. Acad. Sci. USA 90, 1315-1319.

Note Added in Proof

In a recent study, the RuvAB-Holliday junction complex was visualized by electron microscopy. The observed structure confirms the data presented here and in addition demonstrates that the Holliday junction lies in a square-planar configuration, with RuvA sandwiched between two hexameric rings of RuvB: Parsons, C. A., Stasiak, A., Bennett, R. J., and West, S. C. (1995). Structure of a multiprotein complex that promotes DNA branch migration. Nature, in press.