Translesion DNA Synthesis by Yeast DNA Polymerase h on Templates Containing N 2 -Guanine Adducts of 1,3-Butadiene Metabolites*

Yeast DNA polymerase h can replicate through cis-syn cyclobutane pyrimidine dimers and 8-oxoguanine lesions with the same efficiency and accuracy as replication of an undamaged template. Previously, it has been shown that Escherichia coli DNA polymerases I, II, and III are incapable of bypassing DNA substrates containing N 2 -guanine adducts of stereoisomeric 1,3-butadiene metabolites. Here we showed that yeast polymerase h replicates DNA containing the monoadducts ( S )-butadi-ene monoepoxide and ( S , S )-butadiene diolepoxide N 2 guanines albeit at an ; 200–300-fold lower efficiency relative to the control guanine. Interestingly, nucleotide incorporation opposite the ( R )-butadiene monoepoxide and the ( R , R )-butadiene diolepoxide N 2 -guanines was ; 10-fold less efficient than incorporation opposite their S stereoisomers. Polymerase h preferentially incorporates the correct nucleotide opposite and downstream of all four adducts, except that it shows high misincorporation frequencies for elongation of C paired with ( R )-butadiene monoepoxide N 2 -guanine. Additionally, polymerase h does not bypass the ( R , R )- and ( S , S )-butadiene diolepoxide N 2 -guanine- N 2 -guanine intrastrand cross-links, and replication

Various pathways exist in cells to overcome replication blockage caused by DNA lesions. One such pathway, translesion DNA synthesis, involves specialized polymerases that, unlike replicative polymerases, are able to perform DNA synthesis on a damaged DNA template (reviewed in Refs. [1][2][3]. Translesion DNA synthesis can be error-free or error-prone, depending on the chemical structure of the lesion and the polymerase utilized for translesion replication. Among the eukaryotic DNA polymerases, yeast and human DNA polymerases perform efficient and accurate replication past a cis-syn cyclobutane pyrimidine dimer, a predominant DNA lesion formed by ultraviolet irradiation (4 -7). In the yeast Sac-charomyces cerevisiae, deletion of RAD30, which encodes pol , 1 confers moderate sensitivity to UV irradiation and an increase in UV-induced mutagenesis (8).
Mutations in the human RAD30A gene, the counterpart of the yeast RAD30, cause the variant form of xeroderma pigmentosum (9,10). Xeroderma pigmentosum variant cells are hypermutable in response to UV irradiation, and they exhibit a significantly reduced ability to bypass a T-T dimer (reviewed in Ref. 3). Consequently, xeroderma pigmentosum variant individuals suffer from a high incidence of sunlight-induced skin cancers.
7,8-Dihydro-8-oxoguanine is one of the lesions formed by oxidative damage to DNA. Yeast and human pol both efficiently bypass the 7,8-dihydro-8-oxoguanine lesion. Whereas other polymerases insert A opposite this lesion, pol preferentially inserts a C (11). Thus, pol is unique among DNA polymerases in its ability to bypass a T-T dimer and a 7,8dihydro-8-oxoguanine lesion efficiently and accurately.
Here we examined the ability of yeast pol to carry out translesion synthesis on DNA substrates containing N 2 -guanine adducts of stereoisomeric 1,3-butadiene metabolites. 1,3-Butadiene is a potent carcinogen in mice and to a lesser extent in rats (12) and has been classified as a probable human carcinogen. Butadiene-mediated carcinogenesis is initiated through its reactive metabolites: butadiene monoepoxide, butadiene diepoxide, and butadiene diolepoxide. Each of these metabolites is represented by at least two stereoisoforms. The mutagenicity of butadiene and its reactive metabolites has been observed in several biological systems, particularly in yeast (13,14) and mammalian cells (15). Butadiene epoxides can react at numerous sites in DNA, forming a multitude of adducts that differ in their stereochemistry (16,17). Butadiene epoxides are potent inhibitors of synthesis by DNA polymerases. Previously, it has been shown that Escherichia coli DNA polymerases I, II, and III are incapable of bypassing DNA substrates containing (R)-and (S)-BDO N 2 -guanines and (R,R)-and (S,S)-BDE N 2 -guanines (18) as well as (R,R)-and (S,S)-BDE N 2 -guanine-N 2 -guanine intrastrand cross-links (19). Here we examine the action of yeast pol on these two types of the N 2 -guanine epoxide-containing DNA substrates.

MATERIALS AND METHODS
DNA Substrates with Site-specific Lesions-The oligodeoxynucleotides containing butadiene epoxide N 2 -guanine adducts were prepared by the postoligomerization methodology developed by Harris et al. (20). A detailed description of the synthesis of the 11-mer oligonucleotides containing the (R)-and (S)-BDO and (R,R)-and (S,S)-BDE N 2 -guanines has been described previously (18). The 8-mer substrates containing the * This work was supported in part by National Institutes of Health Grants ES05355, S11-ES10018, and ES06676 (to R. S. L.) and GM 19261 (to L. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  (R,R)-and (S,S)-intrastrand BDE N 2 -guanine-N 2 -guanine cross-links were synthesized as published previously (19).
To construct the templates for polymerase reactions, each adducted oligonucleotide was ligated by T4 DNA ligase (New England Biolabs Inc., Beverly, MA) with two flanking oligonucleotides in the presence of the complement 45-mer scaffold. The ligation products were purified via denaturing polyacrylamide gel electrophoresis. The sequences containing the BDO and BDE N 2 -guanine lesions are identical: 5Ј-AGAAT-GTGGAAGATACTGTGGGCAGGTGGTGAATGGTCTGGGCAATGTC-GTTGACTGGGA-3Ј, where the adducted G is underlined. The sequence containing the BDE N 2 -guanine-N 2 -guanine cross-link is as follows: 5Ј-CTAGAATGTGGAAGATACTGTGCATGGTCCAATGGTCTGGGCA-ATGTCGT-3Ј, where the cross-linked guanines are underlined.
Oligodeoxynucleotides of anion-exchange grade purity were used as primers in the polymerase reactions and were obtained from the Midland Certified Reagent Co. (Midland, TX). Their sequences include 5Ј-ACGACATTGCCCAGACCATT-3Ј, which is complementary to the BDO and BDE N 2 -guanine-adducted templates from positions Ϫ6 to Ϫ26 relative to the site of lesion. The same primer was used for the BDE N 2 -guanine-N 2 -guanine cross-link-containing substrates, being complementary from positions Ϫ4 to Ϫ24. 5Ј-ACATTGCCCAGACCATT-GGA-3Ј was used as the Ϫ1 primer for the BDE N 2 -guanine-N 2 -guanine cross-link-containing substrates, and 5Ј-ATTGCCCAGACCAT-TCACCA-3Ј served as the Ϫ1 primer for the DNAs containing the BDO and BDE N 2 -guanine lesions. 5Ј-TTGCCCAGACCATTCACCAC-3Ј and 5Ј-GCCCAGACCATTCACCACC-3Ј served as the 0 and ϩ1 primers, respectively, overlapping the lesion site in the BDO and BDE N 2guanine-adducted substrates.
Primer oligodeoxynucleotides were phosphorylated with T4 polynucleotide kinase (New England Biolabs Inc.) using [␥-32 P]ATP (PerkinElmer Life Sciences). The 32 P-labeled primers were mixed with the oligonucleotide substrates in a molar ratio of 1:2 in the presence of 50 mM Tris-HCl (pH 7.0) and 100 mM NaCl, heated at 90°C for 2 min, and slow cooled to room temperature. The completeness of the primer annealing was confirmed by electrophoresis through a native 7.5% polyacrylamide gel.
pol Purification-The glutathione S-transferase-pol fusion protein was overexpressed and purified as described previously (4).
DNA Polymerase Reaction-The pol polymerase assays were carried out essentially as described by Johnson et al. (4). The reaction mixture (10 l) contained 25 mM potassium phosphate buffer (pH 7.0), 5 mM MgCl 2 , 5 mM dithiothreitol, 100 g/ml bovine serum albumin, 10% glycerol, 100 M dNTPs (each of the four dNTPs or one, as indicated), 5 nM primer annealed to a template, and 2 nM glutathione S-transferasepol . After incubation at room temperature for 20 min, reactions were terminated by the addition of a 10-fold excess loading buffer consisting of 95% (v/v) formamide, 20 mM EDTA, 0.02% (w/v) xylene cyanol, and 0.02% (w/v) bromphenol blue. The pol I (Klenow fragment) polymerase reactions were performed basically under the same conditions as the pol reactions but in the presence of the buffer provided by the enzyme supplier (New England Biolabs Inc.). The reaction products were resolved through a 15% polyacrylamide gel containing 8 M urea. Bands were visualized by autoradiography of the wet gels using Hyperfilm MP x-ray film (Amersham Pharmacia Biotech). Quantitative analyses of the results were performed using a PhosphorImager screen and Image-Quant 5.0 software (Molecular Dynamics, Sunnyvale, CA).
Steady State Kinetic Analysis-Steady state kinetic assays were carried out under the same conditions as the DNA polymerase assays except that 1 nM pol and 10 nM DNA substrates were used with various concentrations of one of the four nucleotides, and reactions were quenched after 5 min. DNA band intensities were quantitated using the PhosphorImager (Molecular Dynamics) and then used to calculate the rate of nucleotide incorporation as described previously (21). The rate of nucleotide incorporation was graphed as a function of nucleotide concentration, and k cat and K m parameters were obtained from the best fit of the data to the Michaelis-Menten equation.

Translesion DNA Synthesis by pol on the (R)-and (S)-BDO and (R,R)-and (S,S)-BDE N 2 -Guanine-adducted DNA Substrates-
The structures of the BDO and BDE N 2 -guanine stereoisomers, which were examined in this study, are shown in Fig. 1. Among the butadiene epoxide guanine species that are formed as a result of the butadiene exposure, the N 2 -guanine adducts are relatively stable (16). In E. coli, replication efficiencies past the BDO and BDE N 2 -guanines are significantly reduced in vivo, and the presence of these lesions in DNA is a complete block to synthesis by E. coli pol I, II, and III in vitro (18).
Primer extension reactions were carried out to test the ability of yeast pol to perform translesion DNA synthesis on the BDO and BDE N 2 -guanine-adducted DNA substrates ( Fig. 2A). Primers were designed that provided "running start" (Ϫ6 primer) and "standing start" (Ϫ1 primer) conditions. As shown in Fig. 2A, yeast pol replicated through all four butadiene lesions, resulting in full-length products. However, pol displays a strong stall site one nucleotide before the DNA lesion (lanes 3-6), suggesting an inhibition of nucleotide incorporation opposite the lesion. Interestingly, the bypass efficiency of pol seems to show stereospecificity. On the BDO N 2 -guaninecontaining substrates, as well as on the BDE N 2 -guanine adducts, translesion DNA synthesis was more efficient in the case of the S stereoisomers.
Primer extension reactions using E. coli pol I (Klenow fragment) were also carried out on substrates containing the BDO and BDE N 2 -guanine adducts (Fig. 2B). These data confirm previous reports that BDO and BDE N 2 -guanines block DNA replication by pol I (18). This polymerase incorporated one nucleotide opposite the lesion but in contrast to the yeast pol , failed to extend the primer further on all four damaged substrates tested. Additionally, heterogeneity in the mobility of the final products (26-mer in the running start assays and 21-mer in the standing start assays) suggested nucleotide misincorporation in these reactions (lanes 3-6 and 9 -12).
Next, the specificity of nucleotide incorporation by pol opposite and downstream of these lesions was examined. To identify the nucleotide that was incorporated by pol opposite the adducted base, single-nucleotide incorporation experiments were carried out using the Ϫ1 primer (Fig. 3). On a nondamaged substrate, pol predominantly incorporated a C opposite G, but some T was also incorporated. In the case of the BDO and BDE N 2 -guanine-containing substrates, a C residue was the only base that was incorporated opposite the lesions. Taking into account the lower efficiency of primer extension by pol on the adducted templates, the substrate to enzyme ratio in the reaction was changed from 5:2 to 1:4. Under these conditions, no nonextended primers were left in the reactions on all five substrates tested when dCTP was in the incubation mixture (data not shown). Again, in the presence of the dTTP, no primer extension was observed on any of the damaged DNA substrates, but primer extension occurred on the nondamaged template.
Although in reactions with the Ϫ1 primer no significant level of misincorporation opposite the lesion was observed, the smearing of bands was noted one nucleotide beyond the lesion, particularly in the case of the (R)-BDO N 2 -guanine-adducted Bypass of N 2 -Guanine Butadiene Adducts by pol substrate (Fig. 2A). To determine whether this smearing was attributable to nucleotide misincorporation past the lesion site, single-nucleotide incorporation studies were performed on (R)-BDO N 2 -guanine-adducted template using a 0 primer, which contains a C opposite the damaged G. On this template, pol extended 93% of the 0 primer with dCTP, 23% with dTTP, 6% with dATP, and 2% with dGTP (Fig. 4). A low level of primer extension was also observed in the presence of dTTP on the (S)-BDO N 2 -guanine-containing template. On the (R,R)-and (S,S)-BDE N 2 -guanine-containing templates, pol incorporated only the C residue.
To test whether any misincorporation occurred beyond one nucleotide downstream of the lesion site, we performed singlenucleotide incorporation experiments using a ϩ1 primer (   damaged substrates examined, and pol synthesized nearly the same amount of DNA on different damaged substrates when all four dNTPs were added to reactions. To quantitate the efficiency of pol -catalyzed synthesis past each of the BDO-and BDE-modified N 2 -guanines, steady state kinetic analyses were performed with both Ϫ1 and 0 primers. As shown in Table I, incorporation of dCTP opposite the (S)-BDO and (S,S)-BDE N 2 -guanines (Ϫ1 primer extension) was 200 -300-fold less efficient than incorporation opposite the unmodified guanine, whereas incorporation opposite the R stereoisomers was 2000 -3000-fold less efficient than incorporation opposite the unmodified guanine. The reduced efficiency for incorporating dCTP opposite BDO-and BDE-adducted N 2guanines is primarily a K m effect, not a k cat effect. Thus, there is a block to inserting dCTP opposite these lesions, as is also demonstrated by the pause site just prior to the adduct in Fig.  2A. The extent of the blockage depended on the stereochemis-try of the adduct, and this result agrees with the data presented in Figs. 2A and 3. Interestingly, there is little block to extending from the C residue paired with the BDO-or BDEmodified N 2 -guanine (kinetics of the dCTP incorporation in reactions with 0 primer), as is also demonstrated by the lack of a pause site at the site of the adduct (Fig. 2A). In contrast to nucleotide incorporation opposite the lesion, no differences in efficiencies of elongation from the resulting base pair were observed. Thus, bypass efficiencies by pol on the BDO-or BDE-modified N 2 -guanines are limited at the step of the nucleotide incorporation opposite the lesion but not at the extension step.
To further evaluate the accuracy of pol replication through BDO and BDE N 2 -guanine adducts, kinetic analyses of nucleotide misinsertion were carried out, and frequencies of misincorporation were calculated as the ratio of k cat /K m of the incorrect nucleotide to the correct nucleotide (21). In reactions with the Ϫ1 primer, frequencies of misincorporation were below the limit of detection under conditions used for all four damaged substrates. Thus, pol incorporates the correct nucleotide C quite accurately opposite N 2 -guanine modified with BDO or BDE. In experiments utilizing the 0 primer, high frequencies of misincorporation were observed in extension from C basepaired with the (R)-BDO N 2 -guanine. On this substrate, the frequencies of misincorporation were 2.0 ϫ 10 Ϫ2 for a T misincorporation and 6.2 ϫ 10 Ϫ4 for an A misincorporation. In all other cases, nucleotide misincorporation was below the limit of detection, which was approximately 5 ϫ 10 Ϫ4 . Thus, kinetic data confirmed the results of the single-nucleotide incorporation experiment (Fig. 4), indicating that pol is less accurate in extension from the base paired with (R)-BDO N 2 -guanine than with (S)-BDO N 2 -guanine.
Lack of Bypass of (R,R)-and (S,S)-BDE N 2 -Guanine-N 2 -Guanine Cross-links by pol -Structures of the BDE N 2 -guanine-N 2 -guanine cross-links are shown in Fig. 6. Cross-linked adducts are believed to contribute to butadiene-mediated carcinogenesis (22,23). Previously, in E. coli, both (R,R)-and (S,S)-BDE N 2 -guanine-N 2 -guanine cross-links were shown to be extremely inhibitory to replicative bypass in vivo, and E. coli DNA pol I, II, and III were shown to be completely blocked on the templates containing these cross-links in vitro (19). To examine whether yeast pol can bypass these lesions, primer extension experiments were performed. As shown in Fig. 7, on the (R,R)-as well as on the (S,S)-BDE N 2 -guanine-N 2 -guanine cross-link-containing substrates, DNA synthesis by pol was  completely blocked just before the lesion, both under standing start and running start conditions.

DISCUSSION
Based on the ability of pol to bypass a T-T dimer efficiently and accurately, it has been suggested that its active site is flexible enough to tolerate the distortion of the Watson-Crick geometry caused by the T-T dimer (6,7,24). However, such a flexibility of the polymerase active site should decrease its overall fidelity. Indeed, steady state kinetics assays of nucleotide incorporation have shown that pol is a low fidelity enzyme. Both yeast and human pol misincorporate nucleotides on undamaged DNA with frequencies of approximately 10 Ϫ2 -10 Ϫ3 (6,24,25). Interestingly, the accuracy of replication by yeast as well as by the human pol opposite a T-T dimer does not differ from that opposite nondamaged DNA (6,7). The fact that both yeast (4,26) and human (25) pol do not possess any intrinsic proofreading exonuclease activity could explain in part the low fidelity of these polymerases. However, pol has a lower fidelity than the other 3Ј3 5Ј exonuclease-deficient DNA polymerases (6,24,25), suggesting that its low fidelity derives from the relaxed requirement of its active site for correct base-pairing geometry. A flexible active site should enable pol to bypass DNA lesions other than the T-T dimer. In agreement with this, both yeast and human pol also bypass a 7,8-dihydro-8-oxoguanine lesion efficiently, and they do so by predominantly inserting a C opposite the lesion (11). In addition, both yeast (26) and human (27) pol preferentially insert the correct nucleotide (C) opposite an N 2 -acetylaminofluorene-guanine. However, yeast pol is unable to further extend DNA synthesis beyond the lesion (26). Human pol can incorporate relatively efficiently one more nucleotide beyond the lesion, but only when the modified guanine is primed with a C (27).
Here it has been shown that yeast pol can bypass (S)-BDO N 2 -guanine as well as (S,S)-BDE N 2 -guanine with 200 -300fold less efficient nucleotide insertion opposite the lesion relative to the nondamaged guanine. pol can also bypass the (R)-BDO N 2 -guanine and (R,R)-BDE N 2 -guanine adducts, but these lesions pose an approximately 10-fold greater block to replication by pol than their S stereoisomers. Thus, the efficiency of translesion DNA synthesis by yeast pol is stereoisomer-specific. Blockage of the pol -catalyzed replication through the BDO and BDE N 2 -guanines occurs at the step of the nucleotide insertion opposite the lesion, not at the exten-sion step. In its ability to effectively extend synthesis past the BDO and BDE N 2 -guanine adducts, yeast pol differs from E. coli pol I, which fails to continue DNA synthesis beyond the lesion. Single-nucleotide incorporation experiments on BDOand BDE N 2 -guanine-containing substrates and steady state kinetic data indicate that lesion bypass by pol can be errorprone at the step of postlesion replication and that the accuracy of translesion DNA synthesis at this step can also be stereoisomer-specific. On three out of four substrates tested, namely on (S)-BDO, (R,R)-BDE, and (S,S)-BDE N 2 -guanine DNA adducts, nucleotide insertion opposite the lesion as well as elongation from the resulting base pair appeared to be quite accurate. On the (R)-BDO N 2 -guanine-containing substrate, pol inserted the correct nucleotide opposite the lesion, but it showed a tendency for nucleotide misincorporation in elongation from the resulting base pair.
Stereoisomeric BDE N 2 -guanine-N 2 -guanine intrastrand cross-links were also examined in this study. However, these lesions were a complete block to synthesis by yeast pol , and in this case, synthesis terminated one base prior to the first adducted guanine. Interestingly, it has been recently demonstrated that human pol is capable of inserting a C opposite the first G of a cisplatin-GG intrastrand cross-link, but incorporation of the second C was highly inefficient, even using higher concentrations of pol in the reaction. When the cisplatin cross-link was primed with a CC opposite the lesion, bypass was achieved (27).
The N 2 -guanine adducts of stereoisomeric 1,3-butadiene metabolites are a complete block to synthesis by E. coli DNA polymerases I, II, and III. In contrast, yeast pol can insert nucleotides opposite these lesions and is able to efficiently extend from the resulting base pair. The ability of yeast pol to bypass N 2 -guanine butadiene adducts provides further support to the hypothesis (6,7,24) that in general, the pol active site tolerates geometric distortions within DNA caused by these and other DNA-damaging agents. However, the inability of pol to bypass an N 2 -guanine-N 2 -guanine intrastrand cross-link suggests that its active site is not flexible enough to adapt to the rather severe distortion imposed upon DNA by the cross-link.