DNA Polymerase β Substrate Specificity

Apurinic/apyrimidinic (AP) sites are continuously generated in genomic DNA. Left unrepaired, AP sites represent noninstructional premutagenic lesions that are impediments to DNA synthesis. When DNA polymerases encounter an AP site, they generally insert dAMP. This preferential insertion is referred to as the A-rule. Crystallographic structures of DNA polymerase (pol) β, a family X polymerase, with active site mismatched nascent base pairs indicate that the templating (i.e. coding) base is repositioned outside of the template binding pocket thereby diminishing interactions with the incorrect incoming nucleotide. This effectively produces an abasic site because the template pocket is devoid of an instructional base. However, the template pocket is not empty; an arginine residue (Arg-283) occupies the space vacated by the templating nucleotide. In this study, we analyze the kinetics of pol β insertion opposite an AP site and show that the preferential incorporation of dAMP is lost with the R283A mutant. The crystallographic structures of pol β bound to gapped DNA with an AP site analog (tertrahydrofuran) in the gap (binary complex) and with an incoming nonhydrolyzable dATP analog (ternary complex) were solved. These structures reveal that binding of the dATP analog induces a closed polymerase conformation, an unstable primer terminus, and an upstream shift of the templating residue even in the absence of a template base. Thus, dATP insertion opposite an abasic site and dATP misinsertions have common features.

DNA polymerases are responsible for template-directed DNA synthesis during genomic replication and repair. Genomic DNA is continuously assaulted with endogenous and environmental stresses that result in steady-state levels of 50,000 -200,000 apurinic/apyrimidinic (AP) 2 sites per eukaryotic cell (1). AP sites are generated through spontaneous or enzymatic hydrolysis of the N-glycosyl bond between the deoxyribose and base. The rate of spontaneous depurination has been estimated to be ϳ10,000/cell/day (2,3). In addition, AP sites are produced as intermediates during the base excision repair of oxidized and alkylated bases. Left unrepaired, AP sites represent noninstructional premutagenic lesions that can block DNA synthesis. When DNA polymerases confront an AP site, they generally insert dAMP. This preferential insertion has become known as the A-rule (4).
The sugar of a natural AP site is in equilibrium between three species as follows: the ␣and ␤-hemiacetals and the open chain aldehyde. The predominant form in solution is the closed hemiacetal with the ring-opened form representing less than 1% of the total AP sites at equilibrium (5). Tetrahydrofuran (THF) (Fig. 1A) is a useful and accepted AP site analog to study the repair of AP sites or the influence of such lesions on polymerase function. Additionally, crystallographic structures of DNA polymerases from various families bound to THF-containing DNA are available (6 -10).
DNA polymerase (Pol) ␤ contributes two enzymatic activities during the repair of simple base lesions in DNA: templatedirected DNA synthesis (nucleotidyltransferase) and deoxyribose 5Ј-phosphate excision (lyase) (11). These activities reside in separate domains as follows: a 31-kDa polymerase domain and an 8-kDa amino-terminal lyase domain. The polymerase domain is structurally organized into functionally distinct subdomains referred to as DNA binding (D, residues 90 -150), catalytic (C, residues 151-260), and nascent base pair binding (N, residues 261-335) subdomains (12). These are referred to as thumb, palm, and fingers subdomains, respectively, for righthanded polymerases 3 that possess a nonhomologous catalytic (palm) subdomain (13). As noted previously, however (14), the catalytic participants (DNA, dNTP, and metals) can be superimposed indicating that all DNA polymerases utilize a common two-metal ion mechanism for nucleotidyl transfer (15). A structure of a complete precatalytic complex of pol ␤ provides compelling evidence for the two-metal chemical mechanism (16).
Crystallographic structures of pol ␤, a family X polymerase, with active site mismatched nascent base pairs indicate that the templating (i.e. coding) base is repositioned upstream of the 3 The hand-like subdomain nomenclature for DNA polymerases belonging to the X family is ambiguous (11). This is because the original subdomain assignment for pol ␤ was based on structural similarity between the palm subdomains of other polymerase families (13). This assignment results in the fingers and thumb subdomains of X family members being functionally opposite to that of other polymerase families; members of the X family are left-handed rather than right-handed DNA polymerases. The structural characterization of most X family DNA polymerases uses this original nomenclature so that careful attention to the specific nomenclature must be recognized before extrapolating between polymerase families.
template binding pocket thereby diminishing interactions with the incorrect incoming nucleotide (17). This effectively produces an abasic site because the templating pocket is devoid of an instructional nucleotide base. However, the templating pocket is not empty; an arginine residue (e.g. Arg-283) occupies the space vacated by the template nucleotide. This arginine residue has previously been shown to modulate fidelity (base substitution and frameshift) (18 -20), misinsertion specificity (19), and catalytic efficiency of pol ␤ (18,21). DNA polymerases prone to produce deletion errors can bypass abasic DNA lesions by utilizing the 5Ј-templating base. When filling a 5-nucleotide DNA gap with a central (position 3) tetrahydrofuran lesion, pol ␤ does not obey the A-rule and inserts a nucleotide complementary to the first downstream templating base (22). When a downstream templating nucleotide, rather than the coding templating base, directs nucleotide insertion, the misaligned template strand is stabilized by dNTP binding. If realignment does not occur after insertion, the product is a deletion error. If realignment does occur, the result is a base substitution error. Two crystallographic structures of a Y family DNA polymerase with a dNTP-stabilized intermediate have been solved (8,23).
In this study, we analyze the kinetics of pol ␤ insertion opposite an abasic site situated in a single-nucleotide DNA gap. Employing a single-nucleotide gapped substrate prevents downstream templating nucleotides from directing DNA synthesis. Additionally, we analyze how nucleotide specificity opposite an abasic site is modulated by divalent metal ions and a mutant pol ␤ known to control dNTP selection. Finally, crystallographic structures of pol ␤ with THF-gapped DNA (binary complex) and with a dATP nonhydrolyzable analog, dAMPCPP (ternary complex), are reported.
Mutagenesis of the Human DNA Polymerase ␤ Gene-The construction and expression of the R283A mutant of human pol ␤ have been described (18). To ensure that the resulting pol ␤ gene contained the desired change, the entire coding sequence of the mutant was confirmed by DNA sequence analysis. The mutant was cloned into pWL-11 (24), a bacterial expression plasmid containing the PL promoter, and overexpressed in Escherichia coli TAP56 cells.
Protein Purification-Wild-type and mutant proteins were purified as described previously (25). Enzyme concentrations were determined by absorbance at 280 nm (⑀ ϭ 23,380 M Ϫ1 cm Ϫ1 ). DNA Preparation-A 34-mer oligonucleotide DNA substrate containing a single-nucleotide gap was prepared by annealing three gel-purified oligonucleotides (IDT, Coralville, IA) to create a single-nucleotide gap at position 16. Each oligonucleotide was resuspended 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 in a PCR thermocycler by incubating a solution of 10 M primer with 12 M each of downstream and template oligonucleotides at 95°C for 5 min followed by 30 min at 65°C and then slow cooling (1°C/min to 10°C). The sequence of the gapped DNA substrate was as follows: primer, 5Ј-CTG CAG CTG ATG CGC-3Ј; downstream oligonucleotide, 5Ј-GTA CGG ATC CCC GGG TAC-3Ј; and template, 3Ј-GAC GTC GAC TAC GCG XCA TGC CTA GGG GCC CAT G-5Ј (X is the template residue in the gap). The primer was 5Ј-labeled with [␥-32 P]ATP using Optikinase (United States Biochemical Corp.), and the free radioactive ATP was removed with either Micro-Spin G-25 or Bio-Spin 6 columns. The downstream oligonucleotide was synthesized with a 5Ј-phosphate.
Kinetic Assays-Steady-state kinetic parameters for singlenucleotide gap-filling reactions were determined by initial velocity measurements as described previously (12). Unless noted otherwise, enzyme activities were determined using a standard reaction mixture (20 l) containing 50 mM Tris-HCl, pH 7.4 (room temperature or 37°C), 100 mM KCl, 5 mM MgCl 2 or MnCl 2 , and 200 nM single-nucleotide gapped DNA. High temperature reactions (i.e. 37°C) were supplemented with 2 g of bovine serum albumin. The core DNA sequence was identical to that used for crystallization. Assays were initiated by mixing enzyme/DNA with dNTP/metal (1:1, v/v). Enzyme concentrations (2-50 nM) 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 0.3 M EDTA and mixed with an equal volume of formamide dye. The substrates and products were separated on 15% dena- turing (8 M urea) polyacrylamide gels and quantified in the dried gels by phosphorimagery. Steady-state kinetic parameters were determined by fitting the rate data to the Michaelis equation.
Crystallization of DNA Polymerase ␤ Substrate Complexes-The DNA substrate consisted of a 16-mer template, a complementary 10-mer primer strand, and a 5-mer downstream oligonucleotide. The annealed 10-mer primer creates a one-nucleotide gap with a templating THF or G residue. The downstream oligonucleotide is 5Ј-phosphorylated. The template sequence was 5Ј-CCG AC(THF or G) GCG CAT CAG C-3Ј. Oligonucleotides were dissolved in 20 mM MgCl 2 , 100 mM Tris-HCl, pH 7.5. Each set of template, primer, and downstream oligonucleotides was mixed in a 1:1:1 ratio and annealed using a PCR thermocycler by heating for 10 min at 90°C and cooling to 4°C (1°C/min) resulting in a 1 mM mixture of gapped duplex DNA. This solution was then mixed with an equal volume of pol ␤ at 4°C, and the mixture was warmed to 35°C and gradually cooled to 4°C.
DNA Polymerase ␤-DNA complexes were crystallized by sitting-drop vapor diffusion. The crystallization buffer was 16% PEG-3350, 350 mM sodium acetate, and 50 mM imidazole, pH 7.5. Drops were incubated at 18°C and streak-seeded after 1 day. Crystals grew in ϳ2-4 days after seeding. The ternary complex was obtained by soaking crystals of binary one-nucleotide (THF) gapped DNA complexes in artificial mother liquor with 50 mM MnCl 2 , 2 mM dAMPCPP, 20% PEG-3350, and 12% ethylene glycol and then flash-frozen to 100 K in a nitrogen stream. All crystals belong to the space group P2 1 .
Data Collection and Structure Determination-Data were collected on a Saturn 92 CCD detector system mounted on a MicroMax-007HF (Rigaku Corp.) rotating anode generator. Data were integrated and reduced for structure refinement with HKL2000 software (26).
Structures were determined by molecular replacement with previously determined structures of pol ␤ complexed with single-nucleotide gapped DNA (binary complex, PDB code 1BPX) (27) and a mismatched complex with an incoming dAMPCPP (ternary complex, PDB code 3C2M) (17). The crystal structures have similar lattices and are sufficiently isomorphous to determine the molecular replacement model position solely by rigid body refinement using CNS (28). Further refinement and model building were carried out using CNS and O (29). Positional refinement was performed using minimization and torsion angle dynamics. Individual B-factor refinement was performed using the maximum likelihood target function with restrained target values for bond lengths and angles. The molecular graphics images were prepared in Chimera (30).

RESULTS
Nontemplated DNA Polymerase ␤ Insertion Specificity-Structural characterization of pol ␤ bound to blunt-end DNA suggested that dATP could be preferentially added in a nontemplated reaction (31). More recently, it was shown that dNTP selection/insertion "opposite" a synthetic abasic site in the context of surrounding single-stranded DNA was influenced by the identity of the adjacent downstream template base. DNA Polymerase ␤ preferentially inserted a nucleotide that was complementary to the downstream templating base indicating that the incoming dNTP could stabilize a template strand misalignment (22). To discourage such template realignments, we examined the dNTP insertion specificity of pol ␤ on a single-nucleotide gapped substrate with a synthetic abasic site lesion analog, tetrahydrofuran, in the gap and the downstream templating bases participating in Watson-Crick hydrogen bonding (i.e. duplex DNA). At room temperature, insertion opposite the nontemplating lesion was very poor, and the reaction could not be saturated. At high dNTP concentrations (Ͼ1 mM), substrate inhibition was evident (data not shown) precluding determination of individual kinetic parameters. However, the dependence of the rate of the reaction on low dNTP concentrations (Ͻ ϽK m ) permitted determination of the catalytic efficiency or specificity constant (k cat /K m ). With gapped DNA, pol ␤ preferentially inserted dATP opposite the synthetic abasic site as follows: dATP Ͼ Ͼ dGTP ϳ dCTP Ͼ Ͼ dTTP (Table 1 and Fig. 2A).
When the reaction temperature was increased to 37°C, the catalytic efficiency of the reactions increased ϳ10-fold. This was enough to get estimates of the insertion rate and binding affinities for the incoming nucleotides. As we have noted previously (12,32), the slow rate of nucleotide insertion simplifies the interpretation of the kinetic constants determined by a steady-state approach. In this situation, K m ϭ K d (equilibrium binding constant for the incoming nucleotide) and k cat ϭ k pol (intrinsic rate constant for nucleotide insertion). The insertion specificity at 37°C was identical to that determined at room temperature. For nontemplated insertion, the natural incoming dNTPs bound weakly (K d Ն 350 M) and were inserted slowly (Table 1). Relative to a gapped substrate with dT in the gap (33), this represents a 150-and 75-fold decrease in the rate of insertion and binding affinity for dATP, 4 respectively. Thus, dATP is inserted opposite THF over 10 4 -fold less efficiently than opposite dT. a When product release is rapid, steady-state kinetic parameters k cat and K m,dNTP are equivalent to k pol and K d (12,61). b Room temperature varied from 21 to 23°C. c Specific kinetic constants were not determined due to weak dNTP binding. The reaction rate was not saturable at experimentally practical dNTP concentrations. At high concentrations (Ͼ1000 M), the reaction rate was inhibited. Thus, catalytic efficiency was determined from the slope of the dNTP concentration dependence of the observed rate at low dNTP concentrations. When ͓dNTP͔ Ͻ Ͻ K m,dNTP , k obs ϭ ͓dNTP͔ (k cat /K m,dNTP ).
The dNTP analog 5-NITP (Fig. 1B) is inserted opposite an abasic site with a catalytic efficiency 1000-fold greater than for dATP with T4 DNA polymerase, a B-family polymerase. This has been attributed to the additional -stacking provided by the 5-nitroindole (34). DNA Polymerase ␤ inserted this analog with efficiency similar to that for dATP ( Table 1).
Effect of Mn 2ϩ on Insertion Specificity-Manganese substitution for magnesium generally lowers DNA polymerase fidelity (35). In contrast, in the absence of a templating base, insertion specificity is unaltered when magnesium is replaced by manganese (Fig. 2B). More importantly, catalytic efficiencies are increased ϳ60-fold in the presence of manganese ( Table  2). As observed for the binding affinities for incorrect incoming nucleotides (17), the binding affinities for the incoming nucleotides opposite an abasic site were also significantly increased in the presence of manganese (K d Ͻ 50 M).
Effect of Alanine Substitution for Arginine 283 on Insertion Specificity Opposite an Abasic Site-In the closed polymerase conformation observed in ternary substrate complex structures, Arg-283 interacts with the minor groove edge of the templating base (16,17). Alanine substitution at this position dramatically lowers the efficiency for correct, but not incorrect, nucleotide insertion and the fidelity of DNA synthesis (18,21). To analyze the effect of this altered side chain on insertion specificity opposite an abasic site, we determined kinetic parameters for insertion of the purine nucleoside triphosphates (dATP and dGTP) ( Table 1). As noted above, wild-type enzyme preferentially inserts dATP relative to dGTP. The R283A mutant inserts dATP into the nontemplated gapped DNA 45-fold less efficiently than wild-type enzyme, whereas dGTP insertion is hardly affected (Fig. 3). The result is that dGTP insertion efficiency is now 3-fold greater than dATP. This altered specificity is nearly completely due to the diminished insertion rate (k pol ) of dATP with the mutant enzyme.
Crystallographic Structures of Substrate Complexes with an Abasic Site-The tight binding of dATP in the presence of manganese suggested that a ternary substrate complex could be formed for structure determination. A 16-mer template strand was annealed with a 10-mer primer strand and a 5-mer 5Ј-phosphorylated downstream oligonucleotide to create a one-nucleotide gapped DNA substrate. The template strand had either a guanine or tetrahydrofuran residue in the gap. Crystals were obtained and analyzed ( Table 3). The binary single-nucleotide DNA complex with a templating guanine is similar to that reported previously (27), but the new data set provides higher resolution information (2.00 Å). Likewise, the binary complex crystals with THF in the gap diffracted to 2.00 Å. Comparison of these two binary complexes indicated that the overall fold of the proteins was identical (root mean square deviation ϭ 0.20 Å; all 327 C ϰ ). Similarly, the DNAs superimpose very well (root mean square deviation ϭ 0.21 Å; upstream and downstream duplex DNA, 610 atoms). As observed with other pol ␤ DNA binary   complex structures, the polymerase is in an open conformation where the N-subdomain is positioned away from the nascent base pair binding pocket (Fig. 4). Previous attempts to obtain structures with incorrect incoming nucleotides resulted in hydrolysis of the incoming nucleotide (36). This problem was circumvented by using a nonhydrolyzable dNTP analog where the bridging oxygen between the ␣and ␤-phosphates was replaced with a methylene group. Crystals of the binary DNA complex, with THF in the gap, were grown and dAMPCPP soaked into the crystal. This crystal diffracted at 2.2 Å, and the structure indicated that the polymerase had adopted a closed conformation (Fig. 4) where the lyase domain and the N-subdomain move toward the incoming nucleotide forming a more compact structure.
Soaking dAMPCPP into crystals of binary single-nucleotide gapped DNA with G or C in the gap resulted in similar closed polymerase conformations (17). To accommodate the mispair in the closed polymerase conformation, the template strand moved upstream so that the coding templating base was moved out of the nascent base pair binding pocket. The structure with THF in the templating position (n) shows a similar upstream shift (ϳ3 Å) in the template strand (Fig. 5A). This new template position is similar to the position of the n Ϫ 1 template nucleotide when a correct base pair is in the nascent base pair binding pocket (Fig. 5B).
The position and conformation of the incoming dAMPCPP are similar to that observed for this analog in the dAMPCPP-dG mismatch structure (Fig. 5A) and dUMPNPP when correctly paired (i.e. template dA) in the nascent base pair binding pocket (Fig. 5B). In contrast to the mismatch ternary complex structure, the primer terminus O3Ј may coordinate the catalytic Mn 2ϩ , and its modeled position relative to P␣ of the incoming nucleotide suggests good geometry for nucleotidyl transfer (d ϭ 3.4 Å and O3Ј(primer terminus)ϪP␣(dNTP)Ϫ CH 2 (dNTP) ϭ 174°). However, the electron density for the primer terminus is poor (B-factor ϭ 72 Å 2 ) indicating that the position of O3Ј is tenuous. Both active site divalent metals on the other hand appear correctly positioned (Fig. 5) and have low B-factors (ϳ27 Å 2 ). The inner coordination sphere of the catalytic metal is incomplete, missing an H 2 O ligand and possibly O3Ј of the primer terminus. Additionally, the coordination distances of both active site metals are ϳ0.1 Å longer than those observed in the ternary complex structures with active site mismatches (17). The base of the primer terminus is still Watson-Crick hydrogen bonded to its templating partner. Because the template strand has moved upstream, the primer terminal base tilts upstream resulting in a loss of stacking interactions with the incoming nucleotide and a minor groove hydrogen bond with Tyr-271. Accordingly, the primer terminus exhibits an elevated B-factor (ϳ70 Å 2 , DNA avg ϭ 32 Å 2 , see Table 3) suggesting that the 3Ј terminus is less stable.
Unique Conformation of Arg-283-In the closed ternary substrate complex with a correct incoming nucleotide, Arg-283 interacts with the minor groove edge of the templating nucleotide (Fig. 5B). Alanine substitution results in a mutant enzyme  with a much lower efficiency for correct nucleotide insertion, although incorrect insertion is hardly affected (21). Alternate residues at this position exhibit alternate misinsertion specificities (19). For example, enzymes with arginine (wild type) or lysine (R283K) at residue 283 show similar misinsertion specificities for dGTP opposite dA and dT; however, the enzyme with an alanine substitution at this position (R283A) exhibits an elevated misinsertion frequency for these misinsertions. The preference for dATP insertion opposite an abasic site is in part due to Arg-283. Alanine substitution resulted in a decrease in catalytic efficiency for dATP, but not dGTP, so that the preference for dATP was lost (Fig. 3). Arg-283 is in identical positions in the open DNA binary complexes with or without a templating nucleotide (i.e. template G or THF; not shown). In the ternary substrate complex without a templating base, Arg-283 is observed in an unexpected position not observed previously. In this case, the arginine side chain is pointed away from the incoming dAMPCPP beneath the templating sugar residue (Fig. 5A and Fig. 6A). The N⑀ of the arginine side chain is within hydrogen bonding distance to the phosphate backbone. The implications of this observation are discussed below.

DISCUSSION
Misinsertion of dATP-Crystallographic structures of premisinsertion complexes of pol ␤ binding an incorrect dATP analog indicate that the templating base (dG or dC) was removed from the coding template position suggesting that misinsertion may occur through an "abasic site" intermediate (17). It had previously been noted that the misinsertion specificity of pol ␤ in a variety of DNA sequence contexts was similar to that observed for DNA polymerases where preferential purine insertion, usually dATP, occurred opposite a noninstructional abasic site (37). The structure of a ternary substrate complex with dAMPCPP situated opposite a synthetic AP site is consistent with this proposal. These structures indicate that the enzyme is in a closed conformation (Fig. 4) and that the template strand has moved upstream (Fig. 5). Even in the absence of the templating base, closure of the N-subdomain results in an upstream shift in the template strand (Fig. 5). This indicates that re-positioning of the N-subdomain to a "closed" position induces template strand slippage and that interactions with the templating base are not required for this upstream shift.
The suggestion that misinsertion of dATP opposite dG and dC occurs in a manner similar to that of an abasic site is also consistent with the kinetics of gap filling. The efficiency of dATP insertion opposite an AP site is similar to that measured for misinsertion opposite dG and dC (Fig. 7). In contrast, the efficiency of dATP misinsertion opposite dA is considerably lower than that observed opposite an AP site (33-fold). Because the decreased efficiency is due to loss in the rate constant for insertion (i.e. K d values for dATP are similar), the results suggest that the templating dA, and/or protein contacts, profoundly distorts the active site geometry in the vicinity of the primer terminus (i.e. primer 3Ј-OH and/or catalytic metal coordination) thereby perturbing misinsertion of dATP.
Active Site Arginine Influences Insertion Specificity-Even though the synthetic abasic site analog tetrahydrofuran does not offer hydrogen bonding potential, polymerase side chains that occupy the template-binding pocket could influence insertion specificity. Arg-283 is situated on ␣-helix N of the N-subdomain. This subdomain closes around the nascent base pair binding pocket after dNTP binding (Fig. 4). In the DNA binary complex (open polymerase conformation), Arg-283 is solventexposed. Upon binding a correct dNTP, the closed polymerase conformation promotes Arg-283 interactions with the minor groove edge of the templating strand (27,38). In contrast, this arginine occupies the template base binding pocket when an incorrect nucleotide binds to pol ␤ (Fig. 5A) forming van der Waals interactions with the incorrect templating base re-positioned upstream (17). Thus, the coding template base has been removed and replaced with a protein side chain. In addition, Arg-283 can hydrogen bond to a misinserted nucleotide (i.e. a mismatched primer terminus situated in the nascent base pair binding pocket where the polymerase has not translocated after misinsertion) (39). Thus, the side chain of Arg-283 dynamically In both instances, the template residue (n) is shifted upstream ϳ3 Å vacating the templating pocket. The incoming dAMPCPP and Mn 2ϩ in these structures superimpose well. In contrast, the 3Ј-OH (O3Ј) of the primer terminus is better positioned in the structure with the abasic site than in the case of a nascent mismatch. Likewise, Arg-283 (R283) is observed in a new conformation not observed previously. This side chain occupies the template base binding pocket vacated in the mismatch structure but lies under the template backbone in the absence of a template base. The solid arrow indicates the upstream direction from the active site, and the dashed arrow indicates the downstream direction. B, active site comparison between superimposed ternary complex structures with a correct active site base pair (dA-dUMPNPP, PDB code 2FMS; gray carbons) (16) and nontemplating abasic site (magenta carbons). The template THF (magenta, n) is moved upstream of the active site and positioned approximately where the n Ϫ 1 (template nucleotide opposite the primer terminus) nucleotide is found with a correct nascent base pair. In this situation, Arg-283 interacts with the minor groove edge of the templating base. The polymerase domain is in a closed conformation as judged by the position of ␣N. NOVEMBER 13, 2009 • VOLUME 284 • NUMBER 46 responds to catalytic cycling altering its interactions upon ligand binding and DNA sequence. The arginine side chain is well suited to provide both van der Waals interactions and hydrogen bonding. Indeed, Arg-324 side chain of deoxycytidyltransferase (Rev1; Y family polymerase) provides two hydrogen bond acceptors for the preferential insertion of dCTP (40). In this case, the templating base is displaced, and Arg-324 occupies the templating pocket.

Side Chain Modulation of the A-rule
It is not unusual for DNA polymerases to occupy the templating base-binding pocket with a protein side chain. It has long been known that A family DNA polymerases (e.g. Klentaq (41,42)) insert an aromatic side chain (phenylalanine or tyrosine) that stacks with the upstream template base in binary DNA complexes displacing the coding template base into an extra-helical major groove position. During correct nucleotide insertion, this side chain is replaced with the coding template base. In the binary pol (X family) single-nucleotide gapped DNA complex, the closed conformation of the N-subdomain inserts Arg-517 (equivalent to pol ␤ Arg-283) in the templating pocket (43) repositioning the templating base upstream like that observed in the pol ␤ mismatch preinsertion ternary complex (17). Arg-517 repositions upon binding a correct nucleotide to interact with the DNA minor groove of the template strand, like Arg-283, when a correct nucleotide binds. In this case, the coding template base moves downstream into the template-binding pocket.
The wild-type enzyme prefers to insert dATP opposite an AP site under varying reaction conditions (temperature and activating divalent metal ion; Tables 1 and 2). The R283A mutant, however, has lost this preference through a differential loss in dATP insertion efficiency, although dGTP insertion is hardly affected (Fig. 3). This result suggests that Arg-283 facilitates dATP insertion opposite an AP site so that loss of this side chain modulates insertion in the absence of a templating base. Surprisingly, the structure of the ternary substrate complex with a synthetic AP site indicates that Arg-283 is in a conformation not observed previously. The guanidinium group sits under the template strand and is pointed away from the incoming dAMPCPP and is within hydrogen bonding distance with the phosphate backbone of the AP site ( Fig. 5A and Fig. 6A).
The loss of dATP insertion efficiency opposite an AP site for the R283A mutant is primarily due to the diminished rate of nucleotide insertion rather than a change in the binding affinity of the incoming purine (Table 1; Fig. 3, legend). This implies that insertion occurs from a poor position or that the concentration of "activated" mutant enzyme is diminished with an incoming dATP. Arg-283 has been suggested to participate in a FIGURE 6. Alternate side chain conformations for key residues involved in active site signaling. A, Arg-283 is observed in the ternary complex structure in a conformation not observed previously (PDB code 3ISD). In the absence of a templating base (n), the arginine side chain points away from the incoming dAMPCPP and makes a single hydrogen bond with the templating phosphate backbone (dotted green line). The immediate upstream and downstream templating nucleotides are illustrated (n Ϫ 1 and n ϩ 1, respectively). Furthermore, the upstream and downstream directions of the template strand are indicated with a solid and dashed arrow, respectively. An F o Ϫ F c simulated annealing electron density omit map (blue) contoured at 2.0 showing electron density corresponding to dAMPCPP, the synthetic abasic site, and Arg-283 is shown. B, F o Ϫ F c simulated annealing electron density omit map (blue) contoured at 3.0 showing electron density corresponding to Phe-272 in the ternary substrate complex with a synthetic abasic site (magenta carbons). Tyr-271 is also shown. Phe-272 was modeled in two conformations (C1 and C2 with 50% occupancy). These residues are compared with a superimposed binary (yellow) (PDB code 3ISB) and ternary substrate complex with a correct base pair (gray; PDB IB 2FMS) (16). Thus, Tyr-271 is observed in an intermediate position between the inactive binary complex and active ternary complex, whereas Phe-272 appears to equilibrate between the two positions. signaling cascade that informs the active site of the position of the N-subdomain (27, 33, 44 -47). Arg-283 stabilizes the closed polymerase conformation that in turn influences side chain conformations of residues in the C-and N-subdomains. Most importantly, these interactions radiate to Asp-192 that coordinates both active site metals. The structural results demonstrate that Arg-283 also responds to the identity of the incoming nucleotide as well as whether the coding template base exists and/or is in the proper position. In addition, Tyr-271 and Phe-272 are observed to be in intermediate or alternate conformations, respectively, in the ternary complex with an abasic site relative to that observed in binary DNA and ternary complexes with a correct base pair (Fig. 6B). Thus, although Arg-283 may not directly interact with the incoming dATP, the altered active site dynamics when alanine replaces the arginine at residue 283 may be expected to translate to a diminished rate of insertion. Such an hypothesis suggests that active site signaling would be sensitive to the identity of the incoming nucleotide. We have shown previously that polymerase/template base interactions are sensitive to the identity of the incoming nucleotide (12).
Polymerase-dependent dATP Insertion Opposite an AP Site-DNA polymerases generally prefer to insert dATP opposite an AP site. In contrast, the Y family pol shows limited discrimination at an abasic site, inserting all four nucleotides with approximately the same low efficiency (48). It should be noted that in all of these cases, the efficiency for insertion opposite an AP site is very poor. Although it is difficult to compare the efficiencies of dATP insertion opposite an AP site for different polymerases due to varying reaction conditions, it is useful to qualitatively examine the propensity alternate DNA polymerases have to insert a nucleotide opposite a synthetic AP site. Y family DNA polymerases have been implicated in the replication bypass of bulky DNA lesions as well as the nonbulky AP site lesion. These polymerases insert dATP opposite an AP site with efficiencies of ϳ50 -200 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 (48 -50). In contrast, exonuclease-deficient variants of T4 DNA polymerase and RB69 (B family members) and the E. coli pol I Klenow fragment (A family member) exhibit considerably higher dATP insertion efficiencies (e.g. T4 exo Ϫ , 4300 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 ) (51-53). DNA lesion bypass requires both nucleotide insertion opposite the lesion and extension of this product. Although high fidelity DNA polymerases can efficiently, relative to other DNA polymerases, insert dATP opposite a simple lesion such as an AP site, its associated proofreading exonuclease activity would quickly remove a nontemplated insertion because extension is a kinetically challenging reaction (i.e. primer terminus is not base paired) thereby resulting in a futile cycle of dATP insertion and removal. DNA Polymerase ␤ exhibits an efficiency similar to that of Y family DNA polymerases (Table 1; ϳ150 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 ).
The ternary substrate complex of pol ␤ with an incoming nucleoside triphosphate opposite an abasic site differs from previously determined structures of representative members of the B (9) and Y family (10) DNA polymerases. In those structures, the incoming nucleotide stacks well within the DNA helix, and the template strand does not exhibit an upstream shift. For pol ␤, although the sugar and triphosphate bind in a similar manner to that observed when a correct templating base is present, the adenine base is severely tilted relative to the primer terminus. This is due to the base of the primer terminus rotating to continue to hydrogen bond to its templating partner that has shifted upstream. The crystallographic structure of the ternary complex indicates that the primer terminus is in an unstable position as evidenced by its poor electron density and elevated B-factor (i.e. 72 Å 2 ). In the binary complex, the B-factor for the primer terminus is significantly lower (26 Å 2 ) suggesting that binding of dAMPCPP destabilizes the primer terminus.
Influence of DNA Sequence on Insertion Specificity Opposite an AP Site-It has long been recognized that the conformations and dynamics of an AP site, in the context of duplex DNA, are dependent on the nucleotide opposite this site and the surrounding DNA sequence (54,55). When pyrimidines are positioned opposite an AP site, the local structure is more perturbed than when purines are opposite this lesion. Additionally, a pyrimidine opposite an AP site permits collapse of this site expelling the pyrimidine and AP site to an extra-helical position and permitting the adjacent base pairs to stack with one another (55). In the case of a template-primer junction, the DNA polymerase could capture an intermediate structure where the AP site is extra-helical due to improved stacking interaction with the downstream template base and the template base opposite the primer terminus. This could be a precursor to the intermediate structure captured with Dpo4 DNA polymerase of the dNTP-stabilized intermediate (8).
Biological Significance-Abasic sites are common nonbulky DNA lesions that are spontaneously generated or may arise as intermediates during base excision repair of simple base lesions. As noted above, replicative DNA polymerases have difficulty bypassing this lesion that can thereby lead to the accumulation of cytotoxic intermediates. To circumvent this replication block, the replication machinery recruits specialized DNA polymerases. Although insertion opposite an abasic site for most DNA polymerases without a proofreading activity is straightforward, further extension relies on the ability of the polymerase to extend a primer terminus that is not base paired. DNA Polymerase ␤ extends such an intermediate 400,000-fold less efficiently than when the primer terminus is properly base paired (32). Accordingly, efficient bypass of an AP site may require two distinct DNA polymerases (48).
The removal of the templating base from a coding position (e.g. extra-helical or upstream shift) to produce an abasic site could be a general strategy employed by DNA polymerases to bypass bulky DNA lesions. If the bulky lesion could be expelled into the major groove, steric constraints in the nascent base pair binding pocket would be relieved creating a noninstructional "coding" pocket that would permit nucleotide insertion, albeit at a much lower efficiency than opposite its Watson-Crick partner. It appears that an exonuclease-deficient mutant of T7 DNA polymerase (A family) can utilize this mechanism to bypass a cis-syn thymine dimer (56). A crystallographic structure of the T7 DNA polymerase binary DNA complex with the 3Ј-thymine in the templating position shows that the thyminedimer is rotated out of the active site (57). Likewise, it has been postulated that pol ␤ could utilize this mechanism to bypass benzo[c]phenanthrene diol epoxide-adducted guanine (58).
Nucleotide Attributes and Polymerase Specificity-DNA polymerases generally follow the A-rule when encountering an AP site. As noted above, however, the efficiency by which they insert dATP varies by several orders of magnitude indicating that the interactions governing insertion are highly dependent on the specific polymerase. At the same time, the preferential insertion of dATP indicates that there must be some property of the base that promotes insertion relative to the other natural nucleotides. NMR characterization of DNA duplexes with each of the four nucleotides opposite an abasic site indicates that adenosine opposite an abasic site was the least perturbing and stacked well within the helix (55,59). The physical properties of adenosine that impart this intra-helical stability are less certain. Compared with the other natural bases, adenine exhibits the lowest dipole moment and can contribute significant stacking interactions with an adjacent base pair (60). T4 DNA polymerase inserts the dNTP analog 5-NITP 1000 times more efficiently than dATP. This has been attributed to the additional -electron density provided by the 5-nitroindole (34). DNA polymerase ␤ inserted this analog with efficiency similar to that for dATP (Table 1). Klenow fragment, an A family polymerase, inserted 5-NITP ϳ6-fold greater than dATP (53). Taken together, these results highlight differences in physical properties of the nascent base pair binding pockets of these DNA polymerases.
Concluding Remarks-Although insertion opposite an AP site is generally referred to as nontemplated, the strong possibility that the polymerase contributes hydrogen bonding groups as well as geometric constraints for the incoming nucleotide indicates that protein templating occurs. The similar kinetics for dATP insertion opposite the AP site and misinsertion opposite dG or dC correlate with the similar global structure of the ternary complexes. It remains to be seen what the structural characterization of mismatch structures with alternate incoming bases might reveal. For example, pol ␤ misinserts dGTP opposite thymidine much more efficiently than opposite an abasic site; 87 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 (17) and 1 ϫ 10 Ϫ6 M Ϫ1 s Ϫ1 (Table 1, 22°C), respectively. Thus, it is expected that either the templating thymidine or a protein side chain facilitates dGTP misinsertion.