The hydrophobic hinge region of rat DNA polymerase beta is critical for substrate binding pocket geometry.

The hydrophobic hinge of DNA polymerase beta facilitates closing and stabilization of the enzyme once the nucleotide substrate has bound. Alteration of the hydrophobic nature of the hinge by the introduction of a hydrophilic glutamine residue in place of isoleucine 260 results in an inaccurate polymerase. The kinetic basis of infidelity is lack of discrimination during the binding of substrate. The I260Q polymerase beta variant has lower affinity than wild type enzyme for the correct substrate and much higher affinity for the incorrect substrate. Our results demonstrate that the hinge is important for formation of the substrate binding pocket. Our results are also consistent with the interpretation that DNA polymerase beta discriminates the correct from incorrect substrate during the binding step.

DNA damage poses a daily threat to the survival of eukaryotic cells. This damage, when left unrepaired, can result in mutations, some of which lead to cancer. Among the repair processes that eukaryotic cells have evolved to ensure prompt repair of DNA damage, base excision repair is the pathway responsible for removal of 10,000 lesions per cell per day that result from oxidative and alkylation damage (1)(2)(3)(4).
Like other repair processes, base excision repair consists of a sequence of finely tuned events that starts with a specific glycosylase recognizing the damaged base (5). Upon removal of the base by the glycosylase, an apurinic/apyrimidinic exonuclease excises the sugar-phosphate backbone and leaves a gap in the DNA (6,7). This gap is filled by DNA polymerase ␤ (Pol␤) 1 in most base excision repair events. In short patch base excision repair, where only a single base is filled in, Pol␤ is able to remove the 5Ј-deoxyribose phosphate with its 5Ј-deoxyribose phosphate lyase activity (8) thus leaving a substrate ready for completion of repair by a DNA ligase (9).
Pol␤ has been associated with a variety of other cellular processes including meiosis where it was found to localize to synaptonemal complexes (10,11). Pol␤ is necessary for development as evidenced by the fact that a homozygous deletion in mice is lethal (12).
Although Pol␤ is considered a fairly low fidelity polymerase (13), it performs nucleotide insertion in a template-directed manner where it is able to discriminate and select the correct dNTP from a pool of very similar nucleotide substrates. The inherently lower fidelity of this enzyme can be further attributed to the fact that it does not have any intrinsic proofreading or nuclease activities. However, the lack of these functions makes it ideal for a direct study of incorporation fidelity. Additionally, several crystal structures of DNA polymerase ␤ are available including that of the enzyme bound to the DNA substrate alone, termed the "open" conformation, and that of the enzyme in complex with both the DNA and dNTP substrates, known as the "closed" conformation (14 -17). The availability of these structures further aids us in understanding fidelity and allows us to relate the functional and structural information. Pol␤ is a relatively small enzyme that is easy to purify. Cumulatively, these features make Pol␤ the ideal candidate for study of enzyme activity and fidelity.
Recently, a variety of Pol␤ mutants have been found associated with numerous types of human cancers. Of the 149 tumor samples examined, 30% contained mutations in DNA polymerase ␤ (18). Given the known role of polymerase ␤ in DNA repair and the possible role in meiosis as well as its implicated roles in small-scale replicative synthesis, it is of crucial importance to understand the types and nature of mutations that could lead to infidelity. This infidelity can result in the accumulation of mutations that could, in turn, result in cancer and/or abnormalities in cellular processes.
Here we present the biochemical characterization of the I260Q variant of Pol␤. Residue Ile-260 is located in the hinge region of DNA polymerase ␤, which is the region that underlies the area between the fingers and the palm domains of the enzyme (16). Ile-260 appears to be of particular significance for its position at the bottom of the hinge, directly underneath the plane where the DNA substrate is positioned. We previously conducted a genetic screen that identified I260Q as a high activity mutant with low fidelity (19). I260Q was shown to have activity levels comparable with wild type in the recA718polA12 heterologous complementation system (20). Because wild type Pol␤ was shown to substitute for Pol I of Escherichia coli in the joining of Okazaki fragments, this complementation system assesses the catalytic activity of Pol␤ mutants that are expressed in the recA718polA12 strain. Furthermore, in the same system, I260Q reverted the trpE56 mutation at a frequency 60-fold higher than wild type indicating that this mutant has a strong in vivo mutator phenotype (21). We confirmed these findings in vitro and found I260Q to have wild type activity levels in a primer extension assay. We also used the in vitro primer extension assay to show that the mutator activity is an intrinsic characteristic of I260Q (19).
In the study reported here, we examined the kinetic and structural basis for the mutator phenotype of the I260Q variant and found that I260Q is able to form all three sets of mispairs opposite both a purine and a pyrimidine. Moreover, the kinetic basis for misincorporations performed by this mutant is in the binding affinity for the dNTP substrate where I260Q has a lower affinity for the correct dNTP and higher affinity for the incorrect dNTP as compared with wild type. Our results suggest that the hinge of Pol␤ is important for discrimination during the binding of nucleotide substrate.

EXPERIMENTAL PROCEDURES
Bacterial Strains-The strain BL21 DE3 with the genotype F Ϫ ompT hsdSB(r b Ϫ m b ) gal dcm (DE3) was used for protein expression. The wild type enzyme as well as the I260Q variant were expressed from a pET28a vector as previously described (22,23).
Chemicals and Reagents-Ultrapure deoxynucleoside triphosphates, ATP, and [␥-32 P]ATP were purchased from New England Biolabs, Sigma, and Amersham Biosciences, respectively. Oligonucleotides were synthesized by the Keck Molecular Biology Center at Yale University and purified by PAGE.
Purification of Pol␤ Wild Type and the I260Q Variant-The I260Q mutant was generated by site-directed mutagenesis from the wild type construct in the pET28a vector (Novagen) as previously described (19) to generate an amino-terminal His 6 tag. The proteins were expressed and purified as previously described (22) using a fast liquid chromatography system (AKTA FPLC; Amersham Biosciences) to perform an affinity purification step followed by ion-exchange purification. Based on a Coomassie Blue-stained SDS-PAGE gel, these proteins were Ͼ90% homogeneous (data not shown). Protein concentrations were calculated based on ⑀ 280 ϭ 21,200 M Ϫ1 cm Ϫ1 and a molecular mass of 40 kDa.
DNA Substrate Preparation for Kinetic Studies-The single base pair-gapped DNA substrates with a 5Ј-phosphate on the downstream oligonucleotide were used for the misincorporation studies. The sequences of the substrates are shown in Table I. The primer oligonucleotide was labeled at the 5Ј end using T4 polynucleotide kinase (New England Biolabs) and [␥-32 P]ATP. The rest of the oligonucleotides were phosphorylated at the 5Ј end with non-radioactive ATP. Labeled and phosphorylated oligonucleotides were purified through a Microspin P-6 or P-30 column (Bio-Rad) and annealed at a primer:template:downstream oligonucleotide ratio of 1:1.3:1.5 in 50 mM Tris, pH 8.0, 250 mM NaCl. The mixture was incubated at 95°C for 5 min, slow cooled to 50°C for 30 min, incubated at 50°C for 20 min, and then immediately transferred to ice. An 18% polyacrylamide gel followed by autoradiography was used to confirm Ͼ90% annealing.
Presteady-state Analyses-Burst reactions were performed in 50 mM Tris-HCl, pH 8.0, 2 mM dithiothreitol, 20 mM NaCl, and 10% glycerol with 300 nM DNA substrate 1B-A and 100 nM enzyme. This buffer was also used for all the other kinetic experiments described in this work. All concentrations given are final concentrations upon mixing. These burst experiments were performed at saturating dNTP concentrations. Reactions were initiated by rapid mixing of the Pol␤/DNA and Mg/ dNTP solutions. The Rapid Quench apparatus (KinTek) (24) was employed in these experiments because of the fast reaction rate. The reactions were quenched with 0.3 M EDTA.
Active Site Titration-A fixed concentration of either wild type Pol␤ or the I260Q variant was preincubated with a range of concentrations of the 1B-A substrate followed by reacting with the correct dNTP substrate, dTTP, for 0.3 s. This time interval was chosen based on the burst reaction and ensures maximum reaction amplitude while minimizing multiple turnovers. The results of these reactions were used to calculate the percentage of active sites of the enzyme and the equilibrium dissociation constant for the DNA substrate, K D , as described below.
Single Turnover Misincorporation Assays-This experimental approach provides a way to compare the I260Q variant with the wild type enzyme for their relative abilities to incorporate correct and incorrect dNTPs into a primer-template by determining the dissociation constant from the dNTP substrate, K d , and the maximum rate of polymerization for each dNTP, k pol . Single turnover kinetic experiments were performed under conditions in which over 90% of the DNA is bound by the enzyme to ensure that the rate of a single catalytic turnover is being measured. These conditions were empirically determined to be at a 15:1 molar ratio of enzyme to DNA for the I260Q variant (data not shown); therefore, 750 nM enzyme (for both I260Q and wild type) and 50 nM DNA were used with varying concentrations of dNTP. The reactions for correct dNTP incorporation were performed in the KinTek apparatus at 37°C by loading 15 l of the Pol␤-DNA complex in one sample loop and 15 l of a single dNTP into the other. Substrate concentrations typically used were 1-100 M and reactions times were 0 -8 s. Misincorporation reactions were performed under the same conditions as for correct dNTP incorporation but manually, with a range of dNTP concentrations between 1 and 1 mM, and reaction times 0 -2700 s.
Product Separation and Data Analysis-All the reactions described above resulted in the addition of one nucleotide to the primer of the DNA substrate used. The n (non-extended) and nϩ1 (extended by one dNTP) DNA were resolved on a denaturing 20% polyacrylamide gel (Sequel NE, American Bioanalytical). To measure product formation as a function of time, the gels were scanned using a Storm 840 PhosphorImager (Amersham Biosciences) and quantified with the accompanying ImageQuant software. The data were analyzed using the Kaleidagraph program (Synergy software). Data from the burst experiments were fit to the burst equation: [product] ϭ A(1 Ϫ exp(Ϫk obs t) ϩ k SS t). Data from active site titrations were fit to the quadratic equation: is the concentration of the Pol␤-DNA complex, [E] 0 is the initial enzyme concentration, [D] 0 is the initial concentration of the gapped DNA substrate, and K D(DNA) is the dissociation constant of the Pol␤-DNA complex. Single turnover kinetic data were fit to the single exponential equation: [product] ϭ A(1 Ϫ exp(Ϫk obs t)), where A is the amplitude, t is the time, and k obs is the observed rate constant. Observed rate constants were then plotted against [dNTP], and the data were fit to the hyperbolic equation: where k pol is the maximum rate of polymerization, and K d is the equilibrium dissociation constant for dNTP. Fidelity values were calculated as follows: fidelity , where c and i represent the correct and incorrect dNTPs, respectively.

I260Q Exhibits a Burst of Product Formation-The I260Q
variant was previously isolated from a genetic screen of in vivo mutator mutants. The mutator tendencies of this variant were previously confirmed using the in vitro missing base primer extension assay (19). In the study described here, we use kinetic analyses to fully characterize this mutant and describe the kinetic basis for its mutator phenotype. First, we wished to determine whether the presteady-state kinetic profile of I260Q is similar to wild type. The reaction conditions are described under "Experimental Procedures." As shown in Fig. 1, I260Q displays biphasic kinetics of presteady-state nucleotide incorporation as does wild type, in which an initial rapid phase is followed by a slower, linear one. This is immediately followed by the rate-limiting step of the catalytic cycle, dissociation from DNA (25). For I260Q, the fast phase occurs at a rate of 11.4 Ϯ 2.5 s Ϫ1 , which is similar to the 13.6 Ϯ 1.3 s Ϫ1 for wild type on  Fig. 2 are the data obtained from these measurements fitted to the quadratic equation as described under "Experimental Procedures." From this experiment, the K D values were calculated to be 35 nM for I260Q and 60 nM for wild type. Thus, the equilibrium dissociation constants are very similar for the two enzymes with I260Q having a slightly greater affinity for the DNA substrate.
I260Q Forms All Three dNTP Mispairs Opposite Both a Purine (A) and a Pyrimidine (C)-We wished to determine the kinetic basis for the mutator phenotype of I260Q previously observed in both in vivo and in vitro assays (19). First, we measured the rates and dissociation constants for the dNTP substrate for correct and incorrect nucleotide incorporation opposite template A in substrate 1B-A. The reaction conditions and molar ratios are listed under "Experimental Procedures." As illustrated in Fig. 3 for incorporation of the correct nucleotide, the results were first fitted to the single exponential equation for each individual dTTP concentration to calculate the k obs for each of the dTTP concentrations used. These rates were then plotted as a function of dTTP concentration (Fig. 3B) and fitted to the hyperbolic equation to obtain the K d and k pol values. As summarized in Table II, the kinetic basis for the increased ability to misincorporate nucleotides opposite purine A is in the dissociation constant, K d , for the dNTP. The dissociation constant for the correct dNTP substrate, dTTP in this case, is ϳ5-fold higher than wild type: 30 M for I260Q and 6 M for wild type, whereas the dissociation constant for the incorrect nucleotide is lower than wild type. For example, when dGTP is incorporated opposite A, the K d for I260Q is 49 M, whereas it is 147 M for wild type. A similar difference between I260Q and wild type is observed for dATP misincorporation opposite A where I260Q has a K d of 30 M and wild type has a k d of 138 M. The K d values for both wild type and mutant are lowered in reactions examining dCTP insertion opposite A most likely because of a slippage phenomenon that has previously been demonstrated for wild type. Because of slippage, the enzyme positions the base immediately 5Ј to the template for nucleophilic attack thus allowing it to pair with the incoming nucleotide. In this case, this would mean incorporation opposite G rather than A, which could then form the Watson-Crick G-C base pair. The rate of incorporation for each dNTP substrate is similar for wild type and I260Q. This ultimately results in a lowered catalytic efficiency for I260Q for the correct dNTP substrate. However, I260Q has a somewhat higher catalytic efficiency than wild type for misincorporation. These data suggest that I260Q misincorporates nucleotides opposite purine A because of a lack of discrimination during substrate binding.
We also wished to examine misincorporation opposite a pyrimidine. We constructed a substrate with a templating C within the same sequence context as for A above. Table III summarizes the results of incorporation opposite pyrimidine C using template 1B-C. As observed with misincorporation opposite A, the I260Q variant also misincorporates opposite C. The kinetic basis for misincorporation opposite template C is K d , the equilibrium dissociation constant. In fact, I260Q appears to prefer the incorrect dNTP substrate. The rates of correct and incorrect incorporation are comparable between wild type and I260Q as observed for insertion opposite template A. The catalytic efficiency of I260Q for misincorporation is slightly higher than that of wild type. This indicates that the basis for the misincorporation phenotype of I260Q and lowered fidelity observed in the in vivo and in vitro assays previously described results from decreased discrimination during dNTP binding by Pol␤. DISCUSSION The I260Q Pol␤ variant was identified in a genetic screen as an enzyme with an overall activity similar to wild type Pol␤ but with a significantly lower fidelity (19). In this study, we describe the kinetic basis for the mutator phenotype of I260Q Pol␤. Strikingly, I260Q binds the correct nucleotide with significantly lower affinity and incorrect nucleotides with 2-6-fold higher affinities than the wild type enzyme. Therefore, the kinetic basis for the mutator phenotype of I260Q lies in its inability to discriminate the correct from the incorrect dNTP during the binding step of nucleotide incorporation. Because residue 260 is located on the inside lining of the hydrophobic pocket within the hinge region that is between the catalytic and the COOH-terminal domains of Pol␤, our kinetic data suggest that this hinge and the associated domain motions are important for polymerase fidelity.
The Hinge Functions in the Geometric Alignment of Substrates-The hinge region is located at the interface between the catalytic domain (residues 149 -262) and the COOH-terminal domain (residues 263-338) of Pol␤. Both domains present residues toward the large active site cleft that are crucial for template binding, nascent base pair recognition, and catalytic turnover (14 -16). Strands ␤2, ␤7, and helix M are located close to the hinge axis, and the network of interactions extending from the residues in this region is likely to moderate the extent of domain motions. Careful analysis of the relative domain motions observed in various Pol␤-substrate complexes identified the loops near Val-177 and Cys-178 and Pro-261, Lys-262, and Asp-263 as hinge boundaries. The flexibility of the two loops allows for the bulk movement of secondary structural elements within the COOH-terminal domain. Other residues, including but not limited to Leu-194, Thr-196, Cys-178, Ile-260, Gly-268, Val-269, and Phe-272, contribute interactions necessary for the maintenance and function of an intricate hydrophobic network along the hinge region. Some of these side chains also display obvious movements upon domain closure. Therefore, it is conceivable that the hinge, as orchestrated by the movement of some of these residues, controls the extent and the dynamics of the enzyme closure around the substrates (15,16). Arg-258, Tyr-271, Arg-283, Asn-294, Glu-295, and Tyr-296 are layered immediately above the hydrophobic hinge region and form a major part of the template binding site. Rotation of the COOH-terminal domain results in enzyme activation by positioning the template, the 3Ј-OH of the primer, the base, and the ␣-phosphate of the incoming dNTP such that rapid and accurate nucleotide incorporation can occur.
Ile-260 Is Critically Involved in Domain Closure-Ile-260 is located near the hinge axis and in the wild type enzyme contributes to the formation of a hydrophobic core. Upon comparison of several high-resolution Pol␤ co-crystal structures, the Ile-260 side chain is consistently found in either of two different conformations. The 2 angle changes such that C␦ is alternating between the "up" and the "down" position, thereby affecting the packing against the adjacent hydrophobic residues. Interestingly, the volume of a small cavity above Ile-260 appears to be altered by the 2 switch as well. This cavity, as depicted in Fig. 4, could feasibly act as a buffer space absorbing local motions and allowing a specific flexing in the hinge between the COOH-terminal and the catalytic domains of Pol␤.
The mutation from Ile-260 to Gln is associated with a slight change in side chain volume as well as a significant change in chemical nature. When altered from Ile to Gln, the small cavity above position 260 disappears. If the side chain amide group moves into the interior of the protein core, the hydrophobic packing would most likely be disrupted. However, analysis of the three most frequent rotamers using PYMOL along with molecular modeling indicate that the amide side chain is most likely oriented away from the adjacent hydrophobic residues and points toward the charged residues on the surface of the template binding cleft. These local changes could result in a polymerase that may not close properly and efficiently, thereby leading to a less stable but active conformation in the I260Q variant as compared with wild type Pol␤. Aberrant positioning of the dNTP-binding residues within the active site of the enzyme could certainly affect dNTP affinity in the mutant enzyme. This suggestion is consistent with the recently published structures of the wild type enzyme in complex with mismatched bases at the active site (26). They indicate that there might be an intermediate so-called "semi-open" complex that occurs after dNTP binding but before chemistry (26). The complexes showed that the bases stagger and form alternate hydrogen bonds with the surrounding polymerase residues. Such staggered bases partially overlap, thus preventing the fingers from properly closing and forming the optimal active site (26). We suggest that the active site of I260Q may also be in a semi-open conformation.
I260Q Is Important for dNTP Binding and Positioning-Molecular modeling suggests that the change of Ile-260 to Gln introduces small but significant shifts in adjacent hydrophobic residues that form part of the dNTP binding pocket, namely Tyr-271 and Phe-272 (15). Reordering of these side chains adjacent to the ribose moiety of the incoming dNTP affects nucleotide binding as has been shown by biochemical studies published previously (27). Alteration of Phe-272 to Leu leads to a decrease in the binding affinity for the dNTP substrate and an enzyme with lower fidelity (28). Asn-279, a direct spatial neighbor of Phe-272, forms a minor groove interaction with the incoming substrate such that incorrect dNTPs most likely protrude into the minor groove and clash with Asn-279. Alteration of Asn-279 to Ala results in an inaccurate enzyme with significantly less affinity for correct dNTP substrate (29). The reshaping of the dNTP binding pocket geometry when Ile-260 is altered to Gln is consistent with the finding that the I260Q Pol␤ shows a 5-fold increase in preference for the incorrect over the correct substrate during the binding step. Thus, we suggest that wild type Pol␤ does discriminate correct from incorrect substrates during dNTP binding and during subsequent steps in the catalytic cycle and not only during transition state chemistry, as has been proposed previously (30).
Ile-260 Is Important for Template Positioning-Based upon our modeling studies, changes in the size and the chemistry of residue 260 would be predicted to cause significant shifts in the position of surface residues of the COOH-terminal domain. As a result, the position of the template DNA could be critically affected, which in turn has long range effects on the geometry of the Watson-Crick binding pocket. The templating base forms an integral part of the Watson-Crick binding pocket and an aberrant geometry could be more permissive for accepting and incorporating the incorrect dNTP. This possibility has been explored and substantiated by molecular modeling studies. In our modeling studies, we have found changes in the side chain positions of residues Arg-258, Glu-295, and Tyr-296 as a result of altering Ile-260 to Gln. Consequently, the template DNA is repositioned relative to where it would be expected to be found in the wild type enzyme, thereby affecting the position of the templating base and changing the geometry of the Watson-Crick binding pocket. The subtle structural changes described above could conceivably explain why the I260Q variant has a lower affinity for the correct dNTP substrate and higher affinity than the parent enzyme for the incorrect dNTP substrate, thus, leading to a significant drop in fidelity as compared with wild type Pol␤.
Summary and Implications-Our data show that changing hinge residue 260 of Pol␤ from Ile to Gln results in a polymerase that has nearly wild type activity but is a mutator mutant. We find that the kinetic basis for misincorporation by I260Q lies in loss of discrimination at the level of dNTP substrate binding. Our modeling studies indicate that the Gln mutant affects the domain closure around the DNA substrate, that the protein matrix around the ribose moiety of the incoming nucleotide is altered, and that the DNA template could be positioned differently in this variant than in the wild type enzyme (19). DNA binding is the first step in the catalytic cycle, and it is likely that repositioning of the DNA substrate would consequently affect the subsequent steps, particularly dNTP binding and chemistry. Based on our finding that the I260Q variant has a decreased affinity for the correct nucleotide and increased affinity for the incorrect one, we propose that this mutant is less well adapted to form a nucleotide binding pocket with a geometry similar to that of wild type Pol␤ that supports catalysis. Subtle changes in a small and compact enzyme such as Pol␤ can have dramatic effects, making it an excellent model to further our understanding of mechanistic aspects of polymerase fidelity.