Efficiency of correct nucleotide insertion governs DNA polymerase fidelity.

DNA polymerase fidelity or specificity expresses the ability of a polymerase to select a correct nucleoside triphosphate (dNTP) from a pool of structurally similar molecules. Fidelity is quantified from the ratio of specificity constants (catalytic efficiencies) for alternate substrates (i.e. correct and incorrect dNTPs). An analysis of the efficiency of dNTP (correct and incorrect) insertion for a low fidelity mutant of DNA polymerase beta (R283A) and exonuclease-deficient DNA polymerases from five families derived from a variety of biological sources reveals that a strong correlation exists between the ability to synthesize DNA and the probability that the polymerase will make a mistake (i.e. base substitution error). Unexpectedly, this analysis indicates that the difference between low and high fidelity DNA polymerases is related to the efficiency of correct, but not incorrect, nucleotide insertion. In contrast to the loss of fidelity observed with the catalytically inefficient R283A mutant, the fidelity of another inefficient mutant of DNA polymerase beta (G274P) is not altered. Thus, although all natural low fidelity DNA polymerases are inefficient, not every inefficient DNA polymerase has low fidelity. Low fidelity polymerases appear to be an evolutionary solution to how to replicate damaged DNA or DNA repair intermediates without burdening the genome with excessive polymerase-initiated errors.

DNA polymerase fidelity or specificity expresses the ability of a polymerase to select a correct nucleoside triphosphate (dNTP) from a pool of structurally similar molecules. Fidelity is quantified from the ratio of specificity constants (catalytic efficiencies) for alternate substrates (i.e. correct and incorrect dNTPs). An analysis of the efficiency of dNTP (correct and incorrect) insertion for a low fidelity mutant of DNA polymerase ␤ (R283A) and exonuclease-deficient DNA polymerases from five families derived from a variety of biological sources reveals that a strong correlation exists between the ability to synthesize DNA and the probability that the polymerase will make a mistake (i.e. base substitution error). Unexpectedly, this analysis indicates that the difference between low and high fidelity DNA polymerases is related to the efficiency of correct, but not incorrect, nucleotide insertion. In contrast to the loss of fidelity observed with the catalytically inefficient R283A mutant, the fidelity of another inefficient mutant of DNA polymerase ␤ (G274P) is not altered. Thus, although all natural low fidelity DNA polymerases are inefficient, not every inefficient DNA polymerase has low fidelity. Low fidelity polymerases appear to be an evolutionary solution to how to replicate damaged DNA or DNA repair intermediates without burdening the genome with excessive polymerase-initiated errors.
The equilibrium between genome stability and instability is tightly regulated since mutations are central to aging, disease, and evolution. Thus, cellular strategies that modulate this equilibrium are of general and immense interest. The structure of DNA was proposed nearly 50 years ago and provided the first clue to how "genetic material" could be replicated faithfully (1). It is now recognized that DNA polymerases play pivotal roles in both genome replication and maintenance (i.e. DNA repair). Polymerases copy the parental (template) strand to generate a new or repaired complementary daughter strand, and accurate DNA synthesis during replication and repair is essential in maintaining genomic integrity. Although DNA polymerases play a central role in these essential processes, the fundamental mechanism(s) by which they select the correct deoxynucleoside 5Ј-triphosphate (dNTP) 1 from a pool of structurally similar molecules to accomplish efficient and faithful polymerization is poorly understood. The intrinsic base substitution error fre-quency for DNA replication and repair polymerases is generally between 10 Ϫ3 and 10 Ϫ6 (2). These frequencies represent one error per thousand or million nucleotides synthesized, respectively. These levels of discrimination are far greater than predicted by free energy differences between matched and mismatched DNA termini (predicted error frequency of ϳ0.4; one error per 3 nucleotides synthesized), indicating that DNA polymerases can enhance fidelity by a large factor (3). However, even this remarkable specificity is inadequate to faithfully replicate a genome of more than 10 9 nucleotides. Thus, replicative DNA polymerases often have an intrinsic proofreading exonuclease to remove misinserted nucleotides, and cells possess a postreplication DNA mismatch repair pathway that can correct misinserted nucleotides that escape proofreading.
DNA polymerase (pol) fidelity, specificity, or discrimination represent relative kinetic terms used to describe the propensity of a polymerase to produce a base substitution error. Polymerase specificity may be quantified in vitro by measuring the insertion kinetics of a single nucleotide (correct or incorrect) opposite a defined templating base. The absolute rate or probability that a pol inserts a correct or incorrect nucleotide follows Michaelis-Menten kinetics. A steady-state kinetic approach defines substrate specificity as catalytic efficiency, k cat / K m,dNTP , for formation of a specific base pair. Substrate specificity determined by a pre-steady-state kinetic approach is k pol /K d where K d is the equilibrium dissociation constant for the incoming dNTP and k pol is the kinetic step that limits insertion of the first nucleotide. This step may be either chemistry (i.e. nucleotidyl transfer) and/or conformational transitions of the ternary polymerase/substrate complex. Although these two kinetic approaches may measure different steps along the catalytic pathway, they measure the same specificity constant (i.e. k cat /K m ϭ k pol /K d ) (4). DNA polymerase specificity is typically quantified by comparing the ratio of catalytic efficiencies for correct and incorrect nucleotide insertion and typically expressed as relative misinsertion efficiency, f ins ϭ (k cat / K m ) incorrect /(k cat /K m ) correct , or fidelity (1/f ins ) (5, 6).
Since fidelity represents a ratio of catalytic efficiencies for correct and incorrect nucleotide insertion, differences in fidelity can be due to changes in one or both specificity constants. We had noted previously that a mutant of pol ␤ had low fidelity and catalytic efficiency for correct insertion (7). By examining the catalytic efficiencies for correct and incorrect nucleotide insertion for wild-type and mutant forms of pol ␤ and comparing them with those reported for other polymerases exhibiting divergent fidelities, we find that the fidelity of natural DNA polymerases is determined primarily by the efficiency for correct nucleotide insertion.

EXPERIMENTAL PROCEDURES
Protein Purification-Wild-type human pol ␤ and mutant derivatives (R283A and G274P) were expressed and purified as described previ-ously (7). Enzyme concentrations were determined by Coomassie dye binding using purified pol ␤ as the standard. The concentration of purified pol ␤ was determined by total amino acid analysis.
DNA Preparation-A 34-mer oligonucleotide DNA substrate containing a single nucleotide gap was prepared by annealing 3 gel-purified oligonucleotides (Oligos Etc., Wilsonville, OR) to create a single-nucleotide gap at position 16. The sequence of the gapped DNA substrate was: primer, 5Ј-CTGCAGCTGATGCGC-3Ј; downstream oligonucleotide, 5Ј-GTACGGATCCCCGGGTAC-3Ј; and template, 3Ј-GACGTC-GACTACGCGGCATGCCTAGGGGCCCATG-5Ј. The resultant singlenucleotide gapped DNA has a templating G in the gap. The oligonucleotides were quantified, resuspended in a buffer, and annealed as described previously (4). The primer was 5Ј-labeled with [␥-32 P]ATP using T4 polynucleotide kinase (New England BioLabs), and radioactive ATP was removed with a MicroSpin G-25 column.
Kinetic Assays-Enzyme activities were determined using a standard reaction mixture containing 50 mM Tris-HCl, pH 7.4 (22°C), 100 mM KCl, and 5 mM MgCl 2 . Steady-state or pre-steady-state kinetic parameters for single-nucleotide gap filling were determined as before (4). Quenched reaction samples were mixed with an equal volume of formamide dye, and the products were separated on 12% denaturing polyacrylamide gels. The dried gels were analyzed using a PhosphorImager (Amersham Biosciences) to quantify product formation.

RESULTS AND DISCUSSION
Inefficient Mutant of Pol ␤ Exhibiting Low Fidelity-Rational or random mutagenesis of the pol active site generally results in mutant enzymes that have moderately reduced or improved specificity. However, alanine substitution for an arginine residue of pol ␤ (R283A, Fig. 1), suggested to stabilize the templating base, results in a dramatic loss of catalytic efficiency for correct insertion and fidelity, implying that catalytic efficiency for correct nucleotide insertion and discrimination were coupled (7). Since those measurements were performed on dissimilar DNA substrates, we reexamined the catalytic efficiency for correct insertion and fidelity on the same single-nucleotide gapped DNA substrate, a substrate preferred by pol ␤ in base excision repair (8). A steady-state kinetic analysis of the R283A mutant of human pol ␤ indicates that there is a 33,000-fold loss in catalytic efficiency for insertion of dCTP opposite a templating guanine within a single-nucleotide gapped DNA substrate relative to the wild-type enzyme. In contrast, the mutant enzyme inserts dTTP opposite guanine 15-fold less efficiently than wild-type enzyme. Since fidelity or relative misinsertion efficiency are relative ratios of catalytic efficiencies, the mutant enzyme has a 2,400-fold lower ability to discriminate against dTTP insertion relative to dCTP ( Fig. 2A).  (27) is shown. The perspective is from the DNA major groove and indicates that the nascent base pair (pink, templating base and incoming nucleotide) is sandwiched between an ␣-helix of pol ␤ (blue) and duplex DNA (i.e. primer-terminus and its complementary base, n-1). Arg-283 makes van der Waals contact with the minor groove edge of the templating base (n) and forms a hydrogen bond with the sugar of the 3Ј-nucleotide (n-1). A cis-peptide bond occurs between Gly-274 and Ser-275, resulting in a sharp bend between two ␣-helices. Gly-274 interacts with the sugar of the incoming nucleotide. The van der Waals surface of Gly-274 and the Arg-283 side chain is illustrated. Additionally, the position of the active site divalent ions is indicated (purple). This figure was made with Molscript (28) and rendered with Raster3D (29).
FIG. 2. Relative misinsertion efficiency for formation of the dG-dTTP mispair as a function of catalytic efficiency for nucleotide insertion. The reported catalytic efficiencies for dCTP or dTTP insertion opposite guanine were analyzed for exonuclease-deficient pols that span five pol families (Table I) For some of these polymerases, several reported catalytic efficiencies are given. These represent alternate determinations from independent laboratories. The dotted line (f ins ϭ 0.6) represents the calculated free energy difference determined for matched and mismatched terminal base pairs (⌬⌬G ϳ0.3 kcal/mol at 37°C) (3). As shown in A, the efficiencies for correct nucleotide insertion (k cat /K m,dCTP ) span 5 orders of magnitude for the DNA polymerases surveyed with a corresponding decrease in the relative misinsertion efficiencies (f ins ). As shown in B, in contrast, the efficiencies for dTTP insertion occurred over a much narrower range for most polymerases and do not correlate with fidelity in a systematic way. This loss in discrimination is entirely due to the loss of the ability to insert the correct nucleotide, dCTP, since dTTP insertion efficiency is reduced in the mutant enzyme.
Low Fidelity DNA Polymerases Are Inefficient-To determine whether the loss of catalytic efficiency for the correct nucleotide generally leads to a loss in discrimination, as observed with the R283A mutant of pol ␤, we analyzed the re-ported insertion kinetics for 12 exonuclease-deficient DNA polymerases from a variety of biological sources that span five pol families (Table I). Replicative and repair DNA polymerases have generally been reported to insert the correct nucleotide with a catalytic efficiency of about 1-10 M Ϫ1 s Ϫ1 with corresponding relative misinsertion efficiencies of 10 Ϫ3 -10 Ϫ6 for a G-T (template base-incoming nucleotide) base pair ( Fig. 2A). Since DNA polymerases can typically misinsert three common nucleotides opposite a templating guanine, albeit with low efficiency, the data points for these mispairs are spread vertically (Fig. 3).
More recently, a new family of DNA polymerases (i.e. Yfamily, which includes pols , , and ) has been described (9). These polymerases lack an intrinsic proofreading exonuclease, exhibit low processivity, replicate DNA with low fidelity, and are believed to assist replication complexes stalled at DNA lesions (for a recent review, see Ref. 10). A survey of the catalytic efficiencies for nucleotide insertion by members of this family indicates that they generally insert dCTP opposite a templating guanine with low efficiency (Ͻ0.2 M Ϫ1 s Ϫ1 ). DNA polymerase inserts dCTP with an efficiency of about 10 Ϫ1 to 10 Ϫ2 M Ϫ1 s Ϫ1 (11,12), whereas the efficiency for pol for this base pair is about 10 Ϫ3 M Ϫ1 s Ϫ1 (13)(14)(15). Thus, the efficiencies for insertion of dCTP opposite guanine can vary 5 orders of magnitude depending on the identity of the pol ( Fig. 2A). As observed for the R283A mutant of pol ␤, discrimination is coupled to the catalytic efficiency for correct insertion, but not for incorrect insertion (e.g. dTTP insertion opposite guanine; Fig. 2B). The loss in the ability to insert the correct nucleotide results in the loss of discrimination (i.e. higher f ins ).
The trend in this analysis suggests that if the catalytic efficiency for correct insertion was reduced further, insertion of  (Table I). The dotted line (f ins ϭ 0.6) represents the calculated free energy difference determined for matched and mismatched terminal base pairs (⌬⌬G ϳ0.3 kcal/mol at 37°C) (3). Relative misinsertion efficiencies that fall below, or above, this line indicate that the pol active site imposes constraints that alter the free energy difference between correct and incorrect base pairing. This can result in an increase, or decrease, in the ability of the pol to discriminate between correct and incorrect base pairs. A, relative misinsertion efficiencies for the misinsertion of dATP (q), dCTP (E), or dGTP (ϫ) opposite adenosine. B, relative misinsertion efficiencies for the misinsertion of dATP (q), dCTP (E), or dTTP (ϫ) opposite cytosine. C, relative misinsertion efficiencies for the misinsertion of dATP (q), dGTP (E), or dTTP (ϫ) opposite guanine. D, Relative misinsertion efficiencies for the misinsertion of dCTP (q), dGTP (E), or dTTP (ϫ) opposite thymine.

TABLE I DNA polymerases and kinetic studies included in the global analysis
of polymerase specificity Reports that express insertion rates (V max ), rather than turnover numbers (k cat ), were excluded from this survey since those catalytic efficiencies (V max /K m ) could not be "normalized" for comparison. In some instances, k cat could be calculated from V max when the molecular concentration of the pol was given (k cat ϭ V max /[pol]). DNA  the incorrect nucleotide may be preferred over that of the correct ((k cat /K m ) incorrect Ͼ (k cat /K m ) correct so that f ins Ͼ 1). There are three examples where this has been observed (Fig. 3, C and  D): pol has been reported to insert dTTP (13,15) and dGTP (13)(14)(15) opposite a templating thymine with a higher efficiency than it inserts dATP. Additionally, pol X from the African swine fever virus inserts dGTP opposite a templating guanine with a greater efficiency than dCTP (16).
As observed for the misinsertion of dTTP opposite a templating guanine ( Fig. 2A), the relative misinsertion efficiency for all 12 mispairs is controlled primarily by the efficiency for correct nucleotide insertion. In all cases, a loss in the ability to insert the correct nucleotide results in a loss of discrimination (higher f ins ). More importantly, the slope of the lines in the log-log plots in Fig. 3 is equal to or greater than Ϫ1 for every mispair. These slopes range from Ϫ0.2 (dCTP insertion opposite adenine) to Ϫ1 (Fig. 4A). The slopes for most mispairs range from Ϫ0.6 to Ϫ1. This corresponds to little or no loss of catalytic efficiency for insertion of the incorrect nucleotide for a corresponding 10-fold loss of catalytic efficiency for correct insertion. However, the slope of the log-log plot for dCTP insertion opposite adenine is Ϫ0.2, corresponding to an 8-fold loss of catalytic efficiency for incorrect insertion with a 10-fold loss for correct insertion. Thus, the slope of these log-log plots may be a measure of the Watson-Crick geometry of each mispair, suggesting that in the confines of the pol active site, dCTP-dA has a geometry more similar to that of a Watson-Crick base pair than the reciprocal mispair (dATP-dC). When all of the mispairs are considered together, the overall slope is Ϫ0.78 Ϯ 0.09, indicating that the catalytic efficiency for incorrect insertion is ϳ1.7fold lower than for a pol that exhibits a 10-fold higher catalytic efficiency for correct insertion.
The catalytic efficiencies for correct insertion range from 56 M Ϫ1 s Ϫ1 , determined for pol ␥ (17), to 5 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 , reported for pol (15), for dATP insertion (Fig. 3). This represents a 10 million-fold range in catalytic efficiencies for correct insertion. Thus, although pol exhibits the lowest fidelity (greatest f ins ) for the dT-dGTP mispair, it also inserts dATP opposite thymine with the lowest efficiency of any polymerase surveyed here. Whereas pol ␥ exhibits a 30,000fold lower f ins than pol (i.e. higher fidelity), its catalytic efficiency for insertion of dGTP opposite thymine is 300-fold greater than pol .
Until recently, the "kinetic space" occupied by replicative and repair DNA polymerases indicated that these enzymes insert the correct nucleotide with an efficiency of about 1-10 M Ϫ1 s Ϫ1 . If f ins is examined for the different mispairs at a catalytic efficiency of 1 M Ϫ1 s Ϫ1 (i.e. 10 0 ), then it is generally observed that pyrimidine-purine mispairs (transition intermediates) are easier to make than pyrimidine-pyrimidine or purine-purine mispairs (transversion intermediates) (Fig. 4B), an observation that has been noted previously (6). Although this generalization is pertinent for DNA polymerases with moderate catalytic efficiencies for correct insertion (i.e. ϳ1 M Ϫ1 s Ϫ1 ), it would not pertain to the lower fidelity polymerases since the relationship between f ins and catalytic efficiency is moderately dependent on the specific mispair.
In general, f ins is strongly correlated with the catalytic efficiency for correct nucleotide insertion. However, the correlation is not perfect, and there are examples where this is clearly not the case. For example, dTTP insertion opposite guanine by human immunodeficiency virus-1 (HIV-1) reverse transcriptase is more efficient (higher f ins ; k cat /K m ϭ 0.03 M Ϫ1 s Ϫ1 ) than would be predicted by taking into account the efficiency that other DNA polymerases produce this mispair (k cat /K m ϳ10 Ϫ4 M Ϫ1 s Ϫ1 ; Fig. 2). This represents the most efficient misinsertion reported (18) in the studies included in this survey. More interestingly, the efficient insertion of dTTP opposite guanine is consistent with the G 3 A hypermutation observed among retroviruses (19,20) and represents an error that would be encouraged by the naturally low dCTP/dTTP pool imbalance (21).
Inefficient Mutant of Pol ␤ Exhibiting High Fidelity-As a first approximation, the correlation between correct nucleotide insertion efficiency and fidelity suggests that if the catalytic efficiency of a mutant pol were greatly diminished, then it may also exhibit a correspondingly low fidelity (e.g. R283A mutant of pol ␤). Traditionally, site-directed mutagenesis of the pol active site (e.g. metal-coordinating carboxy- FIG. 4. Parameters deduced from linear fits of log plots of catalytic efficiency for insertion of a correct nucleotide versus relative misinsertion efficiency for 12 possible mispairs. In A, the slopes are for the linear fits in Fig. 3. Note that if the catalytic efficiency for correct and incorrect insertion were altered in parallel, then the corresponding slope would be zero, and changes in catalytic efficiency would not result in a change in f ins (i.e. fidelity). However, if the catalytic efficiency for correct insertion decreases 10-fold and efficiency for incorrect insertion is not altered, then f ins also decreases 10-fold (i.e. the resulting slope of the log-log plot would be Ϫ1). Additionally, if the catalytic efficiencies for correct and incorrect were observed to move in opposite directions, then a 10-fold decrease in efficiency for correct insertion would result in a slope of the log-log plot of less than Ϫ1 (i.e. more negative). In B, the intercepts (f ins at log(k cat / K m ) correct ϭ 0) are from the linear fits in Fig. 3 and are plotted in ascending order. lates) of the A, B, X, and RT families results in a 100 -1000fold loss in catalytic efficiency (22)(23)(24)(25). Due to this considerable loss of catalytic efficiency, fidelity was not assessed with these "inactive" mutant enzymes. During a steady-state kinetic-screen of site-directed mutants of pol ␤, a G274P mutant exhibited a 10 4 -fold decrease in catalytic efficiency relative to wild-type enzyme. A rare cis-peptide bond is observed between Gly-274 and Ser-275 that creates a sharp turn between two ␣-helices that contribute significant interactions with the nascent base pair. Gly-274 interacts with the sugar of the incoming nucleotide (Fig. 1). Proline substitution for Gly-274 would be expected to sterically clash with the incoming nucleotide and alter helix interactions with the nascent base pair. The enormous loss of catalytic efficiency is consistent with this proposal. To discount the possibility that the mutant enzyme did not fold properly, resulting in a large fraction of inactive protein, we performed a single-turnover analysis (pol Ͼ Ͼ DNA substrate) to directly address the rate of insertion into a single-nucleotide gapped DNA substrate with a templating guanine. Under these assay conditions, the rate of insertion was nearly identical to that determined when substrate concentration is in excess (i.e. steady-state assay), indicating that the poor efficiency was intrinsic to the mutant enzyme (data not shown). To address whether there was a corresponding decrease in fidelity, we attempted to measure dTTP insertion efficiency opposite the templating guanine. However, misinsertion efficiency was very weak, precluding an accurate determination but suggesting that the efficiency for formation of this mispair was reduced significantly relative to wild-type enzyme. By substituting Mn 2ϩ for Mg 2ϩ in the assay mixture, the catalytic efficiency for correct insertion was increased. Under this condition, the fidelity of wild-type enzyme was 16,600 ((k cat /K m ) correct /(k cat /K m ) incorrect ), which is about 4-fold lower than in a reaction utilizing Mg 2ϩ (4). The catalytic efficiencies for correct (dCTP-dG) and incorrect (dTTP-dG) insertion were diminished 4800-and 1300-fold, respectively, for the G274P mutant relative to wild-type enzyme. This translates to a less than 4-fold loss in fidelity for the mutant enzyme, which is far less than the 880-fold loss predicted from the survey of native DNA polymerases. Most importantly, this observation indicates that although low fidelity DNA polymerases insert the correct nucleotide inefficiently, not all polymerases that are inefficient have low fidelity.
Concluding Remarks-Our analysis reveals that the kinetic difference between low and high fidelity DNA polymerases originate primarily in the ability of the pol to insert the correct nucleotide rather than its ability to insert the incorrect nucleotide. Accordingly, the molecular basis for the divergent fidelities requires a kinetic and structural understanding of correct nucleotide insertion. Since naturally occurring DNA polymerases exhibiting divergent fidelities generally insert incorrect nucleotides with similar efficiencies, high fidelity polymerases discriminate against incorrect nucleotides by actively selecting correct nucleotides. A comparison of published crystallographic substrate complexes of DNA polymerases with a Watson-Crick base pair in the confines of their active site suggests that the coordination of the P␣ of the incoming nucleotide has a strong influence on catalytic efficiency, whereas specific interactions with the bases of the nascent base pair do not. 2 These structural observations are consistent with the similar binding affinities, but divergent insertion rates, reported for the correct nucleotide with DNA polymerases exhibiting different fidelities.
Cells have solved the problem of how to replicate structural abnormalities in genomic DNA by evolving enzymes that have the ability to insert nucleotides opposite DNA lesions at the expense of fidelity. Thus, although low fidelity DNA polymerases would theoretically produce an inordinate number of errors per nucleotide synthesized, these polymerases are "kinetically controlled" by virtue of their catalytic efficiency for correct nucleotide insertion. However, the mutagenic threat of a low fidelity pol will be increased if its DNA synthesis capacity increases. This could be achieved through gene induction (increase in pol concentration), polymerase targeting (increase in local pol concentration), or an increase in catalytic efficiency through interaction with accessory proteins. All of these strategies appear to be utilized by the low fidelity Y-family DNA polymerases (for a recent review, see Ref. 26).
We have conducted a comparative study of the insertion efficiency as it relates to fidelity. Equally important subsequent steps concern the fate of a mispair. A mispair can be proofread by either an intrinsic or extrinsic 3Ј-5Ј exonuclease, or it could be extended by the same, or a different, DNA polymerase. Since extension of a mispair is a challenging kinetic event due to the aberrant DNA structure at the interface of the nascent base pair in the pol active site, it is not unexpected that DNA polymerases have also evolved to fulfill this specialized function.