PcrA helicase dismantles RecA filaments by reeling in DNA in uniform steps

Abstract

Translocation of helicase-like proteins on nucleic acids underlies key cellular functions. However, it is still unclear how translocation can drive removal of DNA bound proteins, and basic properties like the elementary step size remain controversial. Using single molecule fluorescence analysis on a prototypical superfamily 1 helicase, Bacillus stearothermophilus PcrA, we discovered that PcrA preferentially translocates on the DNA lagging strand instead of unwinding the template duplex. PcrA anchors itself to the template duplex using the 2B subdomain and reels in the lagging strand, extruding a single stranded loop. Static disorder limited previous ensemble studies of PcrA stepping mechanism. Here, highly repetitive looping revealed that PcrA translocates in uniform steps of 1 nt. This reeling-in activity requires the open conformation of PcrA and can rapidly dismantle a preformed RecA filament even at low PcrA concentrations suggesting a mode of action for eliminating potentially deleterious recombination intermediates.

Introduction

Helicases are ubiquitous enzymes that are involved in nearly all aspects of nucleic acid metabolism in living organisms (Lohman and Bjornson, 1996; Pyle, 2008; Singleton et al., 2007). Most proteins that belong to this class are also ‘translocases’ that move directionally along a DNA or RNA track, although translocation activity is necessary but not sufficient for duplex DNA unwinding (Lohman et al., 2008; Singleton et al., 2007). In fact, some helicases with demonstrated unwinding activity in vitro may have functions in vivo that do not require unwinding activity. For example, E. coli UvrD and yeast Srs2 helicases both display anti-recombinase activity that is related to their abilities to displace RecA and RAD51 recombination proteins, respectively, from single stranded (ss) DNA (Antony et al., 2009; Krejci et al., 2003; Veaute et al., 2005; Veaute et al., 2003). Processive translocation and unwinding require repeated cycles of ATP (or other nucleoside triphosphates) hydrolysis that is coupled to directional movement. A key mechanistic question regarding these enzymes concerns the step size taken during translocation and unwinding.

Helicases/translocases belonging to superfamily 1 (SF1) are among the most extensively characterized of these enzymes (Singleton et al., 2007). Rep, PcrA and UvrD are structurally similar SF1A helicases (Korolev et al., 1997; Lee and Yang, 2006; Singleton et al., 2007; Velankar et al., 1999). The monomeric forms of Rep, UvrD and PcrA are rapid and highly processive translocases that move with 3′ to 5′ directionality along ssDNA (Brendza et al., 2005; Dillingham et al., 2000; Dillingham et al., 2002; Fischer et al., 2004; Myong et al., 2005; Niedziela-Majka et al., 2007). Yet, at least a dimeric form of these enzymes is required to unwind duplex DNA processively in vitro (Lohman et al., 2008; Slatter et al., 2009). Hence, oligomerization and/or interactions with accessory proteins appear to be needed for processive DNA unwinding.

Biochemical studies of PcrA and UvrD monomer translocation on ssDNA have shown that on average one ATP is hydrolyzed per base translocated (Dillingham et al., 2000; Tomko et al., 2007). However, ensemble transient kinetic studies have led to the proposal that UvrD and PcrA monomers translocate with a kinetic step size of ~ 4 nt (Niedziela-Majka et al., 2007; Tomko et al., 2007). As defined, the kinetic step size provides an estimate of how often a recurrent rate-limiting step takes place during processive movement. Similarly, crystallographic analysis led to the proposal that 1 bp of DNA is unwound per single cycle of ATP hydrolysis (Lee and Yang, 2006) while bulk phase (Ali and Lohman, 1997) as well as single-molecule (Dessinges et al., 2004) kinetic studies indicated that the kinetic step size of unwinding is about 4–6 bp. However, because the kinetic step size estimation relies on the variance measurement, it may be influenced by persistent molecular heterogeneities in the reaction rate within an enzyme population (“static disorder”) which has been observed in single-molecule studies of several enzymes (Adelman et al., 2002; Bianco et al., 2001; Neuman et al., 2003; Perkins et al., 2004; Spies et al., 2003; Tan et al., 2003; Zhuang et al., 2002). Our single-molecule analysis presented below shows that the translocation rates of single PcrA proteins differ and these differences persist during our observation time window of minutes. When we bypass the complications arising from the static disorder through molecule-by-molecule analysis, the kinetic step size of PcrA translocation on ssDNA is revealed to be 1 nt.

As non-replicative helicases, the roles of PcrA and UvrD in vivo are yet to be fully defined. PcrA can substitute for several functions of UvrD in vivo while Rep can not (Bidnenko et al., 2006; Lestini and Michel, 2007). One of the suggested functions is removal of RecA at stalled replication forks. UvrD is proposed to act at the fork and remove deleterious RecA filaments formed in the lagging strand. The mechanism by which UvrD (and PcrA) recognizes and functions specifically at the fork structure, given its potential to unwind the template duplex, is unknown.

Using single molecule FRET (Fluorescence Resonance Energy Transfer) analysis (Ha et al., 1996), we discovered that a PcrA monomer induces looping of ssDNA that is tightly coupled to its DNA translocation activity. Looping was observed from the 5′ ssDNA tail of a duplex DNA both in the presence and in the absence of an accompanying 3′ ssDNA tail. In the case of the forked DNA with both tails, contrary to what is expected for a helicase that progresses towards the duplex DNA for its potential unwinding, a PcrA monomer remained in contact with the duplex junction while reeling in the 5′ lagging strand using its motor domains. This robust reeling in activity could induce efficient displacement of RecA filament displacement from the 5′ tail. Our unexpected findings suggest a mechanism that other superfamily 1 translocases may use to control the potentially toxic recombination intermediates at a stalled replication fork.

Results

PcrA induces repetitive looping of a 5′ ssDNA tail of a partial duplex DNA

B. stearothermophilus PcrA is a 3′-5′ helicase that can unwind a duplex DNA possessing a 3′ single stranded tail. Because PcrA in vivo would encounter other types of DNA structures such as a forked DNA and a partial duplex DNA with a 5′ ssDNA tail, we examined how PcrA would behave on these DNA structures. To observe PcrA activities on individual DNA molecules, we immobilized DNA on a polyethylene glycol (PEG)-passivated quartz slide via a biotin-neutravidin linkage (Fig. 1A) (Ha et al., 2002). A fluorescent donor, Cy3, and an acceptor, Cy5, were attached to the two ends of the single stranded part so as to report on the time-averaged end-to-end distance of the ssDNA. The ssDNA is highly flexible, with a persistence length on the order of 1–2 nm and with its conformations being averaged on a much faster time scale than our time resolution (30 ms) (Murphy et al., 2004).

Figure 1. Repetitive looping of ssDNA coupled with PcrA translocation.

A. A dsDNA with 5′-(dT)₄₀ ssDNA tail is immobilized on a PEG surface via the duplex end.

B. FRET between the donor (Cy3) and acceptor (Cy5) reports on the change in the time-averaged distance between the tail end and the partial duplex junction. Single-molecule time traces shown (donor intensity in green, acceptor intensity in red, and FRET efficiency in blue) were obtained in the presence of 5 μM ATP and 100 pM PcrA. See also Figures S1A-S1C.

C. Cy3 intensity increases as PcrA approaches. This property is used to infer distance change between PcrA and the labeled position. Cy3 intensity time trace was obtained under the same conditions as in (a).

D. When Cy3 is attached to the junction no periodic intensity fluctuation was observed upon addition of ATP and PcrA. See also Figures S1D-S1H.

E. Reeling-in model (a) PcrA binds at 5′ partial duplex junction. (b) Translocation begins in 3′->5′ direction while PcrA maintain contact at the junction (c) ssDNA loop formation and its increase in size as PcrA continues to translocate toward the 5′ end. (d) PcrA reaches the 5′ end and runs off the track. (b)->(d) is repeated over in continual cycles.

F. Representative FRET time traces of PcrA looping 5′pdT40 at varying ATP concentrations.

G. Michaelis-Menten fit of repetition rate (1/<Δt>) vs. ATP concentration. Error bars denote STD. Errors in the fit results are shown in SEM.

When the donor and the acceptor are separated by 40 nt of 5′ tail, (dT)₄₀, in a partial duplex (5′pdT40), we observed a FRET efficiency of E_FRET ~0.4 (Fig. S1A). When PcrA and ATP were added, a gradual FRET increase was detected followed by a rapid drop from individual DNA molecules showing that the 5′ tail end approaches the junction gradually and then is released abruptly (Fig. 1B). This asymmetric pattern was repeated multiple times without an observable pause in between and the repetition continued for up to a few minutes. This behavior, which we termed ‘repetitive cycling,’ was also observed when the donor and acceptor positions were swapped (Fig. S1B) and when a ssDNA of mixed sequence was used (Fig. S1C).

The gradual phase of repetitive cycling indicates that two ends of ssDNA are approaching each other gradually. In order to define PcrA’s role during this process, first we used the fact that Cy3 becomes brighter in fluorescence when a protein is nearby (Fischer et al., 2004; Myong et al., 2009). This protein induced fluorescence enhancement is thought to result from protein-induced steric constraints suppressing a non-radiative decay pathway (Aramendia et al., 1994; Luo et al., 2007; Sanborn et al., 2007). When Cy3 is placed at the end of the 5′ tail (with no Cy5 present), its fluorescence intensity rises gradually and drops abruptly in a repetitive manner upon addition of PcrA and ATP (Fig. 1C), showing that PcrA is approaching the 5′ end gradually during translocation, consistent with PcrA monomer translocating in a 3′ to 5′ direction (Dillingham et al., 2000; Dillingham et al., 2002; Niedziela-Majka et al., 2007). In contrast, a substrate that has Cy3 attached at the junction showed no repetitive pattern, indicating that a portion of PcrA remains proximal to the partial duplex junction (Fig. 1D, Fig. S1D). A corroborating result was obtained in bulk solution where PcrA is added to either of the two Cy3 labeled DNA substrates in a cuvette (Fig. S1E and F). Only the 5′-tail end-conjugated Cy3 showed a further intensity increase upon ATP addition, suggesting that PcrA maintains contact with the duplex DNA during its translocation toward the 5′ end. The positioning of PcrA at the junction is also supported by an ExoIII footprint assay where the duplex portion is protected from ExoIII degradation in the presence of PcrA and ATP (Fig. S1G). We confirmed that repetitive cycling still occurs in the buffer conditions used for ExoIII footprinting (Fig. S1H). Finally, Cy3-labeled PcrA showed repetitive FRET changes with the DNA labeled at the 5′ end of the ssDNA tail but not with the DNA labeled at the junction (Fig. 2A and B), further demonstrating that PcrA maintains contacts with the junction during translocation.

Figure 2. Repetitive looping is induced by a PcrA monomer.

A. Cy3-labeled PcrA and ATP are added to DNA labeled with Cy5 at the end of the 5′ tail and an example is shown for binding at 8 sec, repetitive translocation, and dissociation or photobleaching at 17.7 sec.

B. Cy3-labeled PcrA and ATP are added to DNA labeled with Cy5 at the partial duplex junction. An example trace shows only binding and dissociation but no evidence of relative motion between Cy3 and Cy5.

C. ATP and Cy3, Cy5 labeled 5′pdT40 DNA are added to PcrA immobilized on the surface via biotinlyated Anti-His-tag-Antibody. An example time trace shows DNA binding at 7.6 sec followed by repetitive looping. See also Figure S2.

D. Histogram of time interval of repetition (Δt) shown in C and a gamma distribution fit.

The emerging picture is that PcrA can remain anchored to the duplex DNA while using its motor domains to translocate on the 5′ tail, reeling in the 5′ ssDNA tail and progressively extruding a ssDNA loop (Fig. 1E). In this model, when PcrA reaches the end and runs off the track, the 5′ tail is rapidly released back to its flexible conformation, immediately followed by the next cycle of PcrA translocation from the junction, resulting in ‘repetitive looping’.

The time PcrA takes for one cycle of looping, Δt, was determined by visually picking the moments of the sharp FRET decrease after the peak value from the standard construct 5′pdT40 (Fig. 1F). By plotting the inverse of the average Δt vs ATP concentration and fitting to the Michaelis-Menten equation, we obtained K_M=3.5 ± 0.3 μM (Fig. 1G).

PcrA reels in ssDNA as a monomer

To determine the assembly state of PcrA during looping, we conducted several experiments. First, we replaced one native cysteine with alanine (C96A) and conjugated Cy3-maleimide to the remaining cysteine (C247). Upon addition of Cy3-PcrA (250 pM) to the surface-immobilized Cy5-labeled DNA, Cy3-PcrA binding was observed to occur in a single step and later followed by a single step disappearance which is the signature of a monomer (Fig. 2A and B). In between, asymmetric FRET fluctuations were seen consistent with repetitive looping for the majority of molecules (85%). Because the PcrA labeling efficiency is ~45%, if two monomers are required for this activity, 20% of the total traces would have shown multiple binding events. None of the time traces showed such behavior. We conclude that a single PcrA monomer is responsible for repetitive looping. Second, we immobilized PcrA with N-terminal hexa-histidines (1nM) on the surface using biotinylated Anti-His-Tag-Antibody (Fig. 2C). Because PcrA is a monomer by itself, PcrA must exist on the surface as isolated monomers. After washing the chamber thoroughly, fluorescently labeled DNA (5′pdT40 without biotin, 0.3 nM) and 1 mM ATP are added. Fluorescence signal appears as DNA interacts with PcrA immobilized on the surface and more than 65% of DNA molecules showed repetitive looping with an average Δt identical to that of DNA immobilization experiment (Fig. 2D). Because PcrA cannot be recruited from solution, we conclude that a PcrA monomer is responsible for repetitive looping. Third, an additional support for monomeric activity came from our binding time analysis (Fig. S2A). We recorded FRET signals as we flowed in PcrA and ATP and determined the time it took for repetitive looping to begin for each molecule. The inverse of the average initiation time showed a linear dependence on PcrA concentration (R²=0.95 ± 0.06) as expected for a monomeric activity (Fig. S2B–I).

ssDNA length dependence and PcrA translocation rate

We next examined PcrA on pdDNA substrates with 5′ tails of varying lengths (40, 50, 60, 70 and 80nt) (Fig. 3A). The average value of Δt increased linearly with the tail length (Fig. 3B) supporting our interpretation that the gradual FRET increase is due to translocation along the 5′ tail and that the abrupt FRET decrease occurs at the end of the ssDNA track. A linear fit to this result shows a translocation rate of 76±5 nt/sec close to those determined in bulk solution under similar but not identical solution conditions (Dillingham et al., 2000; Dillingham et al., 2002; Niedziela-Majka et al., 2007). The translocation rate is sensitive to the solution conditions. When we used solution conditions that increase the translocation rate of PcrA to 234 (± 2) nt/sec on purely ssDNA in bulk solution (Niedziela-Majka et al., 2007), we observed a similarly accelerated translocation rate in our single-molecule measurements with a rate of 231 (±53) nt/sec (Fig. S3A–D). Therefore, the duplex interaction during looping does not appear to affect the overall ssDNA translocation rate of PcrA. In addition, the y intercept of the fit is very close to zero (0.06±0.05) suggesting that PcrA begins to translocate from near the junction and that it reinitiates translocation immediately after release of the 5′ tail. We have also conducted the same measurements and analysis on DNA labeled only with Cy3 (Fig. S3E and F) and the results agree well with those from the FRET analysis.

Figure 3. ssDNA length dependence and initiation of looping.

A. Representative single-molecule FRET time traces of repetitive looping obtained for 5′ ssDNA tails of various length at 5 μM ATP. See also Figures S3A-S3E.

B. Average time of each translocation cycle (<Δt>) vs. 5′ tail length at saturating ATP concentration (1 mM). The number of molecules analyzed is as follows with the average number of repetitions per molecule shown in parenthesis: 20 molecules for 5′pdT40 (110), 14 molecules for 5′pdT50 (104), 30 molecules for 5′pdT60 (73), 36 molecules for 5′pdT70 (54), and 34 molecules for 5′pdT80 (48). The error bar denotes STD. Errors in the fit result are in SEM. See also Figure S3F.

C. Before repetitive looping begins, an elevated FRET plateau is observed.

D. Histogram of dwell time of the plateau and an exponential fit. The error in the fit result is in SEM.

Initiation of looping

From the single molecule time traces, we could observe the transition from the constant FRET value of DNA-only to the repetitive looping behavior. Interestingly, we detected a transient plateau of slightly elevated FRET value (E_FRET~0.5) before repetitive looping begins (Fig. 3C). The dwell time histogram of the plateau period could be fitted by an exponential decay with a time constant of 1.0 sec (Fig. 3D). The molecular details of this initiation step require further investigations.

Kinetic step size of PcrA translocation

Noting the remarkable regularity in repetitive looping, we asked how many rate-limiting steps occur within a single cycle of duration Δt. In general, a larger number of equivalent steps per cycle will result in a sharper Δt histogram. We used the simplest possible model of N irreversible steps connecting the beginning and the end of a cycle.

$$\text{begin}\stackrel{\text{N}\text{steps}}{\overbrace{\stackrel{k}{\to }\stackrel{k}{\to }\dots \dots \stackrel{k}{\to }\stackrel{k}{\to }}}\text{end}$$

In this model, the histogram of Δt should follow the gamma distribution, (Δt)^N−1 exp(−kΔt). Any additional step, for example dissociation at the end of the 5′ end or re-initiation of translocation, with a rate different from k, would only broaden the distribution of Δt. If such an extra step occurs with the same rate as the stepping rate, this analysis will overestimate N only by a factor of 1/N. Therefore, the simplistic model we present here should provide a lower limit on the number of steps involved that is accurate within a few percent.

An important assumption in this analysis is that the translocation process has a defined rate of stepping, k. If static disorder exists in the translocation rate and the Δt histogram is obtained from an ensemble of single molecules, static disorder will increase the width of the distribution, leading to an underestimation of N. To avoid such a problem, we built a Δt histogram from a single PcrA molecule that showed 256 cycles of repetitive looping (Fig. 4A). Fitting the single molecule Δt histogram to the gamma distribution yielded N=34±5 steps for the 40 nt tail, strongly supporting a 1 nt kinetic step size. We also checked if a larger step size, for example, a 4 nt kinetic step size deduced from ensemble kinetic measurements of PcrA and UvrD (Niedziela-Majka et al., 2007; Tomko et al., 2007), still gives a reasonable fit. Fitting the 40 nt tail data by fixing N = 10 (=40/4) yielded a much poorer fit (Fig. 4A).

Figure 4. One nucleotide kinetic step size of translocation.

A. A Δt histogram obtained from a single PcrA molecule showing 256 repetitive looping events over 140 sec. A fit to the gamma distribution is shown in (red). Also shown is a fit with the value for N fixed at 10.

B. Gamma distribution fits of Δt histograms for 10 different PcrA molecules (R² > 0.97 for each curve). For clarity, the Δt histograms themselves are not shown.

C. N values determined from 10 different molecules shown in B (Error bars denote SEM). The molecules (A to J) are ordered with increasing N values. For the 40 nt translocation, the expected zones for kinetic step sizes of 1, 2, 3 and 4 nt are indicated as shades of different color.

D. Heterogeneity among the PcrA molecules. k vs. N for the 10 molecules analyzed in B (Error bars denote SEM).

E. Δt histograms and gamma distribution fits obtained from single PcrA molecules on 50, 60, 70 and 80 nt 5′ tail of partial duplex DNA.

F. N vs. ssDNA length for the molecules shown in A and E (Error bars denote SEM)..

G. k vs. ssDNA length for the molecules shown in A and E (Error bars denote SEM)..

H. Δt histogram obtained from 77 molecules and gamma distribution fit (R²=0.99).

The same analysis performed on a total of 10 molecules shows that there is a variation of both k and N values among molecules (Fig. 4D). In addition, the average of Δt showed significant variation among the molecules as can be seen in the overlay of the gamma distribution fits (Fig. 4B). Despite the observed heterogeneity, all molecules showed N larger than 20 for the 40 nt tail (Fig 4C) which is inconsistent with any integer step size larger than 1 nt (Fig. 4B–D). We also observed that N values determined from individual molecules increase as the 5′ tail is increased from 50, 60, 70 to 80 nt without any systematic change in the stepping rate k (Fig. 4E, F and G). Overall, our data places an upper limit on the kinetic step size consistently smaller than 2 nt, and assuming that there exists a defined integer value for the kinetic step size, the kinetic step size of ssDNA translocation is 1 nt.

Previous ensemble kinetic analyses reported that PcrA and UvrD takes average kinetic steps of 4 nt (Niedziela-Majka et al., 2007; Tomko et al., 2007). To test if the observed static disorder in the translocation rate can in principle result in an apparent step size larger than 1 nt, we obtained 9,356 Δt values from a population of 77 molecules and built an ensemble histogram (Fig. 4H). Due to the heterogeneity, the resulting distribution is broader than that obtained from a single molecule and the fit showed N=12 for a 40 nt translocation, i.e. 3.3 nt/step. Therefore, kinetic step sizes significantly larger than 1 nt that were estimated from ensemble measurements could be due to the presence of static disorder in the translocation rates among the enzyme population.

PcrA activity on a forked DNA

Thus far, we have presented data only on the 5′ partial duplex DNA which should not be unwound by a 3′ to 5′ helicase. What would happen if a 3′ tail is added in addition to the 5′ tail, that is, will PcrA unwind the duplex DNA instead of displaying translocation on the 5′ tail? We prepared a forked DNA that is identical in structure and fluorescent labeling to the aforementioned 5′ tailed DNA except for the addition of a 20 nt 3′ tail (Fig. 5A). If the 18 bp duplex is efficiently unwound by PcrA, the donor-labeled strand should be released from the surface resulting in a decrease of the number of fluorescent spots in the imaging area. Full unwinding was negligible because the number of fluorescent spots per imaging area did not decrease significantly even after 30 min of reaction (Fig. 5B). Instead, we observed a robust looping behavior (Fig. 5A) likely due to a stable anchoring of PcrA at the duplex junction with concomitant translocation along the 5′ tail. Therefore, a PcrA monomer favors looping of the 5′ tail of a forked DNA over duplex unwinding.

Figure 5. PcrA on a forked DNA and the role of 2B domain of PcrA.

A. A forked DNA substrate labeled with donor and acceptor at the same position as in Fig. 1A. A representative set of traces of fluorescence intensities and FRET efficiency are presented before and after adding PcrA (100 pM) and ATP (5 μM).

B. Number of molecules per imaging area before and after adding PcrA (100 pM) and ATP (1mM) (Error bars denote STD).

C. Representative time traces of 5′pdT40 in the presence of G423T mutant PcrA and 1 mM ATP. See also Figures S4A-S4C.

D. FRET values at every time point of the trajectory are collected and plotted into a histogram (wild type in red, G423T in green).

E. Dual-labeled PcrA is added to unlabeled 5′pdT40 with 1 mM ATP.

F. Representative time trace of fluorescence intensities and FRET for dual-labeled PcrA. Every one second, a 0.1 second pulse of red illumination (633 nm) and no green illumination (532 nm) is applied in order to verify the presence of acceptor. Periods of green illumination are marked with green bars and red illumination with red bars above the time traces as well as with shades over the traces. See also Figures S4D-S4G.

Role of 2B sub-domain and its conformation

PcrA and UvrD monomers crystallized with a 3′ tailed partial duplex DNA showed that the 2B sub-domain is in contact with the duplex region (Lee and Yang, 2006; Velankar et al., 1999). We hypothesized this duplex/2B contact may allow PcrA to anchor itself at the junction during translocation and to re-initiate translocation specifically from one end of the ssDNA that is proximal to the junction. We mutated G423 in the 2B sub-domain of PcrA, which is the second glycine in the “GIG Box” that comprises the conserved helix-hairpin-helix motif that contacts the minor groove of the duplex DNA (Fig. 5C, Fig. S4A). Mutation on first glycine (G421E) of the GIG box was not viable. On our standard DNA construct (5′pdT40), the G423T mutant showed FRET fluctuations, but these were much less regular than those observed with wild type PcrA, and the FRET peak values were lower than observed with wild type (Fig. 5C and D). In contrast, mutating T426, which is proximal to the GIG box, to alanine did not cause any change in the looping activity (Fig. S4B and C). In UvrD, the mutation equivalent to G423T showed a severe decrease in DNA affinity whereas the mutation equivalent to T426A showed only a minor decrease (Lee and Yang, 2006). Therefore, it is likely that PcrA interaction with the duplex via the GIG box of the 2B sub-domain is important for the regularity of the looping activity.

The 2B sub-domain can swivel between two conformations, termed open and closed (Korolev et al., 1997). To determine the PcrA’s conformation during repetitive looping we constructed a PcrA mutant with two cysteines, A533C and C247, that would be in close proximity, ~3.8 nm, in the closed conformation but farther apart, ~5.5 nm, in the open conformation. The mutant was labeled stochastically with a mixture of Cy3 and Cy5 maleimides, and the labeled protein was observed on an unlabeled 5′pdT40 (Fig. 5E). Fig. 5F shows the initial binding of PcrA to the DNA in the high FRET state, quickly followed by a transition to a low FRET state indicating that PcrA changes its conformation from closed to open at the initial stage of DNA binding and remains in an open conformation during looping (additional examples in Fig. S4D–F). Intermittent direct excitation of the acceptor showed that the low FRET state is not due to acceptor photobleaching or blinking. Among the 189 molecules analyzed, about 80% exhibited primarily a low FRET state that we assign to the open state (Fig. S4G). We conclude that PcrA most likely maintains an open conformation during ssDNA translocation coupled with looping.

PcrA dismantles a pre-formed RecA filament

Because PcrA shows specificity for binding at a 5′-ssDNA/duplex junction and continuous interaction with the 5′ ssDNA tail, PcrA may modulate the activity of other proteins that act on the same DNA. In fact, experiments in vivo showed that anti-recombinase activity is restored by introducing Bacillus subtilis PcrA to a UvrD deficient strain (Bidnenko et al., 2006; Lestini and Michel, 2007). In order to test if PcrA can counteract RecA by directly destabilizing RecA filaments, we added PcrA to a preformed RecA filament (Fig. 6). The RecA filament was formed by incubating 5′pdT40 with RecA and ATP (Fig. 6A). With this DNA, filament formation is observed as a FRET decrease (Fig. 6B) because RecA binding stretches the ssDNA portion, resulting in an end-to-end distance increase (Joo et al., 2006). RecA filament is formed within 5 min and a steady state is reached (Fig. S5A,B and C). 10 min after the RecA addition (250 nM), 1 nM PcrA is introduced while keeping the free RecA and ATP levels in solution constant. A significant portion (>50%) of the low FRET DNA population moved to higher FRET values indicating that RecA is removed (Fig. 6C and D). The equilibrium population of DNA that is occupied by RecA further decreased at higher PcrA concentration (Fig. 6D and F). The new equilibrium induced by PcrA could be reached very rapidly, with more than half the reaction completed in less than 0.5 min (Fig. 6E, Fig. S5D,E and FC).

Figure 6. RecA filament removal by PcrA.

A. and B. 5′pdT40 DNA is pre-incubated with 250 μM RecA and 1 mM ATP for 5 min. FRET histograms are shown before and after incubation. See also Figures S5A-S5C.

C. and D. 1–40 nM PcrA is then added to the pre-formed RecA filament maintaining ATP and RecA concentration constant in solution. FRET histograms were obtained 10 min after PcrA addition. See also Figure S5D.

E. Fraction of RecA remaining vs. time for different PcrA concentration conditions. See also Figure S5E-S5F.

F. Fraction of DNA molecules that lost RecA filament 10 min after PcrA addition (Error bars denote SEM from the exponential fit of Fig. 6E).

G. Representative time traces of fluorescence intensities and FRET efficiency upon adding 1 nM PcrA addition to the preformed RecA filament. See also Figures S5G-S5H.

We could also observe the real-time progress of RecA filament dissociation followed by PcrA-induced looping (Fig. 6G). Upon adding PcrA to a preformed RecA filament, we observe a phase in which slow looping occurs several times, 20 s to 40 s, (the slowness is likely due to competition with RecA) which transitions to fast repetitive looping (at ~ 60 s and beyond) when PcrA temporarily triumphs over RecA. During this fast looping period, the looping rates are comparable to that on bare DNA suggesting that RecA is fully removed (Fig. S5G). It is likely that a pseudo-equilibrium is reached where RecA rebinding competes with PcrA in a concentration dependent manner such that FRET time traces show dynamic changes between fast looping, slow looping and no looping periods (Fig. S5H). Our data overall suggest that a PcrA monomer’s looping/translocation activities can efficiently dismantle pre-formed RecA filaments.

Discussion

Function and the conformation of the 2B sub-domain

The role of the 2B sub-domain in SF1 helicases has been controversial. Because of its interaction with the duplex in the crystal structures, it was once thought that the 2B subdomain acts to destabilize the duplex and helps the unwinding process. However, deletion of the 2B sub-domain actually enhances the unwinding activity of Rep (Brendza et al., 2005; Cheng et al., 2002). The looping activity of PcrA we discovered on 5′-tailed partial duplex DNA requires tight binding to the duplex junction via an anchor subdomain in the helicase. The anchor is likely provided by the 2B sub-domain because when the GIG box in the 2B sub-domain is perturbed by mutation, PcrA lost its ability to pull ssDNA all the way to the end of the track. Such an interaction would be suitable for recognizing stalled replication forks or gapped duplexes.

The 2B sub-domain has been found in two different orientations, open and closed, for Rep, PcrA and UvrD. In the closed conformation of PcrA (Fig. 7A) (Velankar et al., 1999), the 5′ ssDNA tail would need to traverse approximately 34 Å to enter the 1A domain with the correct orientation (Fig. 7C) whereas our tail length-dependent data suggests that translocation begins very close to the duplex region. It is thus likely that the 2B orientation of PcrA during looping differs from what is observed in the PcrA (and UvrD) crystal structures, and our single molecule data indeed indicate that the 2B sub-domain takes an open conformation.

Figure 7. Structural model of PcrA translocation on 5′ ssDNA of partial duplex.

A. Crystal structure of PcrA in closed form in complex with a 3′-tailed partial duplex DNA. PcrA is in the Van der Waals representation with the domains colored individually in yellow (1A), green (2A), purple (1B) and orange (2B). DNA is presented with two colors for each strand: brown (3′ tail strand) and pink (5′ complementary strand).

B. Open form of PcrA generated by 130° rotation of 2B domain together with DNA from the structure shown in A.

C. The ssDNA binding sites of PcrA in the closed form as shown in A. Each protein residue is colored according to the color scheme of the domain it belongs to.

D. The ssDNA binding sites of PcrA in the open form as shown in B. Color schemes are the same as those used in C. Inset shows an enlarged image of the region around 5′ end of the DNA placed in proximity to F192.

E. PcrA translocation model at the DNA fork. The closed form of PcrA is shown inside the pink circle. Others are in the open form. Movement of PcrA is indicated with green arrows. The relative movement of the 5′ ssDNA tail (pink) is shown with pink arrow. See also Figure S6.

We used structural modeling to test if the open conformation would allow a proper engagement with the 5′ tail. We started with the closed form of the PcrA structure (PDB I.D. 3pjr), and rotated the 2B sub-domain together with the 3′-tailed DNA by 130°, to mimic the open conformation of Rep (PDB I.D. 1uaa) (Fig. 7B). This resulted in the 3′ tail being positioned in an incorrect polarity relative to the ssDNA binding site of the protein (Fig. 7D). Interestingly, however, the 5′ end of the duplex is in proximity to the ssDNA binding sites (8.5 Å between F192 and the first base of the 5′ DNA end). As a result, an extended 5′ tail could now interact with the ssDNA binding site of the translocase with the correct polarity starting from F192. As translocation proceeds along the 5′ tail, the ssDNA would be reeled in through the 2A sub-domain and then through the 1A sub-domain (Fig. 7E). A single stranded loop would form behind F192 and grow without steric constraints as translocation continues. We also confirmed that PcrA binds to the 5′ tail with the correct polarity (Fig. S6). Overall, our data and structural modeling suggest that an open conformation is indeed the functional conformation necessary for the reeling-in activity of PcrA.

PcrA translocation mechanism

Stopped-flow studies with reaction synchronization and single turnover experimental design suggested a 4 nt kinetic step size for both PcrA and UvrD (Fischer et al., 2004; Niedziela-Majka et al., 2007; Tomko et al., 2007). Combined with the 1:1 ratio between translocation and ATP hydrolysis(Tomko et al., 2007), it was proposed that PcrA and UvrD translocate via a non-uniform stepping mechanism where it moves about 4 nt very rapidly using 1 ATP per nt, then pauses momentarily before taking the next four steps.

However, the ensemble studies used to estimate the kinetic step size were based on variance analysis that is intrinsically susceptible to any static disorder and they were analyzed with the necessary assumption (Fischer et al., 2004; Tomko et al., 2007) that all molecules in the sample share the same average reaction rate. This assumption, however, is not possible to validate in ensemble studies, and the first single molecule helicase study already showed that another SF1 helicase, RecBCD, can display a high degree of variation in the DNA unwinding rate among nominally identical RecBCD molecules (Bianco et al., 2001). We found that the PcrA translocase also displays significant static disorder in its translocation rate on ssDNA, although the origin of this behavior is not currently understood. Therefore, to avoid complications arising from static disorder, a variance-based analysis should be performed on a single molecule. Our discovery that when PcrA initiates translocation from a ss-ds DNA junction it undergoes repetitive looping of DNA that repeats more than 100 times has provided an ideal platform for determining the kinetic step size from a single molecule, not a collection of single molecules. Our analysis shows that PcrA should translocate with a tight coupling of 1 nt translocated per 1 ATP hydrolyzed and that there is no additional hierarchy of larger steps or pauses for ssDNA translocation.

Implications for the kinetic step size of duplex unwinding

The variance-based method used to estimate the kinetic step size was originally developed for studies of DNA unwinding by UvrD (Ali and Lohman, 1997), and was subsequently adopted for other helicases including RecBCD (Lucius et al., 2002), NPH-II (Jankowsky et al., 2000), E. coli DnaB helicase (Galletto et al., 2004), T7 helicase (Jeong et al., 2004), E. coli RecBC (Wu and Lohman, 2008), hepatitis C virus NS3 (Raney and Benkovic, 1995; Serebrov and Pyle, 2004)., T4 Dda (Eoff and Raney, 2006) and NS3h (Levin et al., 2004) and also for single-molecule analysis of UvrD (Dessinges et al., 2004). In all of these studies except for DnaB, kinetic step sizes of unwinding that are significantly larger than 1 bp were reported. Our results here suggest that the presence of static disorder in the unwinding rates can lead to an overestimation of the kinetic step size. Because a direct detection of unwinding steps or a thorough evaluation of static disorder in the unwinding rate are lacking for these helicases except for NS3, we can not yet conclude whether unwinding by these helicases occurs in steps larger than 1 bp or not. However, a translocation step size of 1 nt does not necessarily mean that unwinding occurs in 1 bp steps because NS3 helicase unwinds DNA (Myong et al., 2007) and RNA (Dumont et al., 2006) in ~3 bp increments even though its translocation along the DNA backbone likely occurs in 1 nt steps (Myong et al., 2007).

Functional implications of ssDNA looping by PcrA

The results reported here, along with previous structural and biochemical studies, suggest that a PcrA monomer can bind to a twin-tailed DNA fork in two modes. In one mode it can bind to the leading 3′ ssDNA strand and the duplex as in the crystal structures. In the second mode it can bind to the 5′ ssDNA and the duplex and in this mode it can reel in the 5′ ssDNA tail using a looping mechanism. The persistent interaction of a PcrA monomer with the 5′ tail of a forked DNA construct suggests that the PcrA monomer looping activity may compete with and thus inhibit DNA unwinding at a fork even at higher PcrA concentrations that support processive unwinding. In fact, a study has shown that a ~10 fold higher concentration of Staphylococcus aureus PcrA was needed to unwind a forked DNA compared to a 3′ partial duplex (Anand and Khan, 2004). Unwinding activity of UvrD on a forked structure was also shown to be lower relative to a 3′-ssDNA structure (Ali et al., 1999).

UvrD can help restart replication by Tus removal (Bidnenko et al., 2006) and dismantle recombination intermediates from a ssDNA gap on the lagging strand of a blocked replication fork (Lestini and Michel, 2007). Interestingly, Bacillus subtilis PcrA, but not Rep, is capable of substituting for UvrD in both of these activities. Our finding that a PcrA monomer can stably anchor itself on the duplex while actively translocating along the 5′ tail provides a plausible mechanism for stripping deleterious RecA filaments from the lagging strand (Flores et al., 2005; Lestini and Michel, 2007; Morel et al., 1993; Veaute et al.,; Zieg et al., 1978). Previous studies in vitro showed that UvrD induced removal of RecA filaments on circular ssDNA that lacks dsDNA (Veaute et al., 2005), and over 100 nM of UvrD and 15 min reaction were required to observe effective RecA removal. In our experiments, using partial duplex DNA molecules, however, we observed effective RecA removal at PcrA concentrations as low as 1 nM and within 30 seconds showing that the anti-RecA activity conserved in PcrA is much more effective in the context of a 5′-ss/ds DNA junction, likely mediated by interactions with the duplex This activity is distinct from that of Rep which interferes with the formation of RecA filament but does not dismantle the pre-formed filament (Myong et al., 2005).

Experimental Procedures

PcrA reaction conditions

Standard reaction buffer was 20mM Tris-HCl pH 8.0, 10 mM KCl, 5 mM MgCl₂, with an oxygen scavenging system (1 mg/ml glucose oxidase, 0.4% (w/v) D-glucose, 0.04 mg/ml catalase, and 1% v/v 2-mercaptoethanol). The measurements were performed at room temperature (21±1°C). 1 mM ATP was used in all experiments unless specified otherwise. Bacillus stearothermophilus PcrA, purified as described (Niedziela-Majka et al., 2007), was first mixed at 100 pM with reaction buffer and added to a flow chamber which has 100 pM DNA specifically immobilized on a PEG-coated quartz surface through biotin-neutravidin linkage (Ha et al., 2002).

Single-molecule experiments

Prism type total internal reflection microscopy was used to acquire single molecule data. A Nd:YAG laser with 532nm wavelength was guided through the prism to generate an evanescent field of illumination. A water-immersion objective (60x, numerical aperture 1.2, Olympus) was used to collect the signal and the scattered light was removed using a 550nm long-pass filter. Cy3 and Cy5 signals were separated using a 630nm dichroic mirror and sent to the CCD camera (iXon DV 887-BI, Andor Technology). Data was recorded as a stream of imaging frames and analyzed with scripts written in IDL to generate fluorescence intensity time trajectories of individual molecules. A time resolution of 0.03 sec was used.

Data Analysis

Singe-molecule fluorescence time trajectories were viewed and analyzed using programs written with Matlab. FRET efficiency, E_FRET, was approximated as the ratio between the acceptor intensity and the sum of the acceptor and donor intensities after correction for the crosstalk between the two detection channels. Peak-to-peak analysis to obtain Δt was measured manually from individual FRET time trajectories aided with a program written in Matlab and the resulting histograms were fitted using Origin. Standard 0.1 sec binning was used for fitting to the gamma distribution because varying the binning size did not change the fitting result significantly.

Highlights.

1. PcrA helicase reels in lagging strands rather than unwinding replicating duplex DNA

2. During this reeling-in process, PcrA strips RecA filaments off the DNA efficiently.

3. PcrA activity requires anchoring to the duplex through the 2B subdomain.

4. PcrA translocates on DNA in uniform steps