|Home | About | Journals | Submit | Contact Us | Français|
Coupling nucleic acid processing enzymes to nanoscale pores allows controlled movement of individual DNA or RNA strands that is reported as an ionic current time series. Hundreds of individual enzyme complexes can be examined in single-file order at high bandwidth and spatial resolution. The bacteriophage phi29 DNA polymerase (phi29 DNAP) is an attractive candidate for this technology, due to its remarkable processivity and high affinity for DNA substrates. Here we show that phi29 DNAP-DNA complexes are stable when captured in an electric field across the α-hemolysin nanopore. DNA substrates were activated for replication at the nanopore orifice by exploiting the 3′-5′ exonuclease activity of wild-type phi29 DNAP to excise a 3′-H terminal residue, yielding a primer strand 3′-OH. In the presence of deoxynucleoside triphosphates, DNA synthesis was initiated, allowing real time detection of numerous sequential nucleotide additions that was limited only by DNA template length. Translocation of phi29 DNAP along DNA substrates was observed in real time at Angstrom scale precision as the template strand was drawn through the nanopore lumen during replication.
Single molecule techniques are now used routinely to study nucleic acids in basic science 1–3 and technology 4,5. Methods using nanoscale pores (nanopores) are advantageous because they can report the length, structure and composition of unmodified DNA or RNA molecules that are captured in single file order 6–9. Data are typically reported as a time series of ionic current as each DNA strand is driven by an applied electric field across a single pore controlled by a voltage-clamped amplifier. Hundreds to thousands of molecules can be examined at high bandwidth and spatial resolution.
Recently, the properties of DNA or RNA molecules bound to nucleic acid processing enzymes have been analyzed at a nanopore orifice. The complexes studied include those of single-stranded DNA with Escherichia coli Exonuclease I 10, RNA with the bacteriophage phi8 ATPase, 11, and primer/template DNA substrates bound to the 3′-5′-exonuclease deficient versions of two A-family DNA polymerases, the Klenow fragment of E. coli DNA polymerase (KF(exo-)) and bacteriophage T7 DNA polymerase (T7DNAP(exo-)) 12–16. We have demonstrated that T7DNAP(exo-) could replicate and advance a DNA template held in the α-hemolysin (α-HL) nanopore against an 80 mV applied potential 17. However, due to the low stability of the T7DNAP(exo-)-DNA complex under load, diminished signal to noise ratio at 80 mV potential, and the high turnover rate of the polymerase, it was difficult to detect ionic current steps that reported more than three sequential nucleotide additions during replication.
To overcome these limitations, we are examining other DNA-modifying enzymes whose structural and functional properties might facilitate processive catalysis when positioned at a nanopore orifice. An attractive candidate is the bacteriophage phi29 DNA polymerase (phi29 DNAP). This DNA-dependent DNA replicase from the B family of DNA polymerases contains both 5′-3′ polymerase and 3′-5′ exonuclease functions within a single ~ 66.5 kD protein chain (reviewed in 18,19). Following an initial protein-primed stage that ensures the integrity of the ends of the bacteriophage phi29 linear chromosome 20, phi29 DNAP transitions to a DNA-primed stage and replicates the entire 19.2 kilobase bacteriophage genome without the need for accessory proteins such as sliding clamps or helicases. This highly processive polymerase can catalyze the replication of at least 70 kilobases of DNA in vitro following a single binding event to a DNA-primed substrate 21.
Crystal structures of phi29 DNAP 22,23 revealed the structural basis of this remarkable processivity. The polymerase domain of phi29 DNAP shares the conserved architecture of palm, fingers and thumb sub-domains that resembles a partially open right hand. In addition, a 32 amino acid beta-hairpin insert that is unique to protein-primed DNA polymerases, together with the palm and thumb sub-domains, encircles the primer-template DNA, suggesting that this structure enhances processivity in a manner similar to that achieved by sliding clamp proteins 24. This same beta-hairpin also forms part of a tunnel that surrounds the downstream template DNA. These features indicate that the beta hairpin insert contributes to both the strong DNA binding and processivity of phi29 DNAP. Consistent with this prediction, deletion of the beta-hairpin results in a mutant phi29 DNAP that displays distributive DNA synthesis activity rather than the processive activity of the wild-type enzyme and a markedly diminished binding affinity for primer-template duplex DNA 25.
Experiments using optical tweezers have shown that phi29 DNAP can advance several hundred nucleotides along a template against applied loads of up to ~ 37 pN 26, suggesting that this enzyme could replicate a DNA template held atop the nanopore. Here we show that phi29 DNAP-DNA complexes are three-to-four orders of magnitude more stable than KF(exo-)-DNA complexes when captured in an electric field across the α-HL nanopore. DNA substrates in captured complexes were activated for replication by exploiting the 3′-5′ exonuclease activity of wild-type phi29 DNAP to excise a 3′-H terminal residue, yielding a primer strand 3′-OH. In the presence of deoxynucleoside triphosphates (dNTPs), DNA synthesis was initiated, allowing real time detection of numerous sequential nucleotide additions that was limited only by the length of the DNA template.
The D355A, E357A exonuclease-deficient KF (100,000 U ml−1; specific activity 20,000 U mg−1) was from New England Biolabs. Wild-type phi29 DNAP (833,000 U ml−1; specific activity 83,000 U mg−1) was from Enzymatics. DNA oligonucleotides were synthesized at Stanford University Protein and Nucleic Acid Facility and purified by denaturing PAGE.
A 67 mer, 14 base-pair hairpin DNA substrate labeled with 6-FAM at its 5′ end was self-annealed by incubation at 90°C for four minutes, followed by rapid cooling in ice water. Reactions were conducted with 1 µM annealed hairpin and 0.75 µM phi 29 DNAP(exo+) in 10 mM K-Hepes, pH 8.0, 0.3 M KCl, 1 mM EDTA, 1 mM DTT with MgCl2 added to 10 mM when indicated, and dNTPs added at the concentrations indicated. Reactions were incubated at room temperature for the indicated times and were terminated by the addition of buffer-saturated phenol. Following extraction and ethanol precipitation, reaction products were dissolved in 7 M urea, 0.1X TBE and resolved by denaturing electrophoresis on gels containing 18% acrylamide:bisacrylamide (19:1), 7 M urea, 1X TBE. Extension products were visualized on a UVP Gel Documentation system using a Sybr Gold filter. Band intensities were quantified using ImageJ software (NIH).
The nanopore device and insertion of a single α-HL nanopore into a lipid bilayer have been described 7,12,27. Ionic current flux through the α-HL nanopore was measured using an integrating patch clamp amplifier (Axopatch 200B, Molecular Devices) in voltage clamp mode. Data were sampled using an analog-to-digital converter (Digidata 1440A, Molecular Devices) at 100 kHz in whole-cell configuration and filtered at 5 kHz using a low-pass Bessel filter. For voltage clamped experiments, current blockades were measured at the voltages specified in each figure (trans-positive). Experiments were conducted at 23 ± 0.2 °C in buffer containing 10 mM K-Hepes pH 8.0, 1 mM EDTA, 1 mM DTT, 0.3 M or 0.6 M KCl as indicated, and 10 mM MgCl2 where indicated. DNA hairpin substrates were annealed prior to each experiment by heating at 95°C for 3 minutes and rapidly cooling in an ice bath to prevent intermolecular hybridization.
Active voltage control of DNAP-DNA complexes atop the nanopore was achieved using finite state machine (FSM) logic, which was programmed with LabVIEW software (Version 8, National Instruments) and implemented on a FPGA system (PCI-7831R, National Instruments), as described previously 12,16. Details of the FSM logic applied in the experiments shown in Figures 2 and S2 are given in the figure legends.
Dwell time and amplitudes for KF(exo-)-DNA binary complexes were quantified using software developed in our laboratory that detects and quantifies the dwell time and amplitude of EBS and terminal current steps of capture events 14. Current blockades for phi29 DNAP complexes were quantified using Clampfit 10.2 software (Axon Instruments). Dominant IEBS values for phi29 binary and ternary complexes were obtained by using Clampfit software to determine the peaks of all-points amplitude histograms measured for 1 to 5 second windows in the initial segment of capture events.
To perform nanopore experiments, a single α-HL nanopore is inserted in a lipid bilayer separating two chambers (termed cis and trans) containing buffer solution, and a patch-clamp amplifier applies voltage and measures ionic current (Fig. 1a). To examine binary complexes formed between phi29 DNAP and DNA, we used a 14 base-pair DNA hairpin substrate (Fig. 1b). As demonstrated previously 12,15, when a KF-DNA binary complex formed with this substrate is captured in the α-HL pore, the resulting ionic current signature is characterized by an initial enzyme bound state (EBS). This occurs when KF resides atop the pore 12, holding the double-strand/single-strand junction of the DNA substrate within the confines of the polymerase active site (Fig. 1c, ii). In this KF-bound state, the DNA template strand is suspended through the nanopore lumen, which is wide enough to accommodate single-stranded but not duplex DNA. The amplitude of this state (IEBS) can be selectively augmented by an insert of abasic (1′,2′-H) residues within the template strand positioned so that it resides in the nanopore lumen when the polymerase-DNA complex is perched atop the pore 14,17, such as the 5 abasic residues between template positions +12 to +16 in the DNA hairpin shown in Figure 1b. For KF-DNA binary complexes, the EBS typically lasts a few milliseconds at 180 mV applied potential (Fig. 1c, ii). It is followed by a shorter lower amplitude state (Fig 1c, iii), which occurs when the force pulling on the template strand causes dissociation of KF from the DNA, and the duplex DNA drops into the nanopore vestibule. When this occurs the abasic block that was positioned in the pore lumen during the EBS is displaced to the trans side of the pore, where it has negligible effect on the amplitude of this terminal current step (~ 20 pA at 180 mV). Unzipping of the DNA hairpin within the vestibule followed by electrophoresis of the strand to the trans compartment restores the open channel current (Fig. 1c, iv).
Binary complexes between phi29 DNAP and DNA substrates can be formed in the absence of the divalent cations required for both 5′-3′ polymerase and 3′-5′ exonuclease activity 28. When phi29 DNAP-DNA binary complexes were formed with the hairpin substrate in Figure 1b and captured in the α-HL pore at 180 mV (Fig. 1d, ii), the ~ 35 pA IEBS typically lasted tens of seconds (median = 17.6 s, IQR = 25.6, n = 62). This is approximately 10,000 times longer than KF-DNA binary complexes under the same conditions (median = 1.9 ms, IQR = 2.4 ms, n = 199). In contrast to capture events for KF-DNA complexes, these phi29 DNAP-DNA events did not end in a single terminal step, but instead ended in a series of discrete ionic current steps (Fig 1d, iii) that we termed a “terminal cascade”. The 3′-5′ exonuclease of wild type phi29 DNAP is inhibited under the conditions of the experiment (1 mM EDTA, absent added Mg2+) 29 and thus these current steps are not due to digestion of the primer strand. Therefore we reasoned that the DNA duplex may be unzipping while bound within the confines of the enzyme (Fig.1d, iii). In this scenario, as the template threads out of the complex under tension, the abasic block is drawn out of the lumen in single nucleotide increments that give rise to the sequence of discrete amplitude steps in the terminal cascade (Fig 1d, iii).
This model suggests that the interaction between phi29 DNA and the DNA is strong enough that the DNA secondary structure unzips due to the force pulling on the template strand before the bond between phi29 DNAP and DNA can be broken. It furthermore predicts that reducing the applied voltage during the terminal cascade could allow the DNA duplex to re-anneal within the confines of the enzyme and thus reset the phi29 DNAP-DNA complex to its original position on the DNA template strand, indicated by a return to the ~ 35 pA state. To test this prediction, we compared the ability of complexes captured in the presence or absence of Mg2+ to recover their original EBS amplitude at 180 mV following a controlled voltage drop. A prerequisite for this comparison is a means to ensure that DNA molecules captured in the presence of Mg2+ are intact, so that the nanopore assay compares their fate only after capture. Thus exonucleolytic cleavage of the primer strand in the bulk phase must be miminized during the course of the experiment. We tested whether a 3′-H terminus on the DNA substrate inhibited the rate of 3′-5′ exonucleolytic cleavage by phi29 DNAP, in a gel assay comparing degradation of two 67 mer 5′-6-FAM labeled hairpin substrates (Fig. 2a) bearing either a dCMP (lanes 1–6) or ddCMP (lanes 7–12) terminus. Consistent with the requirement for divalent cations for phi29 DNAP 3′-5′ exonuclease function 29, no cleavage of either DNA substrate was observed after 45 minutes incubation in nanopore buffer containing 1 mM EDTA absent added Mg2+ (Fig. 2a, lanes 1 and 7). With 10 mM Mg2+ present, the extent of DNA digestion for the 3′-H substrate was discernably less than for the 3′-OH substrate. After 10 minutes, while only 24.5% full-length DNA molecules remained for the dCMP-terminated hairpin, 90.5% of the ddCMP-terminated substrate remained intact (Fig. 2a, lanes 4 and 10). After 45 minutes, 4% of the dCMP-terminated substrate and 45% of the ddCMP-terminated substrate remained intact (Fig. 2a, lanes 5 and 11). The protection against excision afforded by a 3′-H terminus is further evidenced by the extent of primer extension in the presence of all four dNTPs. For the 3′-H terminated substrate, the onset of DNA synthesis requires that the ddCMP residue first be excised. Thus while with the 3′-OH terminated hairpin > 80% of the molecules were extended to the full-length 102 mer product in 45 minutes (Fig. 2a, lane 6), with the 3′-H terminated hairpin, 79.8% of the DNA substrate remained intact, with only 20.1% full-length extension product (Fig. 2a, lane 12). Thus 3′-H terminated DNA substrates afforded a window following the addition of Mg2+ during which phi29 DNAP-DNA complexes could be captured with the DNA substrate intact. We therefore used the ddCMP terminated hairpin shown in Figure 1b in a nanopore experiment designed to assess the potential for hairpin refolding following initiation of the phi29 DNAP terminal cascade.
In this experiment, upon capture of a phi29 DNAP-DNA complex at 180 mV, a finite state machine (FSM, see Methods) monitored ionic current in real time until the downward current steps of the terminal cascade were detected (Fig 2b, ii). When the ionic current dropped below 31 pA for at least 0.5 ms (red arrow in Figure 2b), the FSM reduced the applied potential to 70 mV (Fig 2b, iii). After two seconds at 70 mV, the applied potential was restored to 180 mV and the amplitude of the phi29 DNAP-DNA complex was remeasured. In the absence of Mg2+, the IEBS level was reproducibly reset to the original 35 pA level in each of 11 molecules tested. This EBS amplitude is indicative of the initial state in which phi29 DNAP is bound to the base-paired duplex with the n = 0 template residue positioned in the polymerase active site (Fig 2b, iv), and is consistent with re-annealing of the DNA template with an intact primer strand. Importantly, the dominant amplitude during the 70 mV intervals was ~ 10.2 pA, with occasional deflections to ~ 8.5 pA, measurably above the 6.8 pA value determined for unbound DNA at 70 mV in a control experiment (Figure S2). This indicates that the phi29 DNAP complex remained atop the nanopore orifice without dissociating throughout the lower voltage interval, consistent with a model in which hairpin unzipping at 180 mV and refolding at 70 mV occurs within the confines of the enzyme complex atop the pore.
When the refolding experiment was performed in the presence of 10 mM Mg2+, 16 complexes out of 24 captured in the first 12.5 minutes after the addition of Mg2+ had the ~ 35 pA IEBS level indicating they were formed with intact DNA substrate molecules (Fig. 2c, i). This 35 pA state was maintained for several seconds (median = 10.2 s, IQR = 12.7 s, n = 16), before ending with a drop in amplitude (Fig. 2c, ii). The features of the steps that occurred following the 35 pA state differed from those that characterized the terminal cascade in the absence of Mg2+ (compare Fig. 2b, ii to 2c, ii). For these complexes, when the voltage was reduced to 70 mV for two seconds and then restored to 180 mV, the 35 pA IEBS level did not reset for any of the complexes tested (Fig. 2c). This is in contrast to the phi29 DNAP-DNA complexes captured in the absence of Mg2+ and it indicates that the DNA substrates, which had been captured intact, were modified by exonucleolytic cleavage while they were held atop the pore.
Our strategy for detecting DNA synthesis catalyzed by polymerase-DNA complexes held atop the nanopore employs monitoring changes in ionic current as a block of abasic residues in the template strand is drawn into and through the nanopore lumen in single nucleotide increments when the polymerase advances along the template 14,17. This approach permits the recognition of sequential Angstrom-scale movements driven by the enzyme.
As a prelude to DNA replication experiments with phi29 DNAP, we established a reference map that related IEBS to the position of a 5 abasic block within the template strand of DNA hairpin substrates (Fig. 3). To construct this map, phi29 DNAP was bound to each of a series of substrates that contained a block of 5 consecutive abasic residues, sequentially displaced by one nucleotide (Fig. 3a). We measured the IEBS in buffer containing 0.3M KCl for captured complexes under two conditions: i) 1 mM EDTA with no added Mg2+, which permits formation of binary complexes without supporting nucleotide excision or addition (Fig. 3b, lane 1); and ii) 10 mM Mg2+, 400 µM ddCTP, and 100 µM dGTP. These latter conditions maintained the intact status of 98.2 and 96% of 3′-H terminated hairpin molecules in the bulk phase for 10 and 45 minutes, respectively (Fig 3b, lanes 6 and 7). Protection was afforded by ddCTP, which permitted the polymerase function of phi29 DNAP to restore the ddCMP terminus of molecules if it was excised by the exonuclease function (Fig. 3b, lanes 3 and 4). Protection was enhanced by the presence of dGTP, which is complementary to the template residue at n = 0 and can form a phi29 DNAP-DNA-dGTP ternary complex in the presence of the 3′-H terminated DNA substrate 22 that can increase the proportion of time the primer terminus resides in the polymerase domain rather than in the exonuclease domain (Fig. 3b, lanes 6 and 7; Fig. S3). The complex formed in the presence of Mg2+, ddCTP, and dGTP is therefore operationally defined as a ternary complex in this study.
The IEBS maps for phi29 DNAP binary complexes (blue dots) and ternary complexes (red dots) are shown in Figure 3c. Both maps were similar to a map determined for KF(exo-)-DNA-dNTP ternary complexes at 80 mV using a six abasic template insert 17. In 0.3 M KCl at 180 mV, IEBS ranged from 22.3 pA for the ternary complex formed with the 5ab(6,10) substrate (abasic block spanning template positions +6 to +10 measured from n = 0 in the polymerase catalytic site), to 35.4 pA for the binary complexes formed with the 5ab(11,15) and 5ab(9,13) substrates (abasic blocks spanning template positions +11 to +15, and +9 to +13, respectively). This gives a dynamic amplitude range of at least 13 pA for the detection of enzyme movements during polymerization or exonucleolytic reactions.
At all positions within the map, IEBS for the binary and ternary complexes were offset from one another. The direction and the scale of the offset depended in part on the position along the map. For example, at position (i) (Fig. 3c), the change from a binary complex to a ternary complex caused an IEBS increase from 31.5 pA to 34.5 pA. By comparison, at position (ii) (Fig. 3c) the binary to ternary change resulted in a relatively small current increase from 34.4 to 35.2 pA, and at position (iii) (Fig 3c) the binary to ternary transition caused a large IEBS current decrease from 31.5 pA to 25.5 pA. Interestingly, the direction and magnitude of an ionic current flicker within the binary state often predicted the dominant amplitude observed for the ternary complex formed with the same substrate (Fig. 3d).
The results of the mapping experiments permit a prediction based upon the model proposed for the molecular events that give rise to the terminal cascade (Figs. 1 and and2):2): the sequence of current steps in the terminal cascade of binary complex capture events should vary in a manner that is dependent on the initial position of the abasic block in the complex. This was found to be the case. For example, when the duplex segment of the 5ab(6,10) substrate was unzipped during the terminal cascade, the abasic block was drawn from its position proximal to the enzyme towards the trans chamber. This resulted in a series of current steps with a ~ 36 pA peak as the abasic block traversed the pore lumen (Fig. S4, a). In contrast, for binary complexes formed with the 5ab(18,22) substrate, the initial position of the abasic block is distal from the enzyme. When this substrate is unzipped in the terminal cascade, no amplitude peak is observed (Fig. S4, b).
Results from our laboratory have shown that advance of a DNA template in the α-HL nanopore could be detected at single nucleotide precision during replication by T7DNAP(exo-) 17. However, for the majority of complexes with this enzyme only one or two nucleotide addition cycles could be monitored. To determine if phi29 DNAP was more efficient at catalyzing sequential nucleotide additions on the nanopore, we measured phi29 DNAP-driven displacement of synthetic DNA substrates molecules bearing 5 abasic inserts in their template strands. The map in Figure 3 was used to interpret changes in IEBS as single nucleotides were enzymatically added to or removed from the DNA 3′ terminus.
The experiment in Figure 2c showed that the slow excision of a ddNMP residue in the bulk phase could be exploited to capture complexes in the presence of Mg2+ in which the primer strand was intact. Importantly, this experiment also showed that excision of the ddNMP residue could be achieved on the pore, exposing the 3′-OH of the -1 residue and thus yielding a substrate that is potentially competent for synthesis reactions atop the pore in the presence of dNTPs. Consistent with previous findings 30,31, the gel assay in Figure 2a showed that in the presence of dNTPs the polymerization reaction dominated over the exonuclease reaction in bulk phase. These findings were essental to our strategy for DNA replication experiments: capture phi29 DNAP complexes bearing intact 3′-H terminated substrates in the presence of dNTPs, allow the excision reaction to occur on the pore, and use an abasic block marker in the template strand to determine unambiguously whether the polymerization reaction can be observed for complexes held atop the pore. Using this strategy, the majority of complexes captured in the nanopore should initiate replication at the same template position (-1 relative to the original n = 0 position of the starting substrate).
Because dGTP can slow the rate of ddCMP excision due to formation of ternary complexes (Fig. 3b, Fig. S3) we chose to conduct initial nanopore synthesis experiments using 20 µM each of dATP, dCTP, dTTP and 5 µM dGTP. We determined the effect of these conditions on the state of the DNA substrate molecules in bulk phase in a gel assay using the 5′-6-FAM, 3′-H hairpin substrate (Fig. 4b). After 10 minutes, 82.5% of the 67 mer starting substrate remained intact, and 13.6% was extended to the 102 mer product. After 20 minutes, these proportions were 69.4 % and 26.1% extension product, and by 45 minutes almost 30% of the fluorescein labeled hairpin had been extended. We therefore confined our measurements in the nanopore experiments to the first 10 minutes following the addition of Mg2+ and dNTP substrates to the cis chamber.
In initial nanopore replication experiments under these conditions (Fig. 4), we used a DNA substrate with the starting abasic configuration 5ab(15,19) bearing a 3′ ddCMP terminus (Fig. 4a). Typical ionic current traces for capture of phi 29 DNAP-DNA complexes at 180 mV with this substrate in the presence of 10 mM Mg2+, with or without dNTPs, are shown in Figure 4c and 4d, respectively. The dominant initial IEBS upon capture was ~ 29 pA under both conditions, with deflections to ~ 26 pA consistent with an oscillation between the map values for 5ab(15,19) binary and ternary complexes (Fig. 3c). Under both conditions, there was a delay at this starting IEBS level, afforded by the slow excision of the 3′ ddCMP terminus, after which a series of current changes ensued. We interpret the current changes in the experiment conducted in the absence of dNTPs (Fig. 4c) as follows: upon ddCMP excision, the phi29 DNAP exonuclease continued to sequentially cleave nucleotides from the primer terminus, resulting in a progressively shorter duplex segment and greater distance between the enzyme and the abasic insert. The abasic segment was thus moved through the pore toward the trans compartment, causing a progressive ionic current decrease. Eventually, the ionic current returned to the open channel state, consistent with dissociation of the DNA molecule from phi29 DNAP and its subsequent electrophoresis into the trans compartment.
In contrast, when the experiment was conducted in the presence of 20 µM each dATP, dCTP, dTTP and 5 µM dGTP a different ionic current pattern resulted, characterized by a peak at 35.4 pA (Fig. 4d). We hypothesized that these current changes occurred because, following phi29 DNAP excision of the ddCMP residue protecting the DNA 3′ terminus, the presence of dNTPs favored nucleotide additions catalyzed by phi29 DNAP while atop the pore. The duplex DNA segment was lengthened as phi29 DNAP moved progressively closer to the abasic insert within the DNA template, drawing it through the nanopore lumen with the attendant traversal of the major ionic current peak between abasic configurations 5ab(15,19) to 5ab(6,10) in the map in Figure 3b. Several DNA template replication reactions, catalyzed by phi29 DNAP-DNA complexes captured in series during this experiment are shown in Figure 4e.
In the gel experiment shown in Figure 4b, in addition to the starting 67 mer hairpin substrate and the full length extension products, intermediate bands corresponding to partial extension products accumulated with time (Fig. 4b, lanes 6 and 7). These products could arise due to depletion of dNTP pools in the bulk phase, as an increasing fraction of the DNA substrate molecules which are present at 1 µM in both the gel and nanopore assays are replicated. Because this has the potential to affect the extent and rate of synthesis catalyzed by phi29 DNAP complexes atop the pore, we examined whether this could be minimized by using a higher concentration of dNTPs.
We measured the extent of primer extension for the 5′-6-FAM, 3′-H terminated hairpin in the presence of 100 µM each of dGTP, dCTP, dTTP and dATP as a function of time (Fig. 5a and b). Under these conditions the rate of accumulation of the full-length product was slower than in the experiment in Figure 4b (using 20 µM each of dCTP, dTTP, dATP and 5 µM dGTP), likely due to the more efficient inhibition of excision of the ddCMP terminus afforded by the higher dGTP concentration. After 20 minutes, 86.3% of the starting DNA substrate remained intact, and 13.6% was fully extended (Fig. 5a, lane 6, and 5b), compared to 69.4% and 26.1% for these species, respectively, in reactions conducted for the same amount of time with the lower concentrations of dNTPs (Fig. 4b, lane 6). Importantly, even after 30 minutes, accumulation of shorter extension products was below the limit of detection of the assay. We therefore used dNTP substrates at a concentration of 100 µM each in subsequent replication experiments.
To test the model proposed for the ionic current signatures observed in the replication experiment in Figure 4d and e, we used a DNA hairpin substrate in which the first template dTMP residue was at a defined position relative to the abasic insert (Fig. 5). When DNA synthesis reactions are conducted with this substrate in the presence of 100 µM each of dGTP, dCTP, dTTP and ddATP, 12 nucleotides can be added, during which the abasic block will be drawn from its starting position of 5ab(18,22), across the 35.4 pA peak at 5ab(11,15), to position 5ab(6,10). After reaching the dTMP residue at position +12, replication is predicted to stall. In contrast, replication reactions conducted in the presence of 100 µM each dGTP, dCTP, dTTP and dATP should proceed past the +12 position.
When phi29 DNAP complexes formed with this DNA substrate were captured under both of these conditions, an initial period of several seconds occurred during which the dominant current amplitude was ~ 31 A, with oscillations to ~ 27 pA (Fig. 5d and e), similar to the map values for the ternary and binary complexes for this 5ab(18,22) configuration (Fig. 3c). After this state ended, the 35.4 pA ionic current peak was rapidly traversed, indicative of the abasic block being drawn through the lumen. If dGTP, dCTP, dTTP and ddATP were present in the cis chamber, after traversing the peak the polymerase stalled in a state in which the current oscillated between a dominant amplitude of ~ 25 pA to 28 pA for several seconds (Fig 5d). In contrast, in the presence of dATP rather than ddATP, the polymerase advanced without stalling through and beyond the 25 pA state (Fig. 5e). This establishes that the stalled state observed in the presence of ddATP (which indicates replicating complexes have reached the dTMP residue) is attained after the template segment that causes the amplitude peak traverses the lumen. Because reaching this dTMP template residue requires the nucleotide incorporations necessary to traverse the 5ab(17,21) to 5ab(7,11) abasic configurations, these experiments verify that the characteristic amplitude peak is due to replication that ensues following ddCMP excision on the pore.
Experiments using optical tweezers have shown that the rate of replication catalyzed by phi29 DNAP is slowed by tension on the template at forces between ~ 20 and ~ 37 pN 26. This result predicts that the rate of phi29 DNAP replication would be influenced by the voltage applied across the nanopore. However, the voltage regime where this would occur is not known. Figure 6 shows representative events during phi29 DNAP replication reactions along a 25 nt template segment of a DNA hairpin substrate (Fig. 6a), for experiments in which the applied potential was varied in 40 mV increments in the range between 220 mV and 100 mV. The starting abasic configuration for this substrate was 5ab(25,29); therefore during DNA synthesis, the 5 abasic insert will be drawn through the limiting aperture of the nanopore lumen, spanning abasic configurations 5ab(18,22) to 5ab(6,10) and thus the amplitudes mapped in Figure 3c. These peaks were traversed at each voltage, at rates that appeared to increase as applied voltage was decreased (Fig. 6b). We measured the time required to advance between two readily discernible current amplitudes corresponding to positions flanking the major current peak (blue arrows in Figure 6b, i), separated by approximately five nucleotides. At 220 mV, the median time required for replication over this distance was 227 ms (IQR = 174 ms, n = 45); at 100 mV, the median time for replication was 67 ms (IQR = 41 ms, n = 59).
In anticipation of replicating natural DNA templates in the nanopore, we measured phi29 DNAP-dependent replication of a longer segment within a synthetic DNA hairpin substrate. This hairpin substrate had a starting abasic configuration of 5ab(50,54), and up to 50 nucleotides can be added before the enzyme reaches the abasic block (Fig. 7a). When phi29 DNAP-DNA complexes formed with this substrate were captured at 180 mV in buffer containing 0.3 M KCl, there was an initial interval of several seconds during which the current oscillated between a dominant amplitude of ~ 23 pA, with transitions to ~ 25 pA. In 27 out of 47 captured complexes that started with this oscillation, when this period ended, the polymerase proceeded to traverse the mapped amplitude peak (Fig.7b).
We speculate that this oscillating signature corresponds to complexes captured with the ddCMP terminus intact, prior to the ddCMP excision reaction that permits synthesis to ensue, because i) a similar pattern invariably occurred between capture and synthesis for each successful replication reaction that subsequently traversed the abasic 35.4 pA peak in the experiments shown in Figures 4, ,5,5, ,6,6, and and7;7; ii) the upper and lower amplitude levels of the oscillation differ among those experiments in a manner that depends upon the starting abasic configuration of the DNA substrate; iii) those levels closely approximated the amplitudes for the binary and ternary complexes mapped for the abasic configuration for each substrate; and, iv) the proportion of time spent in the upper or lower amplitude state can be modulated as a function of dGTP concentration (data not shown).
We therefore used the end of this oscillating state as a start point to approximate the time required for phi29 DNAP to traverse the ~ 50 nt template segment. We measured from a small but reproducible current dip that occurred just after the oscillation ended (left blue arrow in Figure 7b) to a discernible amplitude state on the distal side of the major map peak (right blue arrow in Figure 7b). The median time required to replicate across this distance in buffer containing 0.3 M KCl was 1.39 s (IQR = 0.57 s; n = 27).
Surprisingly for this mesophilic polymerase, replication of the 5ab(50,54) substrate by phi29 DNAP was also detectable in buffer containing 0.6 M KCl (Fig. 7c). Like the replication reactions in 0.3 M KCl, these events began with a state in which the current oscillated between two levels for several seconds before the onset of synthesis (Fig. 7c). Under these higher ionic strength conditions, the current oscillated between a dominant level of ~ 32 pA, with transitions to ~ 34 pA. Replication that drew the abasic segment through the nanopore lumen, causing the abasic block to traverse the mapped amplitude peak, ensued in 25 out of 41 events that began with this current oscillation. In 0.6 M KCl, the median time required to traverse the distance between the end of the oscillation period (left blue arrow in Figure 7c) and the distal side of the major abasic amplitude peak (right blue arrow in Figure 7c) was 2.41 s (IQR = 1.13 s; n = 25).
The remarkable processivity and robust DNA binding properties of phi29 DNAP led us to predict that it would be a good candidate for observing processive DNA synthesis while held atop the α-HL nanopore in an electric field. This proved to be true; phi29 DNAP-DNA complexes remained associated with the nanopore orifice and readily catalyzed sequential nucleotide additions under 180 mV applied potential. This is in sharp contrast to T7DNAP(exo-), which was difficult to retain atop the pore for sequential additions even at lower voltages 17.
The tenacious binding of phi29 DNAP to DNA is highlighted by the different pathways by which this polymerase and KF(exo-) dissociate from DNA atop the nanopore, under conditions that do not permit exonucleolytic degradation of the DNA by phi29 DNAP. While the bond between KF and DNA can be pulled apart at 180 mV within a few milliseconds (Fig. 1c) with the hairpin duplex base-pairing remaining intact 12, dissociation from the tight binding of phi29 DNAP requires on average ~ 20 seconds, and the force pulling on the template strand suspended through the pore must promote unzipping of base-pairs while the duplex is held within the confines of the enzyme (Figs. 1d and 2b,c).
We exploited two features of the phi29 DNAP 3′-5′ exonuclease in this study. First, we found that a 3′-H terminated DNA substrate was degraded more slowly in bulk phase than a 3′-OH terminated substrate Fig. 2a). To our knowledge this is the first demonstration of discrimination against 3′-H terminated DNA substrates by the 3′-5′ exonuclease activity of phi29 DNAP. This feature provided protection in the bulk phase against both degradation and ddNMP excision-dependent initiation of primer extension of DNA substrate molecules. This protection in turn afforded a window following the addition of Mg2+ to the nanopore chamber during which we could capture numerous phi29 DNAP-DNA complexes in series in which the primer terminus was intact.
Second, we used the phi29 DNAP exonuclease activity to excise the ddNMP terminus of the DNA substrate in complexes while they were held atop the pore in an electric field. In the presence of dNTPs, the polymerization reaction is highly favored over processive degradation 30,31. Therefore excision of the ddCMP residue to yield a primer strand 3′-OH permitted the subsequent initiation of synthesis from a defined DNA template position.
The excision of the ddNMP terminus may be accelerated in complexes by the electric field force atop the pore, as we observed that the time from complex capture to the initiation of synthesis decreased when voltage was increased (not shown). This voltage-promoted excision would nonetheless differ from the processive exonucleolytic regime induced under conditions of high template tension in optical tweezers experiments, in which processive exonucleolytic cleavage dominated even in presence of dNTPs 26. In contrast, while the initiation of synthesis required excision of the ddCMP residue, the polymerization reaction dominated in the nanopore experiments (Figs. 4, ,5,5, ,66 and and7)7) even at 220 mV applied potential (Fig. 6).
Maintenance of a significant pool of intact, unextended DNA substrate in the bulk phase due to the slow exoncleolytic removal of a ddNMP primer terminus allowed us to examine phi29 DNA-catalyzed synthesis in the nanopore under relatively simple conditions. Nonetheless, due to concerns regarding the slow change in the state of the DNA molecules and potential dNTP substrate depletion in the bulk phase over time, this strategy puts constraints on the time frame in which experiments can be conducted. The use of a more robust means of protecting DNA substrate molecules in the bulk phase, such as the blocking oligomers recently employed with KF(exo-) and T7DNAP(exo-) 17, will extend the utility of this enzyme for both DNA sequencing applications and mechanistic studies of polymerase function using the nanopore.
The results of this study demonstrate that phi29 DNAP has properties ideally suited for moving long strands of DNA through nanoscale pores at a rate that is compatible with reliable base detection and identification. In this study we used only chemically synthesized DNA templates, yet the number of sequential nucleotide additions catalyzed by a single enzyme molecule that could be observed was limited only by DNA template length. Features within current traces, such as the ionic current flicker within binary complex events that can predict ternary complex amplitude (Fig. 3), and the oscillation between two amplitudes upon complex capture that precedes replication reactions (Figs. 4, ,5,5, and and7),7), suggest that biochemical processes such as the fingers opening-closing transition 22 and dNTP binding 15 may have discernible signatures. The ability to observe dynamics in complexes under defined substrate conditions and to resolve individual catalytic cycles (Figs. 4, ,5,5, ,6,6, and and7)7) in real time at high bandwidth offers the opportunity to quantify biochemical transformations as a function of applied voltage and dNTP concentration.
We thank Christopher Lam for help with nanopore mapping experiments, Ai Mai for oligonucleotide purification, Robin Abu-Shumays for comments on the manuscript, and Peter Walker (Stanford University Protein and Nucleic Acid Facility) for expert oligonucleotide synthesis. We are grateful to Chris Benoit (Enzymatics) for supplying high specific activity, detergent-free, wild-type phi29 DNA polymerase. This work was supported by grants from NHGRI (1RC2HG005553) to M.A. and NIGMS (1R01GM087484-01A2) to K.R.L. and M.A.
Supporting Information Available: Figure S1 (sequences of 5′-6-FAM, 3′-OH and 5′-6-FAM, 3′-H DNA oligonucleotide substrates used in gel assays); Figure S2 (unbound DNA at 70 mV applied potential); Figure S3 (primer extension gel assays supporting phi2 DNAP-DNA-dGTP ternary complex formation); Figure S4 (amplitude steps in the terminal cascade vary as a function of initial DNA substrate abasic configuration); Complete author list for references 4 and 5. This material is available free of charge via the Internet at http://pubs.acs.org.