Influence of DNA Structure on DNA Polymerase (cid:1) Active Site Function EXTENSION OF MUTAGENIC DNA INTERMEDIATES*

In the ternary substrate complex of DNA polymerase (pol) (cid:1) , the nascent base pair (templating and incoming nucleotides) is sandwiched between the duplex DNA terminus and polymerase. To probe molecular interactions in the dNTP-binding pocket, we analyzed the kinetic behavior of wild-type pol (cid:1) on modified DNA substrates that alter the structure of the DNA terminus and represent mutagenic intermediates. The DNA substrates were modified to 1) alter the sequence of the duplex terminus (matched and mismatched), 2) introduce abasic sites near the nascent base pair, and 3) insert extra bases in the primer or template strands to mimic frameshift intermediates. The results indicate that the nucleotide insertion efficiency ( k cat / K m , dGTP-dC) is highly depend- ent on the sequence identity of the matched ( i.e. Watson-Crick base pair) DNA terminus (template/primer, G/C (cid:1) A/T > T/A (cid:1) C/G). Mismatches at the primer terminus strongly diminish correct nucleotide insertion efficiency but do not affect DNA binding affinity. Transition intermediates are generally extended more easily than transversions. Most mismatched primer termini decrease the rate of insertion and binding affinity of the incoming nucleotide. In contrast, the loss of catalytic efficiency with homopurine mismatches at the duplex DNA terminus is entirely due to the inability to insert the incoming nucleotide, since K d (dGTP) is not affected. Abasic sites and extra nucleotides in

DNA polymerases select (bind and incorporate) a nucleoside triphosphate (dNTP) from a pool of structurally similar molecules to preserve Watson-Crick base pairing rules. DNA polymerase (pol) 1 ␤ is a model polymerase to study mechanisms utilized to assure efficient and faithful DNA synthesis. Its small size, lack of essential accessory proteins, and absence of a proofreading exonuclease have facilitated its biochemical, kinetic, and structural characterization. More importantly, it shares many general structural and mechanistic features exhibited by other DNA polymerases (1).
DNA polymerase ␤ contributes two important enzymatic ac-tivities during single-nucleotide base excision DNA repair. A deoxyribose phosphate lyase activity is associated with the amino-terminal 8-kDa lyase domain. This activity excises a deoxyribose phosphate intermediate during repair of abasic sites and generates a 5Ј-phosphate in a single-nucleotide gap.
The nucleotidyl transferase activity of pol ␤ is associated with the 31-kDa polymerase domain that fills the single-nucleotide gap. In addition, the polymerase activity of pol ␤ is necessary for several alternate repair pathways that require longer gapfilling DNA synthesis (e.g. long patch base excision repair) (2,3). A general feature observed in the structures of all DNA polymerases that include a template overhang (i.e. singlestranded DNA) is that the trajectory of the template strand bends dramatically as it enters the polymerase active site (1). This serves at least two functions. First, it provides the polymerase the ability to assess whether geometrical constraints imposed by correct Watson-Crick hydrogen bonding occur, and second, it discourages the next templating base from prematurely entering the polymerase active site, which could result in the downstream template base coding for nucleotide insertion (deletion mutagenesis) (4). To accurately replicate DNA, polymerases need to stabilize the coding templating base as well as the correct, but not incorrect, incoming nucleotide (5). This is achieved through a series of protein-and substrateinduced conformational changes that result in a dNTP-binding pocket formed by the templating base, DNA duplex terminus, and enzyme. In the absence of an incoming nucleotide, the carboxyl-terminal N-subdomain 2 (residues 262-335) of pol ␤ is in an open conformation so that key polymerase side chains do not interact with the templating nucleotide. Upon binding the correct nucleotide, the N-subdomain closes on the nascent base pair (templating and incoming nucleotides), creating several key interactions with the enzyme (6). Accordingly, the constraints imposed by the dNTP-binding pocket are determined by DNA sequence (i.e. structure) as well as the conformational fluctuations that occur in response to enzyme and substrate binding.
In the pol ␤ closed ternary substrate complex, the nascent base pair is sandwiched between the duplex DNA terminus and ␣-helix N (Fig. 1). Lys 280 and Asp 276 of ␣-helix N contribute van der Waals interactions with the templating and incoming nucleotide bases, respectively, whereas Asn 279 and Arg 283 contribute DNA minor groove interactions. Structure-based sitedirected mutagenesis of these residues has identified interac-tions that contribute to efficient DNA synthesis (7)(8)(9)(10)(11)(12)(13). The specific contribution provided by these interactions appears to be dependent on the identity of the base pair that is formed. For example, decreasing the stacking interactions of residue 280 with the templating base by site-directed mutagenesis resulted in a more dramatic loss in binding affinity for incoming complementary pyrimidines than purines, indicating that the energetic contributions of specific side chain interactions are strongly dependent on the specific insertion (13).
Since the terminal base pair of the growing DNA duplex forms part of the binding pocket of the incoming nucleotide, we now examine the kinetic consequences of modifying the DNA structure at or near the polymerase active site. The structure of the dNTP-binding pocket was altered by varying the DNA sequence or introducing DNA mismatches (i.e. base substitution intermediates), "extra" nucleotides (i.e. frameshift intermediates), or abasic sites in the DNA duplex upstream of the polymerase active site.
Protein Purification-Human DNA polymerase ␤ was purified as described previously (14). Enzyme concentration was determined by Coomassie dye binding using purified pol ␤ as a standard (15). The concentration of purified pol ␤ was determined by total amino acid analysis.
DNA Preparation-The sequence, structure, and nomenclature of the DNA substrates used in this study are illustrated in Fig. 2.
An oligonucleotide DNA substrate containing a single-nucleotide gap at position 16 was prepared by annealing three gel-purified oligonucleotides (Oligos Etc., Wilsonville, OR, or Integrated DNA Technologies, Coralville, IA). Each oligonucleotide was suspended in 10 mM Tris-HCl, pH 7.4, and 1 mM EDTA, and the concentration was determined from their UV absorbance at 260 nm. The annealing reactions were carried out by incubating a solution of 10 M primer with 12 M of downstream and template oligonucleotides at 90 -100°C for 3 min followed by 30 min at 65°C and then slow cooling to room temperature.
Kinetic Assays-Steady-state kinetic parameters for single-nucleotide gap-filling reactions were determined by initial velocity measurements as described previously (13). Unless noted otherwise, enzyme activities were determined using a standard reaction mixture (50 l) containing 50 mM Tris-HCl, pH 7.4, 100 mM KCl, 5 mM MgCl 2 , and 200 nM single-nucleotide gapped DNA. In some instances requiring high dNTP concentrations (e.g. misinsertion reactions), the MgCl 2 concentration was increased to assure that there was at least 5 mM free Mg 2ϩ in the reaction mixture. Enzyme concentrations and reaction time intervals were chosen so that substrate depletion or product inhibition did not influence initial velocity measurements. Reactions were stopped with 20 l of 0.5 M EDTA and 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.
Equilibrium Binding Constants for Gapped Heteropolymeric DNA Substrates-The equilibrium dissociation constants (i.e. K d ) for the binding of heteropolymeric DNA gapped substrates were determined by inhibition of pol ␤ activity on a homopolymeric DNA as described previously (16). Enzyme activities were typically determined using a standard reaction mixture (50 l) containing 50 mM Tris-HCl, pH 7.4, 100 mM KCl, 5 mM MnCl 2 , 30 M [␣-32 P]dTTP, 30 or 300 nM poly(dA)poly(dT) 20 (K m ; expressed as 3Ј-OH primer termini), and varying concentrations of competitor heteropolymeric DNA. Radioactive dTTP incorporation was not typically observed on the competitor DNA substrate because the templating base is not adenine. Reactions were initiated by the addition of 15 nM pol ␤, incubated at room temperature for 10 min, and stopped by the addition of 20 l of 0.5 M EDTA. Quenched reaction mixtures were spotted on Whatman DE-81 filter disks and dried. Unincorporated [␣-32 P]dTTP was removed, and filters were counted as before (17).
FIG. 1. DNA polymerase ␤ stacking interaction with the nascent base pair. The nascent base pair (templating and incoming nucleotides) is sandwiched between ␣-helix N of the N-subdomain of pol ␤ and the growing DNA duplex terminus. In addition to the nascent base pair, two base pairs of duplex DNA are illustrated (stippled gray boxes) and the polarity of each strand (template and primer) indicated. Hydrophobic portions of the Asp 276 and Lys 280 side chains interact with the bases of the incoming and templating nucleotides, respectively. In the closed ternary complex structure, Asp 276 and Lys 280 also hydrogen-bond to Arg 40 and the templating nucleotide 5Ј-phosphate, respectively (not illustrated).

FIG. 2. Sequence and nomenclature of the DNA substrates.
Single-nucleotide gapped DNA substrates were constructed by annealing three oligonucleotides as outlined under "Experimental Procedures." The 5Ј terminus of the primer strand was radioactively labeled with [␥-32 P]ATP. A, correct nucleotide insertion was kinetically characterized on DNA substrates with a matched or mismatched primer terminus. The identity of the terminus is indicated as template/primer: templating base (X/Y:Z). B, correct dGTP insertion was characterized on DNA substrates where an extra nucleotide was strategically included in the oligonucleotide so that proper annealing resulted in a frameshift intermediate. The sequence of the oligonucleotides was generally the same as that indicated in A, except that a cytosine residue was included at specific positions as illustrated. For example, if an extra cytosine residue was situated behind the primer terminus (P Ϫ1 ), this substrate is referred to as P Ϫ2 and represents an addition frameshift intermediate. In this situation, P Ϫ3 can anneal with T Ϫ2 . Extra nucleotides in the template strand annealed to the primer result in deletion frameshift intermediates. In some instances, the identity of the extra nucleotide was altered to remove possible alternate annealing patterns. C, to remove stacking interactions among the bases within the DNA helix, abasic sites (i.e. tetrahydrofuran) were positioned in the primer or template strand, and correct (dGTP) insertion was kinetically analyzed. In this case, a P Ϫ2 DNA substrate indicates that the position of the abasic site is behind the primer terminus, P Ϫ1 .
Data were fitted to Equation 1 for competitive inhibition by nonlinear regression methods.
The Michaelis constant, K m , and k cat for the homopolymeric DNA substrate (S) were determined in the absence of heteropolymeric competitor DNA (C). When competitor DNA binds tightly (K d Ͻ polymerase concentration, E), an alternate form of Equation 1 (Equation 2) is necessary to account for the depletion of free inhibitor and enzyme as the EC complex is formed (18).
For a competitor inhibitor, the apparent inhibitor constant is given by Equation 3.

RESULTS
Influence of the Identity of the DNA Duplex Terminus on Catalytic Efficiency-Steady-state kinetic analysis previously revealed a strong dependence of catalytic efficiency (k cat /K m ) on the identity of the primer terminus for single-nucleotide gap filling by pol ␤ (19). That study examined the efficiency of dCTP insertion into a single-nucleotide gapped substrate with a templating guanine and systematically altered the primer terminus or downstream sequence to characterize the influence of DNA sequence on insertion efficiency. The downstream DNA sequence did not significantly alter the catalytic efficiency of nucleotide insertion. We have reexamined the influence of the identity of the primer terminus on insertion efficiency in a different DNA sequence context and with a different incoming nucleotide (i.e. dGTP rather than dCTP). The results are tabulated in Table I and indicate that the efficiency of dGTP insertion is strongly dependent on the identity of the matched (i.e. Watson-Crick base pair) primer terminus. As observed previously (19), primer termini with a pyrimidine are extended more efficiently than those with a purine situated at the 3Ј terminus (DNA terminus: template/primer, Kinetics of Mispair Extension-Human DNA polymerase ␤ produces base substitution errors at a frequency of 1.1 ϫ 10 Ϫ4 (T/dGTP) to 2.3 ϫ 10 Ϫ7 (C/dCTP) (Table II). For a base substitution error to become a mutation, the mispair must be extended. This is a kinetically challenging event for all DNA polymerases, most likely due to the aberrant nucleotide-binding pocket generated by the mispair. The kinetic analysis of mispair extension is more complex than that to determine the fidelity of DNA synthesis (20 -22). The expression to determine relative mismatch extension efficiency is given by Equation 4 (20).
This expression is the relative rate of adding a correct nucleotide (dNTP) onto a wrong (w, mismatched) and right (r, matched) primer termini at equal concentrations under steadystate conditions. The parameters K d , K m , and P refer to the equilibrium DNA binding affinity, dNTP Michaelis constant, and processivity for correct insertion on a matched or mismatched primer terminus. Processivity is defined as the ratio of rate constants describing nucleotide insertion and DNA dissociation (i.e. k pol /k off ). Thus, the relative efficiency for extending a mismatch is dependent on the correct nucleotide concentration with maximum discrimination occurring at infinitely low dNTP concentrations. This intrinsic mismatch extension efficiency is referred to as f min If the polymerase binds DNA with matched and mismatched primer termini with equal affinities, then f min 0 is simply the  Figure 2A. Except for the G/G terminus, dGTP insertion opposite a templating cytosine was followed. For G/G, the kinetics of dATP insertion was determined (i.e. templating T). The data are tabulated in order of decreasing catalytic efficiencies relative to a matched terminus. The template base of the terminal mismatch is used as the reference nucleotide for the matched terminus.
a Identity of the template (X) primer (Y) terminus adjacent to the templating base (Z) at position n. b As calculated from the ratio of catalytic efficiencies for correct insertion on a mismatched/matched primer terminus. ratio of catalytic efficiencies for insertion on a mismatched relative to that on a matched terminus.
DNA Binding Affinity-We have previously utilized a simple competition assay to assess the equilibrium DNA binding affinity of gapped DNA substrates with pol ␤ (16). The assay follows DNA synthesis on a homopolymeric template/primer system and determines the competitive inhibition constant, K i , for a heteropolymeric DNA substrate that does not support DNA synthesis (i.e. the templating bases on the two competing DNA substrates are different). Fig. 3 illustrates the results from an assay to determine the binding affinity of single-nucleotide gapped DNA that has a G/C or C/C (template/primer) terminus with a templating G or C, respectively. These DNA substrates have similar binding affinities (Table III). The binding affinity of the matched terminus with a templating G is the same as that determined by following the DNA concentration dependence of the pre-steady-state burst amplitude (K d ϳ20 nM) (11). The binding affinities for a series of singlenucleotide gapped DNA substrates with matched or mismatched primer termini were determined and are tabulated in Table III. Although the binding affinities for the onenucleotide gapped DNA substrates are modestly sensitive to DNA sequence, the mean binding affinities for matched and mismatched primer termini are very similar. Likewise, introduction of an abasic site opposite the primer terminus (i.e. T Ϫ1 ) or an extra base directly behind the primer terminus (ϩ1 frameshift intermediate, P Ϫ2 ) does not significantly affect DNA binding affinity.
To determine crystallographic structures of DNA polymerases with bound substrates, an abortive ternary substrate complex is produced (23). This is typically achieved with a polymerase-DNA complex, where the first two templating nucleotides are complementary to the included ddNTP. The polymerase inserts the first ddNTP, resulting in a terminated primer and binds another ddNTP, resulting in an abortive ternary complex. Utilizing a 3Ј-deoxycytosine at the primer terminus, the addition of the complementary dNTP results in a significant increase in the apparent binding affinity for the gapped DNA substrate (Table III), suggesting that important conformational changes have occurred.
Pol ␤-Dependent Mismatch Extension-Since the DNA bind-ing affinities for matched and mismatched primer termini are comparable, the ratio of catalytic efficiencies for correct nucleotide insertion on a mismatched terminus relative to that determined on a matched terminus approximates f min 0 . However, the strong dependence of catalytic efficiency for correct insertion on the identity of the matched primer terminus makes the calculation of f min 0 potentially ambiguous. For example, the catalytic efficiency for insertion on a G/G mismatch needs to be calculated relative to a matched terminus. In this case, a matched terminus could originate with the template base (i.e. G/C) or, alternatively, the primer base (i.e. C/G). Since the catalytic efficiencies for insertion on these matched termini differ by 25-fold, f min 0 will differ by 25-fold, depending on the identity of the matched terminus. We have chosen the template base of the mismatch as the reference nucleotide for comparison with the matched terminus, since the template strand serves as the blueprint for DNA synthesis. Consequently, the FIG. 3. DNA competition assay to access DNA binding affinity. The sensitivity of pol ␤ activity on a homopolymeric template/primer to increasing concentrations of heteropolymeric DNA was followed as described under "Experimental Procedures." The concentration of poly(dA)-p(dT) 20 was 30 nM primer 3Ј-termini. This is equivalent to the K m in the absence of competitor. The data were fitted to Equation 1 (solid line), and the equilibrium dissociation constant for the singlenucleotide gapped DNA substrate with a G/C (F, templating G) or C/C (E, templating C) primer terminus was 16 and 20 nM, respectively.  Fig. 2A. The identity of the primer terminus is G/C (X/Y). The data for each templating base (Z) are tabulated in order of decreasing catalytic efficiencies. mismatched primer nucleotide is the incorrect partner. It should be noted, however, that mismatches can arise through correct nucleotide insertion on a misaligned template strand followed by realignment (i.e. dislocation) (24). In this scenario, the incorrect nucleotide in the primer strand does not represent a misincorporation event. Steady-state kinetic parameters are tabulated in Table I in order of decreasing catalytic efficiencies for the respective mismatches relative to the corresponding matched terminus. Since insertion of dCTP on a G/G mismatch could not be readily measured (see below), the templating base was changed to thymidine. In general, pol ␤ extends transition intermediates more efficiently than transversions. In most instances, this is due to a decrease in k cat and an increase in K m . However, there are situations where changes in one kinetic parameter dominates over the other. For example, k cat for extension of T/G and T/C is hardly affected relative to T/A, but K m is increased 35and 74-fold, respectively. The interpretation of these steadystate kinetic parameters is not straightforward due to the observation that product dissociation (i.e. k off for nicked DNA) and nucleotide insertion (i.e. k pol ) are partially rate-limiting for correct insertion on a matched terminus (11). Thus, k cat can be a reflection of both rate constants. 3 The observation that it is not significantly altered suggests that the increase in K m represents a diminished dNTP binding affinity rather than a change in processivity (k pol /k off ). It is not surprising that a mismatched primer terminus results in diminished dNTP binding and insertion, since the primer terminus forms part of the dNTP-binding pocket. However, the kinetic analysis of extension of an A/A or G/G mismatch indicates that the loss of catalytic efficiency is entirely due to the loss of insertion (k pol ) without altering the binding affinity for the incoming nucleotide. Because k cat for correct insertion is diminished 460-and 2800-fold respectively, k cat is a direct measure of k pol . Since the DNA binding affinity is not significantly altered (Table III), k pol is now completely rate-limiting. In such a situation, K m is equivalent to K d for the incoming nucleotide (25). The K d for the correct incoming nucleotide on a matched terminus is ϳ10 M (11, 13). Thus, the increase in K m for these mismatches reflects the diminished k pol and not a lower binding affinity for the incoming nucleotide.
To verify that the correct nucleotide has a high affinity for the pol ␤ active site when an A/A mismatch forms a portion of the binding pocket, the DNA binding affinity was assessed in the presence of the complementary nucleotide. With a matched primer terminus that cannot be extended (i.e. 3Ј-deoxynucleotide), binding of the correct nucleotide results in an abortive complex that binds DNA more tightly than the binary DNA complex (Table III). As with the matched primer terminus, the binding affinity for single-nucleotide gapped DNA substrate with an A/A terminal mismatch is significantly increased in the presence of dGTP, indicating that a ternary complex is formed. Although dGTP binds well with the A/A duplex terminus, its insertion rate is diminished Ͼ460-fold 4 ( Table I). In contrast, inclusion of dGTP with a C/C mismatched primer terminus, which does not support strong binding of a correct nucleotide, has only a small influence on DNA binding affinity (Table III).
Misalignment of the Free Primer Terminus-As noted above, dGTP insertion opposite a templating deoxycytidine was too low on a G/G mismatch for kinetic characterization. Accordingly, we surveyed other templating nucleotides (Fig. 4). Measurable correct nucleotide insertion could be observed with the other templating bases on this mismatch. With a templating deoxyguanosine, the insertion of two nucleotides occurs. Taken 3 The full expressions for k cat and K m are (k pol k off )/(k pol ϩ k off ) and K d (k off /(k pol ϩ k off )), respectively (25). 4 The turnover number, k cat , is a reflection of different steps for insertion on matched and A/A mismatched termini. For the matched terminus, k cat is partially limited by nucleotide insertion and product dissociation. However, on the A/A mismatch, k cat is completely limited by nucleotide insertion (k pol ). Thus, the difference in k cat for matched and the A/A mismatch (ϳ460-fold relative to A/T) underestimates the intrinsic effect on k pol . It should be noted that the mechanistic step attributed to k pol is either chemistry or an undefined conformational change. Since numerous conformational changes occur upon substrate binding, k pol could be limited by different conformational changes that are substrate-and/or polymerase-dependent. Accordingly, the interpretation of both steady-state and pre-steady-state kinetic parameters should be guarded.  Products were analyzed as described under "Experimental Procedures." Limited product formation can be observed with all of the DNA substrates and suggests that in several instances, the primer terminus realigned to satisfy Watson-Crick hydrogen bonding. For example, with the templating guanine (G; ϩdCTP), correct insertion results in a cytosine at the primer terminus (n ϩ 1 product). However, since this terminus can realign to hydrogen-bond with the preceding guanine, a small amount of single-nucleotide gapped DNA substrate can be formed that would have an extra base (G) in the primer strand. This can serve as the substrate for an additional dCTP incorporation (n ϩ 2) product. See "Results" for further discussion.
together, these results are consistent with a realignment of the primer terminus to satisfy the hydrogen bonding capacity of the terminal primer base. Consequently, when the templating base is cytosine, the terminal deoxyguanosine pairs with the templating base, effectively producing nicked DNA and inhibiting further DNA synthesis. When the templating base is guanine, dCTP insertion results in a primer terminus that can pair with the templating deoxyguanosine or the preceding deoxyguanosine of the G/G mismatch. In this latter case, the realignment produces a one-nucleotide gap with a templating guanine that can support the insertion of a second dCTP (i.e. n ϩ 2). The realignment of the primer terminus was not observed with correct insertion opposite thymidine (dATP) or deoxyadenosine (dTTP).
Influence of Frameshift Intermediates on Catalytic Efficiency-The ability of a polymerase to produce addition or deletion errors is partially dependent on the ability to extend these intermediates. An addition error (ϩ1 frameshift) is characterized by intermediates with an "extra" unpaired base in the primer strand, whereas a deletion error (Ϫ1 frameshift) is characterized by intermediates with an extra base in the template strand (4,24). In vitro, pol ␤ is observed to produce significantly more deletion errors than addition errors (26).
Introduction of frameshift intermediates in the template/ primer stem strongly diminishes the catalytic efficiency for correct nucleotide insertion (Fig. 5). Accordingly, such mutagenic intermediates influence the molecular organization of the polymerase active site. In general, an extra nucleotide in the primer strand was more detrimental than in the template strand in the same position. For example, nucleotide insertion was completely abrogated with an extra cytosine residue behind the primer terminus (i.e. P Ϫ2 ). In contrast, an extra cytosine residue at T Ϫ2 resulted in a 2000-fold loss of catalytic efficiency. Modifying the identity of the extra base (e.g. T) or using a synthetic abasic site (i.e. tertrahydrofuran (THF)) at this position in the primer strand did not restore measurable insertion. However, moving the extra base upstream of the polymerase active site at least 5 nucleotides or 14 base pairs downstream restored catalytic efficiency to nearly that observed with an unmodified gapped DNA substrate.
Effect of an Abasic Site in the Template/Primer Stem on Catalytic Efficiency-Base stacking represents an important energetic source for the stability of duplex DNA. Perturbing these stacking interactions by introducing an abasic site analogue, THF, at strategic positions in the primer or template strands dramatically diminished the catalytic efficiency for insertion of the correct nucleotide (Table IV). Positioning the THF residue opposite the cytosine at the primer terminus (i.e. position T Ϫ1 ) dramatically reduced single-nucleotide gap-filling DNA synthesis. Substituting adenine for cytosine at the primer terminus resulted in a 6-fold increase in catalytic efficiency. Moving the abasic site one nucleotide upstream of the growing DNA terminus resulted in a less detrimental effect on catalytic efficiency. Again, an abasic site in the primer strand was more detrimental to catalytic efficiency than when it was in the opposite template strand. DISCUSSION As observed previously for pol ␤ in a different DNA sequence context (19), correctly matched base pairs at the growing DNA terminus influence the catalytic efficiency for the insertion of the correct nucleotide. Significantly, this result indicates that matched base pairs at the template/primer terminus for pol ␤ are not symmetrical. In contrast, Watson-Crick base pairs in duplex DNA are geometrically symmetrical and have a sequence independent disposition of hydrogen bonding groups in the DNA minor groove (27). Crystallographic structures of several DNA polymerases indicate that duplex DNA deviates from the standard B-form near the polymerase active site (23, 28 -31), suggesting that the polymerase may influence the structure of the primer terminus upon formation of a ternary substrate complex.
As observed with other DNA polymerases (20,21,32), DNA substrates with matched or mismatched primer termini bind to pol ␤ with similar affinities (Table III) (33). The similar affinities suggest that there are few polymerase interactions with the bases of the template/primer terminus in the polymerase-DNA binary complex. This suggestion is supported by the crystallographic structure of the open binary pol ␤-DNA complex (34). In contrast, the structure of the ternary substrate complex where the N-subdomain has closed upon the nascent base pair indicates that Arg 283 and Tyr 271 have been repositioned to form  Fig. 2B. Catalytic efficiencies (k cat /K m ) are expressed relative to that determined for a DNA substrate that is properly annealed (None). No DNA synthesis was observed when the extra base was in the primer strand immediately upstream of the 3Ј-primer (i.e. P Ϫ2 ). Substituting T or THF for the C at P Ϫ2 did not restore DNA synthesis. P ϩ14 represents a control frameshift intermediate substrate where the extra nucleotide is in the downstream oligonucleotide 14 nucleotides downstream of the templating cytosine residue. DNA polymerase ␤ is not believed to significantly interact with this portion of the DNA duplex. 0.004 c 2,500,000 a The nomenclature for the position of the abasic site adduct is outlined in the legend to Fig. 2C.
b The sequence of the nonadducted gapped DNA is given in Fig. 2A, where the identity of the template/primer terminus (X/Y) is G/C and the templating base (Z) is C.
c The efficiency increases 6.5-fold when the abasic site in the template strand is situated opposite A rather than C.
hydrogen bonds with the template/primer terminus. Additionally, Asp 276 and Lys 280 contribute van der Waals interactions with the nascent base pair in the closed complex (Fig.  1). The increase in apparent DNA binding affinity upon formation of an abortive complex is consistent with these additional interactions (Table III).
Base substitution and frameshift errors result in mutagenic DNA intermediates that perturb the polymerase active site. In terms of catalytic efficiency, pol ␤ extends transition intermediates more efficiently than transversions (Table I). Although the absolute insertion efficiency on different mispairs is dependent on the identity of the polymerase, most DNA polymerases generally extend transitions more efficiently than transversions (20,21,32,(35)(36)(37). This reflects the ability of the polymerase active site to easily accommodate a mismatched purine-pyrimidine base pair without perturbing the relative position of the 3Ј-hydroxyl of the primer terminus, active site metals, and ␣P of the incoming dNTP. In contrast to the preferential extension of transition intermediates by most DNA polymerases, human DNA polymerase can extend most mispairs with similar, and high, catalytic efficiencies, implicating it in the extension of aberrant primer termini (38).
A comparison of the relative catalytic efficiencies for extension (Table I) and insertion (Table II) indicates that correct insertion on a mispair is 10 -100-fold more efficient than making that mispair (Fig. 6). The apparent lone exception, A/G, suggests that the catalytic efficiency for insertion of dGTP opposite adenine is greater than the efficiency for insertion of the correct nucleotide on this mispair. However, since the extension efficiency for A/G was determined with a DNA substrate with a templating C, it is likely that the G at the primer terminus interfered with nucleotide insertion by competing with the incoming dGTP for the templating base. Such a scenario was observed with the G/G terminus (Table  I and Fig. 4).
Although the intrinsic relative mispair extension efficiency, f min 0 , characterizes the ability of a DNA polymerase to discriminate among mispairs, it does not indicate whether a specific mispair will be extended more readily than by another polymerase. The ability of a DNA polymerase to extend a mispair depends on 1) dNTP concentration, 2) processivity, and 3) f min 0 (Equation 4) (20). Additionally, since f min 0 represents a comparison of catalytic efficiencies for correct insertion on mismatched and matched termini (Eq. 5), the magnitude of this parameter will be highly dependent on the fidelity of the polymerase. As noted previously (39), DNA polymerase fidelity is strongly coupled to the efficiency for correct, but not incorrect, nucleotide insertion. In other words, two DNA polymerases that exhibit identical insertion efficiencies on a mismatch but differ 100fold in their efficiency to insert a correct nucleotide on a matched terminus would exhibit relative mispair extension efficiencies that differ by 100-fold. For example, pol ␤ inserts a nucleotide on a G/T mismatch with a catalytic efficiency of 0.005 M Ϫ1 s Ϫ1 (Table I), whereas DNA polymerase inserts a correct nucleotide with a catalytic efficiency that is 4-fold lower (35). Since the catalytic efficiencies for inserting a correct nucleotide on a matched G/C template/primer terminus are 1 and 0.037 M Ϫ1 s Ϫ1 for pol ␤ and , respectively, f min 0 would be 0.005 (pol ␤) and 0.033 (pol ). Thus, the elevated f min 0 observed for pol is a reflection of its lower catalytic efficiency on matched termini rather than an intrinsic propensity to extend mispairs as has been suggested (35). This example highlights the care that must be used in attempting to extrapolate the intrinsic relative mispair extension efficiencies among different DNA polymerases.
Since the efficiency of correct nucleotide insertion is dependent on the identity of the terminal mismatched base pair, the polymerase is expected to have a significant influence on the structure(s) of the specific mispair. Molecular dynamics simulations of terminal mismatches in the confines of the pol ␤ active site have suggested that the geometry of transversion mispairs (G/G and C/C) are more distorted than the A/C transition mispair (40), offering a structural origin for the kinetic consequence of these mispairs. DNA sequence is also expected to have a profound influence on the structure of the mismatch and therefore on polymerase-DNA interactions. As discussed above, such an instance was observed with extension of the G/G mismatch when the identity of the templating base was C (Fig. 4).
The structures of DNA mismatches at the template/primer terminus in the confines of an A-family DNA polymerase have recently been reported (41). In many instances, the structure of the mismatch is similar to those reported previously in duplex DNA in the absence of protein. However, several mismatches exhibited a frayed conformation (e.g. A/A) or were disordered (e.g. A/C). As noted by the authors, these later observations may indicate that there are several conformers in equilibrium underscoring the dynamic nature of the polymerase, DNA, and their interactions.
The binding affinities for the incoming correct nucleotide utilizing a mismatched primer terminus (Table I) are generally higher than the binding affinity for the incorrect nucleotide producing that specific mispair (Table II). Surprisingly, the binding affinity for the incoming nucleotide with several mismatched primer termini suggests that the binding pocket for the incoming nucleotide is not significantly perturbed, although the rate of nucleotide insertion is dramatically decreased. For example, the A/A or G/G primer termini do not perturb the K d(dNTP) but strongly inhibit nucleotide insertion. Consistent with this proposal is the observation that the addition of the next correct nucleotide with a gapped DNA substrate with an A/A mismatched terminus results in an apparent stronger DNA binding (Table III). Thus, dGTP binds but is not inserted. The observation that purine-purine mispairs do not occlude the dNTP-binding pocket suggests that one of the glycosidic torsion angles may be in syn-conformation to accommodate the large purine rings. The G/G mismatch situated at the primer terminus in the confines of an A-family DNA polymerase indicates that the primer G base has flipped into a syn-conformation (41). The ability of pol ␤ to bind an incoming FIG. 6. Relative efficiency for misinsertion and mismatch extension. The catalytic efficiencies (Eff) are compared for extending (ext) a base substitution error relative to that for inserting (ins) the incorrect nucleotide. The mismatches are expressed as template-primer or template-incoming nucleotide. The catalytic efficiencies are taken from Tables I and II for extension and misinsertion, respectively. Except for the dA-dG and dC-dA mispairs, the catalytic efficiency for mismatch extension is 10 -100-fold higher than for making the mismatch. nucleotide that is inserted very slowly such as on a mismatched primer terminus (e.g. G/G; Table I) could be biologically significant. In a cellular environment where all the dNTPs are present, correct dNTP binding after a base substitution error would result in a dead end complex. The apparent tighter DNA binding affinity would limit access to the mispair for an extrinsic proofreading exonuclease. Fortunately, transversion intermediates are generally the most difficult base substitution errors for pol ␤ to make (Table II).
Base stacking is a major stabilizing force in DNA. Structural and thermodynamic analysis of abasic sites in duplex DNA indicates that they disturb local stacking interactions with the adjacent base pairs, but not the global DNA conformation (42,43). Structures of a Y-family DNA polymerase, Sulfolobus solfataricus DNA polymerase IV, complexed with DNA containing abasic sites indicate that the conformation of the abasic site was dependent on the surrounding DNA sequence (44). In some instances, the abasic site was observed to be in equilibrium between two conformations, resulting in misaligned template and primer strands. These structural observations coincide with the low frameshift fidelity of this polymerase (45). For pol ␤, the introduction of an abasic site opposite the primer terminus resulted in a profound loss of catalytic efficiency (T Ϫ1 , Table IV). The catalytic efficiency was greater than 6-fold higher when the primer terminus was adenine rather than cytosine, consistent with potentially better base stacking attributes provided by adenine. When the abasic sites were positioned one base pair upstream of the primer terminus (i.e. position Ϫ2), the abasic site in the primer strand was more detrimental to active site function than when it was positioned in the template strand at the equivalent position.
As observed with the introduction of abasic sites near the primer terminus, introducing extra nucleotides in the primer strand appears to have a greater effect than when inserted in the equivalent position in the template strand (Fig. 5). When the extra base was positioned more than 5 nucleotides from the primer terminus, the catalytic efficiency was affected less than 10-fold. The structure of pol ␤ with bound DNA indicates that there are protein-DNA interactions 4 -5 base pairs upstream of the active site (34). The small influence of the extra base on catalytic efficiency beyond the Ϫ5-position may suggest that the extra cytosine produces a local structural change that is transmitted to the polymerase active site.
The observation that the effects on catalytic efficiency are larger when unpaired bases are introduced into the primer strand indicates that the primer terminus is extremely sensitive to primer strand modifications. Molecular dynamics simulations of terminal mispairs in the pol ␤ active site have noted that the geometry of the primer terminus appears less stable than the noncomplementary template nucleotide (40). In the structure of the closed ternary substrate complex, Tyr 271 and Arg 283 hydrogen-bond to the minor groove edge of the primer base and the sugar of the template nucleotide, respectively (34). In the structure of the open binary complex, these interactions are absent, since the N-subdomain (residues 262-335) is positioned away from the nascent base pair binding pocket. Alanine substitution for Arg 283 results in a dramatic decrease in catalytic efficiency for insertion of the correct nucleotide on a matched primer terminus (7,39), precluding characterization with a mismatched terminus. In contrast, alanine substitution for Tyr 271 has little or no effect on insertion efficiency with matched (9) or mismatched termini. 5 Additionally, the Y271A mutant exhibited an identical catalytic efficiency when an abasic site was situated opposite the primer terminus at T Ϫ1 . 5 Unexpectedly, these results indicate that Tyr 271 does not play a critically important role during correct nucleotide insertion independent of whether the primer terminus is properly basepaired. Alternatively, the hydrogen bond provided by Tyr 271 may be very important to catalytic cycling, but the polymerase has compensated for the loss of this hydrogen bond by altering hydrogen bonding networks, as suggested by molecular dynamics simulations (46).
Conclusion-The results presented here demonstrate that DNA sequence and structure have a profound influence on the nascent base pair binding pocket of pol ␤ that translates into altered active site function. In addition, it is not unexpected that the polymerase in turn will also induce local DNA conformational changes to optimize hydrogen bonding and stacking interactions. The conformational flexibility of the enzyme, substrates, and complexes makes it difficult to unambiguously predict the dynamic behavior of a polymerase during catalytic cycling. However, through a multifaceted approach employing structural, kinetic, and modeling techniques, general and distinct strategies utilized by DNA polymerases may be identified.