Active DNA unwinding dynamics during processive DNA replication
- aInstituto Madrileño de Estudios Avanzados, Nanoscience, Cantoblanco, 28049 Madrid, Spain;
- bCentro de Estudios Avanzados de Cuba, Havana 17100, Cuba;
- cDepartamento Física Atómica, Molecular y Nuclear, Universidad Complutense, 28040 Madrid, Spain;
- dCentro de Biología Molecular Severo Ochoa (Consejo Superior de Investigaciones Científicas-Universidad Autónoma de Madrid), 28049 Madrid, Spain; and
- eDepartamento Estructura de Macromoléculas, Centro Nacional de Biotecnología (Consejo Superior de Investigaciones Científicas), Cantoblanco, 28049 Madrid, Spain
See allHide authors and affiliations
Contributed by Margarita Salas, March 20, 2012 (sent for review January 20, 2012)

Abstract
Duplication of double-stranded DNA (dsDNA) requires a fine-tuned coordination between the DNA replication and unwinding reactions. Using optical tweezers, we probed the coupling dynamics between these two activities when they are simultaneously carried out by individual Phi29 DNA polymerase molecules replicating a dsDNA hairpin. We used the wild-type and an unwinding deficient polymerase variant and found that mechanical tension applied on the DNA and the DNA sequence modulate in different ways the replication, unwinding rates, and pause kinetics of each polymerase. However, incorporation of pause kinetics in a model to quantify the unwinding reaction reveals that both polymerases destabilize the fork with the same active mechanism and offers insights into the topological strategies that could be used by the Phi29 DNA polymerase and other DNA replication systems to couple unwinding and replication reactions.
Replication of dsDNA requires overcoming the energetic barrier associated with the base pair melting of its double helix. In many DNA replication systems, replication and unwinding of the fork are carried out by the coordinate action of different proteins, the DNA polymerase holoenzyme and the replicative helicase, respectively (1, 2). In other systems, like the bacteriophage Phi29, these two activities are coupled within the replicative DNA polymerase (3, 4). The Phi29 DNA polymerase presents the common folding and catalytic activities characteristics of the Family B DNA polymerases (4) but in addition, it also presents an amino acid insertion in the polymerization domain, named TPR2 (Fig. S1), which confers the protein a processive strand displacement activity (5, 6). The TPR2 insertion together with the thumb, palm, and exonuclease (exo) domains forms a narrow (10 Å diameter), closed tunnel around the template strand, in comparison with the open channel described in other related DNA polymerases (5). Interestingly, this topological restriction resembles the one imposed by hexameric replicative DNA helicases at the fork junction (7, 8) and requires the dsDNA in front of the polymerase to open in order for the template to enter the active site. Therefore, the Phi29 DNA polymerase works as a hybrid polymerase-helicase and constitutes a good model system to understand the basic mechanistic principles of the coupling between DNA replication and unwinding reactions.
The physical mechanism by which the polymerase replicates and promotes DNA unwinding could be described as passive; (the polymerase will not enter the duplex region until thermal fluctuations transiently open the dsDNA junction), or active (it destabilizes the duplex DNA near the junction shifting the equilibrium of the fork toward opening) (1, 9⇓–11). The rate of a passive motor would be more sensitive than the rate of an active motor to the strength of the fork ahead (11). These two behaviors are the two extremes of a continuous spectra, and the polymerase is expected to present an unwinding mechanisms located between an ideally active and totally passive behavior (11). We used optical tweezers and a detailed kinetic analysis to probe the coupling between DNA unwinding and replication reactions by studying the strand displacement and primer extension activities of the wild-type Phi29 DNA polymerase and a strand displacement deficient variant (12) under the same external (destabilizing) tension on the DNA and sequence context. The particular effect of fork stability on the average unwinding rate and pause kinetics of each polymerase suggests a molecular mechanism for the coupling of DNA replication and unwinding reactions on the Phi29 DNA polymerase which could be extrapolated to understand the coupling dynamics between these two reactions in other DNA replications systems.
Results
Single-Molecule dsDNA Replication-Unwinding Assay.
We used optical tweezers to apply mechanical tension to the ends of the complementary strands of a single 556 bp long DNA hairpin containing a unique 3′ end polymerase loading site and monitor at constant force (below 13 pN where the hairpin is stably folded) the end-to-end distance change of the DNA as the Phi29 DNA polymerase replicates the hairpin in the presence of dNTPs (Δx1 and Δx2, Fig. 1). Initially, during the strand displacement phase, the protein moved through the hairpin using one strand as a template to incorporate the corresponding complementary nucleotide into the 3′ end of the primer and at the same time displacing the complementary strand, so the distance between the beads increased (Δx1, Fig. 1). Upon reaching the end of the hairpin, no more unwinding is required, and the polymerase continues replication in the primer extension mode, converting the displaced complementary single strand (ss-) to dsDNA. This reaction also changes the distance between the beads (Δx2, Fig. 1) due to the different elastic properties of each DNA polymer (13). The Δx1 and Δx2 were converted to the number of nucleotides incorporated as a function of time at a given force (Methods and Fig. S2).
Overview of experimental assay. (A) Schematic representation of the experimental design. A single DNA hairpin was tethered to functionalized beads inside a fluidics chamber. One strand of the hairpin (blue) is attached through a dsDNA handle to a bead held in the laser trap, while the complementary strand (red) is attached to a bead on top of a mobile micropipette (Methods). At a constant tension (f) the strand displacement and primer extension activities of the polymerase (red triangle) are detected as a change in distance between the beads, Δx1 and Δx2, respectively. (B) Representative replication traces of the wild-type (blue) and sdd mutant (green) polymerases showing in detail the distance changes during strand displacement (s.d., Δx1) and primer extension (p.e., Δx2) activities (f ∼ 11 pN). See Fig. S3 for additional experimental traces at different tensions.
To determine the effect of sequence on the polymerase activity, the hairpin includes four GC base pair clusters with one to four repetitions of the GCC sequence, separated by an approximately 100 nucleotide low-GC content sequence, the GC clusters constituting barriers to DNA unwinding (Fig. S1). Therefore, this experimental setup allowed us to study and compare the particular effect of mechanical tension and sequence on both the strand displacement and primer extension reactions of the wild-type and the strand displacement deficient mutant (sdd) polymerases.
Different Effects of Fork Stability on the Unwinding Rates of the Wild-Type and sdd Mutant Polymerases.
The average strand displacement rate of the wild-type polymerase increased moderately with force, from approximately Vsd = 80 nt/s (3.5 pN) to Vsd = 120 nt/s (13 pN), a velocity similar to its primer extension rate along the displaced complementary strand (Vpe = 128 nt/s) (Fig. 2A). In contrast, the average strand displacement rate of the sdd mutant polymerase presented a much stronger force dependence, increasing by a factor of 9, from Vsd = 10 nt/s (4.2 pN) to approximately 90 nt/s (12 pN), also approaching its primer extension rate, Vpe = 128 nt/s (Fig. 2C). Both polymerases present an identical primer extension rate which is independent of tension below 14 pN (Fig. 2 A and C), indicating that template tension within this range does not affect the rate-limiting step of the replication cycle, as reported previously for these polymerases (14). Therefore, these data indicate that, during strand displacement, unwinding of the dsDNA fork is the force dependent step of the reaction, which limits moderately the wild-type and strongly the mutant polymerase advance through the hairpin (Vsd/Vpe = 0.67 vs. 0.08, respectively).
Tension and sequence dependencies of the wild-type and sdd mutant rates. Only activities that replicate the full hairpin length were considered (Fig. S4A). (A) The tension dependency of the average strand displacement rate (Vsd) of the wild-type polymerase with and without pauses (full and empty blue circles, respectively) is well explained by our active model (solid blue line). (B) During strand displacement the wild-type polymerase spends longer times (blue, f ∼ 4 pN, N = 15) at the more stable hairpin positions (dotted lines). This behavior is well predicted by the proposed model (orange). For comparison, the experimental force-unzipping curve of the hairpin is shown in gray. (C) The different tension dependencies of the average strand displacement rate (Vsd) of the mutant with and without pauses (full and empty green circles, respectively) can be explained with the same active model when pause kinetics are included in (solid green line) or excluded from (dashed green line) the model. (D) Representative position versus time traces of the mutant polymerase during strand displacement. Identified pause events are shown in red, long pauses are located at the GC rich positions of the hairpin (gray lines). Insert shows the sdd mutant velocity distribution during strand displacement conditions (Fig. S5C). (A and C) For both polymerases, full and empty red circles represent the average primer extension rates (Vpe) with and without pauses. Solid red lines represent the sequence independent velocity value used in the model (128 nt/s). Error bars represent the standard error.
Next, we measured the effect of sequence on the rates of both polymerases. For the wild-type polymerase during strand displacement, a direct comparison between the average residence time of the protein at each position in the hairpin and the unfolding pattern of the hairpin already shows that this protein spends longer times at the four hairpin positions with the greater GC content (Fig. 2B and Fig. S3B). Quantification of this effect revealed that the unwinding rate of the more stable GC pairs is approximately three times slower than the unwinding of the AT pairs (VsdGC/VsdAT = 0.36), which in turn is close to the primer extension rate (VsdAT/Vpe = 0.86) (Methods and Fig. S3D). Interestingly, the sdd mutant polymerase during strand displacement (but not during primer extension) presents long pauses precisely located at the GC rich positions of the hairpin (Fig. 2D and Fig. S4B). Since we found that template sequence has no significant effect on the primer extension rate of the wild-type and sdd mutant polymerases (Fig. S3 C and D), these data indicate that unwinding of the more stable fork positions has different effects on each polymerase: (i) Slows down the wild-type polymerase translocation and (ii) promotes the entrance of the sdd mutant polymerase to a long-lived inactive pause state. The different effect of fork stability on the unwinding reaction of each polymerase suggests differences in their unwinding mechanisms (11).
The Wild-Type and Mutant Polymerases Present Different Pause Kinetics During DNA Unwinding.
The data show that inactive pause states may be relevant during the unwinding reaction (at least for the sdd mutant). Therefore, to further investigate the unwinding mechanism of each polymerase, we analyzed in detail their characteristic pause behavior and its relation to their primer extension and strand displacement activities. We first checked that for both enzymes pauses do not correspond to the replacement of polymerase molecules at the 3′ end of the primer (Fig. S4).
For the wild-type polymerase only occasional and short-lived pause events were detected during strand displacement and primer extension conditions (Methods and Fig. 3A). During both replication conditions pauses present very similar characteristics: The pause length frequency distribution (SI Text, Methods) is consistent with a single-step Poisson statistics (Fig. 3B). The pause frequency (pause/second) decreases exponentially with tension and increases linearly with the GC density of the DNA (Fig. 3 C and D), while the average pause length (approximately 0.6 s) is tension and sequence independent (Fig. S5). These similarities indicate that the presence of the fork does not alter much the pause kinetics characteristic of DNA replication and does not promote entering an alternative pause state, strongly suggesting that these short pause states are related to the biochemical replication cycle rather than to DNA unwinding reaction. During primer extension conditions the sdd mutant also presents short pauses, with very similar rates, tension, and sequence dependencies than the wild-type (Fig. 4 and Table 1), suggesting an identical nature for the short inactive states during the replication cycle of both polymerases. This is consistent with ensemble experimental observations indicating that this mutant presents the primer extension properties of the wild-type polymerase (15).
Pause kinetics of the wild-type polymerase. (A) Representative position versus time traces of the wild-type polymerase during strand displacement. Identified pause events are shown in red, gray lines show GC rich positions of the hairpin. Insert shows the wild-type velocity distribution during strand displacement conditions (Fig. S5A). (B–D) Red dots represent primer extension and blue dots strand displacement conditions. (B) The pause length frequency distributions (s-2) can be fitted with a single exponential (f ∼ 5 pN). Primer extension, red line (R2 = 0.89, N = 22), strand displacement, blue line (R2 = 0.87, N = 38). (C) The pause frequency (s-1) decreases exponentially with tension (solid lines represent fit to data with equation 1) and (D) increases linearly with the GC content of the DNA (f ∼ 5 pN). Primer extension, red line (R2 = 0.81, N = 38), strand displacement, blue line (R2 = 0.78, N = 22). (E) Proposed kinetic model that explains the behavior of short-lived pauses (Short Pause 1). See Table 1 for rate values and the associated tension dependence, da1. For (B) and (D) error bars calculation is described in SI Text. For (C) error bars represent standard deviation.
Pause kinetics of the sdd mutant polymerase. (A) The pause length frequency distribution (s-2) during primer extension (f ∼ 6.5 pN, red dots) is compatible with a single exponential (solid red line R2 = 0.97, N = 25). For comparison the wild-type distribution from Fig. 3B is shown as a gray line. During strand displacement (f ∼ 6.5 pN, green dots) a double exponential is required to fit the data (green solid line, R2 = 0.93, N = 41). For (B–D), green dots represent short pauses during strand displacement; red dots, short pauses primer extension; black dots, long pauses strand displacement. (B) The average length (seconds) of short pauses is tension independent while the average length of long pauses decreases exponentially with tension (black line fit to the data with Eq. 1). (C) For both pause types the pause frequency (s-1) decreases exponentially with tension (solid lines represent the fit to the data with Eq. 1) and (D) increases linearly with the GC content of the DNA (solid lines represent linear fits, R2 = 0.98, 0.99, 0.92). (E) Proposed kinetic model that explains the mutant pause behavior during strand displacement conditions. See Table 1 for rates and conformational changes values. For (A) and (D) error bars calculation is described in SI Text. For (B) and (C) error bars represent standard deviation.
Rates and conformational changes during primer extension (p.e.) and strand displacement (s.d.) conditions
However, during strand displacement, the mutant’s pause behavior presents different properties (Fig. 2D). In this case, the pause length frequency distribution is better fit to a double exponential, indicating that at least two types of pause events, short- (typically less than 4 s) and long-lived (typically greater than 4 s) coexist (Fig. 4A). Analysis of the short-lived pause events revealed that they present almost identical characteristics to the pauses observed during primer extension and to the ones described for the wild-type polymerase (Fig. 4 and Table 1), suggesting that they correspond to the short-lived pause state characteristic of template replication, and they are not related with unwinding. The long-lived pauses instead present quite different properties from the shorter pauses: They only occur during strand displacement and not during primer extension (Fig. 4A). Their duration is approximately 50 times longer than the short pauses (at low tensions) and more importantly present a strong tension dependence, decreasing exponentially as the destabilization tension at the fork increases (Fig. 4B). In addition, their frequency decreases exponentially with tension (Fig. 4C) and increases linearly with the strength of the GC content of the fork (Fig. 4D). Taken together, our data suggest that long pauses correspond to an inactive polymerase-DNA complex state induced during the unwinding of the more stable dsDNA fork positions (GC clusters) and that the external mechanical destabilization of the fork promotes the exit from this long pause state. These observations explain the inability of the sdd mutant to replicate in vitro the Phi29 dsDNA genome beyond the initial GC clusters (15).
The pause behavior of the wild-type polymerase (both during strand displacement and primer extension) and the mutant polymerase during primer extension conditions can be explained with a simple kinetic model in which the short-lived pauses correspond to a single off-pathway inactive state branching off the main active state of the protein during replication of the GC rich positions of the template (Short Pause 1, Fig. 3E). Instead, the simplest kinetic model that explains the mutant’s pause behavior during unwinding requires the short- and long-lived pause states to branch off independently from the main active state of the polymerase (Short Pause 1 and Long Pause 2, respectively, Fig. 4E). Alternative models were also considered (SI Text). In these models, the tension dependent entry and exit rates to and from the pause states, k(f), change with tension following (16): [1]where f is the applied tension, k(0) the rate at zero tension (Table 1), kB is the Boltzmann constant, T the absolute temperature, and d the distance to the transition state along the mechanical tension coordinate (Table 1). We note that since the tension dependent entry rates to the short-(ka1(f)) and long- (ka2(f)) lived pause states are also sequence dependent (Figs. 3D and 4D), we calculated the particular entrance rates to both pause states at the GC positions, ka1GC(f) and ka2GC(f) (Table 1 and SI Text). The pause exit rate from the short-lived pauses, k1a, can be directly calculated as the inverse of the average tension and sequence independent pause length (Table 1).
Inclusion of Pause Kinetics in a Model to Quantify the Unwinding Mechanism Revealed the Same Active Mechanism for the Wild-Type and Mutant Polymerases.
Our data revealed that pauses strongly modulate the average unwinding rate of the mutant polymerase. Similarly, pauses have been shown also to be part of the unwinding mechanism of several specialized nucleic acid (NA) motors (17⇓⇓–20). Therefore, the kinetics of pause events must be considered for the correct interpretation of tension and sequence dependencies of the unwinding rates. We have extended the physical framework (11, 21) where the unwinding activeness of the NA unwinding motor depends on the interaction energy, Gint, between the protein and the NA fork (Fig. 5A), to include the effect of pauses on the unwinding reaction of the Phi29 DNA polymerase.
DNA unwinding model. (A) Diagram showing notation used for modeling the polymerase movement during strand displacement conditions. M (red circle) defines the range of interaction between the polymerase and the junction; m is the number of ssDNA template nucleotides between the polymerase and the junction; l is the total number of replicated nucleotides and L is the full length of the hairpin. (B) Proposed DNA unwinding mechanism. Figures show the schematic representation of the Phi29 DNA polymerase with the DNA substrate during strand displacement (6). The protein is diagrammed in two levels. The upper level contains the exo (green), the TPR2 (blue) and thumb (orange) domains. The rest of the protein is shown as a gray circle. (1) and (2) Template bending at the TPR2-exo tunnel and steric exclusion of the complementary strand may generate mechanical stress at the fork junction promoting the unwinding of the first 2 bp (blue) of the fork. dNTP binding and hydrolysis fueled the polymerase forward movement. (3) For the sdd mutant, mechanical stress at the junction during unwinding of the more stable fork positions (where m = 0) could lead to destabilization of the template-tunnel interactions, favoring the entrance to the Long Pause 2 state. Fork destabilization by external tension (f) favors DNA unwinding, preventing the entrance to and promoting the exit from the inactive Long Pause 2 state. In (2) and (3) the initial position of the fork is shown in light gray.
In terms of the residence time of the polymerase in each template position, the generalization of the model is direct, the residence time with pauses T(f) is given by , with Ta(f) the residence time without pauses (21) (SI Text), and Ti(f) the residence time contribution due to pauses of type i determined by the pause kinetics. According to our kinetic models (Figs. 3E and 4E), the residence time contribution due to pause kinetics is
, where the additional residence times, T1(f) and T2(f), are determined by the entry and exit rates (ka1(f), ka2(f) and k1a, k2a(f)) to the short and long-lived pause states as
and
(see SI Text for sequence dependencies).
We used our model to quantify the DNA unwinding activity of both polymerases when pauses are included and excluded from the average strand displacement rate. We considered that the polymerase can only move forward when there is ssDNA template available in front of it (m≥1; Fig. 5A). Importantly, in the case of the DNA polymerase, since replication is known to occur in discrete steps of one nucleotide at a time, δ = 1 nt (22) and backtracking is not expected (SI Text) and not observed with our current resolution, only two variables determine the fits; the interaction energy between polymerase and the fork, ΔGint, and the range of this interaction, M (Fig. 5A and SI Text). We found that the best fit of the model to the wild-type polymerase tension dependent strand displacement rate (with and without pauses) was obtained with an interaction potential increase per base of ΔGint = 2kBT and M = 2 (Fig. 2A and Fig. S6). In addition, the same set of parameters also correctly predicted the sequence dependent residence time (and velocity) of the polymerase during unwinding of the hairpin supporting the validity of the model (Fig. 2B and Fig. S3D). Therefore, the tension and sequence dependent unwinding rate of the wild-type Phi29 DNA polymerase (with and without pauses) support an active DNA unwinding model where the polymerase destabilizes at least two bases at the junction facilitating in this way the fork opening.
The strong tension and sequence dependency of the strand displacement rate of the sdd mutant may suggest this protein uses a different or less active unwinding mechanism than the wild-type. However, when pause kinetics are consistently considered, the values ΔGint = 2kBT and M = 2 nicely predict the average strand displacement rate of the mutant (Fig. 2C and Fig. S6). Therefore, our data strongly support that both enzymes use the same active DNA unwinding mechanism, but the mutant’s mechanism fails during unwinding of the more stable fork positions, which promotes entering into the long-lived inactive pause state. The identical values and tension dependencies of the average unwinding rates without pauses found for both polymerases further support this hypothesis (Fig. 2 A and C).
Discussion
Our data show that the rate-limiting step of the polymerization cycle of the Phi29 DNA polymerase is not affected by tension (below 14 pN) or sequence. However, replication along GC rich template sequences promotes the entrance to a short-lived pause state (Short Pause 1), which is more frequent (approximately two- to fourfold) and presents a stronger tension dependence (approximately twofold) during primer extension than during strand displacement conditions (ka1GC and da1GC; Table 1). These observations suggest that translocation along the DNA template could be interrupted by the particular arrangements of the GC rich positions of the template (i.e., GG base stacking), which would occur with less probability and involve fewer bases during strand displacement due to the limited number of bases opened in front of the polymerase. Mechanical extension of the template could prevent base stacking, avoiding in this way the entrance to the short-lived pause state. In fact, other factors known to affect the stability of DNA, as betaine and increasing temperatures, also decrease the pause frequency at GC rich sequences during primer extension conditions for Phi29 and other DNA polymerases, supporting this hypothesis (19, 23, 24).
In contrast, the unwinding of the fork limits the forward translocation of both the wild-type and the sdd mutant polymerases, although in different ways: The more stable fork positions slow down the wild-type but pause the advance of the mutant polymerase. Despite these differences, inclusion of pause kinetics in a model to quantify the polymerase unwinding mechanism revealed that both polymerases actively destabilize the two nearest basepair of the fork (M = 2) with a destabilization energy of ΔGint = 2kBT per basepair (Fig. 2). The extension of the Betterton and Jülicher (11) model we present in this work points out the importance of considering pause kinetics and can be applied to the quantification of the unwinding mechanism of other NA unwinding motors in which pauses are known to be relevant or are expected to occur during their operation (17⇓⇓–20).
To rationalize the above observations, we propose a model for the mechanism of coupling of DNA unwinding and replication. The active destabilization of the fork promoted by the Phi29 DNA polymerase is likely due to the particular polymerase-DNA interactions at the fork junction since, in contrast with other polymerases, the fingers domain seems not to be involved in the DNA unwinding reaction (25⇓–27), and in addition, there is no evidence for the interaction of the displaced strand with the polymerase during unwinding or for different translocation modes between the primer extension and strand displacement activities (6, 22). Structural studies of replicative polymerases during primer extension, including the Phi29 DNA polymerase, showed that during replication the template strand is sharply bent (approximately 90°) inside the polymerase (22) (Fig. 5B). In the case of Phi29 DNA polymerase, this sharp template bending would be maintained during strand displacement activity by the stable thread of the template strand (two nucleotides) through the narrow tunnel formed by the TPR2 and the exo domains (22). At the same time, the small dimensions of the tunnel (10 Å) would promote the steric exclusion of the complementary strand, probably forcing its bending near the tunnel entrance. Due to the elastic properties of ssDNA (13) bending of the two DNA strands may incur in generation of mechanical stress at the fork junction, promoting its destabilization during strand displacement conditions (Fig. 5B).
Our data show that the active unwinding mechanism of the sdd mutant fails (Long Pause 2) at the more stable GC base pairs of the hairpin, which will be closed with higher probability (28). In fact, we found that the tension dependence of the entry rate to the Long Pause 2 state at the GC rich positions, da2GC = 0.5 nm (Table 1), can be explained when considering that entrance occurs when the next basepair is closed (m = 0) (SI Text). The value of da2GC is consistent with the interpretation that tension would prevent entry to long pauses by promoting the opening of the next basepair of the fork, since the average distance between ssDNA nucleotides under our experimental conditions is 0.25 nm/nt (Fig. S2C), 0.5 nm/0.25 nm/nt = 2 nt = 1 basepair. Mutations D12AD66A at the exo domain of the sdd mutant used in this work are thought to disturb the specific arrangement of the TPR2 and the exo domains (29), which in turn would affect the stability of the template tunnel. According to our model, this defect would impede the mutant to hold the mechanical stress at a closed fork junction, which in turn would induce loosening of the correct interactions with the template and would favor the entrance to the long-lived inactive pause state (Fig. 5B). The fork pressure will prevent the functional binding of the template back to the polymerization domain, stopping the polymerase advance. According to our hypothesis the mutant polymerase would exit from the Long Pause 2 inactive state only when the fork is opened (m > 0) and the polymerase has the chance to recover the control on the template strand. Indeed, we checked that the tension dependence of the exit rate from the Long Pause 2 state, d2a = -0.7 nm (Table 1), can be explained when considering m > 0; in other words, when the fork ahead of the polymerase is opened (SI Text). The value of d2a indicates that the unwinding of the first 1 to 2 bp of the fork (0.7 nm/0.25 nm/nt) is required to rescue the polymerase from the Long Pause 2 inactive state (Fig. 5B; the negative sign of d2a simply reflects that k2a increases with growing tension; see Eq. 1).
The Phi29 DNA polymerase presents a more efficient strand displacement activity than other replicative DNA polymerases like the bacteriophage T4, T7 or the prokaryotic Pol III (30, 31). Due to the open template-tunnel characteristic of these polymerases (32, 33), it is possible that they cannot keep the sharp bending of the template against a closed fork junction impeding further polymerization (as observed in our experiments with the sdd mutant). Therefore, to replicate through dsDNA, these polymerases should work in coordination with their respective helicases, which by steric exclusion of the leading strand (34, 35) would increase the probability of the fork to be opened. The steric exclusion of the leading strand by the helicase plus its sharp bending, induced by the polymerase, may account for the efficient fork destabilization during DNA replication. In the case of the Phi29 DNA polymerase the TPR2-exo tunnel, which closely wraps around the template, allows the protein to maintain both the bending of the template and the steric exclusion of the complementary strand against a closed fork promoting its active destabilization. In fact, the fork junction destabilization energy of the Phi29 DNA polymerase (ΔGint = 2 kBT) is apparently higher than the one reported for hexameric DNA helicases, ΔGint = 0.05–1.6 kBT (20, 21, 36, 37) and polymerases, ΔGint = 1.4 kBT (19), studied in isolation. These features plus the energy provided by the dNTPs binding and hydrolysis during replication would allow the Phi29 DNA polymerase to work as a hybrid polymerase-helicase and efficiently couple DNA replication and unwinding activities within the same polypeptide.
Methods
Optical Tweezers Experiments.
A custom-made 515 bp DNA fragment containing four GCC bp clusters of increasing lengths (3, 6, 9, and 12) separated by 97 bp of a low-GC content sequence was purchased from Genscript Corp. The final substrate presents a unique 3′ end loading site for the DNA polymerase and a 556 bp long hairpin (Fig. S1). We used a dual-beam optical tweezers instrument (38) to manipulate individual DNA hairpins. Generation and purification of the wild-type and sdd polymerases was described previously (12, 39). Phi29 DNA polymerase (2, 20, 200 nM) was flowed into the chamber diluted in the reaction buffer [50 mM Tris-HCl (pH 7.5), 20 mM ammonium sulphate, 10 mM MgCl2, 7 mM 2-mercaptoethanol, 4% glycerol (wt/vol), 0.025% Tween20 (wt/vol), 0.1 μg/mL BSA] containing 50 μm dNTPs. Data presented in the main text were taken at a polymerase concentration of 20 nM. Data were collected at 60 Hz at 23 ± 2 °C in the constant force feedback mode, in which the distance between the beads was adjusted to maintain a constant tension in the DNA.
Data Analysis.
The number of nucleotides incorporated as a function of time during strand displacement was obtained by dividing the observed distance change (Δx1) by the change in extension at a given tension accompanying the generation of one new basepair and one single-stranded nucleotide (Fig. S2). The number of replicated nucleotides during primer extension was calculated as described earlier (14). The average rates (with and without pauses) were determined by a line fit to the traces showing the number of replicated nucleotides versus time. The final rate at each tension was obtained by averaging over all of the traces at similar tension values (± 1.6 pN).
The residence time (defined as the inverse of the instantaneous replication rate) was obtained from a linear fit of the number of replicated nucleotides (in each replication condition) over a sliding time window of 0.7 s (50 data points) for all velocities at different tensions. Individual residence time traces within the same tension range were aligned to calculate the average residence time at each DNA position at different tensions (Fig. S3). A moving window of 10 bp was used to obtain the velocity as a function of the GC percentage (Fig. S3).
Pause events were identified following the method described previously (14) with a resolution of 0.4–0.8 s (Figs. 2D and 3A). The accurate alignment of independent extension versus time records allowed us to compute the average pause frequency and duration at each sequence position within a window of approximately 20 bp.
Acknowledgments
We thank F. Moreno and M. Manosas for a careful reading of the manuscript. This work was supported by Grants BFU2011-29038-CO2-1 and CSD2007-00010 (to J.L.C.), BFU2010-15703/BMC (to J.M.V.), BFU2008-00215/BMC and CSD2007-00015 (to M.S.), RYC2007-01765 (to J.R. A-G.), S2009MAT-1507 (to J.L.C., J.M.V., and M.S.), and FIS2010-17440 and GR35/10-A-920GR35/10-A-911 (to F.J.C.).
Footnotes
↵1J.A.M. and F.J.C., contributed equally to this work
- ↵2To whom correspondence may be addressed. E-mail: msalas{at}cbm.uam.es or borja.ibarra{at}imdea.org.
Author contributions: B.I. designed research; J.A.M. and B.I. performed research; J.M.L., J.R.A.-G., J.M.V., J.L.C., and M.S. contributed new reagents/analytic tools; J.A.M., F.J.C., and B.I. analyzed data; and F.J.C., M.S., and B.I. wrote the paper.
The authors declare no conflict of interest.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1204759109/-/DCSupplemental.
References
- ↵
- ↵
- ↵
- Blanco L,
- et al.
- ↵
- ↵
- Rodriguez I,
- et al.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Esteban JA,
- Soengas MS,
- Salas M,
- Blanco L
- ↵
- ↵
- Cheng W,
- Dumont S,
- Tinoco I,
- Bustamante C
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Mytelka DS,
- Chamberlin MJ
- ↵
- Schwartz JJ,
- Quake SR
- ↵
- ↵
- ↵
- Singh K,
- Srivastava A,
- Patel SS,
- Modak MJ
- ↵
- ↵
- ↵
- Canceill D,
- Viguera E,
- Ehrlich SD
- ↵
- Yuan Q,
- McHenry CS
- ↵
- ↵
- ↵
- ↵
- Ahnert P,
- Patel SS
- ↵
- Donmez I,
- Rajagopal V,
- Jeong Y-J,
- Patel SS
- ↵
- Lionnet T,
- Spiering MM,
- Benkovic SJ,
- Bensimon D,
- Croquette V
- ↵
- ↵
Citation Manager Formats
Article Classifications
- Biological Sciences
- Biophysics and Computational Biology