The Steric Gate of DNA Polymerase ι Regulates Ribonucleotide Incorporation and Deoxyribonucleotide Fidelity*

Background: Accurate bypass of DNA damage by translesion DNA polymerases is critical for cell survival. Results: Wild-type human DNA polymerase ι incorporates ribonucleotides, and a steric gate mutant increases both ribonucleotide incorporation and deoxyribonucleotide selectivity. Conclusion: A single amino acid residue in DNA polymerase ι limits incorporation of ribonucleotides into DNA. Significance: DNA polymerase ι may incorporate ribonucleotides during translesion DNA synthesis.

The daily rate of DNA damage is estimated at upwards of 50,000 lesions per cell (1). Although DNA repair mechanisms evolved to contend with this high rate of damage, lesions that persist may stall replication, leading to fork collapse, doublestrand breaks, and genomic instability. Such instability is a hallmark of tumorigenesis. To prevent stalling at sites of damage, cells utilize specialized DNA polymerases (pols) 2 to promote translesion synthesis (TLS) that enables bypass of damage and prevents genomic instability (2). During DNA replication, pausing by the replicase leads to a polymerase switch such that the replicase is replaced by a TLS polymerase capable of bypassing the damage (3). After the damage has been bypassed, the distributive TLS polymerase may synthesize only a handful of nucleotides before dissociating from the DNA, thereby allowing the replicase to resume genome duplication. This mechanism of DNA damage tolerance is critical to maintaining genomic integrity and, ultimately, cell survival.
Many TLS specific polymerases are characterized by lower fidelity and processivity on undamaged DNA than replicases. One of the most error-prone human DNA polymerases is the Y family enzyme, DNA polymerase (pol ). The distributive enzyme lacks exonuclease activity and has a distinct misincorporation signature that is template-dependent (4). For example, misinsertion of G opposite template T occurs with 3-10-fold greater efficiency than insertion of the correct A nucleotide, which causes the enzyme to abort or pause synthesis at template T (5). In contrast, pol exhibits its most efficient correct incorporation opposite template A (5). Unlike other Y family members, pol also possesses intrinsic dRP-lyase activity and as such may participate in specialized forms of base excision repair in vivo, potentially during the somatic hypermutation process (6,7). pol has also been implicated in the TLS of oxidative and ultraviolet-induced DNA lesions (8 -11). Crystal structures of the catalytic domain of pol show that the unique active site of the enzyme readily facilitates Hoogsteen base pairing to accommodate DNA lesions and mispairs (12)(13)(14). pol has been shown to prefer manganese over magnesium as the divalent cation for catalysis and is highly active with low levels (50 -250 M) of manganese (15,16).
Genomic stability depends on DNA polymerases selecting the correct dNTP for the template and the exclusion of NTPs, because cellular NTP levels are much higher than dNTPs (17,18). This imbalance necessitates the critical ability of DNA polymerases to prevent incorporation of NTPs, especially because they are highly susceptible to spontaneous hydrolysis and render the DNA at risk for breaks. Surprisingly, NTPs appear to be tolerated to an extent in DNA, because ribonucleotides have recently been shown to be the most common "lesion" in the mouse genome (19). Studies in yeast on high fidelity replicative polymerases ␣, ␦, and ⑀ estimate that these enzymes incorporate one NTP for every 625-5,000 dNTPs, resulting in over 10,000 NTPs incorporated into the genome during one round of replication (20,21). Human pol ␤, a comparatively lower fidelity enzyme involved in base excision repair, has been reported to insert one NTP per 81 insertion events (22). Removal of ribonucleotides from DNA is accomplished by RNase enzymes in a pathway of ribonucleotide excision repair (23).
Many DNA polymerases contain a single residue nascent to the active site that functions as a "steric gate," thereby preventing incorporation of NTPs by steric hindrance with the 2Ј-OH that does not exist with dNTPs. The location of this residue is highly conserved within polymerase families (24). Escherichia coli pol V is a highly error-prone, TLS polymerase that is homologous to Y family polymerases in eukaryotes. Our laboratory has previously identified critical amino acids in pol V that control sugar discrimination (25). Surprisingly, we found that even the wild-type pol V enzyme efficiently incorporates NTPs into DNA. A single conserved aromatic residue, Tyr 11 , which was identified as the steric gate (26), was shown to limit NTP incorporation into DNA, because substitution with the smaller alanine amino acid further increased NTP incorporation by pol V. In contrast, when the adjacent highly conserved Phe 10 residue was changed to leucine, it had the opposite effect and restricted NTP incorporation and also increased dNTP selectivity.
In this work, we investigated the NTP insertion efficiency and fidelity of the human Y family TLS enzyme, pol . We further examined the structural basis for NTP exclusion using a steric gate mutant, that based upon residues conserved in Y family DNA polymerases is identified as Tyr 39 in pol (Fig. 1A). Wild-type pol and the Y39A variant were evaluated for their ability to bind DNA and both incorporate and extend NTPs in vitro using replication assays, steady-state kinetics, and gel mobility shift assays. Like E. coli pol V, we found that human pol readily incorporates ribonucleotides into DNA. As expected, ribonucleotide incorporation increased in the Y39A variant. Surprisingly, dNTP selectivity also improved in the Y39A mutant. Molecular modeling suggests that when compared with the constricted active site of the wild-type enzyme, which normally favors highly mutagenic base pairing, the Y39A active site is more spacious, thereby allowing normal and accurate Watson-Crick base pairing to occur.

EXPERIMENTAL PROCEDURES
Materials-Ultrapure NTPs and dNTPs were purchased from GE Healthcare and Roche, respectively. All oligonucleotides were synthesized by Lofstrand Laboratories (Gaithersburg, MD) and gel-purified prior to use.
DNA Substrate Preparation-Primers were 5Ј-radiolabeled with 32 P by Lofstrand Laboratories. Annealing reactions consisted of 100 nM radiolabeled primer and 200 nM unlabeled template in 50 mM Tris-HCl, pH 8.0, 5 mM NaCl, 50 g/ml bovine serum albumin, and 1.42 mM 2-mercaptoethanol. Reaction mixtures were heated at 95°C for 5 min and then slowly cooled to room temperature over several hours.
Protein Expression and Purification-Full-length N-terminal His-tagged human wild-type pol was expressed using the E. coli strain RW644 and expression plasmid pJM868 as previously described (27). The Y39A pol mutant was generated by chemically synthesizing a 310-bp NcoI-BglII fragment encoding the N terminus of the E. coli codon optimized POLI gene with the Y39A substitution (Genscript, Piscataway, NJ). The fragment was subsequently subcloned as an NcoI-BglII fragment into pCT14 (a precursor of pJM868 with an NcoI site at the start of the POLI gene), and the resulting plasmid, pKD001, was introduced into the expression strain RW644, and the Y39A pol was protein-purified using the same protocol as wild-type pol (27). Final preparations of both wild-type pol and Y39A were determined to be Ͼ95% pure by SDS-PAGE gel analysis (Fig. 1B). Protein concentrations were determined using the Bio-Rad protein assay designed for 96-well plates and a plate reader. Commercially prepared His-tagged full-length pol (ProteinOne, Gaithersburg, MD) was used to generate the standard curve. Final preparations for both wild-type pol and Y39A were determined to be 1-2 M. Primer Extension Reactions-Standard reaction mixtures (10 l) contained 100 nM pol , 10 nM of radiolabeled DNA substrate, 100 M nucleotide(s) in 1ϫ reaction buffer (0.25 mM MnCl 2 , 40 mM Tris-HCl, pH 8.0, 10 mM dithiothreitol, 250 g/ml bovine serum albumin, 2.5% glycerol). All DNA substrates were confirmed to be Ͼ95% annealed by incubating with wild-type pol and 100 M of all four dNTPs. The reactions were incubated at 37°C for 10 min and quenched by the addition of 10 l of 95% formamide, 10 mM EDTA solution. The samples were heated to 100°C for 5 min and resolved on an 18% polyacrylamide, 8 M urea gel. Reaction products were visualized and quantified using a Fuji FLA-5100 Phosphorimager and ImageGauge software.
Steady-state Kinetic Assays-Incorporation of select dNTPs or NTPs opposite template G and T was evaluated using steadystate kinetics. Steady-state kinetic parameters k cat and K m were measured using standing start reactions as previously described (28). All reactions contained 10 nM DNA substrate and 1ϫ reaction buffer. Reaction conditions were optimized using a time course for a variety of enzyme:DNA ratios to ensure that reactions were in the linear range (Ͻ20% of total primer extension). Steadystate reactions measuring correct dNTP incorporation were performed for 1-2 min using 0.4 -1 nM enzyme. Reactions examining incorrect dNTP incorporation were performed for 1-7.5 min using 1.5-10 nM enzyme. Reactions examining NTP incorporation were performed for 1-20 min using 1-10 nM enzyme. Nucleotide concentrations included in reactions ranged from 0.001 to 100 M. All reactions were initiated by mixing equal volumes of 2ϫ enzyme/DNA solution with 2ϫ nucleotide/Mn 2ϩ solution. All 2ϫ solutions were preincubated at 37°C prior to the reaction. The reactions were quenched, visualized, and quantified as described above. The data were plotted using GraphPad Prism (v6.0) software (San Diego, CA). By plotting velocity (v) versus dNTP concentration and fitting the Michaelis-Menten rectangular hyperbola by least squares (29), we obtain the kinetic parameters V max and K m .
The misincorporation frequency for an incorrect dNTP is calculated as follows, where V max ϭ k cat [total polymerase-primer-template complex] (29). Each reported value is the average of at least three independent experiments. Electrophoretic Mobility Shift Assays-DNA binding constants (K D(DNA) ) for wild type and Y39A were determined by gel electrophoretic mobility shift assay as described (30) with several modifications. Radiolabeled DNA substrate (0.2 nM) was incubated with a dilution series of pol enzyme (0.2-780 nM) in 1ϫ reaction buffer for 20 min at 25°C. Samples were loaded onto a 6% Native Tris acetate PAGE gel that had been prerun at 150 V at 4°C for 1 h. The gels were run at 150V for 4 -5 h at 4°C, then dried, and exposed for 2 h before phosphorimaging. The fraction of bound DNA relative to total DNA was quantified using ImageQuant software. The fraction of bound DNA was plotted as a function of protein concentration, and the DNA binding constant was determined by fitting data to the following equation, where Y is the amount of bound protein, m is a scaling factor, and b is the apparent minimum Y value (31). Replication Foci Assay-Recruitment of Y39A pol to nuclear replication foci following UV treatment was assessed as previously described (32). The peCFP-C1-pol construct encoding the CFP-fusion protein has been previously described (33). To obtain a construct encoding the CFP-fusion protein with the Y39A substitution, a 204-bp BglII-BstEII fragment with the desired sequence was chemically synthesized (Genscript) and subcloned into the wild-type peCFP-C1-pol vector to generate pKD004. Constructs were transfected into MRC5 fibroblasts using TurboFectin 8.0 according to the manufacturer's instructions (Origene, Rockville, MD). For UV treatment, cells were irradiated 20 h post-transfection at 7 J/m 2 and recovered by incubating for a further 6 h. Cells were fixed as noted (33). Localization of pol , as indicated by CFP fluorescence, was evaluated by fluorescence microscopy using a Zeiss Axiophot2 microscope (Carl Zeiss), an Orca ER CCD camera, and Simple PCI software (Hamamatsu). Slides were imaged by excitation at 436 nm and CFP emission at 480 nm. At least 200 nuclei were counted for each cell line and treatment from three independent experiments.
Molecular Modeling-The structure of pol in a ternary complex with a template dG and incoming dCTP has been determined (PDB code 2ALZ) and requires no modeling (14). The structure of pol in a ternary complex with a template dT and incoming dATP was modeled on the BrU/dGTP (PDB code 3H4D) published structures (34). The BrU is replaced by a normal thymine, and the dGTP is replaced by dATP. The misincorporation of dGTP opposite dT was based on the ternary complex structure of pol with a template dT and incoming dGTP (PDB code 3GV8) and requires no remodeling (12). The misincorporation of dGTP opposite dG was based on the ternary complex structure of pol with a template 8-oxo-G and incoming dGTP (PDB code 3Q8R) and incoming dATP (PDB code 3Q8Q), which shows more deviation of the deoxyribose from the standard conformation (8). The 8-oxo-G was converted to a normal G. All four structures were superimposed, and the protein structures are highly similar. Because of the narrow active site for the replicating base pair, large deviations are found in the positions of incoming dNTP. In the majority of the structures, the ␣, ␤, and ␥ triphosphate superimposed well. Deviations occur most often in the deoxyribose moiety (8), as exemplified in Fig. 9E when G:dGTP or A:dGTP mismatch occurs. All ribonucleotides are modeled based on deoxyribonucleotide by replacement of a deoxyribose by a ribose. 39 Functions as a Steric Gate to Impede Ribonucleotide Incorporation and Its Reduced Catalytic Activity Is Rescued by Mn 2ϩ -To assess the ability of full-length wild-type pol and the Y39A steric gate mutant to incorporate ribonucle-otides during DNA synthesis, we initially compared primer extension in the presence of all four dNTPs or NTPs. Because wild-type pol is most active in the presence of low concentrations of either magnesium or manganese (15), we assayed the ability of both wild-type pol and the Y39A variant to incorporate dNTPs/NTPs in the presence of either 500 M Mg 2ϩ or 250 M Mn 2ϩ over time. As previously reported, in the presence of magnesium, wild-type pol efficiently extends the primer by several bases in the presence of dNTPs. In contrast, it exhibits reduced, but nevertheless significant, primer extension when NTPs are present ( Fig. 2A). However, even after extended reactions times, the major product is just one base longer than the primer. In the presence of magnesium, the pol Y39A variant exhibits very low levels of incorporation of dNTPs and even lower levels of incorporation of NTPs ( Fig. 2A). In dramatic contrast, when manganese is the divalent cation in the reaction mixture, pol Y39A efficiently and completely extends the primer to a similar extent with dNTPs or NTPs (Fig. 2B). Similarly, in the presence of manganese, wild-type pol efficiently extends the primer with dNTPs or NTPs, albeit with reduced efficiency compared with pol Y39A (Fig. 2B). We observed a similar lack of dNTP selectivity for wild-type pol in primer extension assays on all four template bases in the presence of manganese (Fig. 3). Extension by at least one nucleotide is observed for each individual dNTP at each template base, consistent with previous reports (15,16). In many cases, primers are extended by several nucleotides, indicating multiple misincorporations and subsequent extension of the mispaired bases by the error-prone polymerase. The pattern of NTPs incorporated by wild-type pol is clearly different from that observed with dNTPs (Fig. 3). The selectivity for misincorporated NTPs and primer extension lengths are much lower than observed with dNTPs. Our results therefore indicate that although wild-type pol has very poor selectivity at the base level for dNTPs, it generally exhibits better selectivity at the base level for NTPs. Incorporation of NTPs by wild-type pol appears to be template-dependent, because full primer extension is observed with at least one single NTP on templates A and G, whereas much less NTP incorporation is observed on templates T and C. These results indicate that undamaged DNA templates containing A or G sequences may be at greater risk in vivo for NTP incorporation (and the resulting mutagenic or destabilizing consequences) by wild-type pol .

Residue Tyr
Under the same assay conditions, we found that pol Y39A incorporates individual NTPs to a much greater extent than wild-type pol (Fig. 3). When provided all four NTPs, Y39A synthesizes the primer with a similar efficiency to reactions with all four dNTPs. Regarding incorporation of individual NTPs, Y39A displays considerably reduced base selectivity compared with the wild-type enzyme on all templates. In contrast, however, when incorporating individual dNTPs, Y39A appears to be more selective than wild-type pol because extension products in the presence of individual dNTPs are shorter than observed with the wild-type enzyme (Fig. 3). Thus, the pol Y39A enzyme exhibits reduced overall discrimination for sugar selectivity but an increased discrimination for the dNTP base.
To facilitate quantitative comparison of nucleotide discrimination differences observed between wild-type pol and pol Y39A, we performed steady-state kinetic analysis for insertion opposite templates T and G with select nucleotides. As indicated by the primer extension assays, pol Y39A exhibits increased discrimination between correct and incorrect dNTPs on both templates (Table 1). At template G, pol Y39A incorporates dGTP 13-fold less frequently than the wild-type enzyme. In contrast, our kinetic data show that pol Y39A is incapable of distinguishing between dCTP and rCTP when incorporating opposite G, with an f inc value of 1 for rCTP. This lack of discrimination makes pol Y39A 1100-fold more efficient than wild-type pol for rCTP misincorporation. However, pol Y39A is capable of discriminating between rCTP and rGTP, and misincorporates rGTP opposite template G only 10-fold more efficiently than wild-type pol ( Table 1). Removal of the steric gate constraint in the pol Y39A enzyme not only makes it a more accurate DNA polymerase but also confers the ability to efficiently incorporate correctly matched NTPs.
With regards to template T, wild-type pol exhibits its hallmark preferential dGTP incorporation (by nearly 2-fold relative to dATP), whereas pol Y39A has lost this unique signature. Indeed, pol Y39A has reduced misincorporation propensity compared with the wild-type enzyme for dG:T misincorporation, as indicated by a 5-fold lower f inc value. When presented with NTPs, pol Y39A incorporates rGTP 850-fold more efficiently than wild-type pol . The base preference of wild-type pol for incorporation of G opposite template T is retained, however, because incorporation was observed at low levels for rGTP but was undetectable for rATP. In contrast, pol Y39A inserts rATP more efficiently than rGTP (Table 1).
Wild-type pol Is Able to Extend a Ribonucleotide-terminated Primer-NTP incorporation was observed with both wildtype pol and pol Y39A enzymes, but less extension was observed when wild-type pol was only provided with NTPs. Therefore, we investigated the ability of wild-type pol to extend a primer with a terminal ribonucleotide (rATP) correctly base-paired with a template T. In these modified primer extension reactions, we sought to determine whether wild-type pol would terminate after NTP incorporation or whether NTPs would be extended with either dNTPs or NTPs and thus persist in the replicated DNA. The reactions were performed with either all four dNTPs or all four NTPs under the same conditions as described above over a 15-min time course. Wild-type pol efficiently extends a ribonucleotide-terminated primer when provided dNTPs, with nearly all primer utilized by 30 s (Fig. 4). In contrast, only a small amount of extension through NTP incorporation by wild-type pol is observed in this context. These results suggest that a ribonucleotide inserted by wild-type pol in vivo may persist in DNA after being efficiently extended by further dNTP incorporation.
Wild-type pol Is Capable of Ribonucleotide Incorporation Opposite Damaged DNA-To assess the ability of wild-type pol and pol Y39A to incorporate NTPs during TLS, we performed primer extension assays using DNA substrates Temp8oxoG and XTA48. We observe marked incorporation of individual NTPs by wild-type pol opposite both abasic site (Fig. 5A) and 8oxoG (Fig. 5B) templates. Interestingly, wildtype pol will readily incorporate rATP, UTP, and rGTP opposite an abasic site, but it does not incorporate rCTP. Without an effective steric gate, the pol Y39A variant readily incorporates all four NTPs opposite the abasic site. Incorporation of NTPs by wild-type pol is also observed on the 8oxoG template and, again, to a greater extent with Y39A. To put these remarkable observations into perspective, the related Y family TLS polymerase pol only incorporates miniscule levels of NTPs opposite either an abasic site or an 8oxoG lesion (Fig. 6), indicating that the ability of wild-type pol to efficiently incorporate ribonucleotides during lesion bypass is not common among TLS polymerases.
The pol Y39A Substitution Does Not Affect DNA Binding Affinity in Vitro or Recruitment to Replication Foci in Vivo-To determine the effect of the single amino acid Y39A substitution on the DNA binding capacity of pol , we determined the DNA binding affinity constants for wild-type pol and Y39A pol using a gel mobility shift assay (Fig. 7). In this assay, wild-type pol has a K D(DNA) of 38 Ϯ 7 nM, which is in accordance with previously reported results obtained using active site titration assays (36). We observed a K D(DNA) of 51 Ϯ 12 nM for pol Y39A; therefore the pol Y39A enzyme binds DNA with similar affinity as the wild-type enzyme. These results indicate that the aberrant incorporation activity of pol Y39A is not a result of altered DNA binding affinity. The Y39A substitution, although disruptive to nucleotide discrimination, does not affect the ability of pol to bind DNA.
To test the hypothesis that the Y39A substitution does not disrupt global structure or cellular properties of the enzyme, we evaluated the recruitment of pol Y39A to damage-induced nuclear replication foci. Wild-type pol has been previously shown to form discrete nuclear foci in response to UV exposure, and this recruitment is likely due to direct interaction with DNA pol (33). MRC5 cells were transfected with a CFP-pol fusion protein coding for either the wild type or Y39A substitution. Cells were subsequently irradiated with 7 J/m 2 UV and allowed to recover for 6 h before evaluating for the presence of nuclear CFP foci. As expected, cells expressing the wild-type fusion protein exhibit a significant increase in foci upon UV treatment (Fig. 8). A similar increase in foci was observed in  cells expressing the CFP-pol Y39A fusion protein. No significant difference was observed across the treated or untreated groups, indicating that pol Y39A exhibits a cellular response to DNA damage similar to wild-type pol . Therefore the Y39A substitution does not affect the recruitment of the enzyme to damage-induced foci, indicating that the enzyme is not grossly misfolded and retains appropriate protein-protein interactions in vivo.

DISCUSSION
Incorporation of ribonucleotides into DNA has mutagenic consequences, because ribonucleotides are unstable and make the DNA susceptible to strand breaks. Replicative polymerases have evolved active site configurations that efficiently exclude the incorporation of NTPs. This active site configuration typically includes an aromatic residue that functions as a steric gate, a physical barrier that provides a steric clash with the 2Ј-OH group that distinguishes NTPs from dNTPs. The steric gate residue has been implicated to have other specialized functions that vary depending on the individual polymerase (37). Depending on the identity and primary function of the polymerase, the steric gate residue may be more or less efficient at excluding NTPs from being incorporated during polymerization. For example, wild-type E. coli pol V can incorporate rNTPs without compromising efficiency, resulting in synthesis of long RNA tracts (25).
In this study, we investigated the mechanism of sugar discrimination employed by the human Y family TLS polymerase, pol . We found that wild-type pol is capable of incorporating NTPs opposite undamaged DNA depending on the sequence context. pol also readily incorporates NTPs opposite damaged DNA, including an abasic site and an 8oxoG lesion. Replicative polymerases generally adhere to the "A rule" by preferentially incorporating dA opposite an abasic site (38 -41). However, pol has been shown to exhibit unique behavior when replicating across an abasic site in the presence of dNTPs (42,43). pol does not follow the A rule, because kinetic data show a preference for pol to insert dG, followed by dT, dA, and dC, and structural data indicate that pol can stabilize the incoming dNTP in the absence of a templating base. Our results with individual NTP incorporation were similar; when incorporating NTPs opposite an abasic site, we found that pol has a preference for rG and U. pol is less efficient at incorporating rA opposite an abasic site, and almost no incorporation of rC is observed. No extension was observed beyond the base incorporated opposite the basic site, consistent with previous dNTP studies (43). Given that in vivo, pol will bypass a lesion and continue to synthesize a short stretch of nucleotides beyond the lesion, our results indicate that certain DNA sequence contexts may be more "at risk" for ribonucleotide incorporation-both at the site of damage itself and the undamaged downstream nucleotides. The ability of wild-type pol to incorporate NTPs has significant in vivo considerations, given that the cellular nucleotide pool is highly biased; roughly 10 -100-fold more NTPs are present compared with the corresponding dNTP (17,18).
To further investigate the structural basis for NTP discrimination, we mutated the putative steric gate residue Tyr 39 of pol to Ala. This alteration effectively removes the steric clash that would result between the 2Ј-OH of the NTP and the aromatic steric gate residue. Removal of the steric gate makes the Y39A enzyme much more efficient than wild-type pol at incorporating and extending NTPs opposite both undamaged and damaged DNA. Surprisingly, in replication assays pol Y39A has reduced sugar selectivity but exhibits increased base selectivity when incorporating dNTPs. Steady-state kinetic analysis confirmed that pol Y39A cannot discern a difference between the dNTP and NTP version of the same base. Moreover, pol Y39A has a stronger preference for traditional Watson-Crick base pairing. When incorporating opposite template T, wild-type pol exhibits a classical 2-fold increase in incorporation efficiency for the T:dG mispair. In contrast, pol Y39A is more efficient at incorporating dA opposite template T, making it less mutagenic than the wild-type enzyme.
Unlike other Y family polymerases, which have a spacious active site, the constricted active site of pol is biased against normal Watson-Crick base pairing. When inserting a dCTP opposite a template G, as shown in Fig. 9A, the template adopts the syn conformation because the narrow active site is too small for a normal Watson-Crick base pair. When the nascent base pair is T:dATP (Fig. 9B), the template T is tilted and nonplanar relative to dATP because of the narrow active site. Orthogonal views of template G (Fig. 9C) and template T (Fig. 9D) highlight the steric gate effect of residue Tyr 39 that is removed when the residue is changed to the much smaller residue, Ala (Y39A). Fig.  9 (A and B) highlights the atypical hydrogen bonds between the G:dCTP and T:dATP base pairs, as well as the hydrogen bond between Gln 59 and the template dT, which may stabilize the tilted template dT. If present, the 2Ј-OH of ribonucleotides CTP or ATP (shown in green carbon and red oxygen) would clash badly with Tyr 39 in the wild-type enzyme. Fig. 9 (C and D) shows that CTP and ATP can be accommodated in the pol Y39A mutant, because the steric clash with the 2Ј-OH has been eliminated.
In the G:dGTP mispair case (Fig. 9E), both template and incoming dGTP are likely to shift from the G:dCTP position (shown in semitransparent gray). The shifts of both alleviate the steric clashes between Tyr 39 and the incoming nucleotide and thereby reduce the base selection, as well as discrimination against an incoming ribonucleotide by the wild-type pol enzyme. The shifts and reduced interactions with the enzyme  . pol Y39A is recruited to replication foci following UV treatment similar to wild-type pol . A, upon UV treatment, both wild-type pol and pol Y39A-expressing cells exhibit a significant increase in nuclear replication foci. B, no significant difference is observed between wild-type pol and pol Y39A under either condition.
via Tyr 39 may also explain the much reduced k cat and increased K m of both wild-type pol and pol Y39A when misincorporating GTP opposite a template G, because neither enzyme efficiently performs this misincorporations ( Table 1). The T:dGTP mismatch, however, is readily accommodated, because the structure of the active site of pol favors mispairing (Fig. 9F). In this case, dGTP forms a hydrogen bond with Gln 59 . For comparison, the normal T:dATP base pair is shown in semitransparent gray overlay. When incorporating dNTPs, base selectivity is increased with Y39A because of the increased size of the active site that can properly accommodate and thus favor an incoming nucleotide, in the correct Watson-Crick base pairing configuration. Gln 59 may hydrogen bond with the template base (T) rather than with the incoming nucleotide (dGTP) (Fig.  9, D, bottom panel, and F). The Y39A substitution may also favor Mn 2ϩ over Mg 2ϩ because the relaxed coordination geometry of Mn 2ϩ is able to stabilize a comparatively less constricted active site. The narrower and more constrained active site of wild-type pol coordinated with Mg 2ϩ provides incoming nucleotide positioning that allows for polymerization but concomitantly favors mutagenic base pairing. Therefore, pol Y39A exhibits a seemingly mixed phenotype, reduced sugar discrimination with increased base discrimination. Previous studies on E. coli pol V have shown that the same two phenotypes can be attributed to two different residues (25). The pol V residue that corresponds to Tyr 39 of pol is pol V Y11. The pol V Y11A mutant has been shown to have a marked reduction in sugar discrimination relative to the wildtype enzyme, but its base selectivity remained unchanged. In contrast, alteration of the neighboring residue, Phe 10 to Leu, had the effect of increasing both base and sugar selectivity for pol V. Our studies with pol suggest that amino acid substitutions at the steric gate of a DNA polymerase can have pronounced effects on both sugar and base selectivity, as well as metal ion coordination.
During DNA replication in yeast, replicative polymerases have been estimated to incorporate ribonucleotides at a rate of 1 per 625-5,000 bases replicated, leading to over 10,000 ribonucleotides incorporated into the genome during a single round of replication (20). Ribonucleotide incorporation may also occur during DNA repair. DNA polymerase ␤, the main polymerase involved in base excision repair, has been estimated to incorporate NTPs at a rate of one per 81 insertion events. Ribonucleotide incorporation by pol during TLS may further contribute to this burden. Mechanisms of NMP removal from DNA have recently been described in yeast and mice and are viewed as essential to maintaining genome integrity (19,23). Although the presence of NMPs in DNA is mutagenic and may also contribute to genomic instability, it has also been recently shown to act as a signal for DNA mismatch repair (35,44). Therefore, a physiological balance may exist between incorporation and excision of ribonucleotides from the genome. Although NTP incorporation during TLS synthesis by pol may contribute to the lesion burden, we acknowledge the intriguing possibility of ribonucleotide incorporation by pol as a signal for further repair.