Kinetics of DNA Unwinding by the RecD2 Helicase from Deinococcus radiodurans

Materials

Synthetic DNA oligonucleotides were purchased from IDT Corp. and were purified by passage over a P6 size exclusion spin column (Bio-Rad Corp.). Stock solutions of oligonucleotides were made in 10 mm Tris-HCl, pH 8.0, and stored at −20 °C. The sequences of the DNA molecules used in this study are given in Table 1. ATP stock solutions were obtained from GE Healthcare.

Purification of D. radiodurans RecD2

D. radiodurans RecD2 was expressed with a C-terminal hexahistidine fusion tag in the vector pET21a (Novagen) using E. coli strain BL21(DE3) (13). Cells were grown at 37 °C with vigorous shaking overnight in 50 ml of LB medium containing 100 μg/ml ampicillin. This culture was inoculated into 1 liter of ZYP-5052 medium (1% bacto-tryptone, 0.5% yeast extract, 1 mm MgSO₄, 25 mm (NH₄)₂SO₄, 50 mm KH₂PO₄, 50 mm Na₂HPO₄, 2.8 mm dextrose, 5.8 mm lactose, and 0.5% glycerol, with 100 μg/ml ampicillin) for autoinduction of RecD2 expression (22). The culture was shaken overnight at 20 °C. The harvested cells were lysed by sonication, and the RecD2 protein was purified by chromatography on Ni²⁺-NTA (Pro-Bond Resin, Invitrogen Corp.) and ssDNA cellulose (Sigma Aldrich), essentially as described (13). Fractions containing the highest amount of protein (by SDS-PAGE) were pooled and dialyzed overnight at 4 °C against 2 liters of enzyme storage buffer (20 mm potassium phosphate, pH 7.5, 250 mm NaCl, 5 mm DTT, 1 mm EDTA, and 50% glycerol). The purified RecD2 concentration was determined by absorbance at 280 nm using a calculated absorption coefficient of 52,060 m⁻¹ cm⁻¹ (13).

DNA 5′-End Labeling

Oligonucleotide substrate DNAs were 5′-end-labeled in reaction mixtures that contained 107 nm [γ-³²P]ATP (Perkin Elmer), 333 nm DNA, and 15 units of T₄ polynucleotide kinase (Fermentas, Inc.), in reaction buffer supplied with the enzyme. The reactions were incubated for 1 h at 37 °C, at which time 333 nm (final concentration) of non-radioactive ATP was added, and the reaction was allowed to proceed for 30 min more. The 5′ ³²P-labeled DNA was purified using a Qiagen nucleotide clean-up kit (Qiagen, Inc.). The concentration of the isolated DNA was determined by liquid scintillation counting of samples removed before and after the purification procedure.

Binding of RecD2 to dsDNA

Varying concentrations of RecD2 were mixed with a fixed concentration of 5′ ³²P-labeled hairpin DNA (HP1, see Table 1) in 50 mm Tris-HCl, pH 7.4, 10 mm MgCl₂, 1 mm DTT, 0.1 mg/ml BSA, and 5% glycerol. The mixtures were allowed to sit at room temperature (23 °C) for 20 min, glycerol was added to a final concentration of 8%, and the samples were loaded onto a non-denaturing 10% polyacrylamide gel (29:1 acrylamide:bis-acrylamide) in TBE buffer (90 mm Tris, 89 mm boric acid, 1.98 mm EDTA, pH 8.3). Gels were run at 50 mA (constant current) for 1 h, dried, and imaged and quantitated using a Storm phosphorimager and ImageQuant (GE Healthcare). The binding data were fit to Y = Y_maxE_T/(K_d + E_T), where Y is the fraction of the DNA bound, E_T is the total RecD2 protein concentration, Y_max is the maximal fraction bound at high RecD2 concentration, and K_d = [RecD2] [DNA]/[RecD2·DNA].

DNA Unwinding

Double-stranded unwinding substrates were prepared by mixing a 5′ ³²P-labeled ssDNA with its unlabeled complement (20 nm of each DNA) in 20 mm Tris-HCl, pH 7.5, 50 mm NaCl, and 8 mm MgCl₂. The mixtures were heated to 96 °C in a water bath for 3 min, cooled to room temperature, and then stored at 4 °C before use.

For DNA unwinding reactions done manually (15 s to 5 min time scale), the reaction mixtures contained (final concentrations in 100 μl) 50 mm Tris-HCl, pH 7.4, 10 mm MgCl₂, 2.5 mm DTT, 0.1 mg/ml BSA, 5% glycerol, and 1 nm 5′ ³²P-labeled dsDNA substrate. RecD2 was added, and the mixture was incubated for 20 min at 23 °C. A 10-μl sample was removed as a zero time point and mixed with 3.5 μl of 4× helicase quench solution (40% glycerol, 2.4% SDS, 100 mm EDTA, and 0.1% bromphenol blue. The quench solution also contained 20 nm of unlabeled ssDNA that was the same sequence as the radiolabeled strand in the unwinding substrate, added to prevent reannealing of the unwound radiolabeled product strand.) ATP was added to 1 mm (final concentration) to initiate DNA unwinding, and samples were removed and quenched. In some reactions, a ssDNA oligonucleotide (2 μm, 40 nt) was added to bind to and inhibit RecD2 that did not bind to the radiolabeled dsDNA in the preincubation. This protein trap ssDNA was either added along with the ATP to start the reaction, included in the preincubation mixture along with the ³²P-labeled dsDNA substrate, or omitted completely.

The quenched unwinding samples were analyzed on 15% polyacrylamide gels (29:1 acrylamide: bis-acrylamide) in TBE. The gels were run for 1 h at 150 V (constant voltage), dried, exposed, and imaged as above. The fraction of DNA unwound at each time point was determined by quantitation of the amount of [³²P]dsDNA and [³²P]ssDNA present in each lane, using ImageQuant. The fraction of [³²P]ssDNA present in the t = 0 lane was subtracted from the fraction unwound at each time point.

Unwinding Reactions Analyzed by Rapid Quench-Flow

Certain substrates were unwound too rapidly by RecD2 to conduct accurate experiments by hand. Unwinding of these substrates was studied using a KinTek model RQF-3 Rapid Quench Flow instrument (KinTek Corp.). One reaction line was loaded with RecD2 and 2 nm 5′ ³²P substrate dsDNA, and the second was loaded with 2 mm ATP. Both lines also contained 50 mm Tris-HCl, pH 7.4, 2 mm DTT, 0.1 mg/ml BSA, 10 mm MgCl₂, and 5% glycerol. In reactions where protein trap ssDNA was added along with the ATP, the 40-mer ssDNA trap was included at 4 μm. The two reaction components were mixed to initiate the reaction, and then mixed with quench solution (400 mm EDTA and 0.7% SDS). The quenched reaction mixture was expelled into a receiving tube that contained enough non-radioactive ssDNA (corresponding to the ³²P-strand in the substrate) to give a final concentration of 5 nm in the quenched mixture. Samples of the quenched reaction mixture (10 μl) were mixed with 3.5 μl of 4× helicase quench solution and then analyzed by electrophoresis as above.

Analysis of Unwinding Kinetics by Simulation

Kinetics data for reactions with the 12-bp substrate in the presence of the protein trap ssDNA were simulated with the mechanism shown in Fig. 4C, using the KinTekSim program (vers. 3.20, KinTek Corp.). We assumed: 1) identical rate constants for the unwinding steps (k_unw), 2) the enzyme can dissociate from the partially unwound E·DNA complexes (k_off), and 3) any enzyme that dissociates from a partially unwound DNA molecule is bound by the protein trap ssDNA and cannot reinitiate unwinding. Thus a dsDNA from which the enzyme dissociates cannot be unwound during the reaction. We assumed a maximum possible extent of unwinding of 85% (fraction unwound = 0.85), the reaction extent observed for the 12-bp substrate when no protein trap ssDNA was added (see Fig. 4A). It may be that some radiolabeled substrates are refractory to unwinding for some reason. In addition, we subtracted out the small amount of ssDNA that was often present in the samples taken before the unwinding reaction was started (12% of the dsDNA), which has the effect of reducing the calculated maximum extent of unwinding.

Kinetics data with the 20-bp substrates in the absence of protein trap ssDNA were simulated with the mechanism in Fig. 5D. For simulations that used the full mechanism in Fig. 5D, values for all rate constants were specified first. Then, concentrations of all species present in the initial mixture (E, E₂, dsDNA, and E·dsDNA) that were consistent with the rate constant values were determined as follows. We used the inferred values for the rate constants k_+E, k_−E, k₊₁, and k₋₁, and successive approximation to find the value of [E] that satisfied the following set of relations: [E₂] = [E]²/(k_+E/k_−E), [E·dsDNA] = [E](D_T)/{(k₊₁/k₋₁) + [E·dsDNA]}, and E_T = [E] + [E·dsDNA] + 2[E₂], where D_T and E_T are the total DNA and enzyme concentrations. These rate constants and the calculated values of [E], [E₂], [E·dsDNA], and [dsDNA] were then used to simulate the reaction time course using KinTekSim. New values of the rate constants were then selected, the values of the concentrations recalculated, and the simulation repeated, until the simulated time course adequately reproduced the data. For simulations that omitted the RecD2 self-association step in Fig. 5D (the steps with rate constants k_+/−E), initial values of [E], [dsDNA], and [E·dsDNA] were calculated using the guessed values for k_+/−1 and K_d = k₋₁/k₊₁, [E·dsDNA] = {(D_T + E_T + K_d) − [(D_T + E_T + K_d)² − 4D_TE_T]^1/2}/2, [dsDNA] = D_T − [E·dsDNA], and [E] = E_T − [E·dsDNA].

Size Exclusion Chromatography

Size exclusion chromatography analysis was done using a Superdex 200 10/300GL column on an Akta FPLC system (GE Healthcare), with 0.25 ml/min flow-rate at 23 °C. The column was equilibrated with 50 mm potassium phosphate, pH 7.5, and 150 mm NaCl. RecD2 (0.5 ml) was injected in storage buffer. Molecular weight standards were from Bio-Rad. Protein elution was monitored by UV absorbance at 280 nm.

Glutaraldehyde Cross-linking

Cross-linking mixtures contained 1 μm RecD2 in 25 mm sodium HEPES, pH 7.5, 10% glycerol, 1 mm EDTA, 1 mm β-mercaptoethanol, and 50 mm NaCl. The 12-nt 5′-tailed hairpin DNA (HP1, 1 μm, see Table 1) was included in one mixture. The mixtures were incubated for 20 min at 23 °C, and then glutaraldehyde (PolySciences, Inc.) was added to 1.4 mm, final concentration. The mixtures were quenched after 20 min at 23 °C by adding 7.5 μl of 1 m Tris-HCl, pH 8.0. Some quenched mixtures (75 μl) were then treated with benzonase nuclease (0.25 units/μl, Novagen Corp.) for 30 min at 37 °C, or with sodium borohydride (2 μl of 1 m NaBH₄) for 5 min at 23 °C, or with both reagents for 30 min at 37 °C. The cross-linked samples were analyzed by SDS-PAGE on 7.5% polyacrylamide gels followed by silver staining.

Unwinding the 20-bp, 12-nt 5′ ssDNA-tailed Substrate

We then sought to measure the rate of DNA unwinding by RecD2 using oligonucleotide substrates. The first substrate tested was a 20-bp duplex DNA with a 12-nt 5′ single-stranded extension (12 nt:20 bp, see Table 1). The DNA was unwound essentially to completion within about 1 min, in assays that contained 1 nm DNA and 100 nm RecD2 (Fig. 2, A and D).

We then attempted to measure DNA unwinding under single turnover conditions. RecD2 was mixed first with the ³²P-labeled dsDNA substrate to allow formation of an enzyme-DNA complex. The reaction was initiated by adding ATP and a large excess of single-stranded DNA to trap and inhibit RecD2 that is not bound to the substrate DNA and limit the bound enzyme to a single turnover. As a control, the protein trap ssDNA was added to the initial mixture of RecD2 and ³²P-labeled dsDNA substrate. We found that 2 μm of a 40-nt ssDNA oligonucleotide inhibited substantially the unwinding reaction when it was included in the preincubation mixture of RecD2 and [³²P]dsDNA substrate (Fig. 2, B and D). The same amount of ssDNA also inhibited to about the same degree when it was added along with the ATP (Fig. 2, C and D). Similar inhibition by the protein trap ssDNA in both protocols was observed in reactions containing higher ATP concentrations (0.5–10 mm ATP, with [MgCl]₂ also varied to maintain constant free Mg²⁺; data not shown).

We found previously that 20-bp DNA substrates with 5′-ssDNA tails shorter than 8 nt were poor unwinding substrates for RecD2, whereas 10- and 12-nt 5′-tailed substrates were unwound equally well (13). Lengthening the ssDNA tail beyond the minimum length required for unwinding has been found to increase the efficiency or extent of DNA unwinding by some helicases (23, 24). We tested whether a longer 5′-tail would affect the RecD2-unwinding reaction, using a 20-bp duplex substrate with a 24-nt 5′-ssDNA terminal extension (24 nt:20 bp, twice the tail-length as on the substrate used in Fig. 2). As observed for the 12-nt 5′-tailed substrate, this extended 5′-tailed DNA was unwound in the reaction with no protein trap, but there was little unwinding when the protein trap DNA was added with the ATP (Fig. 3). Thus a longer ssDNA tail did not affect the RecD2-unwinding kinetics.

Unwinding of a 12-bp Substrate

We then carried out the unwinding reactions with a shorter, 12-bp DNA duplex, with a 20-nt 5′-ssDNA tail (20 nt:12 bp; Fig. 4). This DNA was unwound rapidly in the reaction with no protein trap ssDNA, and the trap ssDNA inhibited significantly when added along with the ³²P-labeled substrate. Now, however, there was significant unwinding of the 12-bp substrate despite the trap ssDNA, when the RecD2 was allowed to bind first to the [³²P]DNA, and the trap was added along with the ATP (Fig. 4A). This result shows that a preformed RecD2-DNA complex is able to unwind the 12-bp DNA with only a small inhibitory effect of the protein trap ssDNA. The results suggest that the low extent of reaction with the longer 20-bp substrates (Figs. 2 and 3) is due to the low processivity of RecD2. The reaction with the 12-bp substrate is quite fast, and so we repeated the single turnover experiment using a rapid-quench device to detect a more complete reaction time course (Fig. 4B).

Unwinding a Forked DNA Substrate

We found previously that a 20-bp duplex with a forked end (having both 5′ and 3′ 12-nt ssDNA extensions) was bound more tightly and unwound more rapidly than the 12-nt 5′-tailed 20-bp substrate (13). We tested this DNA substrate (fork:20 bp) in the single-turnover reaction protocol (Fig. 5). As observed previously, the forked DNA is unwound more rapidly than the 5′-tailed substrate, in the reaction with no protein trap. The protein trap ssDNA inhibited the reaction when premixed with the ³²P-substrate (Fig. 5A), although the inhibition was not as great as with the 12-nt 5′-tailed substrate in Fig. 2. However, the amount of DNA unwound in the reaction where the protein trap was added with the ATP was no greater than that where the ssDNA trap was included in the preincubation mixture (Fig. 5A). (Addition of higher concentrations of the protein trap ssDNA completely suppressed the reaction whether added before or along with the ATP (data not shown).) We verified that RecD2 binds to the forked substrate in the preincubation mixtures, by removing samples before ATP was added and analyzing on a native polyacrylamide gel, as in Fig. 1. The result shows that essentially all of the DNA was bound to RecD2 in samples removed from the mixture that did not contain protein trap ssDNA (Fig. 5C, lane 1). Little RecD2 was bound to the forked substrate DNA when the protein trap ssDNA was present in the preincubation (Fig. 5C, lane 2). Overall, the results show that the apparent processivity with the 20-bp forked substrate is low, as seen with the 5′-tailed substrates in Figs. 2 and 3. Unwinding of the forked DNA is quite rapid, and so we repeated the time course measurement using the rapid quench instrument (with no protein trap ssDNA) to obtain a complete time course (Fig. 5B).

The reaction with the forked substrate was significantly slower at 1 nm RecD2 (Fig. 6) than at 100 nm RecD2 (compare Fig. 6 (1 nm RecD2) and Fig. 5 (100 nm)). The protein trap was also more effective at suppressing the reaction than with 100 nm RecD2, and there was very little DNA unwinding when the trap was added with the ATP (data not shown). The reaction with the 12 nt:20 bp substrate was also slower at 1 nm RecD2 (Fig. 6) than at 100 nm RecD2. The unwinding reaction with the 5′-ssDNA substrate (12 nt:20 bp) was slower than that with the forked DNA, at all RecD2 concentrations, as observed previously (13).

Simulation of RecD2 DNA Unwinding Kinetics: 12-bp Substrate

We simulated the RecD2 unwinding kinetics data to gain insight into the reaction mechanism and quantitative estimates for the rate constants. The single turnover time course data for the 20 nt:12 bp substrate do not fit well to a single exponential, as there is a short lag before unwound DNA appears (see Fig. 4B). We then simulated the data using a mechanism that requires two kinetic steps for complete unwinding of the DNA substrate (Fig. 4C). The data for the 20 nt:12 bp substrate could be reproduced quite well using this mechanism, with k_unw = 5.5 s⁻¹ and k_off = 1.9 s⁻¹ (Fig. 4B). The processivity of the enzyme, expressed as p = k_unw/(k_unw + k_off), is then p = 0.74. (The magnitude of k₋₁ could not be determined from these data, because the simulations work well with k_unw ≫ k₋₁. The value of k₋₁ used for the simulated curve in Fig. 4B (k₋₁ = 0.02 s⁻¹) is the value for k₋₁ used for the 12 nt:20 bp substrate (see below and Fig. 6)).

We also did simulations in which we relaxed the assumption that the rate constants for the two unwinding steps are identical. Not surprisingly, the data could be simulated well if the two steps were allowed to have different values for k_unw. If unwinding of the first step was made faster, then unwinding of the second step had to be correspondingly slower, and the ratio of k_unw for the second step to k_off had to be ∼2.8–3, to reproduce the reaction time course and the final extent of reaction. Moreover, the data could not be simulated well with k_unw for the first step less than about 3 s⁻¹, or greater than about 8 s⁻¹, with k_unw for the second step and k_off varied (not shown).

Simulation of Unwinding the 20-bp Forked Substrate

We then attempted to simulate the unwinding kinetics for the 20-bp substrates using rate constant values that worked well for the 12-bp substrate. We used data only from the reactions without protein trap ssDNA, in which the DNA was unwound to a significant extent. The simulated mechanism therefore includes steps for enzyme binding to and dissociation from the dsDNA substrate (Fig. 5D). The mechanism also includes more unwinding steps than used for the 12-bp substrate. We assumed for simulation that all unwinding steps have the same rate constant, and that a partially unwound DNA from which the enzyme dissociates would re-wind. Thus, the enzyme can bind only to the fully based-paired DNA substrate molecule.

We first simulated the data for the forked substrate obtained at 100 nm RecD2 in the absence of protein trap ssDNA using the mechanism in Fig. 5D, but omitting the dimerization of the enzyme shown in this mechanism (k_+/−E). The data could be reproduced well (see Fig. 5B) with k_unw = 4.5 s⁻¹, k_off = 1.5 s⁻¹, k₊₁ = 4 × 10⁶ m⁻¹ s⁻¹, and k₋₁ = 0.004 s⁻¹. We then attempted to simulate the unwinding time course for the forked substrate with 1 nm RecD2 using these same rate constant values. The observed reaction with 1 nm RecD2 in Fig. 6 was significantly faster than predicted (not shown).

We reasoned that the discrepancy in predicted rates at 1 and 100 nm RecD2 could be explained if the RecD2 self-associates at high protein concentration into a form that is unable to bind to the DNA. We assumed the simple case of reversible dimerization and incorporated this into the simulation mechanism (Fig. 5D). The data for the forked substrate at both 1 and 100 nm RecD2 could be simulated reasonably well using this mechanism, with the following rate constant values: k_+E = 5 × 10⁶ m⁻¹ s⁻¹, k_−E = 0.02 s⁻¹, k₊₁ = 5 × 10⁷ m⁻¹ s⁻¹, k₋₁ = 0.02 s⁻¹, k_unw = 4.5 s⁻¹, and k_off = 1.9 s⁻¹ (Figs. 5B and 6). The values of k_unw and k_off are very similar to those that simulate the data for the 12-bp substrate (see above). The simulated curve for the reaction at 100 nm RecD2 (Fig. 5B) predicts, and reproduces, three phases in the reaction: 1) an initial lag before completely unwound ssDNA product is observed, 2) a rapid burst of DNA unwinding by those enzymes bound initially to the substrate that unwind the entire substrate without dissociating, despite the low processivity (∼20% of the DNA unwound, in the first ∼1 s of the reaction), and 3) a third, slower phase of enzymes binding to and unwinding DNA substrates that were not unwound in the initial burst. A previous study of RecD2 found that the enzyme dissociates slowly from ssDNA (k_diss = 0.14 s⁻¹) (19). We simulated the unwinding data using a kinetic mechanism that included explicitly the dissociation of RecD from the unwound DNA product, with rate constant = 0.14 s⁻¹. We find that including this step makes no difference in the simulated time courses (not shown).

Simulation of Unwinding the 12 nt:20 bp Substrate

Finally, we tested whether we could simulate the results with the 12 nt:20 bp substrate. As for the forked substrate, the data at both 1 and 100 nm RecD2 could not be simulated unless we allowed for reversible dimerization of RecD2 (not shown). The data were simulated reasonably well with this mechanism and k_+E = 5 × 10⁶ m⁻¹ s⁻¹, k_−E = 0.02 s⁻¹, k₊₁ = 1 × 10⁷ m⁻¹ s⁻¹, k₋₁ = 0.02 s⁻¹, k_unw = 4.5 s⁻¹, and k_off = 1.9 s⁻¹ (Figs. 2D and 6). These are the same rate constant values that worked well for the forked substrate (see above), except that the rate constant for RecD2 binding to the DNA was set to k₊₁ = 1 × 10⁷ m⁻¹ s⁻¹ for the 12 nt:20 bp substrate, as compared with k₊₁ = 5 × 10⁷ m⁻¹ s⁻¹ for the forked substrate. Thus the difference in unwinding rate between the forked and singly tailed substrates can be accounted for by a difference in the rate of association between RecD2 and the two dsDNA substrates.

Finally, we used the rate constant values (k_+/−E and k_+/−1) from simulation of the 12 nt:20 bp unwinding kinetics to simulate the equilibrium binding to the 12-nt 5′-tailed DNA hairpin. The values of [E], [E₂], [E·dsDNA], and [dsDNA] were calculated (see “Experimental Procedures”) for mixtures containing 1 nm DNA and the RecD2 concentrations (E_T) used in the binding experiment. A plot of [E·dsDNA] versus E_T was constructed and fit to Y = Y_maxE_T/(K_d + E_T), with a fitted value of K_d = 2.7 (Fig. 1). The simulated binding curve fits the data quite well, and so the results from the kinetics simulations are consistent with the binding results.

RecD2 DNA Unwinding Mechanism

A number of previous studies show that DNA helicases require multiple steps for complete unwinding of short dsDNA substrates (23, 27, 28). The kinetic step size of a helicase is the number of base pairs unwound between two successive rate-limiting (kinetically significant) steps in the unwinding cycle (29). The simulations of DNA unwinding by RecD2 indicate that two steps are required to unwind completely the 12-bp substrate, suggesting 6 bp unwound per step. However, the enzyme presumably need not unwind all the base pairs in the substrate for complete strand separation and appearance of the product observed in a gel, since the two strands will separate spontaneously when only a small number of base pairs remain intact (24, 30). Thus, the actual kinetic step size for RecD is ∼3–4 bp/step, if the last ∼4–6 bp can unwind spontaneously. The rate constant k_unw = 5.5 steps/s then indicates an unwinding rate of about 15–20 bp/s. This is greater than our previous estimate from reactions under steady-state conditions with 20-bp substrates (2 bp/s, see Ref. 13), as expected since the previous estimate was confounded by the requirement for multiple enzyme associations with the substrate for complete unwinding.

RecD2 has been shown to carry out ATP-dependent unidirectional translocation along ssDNA (19). The translocation rate on ssDNA (95 nt/sec) was found to be significantly faster than the dsDNA unwinding rate reported here. The translocation and ATP hydrolysis rates on ssDNA are essentially the same, indicating that RecD2 moves along ssDNA by one nucleotide per ATP hydrolyzed (19). This step size is consistent with the conformation change in the RecD2·ssDNA complex induced by binding of an ATP analogue, determined by x-ray crystallography (19). Several other DNA helicases have been found to move on ssDNA by 1 nt per ATP hydrolyzed, based on kinetics and/or x-ray crystallography (31–33). For many of these helicases, as with RecD2 in this work, the kinetic step size during DNA unwinding is greater than 1 bp unwound per step (34–36). Our results for dsDNA unwinding, along with the data in Ref. 19, suggest that DNA unwinding by RecD2 involves reaction steps in addition to those required for translocation on ssDNA. These could involve a slow DNA unwinding step after several fast ATP hydrolysis/translocation reactions, or slow release of a partially unwound DNA strand from a binding site on the enzyme (34, 36).

RecD2 Properties Compared with Other DNA Helicases

The isolated E. coli RecD protein, like RecD2, is a helicase that can unwind 5′-tailed dsDNA slowly and with low processivity (21), but the poor solubility of this protein has precluded its quantitative characterization (20, 21). Despite its low DNA unwinding activity in isolation, the RecD subunit contributes to rapid DNA unwinding by RecBCD. A RecBCD mutant enzyme in which the RecB subunit is inactive, and which therefore depends on RecD for translocation on the DNA, still unwinds DNA very rapidly (480 bp/s), about one-third as fast as the wild-type enzyme (38). Thus the unwinding activity of the RecD subunit is greatly enhanced by assembly with RecB and RecC to make RecBCD. It is not known whether RecD2 interacts functionally with other proteins, which could also alter its kinetic properties.

After the RecD subunits of RecBCD enzymes, the Dda helicase from phage T4 and the TraI helicase are the next closest relatives to the RecD2 proteins, based on sequence analysis (13). Dda and TraI both require a 5′-terminal ssDNA extension for DNA unwinding (39, 40), but otherwise there is great disparity in their enzymatic properties. Dda is similar to RecD2 in kinetic step size and low processivity, but its unwinding rate is greater than RecD2 (23, 39). TraI, which at 192 kDa is much larger than either RecD2 or Dda, unwinds DNA very rapidly (1100 bp/sec) and with high processivity, and has a 6–8 bp step size (40). Unwinding rates for other SF1 helicases that are more distantly related to RecD2 vary from 30 bp/s for PcrA (32)) and 68 bp/s for UvrD (41)), to 790 bp/s for RecBCD (28). However, the kinetic step sizes of these enzymes are similar to that of RecD2: UvrD (4–5 bp/step (27)), PcrA (4 bp/step (35)), and RecBC and RecBCD (∼3–4 bp/step (28, 37)). The similarity in kinetic step size suggests an underlying mechanistic similarity in DNA unwinding by these helicases, despite the differences in their unwinding rates.

The oligomeric form of a SF1 helicase that is active during DNA unwinding has been somewhat controversial, with evidence for an active monomer in the cases of Dda and TraI (23, 40) and a dimer for UvrD and PcrA (25, 26, 35). The results from size exclusion chromatography and cross-linking are consistent with reversible oligomerization of RecD2, which is reduced when the protein binds to DNA. The dependence of the unwinding kinetics on RecD2 concentration could be simulated only by allowing for association of RecD2 into an inactive oligomer, modeled as a dimer. We did not test any other oligomeric states beyond dimer, and further study of RecD2 is required to determine the active form of RecD2. However, the results show no evidence for requirement of a dimer for DNA unwinding in the protein concentration range tested.