Proc. Natl. Acad. Sci. USAVol. 93, pp. 10051-10056, September 1996Biochemistry

ATPase activity of Escherichia coli Rep helicase crosslinked tosingle-stranded DNA: Implications for ATPdriven helicase translocation

(kinetics/mechanism/protein-DNA crosslinking/protein oligomerization/energy transduction)

ISAAC WONG AND TIMOTHY M. LOHMANtDepartment of Biochemistry and Molecular Biophysics, Washington University School of Medicine, 660 South Euclid Avenue, Box 8231, St. Louis, MO 63110

Communicated by Carl Frieden, Washington University School of Medicine, St. Louis, MO, May 22, 1996 (received for review March 24, 1996)

ABSTRACT To examine the coupling ofATP hydrolysis tohelicase translocation along DNA, we have purified andcharacterized complexes of the Escherichia coli Rep protein, adimeric DNA helicase, covalently crosslinked to a single-stranded hexadecameric oligodeoxynucleotide (S). CrosslinkedRep monomers (PS) as well as singly ligated (P2S) and doublyligated (P2S2) Rep dimers were characterized. The equilibriumand kinetic constants for Rep dimerization as well as thesteady-state ATPase activities of both PS and P2S crosslinkedcomplexes were identical to the values determined for un-crosslinked Rep complexes formed with dT16. Therefore, ATPhydrolysis by both PS and P2S complexes are not coupled to DNAdissociation. This also rules out a strictly unidirectional slidingmechanism for ATP-driven translocation along single-strandedDNA by either PS or the P2S dimer. However, ATP hydrolysis bythe doubly ligated P2S2 Rep dimer is coupled to single-strandedDNA dissociation from one subunit ofthe dimer, although loosely(low efficiency). These results suggest that ATP hydrolysis candrive translocation of the dimeric Rep helicase along DNA by a"rolling" mechanism where the two DNA binding sites of thedimer alternately bind and release DNA. Such a mechanism isbiologically important when one subunit binds duplex DNA,followed by subsequent unwinding.

DNA helicases are motor proteins that use the chemical energyof nucleoside 5'-triphosphate (NTP) hydrolysis to perform themechanical work of disrupting the base pairs between comple-mentary strands of duplex DNA to form single-stranded DNA(ssDNA) intermediates (1, 2). These molecular motors are es-sential components of most DNA metabolic and processingmachineries in all organisms (2), and mutations in helicasesinvolved in DNA repair processes have been linked to a numberof human skin cancers (3-5). Despite the importance of theseenzymes and the large number of DNA helicases that have beenidentified, mechanistic studies on these enzymes have only begunrecently (for reviews, see refs. 1, 6, and 7).DNA helicases must translocate along DNA to unwind DNA

processively and thus have features in common with motorproteins such as cytoplasmic kinesin (8-10). DNA helicasesappear generally to be oligomeric, primarily dimeric or hexameric(1, 6, 7); for example, the Escherichia coli Rep helicase functionsas a dimer (11), with both subunits able to bind DNA and ATP(10, 12, 13). Of particular mechanistic interest is how helicasestranslocate along DNA and how this is coupled to NTP hydro-lysis. In this regard, most helicases display a macroscopic "polar-ity" in DNA unwinding assays in vitro-i.e., unwinding by mosthelicases is stimulated if the DNA duplex possesses a 5' ssDNA"tail" flanking the duplex (a 5' ->3' helicase) or a 3' ssDNA tail(a 3' ->5' helicase). This observation has led to the suggestionthat helicases couple the energy derived from ATP hydrolysis to

translocate unidirectionally along ssDNA, although direct evi-dence in support of this is lacking. On the other hand, we (14)have shown that unidirectional translocation along ssDNA is notessential for DNA unwinding by the E. coli Rep helicase, a3' ->5' DNA helicase. Furthermore, we have proposed an active,rolling model for Rep-catalyzed DNA unwinding (13) based onour finding that one subunit of the Rep dimer binds directly to theduplex region to be unwound while the other subunit binds tossDNA (13, 14), and we noted that such a rolling model could alsobe envisioned as a mechanism for translocation of the Rep dimeralong ssDNA.To distinguish between these mechanisms for helicase trans-

location along ssDNA, we reasoned that any directional bias intranslocation requires coupling of ATP hydrolysis to movement.In particular, a strictly unidirectional sliding mechanism (no"slippage"), in which the same protein subunit maintains contactwith the ssDNA during translocation, would require tight cou-pling to hydrolysis, while a "rolling" mechanism without direc-tional bias would not. Toward this end, we characterized theATPase activities of covalently crosslinked Rep-single-stranded-oligodeoxynucleotide complexes, reasoning that severe inhibitionof the ATPase activities of such complexes would result ifATP-driven translocation occurs by a strictly unidirectional slid-ing mechanism. Such crosslinked Rep-ssDNA complexes willalso provide a useful means for studying putative intermediatesin the DNA unwinding reaction.

MATERIALS AND METHODSReagents and Buffers. [a-32P]ATP (3000 Ci/mmol; 1 Ci = 37

GBq) was obtained from Amersham. Spectrophotometric gradeglycerol, HPLC grade methanol, and 99% triethylamine wereobtained from Aldrich. Sulfo-N-succinimidyl-6-(4'-azido-2'-nitrophenylamino)hexanoate (sulfo-SANPAH) was obtainedfrom Pierce. Amino-modifier-C2-dT phosphoramidite was ob-tained from Glen Research (Sterling, VA). All solutions weremade with reagent grade chemicals, except as noted above, usingMilli-Q H20-i.e., distilled H20 that was de-ionized using aMilli-Q Water Purification System (Millipore). All binding ex-periments and ATPase assays were carried out in BBM buffer [20mM Tris HCl, pH 7.5 at 4°C/6 mM NaCl/5 mM MgCl2/10%(vol/vol) glycerol]. Triethyl-ammonium bicarbonate buffer(TEAB, 1 M) was made by bubbling CO2 (g) derived fromsubliming dry ice into an aqueous solution containing >1 Mtriethylamine at 0°C for 4 hr or until a pH of 7.5 was achieved;Milli-Q H20 was then added to achieve a final concentration of1 M. Kinase buffer was 50mM Tris HCl (pH 7.5 at 25°C), 10mMMgCl2, and 10 mM 2-mercaptoethanol.

Proteins, Enzymes, and Oligodeoxynucleotides. E. coli Repprotein was purified to >99% homogeneity from E. coli MZ-1/

Abbreviations: ssDNA, single-stranded DNA; sulfo-SANPAH, sulfo-N-succinimidyl-6-(4'-azido-2'-nitro-phenylamino)hexanoate.tTo whom reprint requests should be addressed. e-mail:lohman@bcserv.wustl.edu.

10051

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

10052 Biochemistry: Wong and Lohman

pRepO (15) as described (16). Rep concentration was deter-mined spectrophotometrically using s20 = 7.68 x 104 M-l cm-Ifor Rep monomer (14). An ATPase-deficient mutant of Rep,K281, in which Lys-28 was replaced with Ile, was constructed andpurified to >99% purity (unpublished data). T4 polynucleotidekinase was purchased from United States Biochemical. Oligode-oxynucleotides dT16, d(T3C2T12), d(T7C2T8), d(T12C2T3), whereC2 denotes amino-modifier-C2-dT, were synthesized using anApplied Biosystems model PCR-mate 391 DNA synthesizer andwere purified to >99% homogeneity as described (17) anddialyzed (Spectra/Por 7 MWCO 1000) into Milli-Q H20 forstorage. Oligodeoxynucleotide concentrations were determinedspectrophotometrically in 10 mM Tris-HCl (pH 7.5), 1 mMEDTA, and 150 mM NaCl at 25°C using 8260 = 1.29x105M-l cm-1 per molecule.

Synthesis and Purification of d(T7S2T8). Solid sulfo-SANPAHwas added to d(T7C2T8) (350 ,uM) in 0.5 M Na-carbonate(pH 9.0) buffer to a final concentration of 25 mM andallowed to react in the dark at 25°C for .4 h. Three volumesof ice-cold ethanol was added and the sample chilled at-20°C for 1 h to precipitate the DNA. The large brightorange pellet was collected by centrifugation at 4°C andrinsed with 1 volume of 1,4-dioxane, followed by 1 volume of100% ethanol, then dried and resuspended in Milli-Q H20at 55°C for 5 min. The sample was chilled on ice for 30 minand centrifuged for 15 min at 4°C. The supernatant wastransferred to a fresh microfuge tube, being careful to avoidan orange oily residue that clings to the old microfuge tube.The solution was then purified through a 25 ml Bio-Gel P2column in 50 mM Na-carbonate buffer (pH 9.0). The faintlypink band eluting in the void volume was collected andcontained d(T7S2T8). An equal volume of 1 M TEAB buffer(pH 7.5) was added to the pooled fractions and the sampleloaded onto a 900 mg Maxi-Clean C18 cartridge (AlltechAssociates) activated with 1 ml methanol then equilibratedwith 5 ml 10 mM TEAB (pH 7.5). Following washes of 5 ml10 mM TEAB (pH 7.5) and 1 ml of Milli-Q H20, d(T7S2T8)was eluted with 50% (vol/vol) methanol and dried in theSpeed-Vac. The purified DNA was >97% derivatized asdetermined spectroscopically using an extinction coefficientat 260 nm, 8260 = 1.4 x 105 M-1cm-1. d(T3S2T12) andd(T12S2T3) were prepared in the same manner. Care wastaken to avoid buffers that contained ammonia or primaryamines-e.g., Tris or EDTA-and to minimize exposure ofsamples to light. However, the compound is stable in ambientlight for at least 20 min.

Synthesis and Purification of Crosslinked Rep-d(T7S2T8.Rep protein was dialyzed extensively against 5x bindingbuffer (100 mM Tris, pH 7.5/30 mM NaCl/5 mM EDTA/50% glycerol). Crosslinking was carried out at 1 ,uM Repmonomers and S ,tM d(T7S2T8) in 25 ml 2x binding buffer(40 mM Tris, pH 7.5/12 mM NaCl/2 mM EDTA/20%glycerol) in a shallow 3-inch-diameter dish in the cold room.The sample was UV irradiated for 4-6 h. Buffer containing50 mM Tris (pH 7.5) and 20% glycerol without salt wasadded to replace volume loss due to evaporation, and 5 MNaCl added to a final concentration of 100 mM. This wasloaded onto a 2 ml Macro-Prep Hi-Q anion exchange column(Bio-Rad) equilibrated in 50 mM Tris (pH 7.5), 100 mMNaCl, and 20% glycerol, rinsed with 4 ml of the same buffer,and eluted with buffer containing 1 M NaCl. Fractions of sixdrops each were collected. Pink DNA containing fractionswere pooled (-3 ml) and carefully layered onto a 30 mlBio-Gel P-6 column equilibrated with buffer containing 50mM Tris (pH 7.5), 1 M NaCl, and 10% glycerol running ata 5 ml/h. Fractions of 1.5 ml were collected while monitoringthe absorbance of the eluate at both 254 nm and 280 nm.Crosslinked complexes (A280 > A260) eluted in the voidvolume which was collected and pooled. Buffer containing50 mM Tris (pH 7.5) and 20% glycerol was added to the

sample to 25 ml and the sample was dialyzed for 6 h against1 liter of the same buffer containing 100 mM NaCl. Thedialysate was reloaded onto the Macro-Prep Hi-Q columnthat had been reequilibrated with 100 mM NaCl. The columnwas rinsed with 10 ml buffer containing 100 mM NaCl andeluted with 1 M NaCl as before. The fractions containingprotein and DNA were pooled and dialyzed against 5xbinding buffer. Due to the thermodynamic stability of P2Sand the negative cooperativity of binding DNA to P2S toform P2S2, the purified crosslinked mixture contained a signif-icant fraction of uncrosslinked Rep subunits in the form of P2S.However, all uncrosslinked DNA has been removed.ATPase Assay. ATPase activity was determined at 4°C by

measuring the initial rate of conversion ofATP to ADP using[a-32P]ATP (Amersham) in BBM buffer. Reactions weretypically initiated by addition of 10 ,ul of 10 mM ATP to 90,ul protein. At regular time intervals, 10 ,lI aliquots wereremoved and quenched into 10 ,ul 0.5 M EDTA (pH 8.5).Intervals between time points ranged from S to 20 s depend-ing on the anticipated rate of hydrolysis. Eight to 10 timepoints were taken per assay, but only the linear portion ofeach time course, <60% product formation, was used todetermine the initial velocity. The extent of ADP formationwas monitored by spotting 1 ,lI onto polyethyleneimine-cellulose TLC plates (Merck). TLC plates were developedusing 0.3 M potassium phosphate (pH 7.0) as the mobilephase, dried, and imaged on a Betascope 603 direct ,3-emis-sion imager (Betagen). Spots corresponding to radiolabeledATP and ADP were quantitated using software supplied bythe manufacturer.

Concentration Dependence of the ATPase Activity ofCrosslinked Rep-d(T7S2T8). Crosslinked Rep-d(T7S2T8) was se-rially diluted to span a range of concentrations from 2 ,uM to 27nM. The solutions were allowed to equilibrate at 4°C in the coldroom for 30 min and then assayed for ATPase activity asdescribed above. The resulting isotherm was analyzed by nonlin-ear least-squares methods (18) and the following three equilibria:

K2

K3PS* + P'=±P2S

K52PS*;=±P2S2

For a mixture of native and crosslinked Rep monomers, theconcentration of PS at any concentration of total Rep pro-tomer, PT, is given by the real nonnegative root of Eq. 1:

[p1K2(K2K3 +3K3 + 2K2K5)[pS12[PS]3 + 2K2K3K5

+K2[1 - K3PT(2r - 1)] + 1[PS]2K2K3,K5 [S

rPT2K32K,K5° [1]

where r = 0.61 is the fraction of crosslinked subunits. Theconcentrations of the other species are given by Eqs. 2a-d.

(1 - r)(2K2K4[PS] + K2 + 1)[PS]-P]= (2r - 1)K2K3[PS] + r

[PS*] = K2[PS]

[P2S] = K2K3[PS][P][P2S2] = K22K[PS]2

[2a]

[2b]

[2c]

[2d]

The fit is not sensitive to the isomerization between PS andPS*; however, the equilibrium was included, constrainingK2 =13 as determined by Bjornson et al. (19). The specific ATPase

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 10053

activity of PS and PS* in the fit was constrained at 2 s-1 asdetermined for native uncrosslinked Rep (20).

Kinetics of Transient Formation of P2S2. For the experi-ments measuring the transient formation of P2S2, 75 ,ul of aRep-d(T7S2T8) solution (2 ,AM or 1.66 ,uM) preincubated with0.34 ,uM Rep was mixed with 75 Al of 2x [dT16] at t = 0. At20-s intervals, 10 Al aliquots were removed and mixed with 10,ul 2x [ATP] and allowed to react for 10 s. Ten microliters wasthen transferred into 10 p,l of '0.5 M EDTA to quench thereaction. In this experiment, the time courses were not linearbeyond 10 s. However, within the first 10 s of the reaction, therate of ATP hydrolysis was constant as determined by rapid

k, k2 ki k22P+2Sw- 'PS+P+S tPS*+P+S rPS*+PS ' 2PS*

ki k2 k2 'ji

M4

P2S+S of

k-4P2S2

Scheme I

quench-flow experiments (data not shown). The data wereanalyzed to obtain the rate constants for Scheme I and kcat forP2S2 using FITSIM (21).

All Rep species bound to dT16 were treated as beingequivalent to the crosslinked Rep-d(T7S2T8) species with theexception that the rate of dissociation of dT16 from P2S2 whenthe other Rep subunit was crosslinked to d(T7S2T8) was halvedrelative to when neither subunit was crosslinked. Because ourversion of FITSIM does not allow output factors to be floated,the best fit value of kcat for P2S2 was obtained by executing fitsusing different values of kcat for P2S2 and manually minimizingthe sums of squares of differences.Data Analysis. FITSIM (21) was executed on an IBM PS/2

Model 76. NONLIN (18) was executed on a Hewlett-Packard715. All other data analyses were performed using KALEIDA-GRAPH (Synergy Software, Reading, PA) on an Apple Macin-tosh Quadra 700.

RESULTSBecause only one Rep monomer or one subunit of a dimer canbind to a 16-base oligonucleotide (11, 12), we synthesized threeanalogs of dT16: d(T3S2T12), d(T7S2T8), and d(T12S2T3), eachcontaining a single thymidine with a photoreactive azido-moietyattached to its C5 position (designated S2) as described in Fig. LA.Pilot crosslinking studies were performed, the reaction productswere separated by denaturing SDS/PAGE, and the radioactivebands corresponding to free and crosslinked DNA were quanti-tated (Fig. 1B). These studies indicated that the three oligode-oxynucleotides can be crosslinked to Rep; however, they showeddifferent crosslinking efficiencies, with d(T3S2T12) being the leasteffective (-35% Rep crosslinked) and d(T7S2T8) the most ef-fective (80-90%). We therefore purified mg quantities ofcrosslinked Rep-d(T7S2T8).Although all of the DNA in the "purified" samples of Rep-

d(T7S2T8) was crosslinked to Rep subunits, the sample alsocontained free Rep protein dimerized (noncovalently) tocrosslinked Rep, because DNA binding induces Rep dimeriza-tion. We determined the fraction of crosslinked Rep subunits bydirectly titrating an aliquot of the purified sample, containing 200nM total Rep protomers, with free Rep monomers (containingno DNA) while monitoring the ATPase activity (Fig. 2). TheATPase activity increased linearly on addition of free Repmonomers to a concentration of 42 nM and then plateaued at anactivity of 9.5 s-1 per total Rep monomer, corresponding to an

activity of 19 s-1 per crosslinked P2S. This plateau value isidentical within error to the rate of ATP hydrolysis by theuncrosslinked P25, indicating that the ATPase activity of P2S was

B

_ _ Rep-DNA

4_ free DNA

1 2 3 4FIG. 1. Synthesis of photoreactive crosslinkable single-stranded oli-

godeoxynucleotides. Three analogues of dT16, d(T3C2T12), d(T7C2T8),and d(T12C2T3), were synthesized (A) with amino-C2-modifier-dT phos-phoramidite (denoted C2) containing a primary amine attached via alinker arm to the C5 of thymidine incorporated at positions 4, 8, and 13from the 5' end. These were then chemically coupled to the hetero-bifunctional crosslinker sulfo-SANPAH in the dark via its amino-reactivesuccinimidyl functionality to yield the corresponding photoreactivecrosslinking hexadecamers, d(T3S2Tu2), d(T7S2T8), and d(T12S2T3). In-cubation with Rep in the presence of UV irradiation at 366 nm yieldscovalently crosslinked Rep-DNA complexes as shown by SDS/PAGE(B). Lanes: 1, control without irradiation; 2-4, irradiation products withd(T3S2T12), d(T7S2T8), and d(T12S2T3), respectively.

unaffected by the crosslinking. Under these reaction conditions,the only species present are P2S, PS (including PS*), and P. Theaddition of 42 nM P converted all of the PS to P2S, which led tothe increased ATPase activity. Therefore, 121 nM (one half of200 nM + 42 nM), or 61% of the original 200 nM Rep protomerswere crosslinked to d(T7S2T8).Rep dimerization is induced upon DNA binding (11) such that

the dimerization equilibrium constant increases by at least 4-5orders of magnitude on binding DNA (12, 13) and dimerizationincreases Rep's ATPase activity dramatically (19, 20, 22). Con-sequently, Rep can form multiple complexes with a 16-baseoligodeoxynucleotide, depending on the DNA and Rep concen-

trations. Rep monomers can be free (P) or bound to ssDNA (PS),and two types of Rep dimers can form, P2S and P2S2, where one

or both subunits are bound to ssDNA, respectively (12, 13). These

10.0

(-

uo

-

2

6-

7.5

5.050 100 150

(Rep] added(nM monomers)

FIG. 2. Direct titration to determine fraction of crosslinked Repsubunits. The "purified" Rep-d(T7S2T8) sample contains a mixture ofcrosslinked and uncrosslinked Rep subunits. To determine the fraction ofcrosslinked Rep protomers in this mixture, Rep-d(T7S2T8) (200 AM totalprotomers) was mixed with increasing amounts of free Rep monomers

(BBM buffer at 4°C) and the ATPase activity was monitored.

A

Biochemistry: Wong and Lohman

10054 Biochemistry: Wong and Lohman

are shown schematically in Table 1. Under the conditions of ourexperiments (BBM buffer at 4°C), Bjornson etal. (19) have shownthat ssDNA-induced Rep dimerization proceeds according to themechanism in Scheme I with kinetic and equilibrium constantsshown in Table 2 (19, 22).To determine if DNA crosslinking influenced the energetics of

Rep dimerization, we first examined the kinetics of dissociationof the crosslinked P2S dimer to form P and PS by monitoring thetime-dependent loss of ATPase activity following the addition ofa mutant Rep protein, K281, to trap dissociated crosslinked Repmonomers (PS). K281 is defective in ATPase activity due to a

Lys-28 to Ile substitution in the conserved GX4GKT sequence ofthe putative ATP binding site (I.W., unpublished experiments).This mutant serves as an efficient trap for dissociated PS speciesbecause the heterodimer formed between a wild-type and a K281protomer shows no steady-state ATPase activity (I.W., unpub-lished experiments). Since P2S first undergoes rate-limiting dis-sociation to form PS and P before PS can then dimerize withK281, the time-dependent loss of activity reflects the rate ofdissociation of the dimer. The time course of the ATPase activityof Rep-d(T7S2T8) (0.4 ,uM total Rep protomer) mixed with excess

K281 (2.0 ,uM monomer) is shown in Fig. 34 fitted to a singleexponential with an observed rate constant of 2.5 + 0.04 x10-3s- . This rate constant is identical to that measured fordissociation of native P2S dimer (2.7 + 0.8 x 10-3s-1) byfluorescence stopped-flow experiments under identical solutionconditions (19).The change in the ATPase activity of the crosslinked Rep-

d(T7S2T8) mixture as a function of concentration (27 nM to 2.0,uM total protomer) (Fig. 3B) showed further that the equilibriumconstants for Rep dimerization were unaffected by ssDNAcrosslinking. The increase in ATPase activity with increasingprotein concentration was biphasic with an intermediate plateauregion between 100 to 300 nM. Consistent with the negativecooperativity for DNA binding (12, 13, 22), the intermediateplateau region represents formation of P2S while the secondtransition reflects formation of the P2S2 dimer. Due to the lowsolubility limits of Rep, it was not possible to saturate this secondphase. From nonlinear analysis of the data, we determined a

dimerization constant, K3 = 1.7 + 0.1 x 108 M- 1, for theformation of P2S from PS* and P and an apparent kcat = 18 + 1.2s-1 for ATP hydrolysis by P2S. We were not able to resolve thedimerization constant, K5, for forming P2S2 from 2 PS*, due to thelack of an end-point for the second transition. We were, however,able to obtain a best fit value of kcat = 77 + 9 s-I for ATPhydrolysis by P2S2 by constraining K5 to 1.6 x 105 M-1 as

measured for native Rep (20). The population distribution of PS,PS*, P2S, and P2S2 calculated from these values of K2, K3, and K5are shown overlaid on the fit to the data.We then measured the ATPase activities of each ssDNA

ligation state of the crosslinked Rep to compare with thosedetermined for uncrosslinked complexes (20). The monomer

species, PS, was formed at 1 nM total subunit concentration;based on our previous studies of uncrosslinked Rep-dT16 com-

plexes as well as the value ofK3 determined in the above titration,only monomers, P and PS, are present at this concentration. Since

Table 2. Kinetic and equilibrium constants for proteindimerization of crosslinked versus native Rep

Crosslinked Native

PS* + P 4 P2Sk3, M-'s-' 4.3 ± 0.3 x 105 4.5 + 0.3 x 105*k-3, s-1 2.5 ± 0.04 x 10-3 2.7 + 0.08 x 10-3*K3, M-1 1.7 0.1 x 108 1.7 0.5 x 108*

P2S + S 4 P2S2k4, M-'s-I 394 ± 6 387 ± 3tk-4, 51 <1.1 x 10-3 <1.1 X 10-3t

2PS* 4 P2S2k5, M1 s 1.5 + 0.1 x 103 1.3 ± 0.2 x 103tk-5, 51 6.9 ± 0.1 x 10-3 7.9 ± 0.1 x 10-3tK5, M-1 2.2 ± 0.2 x 105 1.65 ± 0.25 x 105tData are for BBM buffer at 4°C.

*From Bjornson et al. (19).tFrom Wong et al. (20).

the ATPase activity of unligated Rep monomer, P, is 1000-foldlower than that of PS, its ATPase activity is not detectable (20).The ATP concentration dependence of the ATPase activity(BBM buffer at 4°C) of crosslinked PS normalized with respectto total PS concentration (Fig. 44, 0) is well described by arectangular hyperbola with best fit values of kcat = 2.29 + 0.07 s- I

and KM = 2.3 + 0.19 ,uM, and is virtually indistinguishable fromdata obtained with uncrosslinked PS (a) (kcat = 2.17 + 0.04 s-and KM = 2.05 + 0.1 ,uM) (20).The steady-state ATPase activity of the crosslinked P2S dimer,

normalized to the total P2S concentration is shown in Fig. 4B (0).The data was fitted to a rectangular hyperbola, with best fit valuesof kcat = 18.3 ± 0.3 s-I and KM = 3.1 ± 0.2 ,_tM. Also shown aredata previously obtained for uncrosslinked P2S (0) fitted to kcat =16.5 ± 0.2 s-I and KM = 2.7 ± 0.2 ,uM (20). Therefore,crosslinked and uncrosslinked P2S also have identical ATPaseactivities within experimental uncertainty.

Lastly, we measured kcat for ATP hydrolysis by the fullyligated Rep dimer, P2S2, with one or both of the Rep subunitscrosslinked to ssDNA. Due to the negative cooperativity forDNA binding to the Rep dimer, it is difficult to populate asufficiently large fraction of P2S2 at equilibrium. However, wehave previously shown that the ATPase activity of P2S2 can beaccurately determined by measuring the ATPase activity as afunction of time following the addition of excess dT16 to asolution of P2S (20) since this results in a transient accumu-lation of P2S2 prior to re-equilibration to PS (Fig. 5). Theresults from a series of such experiments performed at 0.5 ,uMP2S and 6.25, 12.5, 25, 50, and 100 ,uM dT16 are shown in Fig.5A. The observed transient increase in ATPase activity wasdependent on [dTI6] and correlates well with transient forma-tion of P2S2. The subsequent loss of activity corresponds todissociation of the dimer to yield an equilibrium mixture ofPS*, PS, and P2S2 as shown previously (20). The [dTI6]dependence of the rate of the transient increase gives theapparent association rate constant, k4, for binding ssDNA toP2S. The amplitude of this transient is directly proportional tothe difference between the kcat values of P2S and P2S2. The rate

Table 1. ATPase activities of ssDNA ligation states of crosslinked versus native Rep

kcat, S- KM, AMCrosslinked Native Crosslinked Native

O P 2 x 10-3* 7 x 10-3*o~- PS 2.30 + 0.07 2.17 ± 0.04t 2.30 + 0.19 2.05 +0.1t

8 P2S 18.3 + 0.3 16.5 ± 0.2t 3.1 ± 0.2 2.7 + 0.2tP2S2 68 2 71 2.5t

Data are for BBM buffer at 4°C.*From Moore and Lohman (23).tFrom Wong et al. (20).

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 10055

6.0

0

1 3.00

0.0

10.0

0

1

&-

5.0o

o.o10-3 10-1 101

(Rep"t*, (txM)103

2.5

', 2.00-3

cn 1.0

0.5

0.0

20

.. 15

Ir- 100

l. 5

>

0

0.8 Cw

Iari,

0.4 aS

cr

0o0.0

FIG. 3. Kinetics and thermodynamics of crosslinked dimers. (A) Anexcess (2.0 AM) of Rep mutant, K28I, was added to Rep-d(T7S2T8) (0.4,uM). PS formed from the dissociation of P2S to PS and P is trapped byrapid rebinding to this mutant to form heterodimers which are defectivein ATPase activity. The time dependence of the loss of activity thereforecorresponds to the rate of P2S dimer dissociation and fits to a singleexponential with a rate constant of 2.5 ± 0.04 X 10-3-S1. (B) TheATPase activity of Rep-d(T7S2T8) is very sensitive to protein concentra-tion due to the linkage to Rep dimerization. The solid line represents thenonlinear least-square best fit of the ATPase activities at different proteinconcentrations using kcat values of 2 s-1, 18 s-1, and 77 + 9 s-1 for theATPase activities of PS + PS*, P2S, and P2S2, respectively. Equilibriumdimerization constants K3 and K5 for formation of P2S from PS* + P andP2S2 from 2PS* were 1.7 ± 0.1 x 108 M-1 and 1.5 ± 0.1 x 105 M-1,respectively. The isomerization equilibrium for PS to PS* was fixed at K2= 13. The species distributions are also shown.

of decay of the transient gives the rate of dissociation of P2S2to form 2PS*. These data were globally fitted together withdata obtained at a different protein concentration (not shown)by FITSIM (21) according to Scheme I. Solid lines represent bestfits using k4 = 3.94 0.06 x 102 M-1's-s, k5 = 1.5 + 0.1 x 103M-1-s'1, k-5 = 6.9 0.1 X 10-3s-1, and kcat = 68 2 s-1. Thedissociation rate of DNA from P2S2, k-4, was not resolvedbeyond an upper limit of 1.1 x 10-3 s-'. The rate constant forthe dissociation of P2S to form PS* and P, k-3, was fixed at0.0025 s-1 as determined in the mutant trapping experiment,and the association rate constant, k3, was fixed at 4.3 x 105M-1Ls-, calculated as k_3K3 using a value of K3 = 1.7 x 108M-1 as determined in the titration described above. Fig. SBshows the predicted species population distribution during thetime course of the reaction shown in the top curve of Fig. SAbased on these rate constants, which confirms that the tran-sient rise in ATPase activity results from the transient accu-mulation of P2S2 which then dissociates to form 2PS.

DISCUSSIONTables 1 and 2 summarize the kinetic and equilibrium con-stants for Rep dimerization and the kcat values for ATPhydrolysis by the crosslinked Rep-d(T7S2T8) complexes, PS,P2S, and P2S2, and compares them with those measured for theuncrosslinked complexes formed with dT16 (19, 22). All pa-rameters measured for the crosslinked and the uncrosslinkedspecies are identical within experimental error. Neither theRep dimerization energetics nor the initial velocity steady-state ATPase activities of any of the ssDNA ligation states are

perturbed by DNA crosslinking. The ATPase activity of Repwas also not changed upon crosslinking to either d(T3S2T12) or

d(T12S2T3) (data not shown). These results demonstrate thatATP hydrolysis by Rep is not tightly coupled to DNA disso-

10-' 1U(ATPJ (11M)

103

FIG. 4. Steady-state ATPase activity of crosslinked Rep monomer(PS) and singly ligated Rep dimer (P2S). (A) Crosslinked PS was formedby performing experiments at 1 nM total protomers and ATPase activitywas assayed as a function of [ATP]. The solid line represents best fit ofthe data (0) to a rectangular hyperbola with kcat = 2.3 ± 0.07 s-1 and KM= 2.3 ± 0.19 ,uM. 0, Data for uncrosslinked PS obtained at 1 nM Rep and50 nM dT16. The dotted line represents best fit of the uncrosslinked datato kcat = 2.17 ± 0.04 S-1 and KM = 2.05 ± 0.1 ,uM. (B) Free Repmonomers (0.07 ,uM) were preincubated with 0.33 ,M crosslinkedmixture to form 0.2 uM crosslinked P2S and assayed for ATPase activityas a function of [ATP]. The solid line represents best fit of the data (-)to a rectangular hyperbola with kcat = 18.3 ± 0.08 S-1 and KM = 3.1 ±0.2 ,uM. 0, Data for uncrosslinked P2S obtained at 0.5 ,uM Rep monomersand of 3 nM dT16 (>99% Rep is P2S under these concentrations). Thedotted line represents the best fit of the uncrosslinked data to kt =

16.5 ± 0.2 s-1 and KM = 2.7 + 0.2 ,uM.

ciation or translocation, thus ruling out any mechanisms forhelicase translocation that require tight coupling.We also have shown that the time courses ofATP hydrolysis

by both crosslinked and uncrosslinked P2S dimers remainlinear until >60% of the starting ATP has been converted toADP. This result is consistent with the hypothesis that thessDNA does not dissociate from the P2S dimer during steady-state ATP hydrolysis. Net dissociation of ssDNA from P2Swould result in dimer dissociation and a resulting loss oflinearity in the time course ofATP hydrolysis because the rateof dimerization to reform P2S would be rate-limiting under

25

I-

0

0.,

00

gr

20

15

10

5

25

20

15

10

5

1.0

0.8

0.6

0.4

0.2

0.0

CD

aC.

_.o

o0~

Ct

0

0 50 100 150 200 250Time (s)

FIG. 5. Transient kinetic formation of P2S2 and determination of itsATPase activity. (A) Crosslinked P2S (0.5 ,uM) was mixed with 6.25 (A),12.5 (v), 25 (v), 50 (0), and 100 AM (0) dT16 to form transientlycrosslinked P2S2 at t = 0. At 20-s intervals, the ATPase activity was

measured and was observed to increase transiently followed by a slowerdecay. Solid lines represent best fits to all data sets using the rate constantsshown in Tables 1 and 2 according to Scheme I. (B) The simulatedpopulation distribution of species is displayed for the top curve in Ashowing transient formation of P2S2 from P2S and its dissociation to PS.

B -- ~P2S~*PS* /

IN.

I PS

PS

............

.......

PS

p2

't ''.P2S PS-%_

B. i

Biochemistry: Wong and Lohman

10056 Biochemistry: Wong and Lohman

these reaction conditions. However, because of the steady-state nature of the ATPase assay, this observation of linearityalone does not constitute proof that dissociation does notoccur as there could be other compensating effects such asfrom changes in the protein oligomeric state (20). Thecrosslinking results, however, provide strong corroboratingevidence that the ATPase activity of P2S is not coupled toDNA dissociation. The results are therefore most consistentwith the ssDNA remaining bound to Rep in both its PS and P2Scomplexes during multiple rounds of ATP hydrolysis.Because we have ruled out ATP coupled dissociation of

ssDNA from both P2S and PS, the only remaining possiblemode of translocation in these ligation states, if it occurs at all,would be via sliding (i.e., translocation while the same proteinsubunit maintains contact with the DNA). A related questionis whether ATP hydrolysis fuels unidirectional sliding as amechanism for helicase translocation along ssDNA (24-26).The extent of any directional bias during translocation must bedirectly related to the degree of coupling between ATPhydrolysis and movement. In one extreme, strictly unidirec-tional translocation without "slippage" would require a highcoupling efficiency between ATP hydrolysis and translocationsuch that tethering of the DNA to the helicase by crosslinkingwould be expected to severely inhibit its ATPase activity. In theother extreme, if ATP hydrolysis is independent of DNAmovement, then it could not possibly fuel a directionally biasedtranslocation. Therefore, our results indicate that transloca-tion does not occur by a strictly unidirectional sliding mech-anism since the tethered single-stranded oligodeoxynucleotidewould restrict translocation and thus inhibit ATP hydrolysis.Although we cannot completely rule out a biased directionalsliding mechanism, this would have to occur with low effi-ciency. Furthermore, the fact that the same ssDNA stimulatedATPase activity can be obtained with either dT16 or poly(dT)(20) also indicates that biased directional translocation is notrequired for maximal ATPase activity.

If ATP hydrolysis by a P2S Rep dimer does not stimulateDNA dissociation or translocation by sliding, then how mighta Rep dimer translocate along DNA? Unlike the P2S dimer,ATP hydrolysis by the doubly ligated P2S2 Rep dimer is linearfor only the first "10 s, after which it decreases significantly,indicating that continued ATP hydrolysis by the P2S2 dimerleads to changes in its DNA ligation and/or dimerization states(20, 27). In fact, recent stopped-flow fluorescence measure-ments show that dissociation of a single-stranded oligode-oxynucleotide from one subunit of the P2S2 Rep dimer isenhanced -60-fold during the course of ATP hydrolysis (27).Therefore, the rate of DNA dissociation from one subunit ofP2S2, but not P2S or PS, is enhanced during ATP hydrolysis,indicating that ATP hydrolysis is coupled to DNA dissociationfrom P2S2. However, even for this reaction, an average of 150ATPs are turned over for each net dissociation of DNAindicating a low coupling efficiency (<1%) under our condi-tions (BBM buffer at 4°C). Therefore, DNA dissociation fromP2S2 is not obligatory during each ATPase cycle. This isconsistent with our result that the initial steady-state ATPaseactivity of the P2S2 dimer is unaffected by DNA crosslinking.Based on these results we propose a "rolling" model of

translocation for the dimeric Rep helicase that takes advantage ofthe two distinct DNA binding sites of the dimer (12, 13). In thismodel, net movement results from transient binding ofssDNA tothe unligated subunit of P2S followed by release of ssDNA fromthe first subunit. This is the same type of translocation mechanismthat we previously proposed for the dimeric Rep helicase duringits DNA unwinding reaction (13), the only difference being thattranslocation during unwinding occurs upon binding duplexDNA, D, into the free Rep subunit to form a P2SD intermediate;this duplex DNA is then unwound to transiently form a P2S2complex. Translocation along ssDNA by this mechanism need notoccur with any net directionality. As we have previously proposed

(13), the macroscopic 3' -*5' "polarity" of DNA unwinding bythe Rep helicase may originate from the asymmetry inherent atthe unwinding junction. This is supported by the results reportedhere because a random diffusion process would not require a tightcoupling of energy derived from ATP hydrolysis. Such a rollingmechanism also provides a means for processive translocation ofthe dimeric helicase. Whereas the increase in the rate of disso-ciation of ssDNA from one subunit of the P2S2 Rep dimer uponATP hydrolysis would result directly in an increased rate oftranslocation, the lack of an ATP-stimulated DNA dissociationfrom the P2S dimer would ensure that one subunit of the dimer(although not always the same subunit) remains tightly bound tothe ssDNA lattice throughout the translocation cycle.There is currently no direct evidence to support the common

assumption that unidirectional sliding of helicases along ssDNAis crucial to their mechanism of DNA unwinding. Interestingly,observations that an increase in the length of the ssDNA latticeresults in a net increase in the ATPase activity of DNA helicaseshave been cited as support for ATP-driven unidirectional trans-location (25). In this context, the crosslinked Rep-ssDNA com-plexes described here would approximate the limit of Rep boundto an infinitely long ssDNA lattice, yet we observe no net increasein ATPase activity in these crosslinked complexes relative touncrosslinked complexes of Rep bound to dT16, a single-strandedoligodeoxynucleotide approximately equal in length to the sitesize for the Rep monomer (12). Furthermore, we have also shownthat the #2-fold increase in ATPase activity of Rep bound topoly(dT) versus dT16 is due to an increase in the transientpopulation of the P2S2 species (20).We have previously proposed an active "rolling" model for

Rep helicase-catalyzed DNA unwinding in which the dimersimultaneously binds to both duplex and ssDNA (13). Ourcurrent findings also rule out a strict unidirectional slidingalong ssDNA as a mechanism for a targeted search by the Rephelicase for duplex DNA. We now propose, based on thesefindings, that the Rep helicase can translocate along ssDNAvia a similar rolling model without need for an explicitdirectional bias, although this may occur by an as yet unknownmechanism. The same rolling or subunit switching mechanismcan be readily extended to account for translocation of thering-like hexameric helicases with each subunit providing apotential DNA binding site (1).We thank Bill Van Zante for oligodeoxynucleotide synthesis and

purification, and Keith Bjornson and Janid Ali for critical discussionsand Carl Frieden and Peter Burgers for comments on the manuscript.This work was supported by grants from the National Institutes ofHealth (GM 45948) and the American Cancer Society (NP-756B). I.W.received partial support from an American Cancer Society Postdoc-toral Fellowship (PF-3671).

1. Lohman, T. M. & Bjornson, K P. (1996) Annu. Rev. Biochem. 65, 169-214.2. Matson, S. W. & Kaiser-Rogers, K. A. (1990) Annu. Rev. Biochem. 59, 289-329.3. Friedberg, E. C. (1992) Cell 71, 887-889.4. Sancar, A. (1994) Science 266, 1954-1956.5. Modrich, P. (1994) Science 266, 1959-1960.6. Lohman, T. M. (1992) Mol. Microbiol. 6, 5-14.7. Lohman, T. M. (1993) J. Biol. Chem. 268, 2269-2272.8. Hackney, D. D. (1994) Proc. Natl. Acad. Sci. USA 91, 6865-6869.9. Gilbert, S., Webb, M. R., Brune, M. & Johnson, K. A. (1995) Nature (London) 373,671-676.

10. Moore, K. J. M. & Lohman, T. M. (1995) Biophys. J. 68, 180s-185s.11. Chao, K. & Lohman, T. M. (1991) J. Mol. Biol. 221, 1165-1181.12. Wong, I., Chao, K. L., Bujalowski, W. & Lohman, T. M. (1992) J. Biol. Chem. 267,

7596-7610.13. Wong, I. & Lohman, T. M. (1992) Science 256, 350-355.14. Amaratunga, M. & Lohman, T. M. (1993) Biochemistry 32, 6815-6820.15. Colasanti, J. & Denhardt, D. T. (1987) Mol. Gen. Genet. 209, 382-390.16. Lohman, T. M., Chao, K., Green, J. M., Sage, S. & Runyon, G. (1989)J. Biol. Chem. 264,

10139-10147.17. Lohman, T. M. & Bujalowski, W. (1988) Biochemistry 27, 2260-2265.18. Johnson, M. L. & Frasier, S. G. (1993) Methods Enzymol. 117, 301-342.19. Bjornson, K. P., Moore, K. J. M. & Lohman T. M. (1996) Biochemistry 35, 2268-2282.20. Wong, I., Moore, K. J. M., Bjomson, K. P., Hsieh, J. & Lohman, T. M. (1996) Biochem-

istry 35, 5726-5734.21. Zimmerle, C. T. & Frieden, C. (1989) Biochem. J. 258, 381-387.22. Wong, I., Amaratunga, M. & Lohman, T. M. (1993) J. Biol. ChenL 268, 20386-20393.23. Moore, K. J. M. & Lohman, T. M. (1994) Biochemistry 33, 14550-14564.24. Brown, W. C. & Romano, L. J. (1989) J. Biol. Chem. 264, 6748-6754.25. Young, M. C., Schultz, D. E., Ring, D. & von Hippel, P. H. (1994) J. Mol. Biol. 235,

1447-1458.26. Raney, K. D. & Benkovic, S. J. (1995) J. Biol. Chem. 270, 22236-22242.27. Bjornson, K. P., Wong, I., & Lohman, T. M. (1996) J. MoL. Biol., in press.

Proc. Natl. Acad. Sci. USA 93 (1996)