Preparation of Site-specific T=mCG cis-syn Cyclobutane Dimer-containing Template and Its Error-free Bypass by Yeast and Human Polymerase η*

Background: The cis-syn cyclobutane dimer of TmCG is responsible for C-to-T mutations in skin cancer. Results: The mC of the dimer codes as C when bypassed by the DNA damage bypass polymerase η. Conclusion: The coding properties of the mC are consistent with a water-exposed polymerase active site. Significance: The mC of the dimer must deaminate to T to become mutagenic. C-to-T mutations are a hallmark of UV light and, in humans, occur preferentially at methylated PymCG sites, which are also sites of preferential cyclobutane pyrimidine dimer (CPD) formation. In response, cells have evolved DNA damage bypass polymerases, of which polymerase η (pol η) appears to be specifically adapted to synthesize past cis-syn CPDs. Although T=T CPDs are stable, CPDs containing C or 5-methylcytosine (mC) are not and spontaneously deaminate to U or T at pH 7 and 37 °C over a period of hours or days, making their preparation and study difficult. Furthermore, there is evidence to suggest that, depending on solvent polarity, a C or an mC in a CPD can adopt three tautomeric forms, one of which could code as T. Although many in vitro studies have established that synthesis past T or U in a CPD by pol η occurs in a highly error-free manner, the only in vitro evidence that synthesis past C or mC in a CPD also occurs in an error-free manner is for an mC in the 5′-position of an mC=T CPD. Herein, we describe the preparation and characterization of an oligodeoxynucleotide containing a CPD of a TmCG site, one of the major sites of C methylation and C-to-T mutations found in the p53 gene of basal and squamous cell cancers. We also demonstrate that both yeast and human pol η synthesize past the 3′-mC CPD in a >99% error-free manner, consistent with the highly water-exposed nature of the active site.

C in the 5-position of an m C‫؍‬T CPD. Herein, we describe the preparation and characterization of an oligodeoxynucleotide containing a CPD of a T m CG site, one of the major sites of C methylation and C-to-T mutations found in the p53 gene of basal and squamous cell cancers. We also demonstrate that both yeast and human pol synthesize past the 3-m C CPD in a >99% error-free manner, consistent with the highly water-exposed nature of the active site.
One of the primary causes of basal and squamous cell skin cancer is exposure to sunlight. Sequencing of the p53 gene of these skin cancers has revealed a predominance of C-to-T and CC-to-TT mutations at methylated dipyrimidine sites, Py m CG (1,2). These sites are also hotspots for cis-syn cyclobutane pyrimidine dimer (CPD) 3 formation induced by the UV in sunlight (3,4). Unlike CPDs of TT sites, C-or 5-methylcytosine ( m C)-containing CPDs are unstable and deaminate to U or T (5)(6)(7)(8)(9)(10)(11). The major responses to UV damage are global genomeand transcription-coupled nucleotide excision repair (12)(13)(14) and translesion synthesis (15). CPDs do not distort DNA structure or duplex stability greatly (16 -18), making them difficult to be detected and repaired by nucleotide excision repair (19). Although they block RNA polymerases and can thus be removed by transcription-coupled repair, this cannot occur for the non-transcribed strand or for either strand of inactive genes. Thus, many CPDs may go unrepaired before replication is initiated, and because they block replicative polymerases, DNA damage bypass polymerases are recruited to carry out translesion synthesis. Of these, polymerase (pol ) is the most efficient at synthesizing past CPDs and appears to be the principal polymerase involved in CPD translesion synthesis (20 -22). Early in vitro studies showed that pol could synthesize past a TϭT CPD in an essentially error-free manner, suggesting that the two Ts of the CPD were capable of directing the insertion of As. It was only by investigating DNA synthesis past N 3 -methyl derivatives of TϭT CPDs that it was possible to demonstrate that both the 5Ј-and 3Ј-Ts of the CPD were directing the insertion of As and that the As were not being inserted by a non-templated mechanism, otherwise known as the "A rule" (23). Furthermore, studies with 7-deaza-ATP demonstrated that a Watson-Crick like intermediate was being used to direct the insertion of the A, and not a Hoogsteen-type base pair as found in the crystal structure of another Y-family polymerase (24). These conclusions were later confirmed through x-ray crystallography of intermediates in the synthesis past a TϭT CPD by human pol (25).
Unlike T in native DNA or in a CPD, which is known to preferentially adopt a keto tautomer and base pair with A, a C or an m C in a CPD could, in principal, adopt amino and/or imino tautomeric forms with E-or Z-stereochemisty (the imino hydrogen is trans or cis with respect to N3), some of which could thereby direct the insertion of G and/or A ( Fig. 1) (26,27). The possibility that C or m C could exist in an imino tautomeric form comes from early work demonstrating that dihydrocytosine adopts the amino tautomeric form in water but an imino tautomeric form in the much less polar solvent chloroform (28).
Gas phase theoretical calculations have reproduced the preference for the imino tautomeric form in non-polar environments (29,30) but failed to reproduce the preference for the amino form in water (30). The same type of calculations also predicted that the Cs in CPD should exist almost exclusively in the imino tautomeric state in water (27). However, evidence that Cs in CPDs exist in their amino tautomeric state comes from the highly preferential insertion of G opposite the Cs by pol from various genetic experiments in yeast and human cells (31,32) and from an in vitro translesion experiment with an m CϭT CPD-containing template and yeast pol (33). No work has yet been carried out on a site-specific CPD of an m CG site. Because we have previously observed that the deamination rate of an m C in a CPD is 3-4 times slower for Pu m CTPu than for PuT m CPu (where Pu ϭ A or G) and depends on flanking sequence (34), it was possible that the relative proportion of tautomers may likewise depend on position and flanking sequence.
Part of the difficulty in working with C-or m C-containing CPDs in vitro or in vivo is their spontaneous deamination to U or T within hours to days (34) compared with ϳ50,000 years for canonical C or m C in duplex DNA (35,36). This lability has also impeded the development of a phosphoramidite building block for site-specific incorporation of C-containing CPDs into DNA templates by automated synthesis. At the moment, the only route to such templates involves UV irradiation of an oligodeoxynucleotide (ODN) followed by HPLC purification, as was originally developed for the preparation of a TϭC CPD-containing 11-mer (37). We used this approach to prepare a 14-mer containing the CPD of an m CTA site (33). An m CTA site was chosen for our initial studies because the CPD was found to have a much lower deamination rate and to be produced in a higher yield than at the more biologically relevant T m CG site (34). Herein, we report the preparation and characterization of a CPD of a T m CG site, a known hotspot for C methylation, CPD formation, and UV light-induced C-to-T mutations, and dem-onstrate the highly error-free, non-mutagenic insertion of G opposite the m C in the CPD by yeast and human pol . We also report the deamination rate of the m C in the CPD and confirm the error-free but mutagenic insertion of A opposite the resulting T.

EXPERIMENTAL PROCEDURES
Enzymes, Substrates, and Equipment-T4 polynucleotide kinase was purchased from New England Biolabs, [␥-32 P]ATP from Amersham Biosciences, and dNTPs from Invitrogen. The catalytic core of yeast pol with an N-terminal His 6 tag was prepared as described previously (38). Full-length human pol with a C-terminal His 6 tag was also prepared as described previously (39). The m C-containing template ODN and primer ODNs were synthesized by Integrated DNA Technologies, Inc. Mass spectrometry was carried out on an LCQ Classic ion trap mass spectrometer (Thermo Finnigan, San Jose, CA).
Preparation and Deamination of cis-syn Tϭ m C CPD-containing 14-mer DNA Template-100 g of T m C 14-mer in 200 l of 20 mM Tris (pH 8.8) on a piece of Saran Wrap was irradiated for 1 h on top of a 302-nm transilluminator (ϳ1.9 milliwatts/cm 2 ) at 4°C in a cold room. HPLC purification was carried out with 50 mM triethylamine acetate at pH 8.5 to minimize deamination to yield ϳ5.2 g of the cis-syn dimer eluting in 1 ml of ϳ10% acetonitrile. This sample was immediately stored at Ϫ20°C prior to use. A sample was also completely deaminated by adjusting the pH of the Tϭ m C 14-mer sample to pH 6.5 by adding 0.5 M Mes buffer and heated overnight at 67°C.
DNA Photoproduct Identification and Characterization by HPLC and Mass Spectrometry-The fractions corresponding to the major photoproduct peaks from the HPLC of the T m C 14-mer irradiation mixture, with and without heating, were analyzed by an enzyme-coupled mass spectrometry assay. In a typical assay, 1 l of a 1 unit/l aqueous solution of nuclease P1 was added to a 10-l aliquot of the ODN sample, incubated at

Error-free Bypass of T‫؍‬ m C CPD by pol
room temperature for 3 min, cooled on ice, and immediately analyzed by electrospray ionization coupled to MS/MS. MS/MS data were acquired on the selected [M Ϫ H] ions. To select both the deaminated and undeaminated components for fragmentation, the mass width for precursor selection was set at 5-6 m/z units Single-hit Competitive Insertion Assay-In a cold room (4°C), a 5-l aliquot of the Tϭ m C 14-mer template (120 nM) was added to 5 l of 40 nM 5Ј-32 P end-labeled primer, 20 mM Tris-HCl (pH 7.5), and 10 mM DTT and allowed to anneal for 10 min before the addition of pol (500 nM). Each sample was incubated for an additional 7 min before initiating single-hit synthesis by the addition of 30 l of all dNTPs (200 M each) containing 10 mM MgCl 2 , 10 mM Tris-HCl (pH 7.5), and 10 mg/ml sonicated/denatured salmon sperm DNA as a polymerase trap. The synthesis reaction was quenched after 10 s by the addition of 80 l of formamide containing 0.1% xylene cyanol, 0.2% SDS, 25 mM EDTA, and 1 g of unlabeled primer (stop mixture). The samples were then heated at 100°C for 10 min before loading onto a 10% polyacrylamide gel (40 cm ϫ 0.8 mm) containing 25 mM citrate (pH 3.5), which was the same as the reservoir buffer. The gel was polymerized by adding 1.3 ml of ferrous sulfate (250 mg/100 ml), 1.3 ml of 10% ascorbate, and 300 l of 3% hydrogen peroxide. The gel was run at 2000 V until the xylene cyanol dye marker reached 25 cm (ϳ3 h). For biased pool experiments, the dNTP of interest was held at 100 M while the dGTP concentration was varied as indicated.
Multiple-hit Competitive Nucleotide Insertion Assays-In a cold room (4°C), primer-templates were prepared by annealing 20 nM primer with 60 nM template, to which 200 M each dNTP, 10 mM MgCl 2 , 5 mM DTT, 10 mM Tris-HCl (pH 7.5), and 500 nM pol were added. After 2 min, a 10-l aliquot was removed and added to a tube containing 300 g of sonicated and denatured salmon sperm DNA before quenching the reaction with 80 l of stop mixture. Aliquots from extension times that yielded almost exclusively full-length product were then assayed by electrophoresis on a 20% polyacrylamide gel containing 25 mM citrate (pH 3.5).
Deamination Rates by Nucleotide Insertion Assay-The Tϭ m C 14-mer template (120 nM) was adjusted to pH 7.5 with Mes buffer (0.5 M) and incubated at 37 or 50°C. Aliquots (5 l) were removed at various times and quickly frozen on dry ice before storing overnight at Ϫ70°C. All subsequent steps were performed in a cold room (4°C). To each of the 5-l aliquots was added 5 l of 40 nM 5Ј-32 P end-labeled primer, 1 mM each dNTP, 20 mM Tris-HCl (pH 7.5), and 10 mM DTT, followed by annealing for 10 min before the addition of yeast pol (500 nM). Each sample was incubated for an additional 5 min to complete multiple-hit synthesis, after which 300 g of sonicated and denatured salmon sperm DNA was added before quenching the reaction with 80 l of stop mixture. The samples were then heated at 95°C for 10 min before loading onto a 25% polyacrylamide gel (40 cm ϫ 0.8 mm) containing 25 mM citrate (pH 3.5), which was the same as the reservoir buffer, and the gel was run at 2000 V until the xylene cyanol dye marker reached 18 cm (ϳ4 h). The gel had been polymerized by adding 1.3 ml of ferrous sulfate (250 mg/100 ml), 1.3 ml of 10% ascorbate, and 300 l of 3% hydrogen peroxide.
Analysis of Deamination Rate Data-The deamination rate constant was obtained from the slope of a nonlinear least squares fit of the natural log of the fraction of dGMP inserted, G/(G ϩ A), versus time to ln(G/(G ϩ A)) 0 Ϫk*t where (G/(G ϩ A)) 0 equals the fraction of dGMP inserted at time 0.

RESULTS
Template Design-Because there is still no phosphoramidite building block available for incorporating a T m C CPD into an ODN, we resorted to preparing the T m C CPD-containing template by UV irradiation of an ODN. On the basis of prior experience with the preparation of an m CϭTA CPD-containing 14-mer (33), we designed the T m C-14-mer ( Fig. 2 and supplemental Fig. S1). This sequence is devoid of other dipyrimidine sites, so the Tϭ m C CPD-containing product could be cleanly separated from other photoproducts of the T m C site and unreacted ODN by reverse phase HPLC. The sequence was long enough, however, to accommodate a 10-mer for primer extension reactions by positioning the T m C site toward the 5Ј-end. However, the exact sequence was modified from what was used before to accommodate the T m CG sequence and to optimize separation of the expected primer extension products by citrate-PAGE according to which nucleotides were inserted opposite the CPD by pol (Fig. 2).
Preparation and Characterization of CPD-containing 14-mer Templates-The T m C-14-mer was irradiated with 302 nm light at 4°C at pH 8.5 to suppress deamination for 1 h, after which half of the sample was allowed to completely deaminate by heating at 67°C for 3 h at pH 6.5. The reaction mixtures were then subjected to reverse phase HPLC to separate the different photoproducts from the starting T m C-14-mer. The fractions collected by HPLC were immediately cooled on dry ice to suppress deamination prior to further analysis and use. The desired cis-syn CPD-containing product, T[c,s] m C-14-mer, was initially identified by its conversion to its deamination product, T[c,s]T-14-mer, which eluted with a longer retention time. together with sonicated/denatured salmon sperm DNA was added to a preincubated mixture of the indicated 9-mer primer-14-mer template and pol . The reactions were terminated after 10 s with EDTA and unlabeled primer, and the products were electrophoresed on a citrate (pH 3.5)-10% polyacrylamide gel. The standards corresponding to extension of the 9-mer by A or G were prepared by automated synthesis.
Thus, UV irradiation of the T m C-14-mer resulted in one major product peak with a retention time of 30.5 min (supplemental Fig. S2A), which shifted to a peak with a retention time of 33.2 min after heating to effect deamination (supplemental Fig.  S2B). To further confirm the assignment of the Tϭ m C-14-mer as the cis-syn cyclobutane dimer, we analyzed the sample by nuclease P1-coupled electrospray ionization-MS. Nuclease P1 digests CPD-containing DNA to mononucleotides and a CPDcontaining trinucleotide, pPyϭPyN, which results in a major fragment, pPyϭPyab, corresponding to loss of the base from the 3Ј-nucleotide to yield an abasic site (ab) (40). MS/MS of the nuclease P1 digestion products of the HPLC fraction that eluted at 30.5 min showed an ion at m/z 953.4, corresponding to a pTϭ m CG photoproduct, which fragmented to a major ion at m/z 802 (supplemental Fig. S3), corresponding to pT[c,s] m Cab. MS/MS of the digestion products of the fraction eluting at 33.2 min showed an ion at m/z 954.3, corresponding to pTϭTG, which fragmented to an ion at m/z 803, corresponding to pT[c,s]Tab (supplemental Fig. S4).
Single-hit Competitive Nucleotide Insertion Assay Opposite CPDs by pol -To determine the extent to which the m C in the Tϭ m C CPD codes as C or T, we first carried out a single-hit competition assay with dATP and dGTP. In this assay, yeast pol was first incubated with the 9-mer primer-TX 14-mer templates and then added to a solution containing equimolar dATP and dGTP and a large excess of sonicated and denatured salmon sperm DNA to trap the polymerase once it dissociated from the primer-template. To resolve the products of insertion, we used of a 25 mM citrate gel (pH 3.5), which can easily separate products according to their nucleotide composition as evidenced from the independently synthesized standards in Fig. 2. Primer extension under these single-hit conditions yielded primarily the products in which the primer was extended by two nucleotides, corresponding to the almost exclusive extension by G followed by A for the native and CPD-containing T m C templates and to extension by AA for the TT templates.
Multiple-hit Competitive Nucleotide Insertion Assay Opposite CPDs-We also examined insertion opposite the m C site in the Tϭ m C CPD using a multiple-hit competition assay in which the full-length translesion synthesis products could also be separated according to nucleotide composition on a low pH citrate gel. In this assay, primer extension was carried out by both yeast and human pol on the Tϭ m C-14-mer template in the presence of equal amounts of all dNTPs. The primer extension products were then electrophoresed on a 20% Tris borate/ EDTA-polyacrylamide gel to identify the full-length synthesis products in comparison with authentic 14-mer standards AA-14-mer and GA-14-mer, corresponding to insertion of dAMP or dGMP opposite the m C site, respectively. We found that both yeast and human pol were able to fully extend the primer to the end of the Tϭ m C and TϭT-14-mer templates and that the full-length products had the same mobility on a denaturing polyacrylamide gel as the primer extension products opposite the undimerized T m C and TT-14-mer templates and both the GA-14-mer and AA-14-mer standards (supplemental Fig. S5). To determine the nucleotide insertion selectivity opposite the dipyrimidine sites, we used a 20% citrate-polyacrylamide gel to separate the full-length 14-mer synthesis products according to sequence composition (Fig. 3). As shown, the full-length synthesis products from both yeast and human pol opposite the Tϭ m C-14-mer were found to have the same gel mobility as those opposite the undimerized T m C-14-mer, as well as an authentic 14-mer ODN containing G (Fig. 3, lanes 1-3), indicating the preferential insertion of G opposite the m C of the CPD. Conversely, the full-length synthesis product opposite the completely deaminated Tϭ m C-14-mer had the same mobility as that opposite the undimerized TT template and the authentic 14-mer product containing A (lanes 1, 2, and 5). These results demonstrate that the Tϭ m C-14-mer template did not undergo any detectable deamination during its purification and subsequent handling, which would have resulted in the production of an A-containing 14-mer bypass product.
Selectivity of Nucleotide Insertion Opposite Tϭ m CG CPD-To determine the selectivity of nucleotide insertion opposite the m C of the dimer in comparison with undamaged DNA, we carried out the single-hit primer extension experiment in the presence of a biased nucleotide pool in which the indicated nucleotide was present in a 32-fold excess over dGTP (Fig. 4). As shown, the 9-mer primer was extended only by G opposite both T m C and Tϭ m C by both yeast and human pol , even in the presence of 32-fold excess dCTP or dTTP. Extension of the primer by C or T would have produced bands moving slower than the 9-mer and faster than 9G, respectively (33), which was not observed. On the other hand, in the presence of 32-fold excess dATP, a small amount of a band corresponding to extension by AA opposite Tϭ m C was observed in addition to extension by GA. This band was not observed for the undamaged T m C site, suggesting that insertion opposite the m C in a dimer proceeds with a lower fidelity of insertion than that opposite an undamaged m C.
To more accurately determine the selectivity of nucleotide insertion, primer extension was carried out with different ratios of dATP to dGTP (Fig. 5). The selectivity of an A insertion relative to G could then be determined accurately from the slope of a linear fit of the ratio of the primer extension product incorporating A compared with G versus the ratio of the concentrations of dATP versus dGTP. Both yeast and human pol showed similar selectivities for A versus G insertion of 6.6 ϫ

Error-free Bypass of T‫؍‬ m C CPD by pol
10 Ϫ3 and 6.1 ϫ 10 Ϫ3 , respectively, corresponding to G versus A selectivities of 152 Ϯ 7 and 165 Ϯ 6. The non-zero intercepts of 0.012 Ϯ 0.008 and 0.007 Ϯ 0.006 for yeast and human pol suggest the presence of ϳ1% of the deaminated Tϭ m C CPD and/or the E-imino tautomer.
Deamination Kinetics of Tϭ m C CPD-The ability of pol to faithfully insert A opposite T and G opposite m C in a CPD was then used to study the deamination kinetics of the Tϭ m C CPD at two temperatures. The fraction of m C remaining in the CPD as a function time was determined by monitoring the fraction of G inserted opposite the m C by pol in the full-length primer extension reaction (supplemental Fig. S6). The rate constants were then determined from the slope of linear fits of the data to a first-order rate process (supplemental Fig. S7). In this way, the deamination rate constants were determined to be 0.0014 min Ϫ1 at 37°C and 0.0054 min Ϫ1 at 50°C, corresponding to deamination half-lives of 8.25 and 2.1 h, respectively.

DISCUSSION
We have demonstrated that DNA synthesis by both yeast and human pol opposite a m C in a cis-syn CPD of a T m CG site proceeds in a Ͼ99% error-free and hence non-mutagenic manner. Our original expectation was that the m C might also exist in an imino tautomeric form based on an early study showing that 5,6-dihydrocytosine could exist in either an amino or imino tautomeric form depending on the solvent (28). In the relatively non-polar solvent chloroform, dihydrocytosine was found to exist primarily in an imino tautomeric form (E and/or Z), whereas in water, it was found to exist primarily in the amino tautomeric form. If a polymerase were to surround the nascent base pair involving C or m C of a CPD and exclude water, as it does for a representative replicative polymerase, yeast pol ␦ (Fig. 6A) (41), one might expect an increased preference for an imino tautomer.
The observation of only a dGMP insertion opposite the m C of the T m C CPD by either yeast or human pol in the presence of equal concentrations of all four dNTPs indicates that the m C is not adopting a significant amount of the E-imino tautomer form, which would have templated the insertion of dAMP (Fig.  1). An upper limit to the amount of the E-imino tautomer comes from the non-zero intercept of ϳ1% observed in the plots of dAMP versus dGMP insertion as a function of [dATP]/ [dGTP], which could also more likely be attributed to the presence of the deaminated Tϭ m C CPD. We do not have any direct way of knowing whether or not any of the Z-imino tautomer is present, as it is not Watson-Crick complementary to any of the canonical DNA bases. If it is present, it is either very poor at directing nucleotide insertion or directs the insertion of only G. Results from the single-hit primer extension experiments (Fig.  2) did not show any premature dissociation from the template compared with the undamaged template that would be indicative of significant impediment to nucleotide insertion. If the Z-imino isomer were directing insertion of G, it would likely be occurring via the equivalent of a G⅐T wobble base pair. Both insertion and extension of a G opposite the 3Ј-T of a TϭT CPD by yeast pol are known, however, to be much less efficient than for a Watson-Crick base-paired A (42,43). Thus, the comparably efficient extension opposite the 3Ј-m C in the dimer and the undamaged template observed in the single-hit primer extension reaction suggest that insertion is occurring largely if not exclusively via the amino tautomer in the CPD.
The selectivity of dGMP insertion relative to dAMP by yeast and human pol opposite the m C of the Tϭ m C CPD is less, however, than that opposite an undamaged m C, as shown in Fig.  3, in which 32-fold excess of dATP over dGTP was used. In Fig.  3, a band for dAMP insertion opposite the m C in the CPD is clearly seen, whereas it is not seen for undamaged m C. This is FIGURE 4. Single-hit primer extension biased nucleotide pool competition experiment opposite 14-mer templates. A 32:1 mixture of the indicated dNTP/dGTP (100 M in the dNTP) together with sonicated/denatured salmon sperm DNA was added to a preincubated mixture of the indicated 9-mer primer-14-mer template and pol . The reaction was terminated after 10 s with EDTA and unlabeled primer, and the products were electrophoresed on a citrate (pH 3.5)-10% polyacrylamide gel. The standards corresponding to extension of the 9-mer by A or G were prepared by automated synthesis. FIGURE 5. Selectivity of dGMP versus dAMP insertion opposite m C of T‫؍‬ m C CPD via single-hit assay. Varying ratios of dATP to dGTP (with dATP fixed at 100 M) together with sonicated/denatured salmon sperm DNA were added to a preincubated mixture of the 9-mer primer-14-mer template and pol . The reaction was terminated after 10 s with EDTA and unlabeled primer, and the products were electrophoresed on a citrate (pH 3.5)-10% polyacrylamide gel. The standards corresponding to extension of the 9-mer by A or G were prepared by automated synthesis.
consistent with experiments that determined that the selectivity for dGMP insertion relative to dAMP opposite a C in undamaged DNA was 960 for the same yeast pol used in these experiments (44) and 909 for human pol (45) compared with 152 and 165, respectively, which we observed for insertion opposite the m C in the Tϭ m C CPD. The lower fidelity may have to do with the greater ability of an m C in a CPD to adopt the E-imino tautomer compared with an undamaged m C due to the loss of aromatic stabilization of the amino tautomer (27,29,30). Interestingly, we had observed a selectivity of only 120 for insertion opposite the m C in an m CϭT CPD (33), suggesting that the 5Ј-m C may be more prone to tautomerization or that insertion opposite the amino tautomer is less specific opposite the 5Ј-m C of a CPD.
The almost exclusive presence of the amino tautomeric form of the m C in the CPD during nucleotide insertion suggests that it is in a polar and/or aqueous environment (28), which is in accord with a recent crystal structure of a ternary complex of human pol with dATP opposite the 3Ј-T of a TϭT CPD (25). In this structure, the O4 carbonyl of the 3Ј-T of the TϭT CPD, corresponding to the position of the N4 of the 3Ј-C of a Pyϭ m C CPD, is completely exposed to water during insertion of dATP (Fig. 6B). Likewise, the O4 carbonyl of the 5Ј-T of the CPD, corresponding to the 5Ј-m C of an m CϭPy CPD, is also highly exposed to water during dATP insertion (Fig. 6C). The 5Ј-m C of an m CϭT CPD would therefore also be expected to adopt the amino tautomeric form, in accord with our previous study showing that only dGMP is inserted opposite this m C in the presence of a equal concentration of A (33). These results suggest that pol may have evolved to bypass CPDs in an errorfree manner by increasing the exposure of the N4 of a C or an m C in either or both of the 3Ј-and 5Ј-sites of a cis-syn CPD to maximize the amount of the amino tautomer.
The active site of pol also had to evolve to accommodate the 5Ј-pyrimidine of the CPD when the 3Ј-pyrimidine of the CPD is in the templating position, otherwise the 3Ј-pyrimidine could not template insertion of the complementary nucleotide. Replicative polymerases achieve high selectivity for canonical bases in part by extensive contacts with the 5Ј-face of the templating nucleotide (Fig. 6A), which would prevent the CPD from entering the active site when insertion opposite the 3Ј-pyrimidine of the CPD is to take place. As a result, the polymerase would become arrested at a CPD, which would allow for the recruitment of DNA damage bypass polymerases. It has been shown, however, that in the absence of other polymerases, the replicative T7 DNA polymerase will preferentially insert A opposite the 3Ј-pyrimidine of all types of dipyrimidine photoproducts, irrespective of their base pairing properties, by a nontemplated mechanism (46 -48). If non-templated insertion were to occur opposite the 3Ј-C or m C of a CPD, a C-to-T mutation would result and may represent an alternate pathway for UV light-induced C-to-T mutations. Thus, the Tϭ m C CPD could serve as a useful probe for determining if a polymerase synthesizes past CPDs by a templated or non-templated mechanism.
Although bypass of a C-or an m C-containing CPD by pol occurs in an error-free and hence non-mutagenic manner, once the C or m C in a CPD deaminates, error-free insertion by pol opposite the resulting U or T leads to a mutation. Thus, the mutagenicity of pol bypass of a CPD will depend on the presence and deamination rate of a C or an m C within the CPD, which we have recently shown to be highly dependent on sequence (34), protein binding (49), and nucleosome position (50). As we have shown herein, the Tϭ m C CPD in a singlestrand AT m CG sequence context deaminates with a half-life of 8.25 h at 37°C, which is not much different from what we previously found for a CPD with the same flanking sequence in duplex DNA at low salt concentration (7.7 h) (34). On the other hand, the methyl-CpG-binding protein MeCP2 was found to completely suppress deamination of a Tϭ m CG CPD (49) and hence the mutagenic potential of the CPD if bypassed by pol . In a nucleosome core particle, the deamination rate of a Tϭ m CG CPD whose backbone faced away from the histone surface was accelerated by a factor of 9, whereas one whose backbone faced toward the surface was retarded by ϳ5-fold (50). Thus, as long as pol synthesizes past a C/ m C-containing CPD prior to deamination, the bypass event will be non-mutagenic, but as soon as the CPD deaminates, the bypass will be mutagenic. Therefore, although pol synthesizes past CPDs in an error-free manner, in the sense of faithfully inserting nucleotides that are complementary to the pyrimidines present in the CPD, if a U or T arises from deamination of a C or an m C, translesion synthesis will be mutagenic and can explain the origin of UV light-induced C-to-T mutations.