Autoinhibition of Escherichia coli Rep monomer helicase activity by its 2B subdomain

Abstract

DNA helicases catalyze separation of double-helical DNA into its complementary single strands, a process essential for DNA replication, recombination, and repair. The Escherichia coli Rep protein, a superfamily 1 DNA helicase, functions in DNA replication restart and is required for replication of several bacteriophages. Monomers of Rep do not display helicase activity in vitro; in fact, DNA unwinding requires Rep dimerization. Here we show that removal of the 2B subdomain of Rep to form RepΔ2B activates monomer helicase activity, albeit with limited processivity. Although both full length Rep and RepΔ2B monomers can translocate with 3′ to 5′ directionality along single-stranded DNA, the 2B subdomain inhibits the helicase activity of full length Rep. This suggests an autoregulatory mechanism for Rep helicase, which may apply to other nonhexameric helicases, whereby helicase activity is regulated by the rotational conformational state of the 2B subdomain; formation of a Rep dimer may relieve autoinhibition by altering the 2B subdomain orientation.

DNA helicases are a ubiquitous class of enzymes that use the binding and hydrolysis of nucleoside triphosphates to catalyze the separation of the DNA double helix into its complementary single strands. This process is essential for DNA replication, recombination, and repair (1-3), and defects in some DNA helicases are linked to human diseases (4-6). DNA helicases are classified into superfamilies based on their primary structure, with the majority belonging to superfamilies (SF)1 and SF2 (7). Some DNA helicases function as hexamers (8); others, such as the Escherichia coli SF1 helicases Rep (9) and UvrD (10) and the SF2 hepatitis C viral (HCV) NS3 helicase (11), function as dimers in vitro, whereas others, such as the SF1 phage T4 Dda helicase (12), show limited activity as monomers in vitro. The SF1 Bacillus stearothermophilus PcrA helicase has been proposed to function as a monomer (13), although this has not been demonstrated experimentally. Possible roles for oligomerization in helicase activity have been discussed (3, 8, 14).

The E. coli Rep protein (673 amino acids), a 3′ to 5′ SF1 DNA helicase (3, 14), is involved in replication restart (15) and is required for replication of some bacteriophages (16). Rep exists as a monomer in solution in the absence of DNA; however, in vitro, Rep monomers are inactive as helicases, and Rep dimerization is required for processive DNA unwinding (9, 17). E. coli Rep is structurally homologous (18) to B. stearothemophilus PcrA (19) and E. coli UvrD (S. Korolev, N. K. Maluf, T.M.L., and G. Waksman, unpublished results), both of which are also 3′ to 5′ SF1 DNA helicases. Rep monomer is composed of two domains (1 and 2), each with two subdomains (1A, 2A, 1B, and 2B) (Fig. 1). In the asymmetric unit of the Rep-(dT₁₆) crystal structure, two molecules of Rep are observed that differ in the orientation of the 2B subdomain (18). These two orientations (“open” vs. “closed”) differ by a ≈130° rotation of the 2B subdomain (≈168 amino acids) about a hinge region connecting it to the 2A subdomain. Two hypotheses have been suggested for the role of the 2B subdomain. It was suggested that the 2B subdomain of Rep might be involved in protein dimerization (18), and it was proposed to form a part of the duplex DNA-binding site in B. stearothermophilus PcrA (20).

Fig. 1.

Ribbon diagrams of the “open” and “closed” forms of wtRep protein complexed with DNA (blue) (18). The subdomains of Rep are colored (1A, yellow; 1B, green; 2A, red; and 2B, cyan).

We have shown (21) that RepΔ2B, a mutant in which the 2B subdomain has been deleted and replaced by three glycines, retains helicase activity both in vitro and in vivo, suggesting that this domain does not function enzymatically in Rep but rather might be regulatory. When wild-type Rep (wtRep) is directly mixed with a DNA substrate in vitro, the rate of initiation of DNA unwinding is limited by dimerization of the Rep protein on the DNA (9). However, the rate of initiation of DNA unwinding by RepΔ2B is ≈120-fold faster than for wtRep (22), suggesting that either: (i) removal of the 2B subdomain increases the rate of Rep dimerization by ≈120-fold or (ii) the RepΔ2B protein already exists as a functional helicase. The latter would apply if RepΔ2B already exists as a functional dimer in the absence of DNA or if a RepΔ2B monomer is an active helicase. To examine this, we have characterized the assembly states of purified RepΔ2B in the absence and presence of a DNA substrate and have compared DNA unwinding of RepΔ2B and wtRep monomers by using single-turnover (STO) kinetics experiments under the same solution conditions used to examine oligomerization.

Materials and Methods

Buffers. Buffer M is 20 mM Mops/20% (vol/vol) glycerol/5 mM 2-mercaptoethanol, pH 6.5, at 25°C, plus the indicated [NaCl]. RepΔ2B protein has higher solubility in this buffer than in the one used previously (21), facilitating sedimentation equilibrium studies with both wtRep and RepΔ2B proteins under the same solution conditions. Heparin was prepared as described (23).

Proteins and DNA. RepΔ2B protein was expressed from plasmid pRepOΔ2B as described and was untagged (21). The wtRep and RepΔ2B proteins were purified, and DNA substrates were prepared as described (9, 21). Sequences of the DNA substrates are given in Table 1, which is published as supporting information on the PNAS web site.

Sedimentation Equilibrium. Sedimentation equilibrium experiments were performed using an Optima XL-A analytical ultracentrifuge (Beckman Instruments, Fullerton, CA) and analyzed as described (10). See details in Supporting Text, which is published as supporting information on the PNAS web site.

DNA Unwinding Experiments. Stopped-flow experiments were performed using an Applied Photophysics SX18MV instrument (Leatherhead, U.K.), as described (9). Reactions were initiated by mixing a RepΔ2B-DNA solution with Buffer M (plus the indicated [NaCl]) containing 3 mM ATP, 4.2 mM MgCl₂, and 12 μM HP10T40 (concentration before mixing).

STO quenched-flow DNA unwinding experiments were performed essentially as described (9, 24) by using an RQF-3 quenched-flow instrument (KinTek Instruments, University Park, PA). DNA II (Fig. 2 and Table 1) was labeled at the 5′ end with ³²P as described (24). Reactions were initiated by mixing preformed RepΔ2B-DNA complex with Buffer M (plus the indicated [NaCl]) containing 3 mM ATP, 4.2 mM MgCl₂, 12 μM HP10T40, and 10 μM unlabeled “top” strand of DNA II (25). HP10T40 is a DNA that possesses a 3′-(dT₄₀) tail and a 10-bp hairpin and serves as a trap for free protein (21), which ensures that DNA unwinding time courses are STO and thus that any DNA unwinding results exclusively from RepΔ2B that is prebound to the DNA. The unlabeled “top strand” serves as a DNA trap to prevent reannealing of released unwound strands. Control experiments showed that DNA unwinding was ATP-dependent, and RepΔ2B does not unwind a blunt-ended DNA under these conditions (data not shown).

Fig. 2.

Schematic of DNA substrates. All substrates possess a (dT₂₀) tail. DNA II and III represent a series of substrates with varying duplex length, L. The top strand in DNA II is labeled on its 5′ end with ³²P. DNA sequences of the substrates used are given in Table 1.

DNA unwinding time courses were analyzed by using Eq. 1 (25, 26), based on Scheme 1:

[1]

Scheme 1.

where f_ss(t) is the fraction of DNA molecules unwound at time, t; A is the total unwinding amplitude, including both phases of the reaction; n is the number of steps; k_obs = (k_U + k_d); and (1 - x) is the fraction of DNA bound by helicase in nonproductive complexes, which subsequently isomerize with rate constant k_NP to form productive complexes. Γ(n) is the gamma function of n, and Γ(n, k_obs t), Γ(n, (k_obs-k_NP)t) are the incomplete gamma functions of their arguments, respectively. In our analyses, we determine only k_obs, because we have insufficient information to determine k_d.

Single-Stranded DNA (ssDNA) Translocation Experiments. Rates of ssDNA translocation by wtRep and RepΔ2B monomers were determined as described (23). Briefly, a fluorophore (Cy3) is attached to either the 3′ or the 5′ end of a series of synthetic oligodeoxythymidylates (dT_L) of varying lengths, L, and arrival of the protein at the end of the DNA is monitored by the increase in Cy3 fluorescence as the protein interacts with the Cy3. By using a molar excess of DNA (200 nM), monomers of wtRep or RepΔ2B (50 nM) were prebound to the DNA labeled at either the 3′ or the 5′ end with Cy3 and then rapidly mixed with a solution containing 3 mM MgCl₂, 3 mM ATP, and 8 mg/ml heparin (Buffer M plus 200 mM NaCl, 25°C). The heparin serves as a trap for free protein, thus preventing rebinding of free protein to the DNA upon dissociation and ensuring translocation reactions that are STO (single round) (23). Nonlinear least-squares (NLLS) analyses of the translocation time courses were performed as described (23) (see details in Supporting Text).

Results

Assembly State of RepΔ2B. The assembly states of RepΔ2B alone and when bound to a short DNA substrate were examined by sedimentation equilibrium in Buffer M [20 mM Mops/20% (vol/vol) glycerol, pH 6.5] plus 200 mM NaCl at 25°C. Experiments performed at three RepΔ2B concentrations (0.5-2 μM) at rotor speeds of 23,000, 28,000, and 34,000 rpm in the absence of DNA (see Supporting Text; Fig. 6A, which is published as supporting information on the PNAS web site), showed that RepΔ2B behaved as a single ideal species with molecular mass, M = 57.4 (±2.3) kDa, indicating that RepΔ2B is monomeric under these conditions. Sedimentation equilibrium experiments performed under the same solution conditions indicate that wtRep is monomeric with M = 76.8 (±4.8) kDa (Fig. 6B). Therefore, both wtRep and RepΔ2B exist as monomers in Buffer M plus 200 mM NaCl at 25°C, at least up to concentrations of 2 μM, in agreement with previous studies of wtRep (27).

To examine the helicase activity of RepΔ2B monomers, STO DNA unwinding experiments were performed under conditions such that no more than one RepΔ2B monomer is prebound to each DNA substrate. To determine these conditions, we performed sedimentation equilibrium experiments by using an 18-bp DNA substrate containing a 3′-(dT₂₀) ssDNA tail and a Cy3 chromophore attached covalently to its blunt end via the 5′ end of one strand of the DNA duplex (DNA I; Figs. 2 and 3A; Table 1). By monitoring Cy3 absorbance (550 nm), the DNA concentration profile can be observed without interference from the RepΔ2B. Sedimentation equilibrium experiments were performed at different ratios of RepΔ2B to DNA (Buffer M plus 200 mM NaCl at 25°C), and Fig. 3A shows a representative set of results (2 μM RepΔ2B and 4 μM DNA). NLLS analysis of the absorbance profiles (see details in Supporting Text) indicates two DNA species with molecular masses, M = 17.6 (±0.4) kDa and 77.9 (±5.4) kDa, corresponding to free DNA and DNA bound by one RepΔ2B monomer, respectively. Experiments at varying RepΔ2B/DNA molar ratios indicate that as long as the input molar ratio of RepΔ2B/DNA <0.6, no more than one RepΔ2B is bound per DNA substrate at equilibrium (Fig. 7, which is published as supporting information on the PNAS web site). Experiments performed with wtRep protein in Buffer M plus 200 mM NaCl at 25°C (Fig. 8, which is published as supporting information on the PNAS web site) showed that no more than one wtRep is bound per DNA substrate at wtRep/DNA <1.

Fig. 3.

RepΔ2B monomer is an active helicase. (A) Sedimentation equilibrium DNA concentration profiles (monitoring Cy3 absorbance of DNA I) at 18,000 (red), 22,000 (blue), and 27,000 (green) rpm. The solid curves are simulations based on global NLLS fits of the data to Eq. 2 (see Supporting Text), with residual plots below. (B) STO kinetics of unwinding of DNA III catalyzed by RepΔ2B [increase in Cy3 fluorescence (red) and decrease in Cy5 fluorescence (blue)]; no DNA unwinding (no Cy3 fluorescence increase) was catalyzed by wtRep monomer (dark green).

We were unable to characterize the assembly state(s) of the RepΔ2B protein alone at NaCl concentrations below 200 mM by using sedimentation equilibrium, because stable equilibrium distributions of protein were not obtainable at these lower [NaCl]. However, RepΔ2B-DNA complexes could be examined at 50 mM NaCl in the presence of a molar excess of DNA, and we find that RepΔ2B binds to the DNA substrate as a monomer when the DNA is in molar excess over RepΔ2B (Fig. 9, which is published as supporting information on the PNAS web site).

Removal of the 2B Subdomain Activates Rep Monomer Helicase Activity. We performed a series of STO DNA unwinding kinetics experiments to compare the DNA helicase activity of RepΔ2B monomers vs. wtRep monomers. DNA unwinding was first monitored by using a fluorescently labeled DNA substrate possessing a 3′-(dT₂₀) tail and an 18-bp duplex, with Cy3 and Cy5 fluorophores covalently attached to each strand at the blunt end of the duplex (DNA III; Fig. 2; Table 1). This DNA substrate allows the DNA unwinding time course to be followed in real time in a stopped-flow experiment by monitoring the changes in fluorescence resonance energy transfer (FRET) between the donor fluorophore, Cy3, and the acceptor fluorophore, Cy5 (9, 17, 28). A loss of FRET accompanies DNA unwinding and strand separation resulting in an increase in Cy3 fluorescence and a concomitant decrease in Cy5 fluorescence.

In the STO DNA unwinding kinetics experiments, RepΔ2B (80 nM) and DNA III (274 nM) were premixed in one syringe, in Buffer M plus 200 mM NaCl (25°C) under conditions of a molar excess of DNA such that no more than one RepΔ2B monomer is bound per DNA. Reactions were initiated by mixing the RepΔ2B-DNA complex with a solution containing MgCl₂, ATP, and trap for free protein in Buffer M plus 200 mM NaCl (see Materials and Methods for details). Such an experiment performed with the RepΔ2B monomer displayed a short lag phase followed by an increase in Cy3 fluorescence and a concomitant decrease in Cy5 fluorescence indicating DNA unwinding (Fig. 3B). The observation of lag kinetics in such an “all or none” unwinding assay reflects the fact that the helicase proceeds through multiple intermediate steps, each with similar rate constants, before the DNA is unwound completely, as shown in Scheme 1 (24). NLLS fits of the Cy3 time course to Scheme 1 (Eq. 1) yields k_obs = (k_U + k_d) = 50.9 (±1.5) s^-1. In this analysis, we assumed a kinetic step size, m = L/n = 4 bp, where L is the length of the duplex DNA in base pairs, and n is the number of steps in the kinetic scheme.

In contrast, Fig. 3B also shows that no DNA unwinding is detected in an identical experiment performed with wtRep protein when no more than one wtRep monomer is bound per DNA (200 nM wtRep preincubated with 600 nM DNA III). Therefore, a wtRep monomer is unable to unwind even this 18-bp DNA duplex. However, when the molar ratio of wtRep/DNA is increased to >2, then DNA unwinding is observed (Fig. 10, which is published as supporting information on the PNAS web site) consistent with the need for dimerization of wtRep to activate its helicase activity (9).

We also examined the effect of [NaCl] on DNA unwinding by the RepΔ2B monomer (Fig. 4A). STO DNA unwinding experiments were performed with 80 nM RepΔ2B and 320 nM DNA III in the stopped-flow, and DNA unwinding was monitored as changes in Cy3 fluorescence signal. Fig. 4A shows that the amplitude of DNA unwinding increases as [NaCl] is decreased. NLLS fits of these time courses to Scheme 1 (Eq. 1; Table 2, which is published as supporting information on the PNAS web site) show that k_obs = (k_U + k_d) increases with increasing [NaCl] from 37.3 (±1) s^-1 at 30 mM NaCl to 56.6 (±7) s^-1 at 200 mM NaCl (assuming a step size, m = 4 bp). Although we are unable to obtain separate estimates for k_U and k_d from these fits, the increase in k_obs with increasing [NaCl] may be due partly to an increase in the dissociation rate constant, k_d, with increasing [NaCl]. These results suggest that the processivity of RepΔ2B monomer increases with decreasing [NaCl]. Identical experiments performed with wtRep showed no detectable DNA unwinding by wtRep monomer at any [NaCl] from 10 to 200 mM.

Fig. 4.

STO DNA unwinding kinetics catalyzed by RepΔ2B monomer. (A) DNA unwinding monitored by the increase in Cy3 fluorescence at several [NaCl]. Identical experiments performed with wtRep monomer (orange). (B and C) DNA unwinding dependence on duplex length. STO quenched-flow experiments at 200 mM NaCl (B). STO stopped-flow experiments at 50 mM NaCl (C). Solid lines are simulations based on Eq. 1 and the kinetic parameters from a global NLLS fit of the unwinding time courses (Table 2).

Unfortunately, because the DNA unwinding processivity of the RepΔ2B monomer is so low, we were able to examine it only qualitatively. We performed STO DNA unwinding experiments by using a series of DNA substrates with increasing duplex lengths, all possessing a 3′-(dT₂₀) tail. Rapid chemical quenched-flow experiments similar to those described (24, 29) were performed (Buffer M plus 200 mM NaCl, 25°C). RepΔ2B (80 nM) was preincubated with DNA II (320 nM) radiolabeled with ³²P on the 5′ end of the top strand (see Materials and Methods for details). Fig. 4B shows that the unwinding amplitude drops from ≈50% for a DNA substrate with a 12-bp duplex to <2% for DNA substrates with duplex lengths of 21 or 24 bp. A global NLLS fit of the data to Scheme 1 (Eq. 1), assuming m = 4 bp, yields k_obs = (k_U + k_d) = 31.5 (±13.8) s^-1 at 200 mM NaCl. Although these experiments do not provide a quantitative measure of processivity, it is clear that a RepΔ2B monomer is not highly processive, being unable to fully unwind even a 21-bp duplex under these solution conditions.

Stopped-flow DNA unwinding experiments were also performed with RepΔ2B as a function of duplex DNA length at 50 mM NaCl. Fig. 4C shows that decreasing the [NaCl] from 200 to 50 mM increases the unwinding amplitude of all duplex lengths examined, suggesting an increase in unwinding processivity. RepΔ2B monomer is able to unwind a 30-bp duplex at 50 mM NaCl, although no detectable unwinding of a 40-bp duplex is observed (Fig. 4C). Thus, the RepΔ2B monomer is not a highly processive helicase in vitro, even at 10 mM NaCl, which is the lowest [NaCl] examined (data not shown).

ssDNA Translocation of wtRep and RepΔ2B Monomers. We next examined the ability of wtRep and RepΔ2B monomers to translocate along ssDNA to determine whether the lack of helicase activity of a wtRep monomer reflects an inability of the monomer to translocate. We used a fluorescence stopped-flow assay as described (23, 30) (see Materials and Methods and Supporting Text for details). The resulting time courses for wtRep monomers (Fig. 5A) and RepΔ2B monomers (Fig. 5B) on dT_L (L = 64, 84, 104, and 124 nucleotides), labeled with Cy3 at the 5′ end, show an initial ATP-dependent enhancement of Cy3 fluorescence associated with the accumulation of protein at the 5′ end of the DNA, followed by a slower decrease in fluorescence reflecting protein dissociation from the 5′ end of the DNA. Furthermore, the peak in the Cy3 fluorescence time course moves to longer times as the length of the DNA increases. In contrast, control experiments performed with dT_L (L = 59 and 79 nucleotides), labeled with Cy3 at the 3′ end, show a length-independent decrease in Cy3 fluorescence (Fig. 5). These results indicate that both wtRep and RepΔ2B monomers translocate with biased 3′ to 5′ directionality along ssDNA in ATP-dependent reactions (30).

Fig. 5.

ssDNA translocation by wtRep and RepΔ2B monomers. Stopped-flow kinetics experiments were performed as described (23) with a series of 5′-Cy3-(dT_L) (L = 54, 64, 74, 84, 94, 104, 114, and 124 nt), monitoring Cy3 fluorescence with wtRep (A), and RepΔ2B (B). For clarity, only the time courses for L = 64, 84, 104, and 124 nt and control experiments with 5′-(dT₇₉)-Cy3 (black), and 5′-(dT₅₉)-Cy3 (orange) are shown. Solid lines are simulations using Eq. 3 (RepΔ2B) or Eq. 4 (wtRep) (see Supporting Text) and the kinetic parameters from a global NLLS fit of time courses obtained with all DNA lengths.

As described (23), one can analyze the time courses in Fig. 5 to obtain quantitative estimates of the macroscopic rate of translocation by using the minimal sequential “n-step” mechanism shown in Scheme 2, by using Eq. 3 (see Supporting Text). The macroscopic translocation rate determined from NLLS analysis of RepΔ2B monomer translocation time courses (Fig. 5B) is 530 (±10) nucleotides s^-1 (Buffer M, 200 mM NaCl, 25°C). Interestingly, the minimal sequential “n-step” mechanism needed to describe the wtRep translocation time courses requires the inclusion of an additional step (k_a in Scheme 3). The macroscopic translocation rate determined from NLLS analysis of wtRep monomer translocation time courses (Fig. 5A) according to Scheme 3 (Eq. 4 in Supporting Text) is 298 (±2) nucleotides s^-1 (Buffer M, 200 mM NaCl, 25°C). Translocation experiments performed at 50 mM NaCl (Buffer M, 25°C) gave similar results (630 (±20) nucleotides s^-1 for RepΔ2B monomers vs. 279 (±2) nucleotides s^-1 for wtRep monomers). Therefore, both wtRep and RepΔ2B monomers can translocate rapidly and with 3′ to 5′ directionality along ssDNA, although RepΔ2B monomers translocate with a slightly higher rate.

Scheme 2.

Scheme 3.

Discussion

Our results show that a RepΔ2B monomer is able to unwind DNA in vitro, although with low processivity. A RepΔ2B monomer is able to translocate along ssDNA [poly(dT)] at a rate of 530 (±10) nucleotides s^-1, with a reasonably high processivity of 800 (±200) nucleotides, but its net rate of duplex DNA unwinding, mk_U, is no greater than mk_obs = [4×(56.6 (±7)] = 226 (±28) bp s-¹ (Buffer M, 200 mM NaCl, 25°C). In contrast, no DNA unwinding is detectable by a wtRep monomer, consistent with previous reports (9, 17), even though under the same conditions, a wtRep monomer can translocate along ssDNA [poly(dT)] at a rate of 298 (±2) nucleotides s^-1 with a processivity of 700 (±50) nucleotides. These results also indicate that the inability to detect DNA unwinding by wtRep monomers is not due to artifactual problems related to reannealing of the DNA duplex behind the helicase, because unwinding of the identical DNA substrate by the RepΔ2B monomer is readily detectable. The lack of wtRep monomer helicase activity is also not due to an inability to translocate rapidly and unidirectionally along ssDNA. Rather, the 2B subdomain must serve as a “kinetic block” preventing a Rep monomer from invading the DNA duplex. Thus, as we have observed with UvrD monomers (23), the simple ability of a wtRep monomer to translocate along ssDNA with directional bias is not sufficient to confer helicase activity.

These results indicate that the 2B subdomain is autoinhibitory for Rep monomer helicase activity and suggest this subdomain may play a role in regulating helicase activity. Activation of the helicase activity of full-length wtRep protein in vitro requires oligomerization (9). Inhibition of Rep monomer helicase activity by its 2B subdomain provides a mechanism by which DNA unwinding activity can be regulated through a requirement for protein-protein interactions. This type of regulation is an elegant yet simple method to control cellular processes and is also used in a number of other systems (31).

The 2B subdomain of E. coli Rep and other SF1 DNA helicases contains none of the seven conserved helicase motifs; in fact, this subdomain is found as an insert between helicase motifs IV and V. Furthermore, the size and sequences of the 2B subdomains of members of the SF1 superfamily are not highly conserved (21). Thus it is not surprising that the 2B subdomain is not directly required for helicase activity. The 2B subdomain in Rep, and likely in other SF1 helicases, is rotationally flexible (18, 32) and thus can adopt different orientations within the monomer. This flexibility may directly contribute to its autoinhibitory function and be key to its regulatory role. That DNA unwinding by a Rep monomer can be activated by deleting the 2B subdomain suggests that the relative position of the 2B subdomain within Rep controls activation. In this regard, it is interesting that Rasnik et al. (32) have shown that when a Rep monomer is bound at a 3′-ssDNA/duplex junction, the 2B subdomain is found primarily in the “closed” orientation, suggesting that this orientation may promote inhibition. Hence, activation of helicase activity through self association of wtRep monomers may involve a reorientation of the 2B subdomain. Like wtRep and UvrD, the full length HCV NS3 protein (containing both the helicase and the protease domains) seems to also require dimerization to be active on duplex RNA in vitro (11). However, upon removal of the protease domain of NS3, the HCV NS3 helicase domain (33) shows limited helicase activity as a monomer on duplex DNA (34), similar to RepΔ2B. Therefore, it is possible that the protease domain of the full length NS3 monomer may inhibit its helicase activity in a manner similar to that observed for the 2B subdomain within the E. coli Rep monomer.

Whether Rep protein oligomerization is needed to activate its helicase activity in vivo is not known. It is possible that helicase activity could be activated through interactions of Rep monomer with other accessory proteins. Interestingly, structural studies of the heterotrimeric E. coli RecBCD helicase suggest that such a mechanism may be used to activate the helicase activity of the RecB subunit, a 3′ to 5′ SF1 helicase (35). A structure of RecBCD in complex with a DNA duplex shows that the 2B subdomain of RecB contacts the RecC subunit, but not the DNA. Although RecB protein alone has been shown to possess helicase activity in vitro, the activity of the RecB monomer has not been examined. Thus, although it is not known whether the 2B subdomain of RecB also is autoinhibitory, it seems likely that the interactions with RecC would restrict rotational flexibility of the 2B subdomain of RecB and thus might serve to relieve any autoinhibition. The RecC subunit is structurally homologous to the SF1 helicases, Rep, PcrA, and RecB, even though it shares little sequence similarity and has no helicase or ATPase activity (35). Furthermore, the RecB and RecC subunits also interact partially through their respective 2B subdomains (35). Interestingly, just such a 2B-2B interaction was hypothesized to be involved in Rep dimerization (18). In this respect, the active Rep dimer helicase may share similarities to the active RecB-RecC heterodimeric helicase (36).

The results reported here have broad implications for the mechanisms of DNA unwinding by nonhexameric helicases in general and suggest a novel mechanism for the regulation of their helicase activities inside the cell. Autoinhibition of helicase activity of SF1 monomers would provide a mechanism for controlling DNA unwinding activity during DNA metabolism. With activation requiring self association or heterointeractions with accessory proteins, DNA unwinding activity could be regulated through changes in protein levels at different points during the cell cycle.