Structure, Vol. 9, 47–55, January, 2001, 2001 Elsevier Science Ltd. All rights reserved. PII S0969-2126(00)00552-9

Phe217 Regulates the Transfer of AllostericInformation across the Subunit Interfaceof the RecA Protein Filament

UvsX from bacteriophage T4 [4] and both yeast andhuman Rad51 [5, 6] show that such an oligomeric fila-ment is a common structure found in RecA homologsamong a wide variety of prokaryotic and eukaryotic or-ganisms. RecA function is regulated in classic allosteric

Julie Kelley De Zutter, Anthony L. Forget,Karen M. Logan, and Kendall L. Knight*Department of Biochemistry and Molecular

PharmacologyUniversity of Massachusetts Medical School55 Lake Avenue North fashion, whereby the binding of ATP greatly enhances

the protein’s affinity for ssDNA binding and activates itWorcester, Massachusetts 01655for its catalytic activities [7, 8]. Recent work suggeststhat this effect of ATP on DNA affinity is a result of anincrease in the cooperative assembly of the protein itselfSummaryrather than an increase in the intrinsic affinity for DNA[9]. Analysis of the RecA crystal structure shows thatBackground: ATP-mediated cooperative assembly of

a RecA nucleoprotein filament activates the protein for 55 of the 303 residues in the structure participate insubunit-subunit interactions [10]. Our studies of differ-catalysis of DNA strand exchange. RecA is a classic

allosterically regulated enzyme in that ATP binding re- ent areas of the subunit interface have addressed twofundamental questions: (i) which of the intersubunit con-sults in a dramatic increase in ssDNA binding affinity.

This increase in ssDNA binding affinity results almost tacts observed in the structure are actually importantfor RecA function and oligomer stability and (ii) whichexclusively from an ATP-mediated increase in coopera-

tive filament assembly rather than an increase in the specific residues are important for the transmission ofallosteric information across the protein’s oligomeric in-inherent affinity of monomeric RecA for DNA. Therefore,

certain residues at the subunit interface must play an terface. Initial work focused on a region defined by resi-dues 213–222, which shows five side chains that makeimportant role in transmitting allosteric information

across the filament structure of RecA. specific contacts with positions in the neighboring sub-unit. Through mutational analyses and the use of engi-neered disulfides [11–13] we have demonstrated theResults: Using electron microscopic analysis of RecA

polymer formation in the absence of DNA, we show that importance of one particular residue within this region,Phe217, for maintenance of both the structural and func-while wild-type RecA undergoes a slight decrease in

filament length in the presence of ATP, a Phe217Tyr tional integrity of the protein. While a Phe217Tyr muta-tion maintains wild-type recombination activities in vivo,substitution results in a dramatic ATP-induced increase

in cooperative filament assembly. Biosensor DNA bind- all other substitutions result in a significant decrease orcomplete inhibition of RecA function [11]. In the presenting measurements reveal that the Phe217Tyr mutation

increases ATP-mediated cooperative interaction be- study, we compare various biochemical properties ofwild-type RecA and the Phe217Tyr mutant protein.tween RecA subunits by more than 250-fold.These data identify Phe217 as a key residue within theRecA subunit interface required for the transmissionConclusions: These studies represent the first identifi-

cation of a subunit interface residue in RecA (Phe217) of ATP-mediated allosteric information throughout theoligomeric structure of the protein.that plays a critical role in regulating the flow of ATP-

mediated information throughout the protein filamentstructure. We propose a model by which conformational Resultschanges that occur upon ATP binding are propagatedthrough the structure of a RecA monomer, resulting in Electron Microscopy

Negative stained samples of both wild-type RecA andthe insertion of the Phe217 side chain into a pocket inthe neighboring subunit. This event serves as a key step the Phe217Tyr mutant are shown in Figure 1. In the

absence of both ATPgS and DNA, the wild-type proteinin intersubunit communication leading to ATP-mediatedcooperative filament assembly and high affinity binding forms short filaments (Figure 1a), whereas the mutant

forms no detectable oligomeric structures (Figure 1d).to ssDNA.In the presence of DNA and ATPgS, both proteins formwell defined nucleoprotein filaments (Figures 1c and 1f).IntroductionStrikingly, in the absence of DNA, the mutant shows adramatic increase in filament length with addition ofThe bacterial RecA protein is a multifunctional enzyme

that plays a central role in recombinational DNA repair, ATPgS (Figure 1e), whereas the length of wild-type fila-ments clearly decreases (Figure 1b). Results for wild-homologous genetic recombination and induction of the

cellular SOS response [1, 2]. Each of these activities type RecA are consistent with previous electron micro-scopic and hydrodynamic analyses [14, 15], whereasrequires a common initial step, formation of a RecA-

ATP-ssDNA nucleoprotein filament [1–3]. Studies of re- the ATP-dependent increase in polymer length for thelated DNA repair and recombination proteins including

Key words: RecA protein; cooperative filament assembly; coopera-tivity mutant; ATP-mediated allosteric changes; electron micros-* To whom correspondence should be addressed (e-mail: kendall.

knight@umassmed.edu). copy; IAsys biosensor DNA binding

Structure48

Figure 1. Negative Stained Electron Micro-graphs of Protein and Protein–ssDNA Com-plexes

Wild-type RecA (A-C) and the Phe217Tyr pro-tein (D-F) were incubated as described in Ex-perimental Procedures in the absence ofATPgS and ssDNA (A, D; protein 5 4.0 mM),in the presence of 1.0. mM ATPgS (B, E; pro-tein 5 4.0 mM) or in the presence of bothATPgS and 0.5 mM φX174 single-strandedcircular DNA (C, F; protein 5 2.0 mM). Insetin panel F shows a 3.6-fold magnification of afree protein filament (top) and a nucleoproteinfilament (bottom). Black bar in each panelequals 0.1 mm.

Phe217Tyr mutant is the first demonstration of such of protein concentration. Calculated values for Bmax, K9,napp and S0.5, appear in Table I. Values for wild-type RecAan occurrence for a mutant RecA protein. These data

suggest that position 217 plays a key role in regulating are consistent with those determined previously [9].The mutant protein shows a slight decrease in Bmax andpolymerization of the RecA protein filament. To provide

a quantitative understanding of the effect of the Tyr a 2.5-fold increase in affinity (S0.5 5 120 and 45.2 nM forsubstitution on RecA polymerization, we performed aseries of biosensor DNA binding studies.

DNA Binding: Equilibrium ParametersBinding of both wild-type RecA and the Phe217Tyr mu-tant to ssDNA was measured using an IAsys biosensor.We have previously demonstrated that this techniqueprovides a reliable and quantitative measurement of theDNA binding properties of both RecA and the humanRad51 protein [9]. Binding time courses for wild-typeRecA and the mutant protein in the presence of ATPgSare shown in Figure 2. Both proteins showed similar lowlevels of DNA binding in the absence of ATPgS (notshown). From these curves the extent of binding (bindingresponse at equilibrium) was calculated for each proteinconcentration. Data were fit to the form of the Hill equa-tion shown below.

(extent bound) b 5Bmax Sn (app)

K9 1 Sn (app)

Bmax is the maximum amount of protein-DNA complexat equilibrium (reported as the maximum extent of bind-ing in arc seconds), S is protein concentration (free) andK9 is a complex constant comprising interaction factorswhich reflect successive binding and dissociation stepsalong the cooperative binding pathway [16]. Assumingthat RecA polymer assembly progresses by addition ofindividual subunits to the growing chain [8], we haveinterpreted the Hill coefficient (napp) as the minimumnumber of interacting subunits during growth of the nu-cleoprotein filament which give rise to the observedcooperative binding parameters. S0.5, the protein con-centration at the mid-point of the titration curve, is calcu-

Figure 2. Biosensor DNA Binding Time Courseslated as S0.5 5 n app√K9 [16]. The sigmoidal curves in Fig-

Reactions included the indicated concentrations of protein and 2ure 3 indicate cooperative binding of both proteins to mM ATPgS, and were incubated at 248C for 25–30 min as describedssDNA, with the mutant showing a dramatic transition in Experimental Procedures. All time courses were repeated a mini-

mum of three times.from 10%–90% binding over an extremely narrow range

Phe217 Regulates Subunit Interactions in RecA49

Table 1. Parameters for Binding to ssDNA

Wild type RecA Phe217 → Tyr

Hill analysis

Bmax 1120.0 765.0napp 5.8 5.8K9 7.2 3 10241 2.5 3 10243

S0.5 (nM) 120.0 45.2

McGhee/von Hippel analysis

K (M21) 1.1 3 105 1.0 3 103

v 75.8 2.0 3 104

(Kv)21 (nM) 120.0 50.0

Parameters for DNA binding were calculated as described in thetext. DNA binding time courses were repeated a minimum of threetimes and errors for all parameters shown ranged from 5%–20%.

cepted model for RecA filament assembly onto ssDNA,it proceeds by (i) directional binding to an establishedsite [20] and by (ii) addition of protein monomers to thegrowing filament [8]. Best fits for K and v were obtainedby iterative non-linear regression analysis using the soft-Figure 3. DNA Binding as a Function of Protein Concentrationware package KaleidaGraphe 3.08d (Synergy Software,Data from time courses in Figure 2 were fit to the Hill equation asReading, PA). Values for K, v, and (Kv)21 are shown indescribed in the text. Extent is the calculated maximum binding atTable 1. As expected (Kv)21 approximates the values ofequilibrium in arc seconds for each protein concentration.S0.5 for both the wild-type and mutant proteins. However,the analysis reveals a dramatic increase in the coopera-tive nature of filament assembly for the mutant protein.wild-type and mutant, respectively). Additionally, napp 5

5.8 for wild-type RecA and the mutant. This suggests The Phe217Tyr substitution increases v by more than250-fold (v 5 75.8 and 20,000 for the wild-type andthat the DNA binding properties of the entire RecA fila-

ment are defined by the interaction of the initial 6 sub- Phe217Tyr proteins, respectively).Although the Phe217Tyr mutant also shows a signifi-units for both wild-type and mutant proteins. Therefore,

when the data is analyzed using this standard form of cant decrease in K, we feel that this does not accuratelyreflect a true measure of the inherent binding affinity forthe Hill equation the Phe217Tyr substitution seems to

impose no significant change in the DNA binding proper- the following reason. In contrast to wild-type RecA, thePhe217Tyr protein polymerizes readily in the presenceties of the protein. However, analysis of the binding data

using a procedure that separates intrinsic DNA binding of ATP (see Figures 1e and 1f), and it is likely that someportion of the protein in the DNA binding assay existsaffinity from cooperative protein-protein interactions

[17] reveals that the Phe217Tyr mutation has a dramatic as free protein filaments. Because free RecA filamentsand nucleoprotein filaments are not directly intercon-effect on RecA polymerization.vertible [21–23], the free filaments must disassembleprior to reassembly on DNA. Therefore, the actual con-DNA Binding: Determination of K and v

To investigate potential changes in cooperative filament centration of protein that is immediately available forbinding to DNA may be overestimated, resulting in anassembly resulting from the Phe217Tyr mutation we an-

alyzed our data as previously described [9, 17] using underestimation of the DNA binding affinity. Experi-ments are currently in progress in an effort to resolvethe equation below.

In this expression, L is the concentration of free ligand, these competing filamentation events.RecA filaments have been shown to pack in a side-n is the number of bases bound by a single RecA subunit

(binding site size), v is the binding density of RecA on by-side manner in the absence of DNA to create orderedbundles [24]. Because biosensor measurements simplyDNA (fractional maximum extent of binding/n), K is the

intrinsic binding constant of a single RecA monomer for record the accumulation of mass within the optical win-dow, experiments were designed to ask whether ourDNA and v is the cooperativity parameter. v is defined

as the relative affinity of an incoming ligand for a singly measurements were also recording fortuitous interac-tions between free protein filaments and RecA-ATPgS-contiguous site vs. an isolated site [17, 18]. A value of

n 5 3 was used as the binding site size [2, 19]. This ssDNA complexes. Electron microscopic analyses werecarried out using an excess of protein relative to ssDNAexpression accommodates two aspects of a widely ac-

Structure50

unit interface revealed that Phe217 was particularly sen-sitive to mutation [11]. The RecA crystal structure showsthat the Phe217 side chain is within van der Waals con-tact distance of the Thr150 and Ile155 side chains in theneighboring subunit (Figures 5a and 5b). However, aThr150→Ser substitution imposes no defect in RecAfunction or structure, and Ile155 can tolerate a variety ofpolar and non-polar substitutions [26]. Therefore, thesecross-subunit hydrophobic interactions seen in thestructure are not important, and we suggested thatPhe217 can form stable interactions with positions fur-ther within the neighboring subunit than is apparent inthe structure [11, 13]. In the present study, we find thata Tyr substitution at position 217 results in a dramaticincrease in ATP-dependent cooperative filament assem-bly, and we propose that ATP-mediated allosteric infor-Figure 4. ATPase Activity of Both Wild-type RecA and the Phe217-mation is transmitted across the RecA filament via theTyr Mutant Protein as a Function of ATP Concentration

following mechanism. Gln194 serves as the g-phos-Reactions included 2 mM protein, 6 mM φX174 single-stranded cir-cular DNA, the indicated concentrations of ATP and were incubated phate sensor within the ATP binding site [27, 28], andas described in Experimental Procedures. Data were fit to the Hill its interaction with bound ATP triggers a conformationalequation as described in Experimental Procedures. change in disordered loop 2 (L2, residues 195–209; Fig-

ure 5b). Phe217 is within a small helix (helix G) immedi-(protein monomers:DNA bases 5 4:1). The results showed ately downstream of the L2 region, and a pair of Glynormal nucleoprotein filament formation for both wild- residues at positions 211 and 212 that are critical fortype RecA (Figure 1c) and the mutant protein (Figure RecA function [29] link L2 with helix G. Therefore, ATP-1f). For wild-type RecA occasional short filaments of mediated conformational changes occurring within L2free protein were seen (arrow in Figure 1c), whereas a will propagate through helix G resulting in the insertionmuch larger percentage of the unbound mutant protein of the Phe217 side chain further into a pocket in theformed elongated filaments (arrows in Figure 1f). Differ- neighboring subunit. This insertion may be stabilized byences in the helical pattern of free protein versus nucleo- hydrophobic interactions with several residues that lieprotein filaments (inset in Figure 1f) reflect the fact that within this pocket (Ile93, Ala95 and Ala148; Figures 5when bound to DNA the pitch of the RecA filament is and 6), and likely results in additional changes in nearby<95 A, whereas in the absence of DNA it is <65 A cross-subunit interactions that enhance cooperative fil-[21, 25]. There was no appearance of free RecA protein ament assembly. Upon ATP hydrolysis, Gln194 returnsbundling in a side-by-side manner with protein already to its position in the inactive form of the enzyme resultingin complex with DNA. Therefore, our biosensor measure- in a relaxation of cross-subunit interactions. The signifi-ments are recording only the growth of single RecA cant increase in cooperative filament assembly affordedfilaments on the immobilized ssDNA substrate. by the Tyr substitution likely arises from its ability to

form hydrogen bonds with positions in the neighboringATPase Activity subunit.We analyzed ssDNA-dependent ATP hydrolysis as afunction of ATP concentration for both wild-type andthe Phe217Tyr mutant protein. As shown in Figure 4 the Alternative Cross-Subunit Contacts Made

by a Tyr Side Chain at Position 217mutant protein has both a higher Vmax and a higher affinityfor ATP. Analysis of the data using the Hill equation (see Modeling a Tyr substitution at position 217 (Figure 5b)

shows that the Tyr–OH is within hydrogen bonding dis-Experimental Procedures) shows that the mutant proteindisplays an approximate 1.2-fold increase in Vmax and a tance of 3 positions in the neighboring subunit, the main

chain carbonyl oxygens of Ala95 and Ala148, and the1.7-fold increase in affinity for ATP (S0.5 5 36.6 and 62.2mM for the mutant and wild-type proteins, respectively). side chain –OH of Thr150. Also, the RecA structure

shows that Lys216 lies immediately underneath Phe217,The Hill coefficient for ATP binding was not changedsignificantly by the mutation, (napp 5 2.1 6 0.1 and 2.9 6 close enough to provide stabilization of the position of

the Phe side chain via a cation-p interaction [30]. The0.1 for the mutant and wild-type proteins, respectively).Both proteins showed an equivalent basal level of ATP strength of a cation-p interaction in a Tyr/Lys pair in-

creases if the Tyr–OH serves as a hydrogen bond donorhydrolysis in the absence of ssDNA (Vi/E < 0.1). Togetherwith the DNA binding data, these results show that the [31]. Given the proximity of residues in this region (Fig-

ures 5b and 6a), this may also contribute to the increasedPhe217Tyr mutation increases both cooperative pro-tein-protein interactions and affinity for ATP. efficiency of filament formation by the mutant protein.

The molecular surface shown in Figure 6a providesan image of the pocket in the neighboring subunit intoDiscussionwhich residue 217 may insert. An area in the neighboringsubunit immediately in front of position 217 (Figure 6a)Mechanism of ATP-Mediated Information

Transfer across the RecA Filament shows no steric obstruction and could allow insertion ofa Phe or Tyr side chain. Again, stabilization of a transientPrevious genetic analyses of a number of mutant RecA

proteins carrying substitutions in one region of the sub- insertion by the wild-type Phe side chain may occur by

Phe217 Regulates Subunit Interactions in RecA51

Figure 5. Structure of the RecA Protein andSpecific Interactions at the Subunit Interface

(a) An a-carbon trace of dimeric RecA takenfrom the crystal structure of the helical pro-tein filament [10] with the main chain of sub-unit 1 in yellow and subunit 2 in white. Gln194,Phe217 and ADP are shown in both subunits.In subunit 2 side chain carbons of E96, D144,T150 and 155 are dark gray, with side chainoxygens in red.(b) Detail of the region of the subunit interfaceinvolved in the proposed mechanism for ATP-mediated transfer of allosteric information.Phe217 (here, replaced by Tyr) is within vander Waals distance of Thr150 and Ile155 inthe neighboring subunit. The model proposesthat ATP binding in subunit 1 results in con-formational changes that are propagatedthrough the L2 region (residues 195–209; dis-ordered in the current RecA structure) andinto helix G (residues 212–219), thereby re-sulting in insertion of the Phe217 side chainfurther into a pocket in subunit 2 (see text).Residues shown in subunit 1 include 191–194, 210–222 and ADP, and those in subunit2 include 92–97, 144–155, 191–194 and ADP.Positions in the neighboring subunit withinhydrogen bonding distance of the Tyr–OH in-clude the main chain carbonyl oxygens ofAla95 and Ala148, and the –OH of Thr150. Alsoshown is Lys216 which may stabilize the posi-tion of either Phe or Tyr at position 217through a cation-p interaction (see text).These images were created using the pro-gram Midas (U.C.S.F.).

interaction with hydrophobic residues in this pocket. of the active site Mg21 [27]. The idea that conformationalchanges in this region of the interface contribute to theDepending on the degree to which the mutant Tyr side

chain inserts into this pocket several other positions cooperativity observed for ATP binding and hydrolysis[32] is supported by our observation that both the affinityappear as candidates for hydrogen bonding interac-

tions, e.g., main chain nitrogens at positions 122 and for ATP and the rate of ATP hydrolysis is increased bythe Phe217Tyr mutation.150 (G122N and T150N in Figure 6b).

The RecA structure also shows that Phe217 lies veryclose to the ATP binding and catalytic site of the neigh- Supporting Evidence for Proposed

Allosteric Mechanismboring subunit (Figures 5b and 6a). Glu96 is proposedto activate a water molecule for an in-line attack of the The key roles played by each of the structural elements

in this mechanism (i) Gln194, (ii) Gly211 and Gly212, andg-phosphate of ATP and Asp-144 may stabilize binding

Structure52

Figure 6. Molecular Surface of the Neigh-boring Subunit with which Residue 217 In-teracts

(a) The surface is colored by atom type (car-bon 5 gray; oxygen 5 red; nitrogen 5 blue).In this view, looking from position 217 towardthe neighboring subunit, the Tyr substitutionat position 217 is seen to lie immediatelyacross from a pocket in the adjacent subunit.

(b) A top view of the pocket showing positionsthat may promote stable interactions with aTyr side chain at position 217 (see text). Addi-tional positions that may participate in hydro-gen bond formation with the Tyr–OH are mainchain nitrogens of Gly122 (G122N) and Thr150(T150N). Figures were created using the pro-gram GRASP [54].

(iii) Phe217 are supported by previous studies. Gln194 bone at these positions is critical for the propagationof conformational changes transmitted from the ATPis non-mutable [28, 29] and three mutant proteins,

Gln194Asn, Gln194Glu or Gln194Ala are completely in- binding site to the subunit interface at position 217.Position 217 tolerates only a conservative Tyr substitu-hibited for all ATP-dependent activities [28]. Both Gly211

and Gly212 lie immediately next to L2 at the N terminus tion while a Phe217 Cys change maintains only a lowpartial activity (approximately 10%–20% recombinationof helix G and neither position tolerates any substitution

[29]. Both Gly residues are invariant among eubacterial function). All other substitutions, including Trp and otherhydrophobic residues result in a rec2 phenotype [11].RecA proteins [2, 33] and have been proposed to be

involved directly in DNA binding or in mediating ATP- These results argue that the specific geometry of thephenyl ring of Phe or Tyr is required at position 217,induced conformational changes [27]. Our data sup-

ports the idea that the flexibility of the polypeptide back- and supports our idea that ATP-mediated insertion of

Phe217 Regulates Subunit Interactions in RecA53

this side chain into a specifically designed pocket in the Gln254 in the B. stearothermophilus PcrA helicase alignswith Gln194 in RecA [41], and the PcrA structure showsneighboring subunit is important to RecA function. Other

amino acids in this region of the interface that contact that Gln254 is likely to be the switch that communicatesATP binding and hydrolytic events to a DNA bindingpositions in the neighboring subunit (Asn213, Lys216,

Tyr218, and Arg222) have a greater tolerance for muta- site immediately downstream (residues 254–264; [41]).Likewise, Sawaya et al. [42] propose that His465 is thetion than does Phe217 [11].g-phosphate sensor in the NTP binding site of the bacte-riophage T7 gene 4 protein and may serve to propagateSubstantial Differences in Subunit Contactsconformational changes to the adjacent DNA bindingbetween the Active versus Inactive Formsmotif H4. In fact, in all other DnaB helicase family mem-of RecA Occur in a Region of the Structurebers residue 465 is Gln [42]. The L2 region in RecA,Other Than that Containing Phe217which is immediately adjacent to Gln194, has been pro-Given that the helical pitch of free RecA filaments isposed to be the protein’s primary DNA binding site [29,<65 A (the inactive form) and that of nucleoprotein fila-43, 44]. Therefore, communication between the ATP andments is <95 A (the active form), differences in cross-primary DNA binding sites may share mechanistic simi-subunit interactions likely exist between each of theselarities with helicases. However, other evidence sug-two oligomeric states [21, 25]. The data in this studygests that disordered region L1 (residues 156–164)show that ATP-mediated cooperative filament assemblymakes up the primary DNA binding site in RecA [45, 46].is dramatically increased by virtue of a single, conserva-We note that the effect of ATP on the polymerizationtive mutation within one region of the RecA subunitof the Phe217Tyr mutant is independent of DNA and,interface. The fact that this increase occurs both in thetherefore, our proposed mechanism can accommodatepresence and absence of DNA suggests that the posi-either of the disordered regions, L2 or L1, as being thetion of residue 217 within the protein filament is similarprimary DNA binding site. Additionally, the ATP-inducedwhether RecA polymerizes on its own or on DNA. Thisincrease in the DNA binding affinity of RecA results moreidea is supported by a recent study showing that thefrom an increase in cooperative protein-protein interac-efficiency of cross-subunit disulfide formation betweentions between RecA monomers rather than an increasea Phe217Cys substitution and several Cys substitutionsin the protein’s inherent DNA affinity [9]. This result ison the surface of the neighboring subunit was similarconsistent with the model presented here rather than awhether RecA was in the “active” or “inactive” formmodel in which ATP binding has a direct effect on the[13]. Additionally, studies of another region of the RecAintrinsic affinity for DNA. Recent crystallographic studiessubunit interface (the N-terminal 28 residues in one sub-of the phage T7 g4p helicase shows that a fully activeunit and residues 112–139 in the neighboring subunit)form of the enzyme crystallizes as a closed hexamericshow that certain mutations destabilize free RecA poly-ring [47]. Therefore, the subunit interface in hexamermers to a far greater extent than nucleoprotein com-helicase structures is very different from RecA and itplexes [34]. These results argue that significant differ-is unlikely that ATP regulates subunit interactions inences in cross-subunit interactions between free RecAhelicases in a way related to that in RecA. While theoligomers vs. RecA nucleoprotein complexes occur inhelicase DnaB family may be derived from a recA genethis latter area of the interface rather than in the regionduplication [48], RecA appears to have evolved to ac-containing Phe217.commodate the fact that its functional form is a helicalprotein filament whose assembly and disassembly on

Relationship to Allosteric Mechanisms in Related DNA substrates is mediated by ATP-induced allostericEnzymes: Eukaryotic RecA Homologues events. Therefore, a mechanistic relationship betweenand DNA Helicases RecA and helicases remains to be clarified.Published alignments comparing the bacterial RecA se-quence with the yeast and human Rad51 and DMC1proteins show that Gln194, Gly211 and Gly212 are con- Biological Implicationsserved [35–37]. This suggests that ATP may regulateallosteric events in these 4 eukaryotic proteins using Regulated cooperative interaction between subunits in

multi-protein complexes is a fundamental mechanisma mechanism similar to that described here for RecA.However, residues in RecA helix G, including Phe217, by which protein function is controlled in all biological

systems. Point mutations within subunit interfaces thatare not conserved in either of the Rad51 or DMC1 pro-teins. As with RecA, binding of S. cerevisiae Rad51 to give rise to increases in cooperative behavior have been

identified in various proteins. One of the most remark-ssDNA is dependent on ATP [38]. Unexpectedly, how-ever, the human Rad51 protein binds ssDNA coopera- able, of course, is the b6Glu to Val change in HbS [49].

Other examples include an Arg152Lys substitution thattively and with high affinity in the absence of ATP [9].Therefore, despite the fact that these residues play im- creates a “hypercooperativity” mutant in E. coli phos-

phofructokinase [50], and similarly, aThr72 Val mutationportant roles in the allosteric mechanism of RecA, ATP-mediated allosteric events and key residues involved in in S. inaequivalvis hemoglobin that results in an increase

in cooperative oxygen binding [51]. Our studies of thepropagating information across the protein filament maybe quite different for human Rad51. Phe217Tyr mutant reveal a striking increase in ATP-

mediated cooperative assembly of a RecA protein fila-Functional and structural studies of helicase enzymessuggest similarities with the RecA protein regarding the ment. Together with previous mutational analyses our

data identifies Phe217 as a critical component withinuse of ATP as an allosteric effector [39, 40]. For example,

Structure54

re-equilibrated in PBS-Tween-Mg buffer containing NTP, where indi-the subunit interface that regulates the transmission ofcated.allosteric information throughout the protein filament.

Free protein concentration (S) was calculated by subtracting theInterestingly, studies of the RecA-like human Rad51 pro-amount of protein-DNA complex from total protein. The amount of

tein suggest that the allosteric mechanism mediated by complex was determined using a conversion of 600 arc seconds 5ATP binding and hydrolysis is significantly different than 1 ng protein/mm2, with a useable surface area in the biotinylated

cuvette of 4 mm2. Binding is measured in “arc seconds,” whichthat described here for RecA [9]. Therefore, despite thecorresponds to the accumulation of mass within the optical windowstructural and catalytic similarities between these twoat the binding surface, and “extent” refers to the calculated maxi-proteins, RecA may serve in only a limited way as amum binding at equilibrium for any given protein concentration.paradigm for understanding the mechanistic propertiesBinding data were analyzed using the Fastfit software supplied with

of its human counterpart. Further studies will be required the instrument.to complete the molecular description of the allostericmechanism of RecA, i.e., which residues in the neigh- Acknowledgmentsboring subunit “sense” the insertion of Phe217, as well

The authors wish to thank Drs. Tony Carruthers, Bill Royer, and Celiaas to clearly understand important mechanistic differ-Schiffer for comments on the manuscript, and Jennifer Foulkes andences between RecA and related eukaryotic strand ex-Celia Schiffer for assistance with the graphics figures. This workchange enzymes.was supported by NIH grant GM44772 to K. L. K.

Experimental Procedures Received September 13, 2000.Revised October 27, 2000.

Plasmids, Proteins, and DNA Substrates Accepted November 1, 2000.Wild-type and Phe217Tyr recA genes are carried on plasmids inwhich recA expression is regulated by ptac [11]. Purification of both

Referencesproteins was performed as previously described [52]. Concentra-tions of proteins in all assays refer to RecA monomers. 59-biotinylated 1. Eggleston, A.K., and West, S.C. (1996). Exchanging partners:phosphoramidite was from Glen Research and the biotinylated oligo- recombination in E. coli. Trends Genet. 12, 20–26.nucleotide used in DNA binding assays was made using an ABI 392 2. Roca, A.I., and Cox, M.M. (1997). RecA protein: structure, func-RNA/DNA synthesizer [9]. The substrate for electron microscopy tion, and role in recombinational DNA repair. Prog. Nucleic Acid(single-stranded circular φX174 DNA) was from New England Bio- Res. Mol. Biol. 56, 129–222.labs. Phage M13 RV-1 single-stranded circular DNA was prepared 3. Radding, C.M. (1989). Helical RecA nucleoprotein filaments me-as described [53]. diate homologous pairing and strand exchange. Biochem. Bio-

phys. Acta 1008, 131–145.4. Griffith, J., and Formosa, T. (1985). The uvsX protein of bacterio-ATPase Assay

phage T4 arranges single-stranded and double-stranded DNADNA dependent ATPase activity was measured as described [52].into similar helical nucleoprotein filaments. J. Biol. Chem. 260,PEI chromatography plates were analyzed using a Molecular Imager4484–4891.FX and QuantityOne software (Bio-Rad). Data are derived from initial

5. Ogawa, T., Yu, X., Shinohara, A., and Egelman, E.H. (1993).rates of ATP hydrolysis determined using various ATP concentra-Similarity of the yeast Rad51 filament to the bacterial RecA

tions (1–500 mM) and fit to the Hill equation, v 5Vmax Sn (app)

K9 1 Sn (app), using filament. Science 259, 1896–1899.

6. Benson, F.E., Stasiak, A., and West, S.C. (1994). PurificationKaleidaGraph 3.08 e (Synergy Software, Reading, PA).and characterization of the human Rad51 protein, an analogueof E. coli RecA. EMBO J. 13, 5764–5771.

Electron Microscopy 7. Silver, M.S., and Fersht, A.R. (1982). Direct observation of com-Samples were prepared for electron microscopy by incubating wild- plexes formed between recA protein and a fluorescent single-type or mutant RecA protein at 4.0 mM in the absence of DNA, or stranded deoxyribonucleic acid derivative. Biochemistry 21,2.0 mM in the presence of 0.5 mM phage φX174 single-stranded 6066–6072.circular DNA. All reactions containing DNA included E. coli SSB 8. Menetski, J.P., and Kowalczykowski, S.C. (1985). Interaction of(0.03 mM) and were pre-incubated at 378C for 5 min. Reaction buffer recA protein with single-stranded DNA: Quantitative aspects ofincluded 25 mM triethanolamine-HCl (pH 7.5), 50 mM KCl and 5 mM binding affinity modulation by nucleotide cofactors. J. Mol. Biol.MgCl2, as well as 1 mM ATPgS where appropriate. Upon addition 181, 281–295.of RecA, reactions were incubated for 15 min at 378C. Reactions 9. De Zutter, J.K., and Knight, K.L. (1999). The hRad51 and RecAwere spread onto thin carbon films on holey carbon grids (400 mesh) proteins show significant differences in cooperative binding toand stained with 1% uranyl acetate. Samples were visualized by single-stranded DNA. J. Mol. Biol. 293, 769–780.transmission electron microscopy using either a Philips CM10 or a 10. Story, R.M., Weber, I.T., and Steitz, T.A. (1992). The structurePhilips EM400 microscope. of the E. coli recA protein monomer and polymer. Nature 355,

318–325.11. Skiba, M.C., and Knight, K.L. (1994). Functionally important resi-IAsys DNA Binding

Cuvettes containing immobilized ssDNA were prepared using a dues at a subunit interface site in the RecA protein from Esche-richia coli. J. Biol. Chem. 269, 3823–3828.59-biotinylated 86mer oligonucleotide as described [9]. Reactions

were initiated with addition of protein, and performed at 248C in 12. Logan, K.M., Skiba, M.C., Eldin, S., and Knight, K.L. (1997).Mutant RecA proteins which form hexamer-sized oligomers. J.PBS-Tween-Mg buffer (10 mM Na2HPO4/NaH2PO4 [pH 7.4], 138 mM

NaCl, 27 mM KCl, 0.05% Tween20, and 5 mM MgCl2) in the absence Mol. Biol. 266, 306–316.13. Skiba, M.C., Logan, K.M., and Knight, K.L. (1999). Intersubunitor presence of 2 mM ATPgS. Protein was preincubated briefly with

ATPgS when appropriate (<1 min) prior to addition to the cuvette. proximity of residues in the RecA protein as shown by engi-neered disulfide crosslinks. Biochemistry 38, 11933–11941.Binding was measured for 30 min. We performed a number of bind-

ing experiments over various times and showed that 25–30 min was 14. Brenner, S.L., Zlotnik, A., and Griffith, J.D. (1988). RecA proteinself-assembly. Multiple discrete aggregation states. J. Mol. Biol.required to calculate a true equilibrium binding constant for RecA.

Following binding, the cuvette was rapidly washed three times with 204, 959–972.15. Wilson, D.H., and Benight, A.S. (1990). Kinetic analysis of thebinding buffer and dissociation of the protein from ssDNA was mea-

sured for 5 min. The immobilized DNA surface was regenerated by pre-equilibrium steps in the self-assembly of RecA protein fromEscherichia coli. J. Biol. Chem. 265, 7351–7359.addition of 2 M MgCl2 for 2 min to strip protein, then washed and

Phe217 Regulates Subunit Interactions in RecA55

16. Segel, I.H. (1975). Enzyme Kinetics, T. Hoffman, ed. (New York: Structural relationship of bacterial RecA proteins to recombina-tion proteins from bacteriophage T4 and yeast. Science 259,John Wiley & Sons), pp. 346–385.

17. McGhee, J.D., and von Hippel, P.H. (1974). Theoretical aspects 1892–1896.38. Namsaraev, E., and Berg, P. (1998). Binding of Rad51p to DNA:of DNA-protein interactions: co-operative and non-co-operative

binding of large ligands to a one-dimensional homogeneous Interaction of Rad51p with single- and double-stranded DNA.J. Biol. Chem. 273, 6177–6182.lattice. J. Mol. Biol. 86, 469–489. (erratum appears in 1976. J.

Mol. Biol. 103, 679. 39. Egelman, E.H. (1998). Bacterial helicases. J. Struct. Biol. 124,123–128.18. Kowalczykowski, S.C., Lonberg, N., Newport, J.W., and von

Hippel, P.H. (1981). Interaction of bacteriophage T4-coded gene 40. Soultanas, P., and Wigley, D.B. (2000). DNA helicases: ‘inchingforward’. Curr. Opin. Struct. Biol. 10, 124–128.32 protein with nucleic acids: I. characterization of the binding

interactions. J. Mol. Biol. 145, 75–104. 41. Subramanya, H.S., Bird, L.E., Brannigan, J.A., and Wigley, D.B.(1996). Crystal structure of a DExx box DNA helicase. Nature19. Zlotnik, A., Mitchell, R.S., Steed, R.K., and Brenner, S.L. (1993).

Analysis of two distinct single-stranded DNA binding sites on 384, 379–383.42. Sawaya, M.R., Guo, S., Tabor, S., Richardson, C.C., and Ellen-the recA nucleoprotein filament. J. Biol. Chem. 268, 22525–

22530. berger, T. (1999). Crystal structure of the helicase domain from20. Register, J.C., and Griffith, J. (1985). The direction of RecA the replicative helicase-primase of bacteriophage T7. Cell 99,

protein assembly onto single strand DNA is the same as the 167–177.direction of strand assimilation during strand exchange. J. Biol. 43. Voloshin, O.N., Wang, L., and Camerini-Otero, R.D. (1996). Ho-Chem. 260, 12308–12312. mologous DNA pairing promoted by a 20-aminio acid peptide

21. Egelman, E.H., and Stasiak, A. (1993). Electron Microscopy of derived from RecA. Science 272, 868–872.RecA-DNA Complexes: Two Different States, Their Functional 44. Wang, L., Voloshin, O.N., Stasiak, A., and Camerini-Otero, R.D.Significance and Relation to the Solved Crystal Structure. Mi- (1998). Homologous DNA pairing domain peptides of RecA pro-cron 24, 309–324. tein: intrinsic propensity to form beta-structures and filaments.

22. Yu, X., and Egelman, E.H. (1992). Structural data suggest that J. Mol. Biol. 277, 1–11.the active and inactive forms of the RecA filament are not simply 45. Wang, Y., and Adzuma, K. (1996). Differential proximity probinginterconvertible. J. Mol. Biol. 227, 334–346. of two DNA binding sites in the Escherichia coli RecA protein

23. Morrical, S.W., and Cox, M.M. (1985). Light scattering studies using photo-cross-linking methods. Biochemistry 19, 3563–of the RecA protein of Escherichia coli: relationship between 3571.free RecA filaments and RecA-ssDNA complex. Biochemistry 46. Cazaux, C., Blanchet, J.-S., Dupuis, D., Villani, G., Defais, M.,24, 760–767. and Johnson, N.P. (1998). Investigation of the secondary DNA-

24. Egelman, E.H., and Stasiak, A. (1986). Structure of helical RecA- binding site of the bacterial recombinase RecA. J. Biol. Chem.DNA complexes: complexes formed in the presence of ATP- 273, 28799–28804.gamma-S or ATP. J. Mol. Biol. 191, 677–692. 47. Singleton, M.R., Sawaya, M.R., Ellenberger, T., and Wigley, D.B.

25. Egelman, E.H. (1993). What do X-ray crystallographic and elec- (2000). Crystal structure of T7 gene 4 ring helicase indicates atron microscopic structural studies of the RecA protein tell us mechanism for sequential hydrolysis of nucleotides. Cell 101,about recombination? Curr. Opin. Struc. Biol. 3, 189–197. 589–600.

26. Nastri, H.G., and Knight, K.L. (1994). Identification of residues 48. Leipe, D.D., Aravind, L., Grishin, N.V., and Koonin, E.V. (2000).in the L1 region of the RecA protein which are important to The bacterial replicative helicase DnaB evolved from a RecArecombination and coprotease activities. J. Biol. Chem. 269, duplication. Genome Res. 10, 5–16.26311–26322. 49. Eaton, W.A., and Hofrichter, J. (1990). Sickle cell hemoglobin

27. Story, R.M., and Steitz, T.A. (1992). Structure of the recA protein/ polymerization. Advan. Protein Chem. 40, 63–279.ADP complex. Nature 355, 374–376. 50. Auzat, I., Le Bras, G., and Garel, J.-R. (1995). Hypercooperativity

28. Kelley, J.A., and Knight, K.L. (1997). Allosteric regulation of RecA induced by interface mutations in the phosphofructokinase fromprotein function is mediated by Gln194. J. Biol. Chem. 272, 25778– Escherichia coli. J. Mol. Biol. 246, 248–253.25782. 51. Royer, W.E., Jr., Pardanani, A., Gibson, Q.H., Peterson, E.S.,

29. Hortnagel, K., Voloshin, O.N., Kinal, H.H., Ma, N., Schaffer- and Friedman, J.M. (1996). Ordered water molecules as keyJudge, C., and Camerini-Otero, R.D. (1999). Saturation muta- allosteric mediators in a cooperative dimeric hemoglobin. Proc.genesis of the E. coli RecA Loop L2 homologous DNA pairing Natl. Acad. Sci. USA 93, 14526–14531.region reveals residues essential for recombination and recom- 52. Konola, J.T., Nastri, H.G., Logan, K.M., and Knight, K.L. (1995).binational DNA repair. J. Mol. Biol. 286, 1097–1106. Mutations at Pro67 in the RecA protein P-loop motif differentially

30. Gallivan, J.P., and Dougherty, D.A. (1999). Cation-pi interactions modify coprotease function and separate coprotease from re-in structural biology. Proc. Natl. Acad. Sci. USA 96, 9459–9464. combination activities. J. Biol. Chem. 270, 8411–8419.

31. Mecozzi, S., West, A.P., Jr., and Dougherty, D.A. (1996). Cation- 53. Zagursky, R.J., and Berman, M.L. (1984). Cloning vectors thatpi interactions in aromatics of biological and medicinal interest: yield high levels of single-stranded DNA for rapid DNA sequenc-electrostatic potential surfaces as a useful qualitative guide. ing. Gene 27, 183–191.Proc. Natl. Acad. Sci. USA 93, 10566–10571.

54. Nicholls, A., Bharadwaj, R., and Honig, B. (1993). GRASP: a32. Mikawa, T., Masui, R., and Kuramitsu, S. (1998). RecA protein

graphical representation and analysis of surface properties. Bio-has extremely high cooperativity for substrate in its ATPase

phys. J. 64, A166.activity. J. Biochem. 123, 450–457.

33. Karlin, S., and Brocchieri, L. (1996). Evolutionary conservationof RecA genes in relation to protein structure and function. J.Bacteriol. 178, 1881–1894.

34. Eldin, S., Forget, A.L., Lindenmuth, D., Logan, K.M., and Knight,K.L. (2000). Mutations in the N-terminal region of the RecA pro-tein that disrupt stability of free protein filaments but not nucleo-protein filaments. J. Mol. Biol. 299, 91–101.

35. Shinohara, A., Ogawa, H., Matsuda, Y., Ushio, N., Ikeo, K., andOgawa, T. (1993). Cloning of human, mouse and fission yeastrecombination genes homologous to RAD51 and recA. Nat.Genet. 4, 239–243.

36. Yoshimura, Y., Morita, T., Yamamoto, A., and Matsushiro, A.(1993). Cloning and sequence of the human RecA-like genecDNA. Nucleic Acids Res. 21, 1665.

37. Story, R.M., Bishop, D.K., Kleckner, N., and Steitz, T.A. (1993).