Advertisement

Kinetics of DNA strand transfer between polymerase and proofreading exonuclease active sites regulates error correction during high fidelity replication

Open AccessPublished:November 24, 2022DOI:https://doi.org/10.1016/j.jbc.2022.102744

      Abstract:

      We show that T7 DNA polymerase and exonuclease domains contribute to selective error correction during DNA replication by regulating bidirectional strand transfer between the two active sites. To explore the kinetic basis for selective removal of mismatches we used a fluorescent cytosine analog (tCo) to monitor the kinetics of DNA transfer between the exonuclease and polymerase sites. We globally fit stopped-flow fluorescence and base excision kinetic data, and compared results obtained with ssDNA versus duplex DNA to resolve how DNA transfer governs exonuclease specificity. We performed parallel studies using hydrolysis-resistant phosphorothioate oligonucleotides to monitor DNA transfer to the exonuclease site without hydrolysis—ssDNA binds to the exonuclease site at the diffusion limit (109 M-1s-1) but is relatively weak (40 nM) although hydrolysis of the 3’-terminal nucleotide is fast (>5,000 s-1). Analysis using duplex DNA with a 3’-terminal mismatch, or a buried mismatch exposed a unique intermediate state between polymerase and exonuclease active sites and revealed that transfer via the intermediate to the exonuclease site is stimulated by free nucleoside triphosphates. Transfer from the exonuclease site back to the polymerase site after cleavage is fast and efficient. We propose a model to explain why buried mismatches are removed faster than single 3’-terminal mismatches and thereby provide an additional opportunity for error correction. Our data provide the first comprehensive model to explain how DNA transfer from polymerase to exonuclease active sites and back again after base excision allow efficient, selective mismatch removal during DNA replication to improve fidelity by more than 1000-fold.

      Keywords

      Introduction

      DNA polymerase fidelity is finely tuned to support the survival of most organisms. For many viral enzymes such as HIV reverse transcriptase, a polymerase with only moderate fidelity (1 error in 104) strikes the right balance between replication that is accurate enough to propagate the virus, but with a sufficient error rate to rapidly evade the host’s immune response. For high fidelity enzymes the net accuracy is a product of the intrinsic fidelity during nucleotide incorporation at the polymerase active site and the contribution of a 3’-5’ proofreading exonuclease that selectively removes mismatched nucleotides to further improve fidelity. For the well-characterized high fidelity T7 DNA polymerase, the proofreading exonuclease domain binds single stranded DNA and is 25Å away from the polymerase active site. Therefore, a major challenge is to understand how the exonuclease achieves specificity to remove mismatches when the DNA is no longer base-paired with the template strand when in the exonuclease active site. The enzyme solves the problem by modulating exonuclease hydrolysis by kinetic partitioning of the primer strand between forward polymerization versus transfer of the primer strand to the exonuclease site. Following incorporation of a mismatch, the partitioning changes due to much slower forward polymerization and slightly faster transfer of the primer stand to the exonuclease site (
      • Wong I.
      • Patel S.S.
      • Johnson K.A.
      An induced-fit kinetic mechanism for DNA replication fidelity: direct measurement by single-turnover kinetics.
      ,
      • Donlin M.J.
      • Patel S.S.
      • Johnson K.A.
      Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
      ). Once in the exonuclease active site, the 3’-terminal nucleotide is rapidly excised, then the primer strand transfers back to the polymerase active site to allow continuing processive polymerization. Thus, the proofreading exonuclease achieves specificity indirectly based on the events occurring at the polymerase active site that regulate the partitioning between polymerization to extend versus hydrolysis to remove of the 3’-terminal base pair. Although this model was derived three decades ago and provides a satisfying resolution of the conundrum, little is known regarding the kinetics of primer strand transfer between the polymerase (pol) and exonuclease (exo) active sites despite numerous studies (
      • Dodd T.
      • Botto M.
      • Paul F.
      • Fernandez-Leiro R.
      • Lamers Meindert H.
      • Ivanov I.
      Polymerization and editing modes of a high-fidelity DNA polymerase are linked by a well-defined path.
      ,
      • Black C.B.
      • Cowan J.A.
      A critical evaluation of metal-promoted Klenow 3′-5′ exonuclease activity: calorimetric and kinetic analyses support a one-metal-ion mechanism.
      ,
      • Lamichhane R.
      • Berezhna S.Y.
      • Gill J.P.
      • Van der Schans E.
      • Millar D.P.
      Dynamics of Site Switching in DNA Polymerase.
      ,
      • Johnson A.A.
      • Johnson K.A.
      Exonuclease Proofreading by Human Mitochondrial DNA Polymerase.
      ,
      • Beechem J.M.
      • Otto M.R.
      • Bloom L.B.
      • Eritja R.
      • Reha-Krantz L.J.
      • Goodman M.F.
      Exonuclease−Polymerase Active Site Partitioning of Primer−Template DNA Strands and Equilibrium Mg2+ Binding Properties of Bacteriophage T4 DNA Polymerase.
      ).
      We previously reported methods to label the T7 DNA polymerase with a fluorescent artificial amino acid that gives a fluorescence change upon nucleotide binding (
      • Dangerfield T.L.
      • Johnson K.A.
      Optimized incorporation of an unnatural fluorescent amino acid affords measurement of conformational dynamics governing high-fidelity DNA replication.
      ). We then used this variant to characterize the conformational dynamics that govern correct nucleotide incorporation (
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ) and compared that to mismatched nucleotide incorporation (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Conformational dynamics during misincorporation and mismatch extension defined using a DNA polymerase with a fluorescent artificial amino acid.
      ) to define the mechanistic basis for the extraordinary fidelity of this enzyme. We also previously characterized the DNA substrate specificity for the proofreading exonuclease of T7 DNA polymerase using rapid-quench methods and showed that the rate of mismatch excision increased with the single stranded nature of the DNA substrate; that is, ssDNA was hydrolyzed the fastest, while the 3’-terminal base in duplex DNA was removed at a rate that increased with the number of mismatched base pairs (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ). Surprisingly, we found that a mismatch buried by a single correct base pair was removed much more efficiently than a single terminal mismatch, demonstrating that the enzyme affords multiple opportunities to remove a mismatch. We also presented a structural model for the proofreading complex for T7 DNA polymerase based on homology to a crystal structure of a Klenow Fragment of E. coli DNA polymerase-I editing complex, which showed that three bases must melt for the primer strand to reach fully into the exonuclease active site. Based on the observation that the stereochemistry of the exonuclease hydrolysis reaction is the same as that for similar polymerases that have been characterized we proposed a catalytic mechanism for exonuclease hydrolysis (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Conformational dynamics during misincorporation and mismatch extension defined using a DNA polymerase with a fluorescent artificial amino acid.
      ,
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ). This study, however, left unanswered questions about the detailed kinetics governing partitioning of the DNA between the two active sites. Because hydrolysis of ssDNA at least 100-fold faster than removal of a 3’-terminal mismatch from duplex DNA, the rate of transfer of the DNA from the polymerase to the exonuclease site appears to limit the rate of hydrolysis.
      In this paper we expand on our previous studies on exonuclease proofreading by T7 DNA polymerase through characterization of exonuclease hydrolysis on three physiologically relevant substrates: ssDNA, DNA containing a single 3’ mismatch, and DNA containing a buried mismatch. To measure DNA binding and dissociation kinetics at the exonuclease active site, we enzymatically synthesized hydrolysis-resistant DNA substrates containing phosphorothioate nucleotides—this is preferred to the alternative method of using exonuclease-deficient mutants derived by removal of charged residues that bind Mg2+ and thereby perturb the electrostatics at the active site. We also introduce a DNA substrate containing the fluorescent cytosine analog 1,3-diaza-2-oxophenoxazine (tCo) that provides a large fluorescence signal upon transitioning from double stranded in the polymerase active site to single stranded DNA in the exonuclease active site. We provide new data to define the kinetics of the transfer of DNA between the pol (polymerase) and exo (exonuclease) active sites.
      We first show that excision of ssDNA in the exo site is very fast, so in subsequent experiments with dsDNA the observed rates of excision are limited by, and thereby serve as a readout for, the rates of transfer of the primer strand to the exonuclease site. Fluorescence signals are used to monitor DNA binding (FAM label anisotropy) and unwinding (tCo signal), while rapid quench methods define the kinetics of nucleotide excision. We use two protocols to explore DNA binding to the pol and exo sites: (
      • Wong I.
      • Patel S.S.
      • Johnson K.A.
      An induced-fit kinetic mechanism for DNA replication fidelity: direct measurement by single-turnover kinetics.
      ) Mixing enzyme and DNA to initiate the reaction provides evidence to define kinetics of binding to pol and exo sites; (
      • Donlin M.J.
      • Patel S.S.
      • Johnson K.A.
      Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
      ) By preincubating enzyme and DNA without Mg2+, then initiating the reaction by adding Mg2+ gives data to define the equilibrium distribution between binding states and the kinetics of partitioning between the sites, as evidenced by fluorescence and rapid-quench kinetic data; (
      • Dodd T.
      • Botto M.
      • Paul F.
      • Fernandez-Leiro R.
      • Lamers Meindert H.
      • Ivanov I.
      Polymerization and editing modes of a high-fidelity DNA polymerase are linked by a well-defined path.
      ) By adding nucleotides that extend the DNA only after removal of a mismatch we obtain information to define the kinetics of transfer to the exo site and back to the pol site. In these experiments, we know that polymerization is very fast and so the observed rates appear to be limited by the transfer between the active sites. Naturally, each of these experiments provides information on the multiple binding states and kinetics of transfer between the states, but no single experiment can be interpreted directly. Rather, it is only through global data fitting that the relationships between DNA binding and unwinding can be correlated with the kinetics of strand transfer and nucleotide excision to define a comprehensive model to define the molecular basis for selective mismatch removal by the proofreading exonuclease.

      Results:

      Use of the fluorescent cytosine analog (tCo) to measure DNA binding and melting of double stranded DNA

      We previously described kinetics of exonuclease hydrolysis on various DNA substrates using rapid-quench methods (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ). These studies, while useful to determine relative observed rates of excision on different DNA substrates, do not provide sufficient information to determine rate constants for individual steps governing the exonuclease proofreading reaction. In particular, to monitor transfer of the DNA between the pol and exo active sites, we prepared a DNA substrate containing the fluorescent cytosine analog tCo at the n-2 position from the 3’ end of the primer strand (Figures 1 and 2). The analog tCo has some unique properties that make it an ideal fluorescent probe relative to other available fluorescent nucleoside analogs, such as 2-aminopurine—it has a quantum yield of 0.3 when in ssDNA oligonucleotides, which is reduced to 0.22 in double stranded DNA substrates, making tCo the brightest DNA base analog so far reported and it provides a significant change in signal to monitor DNA unwinding (
      • Sandin P.
      • Börjesson K.
      • Li H.
      • Mårtensson J.
      • Brown T.
      • Wilhelmsson L.M.
      • Albinsson B.
      Characterization and use of an unprecedentedly bright and structurally non-perturbing fluorescent DNA base analogue.
      ). Furthermore, studies have shown that tCo base pairs almost exclusively with guanine, causes minimal perturbations to the native structure of DNA (
      • Sandin P.
      • Börjesson K.
      • Li H.
      • Mårtensson J.
      • Brown T.
      • Wilhelmsson L.M.
      • Albinsson B.
      Characterization and use of an unprecedentedly bright and structurally non-perturbing fluorescent DNA base analogue.
      ), and Circular Dichroism (CD) spectra showed that all tCo containing oligonucleotide duplexes form the characteristic B-form DNA. Other fluorescent DNA base analogs share a common feature in that the emission intensity is highly sensitive to microenvironment, while tCo is relatively insensitive to its local chemical environment (
      • Sandin P.
      • Börjesson K.
      • Li H.
      • Mårtensson J.
      • Brown T.
      • Wilhelmsson L.M.
      • Albinsson B.
      Characterization and use of an unprecedentedly bright and structurally non-perturbing fluorescent DNA base analogue.
      ) but is quenched by base-stacking interactions. Furthermore Klenow Fragment of E. coli DNA polymerase I and human DNA polymerase α have been shown to incorporate the tCo analog with high efficiency (
      • Stengel G.
      • Purse B.W.
      • Wilhelmsson L.M.
      • Urban M.
      • Kuchta R.D.
      Ambivalent Incorporation of the Fluorescent Cytosine Analogues tC and tCo by Human DNA Polymerase α and Klenow Fragment.
      ,
      • Sandin P.
      • Stengel G.
      • Ljungdahl T.
      • Börjesson K.
      • Macao B.
      • Wilhelmsson L.M.
      Highly efficient incorporation of the fluorescent nucleotide analogs tC and tCO by Klenow fragment.
      ).
      Figure thumbnail gr1
      Figure 1Fluorescent cytosine analog, tCo, as a probe for DNA melting to monitor the transfer between exonuclease and polymerase active sites. Base pairing between G (grey) and tCo (green) is shown, with 3 hydrogen bonds as expected for a G:C base pair. The fluorescence of tCo increases when the duplex DNA melts to form to single stranded DNA.
      Figure thumbnail gr2
      Figure 2Kinetics of ssDNA binding and excision. DNA substrate: The 27nt ssDNA substrate used in these experiments is shown at the top of this figure. Experiments with 6-FAM labeled DNA contained a 5’ [6-FAM] label and a C rather than tCo. Experiments in (A) through (C) were performed using a hydrolysis resistant phosphorothioate (Sp isomer) ssDNA (see Experimental Procedures). Below we describe approximate estimates of kinetic parameters derived from equation-based data fitting, but the solid lines through the data show the best global fits with KinTek Explorer using the model at the bottom of the figure and parameters listed in . A) Stopped flow ssDNA binding rate. A solution of 50 nM FAM-phosphorothioate ssDNA (FAM-27-PThio) and 12.5 mM Mg2+ was mixed with varying concentrations of T7 DNA polymerase (100-400 nM) to start the reaction and the change in anisotropy as a function of time was monitored. This experiment provides an estimate of the ssDNA binding rate constant (k1) of ∼1,000 μM-1s-1, and a significant dissociation rate constant (k-1 ∼ 40 s-1), defined by the amplitude dependence of the reaction. B) Stopped flow ssDNA off-rate experiment. A solution of 50 nM FAM phosphorothioate ssDNA (FAM-27-PThio), 75 nM T7 DNA polymerase, and 12.5 mM Mg2+ was mixed with 1 μM unlabeled phosphorothioate ssDNA as a trap to start the reaction and the anisotropy as a function of time is shown. The data best fit a single exponential function with an observed rate of 47.5 ± 1.5 s-1. C) Stopped flow ssDNA equilibrium titration. A solution of 25 nM tCo phosphorothioate ssDNA (tCo-27-PThio) and 12.5 mM Mg2+ was titrated with T7 DNA polymerase over the course of 5 minutes with constant stirring using a micro stir bar. The fluorescence was corrected for the small dilution and an inverse inner filter effect as described in the Experimental Procedures section. The data fit a hyperbola giving a Kd ∼ 60 nM. D) Rapid quench ssDNA excision kinetics – gel. A solution of 150 nM FAM ssDNA (FAM-27) and 12.5 mM Mg2+ was mixed with varying concentrations of T7 DNA polymerase (0.25 – 1.25 μM) and 0.1 mg/ml BSA to start the reaction. Samples were quenched with EDTA at various times and products were resolved by denaturing PAGE (15% acrylamide). Lengths of ssDNA are given to the right of the gel. The gel from the experiment at 0.25 μM T7 DNA polymerase is shown as a representative data set. E) Rapid quench ssDNA excision kinetics – concentration versus time. Reaction conditions are given in (D). Data corresponding to 27, 26, 25, and 24 nt ssDNA are shown in red, green, blue, and magenta, respectively. Other bands are omitted for clarity. The experiment at 1.25 μM T7 DNA polymerase is shown as a representative data set. F) Rapid quench ssDNA excision, inhibition by dCMP. A solution of 1 μM T7 DNA polymerase, 0.1 mg/ml BSA, and 12.5 mM Mg2+ was mixed with 150 nM FAM ssDNA (FAM-27) and either 0 or 4 mM dCMP to start the reaction. Samples were quenched with EDTA and products resolved by PAGE as above. Data at 4 mM dCMP are shown as open circles with dashed lines for the fit by simulation, while data at 0 mM dCMP are shown with filled circles and solid lines for the fit by simulation. Scheme: Kinetic model for ssDNA binding and excision. The x superscript following the D indicates binding at the exo site and the numbered subscript indicates the length of the DNA substrate.
      We chose the n-2 position in the primer strand to place the tCo label based on our homology model for the exonuclease proofreading complex that we described previously (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ). This model clearly showed that three bases must melt for the primer strand to partition completely into the exonuclease active site, so the n-2 position where we incorporated tCo should undergo a double strand to single strand transition upon complete partitioning into the exonuclease active site, giving rise to a fluorescence change. In addition, by placing the fluorescent base analog as the third base from the primer terminus, we reasoned that the analog is less likely to cause significant perturbations affecting DNA binding in the exo site and partitioning of the primer strand. We also carefully selected the surrounding sequences using a table from a previous study showing that tCo surrounded by an A on either side has negligible effect on the melting temperature of the duplex (
      • Sandin P.
      • Börjesson K.
      • Li H.
      • Mårtensson J.
      • Brown T.
      • Wilhelmsson L.M.
      • Albinsson B.
      Characterization and use of an unprecedentedly bright and structurally non-perturbing fluorescent DNA base analogue.
      ). As a control experiment to test whether the tCo label affected normal polymerization kinetics, we performed a rapid-quench pre-steady state burst experiment (Figure S1) where the next correct nucleotide was added to the primer strand. This experiment demonstrated that T7 DNA polymerase is still able to rapidly add bases to a primer containing tCo at rates exceeding 300 nt s-1 and with full amplitude, as defined in studies using normal nucleotides (
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ).

      Hydrolysis-resistant phosphorothioate oligonucleotides to study DNA binding

      In our previous paper on exonuclease proofreading (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ), we determined the stereospecificity of the T7 DNA exonuclease using phosphorothioate DNA and showed that the Sp isomer was resistant to hydrolysis. This is consistent with the stereochemistry of the reactions for T4 DNA polymerase (
      • Otto M.R.
      • Bloom L.B.
      • Goodman M.F.
      • Beechem J.M.
      Stopped-Flow Fluorescence Study of Precatalytic Primer Strand Base-Unstacking Transitions in the Exonuclease Cleft of Bacteriophage T4 DNA Polymerase.
      ) and Klenow Fragment of E. coli DNA polymerase I (
      • Brautigam C.A.
      • Steitz T.A.
      Structural principles for the inhibition of the 3′-5′ exonuclease activity of Escherichia coli DNA polymerase I by phosphorothioates 11Edited by R. Huber.
      ,
      • Brautigam C.A.
      • Sun S.
      • Piccirilli J.A.
      • Steitz T.A.
      Structures of Normal Single-Stranded DNA and Deoxyribo-3‘-S-phosphorothiolates Bound to the 3‘-5‘ Exonucleolytic Active Site of DNA Polymerase I from Escherichia coli.
      ). Using this information, we developed methods to enzymatically degrade the Rp isomer in a racemic mixture of both diastereomers (Experimental Procedures), followed by purification of the resulting hydrolysis-resistant oligonucleotide by ion exchange chromatography. The presence of full-length oligonucleotide was then confirmed by MALDI-MS. Although previous studies on other enzymes (
      • Brautigam C.A.
      • Steitz T.A.
      Structural principles for the inhibition of the 3′-5′ exonuclease activity of Escherichia coli DNA polymerase I by phosphorothioates 11Edited by R. Huber.
      ,
      • Brautigam C.A.
      • Sun S.
      • Piccirilli J.A.
      • Steitz T.A.
      Structures of Normal Single-Stranded DNA and Deoxyribo-3‘-S-phosphorothiolates Bound to the 3‘-5‘ Exonucleolytic Active Site of DNA Polymerase I from Escherichia coli.
      ) have shown that the Sp phosphorothioate is not a perfect substrate for exonuclease domains as it interferes with metal binding at the active site, we opted to use the phosphorothioate approach to measure DNA binding rather than the alternative of performing studies in the absence of Mg2+ or with active site mutants. Preliminary studies with the exo- T7 DNA polymerase variant showed large differences in DNA binding and dissociation kinetics between experiments with and without Mg2+ (not shown), while the difference between normal and phosphorothioate DNA in the presence of Mg2+ was minimal.

      Kinetics of ssDNA binding

      Since single stranded DNA (ssDNA) is the preferred substrate for the proofreading exonuclease (
      • Donlin M.J.
      • Patel S.S.
      • Johnson K.A.
      Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
      ), we began our detailed kinetic studies on exonuclease proofreading using this substrate. The physiologically relevant substrate is duplex DNA with a melted single-stranded DNA 3’-strand projecting into the exonuclease site; therefore, ssDNA provides a useful model for understanding the reactions of DNA in the exonuclease site. Before investigating the kinetics of excision, we used hydrolysis-resistant phosphorothioate (Sp isomer) single stranded oligonucleotides to measure binding kinetics without hydrolysis. We first measured the ssDNA binding rate by stopped-flow fluorescence using the anisotropy signal from a 6-carboxyfluorescein (6-FAM) labeled phosphorothioate ssDNA substrate (FAM-27-PThio in Table 1). In this experiment, a fixed concentration of FAM-labeled phosphorothioate ssDNA was mixed with varying concentrations of T7 DNA polymerase (with enzyme in excess) to start the reaction and we then monitored the change in anisotropy as a function of time (Figure 2A). The rapid timescale of the experiment indicates that the binding rate is extremely fast, with a second order rate constant of ∼1,000 μM-1 s-1. The observed amplitude dependence on enzyme concentration suggests that the equilibrium dissociation constant for single stranded DNA is comparable to the enzyme concentrations covered in the experiment. We then directly measured the single stranded DNA dissociate rate using the anisotropy signal from the FAM-labeled phosphorothioate ssDNA (Figure 2B). A solution of T7 DNA polymerase and FAM-labeled phosphorothioate ssDNA was mixed with a large excess of unlabeled phosphorothioate ssDNA in the stopped-flow to start the reaction. The unlabeled phosphorothioate ssDNA acts as a trap for any enzyme that has dissociated from the T7 DNA polymerase/FAM-labeled ssDNA complex. The data best fit a single exponential function with an observed rate of 48 ± 1.5 s-1 (fit not shown). This fast off-rate is consistent with the amplitude dependence observed in the ssDNA binding rate experiment. Simultaneous global fitting of binding and dissociation data resolves both constants (Table 2). Finally, to ensure all binding steps have been accounted for, an equilibrium titration experiment was performed using the tCo phosphorothioate ssDNA. In this experiment, a solution of tCo phosphorothioate ssDNA was incubated in a temperature-controlled cuvette and T7 DNA polymerase was titrated over the course of 5 minutes with constant mixing from a micro stir bar. Since the signal for ssDNA binding using the tCo phosphorothioate substrate was small when exciting tCo directly at 365 nm, we opted to use the FRET signal between tryptophan residues in T7 DNA polymerase and the tCo label by exciting the sample at 295 nm which produced a large signal upon DNA binding. The titration data were corrected for the small dilution and the fluorescence artifact from titration of the enzyme (FRET donor) as described in the Experimental Procedures section to give the corrected data shown in Figure 2C. The titration curve best fits a hyperbola with a Kd for ssDNA binding of approximately 60 nM. All binding data were fit globally in KinTek Explorer, and the resulting fits are shown as the solid lines through the data in Figure 2A-2C with rate constants given in Table 2 with k1 = 1100 ± 20 μM-1 s-1, k-1 = 44 ± 1.3 s-1 and the Kd = 40 ± 1.4 nM.
      Table 1Oligonucleotides used in this study
      Oligo NameSequence 5’→3’Extinction Coefficient, 260 nm (M-1 cm-1)
      FAM-27[6-FAM]-CCGTCGCAGCCGTCCAACCAACTCAAC266,660
      FAM-27-PThio[6-FAM]-CCGTCGCAGCCGTCCAACCAACTACA*C267,660
      27CCGTCGCAGCCGTCCAACCAACTACAC246,700
      27-PThioCCGTCGCAGCCGTCCAACCAACTACA*C246,700
      tCo-27CCGTCGCAGCCGTCCAACCAACTA[tCo]AC250,100
      32P-tCo-2732P-CCGTCGCAGCCGTCCAACCAACTA[tCo]AC250,100
      tCo-27-PThioCCGTCGCAGCCGTCCAACCAACTA[tCo]A*C250,100
      45-P/TGGACGGCATTGGATCGATGTGTAGTTGGTTGGACGGCTGCGACGG435,100
      45-3’MMGGACGGCATTGGATCGATCTGTAGTTGGTTGGACGGCTGCGACGG431,600
      45-3’MM-PThioGGACGGCATTGGATCGATCTGTAGTTGGTTGGACGGCTGCGACG*G431,600
      45-BMMGGACGGCATTGGATCGATGAGTAGTTGGTTGGACGGCTGCGACGG439,600
      45-BMM-PThioGGACGGCATTGGATCGATGAGTAGTTGGTTGGACGGCTGCGACG*G439,600
      We use * to designate the phosphorothioate linkage.
      Table 2Kinetic Parameters for ssDNA
      ParameterBest-Fit95% Confidence Interval
      k11,100 μM-1 s-1(1070, 1130)
      k-144.1 s-1(42.8, 45.5)
      k25,000 s-1Locked
      k-21,670 s-1(1470, 1920)
      k3443 s-1(415, 480)
      k-30.0518 μM-1 s-1(0.0374, 0.0648)
      k3,2546 s-1(516, 584)
      k3,3631 s-1(591, 675)
      KI5.65 mM(4.52, 7.06)
      aRate constants derived in Figure 2 are summarized with confidence intervals derived by contour analysis. Because the contours are symmetrical, we can estimate the standard errors directly from the confidence intervals, for example, k1 = 1100 ± 30 μM-1 s-1. The rate constant k2 was locked at a lower limit of 5000 s-1. Rate constants k3, k3,2, and k3,3 are for the release of sequential dNMP products from the 3’-end of substrate FAM-27 shown in Table 1 (i.e. dCMP, dAMP, and dAMP).

      Kinetics of ssDNA hydrolysis:

      Next, we determined the kinetics of ssDNA hydrolysis using rapid-quench methods. A solution of FAM-ssDNA was mixed with varying concentrations of T7 DNA polymerase (with the enzyme in excess and therefore determining the observed rate) to start the reaction. Samples were quenched with EDTA at various times and products were resolved by denaturing PAGE. A representative gel is shown in Figure 2D. Preliminary inspection of the gel clearly reveals a fast rate of excision of ssDNA by T7 DNA polymerase, with 8 bases removed in less than 20 milliseconds. By performing the experiment at multiple enzyme concentrations, we obtained estimates of the rate constant for DNA binding in agreement with the measurements made using the phosphorothioate DNA. This analysis afforded resolution of the rate constants for DNA binding and chemistry (k1, and k2, respectively, in the scheme in Figure 2).
      A representative plot of concentration of product versus time at a single enzyme concentration is given in Figure 2E, showing products from 27 – 24 nt in length (shorter products were omitted for clarity). The data best fit a model with fast ssDNA binding (∼1,000 μM-1 s-1), and an even faster rate of chemistry (> 5,000 s-1). In other words, the observed excision kinetics for the first hydrolyzed base at all enzyme concentrations tested was limited by the rate of ssDNA binding. Fitting the data also demonstrated that the first nucleotide was hydrolyzed faster than subsequent bases, leading to a model where ssDNA binds, the terminal base is rapidly hydrolyzed, and release of the dNMP product (k3 in the scheme in Figure 2) limits the rate of subsequent excision reactions (∼500 s-1) but the cleavage rate varies depending on the identity of the terminal base (440 - 630 s-1). Rates of dNMP release extracted from the fitting for excision of the first 3 bases are given in Table 2.
      In our previous work on correct nucleotide insertion for T7 DNA polymerase (
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ), we were able to measure rates of the pyrophosphorolysis reaction (reverse of nucleotide incorporation) which gave insights into the translocation step that was not observable when monitoring the reaction in the forward direction. Since the reverse reaction for exonuclease hydrolysis is not observable even at long time scales (data not shown), we performed the exonuclease hydrolysis experiment (forward reaction) in the presence of dCMP in an attempt to extract the second order rate constant for dNMP binding (k-3 in the scheme in Figure 2) and the reverse of chemistry (k-2 in the scheme in Figure 2). The results are shown in Figure 2F (dashed lines). In this experiment a solution of FAM-ssDNA was mixed with an excess of T7 DNA polymerase and either 0 or 4 mM dCMP to start the reaction in the rapid quench-flow instrument and samples collected by quenching at various times were processed as described above. The data indicate that dCMP binding is relatively slow (k-3 = 0.052 μM-1 s-1) but the data afford an estimate for the equilibrium constant for the chemistry step. Since the rate constant for chemistry is too fast to measure, we locked this rate constant at a lower limit of 5,000 s-1 and fit the data to estimate the rate constant for the reverse of chemistry (k-2 = 1,670 s-1) resulting in an equilibrium constant for the chemistry step, K2 = 3.4. Fitting the data also required including a step where dCMP binds to the free enzyme, acting as a weak competitive inhibitor for ssDNA binding (Ki in the scheme in Figure 2). To estimate the equilibrium constant, we locked the rate of dNMP binding to the free enzyme at 100 μM-1 s-1 and allowed the dNMP off-rate to float to extract a Ki of 5.6 mM.
      All experiments for ssDNA were fit globally using the model in Figure 2, and we then tested whether each parameter was well constrained by the data based on confidence contour analysis with the FitSpace feature of KinTek Explorer (
      • Johnson K.A.
      • Simpson Z.B.
      • Blom T.
      FitSpace Explorer: An algorithm to evaluate multidimensional parameter space in fitting kinetic data.
      ) as shown in Figure 3. This statistical analysis is the most rigorous test to determine whether parameters in the model are well constrained by the data and circumvents many of the pitfalls of error analysis from nonlinear regression when parameters are linked or are not well-constrained (

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ). The contours are obtained by measuring the dependence of the χ2 on each parameter while allowing all remaining parameters to float in finding the optimal fit. The parabolic confidence contour curves outline the space over which parameters can vary and indicate that each parameter was well constrained by the data—the dashed line each panel gives the threshold corresponding to the 95% confidence interval for each parameter, calculated based on the number of parameters and number of data points in the fitting (

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ).
      Figure thumbnail gr3
      Figure 3Confidence contours for ssDNA binding and excision kinetics. Confidence contours from the FitSpace calculation in KinTek Explorer are shown for the global fit of all ssDNA experiments in . k2 was locked at 5,000 s-1 in the fitting. The dashed line gives the 99.5% threshold, corresponding to the 95% confidence interval based on the number of data points and parameters in the fitting, as calculated by KinTek Explorer. Best fit parameters and confidence intervals are given in .

      Kinetics of 3’ mismatch hydrolysis

      Although ssDNA hydrolysis is important mechanistically and puts estimates on the rate of chemistry occurring at the exonuclease active site, double stranded DNA (dsDNA) containing mismatch(es) at the 3’-end of the primer strand in duplex DNA are mechanistically more important as the enzyme must sense the mismatch and repair it faithfully to improve DNA replication fidelity. Therefore, we investigated the kinetics of hydrolysis of a single 3’ mismatch, the product formed after the enzyme misincorporates a base during replication at the polymerase active site. To build the kinetic model from the ground up, we begin by performing experiments to measure binding using hydrolysis-resistant phosphorothioate dsDNA substrates labeled with FAM or tCo. First, we measured the DNA binding rate using fluorescence anisotropy with a double stranded FAM phosphorothioate 3’ mismatch DNA substrate shown at the top of Figure 4. A solution of non-hydrolyzable FAM-DNA was mixed with varying concentrations of T7 DNA polymerase (with E in excess) to start the reaction in the stopped-flow and the increase in anisotropy upon DNA binding was monitored (Figure 4A). Initial inspection of the traces shows that, unlike for ssDNA, the traces all approach the same endpoint, suggesting DNA binding is tight relative to the fixed DNA concentration (25 nM). The binding rate is also fast, occurring on a 25 ms timescale for the concentrations used in the experiment, which suggests a net diffusion-limited DNA binding rate of ∼ 1000 μM-1 s-1, similar to the ssDNA binding kinetics. As we will show with subsequent experiments, the kinetic model for exonuclease hydrolysis on dsDNA substrates is complex, requiring multiple techniques to monitor different aspects of the reaction to provide a full model derived by global data fitting. We initially performed a DNA binding rate experiment using the tCo fluorescence with the phosphorothioate DNA substrate, but the signal was too complicated to interpret (data not shown). The advantage of anisotropy in this experiment is that the signal change is known to be due to DNA binding to the enzyme. Thus, this experiment defines the net binding rate for all sites on the enzyme which is an important constraint, although this experiment does not provide information on the fraction that binds at each site (pol versus exo). Next, we used the tCo phosphorothioate 3’ mismatch substrate to measure the DNA dissociation rate (Figure 4B). The data best fit two exponentials, with rates of approximately 1.2 s-1 and 0.27 s-1, although interpretation of this experiment alone or with the binding data does not define the dissociation rate from each site. To include equilibrium constraints on the fitting, an equilibrium titration was performed with increasing concentrations of T7 DNA polymerase at a fixed concentration of tCo labeled DNA (Figure 4C). Compared to the ssDNA titration curve in Figure 2C, the 3’ mismatch dsDNA titration clearly shows that this substrate binds the T7 DNA polymerase with much higher affinity than ssDNA, and fitting to a quadratic equation estimates a net Kd of less than 1 nM. Because of the complex nature of the reaction with multiple DNA binding sites on the enzyme, even global fitting of the binding data was insufficient to produce a unique model. These experiments however provide important constraints on the rate constants for DNA binding and dissociation that, when fit with the excision data described below, is a prime example of a case where the whole is greater than the sum of the parts (

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ).
      Figure thumbnail gr4
      Figure 4Kinetics of binding and excision of a 3’ mismatch. DNA substrate: The DNA substrate used in the experiments shown in this figure is given at the top. For experiments with FAM-labeled DNA, the DNA primer strand contained a 5’-[6-FAM] label and the tCo was a substituted with C in the primer. Solid lines through the data in this figure show the best fits by simulation using the kinetic scheme at the bottom of this figure with parameters listed in . For each experiment we quote values obtained by equation-based data fitting, but we show only the curves derived by global data fitting. Experiments in (A) through (C) were performed to measure DNA binding using hydrolysis-resistant phosphorothioate (Sp isomer) 3’-terminal mismatch DNA. A) Stopped-flow 3’ mismatch DNA binding rate experiment. A solution of 25 nM FAM-labeled phosphorothioate 3-terminal mismatch DNA (FAM-27-PThio/45-3’MM-PThio) (12.5 mM Mg2+) was mixed with varying concentrations of T7 DNA polymerase (100-400 nM) and 2 μM thioredoxin to start the reaction. Measuring fluorescence anisotropy gives the net binding rate for all DNA binding sites on the enzyme and this experiment estimates a net second order binding rate constant around 1,000 μM-1 s-1. B) Stopped flow 3’ mismatch DNA off-rate experiment. Fluorescence was monitored after mixing a solution of 25 nM enzyme-DNA complex (tCo phosphorothioate DNA with a 3’-terminal mismatch (tCo-27-PThio/45-3’MM-PThio), 30 nM T7 DNA polymerase, 1 μM thioredoxin, 12.5 mM Mg2+) with 500 nM unlabeled trap DNA (phosphorothioate DNA with a 3’-terminal mismatch, 27-PThio/45-3’MM-PThio). The data best fit two exponentials with rates of 1.16 ± 0.02 s-1 and 0.270 ± 0.003 s-1 and 65% of the total amplitude occurred during the fast phase. C) Equilibrium DNA binding titration experiment. A solution of 20 nM tCo 3’ mismatch phosphorothioate DNA (tCo-27-PThio/45-3’MM-PThio) and 2 μM thioredoxin in a cuvette was titrated with T7 DNA polymerase with constant mixing from a micro star bar over the course of 5 minutes. The data were corrected for the small dilution and inverse inner filter effect and the corrected data are shown. Conventional fitting of the experiment to a quadratic equation estimates a Kd for DNA binding of 0.7 ± 0.1 nM. D) Rapid-quench 3’ mismatch DNA binding and excision experiment. A solution of 150 nM 32P-tCo 3’ mismatch DNA (32P-tCo-27/45-3’MM) (12.5 mM Mg2+) was mixed with 1 μM T7 DNA polymerase, 10 μM thioredoxin, and 0.1 mg/ml BSA to start the reaction. Aliquots were quenched with EDTA at various times, products were resolved by denaturing PAGE, and the gel was visualized by phosphorimaging (shown in (I)). The concentration of 27nt primer is shown in red and concentration of products less than or equal to 26nt are shown in green. The data best fit 3 exponentials: a fast phase at > 200 s-1 with an amplitude of 19 ± 2 nM, a second phase at 9 ± 5 s-1 with an amplitude of 26 ± 17 nM, and a third phase at 2.0 ± 0.6 s-1 with an amplitude of 52 ± 17 nM. E) Rapid-quench Mg2+ addition to pre-incubated E-DNA complex experiment. A solution of 1 μM T7 DNA polymerase, 75 nM 32P-tCo 3’ mismatch DNA (32P-tCo-27/45-3’MM), 10 μM thioredoxin, and 0.1 mg/ml BSA was mixed with 12.5 mM Mg2+ to start the reaction and samples were processed as in (D). The data best fit 3 exponentials: a fast phase at 168 ± 59 s-1 with an amplitude of 7.1 ± 1.2 nM, a second phase at 7 ± 2 s-1 with an amplitude of 22 ± 6 nM, and a third phase at 1.53 ± 0.46 s-1 with an amplitude of 21 ± 6 nM. F) Stopped-flow Mg2+ addition to preincubated E-DNA complex experiment. A solution of 100 nM T7 DNA polymerase, 2 μM thioredoxin, and 40 nM tCo-3’ mismatch DNA (tCo-27/45-3’MM) was mixed with 12.5 mM Mg2+ to start the reaction. The data best fit two exponentials, with the fast phase having 88% of the total amplitude at a rate of 7.24 ± 0.04 s-1 and the slow phase having a rate of 1.86 ± 0.05 s-1. G) Rapid quench 3’ mismatch excision/extension experiment. A solution of 75 nM 32P-tCo 3’ mismatch DNA (32P-tCo-27/45-3’MM), 400 nM T7 DNA polymerase, 8 μM thioredoxin, 0.1 mg/ml BSA, and 1 mM EDTA, was mixed with 12.5 mM Mg2+ and 1 μM unlabeled 3’ mismatch DNA as a trap, 100 μM dGTP, and 100 μM dATP to start the reaction. Aliquots were quenched with EDTA, processed as in (D) and (E)—the corresponding gel is shown in (J). Data corresponding to the 27, 26 and 28nt products are colored in red, green and blue, respectively. Data for the 27nt DNA best fit a 2-exponential function with rates of 8.3 ± 2.2 s-1 and 2.6 ± 0.73 s-1 with amplitudes of 30 ± 13 nM and 32 ± 13 nM for the fast and slow phases, respectively. H) Stopped flow 3’ mismatch excision then polymerization experiment. A solution of 40 nM tCo 3’ mismatch DNA (tCo-27/45-3’MM), 100 nM T7 DNA polymerase, 2 μM thioredoxin, and 1 mM EDTA, was mixed with 12.5 mM Mg2+, 1 μM unlabeled 3’ mismatch DNA, 100 μM dGTP, and 100 μM dATP to start the reaction. Fitting the data to two exponentials gives a fast phase with 91% of the total amplitude at 7.63 ± 0.04 s-1 and the slow phase at an observed rate of 0.75 ± 0.04 s-1. I) Gel for rapid-quench 3’ mismatch DNA binding and excision experiment. Gel from the experiment in (D). The black vertical line indicates where two gel images were spliced together to make the continuous gel figure. DNA product lengths are labeled to the right of the gel. J) Gels for rapid quench 3’ mismatch excision/extension experiment. Gels from experiment in (G). DNA product lengths are labeled to the right of the gel. The space between the lanes indicates that samples were run on two different gels and combined here for the figure. Scheme: Kinetic model for 3’ mismatch excision by T7 DNA polymerase. Solid arrows indicate rate constants defined by the experiments without dNTPs present. The dashed lines represent the rate constants that change when dNTPs are added to the reaction. The rate of chemistry (k6) was locked at 5,000 s-1 based on results for ssDNA excision.
      Next, we performed rapid-quench and stopped-flow assays on the hydrolyzable substrate containing a 3’ mismatch to accompany the binding data from the previous section. In the first experiment, we measured the rate of mismatch excision upon mixing the enzyme and DNA to start the reaction in the quench-flow (Figure 4D). To ensure continuity between the stopped-flow and rapid-quench experiments, we opted to label the tCo dsDNA substrate with a 5’-32P to use in the rapid-quench experiments. A solution of T7 DNA polymerase (consisting of the T7 gp5 catalytic subunit and the thioredoxin accessory protein) was mixed with 32P-tCo 3’ mismatch dsDNA (32P-tCo-27/45-3’MM) to start the reaction in the quench-flow. Samples were quenched at various times with EDTA and products were resolved by denaturing PAGE (Figure 4I). Preliminary inspection of the gel shows that the enzyme removes the terminal mismatch relatively quickly, then much more slowly removes properly base paired bases in the DNA duplex. Figure 4D shows the concentration of the different DNA species versus time for this experiment with best fits to three exponentials. Previous studies on this enzyme (
      • Donlin M.J.
      • Patel S.S.
      • Johnson K.A.
      Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
      ) only showed two exponentials for hydrolysis of similar substrates. Here, fitting the data by simulation afforded a rationale for three kinetically significant steps and provided a justification for the additional phase. The data best fit a model with three DNA binding sites on the enzyme. The high enzyme concentration used in our experiment (1 μM) relative to the concentration used previously (300 nM) (
      • Donlin M.J.
      • Patel S.S.
      • Johnson K.A.
      Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
      ) makes the second order rate constant for binding directly at the exonuclease active site faster (net DNA binding rate is known from anisotropy binding experiment), revealing the fraction that binds directly to the exonuclease active site and is rapidly hydrolyzed. The slightly slower phase defines DNA binding to an intermediate partially melted state on the enzyme that is then transferred to the exonuclease active site for hydrolysis. The slowest phase represents the fraction of DNA that binds to the polymerase active site, slowly dissociates, and rebinds at one of the other two sites on the enzyme before being hydrolyzed. The smooth curves in Figure 4 show the global data fitting according to the scheme and rate constants shown at the bottom of the figure. The flux calculation feature in KinTek Explorer, derived from partial derivatives computed by numerical integration, was used to determine the fraction of DNA that binds from solution at each site for this substrate (Figure S2), showing that approximately 40% of the DNA binds at the polymerase active site, 47% binds at the intermediate site, and 12% binds directly to the exonuclease active site.
      Next, a similar rapid-quench experiment was performed, except the enzyme and DNA were preincubated in the absence of Mg2+ and then Mg2+ was then added to start the reaction. Therefore, the DNA pre-equilibrates between the different enzyme sites before the reaction is initiated with Mg2+. The results of this experiment are shown in Figure 4E. The data are almost identical to the experiment in Figure 4D and fitting by simulation to the model in Figure 4 estimates that approximately 34%, 57%, and 8% of the DNA is in the polymerase, intermediate, and exonuclease active sites, respectively at equilibrium in the absence of Mg2+. These fractions are similar to the fraction that binds from solution at each site in the presence of Mg2+ and provide important constraints to interpret stopped-flow fluorescence signals where it is impossible to accurately determine the fraction in each site without this information. With these constraints, we performed the same Mg2+ addition to the pre-incubated E-DNA complex experiment in the stopped-flow to monitor changes in hybridization state of the DNA upon Mg2+ addition (Figure 4F). The decrease in fluorescence arises from the transfer of the trimmed primer strand back to the polymerase active site after mismatch excision. Conventional data fitting to equations estimates an observed rate of ∼ 7 s-1 for this step but fitting by simulation reveals that the observed rate is limited by transfer to the exonuclease active site before hydrolysis and the rate constant for transfer to the polymerase active site after mismatch hydrolysis (k7 in the scheme in Figure 4) is approximately 60 s-1.
      Since the scaling factors in fluorescence studies are difficult to interpret without additional information, we performed further experiments where mismatches were excised in the presence of added nucleotides to allow extension of the primer after the mismatch is removed (Figures 4G, H, J). After the mismatch is removed at the exonuclease active site, the primer must transfer back to the polymerase active site before additional nucleotides can be added. Because polymerization is much faster than strand transfer, any fluorescence signal that correlated with the rate of primer extension could be attributed to the DNA transfer step. First, we used rapid-quench methods to monitor the progress of the reaction, where the enzyme-32P-tCo 3’ mismatch dsDNA complex was mixed with Mg2+, dGTP, dATP, and an excess of unlabeled 3’ mismatch DNA as a trap to start the reaction (Figure 4G). Time points were analyzed as before and the resulting gel is shown in Figure 4J. As shown in Figure 4G, the 26 nt intermediate does not accumulate. Rather, the band for the 26 nt intermediate forms and decays rapidly as the enzyme extends the 26 nt primer by 2 bases to form the 28 nt product. These data support the notion that transfer from the exonuclease active site to the polymerase active site after mismatch excision is fast (∼60 s-1) relative to the rate of excision.
      Since a trap (unlabeled DNA) was included in the reaction, all 28 nt product that is formed results from excision of the mismatch, transfer to the pol site, and extension without dissociation from the enzyme. Note that the amplitude of product formation in the mismatch excision/extension experiment (Figure 4G) is higher than the excision only experiments (Figures 4D and E), even though a trap was included in the excision/extension experiment.
      We will later show that the nucleotide-dependent increase in amplitude is even more pronounced for the buried mismatch substrate. To account for these observations based on the same base model as for the experiments without nucleotide, the model required modification to include a nucleotide-dependent transfer from the EDp to the EDi state (k4,act in the scheme in Figure 4), and a nucleotide dependent increase in the rate of transfer from EDi to EDx (k5,act in the scheme in Figure 4). For this substrate, it is interesting that without dNTPs there is no significant transfer from EDp to EDi, but in the presence of dNTPs, the transfer occurs readily. Finally, we repeated the same experiment in the stopped-flow to monitor the rate of transfer from the exonuclease active site to the polymerase active site after mismatch excision (Figure 4H). Conventional analysis of the data in this figure shows that the observed rates are similar to those in the experiment in Figure 4F, albeit slightly faster. Furthermore, based on our model, the slow linear phase represents the small fraction of DNA that was bound to the EDp state and is slowly transferred to the exonuclease active site in the presence of dNTPs.
      Global data fitting by simulation in KinTek Explorer was essential to develop a model that accounted for all the experimental data shown in Figure 4. Due to the large number of parameters in the complex model, it was critical to perform confidence contour analysis with the FitSpace function of KinTek Explorer to determine whether each parameter was supported by the data—the results are shown in Figure 5. Parabolic curves indicate that each parameter is well constrained by the data, including the “activated” steps where some rate constants are stimulated in the presence of dNTPs. The main features of the model include 3 binding sites for the DNA on the enzyme, one site being the exonuclease active site, another being the polymerase active site, and the third being an intermediate with partially melted DNA. The identity of the intermediate state is unknown but the fluorescence data show that it has a higher fluorescence than complexes at the polymerase active site, suggesting it is at least partially melted. Our model also includes the fast rate of chemistry (∼5,000 s-1) estimated from the ssDNA experiments. Finally, in the presence of nucleotides, the kinetics of transfer between the sites changes. In the absence of dNTPs, there is no evidence for direct transfer from the polymerase active site to the exonuclease active site. However, in the presence of nucleotides data fitting requires a transfer step from the polymerase to the intermediate state and a nucleotide-dependent stimulation of the rate of transfer from the intermediate state to the exonuclease active site. Unfortunately, we cannot address whether the nucleotide-dependent transfer rate requires proper base pairing with the template strand because our method of measurement requires fast template-dependent extension of the 26 nt product of mismatch excision. With a model for excision of the 3’ mismatch substrate, we next investigate the kinetics of excision on a mismatch buried by a correct base pair.
      Figure thumbnail gr5
      Figure 5Confidence contours for global fit of 3’ mismatch DNA. Confidence contours were calculated with the FitSpace feature of KinTek Explorer. Note that k6 was locked at 5,000 s-1 in the fitting. A 0.995% confidence threshold, shown as the dashed grey line, represents the 95% confidence interval, based on the number of data points and the number of parameters in the fitting, as calculated by the software. Best fit parameters are given in .

      Kinetics of buried mismatch excision

      In our previous paper on substrate specificity for the exonuclease proofreading domain of T7 DNA polymerase (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ), we showed that buried mismatches are excised much more rapidly than single 3’ mismatches. This reaction provides an additional mechanism for error correction by first extending the mismatch with the correct base before highly efficient excision of two nucleotides including the mismatch and the added nucleotide. Here we investigate the detailed kinetics governing how the buried mismatch is more efficiently hydrolyzed than the single 3’ mismatch substrate. As for the other substrates investigated in this paper, we begin with binding measurements using the hydrolysis-resistant phosphorothioate oligonucleotide substrates. First, we measured the net DNA binding rate using fluorescence anisotropy by stopped-flow after mixing a solution of FAM phosphorothioate buried mismatch DNA (FAM-27-PThio/45-BMM-PThio) with varying concentrations of T7 DNA polymerase and thioredoxin (Figure 6A). As for the 3’ mismatch substrate, the net DNA binding rate was fast, approaching a second-order rate constant of 1,000 μM-1 s-1. All traces approach the same endpoint, indicating tight binding relative to the fixed DNA concentration (25 nM). As above, this experiment does not define individual rate constants by itself because it represents the sum of binding to two or three enzyme sites but provides important constraints on the net binding rate for the sum of all DNA binding sites on the enzyme. We next measured the DNA dissociation rate in the stopped-flow by mixing a pre-incubated solution of tCo phosphorothioate buried mismatch DNA (tCo-27-PThio/45-BMM-PThio) and T7 DNA polymerase with a large excess of unlabeled phosphorothioate buried mismatch DNA (27-PThio/45-BMM-PThio) as a trap for free enzyme that has dissociated from the labeled DNA. This experiment provides important constraints that become meaningful when globally fit with the other experiments in Figure 6. We next performed an equilibrium titration experiment by titrating the T7 DNA polymerase complex into a solution of tCo phosphorothioate buried mismatch DNA over the course of 5 minutes with constant mixing from a micro stir bar—the results are shown in Figure 6C. Fitting the titration curve to a quadratic equation provides a net Kd of approximately 0.5 nM, providing an additional equilibrium constraint on the binding parameters. Also, this experiment demonstrates that even though the DNA contains mismatches, the Kd is similar to the Kd for DNA containing a single mismatch, suggesting the binding is not weaker; however, the partitioning of the DNA between sites is different, as described below.
      Figure thumbnail gr6
      Figure 6Buried mismatch binding and excision kinetics. DNA Substrate. The DNA substrate used in these experiments is shown at the top of the figure. For experiments with FAM-labeled DNA, the location given for tCo in the substrate was changed to a C and the primer contained a 5’-6-FAM label. Experiments in (A) through (C) were performed with hydrolysis-resistant phosphorothioate (Sp isomer) DNA to measure binding. In all experiments, the solid line through the data shows the best global fit by simulation in KinTek Explorer using the model at the bottom of the figure. A) Stopped-flow buried mismatch DNA binding-rate measurement. A solution of 25 nM FAM-buried mismatch phosphorothioate DNA (FAM-27-PThio/45-BMM-PThio) and 12.5 mM Mg2+ was mixed with 2 μM thioredoxin and varying concentrations of T7 DNA polymerase (100 – 400 nM) to start the reaction. The data fitting provided an estimate a net second order DNA binding rate constant of approximately 1,000 μM-1 s-1. B) Stopped flow buried mismatch DNA off-rate experiment. A solution of 25 nM tCo-buried mismatch phosphorothioate DNA (tCo-27-PThio/45-BMM-PThio), 30 nM T7 DNA polymerase, 600 nM thioredoxin, and 12.5 mM Mg2+ was mixed with 500 nM unlabeled buried mismatch phosphorothioate DNA (27-PThio/45-BMM-PThio) to start the reaction. The data fit two exponentials with rates of 0.60 ± 0.02 s-1 and 0.25 ± 0.01 s-1 with 60% of the total amplitude corresponding to the fast phase. C) Equilibrium buried mismatch DNA binding titration. A solution of 25 nM tCo buried mismatch phosphorothioate DNA (tCo-27-PThio/45-BMM-PThio), 2 μM thioredoxin, and 12.5 mM Mg2+ in a cuvette was titrated with T7 DNA polymerase with constant mixing from a micro stir bar. The signal was corrected for the small dilution and the inverse inner filter effect and the corrected data are shown. Fitting the titration curve to a quadratic equation estimates a Kd of approximately 0.5 nM. D) Rapid quench buried mismatch binding and excision experiment. A solution of 150 nM 32P-tCo buried mismatch DNA (32P-tCo-27/45-BMM) and 12.5 mM Mg2+ was mixed with 1 μM T7 DNA polymerase, 10 μM thioredoxin, and 0.1 mg/ml BSA to start the reaction. Time points were quenched with EDTA, separated by denaturing PAGE, and visualized by phosphorimaging. The gel is shown in (I). Products of 27, 26, and ≤ 25 nt in length are colored in red, green, and blue, respectively. The data for the loss of 27 nt DNA best fit two exponentials with a fast phase with an amplitude of 45 ± 3 nM at 350 ± 115 s-1, and a slow phase with an amplitude of 105 ± 4 nM at a rate of 7.3 ± 0.6 s-1. E) Rapid quench Mg2+ addition to pre-incubated E-DNA complex. A solution of 75 nM 32P-tCo buried mismatch DNA (32P-tCo-27/45-BMM), 1 μM T7 DNA polymerase, 10 μM thioredoxin, 0.1 mg/ml BSA, and 1 mM EDTA, was mixed with 12.5 mM Mg2+ to start the reaction. Samples were processed as in (D). Products of 27, 26, and ≤ 25 nt in length are colored in red, green, and blue respectively. The data for the loss of the 27 nt DNA best fits a double exponential function. The fast phase has an amplitude of 12 ± 2 nM at a rate of 84 ± 31 s-1, a slower phase with an amplitude of 56 ± 2 nM and a rate of 3.5 ± 0.3 s-1. F) Stopped flow Mg2+ addition to pre-incubated E-DNA complex. A solution of 40 nM tCo-buried mismatch DNA (tCo-27/45-BMM), 100 nM T7 DNA polymerase, 2 μM thioredoxin , and 1 mM EDTA was mixed with 12.5 mM Mg2+ to start the reaction. The data best fit two exponentials with a fast phase with 82% of the total amplitude at 6.73 ± 0.03 s-1, and a slower phase at 0.35 ± 0.02 s-1. G) Rapid quench buried mismatch excision/extension experiment. A solution of 75 nM 32P-tCo buried mismatch DNA (32P-tCo-27/45-BMM), 400 nM T7 DNA polymerase, 8 μM thioredoxin, 0.1 mg/ml BSA, and 1 mM EDTA, was mixed with 1 μM unlabeled buried mismatch DNA as a trap, 100 μM dATP, 100 μM dCTP, 100 μM dTTP, and 12.5 mM Mg2+ to start the reaction. Samples were processed as in (D) and (E) and the gel is shown in (J). Products of 30, 27, 26, and ≤ 25 nt in length are shown in yellow, red, green, and blue, respectively. Data for the loss of the 27 nt DNA best fit 2 exponentials with a fast phase with an amplitude of 18 ± 4 nM at a rate of 19 ± 5 s-1 and a slower phase with a rate of 3.5 ± 0.3 s-1 and an amplitude of 48 ± 4 nM. H) Stopped flow buried mismatch excision/extension experiment. A solution of 25 nM tCo buried mismatch DNA (tCo-27/45-BMM), 40 nM T7 DNA polymerase, and 800 nM thioredoxin was mixed with 600 nM unlabeled buried mismatch DNA as a trap, 100 μM dATP, 100 μM dCTP, 100 μM dTTP, and 12.5 mM Mg2+ to start the reaction. The data best fit two exponentials with a fast phase at a rate of 6.66 ± 0.04 s-1 with 82% of the total amplitude, and a slower phase at 0.50 ± 0.03 s-1. I) Gel for rapid quench buried mismatch binding and excision experiment. Gel from the experiment in (D). DNA product lengths are labeled to the left and right of the gel. J) Gel for rapid-quench buried mismatch excision/extension experiment. Gel from the experiment in (G). DNA product lengths are labeled to the left and right of the gel. Scheme: Kinetic model for buried mismatch excision. Rate constants locked during the fitting are shown in red. Dashed dark blue arrows indicate states from which the DNA can dissociate from the enzyme. These dissociation rate constants were mostly locked at values obtained for the 3’ mismatch substrate.
      Next, we measured the kinetics of buried mismatch excision using experiments similar to those performed for the 3’ mismatch substrate. In the first rapid-quench experiment, a solution of T7 DNA polymerase was mixed with 32P-labeled tCo buried mismatch DNA (32P-tCo-27/45-BMM) to start the reaction and samples were collected and processed as described for the 3’ mismatch experiments and shown in Figure 6D. The data for the loss of the 27 nt starting material best fits at least 2 exponentials, with a very fast phase that corresponds to the fraction of DNA that binds directly to the exonuclease active site from solution and is rapidly hydrolyzed, followed by a slower phase that likely represents transfer of the primer strand from another site on the enzyme to the exonuclease active site. Relative to the 3’ mismatch substrate, the amplitude of the fast phase is greater and the rate of the slower phase is faster, supporting the model where the buried mismatch has more single stranded character and favors partitioning and transfer into the exonuclease active site. The gel for this experiment is shown in Figure 6I and shows that the enzyme removes the terminal two bases of the primer strand then pauses and very slowly excises the remaining correctly base paired nucleotides of the primer strand. Flux calculations (Figure S3) estimate that approximately 2%, 64%, and 34% of the DNA binds from solution at the polymerase active site, intermediate site, and exonuclease active site, respectively. Next, we performed the experiment where 32P-tCo buried mismatch DNA, preincubated with T7 DNA polymerase, was mixed with Mg2+ to start the reaction in the quench-flow (Figure 4E). In this experiment, global data fitting indicates that when preincubated in the absence of Mg2+, approximately 27%, 64%, and 9% of the DNA occupies the polymerase active site, intermediate site, and exonuclease active site, respectively. In our preliminary studies, we discovered that Mg2+ has a large effect on the binding and partitioning kinetics for various substrates and the fractions in each site in the absence of Mg2+ versus the fraction that binds from solution for this experiment varies greatly (data not shown). Because Mg2+ has large effects on DNA binding (
      • Gong S.
      • Kirmizialtin S.
      • Chang A.
      • Mayfield J.E.
      • Zhang Y.J.
      • Johnson K.A.
      Kinetic and thermodynamic analysis defines roles for two metal ions in DNA polymerase specificity and catalysis.
      ), the fraction of DNA in each site in the absence of Mg2+ may not be physiologically relevant. However, this experiment provides important constraints for the next stopped-flow experiment that provides additional information on the rates of transfer between sites. To monitor the kinetics of transfer to the polymerase active site after Mg2+ addition and mismatch excision, we performed a stopped-flow experiment where a solution of tCo buried mismatch DNA (tCo-27/45-BMM) and T7 DNA polymerase was mixed with Mg2+ to start the reaction (Figure 6F). The data are biphasic, with 82% of the total amplitude at 6.73 ± 0.03 s-1, and a slower phase observed at 0.35 ± 0.02 s-1
      To provide additional data to interpret the fluorescence signal from the tCo label, we performed excision/extension experiments, as were performed above for the 3’ mismatch substrate. First, we performed an experiment by rapid-quench to define the kinetics of DNA excision and extension (Figure 6G). A solution of T7 DNA polymerase and 32P-tCo buried mismatch DNA was mixed with Mg2+, dATP, dCTP, dTTP, and an excess of unlabeled buried mismatch DNA as a trap to start the reaction. Samples were process as described above and the resulting gel is shown in Figure 6J. From the gel, we see rapid excision of the terminal base, but little accumulation of the 25 nt intermediate, indicating that as soon as the 3’-terminal base is excised, the primer partitions back into the polymerase active site and nucleotides are rapidly added to the primer strand to form the 30 nt product. When comparing this experiment in Figure 6G to the similar experiment in Figure 6E, it is clear that the excision kinetics change significantly in the presence of dNTPs. Since the experiment was performed in the presence of excess unlabeled DNA, all observed excision kinetics occur within a single binding event. When looking at the rate of loss of the starting 27 nt primer, the rate is visibly faster than the corresponding curve in Figure 6E. Furthermore, the 26 nt product is much more rapidly hydrolyzed to the 25 nt product, whereas without dNTPs, the 26 nt product is slowly trimmed to the 25 nt product. We then performed a similar experiment in the stopped-flow by mixing a solution of T7 DNA polymerase and tCo buried mismatch DNA with Mg2+, dATP, dCTP, dTTP and an excess of unlabeled buried mismatch DNA as a trap to start the reaction (Figure 6H). The observed rate is slightly faster than the experiment in Figure 6F without dNTPs; however, the rate of transfer from the exonuclease to polymerase active site determined from data fitting does not appear to change.
      As for the other substrates, global data fitting by simulation was crucial to making sense of the data and is another example of a case where the whole is greater than the sum of the parts. The final model used for data fitting is shown in the scheme in Figure 6. There can be no question that the model accounts for the data based on the goodness of fit. The remaining question is whether the model is too complex, which we address by confidence contour analysis to assess the extent to which each parameter is well-constrained by the data. Confidence contours in Figure 7 demonstrate that fitted parameters are well constrained by the data with well-defined upper and lower limits, except for two parameters which only have lower or upper limits. It is understandable that there are no data to define the rate of dNMP release (k11 > 100 s-1). Similarly, the data only define an upper limit for DNA binding to the pol site (k3 < 6.7 μM-1s-1). In addition, several rate constants were locked, based on preliminary fitting attempts revealing that only a lower limit could be set for each of these constants, which are given in red text in the scheme in Figure 6 and in Table 4. Despite the complexity of the model, the quality of the global fit and the significant number of well-constrained parameters gives confidence in the validity of the model to represent the reactions governing DNA partitioning between exo and pol sites.
      Figure thumbnail gr7
      Figure 7Confidence contours for buried mismatch data. Confidence contours were calculated with the FitSpace feature of KinTek Explorer. A threshold of 0.99 in the normalized χ2 ratio, shown as the dashed grey line, represents the 95% confidence interval, calculated using the F-distribution based on the number of data points and the number of parameters in the fitting as calculated by the software.
      Table 3Kinetic Parameters for Mismatched DNA
      Rate constants derived in Figure 4 are summarized with confidence intervals derived by contour analysis. Because the contours are symmetrical, we can estimate the standard errors directly from the confidence intervals, for example, k1 = 114 ± 12 μM-1 s-1 with lower and upper limits of 101 and 125. The value of k6 = 5000 s-1 (shown in red text) for the rate of hydrolysis of the 3’-terminal nucleotide was locked at a lower limit sufficient to account for the observed rates.
      ParameterBest-Fit95% Confidence Interval
      k1114 μM-1 s-1(101, 125)
      k-10.372 s-1(0.337, 0.402)
      k2451 μM-1 s-1(426, 477)
      k-22.44 s-1(2.31, 2.58)
      k3382 μM-1 s-1(362, 402)
      k-30.231 s-1(0.228, 0.234)
      k54.87 s-1(4.79, 4.95)
      k-52.94 s-1(2.68, 3.27)
      k65,000 s-1Locked
      k759.6 s-1(55.7, 63.8)
      k4,act2.81 s-1(2.57, 3.07)
      k5,act7.83 s-1(7.63, 8.04)
      a Rate constants derived in Figure 4 are summarized with confidence intervals derived by contour analysis. Because the contours are symmetrical, we can estimate the standard errors directly from the confidence intervals, for example, k1 = 114 ± 12 μM-1 s-1 with lower and upper limits of 101 and 125. The value of k6 = 5000 s-1 (shown in red text) for the rate of hydrolysis of the 3’-terminal nucleotide was locked at a lower limit sufficient to account for the observed rates.
      Table 4Buried mismatch kinetic parameters for buried mismatch excision
      ParameterBest-Fit95% Confidence Interval
      k1278 μM-1 s-1268, 285
      k-10.193 s-10.186, 0.196
      k2701 μM-1 s-1686, 714
      k-26.23 s-15.65, 6.5
      k36.66 μM-1 s-1< 6.66
      k-30.2 s-1Locked
      k41.48 s-11.38, 1.51
      k-40.722 s-10.627, 0.777
      k54.56 s-14.46, 4.71
      k-50.357 s-10.342, 0.392
      k65,000 s-1Locked
      k743.1 s-141.9, 44.5
      k84.93 s-14.49, 5.16
      k-816,200 s-115,100, 16,700
      k90.372 s-1Locked
      k102.44 s-1Locked
      k11101 s-1> 101
      k1260 s-1Locked
      k134.74 s-14.46, 5.21
      k140.2 s-1Locked
      k4,act1.5 s-11.44, 1.57
      k5,act7.08 s-16.99, 7.15
      k8,act43.5 s-140.4, 44.6
      aRate constants derived from fitting the data in Figure 6 are summarized with confidence intervals derived by contour analysis, with lower and upper limits given in parentheses. Parameters in red text were locked during final fitting base on analysis showing that these rate constants had only lower limits in preliminary data fitting.
      The main features of our model for the buried mismatch DNA substrate are as follows: Just like for the 3’ mismatch DNA substrate, the best fit model contains three DNA binding sites on the enzyme, the polymerase active site, the exonuclease active site, and a partially melted intermediate state. Unlike the 3’ mismatch substrate, transfer from the polymerase active site to the intermediate state readily occurs with this substrate, although only a small fraction binds directly into the polymerase active site. Interestingly, we had to add a rate limiting dNMP release step following hydrolysis of the first base, following fast excision at 5,000 s-1. In the absence of nucleotides, most of the DNA partitions out of the exonuclease active site before hydrolysis of the second base of the primer strand, as demonstrated by the slow rates of excision of bases after the first, shown in Figures 6D and E. In contrast, in the presence of nucleotides, the majority of the DNA partitions to remove the second base after excision of the first, and almost the full amplitude of the starting material is hydrolyzed and then extended in a single turnover (in the presence of a large excess of unlabeled DNA as a trap). Both in the presence and absence of nucleotides, dNMP release after the second base is hydrolyzed is fast, only having a lower limit in the confidence contour analysis. This is consistent with the data for the 3’ mismatch where the data show that chemistry and dNMP release appear to occur simultaneously.

      Discussion:

      The observed kinetics of DNA binding and excision are complex because the DNA can bind to either the exonuclease or polymerase sites and can partition between the two sites. In the exo site the 3’-terminal nucleotide of the DNA can be rapidly excised and then the 3’-terminus can return to the pol site for continued polymerization. After binding at the pol site the DNA can be extended by polymerization or slip into the exo site to be shortened. Any one measurement attempting to address the pathway of base excision is a complex function of all of the alternate pathways, and for this reason, no one experiment can be interpreted unambiguously. However, global data fitting can resolve a single unifying mechanism by including information from multiple experiments performed with different starting states and monitoring either fluorescence or chemical conversion steps (
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ,

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ). In these studies, each experiment provides unique information, but also lacks specific details the are filled in by parallel experiments. For example, it is often asserted that fluorescence signals are difficult to interpret because one does not know a priori what step in a reaction pathway produces a change in fluorescence and the scaling factors relating signal intensity to concentration are unknown. Analysis of the concentration dependence resolves signals associated with binding steps, while measurements of the rates of the chemical reaction can be included to aid in understanding observable fluorescence signals. Globally fitting all of the experiments resolves a single model to account for all of the available data, and confidence contour analysis provides quantitative data to address whether the model is fully supported by the data (
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ,

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ,
      • Johnson K.A.
      Fitting enzyme kinetic data with KinTek global kinetic explorer.
      ,
      • Johnson K.A.
      • Simpson Z.B.
      • Blom T.
      Global kinetic explorer: a new computer program for dynamic simulation and fitting of kinetic data.
      ).
      In this paper we have given a detailed kinetic analysis of proofreading for a high-fidelity DNA polymerase on three physiologically relevant substrates. We introduced a fluorescent DNA base analog tCo, which provided a large fluorescence signal upon DNA binding to the enzyme and from DNA melting required for binding in the exonuclease active site. Careful selection of the sequence surrounding tCo (
      • Sandin P.
      • Börjesson K.
      • Li H.
      • Mårtensson J.
      • Brown T.
      • Wilhelmsson L.M.
      • Albinsson B.
      Characterization and use of an unprecedentedly bright and structurally non-perturbing fluorescent DNA base analogue.
      ) and its position resulted in a DNA substrate that we expect to be much less perturbing of DNA structure than observed in similar studies using 2-aminopurine (
      • Bessman M.J.
      • Muzyczka N.
      • Goodman M.F.
      • Schnaar R.L.
      Studies on the biochemical basis of spontaneous mutation: II. The incorporation of a base and its analogue into DNA by wild-type, mutator and antimutator DNA polymerases.
      ,
      • Clayton L.K.
      • Goodman M.F.
      • Branscomb E.W.
      • Galas D.J.
      Error induction and correction by mutant and wild type T4 DNA polymerases. Kinetic error discrimination mechanisms.
      ,
      • Hopkins R.
      • Goodman M.F.
      Asymmetry in forming 2-aminopurine · hydroxymethylcytosine heteroduplexes; a model giving misincorporation frequencies and rounds of DNA replication from base-pair populations in vivo.
      ,
      • Sowers L.C.
      • Eritja R.
      • Chen F.M.
      • Khwaja T.
      • Kaplan B.E.
      • Goodman M.F.
      • Victor Fazakerley G.
      Characterization of the high pH wobble structure of the 2-aminopurine·cytosine mismatch by N-15 NMR spectroscopy.
      ,
      • Sowers L.C.
      • Fazakerley G.V.
      • Eritja R.
      • Kaplan B.E.
      • Goodman M.F.
      Base pairing and mutagenesis: observation of a protonated base pair between 2-aminopurine and cytosine in an oligonucleotide by proton NMR.
      ,
      • Watanabe S.M.
      • Goodman M.F.
      On the molecular basis of transition mutations: frequencies of forming 2-aminopurine.cytosine and adenine.cytosine base mispairs in vitro.
      ,
      • Pless R.C.
      • Bessman M.J.
      Influence of local nucleotide sequence on substitution of 2-aminopurine for adenine during deoxyribonucleic acid synthesis in vitro.
      ,
      • Bloom L.B.
      • Otto M.R.
      • Eritja R.
      • Reha-Krantz L.J.
      • Goodman M.F.
      • Beechem J.M.
      Pre-Steady-State Kinetic Analysis of Sequence-Dependent Nucleotide Excision by the 3'-Exonuclease Activity of Bacteriophage T4 DNA Polymerase.
      ). In some studies, we wanted measure DNA binding to the exonuclease site without complications from the subsequent chemistry step. Here there are three alternatives: mutate the enzyme active site, alter the substrate, or omit Mg2+, which is essential for catalysis. Experiments to restrict exonuclease activity by removing Mg2+ can be difficult to interpret because we know that Mg2+ plays an important role in shielding the negative charge on the DNA backbone to attenuate protein-DNA interactions. Accordingly, Mg2+ has been implicated in significantly changing the activity of many nucleic acid enzymes (
      • Bellamy S.R.W.
      • Kovacheva Y.S.
      • Zulkipli I.H.
      • Halford S.E.
      Differences between Ca2+ and Mg2+ in DNA binding and release by the SfiI restriction endonuclease: implications for DNA looping.
      ,
      • Soulière M.F.
      • Perreault J.-P.
      • Bisaillon M.
      Magnesium-binding studies reveal fundamental differences between closely related RNA triphosphatases.
      ,
      • Green N.H.
      • Williams P.M.
      • Wahab O.
      • Davies M.C.
      • Roberts C.J.
      • Tendler S.J.B.
      • Allen S.
      Single-Molecule Investigations of RNA Dissociation.
      ,
      • Guéroult M.
      • Picot D.
      • Abi-Ghanem J.
      • Hartmann B.
      • Baaden M.
      How Cations Can Assist DNase I in DNA Binding and Hydrolysis.
      ). For example, studies using HIV reverse transcriptase kinetics revealed that one Mg2+ ion enters the active site chelated by the incoming nucleotide, while the second Mg2+ binds after the nucleotide-induced conformational change to facilitate catalysis (
      • Gong S.
      • Kirmizialtin S.
      • Chang A.
      • Mayfield J.E.
      • Zhang Y.J.
      • Johnson K.A.
      Kinetic and thermodynamic analysis defines roles for two metal ions in DNA polymerase specificity and catalysis.
      ). This study also included MD simulations to illustrate the dynamic role of Mg2+ in electrostatic shielding the phosphates on the DNA backbone with local concentrations of ∼4 M. For T4 DNA polymerase, a study showed that Mg2+ was a key thermodynamic switch and that in the absence of Mg2+ a hydrophobic exonuclease site is preferred over the polymerase site for binding the primer terminus (
      • Beechem J.M.
      • Otto M.R.
      • Bloom L.B.
      • Eritja R.
      • Reha-Krantz L.J.
      • Goodman M.F.
      Exonuclease−Polymerase Active Site Partitioning of Primer−Template DNA Strands and Equilibrium Mg2+ Binding Properties of Bacteriophage T4 DNA Polymerase.
      ). In contrast, exonuclease-deficient mutants and polymerase switching mutants shift this equilibrium toward the polymerase active site. For these reasons, it was important for us to avoid using active site mutants in T7 DNA polymerase to study oligo binding the presence of Mg2+ or to chelate Mg2+ to allow binding but not chemistry. We opted to create hydrolysis-resistant substrates by developing methods to enzymatically degrade the Rp isomer of a racemic mixture of phosphorothioate oligonucleotides and then purifying the hydrolysis-resistant Sp isomer. These sulfur containing oligos have been useful for mechanistic studies on other enzymes as well as in biotechnology (
      • Brautigam C.A.
      • Steitz T.A.
      Structural principles for the inhibition of the 3′-5′ exonuclease activity of Escherichia coli DNA polymerase I by phosphorothioates 11Edited by R. Huber.
      ,
      • Brautigam C.A.
      • Sun S.
      • Piccirilli J.A.
      • Steitz T.A.
      Structures of Normal Single-Stranded DNA and Deoxyribo-3‘-S-phosphorothiolates Bound to the 3‘-5‘ Exonucleolytic Active Site of DNA Polymerase I from Escherichia coli.
      ,
      • Stec W.J.
      • Zon G.
      • Egan W.
      Automated solid-phase synthesis, separation, and stereochemistry of phosphorothioate analogs of oligodeoxyribonucleotides.
      ,
      • Thorogood H.
      • Grasby J.A.
      • Connolly B.A.
      Influence of the Phosphate Backbone on the Recognition and Hydrolysis of DNA by the EcoRV Restriction Endonuclease: A STUDY USING OLIGODEOXYNUCLEOTIDE PHOSPHOROTHIOATES.
      ,
      • Grasby J.A.
      • Connolly B.A.
      Stereochemical outcome of the hydrolysis reaction catalyzed by the EcoRV restriction endonuclease.
      ). Here, they provide the ability to perform kinetic measurements to isolate the binding steps in the presence of Mg2+ without DNA cleavage. We believe that the phosphorothioate-containing substrates offer the best alternative to prevent chemistry when measuring binding to the exo site.
      We began our kinetic studies in this paper using a ssDNA substrate since it can bind directly into the exonuclease active site without first undergoing a melting step as with dsDNA (
      • Donlin M.J.
      • Patel S.S.
      • Johnson K.A.
      Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
      ). First, we measured binding by fluorescence anisotropy with FAM-labeled DNA using the hydrolysis-resistant phosphorothioate ssDNA to provide an estimate of the second order rate constant for ssDNA binding of 1,100 μM-1 s-1. While this second order rate constant is faster than some association constants measured for enzyme-nucleic binding (
      • Jia Y.
      • Kumar A.
      • Patel S.S.
      Equilibrium and Stopped-flow Kinetic Studies of Interaction between T7 RNA Polymerase and Its Promoters Measured by Protein and 2-Aminopurine Fluorescence Changes.
      ,
      • Ferreiro D.U.
      • de Prat-Gay G.
      A Protein–DNA Binding Mechanism Proceeds Through Multi-state or Two-state Parallel Pathways.
      ), it is within the diffusion limit (
      • Bar-Even A.
      • Noor E.
      • Savir Y.
      • Liebermeister W.
      • Davidi D.
      • Tawfik D.S.
      • Milo R.
      The Moderately Efficient Enzyme: Evolutionary and Physicochemical Trends Shaping Enzyme Parameters.
      ) and can be rationalized by considering the high concentration of negative charge in the DNA backbone and the highly positive DNA binding clefts on the polymerase (
      • Ferreiro D.U.
      • de Prat-Gay G.
      A Protein–DNA Binding Mechanism Proceeds Through Multi-state or Two-state Parallel Pathways.
      ,
      • Nickell C.
      • Prince M.A.
      • Lloyd R.S.
      Consequences of molecular engineering-enhanced DNA binding in a DNA repair enzyme.
      ). For comparison, the second order rate constant for the binding of ssDNA to the single strand DNA binding protein ranges from 1000 to 2000 μM-1 s-1 for E. coli and the human mitochondrial proteins, respectively (
      • Kozlov A.G.
      • Lohman T.M.
      Stopped-flow studies of the kinetics of single-stranded DNA binding and wrapping around the Escherichia coli SSB tetramer.
      ,
      • Qian Y.
      • Johnson K.A.
      The human mitochondrial single-stranded DNA-binding protein displays distinct kinetics and thermodynamics of DNA binding and exchange.
      ). We then measured the dissociation rate by competition to provide an estimate of 44 s-1, resulting in a Kd for ssDNA binding of 40 nM (Figure 2, Table 2). Subsequent studies with normal (no phosphorothioate) DNA supported the fast-binding rate and additionally set a lower limit of 5,000 s-1 for the rate of the chemistry step, to account for the observed rate that appeared coincident with by the rate of diffusion-limited ssDNA binding. The lower limit for the rate of chemistry was set by global data fitting and confidence contour analysis of the DNA cleavage kinetics (Figures 2 and 3). Because the observed rates of ssDNA cleavage are orders of magnitude faster than seen with duplex DNA, our kinetic analysis with ssDNA is used to define the rate of cleavage of a ssDNA segment melted from duplex DNA after it enters the exonuclease active site. Analysis of the forward excision reaction in the presence of dCMP product indicated that the reaction is almost completely irreversible due to weak binding (and fast release) of the dCMP product at the exonuclease active site but afforded an estimate of the equilibrium for the chemistry step at the active site (K2 = 3) favoring product formation. The majority of inhibition observed in this experiment (Figure 2F) came from dCMP competing with DNA binding at the exonuclease active site of the free enzyme, which was resolved in our global data fitting.
      Next, we examined the detailed kinetics of hydrolysis of a single 3’ mismatched DNA substrate, which would form if the polymerase made a mistake (at the polymerase active site) during DNA replication. For an exonuclease deficient T7 DNA polymerase variant, we previously estimated misincorporation to occur somewhere on the order of 1 in a million nucleotides (
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ,
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Conformational dynamics during misincorporation and mismatch extension defined using a DNA polymerase with a fluorescent artificial amino acid.
      ) which is in line with estimates for other high fidelity DNA polymerases (
      • Dodd T.
      • Botto M.
      • Paul F.
      • Fernandez-Leiro R.
      • Lamers Meindert H.
      • Ivanov I.
      Polymerization and editing modes of a high-fidelity DNA polymerase are linked by a well-defined path.
      ). Outstanding questions in the literature ask how efficiently the mismatch is excised versus extended, whether the primer strand gets to the exonuclease active site by direct transfer from the polymerase active site or by dissociation from the enzyme and rebinding at the exonuclease active site, and, finally, whether the DNA directly partitions back to the polymerase active site after mismatch excision without dissociating from the enzyme (
      • Johnson A.A.
      • Johnson K.A.
      Exonuclease Proofreading by Human Mitochondrial DNA Polymerase.
      ,
      • Beechem J.M.
      • Otto M.R.
      • Bloom L.B.
      • Eritja R.
      • Reha-Krantz L.J.
      • Goodman M.F.
      Exonuclease−Polymerase Active Site Partitioning of Primer−Template DNA Strands and Equilibrium Mg2+ Binding Properties of Bacteriophage T4 DNA Polymerase.
      ,
      • Bloom L.B.
      • Otto M.R.
      • Eritja R.
      • Reha-Krantz L.J.
      • Goodman M.F.
      • Beechem J.M.
      Pre-Steady-State Kinetic Analysis of Sequence-Dependent Nucleotide Excision by the 3'-Exonuclease Activity of Bacteriophage T4 DNA Polymerase.
      ,
      • Brutlag D.
      • Kornberg A.
      Enzymatic Synthesis of Deoxyribonucleic Acid: XXXVI. A PROOFREADING FUNCTION FOR THE 3′ → 5′ EXONUCLEASE ACTIVITY IN DEOXYRIBONUCLEIC ACID POLYMERASES.
      ,
      • Wang J.
      • Yu P.
      • Lin T.C.
      • Konigsberg W.H.
      • Steitz T.A.
      Crystal Structures of an NH2-Terminal Fragment of T4 DNA Polymerase and Its Complexes with Single-Stranded DNA and with Divalent Metal Ions.
      ). To address these questions, we measured the complete kinetic pathway for 3’ mismatch hydrolysis using both rapid-quench experiments and fluorescence experiments with DNA containing the fluorescent cytosine analog, tCo. All of the DNA binding and cleavage reactions are kinetically linked and cannot be deconvoluted by any single measurement. Therefore, global data fitting was essential to resolve individual steps in the pathway, which led to a model where the DNA can bind to three distinct sites on the enzyme (Figure 4, Table 3). First is the exonuclease active site where the 3-terminal nucleotide of mispaired DNA (or ssDNA) is rapidly hydrolyzed and exhibits a characteristic high fluorescence state in studies with tCo due to the melting of dsDNA to allow access into the exo active site. A second site is the polymerase active site where nucleotides can be rapidly added (300-500 s-1) and has a low fluorescence state in studies with tCo. The third is an intermediate site for which the structural identity is not known but has an intermediate fluorescence suggesting the DNA is at least partially melted and is an important kinetically defined intermediate in the partitioning of DNA between the exonuclease and polymerase active sites. Interestingly, in a study on exonuclease proofreading for T4 DNA polymerase the authors also suggested an intermediate state where the primer strand is partly denatured (
      • Baker R.P.
      • Reha-Krantz L.J.
      Identification of a transient excision intermediate at the crossroads between DNA polymerase extension and proofreading pathways.
      ); therefore, the presence of such an intermediate may be generally applicable to DNA polymerases with a proofreading function. The presence of an intermediate state is consistent with the large (>25 Å) distance between the pol and exo active sites, the required melting of at least three base-pairs, coupled to the rotation of the DNA duplex to reach from the pol to the exo site as implied by MD simulations (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ).
      We previously showed that after a mismatch is incorporated at the polymerase active site, the next correct base can be added on top of the mismatch, effectively burying the mismatch, although this occurs at a slow rate (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Conformational dynamics during misincorporation and mismatch extension defined using a DNA polymerase with a fluorescent artificial amino acid.
      ). However, further extension by the addition of a second nucleotide on top of the buried mismatch is severely inhibited, increasing the likelihood that the polymerase will correct the mistakes before continued DNA synthesis (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ). We also showed that buried mismatches are excised more efficiently than mismatches at the 3’-primer terminus. Faster excision of buried mismatched provides an additional opportunity to correct the mistake by first extending the mismatch and thereby enhancing the net DNA replication fidelity. A similar phenomenon has been shown for the human mitochondrial polymerase γ which removes buried mismatches faster than single 3’-terminal mismatches (
      • Johnson A.A.
      • Johnson K.A.
      Exonuclease Proofreading by Human Mitochondrial DNA Polymerase.
      ) so this observation is likely not unique to T7 DNA polymerase. We therefore performed detailed kinetic experiments to measure the rate constants governing buried mismatch excision by T7 DNA polymerase. The rate constants tell the story behind why the buried mismatch is more efficiently excised. When binding from solution, almost no buried mismatch substrate binds at the polymerase active site. In contrast, almost a third of DNA with a 3’ mismatch binds at the pol site, while for the buried mismatch the majority binds at the exo and intermediate sites. Once bound to the enzyme, the equilibrium constant for transfer of the buried mismatch primer from the pol site to intermediate state increases (K4 ∼ 2), and the equilibrium constant for the transfer from the intermediate state to the exonuclease active site is around 5 times greater than for the 3’ mismatch substrate. For both DNA substrates, after excision of the mismatch, our data show that the trimmed primer is transferred back to the polymerase active site rapidly (60 s-1), without dissociating from the enzyme.
      When we investigated the kinetics of excision in the presence of nucleotides to allow for polymerization following mismatch excision, we noticed that the rates and amplitudes of the curves changed showing that excision was even more efficient in the presence of added nucleotides. These experiments were performed in the presence of a trap for free enzyme (large excess of unlabeled DNA), indicating that the difference was not due to faster dissociation of the DNA and rebinding at the exonuclease active site, but rather direct transfer between sites on the enzyme without dissociation of the DNA from the enzyme. This phenomenon is not unique to T7 DNA polymerase as it has been shown for human DNA polymerase γ, where addition of nucleotide to the reaction that is identical to the 3’ base increased the rate of excision 7-fold. The authors proposed that the free nucleotide may enhance the rate of transfer of the DNA to the exonuclease active site by interrupting the correct 3’ base pairing through interaction with the template base (
      • Johnson A.A.
      • Johnson K.A.
      Exonuclease Proofreading by Human Mitochondrial DNA Polymerase.
      ). This phenomenon has also been shown for Klenow Fragment of E. coli DNA polymerase I where the authors used single molecule Fӧrster resonance energy transfer (FRET) to measure the kinetics of site switching. The authors showed that dNTPs specifically accelerate pol to exo switching of a mispaired DNA terminus using active site mutants to prevent hydrolysis (
      • Lamichhane R.
      • Berezhna S.Y.
      • Gill J.P.
      • Van der Schans E.
      • Millar D.P.
      Dynamics of Site Switching in DNA Polymerase.
      ), but Klenow has a very inefficient exonuclease reaction that may not relate to high fidelity polymerases.
      The structural basis for stimulation of the rate of transfer by free nucleotide remains unknown but poses an interesting question for structural or computational studies. Future studies could also be designed to measure the concentration dependence of nucleotide on the rates of switching to further understand the proofreading pathway for T7 DNA polymerase under physiological conditions or conditions of nucleotide starvation. To account for the different kinetics in the presence of nucleotides, we added “activation” steps to our pathways that occur in the presence of nucleotide. For the 3’ mismatch substrate, this had the effect of increasing the rate of transfer from the pol to intermediate site from close to 0 to over 2 s-1 and increasing the rate of transfer from the intermediate state to the exonuclease active site by almost a factor of 2. For the buried mismatch substrate, the already appreciable rate of transfer from the pol to intermediate site was increased slightly, along with the already fast rate of transfer from the intermediate site to the exo site. The bigger effect however was to greatly reduce the rate of dissociation following excision of the terminal base, resulting in much faster hydrolysis of the subsequent mismatched base followed by transfer back to the pol site for continued polymerization.
      With all the data combined, the answer to the question regarding direct pol to exo transfer versus dissociation and rebinding becomes more complicated. In the absence of nucleotides our data suggest that the preferred route is through dissociation and rebinding, at least for the 3’ mismatch substrate. However, in the presence of nucleotides, the direct transfer pathway becomes more efficient with over 90% going through the direct transfer pathway. For the buried mismatch, direct transfer to the exo site is the favored pathway in the absence of nucleotides, however the enzyme is most likely to remove only one base then dissociation from the DNA, rebind at the exo site, then remove the next base. The presence of nucleotides shifts the equilibrium after excision of the first base towards excision of the mismatched base allowing proofreading of both bases in a single binding event. For both substrates in the presence and absence of nucleotides, after mismatch excision the DNA is rapidly transferred back to the polymerase active site without dissociation. This can be rationalized structurally by favorable energetics of re-annealing of the melted primer strand with the template strand. The results presented in this paper define the kinetic pathway for proofreading for T7 DNA polymerase and are the overall themes are likely applicable to other high fidelity, replicative DNA polymerases.

      Experimental Procedures

      Enzymes, oligonucleotides, reagents

      Wild-type (x+) T7 DNA polymerase was reconstituted by combining the catalytic subunit (T7 gp5) with the E. coli thioredoxin accessory protein (
      • Patel S.S.
      • Wong I.
      • Johnson K.A.
      Pre-Steady-State Kinetic-Analysis of Processive Dna-Replication Including Complete Characterization of An Exonuclease-Deficient Mutant.
      ,
      • Akabayov B.
      • Akabayov S.R.
      • Lee S.J.
      • Tabor S.
      • Kulczyk A.W.
      • Richardson C.C.
      Conformational dynamics of bacteriophage T7 DNA polymerase and its processivity factor, Escherichia coli thioredoxin.
      ) and purified as previously described (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Conformational dynamics during misincorporation and mismatch extension defined using a DNA polymerase with a fluorescent artificial amino acid.
      ). Active polymerase was reconstituted using at least a twofold molar excess of thioredoxin relative to the polymerase. Bovine serum albumin (BSA), T4 polynucleotide kinase (T4 PNK), and dNTPs were purchased from New England Biolabs. Polyethyleneimine (PEI)-cellulose Thin Layer Chromatography (TLC) plates and dCMP were purchased from Sigma Aldrich. Before using TLC plates, plates were developed with dH2O until the water reached the top of the plate. The plates were dried, the contaminants at the top were cut off, and the plates were stored at 4°C until use. Oligonucleotides not containing tCo were purchased from Integrated DNA Technologies with standard desalting and then further purified by denaturing PAGE in house to achieve greater than 98% full length oligonucleotide, as determined by capillary electrophoresis. Oligonucleotides containing tCo were purchased from Bio-Synthesis corporation (www.biosyn.com) with PAGE purification. Phosphorothioate oligonucleotides were purchased as a racemic mixture of the two diastereomers and enzymatically processed to leave only the non-hydrolyzable Sp isomer (described below). All PAGE reagents and buffer components were purchased from either Fisher Scientific or Sigma Aldrich.

      Enzymatic synthesis of hydrolysis-resistant phosphorothioate oligos

      Hydrolysis-resistant oligonucleotides containing an Sp phosphorothioate linkage were prepared by mixing 0.5 μM T7 DNA polymerase, 0.1 mg/ml BSA, 25 μM phosphorothioate oligo, and 12.5 mM Mg2+ in T7 Reaction Buffer to start the reaction and the reaction was incubated at 20°C for 45 minutes. The reaction was then quenched by adding EDTA to a final concentration of 0.25 M. Processing of the reactions followed different protocols for oligonucleotides containing tCo and oligonucleotides without tCo.

      Purification of enzymatic degradation reactions for oligos containing tCo

      Proteins were inactivated and removed by incubating the quenched sample at 75°C for 20 minutes then filtering the sample through a 0.2 μm Whatman syringe filter. Buffer A-1 (25 mM Tris-HCl, pH 8.9) was added at a ratio of 2:1 Buffer A-1 to sample. Samples were loaded onto a BioRad Uno Q-1 column connected to an AKTA Pure FPLC system (GE Healthcare), equilibrated in 10% Buffer B-1 (Buffer A-1 with 1 M NaCl). Absorbance was monitored at 260 nm. The column was washed with 10 ml of 30% Buffer B-1, then bound oligos were eluted with a gradient from 30 – 60% Buffer B-1 over 40 ml, collecting 1 ml fractions. Fractions containing the full length non-hydrolyzable oligo were buffer exchanged into dH2O by ultrafiltration with GE Healthcare Vivaspin 20 centrifugal concentrators (5 kDa MWCO). The concentration of the purified, buffer exchanged oligonucleotides were determined by absorbance at 260 nm using the extinction coefficients given in Table 1 and stored at -20°C.

      Purification of enzymatic degradation reactions for oligos without tCo

      Proteins were inactivated by incubation at 85°C for 10 minutes and the precipitated protein was removed by filtration through a 0.22 μM syringe filter (Whatman). The sample was diluted 20-fold with Buffer A-2 (10 mM NaOH, pH 12) and loaded onto an Uno Q-1 anion exchange column (BioRad) equilibrated in Buffer A-2.The oligo was eluted with a gradient of 0 – 90% Buffer B-2 (Buffer A-2 with 1 M NaCl) over 40 ml. Fractions containing the purified hydrolysis-resistant oligonucleotide were neutralized, then extensively buffer exchanged into TE buffer on Vivaspin 20 centrifugal concentrators (GE Healthcare) before determining the concentration using the extinction coefficients at 260 nm in Table 1 and storage at -20°C. MALDI-MS was used to verify full length phosphorothioate DNA was obtained after the enzymatic degradation reaction using previously established methods (
      • Dangerfield T.L.
      • Kirmizialtin S.
      • Johnson K.A.
      Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
      ).

      32P Labeling of tCo oligonucleotide

      Oligonucleotides containing tCo were labeled with 32P for rapid-quench assays with T4 PNK by mixing 40 μM oligo, [γ-32P]-ATP, and T4 polynucleotide kinase (25 units) in T4 reaction buffer (supplied by NEB) in a volume of 40 μl at 37°C for 1 hour. T4 PNK was inactivated by incubation at 65°C for 20 minutes. One microliter of the reaction was spotted on a PEI-cellulose TLC plate, and excess ATP was removed from the sample with Oligo Clean and Concentrator spin columns (Zymo). One microliter of the cleaned-up reaction was spotted on a separate lane of the TLC plate and dried. The TLC plate was developed in 375 mM phosphate buffer, pH 4.0, dried, then visualized with phosphorimaging. The concentration of labeled oligo was determined with the formula C1V1=C2V2 and the purified oligo was stored at -20°C until annealing with the template strand (described above) for use in kinetics assays.

      Kinetic Experiments

      Kinetic experiments were performed in T7 Reaction Buffer (40 mM Tris-HCl pH 7.5, 50 mM NaCl, 1 mM EDTA, 1 mM DTT) (
      • Dangerfield T.L.
      • Johnson K.A.
      Optimized incorporation of an unnatural fluorescent amino acid affords measurement of conformational dynamics governing high-fidelity DNA replication.
      ,
      • Dangerfield T.L.
      • Johnson K.A.
      Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
      ) at 20°C. All reactions contained a final concentration of 12.5 mM Mg2+, however some experiments had Mg2+ in both syringes while other experiments had the enzyme and DNA in one syringe and Mg2+ was mixed in from the other syringe. In either case the final Mg2+ concentration during the reaction was 12.5 mM. Concentrations of reaction components given in the text are final concentrations after mixing unless otherwise noted.

      Rapid-quench experiments

      Rapid-quench experiments were performed with T7 Reaction Buffer without Mg2+ in the drive syringes and 0.6 M EDTA in the quench syringe, with the circulating water bath set to 20°C. Formamide Loading Buffer (5% [w/v] sucrose, 90% [v/v] formamide, 10 mM EDTA, 0.025% [w/v] bromophenol blue, 0.025% [w/v] xylene cyanol) was added to each sample at a 1:2 ratio of loading buffer to sample. Samples were heated at 98°C for 5 minutes and separated on 15% polyacrylamide sequencing gels containing 7 M urea for approximately 3 hours at 50°C. For samples with 32P labeled oligos, gels were dried then visualized by phosphorimaging after overnight exposure. For samples with FAM labeled DNA, gels were scanned on a Typhoon FLA 9500 laser scanner (GE Healthcare) with the FAM fluorescence filter. For both 32P experiments and FAM experiments, gels were quantified with Image Quant software (GE Healthcare). All rapid-quench experiments with FAM labeled DNA were performed on at least two separate occasions to ensure reproducibility. Rapid quench experiments with 32P labeled tCo oligos were reproduced with FAM-labeled substrates on at least 2 separate occasions with similar results.

      Stopped-flow experiments

      Stopped-flow experiments were performed on a SF-300X instrument (KinTek Corp.) with a dead time of 1.3 milliseconds using a 150-watt xenon lamp (Hamamatsu) as the light source. All stopped-flow traces in the main text are an average of at least 8 individual traces and each experiment was repeated on at least 3 separate occasions to ensure reproducibility. Fluorescence anisotropy experiments with FAM labeled oligos were performed with the anisotropy kit for the SF-300X instrument (KinTek Corp.), including vertical and horizontal emission polarizers and a Glan-Taylor excitation polarizer. Polarizers were calibrated with a solution of glycogen in water using previously established methods (

      Lakowicz, J. R. (1999) Principles of fluorescence spectroscopy, Second edition. New York : Kluwer Academic/Plenum, [1999] ©1999

      ). Before data collection, the G-factor was measured and applied to subsequent traces. Anisotropy experiments were performed with an excitation wavelength of 470 nm with a slit width of 3.16 mm, and emission was monitored with 520 nm filters with a 15 nm bandpass. Fluorescence experiments with tCo labeled oligonucleotides were performed with excitation at 295 nm with a slit width of 1.56 mm. Emission was monitored at 445 nm with a 45 nm bandpass. All fluorescence experiments were performed on at least two separation occasions to ensure reproducibility.
      Equilibrium titrations were performed with the TMX titration module for the SF-300X instrument (KinTek Corp.). Titrations were performed with 280 μl of a solution of tCo DNA in a cuvette in a temperature-controlled cuvette chamber and 20.5 μl of titrant (enzyme) was added from a Hamilton syringe over the course of 5 minutes. Excitation was at 295 nm with a slit width of 1.56 mm. Emission was monitored with a 445 nm filter with a 45 nm bandpass. The reaction was mixed continuously with a micro stir bar in the cuvette and the titration data were corrected for the small dilution before further manipulation. Because the sample was excited at 295 nm and the enzyme (FRET donor) was the titrant, the raw data showed the inverse of an inner filter effect, with increasing fluorescence with added enzyme due to the tryptophan residues acting as the FRET pair donor. To correct for the fluorescence artifact, first the data corrected for the small dilution were fit in KinTek Explorer using the following observable, where a is the overall scaling factor, b is the scaling factor for the increase in fluorescence upon binding DNA, and ε is the extinction coefficient of the titrant.
      observable=a*(D+(b*ED))*110ε*(E+ED)ln(10)*ε*(E+ED)


      The fit from KinTek Explorer gave a value for e of 0.805 which was then used to correct the raw data using the following equation:
      CorrectedFluorescence=RawFluorescence110ε*[T7DNAPAdded]ln(10)*ε*[T7DNAPAdded]


      The corrected data were then fit in KinTek Explorer using the observable:
      Observable=a*(D+(b*ED))


      The corrected data are shown in the main text.

      Software and Figures

      Data analysis was performed with the simulation software KinTek Explorer v10 (kintekexplorer.com) (
      • Johnson K.A.
      Fitting enzyme kinetic data with KinTek global kinetic explorer.
      ,
      • Johnson K.A.
      • Simpson Z.B.
      • Blom T.
      Global kinetic explorer: a new computer program for dynamic simulation and fitting of kinetic data.
      ). Kinetics figures were prepared with KinTek Explorer software. Model scheme images were prepared with Inkscape software. Structure figures were prepared with Pymol (

      Schrodinger, LLC. (2015) The PyMOL Molecular Graphics System, Version 1.8.

      ).

      Data analysis in KinTek Explorer

      Mechanism files with global fits and confidence contours derived in fitting the data for the three substrates described in this paper are provided, as described in the Supplemental Information. These files can be opened without needing a license using KinTek Explorer software available at https://kintekexplorer.com/downloads/. Models used for data fitting are given in the Figures for the corresponding substrates. For phosphorothioate substrates, parallel models without hydrolysis steps were used with rate constants linked to rate constants for the kinetic schemes for the hydrolysable substrates. Flux calculations described in the main text were performed using a dynamic partial derivative analysis during data fitting based on numerical integration of the rate equations using KinTek Explorer software (

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ,
      • Johnson K.A.
      Fitting enzyme kinetic data with KinTek global kinetic explorer.
      ).
      For conventional fitting of data using equations described in the figure legends, the analytical fit (aFit) function of KinTek Explorer was used. The equation y=A0+A1(1eb1t) was used for fits to a single exponential function, where A0 is the y-intercept, A1 is the amplitude, and b1 is the observed rate. The equation y=A0+A1(1eb1t)+A2(1eb2t) was used for data fitting two exponentials where A0 is the y-intercept, A1 and A2 are the amplitudes and b1 and b2 are the observed rates of the 1st and 2nd phases, respectively. For data fitting three exponentials, the following equation was used: y=A0+A1(1eb1t)+A2(1eb2t)+A3(1eb3t) where A0 is the y-intercept, An is the amplitude and bn is the observed rate of the nth exponential. The quadratic equation used was =A0+AKd+E+S(Kd+E+S)2+4ES2E , where A0 is the y-intercept, A is the amplitude, Kd is the equilibrium dissociation constant, E is the enzyme concentration, and S is the substrate concentration. The equation used for the hyperbolic fit was y=A0+A1SKd+S where A0 is the y-intercept, A1 is the amplitude, S is the substrate concentration, and Kd is the equilibrium dissociation constant. The results from equation-based fitting are given in the figure legends but were used only to guide the development of the complete models derived by global data fitting. All lines given in Figures 2, 4 and 6 were calculated by numerical integration from the global fits using the complete models.

      Data Availability

      All data are available in Supplemental Information in the form of KinTek Explorer mechanism files. These files can be opened (without needing a license) using KinTek Explorer software available at https://kintekexplorer.com, where video tutorials are also available to illustrate use of the software. The mechanism files also show how the data were modeled, and users can explore alternative models (

      Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

      ,
      • Johnson K.A.
      • Simpson Z.B.
      • Blom T.
      Global Kinetic Explorer: A new computer program for dynamic simulation and fitting of kinetic data.
      ).

      Conflict of Interest

      K. A. J. is president of KinTek Corporation, which provided the RQF-3 rapid quench flow and SF 300-X stopped-flow instruments and KinTek Explorer software used in this study.

      Acknowledgements

      This work was supported by grants from the National Institute of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health (NIH) R01AI110577 (to K. A. J) and National Institute of General Medical Sciences (NIGMS) of the National Institutes of Health (NIH) R01114223 (to K. A. J.).

      References

        • Wong I.
        • Patel S.S.
        • Johnson K.A.
        An induced-fit kinetic mechanism for DNA replication fidelity: direct measurement by single-turnover kinetics.
        Biochemistry (Mosc). 1991; 30: 526-537
        • Donlin M.J.
        • Patel S.S.
        • Johnson K.A.
        Kinetic partitioning between the exonuclease and polymerase sites in DNA error correction.
        Biochemistry. 1991; 30: 538-546
        • Dodd T.
        • Botto M.
        • Paul F.
        • Fernandez-Leiro R.
        • Lamers Meindert H.
        • Ivanov I.
        Polymerization and editing modes of a high-fidelity DNA polymerase are linked by a well-defined path.
        Nature communications. 2020; 11 (5379): 5379
        • Black C.B.
        • Cowan J.A.
        A critical evaluation of metal-promoted Klenow 3′-5′ exonuclease activity: calorimetric and kinetic analyses support a one-metal-ion mechanism.
        JBIC Journal of Biological Inorganic Chemistry. 1998; 3: 292-299
        • Lamichhane R.
        • Berezhna S.Y.
        • Gill J.P.
        • Van der Schans E.
        • Millar D.P.
        Dynamics of Site Switching in DNA Polymerase.
        Journal of the American Chemical Society. 2013; 135: 4735-4742
        • Johnson A.A.
        • Johnson K.A.
        Exonuclease Proofreading by Human Mitochondrial DNA Polymerase.
        Journal of Biological Chemistry. 2001; 276: 38097-38107
        • Beechem J.M.
        • Otto M.R.
        • Bloom L.B.
        • Eritja R.
        • Reha-Krantz L.J.
        • Goodman M.F.
        Exonuclease−Polymerase Active Site Partitioning of Primer−Template DNA Strands and Equilibrium Mg2+ Binding Properties of Bacteriophage T4 DNA Polymerase.
        Biochemistry. 1998; 37: 10144-10155
        • Dangerfield T.L.
        • Johnson K.A.
        Optimized incorporation of an unnatural fluorescent amino acid affords measurement of conformational dynamics governing high-fidelity DNA replication.
        J Biol Chem. 2020; 295: 17265-17280
        • Dangerfield T.L.
        • Johnson K.A.
        Conformational dynamics during high-fidelity DNA replication and translocation defined using a DNA polymerase with a fluorescent artificial amino acid.
        J. Biol. Chem. 2021; 296100143
        • Dangerfield T.L.
        • Kirmizialtin S.
        • Johnson K.A.
        Conformational dynamics during misincorporation and mismatch extension defined using a DNA polymerase with a fluorescent artificial amino acid.
        J. Biol. Chem. 2022; 298101451
        • Dangerfield T.L.
        • Kirmizialtin S.
        • Johnson K.A.
        Substrate specificity and proposed structure of the proofreading complex of T7 DNA polymerase.
        J. Biol. Chem. 2022; 298101627
        • Sandin P.
        • Börjesson K.
        • Li H.
        • Mårtensson J.
        • Brown T.
        • Wilhelmsson L.M.
        • Albinsson B.
        Characterization and use of an unprecedentedly bright and structurally non-perturbing fluorescent DNA base analogue.
        Nucleic Acids Res. 2008; 36: 157-167
        • Stengel G.
        • Purse B.W.
        • Wilhelmsson L.M.
        • Urban M.
        • Kuchta R.D.
        Ambivalent Incorporation of the Fluorescent Cytosine Analogues tC and tCo by Human DNA Polymerase α and Klenow Fragment.
        Biochemistry. 2009; 48: 7547-7555
        • Sandin P.
        • Stengel G.
        • Ljungdahl T.
        • Börjesson K.
        • Macao B.
        • Wilhelmsson L.M.
        Highly efficient incorporation of the fluorescent nucleotide analogs tC and tCO by Klenow fragment.
        Nucleic Acids Res. 2009; 37: 3924-3933
        • Otto M.R.
        • Bloom L.B.
        • Goodman M.F.
        • Beechem J.M.
        Stopped-Flow Fluorescence Study of Precatalytic Primer Strand Base-Unstacking Transitions in the Exonuclease Cleft of Bacteriophage T4 DNA Polymerase.
        Biochemistry. 1998; 37: 10156-10163
        • Brautigam C.A.
        • Steitz T.A.
        Structural principles for the inhibition of the 3′-5′ exonuclease activity of Escherichia coli DNA polymerase I by phosphorothioates 11Edited by R. Huber.
        Journal of Molecular Biology. 1998; 277: 363-377
        • Brautigam C.A.
        • Sun S.
        • Piccirilli J.A.
        • Steitz T.A.
        Structures of Normal Single-Stranded DNA and Deoxyribo-3‘-S-phosphorothiolates Bound to the 3‘-5‘ Exonucleolytic Active Site of DNA Polymerase I from Escherichia coli.
        Biochemistry. 1999; 38: 696-704
        • Johnson K.A.
        • Simpson Z.B.
        • Blom T.
        FitSpace Explorer: An algorithm to evaluate multidimensional parameter space in fitting kinetic data.
        Anal. Biochem. 2009; 387: 30-41
      1. Johnson, K. A. (2019) Kinetic Analysis for the New Enzymology, KinTek Corporation, Austin

        • Gong S.
        • Kirmizialtin S.
        • Chang A.
        • Mayfield J.E.
        • Zhang Y.J.
        • Johnson K.A.
        Kinetic and thermodynamic analysis defines roles for two metal ions in DNA polymerase specificity and catalysis.
        J Biol Chem. 2021; 296100184
        • Johnson K.A.
        Fitting enzyme kinetic data with KinTek global kinetic explorer.
        Methods in enzymology. 2009; 467: 601-626
        • Johnson K.A.
        • Simpson Z.B.
        • Blom T.
        Global kinetic explorer: a new computer program for dynamic simulation and fitting of kinetic data.
        Analytical biochemistry. 2009; 387: 20-29
        • Bessman M.J.
        • Muzyczka N.
        • Goodman M.F.
        • Schnaar R.L.
        Studies on the biochemical basis of spontaneous mutation: II. The incorporation of a base and its analogue into DNA by wild-type, mutator and antimutator DNA polymerases.
        Journal of Molecular Biology. 1974; 88: 409-421
        • Clayton L.K.
        • Goodman M.F.
        • Branscomb E.W.
        • Galas D.J.
        Error induction and correction by mutant and wild type T4 DNA polymerases. Kinetic error discrimination mechanisms.
        Journal of Biological Chemistry. 1979; 254: 1902-1912
        • Hopkins R.
        • Goodman M.F.
        Asymmetry in forming 2-aminopurine · hydroxymethylcytosine heteroduplexes; a model giving misincorporation frequencies and rounds of DNA replication from base-pair populations in vivo.
        Journal of Molecular Biology. 1979; 135: 1-22
        • Sowers L.C.
        • Eritja R.
        • Chen F.M.
        • Khwaja T.
        • Kaplan B.E.
        • Goodman M.F.
        • Victor Fazakerley G.
        Characterization of the high pH wobble structure of the 2-aminopurine·cytosine mismatch by N-15 NMR spectroscopy.
        Biochemical and Biophysical Research Communications. 1989; 165: 89-92
        • Sowers L.C.
        • Fazakerley G.V.
        • Eritja R.
        • Kaplan B.E.
        • Goodman M.F.
        Base pairing and mutagenesis: observation of a protonated base pair between 2-aminopurine and cytosine in an oligonucleotide by proton NMR.
        Proc Natl Acad Sci U S A. 1986; 83: 5434-5438
        • Watanabe S.M.
        • Goodman M.F.
        On the molecular basis of transition mutations: frequencies of forming 2-aminopurine.cytosine and adenine.cytosine base mispairs in vitro.
        Proc Natl Acad Sci U S A. 1981; 78: 2864-2868
        • Pless R.C.
        • Bessman M.J.
        Influence of local nucleotide sequence on substitution of 2-aminopurine for adenine during deoxyribonucleic acid synthesis in vitro.
        Biochemistry. 1983; 22: 4905-4915
        • Bloom L.B.
        • Otto M.R.
        • Eritja R.
        • Reha-Krantz L.J.
        • Goodman M.F.
        • Beechem J.M.
        Pre-Steady-State Kinetic Analysis of Sequence-Dependent Nucleotide Excision by the 3'-Exonuclease Activity of Bacteriophage T4 DNA Polymerase.
        Biochemistry. 1994; 33: 7576-7586
        • Bellamy S.R.W.
        • Kovacheva Y.S.
        • Zulkipli I.H.
        • Halford S.E.
        Differences between Ca2+ and Mg2+ in DNA binding and release by the SfiI restriction endonuclease: implications for DNA looping.
        Nucleic Acids Res. 2009; 37: 5443-5453
        • Soulière M.F.
        • Perreault J.-P.
        • Bisaillon M.
        Magnesium-binding studies reveal fundamental differences between closely related RNA triphosphatases.
        Nucleic Acids Res. 2007; 36: 451-461
        • Green N.H.
        • Williams P.M.
        • Wahab O.
        • Davies M.C.
        • Roberts C.J.
        • Tendler S.J.B.
        • Allen S.
        Single-Molecule Investigations of RNA Dissociation.
        Biophysical Journal. 2004; 86: 3811-3821
        • Guéroult M.
        • Picot D.
        • Abi-Ghanem J.
        • Hartmann B.
        • Baaden M.
        How Cations Can Assist DNase I in DNA Binding and Hydrolysis.
        PLOS Computational Biology. 2010; 6e1001000
        • Stec W.J.
        • Zon G.
        • Egan W.
        Automated solid-phase synthesis, separation, and stereochemistry of phosphorothioate analogs of oligodeoxyribonucleotides.
        Journal of the American Chemical Society. 1984; 106: 6077-6079
        • Thorogood H.
        • Grasby J.A.
        • Connolly B.A.
        Influence of the Phosphate Backbone on the Recognition and Hydrolysis of DNA by the EcoRV Restriction Endonuclease: A STUDY USING OLIGODEOXYNUCLEOTIDE PHOSPHOROTHIOATES.
        Journal of Biological Chemistry. 1996; 271: 8855-8862
        • Grasby J.A.
        • Connolly B.A.
        Stereochemical outcome of the hydrolysis reaction catalyzed by the EcoRV restriction endonuclease.
        Biochemistry. 1992; 31: 7855-7861
        • Jia Y.
        • Kumar A.
        • Patel S.S.
        Equilibrium and Stopped-flow Kinetic Studies of Interaction between T7 RNA Polymerase and Its Promoters Measured by Protein and 2-Aminopurine Fluorescence Changes.
        Journal of Biological Chemistry. 1996; 271: 30451-30458
        • Ferreiro D.U.
        • de Prat-Gay G.
        A Protein–DNA Binding Mechanism Proceeds Through Multi-state or Two-state Parallel Pathways.
        Journal of Molecular Biology. 2003; 331: 89-99
        • Bar-Even A.
        • Noor E.
        • Savir Y.
        • Liebermeister W.
        • Davidi D.
        • Tawfik D.S.
        • Milo R.
        The Moderately Efficient Enzyme: Evolutionary and Physicochemical Trends Shaping Enzyme Parameters.
        Biochemistry. 2011; 50: 4402-4410
        • Nickell C.
        • Prince M.A.
        • Lloyd R.S.
        Consequences of molecular engineering-enhanced DNA binding in a DNA repair enzyme.
        Biochemistry. 1992; 31: 4189-4198
        • Kozlov A.G.
        • Lohman T.M.
        Stopped-flow studies of the kinetics of single-stranded DNA binding and wrapping around the Escherichia coli SSB tetramer.
        Biochemistry. 2002; 41: 6032-6044
        • Qian Y.
        • Johnson K.A.
        The human mitochondrial single-stranded DNA-binding protein displays distinct kinetics and thermodynamics of DNA binding and exchange.
        J. Biol. Chem. 2017; 292: 13068-13084
        • Brutlag D.
        • Kornberg A.
        Enzymatic Synthesis of Deoxyribonucleic Acid: XXXVI. A PROOFREADING FUNCTION FOR THE 3′ → 5′ EXONUCLEASE ACTIVITY IN DEOXYRIBONUCLEIC ACID POLYMERASES.
        Journal of Biological Chemistry. 1972; 247: 241-248
        • Wang J.
        • Yu P.
        • Lin T.C.
        • Konigsberg W.H.
        • Steitz T.A.
        Crystal Structures of an NH2-Terminal Fragment of T4 DNA Polymerase and Its Complexes with Single-Stranded DNA and with Divalent Metal Ions.
        Biochemistry (Easton). 1996; 35: 8110-8119
        • Baker R.P.
        • Reha-Krantz L.J.
        Identification of a transient excision intermediate at the crossroads between DNA polymerase extension and proofreading pathways.
        Proc Natl Acad Sci U S A. 1998; 95: 3507-3512
        • Patel S.S.
        • Wong I.
        • Johnson K.A.
        Pre-Steady-State Kinetic-Analysis of Processive Dna-Replication Including Complete Characterization of An Exonuclease-Deficient Mutant.
        Biochemistry (Mosc). 1991; 30: 511-525
        • Akabayov B.
        • Akabayov S.R.
        • Lee S.J.
        • Tabor S.
        • Kulczyk A.W.
        • Richardson C.C.
        Conformational dynamics of bacteriophage T7 DNA polymerase and its processivity factor, Escherichia coli thioredoxin.
        Proc Natl Acad Sci U S A. 2010; 107: 15033-15038
      2. Lakowicz, J. R. (1999) Principles of fluorescence spectroscopy, Second edition. New York : Kluwer Academic/Plenum, [1999] ©1999

      3. Schrodinger, LLC. (2015) The PyMOL Molecular Graphics System, Version 1.8.

        • Johnson K.A.
        • Simpson Z.B.
        • Blom T.
        Global Kinetic Explorer: A new computer program for dynamic simulation and fitting of kinetic data.
        Anal. Biochem. 2009; 387: 20-29