Preferential Misincorporation of Purine Nucleotides by Human DNA Polymerase η Opposite Benzo[a]pyrene 7,8-Diol 9,10-Epoxide Deoxyguanosine Adducts*

Human DNA polymerase η was used to copy four stereoisomeric deoxyguanosine (dG) adducts derived from benzo[a]pyrene 7,8-diol 9,10-epoxide (diastereomer with the 7-hydroxyl group and epoxide oxygen trans (BaP DE-2)). The adducts, formed by either cis or trans epoxide ring opening of each enantiomer of BaP DE-2 by N 2 of dG, were placed at the fourth nucleotide from the 5′-end in two 16-mer sequence contexts, 5′∼CG*A∼ and 5′∼GG*T. polη was remarkably error prone at all four diol epoxide adducts, preferring to misincorporate G and A at frequencies 3- to more than 50-fold greater than the frequencies for T or the correct C, although the highest rates were 60-fold below the rate of incorporation of C opposite a non-adducted G. Antito syn rotation of the adducted base, consistent with previous NMR data for a BaP DE-2 dG adduct placed just beyond a primer terminus, provides a rationale for preferring purine misincorporation. Extension of purine misincorporations occurred preferentially, but extension beyond the adduct site was weak withV max/K m values generally 10-fold less than for misincorporation. Mostly A was incorporated opposite (+)-BaP DE-2 dG adducts, which correlates with published observations that G → T is the most common type of mutation that (+)-BaP DE-2 induces in mammalian cells.

Somatic mutation of proto-oncogenes and tumor suppressor genes is a key component in the initiation of cancer (1). Any DNA damage that escapes repair could lead to mutations. However, replicative DNA polymerases are blocked in a potentially lethal manner when they encounter bulky adducts (2). Thus, they are unable to convert adducts to mutations. Recently, a growing number of DNA polymerases capable of conducting translesion DNA synthesis have been discovered (3)(4)(5)(6)(7)(8)(9)(10)(11); these likely hold the key to mutagenesis and the initiation of cancer induced by bulky adducts. One of these lesion-bypassing DNA polymerases, human DNA polymerase eta (pol) 1 (12,13) is a member of the UmuC/DinB/Rev1/Rad30 superfamily (now called the Y family (14)) of DNA polymerases. In humans it is a product of the XPV skin cancer susceptibility gene; inactivation of pol results in a variant form of xeroderma pigmentosum (12,15). pol incorporates mostly correctly at cis-syn T-T cyclobutane dimers (16,17), N-(deoxyguanosin-8yl)-acetylaminofluorene (18), and cis-Pt G-G adducts (18,19); small amounts of incorrect nucleotides are also incorporated, mostly at the level seen with undamaged DNA. With  T-T photoproducts, however, a single G is preferentially incorporated (20).
Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous environmental carcinogens (21). Bay-region diol epoxides, formed on an angular benzo-ring, have been shown to be the ultimate carcinogenic metabolites of the PAHs (21)(22)(23)(24)(25). Four optically active bay-region diol epoxide isomers (enantiomers of a pair of diastereomers) are formed metabolically from a given hydrocarbon. In the DE-1 ("syn") diastereomer, the epoxide oxygen and benzylic 7-hydroxyl groups are cis, whereas in the DE-2 ("anti") diastereomer these groups are trans (see Fig. 1A below). For bay-region diol epoxides, the DE-2 isomer with R,S,S,R absolute configuration exhibits by far the greatest carcinogenicity compared with the other three optically active isomers, whereas for fjord-region diol epoxides significant carcinogenic activity is not limited to this isomer (26).

MATERIALS AND METHODS
Oligonucleotides-Oligonucleotides containing a single specific cis or trans BaP DE-2 dG adduct placed on the fourth nucleotide from the 5Ј-end in either sequence context III(G) (5Ј-TTCG*AATCCTTCCCCC-3Ј) or IV(G) (5Ј-GGGG*TTCCCGAGCGGC-3Ј) were synthesized as described previously (29,30). The basis for selection of these sequences has been described previously (30). Polyacrylamide gel-purified, nonadducted oligomers were purchased from either Lofstrand Laboratories Limited (Gaithersburg, MD) or Midland Certified Reagent Co. (Midland, TX) and were purified further by reverse-phase high performance liquid chromatography if necessary.
DNA Polymerase Reactions-His-tagged recombinant DNA polymerase was produced as described previously (18). Polymerase reactions contained final concentrations of 40 mM Tris-HCl (pH 8.0), 1 mM MgCl 2 , various concentrations of the four dNTPs, 10 mM dithiothreitol, 250 g/ml bovine serum albumin, 60 mM KCl, 2.5% glycerol, 40 nM 5Ј-32 P-labeled primer previously annealed to 60 nM 16-mer template (by heating at 95°C for 5 min, and slowly cooling), and 2 nM DNA polymerase in a 10-l volume. Reactions proceeded at 37°C for 3-15 min. Reactions were stopped by addition of 10 l of 90% aqueous formamide containing EDTA (25 mM) and gel sequencing dye solution, followed by heating in a boiling water bath.
Kinetic Analysis of Incorporation-Reaction products were subjected to electrophoresis on 20% polyacrylamide-7 M urea gels run at 60 watts. After drying the gels, the extent of incorporation was quantitated with a Fujifilm Fluorescent Image Analyzer and Image Gauge V3.12 software. Steady-state kinetic experiments followed the procedures of Creighton et al. (31) whereby maximum incorporation is kept below 20%. Rectangular hyperbolic fits yielding apparent V max and K m values were calculated using SigmaPlot 2000. Relative values of V max /K m for incorrect versus correct nucleotide incorporation were used to calculate the misincorporation efficiency (f inc ).

RESULTS
Misincorporation on the Non-adducted Sequence Context III(G) DNA Template-Previous studies of pol have used templates that ranged from 30 to 75 bases in length (17,18,28,32). We first examined our shorter context III(G) 16-mer template to determine its efficacy as a substrate. In standing start assays with a 12-mer primer, non-adducted 16-mer template of context III(G) DNA (12/16 DNA) and dCTP at 1 mM concentration, the primer was mostly extended by one nucleotide (Fig.  2A). When all four dNTPs were present at 1 mM further extension occurred, with limited procession to the end of the template ( Fig. 2A). Misincorporation efficiencies (f inc ) were measured in standing start, steady-state kinetic assays. Table I shows similar misincorporation efficiencies of around 2 ϫ 10 Ϫ3 for all three incorrect nucleotides. These misincorporation efficiencies are somewhat lower than those reported by Johnson et al. (17) for the three nucleotides and Matsuda et al. (32) for G. Matching f inc values are not expected because of differences in assay conditions and templates. However, observation of misincorporation frequencies that fall within the range of 10 Ϫ2 to 10 Ϫ3 previously observed for pol increases our confidence that correct f inc values are likely to be obtained from kinetic studies with 12/16 oligomers containing adducts at the fourth nucleotide (dG) from the 5Ј-end of the 16-mer template.
Patterns of Incorporation in Sequence Context III(G) with BaP DE-2 dG Adducts- Fig. 2B shows dNTP incorporation across from each of the four stereoisomeric BaP DE-2 dG adducts, compared with the non-adducted template. Assays were Cis opened adducts retain configuration at C-10, whereas trans adducts have inverted configuration. C, the DNA 16-mers, contexts III(G) and IV(G), used as templates in the present study. The adduct is indicated as G*. performed with 0.2 mM dNTP concentrations. It is clear that when any one of the stereoisomeric BaP DE-2 adducts is present, purine nucleotides are incorporated far more than either T or the correct C. Misincorporation of A is most extensive with the trans S adduct.
For each dNTP, more incorporation is seen with the transring opened adducts compared with the cis adducts with the same R or S configuration at C-10. Less purine misincorporation occurs at cis adducts than trans adducts, particularly the cis S adduct, but incorporation of the correct C is also lower. Incorporation at cis R adducts formed from the carcinogenic (ϩ)-BaP DE-2 is greater than incorporation at cis S adducts formed from the non-carcinogenic (Ϫ)-BaP DE-2.
The results shown in Fig. 2B for the BaP DE-2 trans S adduct agree with those obtained by Zhang et al. (28), who examined only this adduct, in that A is incorporated the best, but differ significantly in that we observed much higher G incorporation and much lower T incorporation. Differences between the assays, such as DNA sequence context or assay buffer composition and pH, could underlie the differences. Table II shows V max /K m data for incorporation of individual nucleotides by pol opposite BaP DE-2 dG adducts in sequence context III(G). In the presence of these adducts, the highest V max /K m values observed are about two orders of magnitude lower than the V max /K m for incorporation of the correct C at non-adducted G ( Table I). The adducts constitute a fairly strong block to the enzymatic activity of pol. Notably, however, the adducts have a blocking effect only on incorporation of a correct C, whereas incorrect base incorporation is either changed very little or actually enhanced in the presence of adducts. Thus, the presence of an adduct lowers V max /K m for C incorporation by factors of 1600 to more than 18,000 whereas V max /K m for T incorporation decreases by 2.2to 16-fold, G incorporation rises up to 3-fold and A incorporation rises 1.5-to 7.4-fold.

Kinetics of Misincorporation at BaP DE-2 dG Adducts in Sequence Context III(G)-
Typical kinetic data for the trans S dG adduct are shown in Fig. 3. For all four adducts, incorporation of the correct C is the lowest of the four dNTPs. In the assays, there is evidence for inhibition by dNTP concentrations greater than 2 mM, which limits the amount of dCTP that can be used to increase incorporation. This makes it difficult to determine a precise V max /K m value for C incorporation at cis S adducts and precise f inc values for G, A, and T misincorporation. Still, it is clear that incorporation of purines far exceeds that of pyrimidines opposite all four adducts in sequence context III(G). The preference for G misincorporation over correct incorporation of C ranges from 5.5-fold to more than 50-fold, whereas the preference for A over C ranges from 20-fold to more than 50-fold.
With all four dNTPs, V max /K m values for purine and T misincorporation are generally greater for trans ring-opened ad-ducts than cis ring-opened adducts, whether comparing the pairs trans R versus cis R or trans S versus cis S (members of each pair derived from opposite enantiomers of the diol epoxide) or the pairs trans R versus cis S or trans S versus cis R (members of each pair derived from the same diol epoxide enantiomer so that the configuration differs only at C-10). V max /K m values for A, T, and C incorporation are greatest at trans S adducts, whereas V max /K m for G incorporation is greatest at trans R adducts. The overall rate of incorporation (sum of V max /K m values) follows the pattern trans R Ն trans S Ն cis R Ͼ Ͼ cis S. Values of f inc for A and G, however, are much greater for cis adducts than trans adducts, because of far lower incorporation of C at cis adducts. Values of f inc for G versus A are similar with all of the adducts except trans S, whose f inc for A is four times larger than f inc for G.
Kinetic Comparisons of Incorporation Across from BaP DE-2 dG Adducts in Sequence Contexts III(G) and IV(G)-Preferential incorporation of purine nucleotides over pyrimidine nucleotides is seen with both sequence contexts III(G) (5ЈϳCG*Aϳ) (Table II) and IV(G) (5ЈϳGG*Tϳ) (Table III). But, except for the trans S isomer, the V max /K m for misincorporation is far lower and V max /K m for C is unchanged or higher, with the result that f inc values are far lower in context IV(G). This is particularly so with the cis isomers. With the trans R, cis R and cis S isomers, the decreased preference for purine misincorporation is due to the decline in V max /K m values in context IV(G) compared with context III(G) being greater for the purines than the pyrimidines. With the change in sequence context, the V max /K m values for purine misincorporation drop an order of magnitude for trans R and cis R adducts and 3-to 5-fold for cis S adducts. With trans S adducts, V max /K m values are greater in sequence context IV(G), substantially so with G and T incorporation. At the same time, when the sequence is changed to IV(G), C incorporation efficiency declines only 4-fold for trans R adducts and either stays about the same or rises for the other three adducts. Consequently, A incorporation at trans S adducts has the highest efficiency of misincorporation (f inc ) of all.
Patterns emerge from trans versus cis comparisons of V max /K m values for incorporation in sequence context IV(G) that are like the ones seen in sequence context III(G). V max /K m values for purine misincorporation are appreciably larger for trans adducts. However, because C incorporation is so different at context IV(G) adducts, f inc values for G and A for trans adducts are no longer much smaller than for cis adducts, as a In units of micromolar primer extended/M dNTP/min; means and standard deviations generated from at least three kinetic assays.
d Template G at position 21 (from 3Ј) in 30-mer with local sequence 5Ј-TGG. was seen in context III(G). Patterns based on R versus S comparisons in sequence context IV(G) are not evident. The effect of changing the sequence from III(G) to IV(G) on T misincorporation parallels the effect seen with the purines. V max /K m for T misincorporation is higher at trans S adducts, and lower at the other three. Combined with the aforementioned changes in C incorporation, the result is a decrease in f inc for T at cis adducts, little change at trans R adducts and a rise at trans S adducts. Table IV summarizes the V max /K m data for extension of 13-mer primers containing either a correct C or a mispaired base opposite the adducts by incorporation of a correct G opposite the C immediately 5Ј to the adduct position in template sequence III(G). Data for the control, unadducted template are included for comparison. Notably, extension beyond the adducts is very inefficient, especially for both the trans adducts as well as the cis S adduct. In the absence of an adduct, extension of the correctly paired C-containing primer is favored by a factor of 30 to 100. Strikingly, however, for all but the cis R adduct, extension of mispaired primers containing a purine opposite the adduct is favored (Ն 3-fold) relative to extension of primers containing the correctly paired C. Thus, once an incorrect base is inserted opposite these adducts, pol exhibits a bias in favor of extending the mispaired primer. The cis R adduct constitutes an exception, in that extension of the correctly paired primer is substantially preferred over that of primers containing a mispair (although only 2-fold over extension of a misincorporated A), and this correct extension occurs with the highest efficiency of any of the adducts. DISCUSSION A Mechanism for Preferential Purine Nucleotide Misincorporation Across from BaP DE-2 dG Adducts-All four BaP DE-2 dG adducts represent significant blocks to C incorporation. In  22-28). B, hyperbolic Michaelis-Menten plots of the rates of incorporation using the data obtained from A; units of V max /K m are as in Table II.

TABLE III Fidelity of pol at BaP DE-2 adducts in sequence context IV(G)
For template sequence see Figure 1C. sequence context III(G), incorporation of the correct C drops as much as 20,000-fold relative to the corresponding rate in the absence of adduct. In contrast, the rates of incorporation of the purine nucleotides rise slightly (less than 10-fold compared with their rates in the absence of adduct). Consequently, the efficiency (f inc ) of A misincorporation at BaP DE-2 dG adducts in sequence context III(G) ( Table II), relative to A misincorporation in normal DNA (Table I), jumps over four orders of magnitude from 1.8 ϫ 10 Ϫ3 to a range of 20 to over 50 times C incorporation (Table II), and the efficiency of G misincorporation jumps nearly as much.
We propose the following mechanism for preferential purine misincorporation opposite the BaP DE-2 dG adducts: 1) Stacking of the hydrocarbon with the preceding base pair causes the adducted dG to assume the unusual syn rather than the usual anti torsion angle, with the stacked hydrocarbon and the syn dG conformation being accommodated by the open pol active site; 2) the enzyme misincorporates purine nucleotides opposite this syn dG adduct by Topal and Fresco (33) purine-purine base pairing, and 3) after purine misincorporation the adduct remains intercalated in the DNA. Evidence supporting this mechanism comes from the structural information that follows.
pol exhibits unusually high misincorporation frequencies of 10 Ϫ2 to 10 Ϫ3 in normal, non-adducted DNA (17,32). This indicates an active site whose discrimination in favor of Watson-Crick and other base pairs with similar geometry is considerably relaxed compared with many other non-proofreading polymerases but is still 100-to 1000-fold. Crystal structures for members of the Y family of DNA polymerases (34 -37), including the catalytic core of pol from yeast (35), have been reported. Although Y family polymerases have a classic DNA polymerase palm structure, the finger and thumb structures are smaller than usual. An additional structure called a polymerase-associated domain (35) or little finger (36) is present. Despite the latter, the Y-family DNA polymerase active site is more open where the incoming deoxyribonucleotide binds, and the degree of interaction with the primer-template complex is less than what is seen with high fidelity DNA polymerases. These features, which are inferred by homology modeling with T7 DNA polymerase in two studies (34,35) and are observed directly in a third (36), could underlie the relative lack of fidelity exhibited toward non-adducted DNA, and the ability to accommodate bulky adducts attached to syn-rotated dG and incorporate a nucleotide across from such lesions (34 -37).
The solution NMR structure of a BaP DE-2 trans S dG adduct bound to the fourth base from the 5Ј-end of a 13-mer template (5Ј-AACG*Cϳ) adjacent to the terminus of a 9-mer primer strand provides a model for a partially extended DNA duplex as "seen" by a polymerase. This structure shows that prior to insertion of the complementary nucleotide, the adducted dG is rotated anti to syn to allow the hydrocarbon moiety to stack with the base pair formed by the 3Ј-terminal primer base and the base immediately 3Ј to the modified dG on the template strand (38). After a complementary C is incorporated opposite the BaP DE-2 trans S dG adduct, the adducted dG assumes the normal anti conformation and the adduct swings out into the minor groove (39). This same groove-bound, anti conformation of the adduct is maintained upon further extension of the primer chain beyond the adduct site, as is shown by the structure of a fully complementary 11-mer duplex (40) with the same local ϳCG*Cϳ sequence around the adduct.
There are no NMR data for mismatched duplex structures corresponding to incorrect purine nucleotide insertion opposite BaP dG adducts; however, substantial UV red shifts are observed for the long-wavelength pyrene band of duplexes containing purine mismatches opposite BaP trans R and trans S dG adducts in the same ϳCG*Cϳ context (41), from 346 nm (when C is paired with the adducted G) to around 350 nm in the G and A mispaired duplexes. These shifts are indicative of base stacking, and suggest that the hydrocarbon remains intercalated (41), with the adducted dG presumably still in the syn conformation, after an incorrect purine nucleotide is inserted opposite the adduct and the primer is further extended.
If the adducted dG is in the syn conformation before and after polymerization then presumably it is also syn when the incoming nucleotide binds and is incorporated. A syn glycosidic torsion angle of the adducted dG in the template allows purinepurine base pairing of the type proposed by Topal and Fresco (33). Molecular modeling (42) of the BaP DE-2 trans S adduct in a 13-mer template, 10-mer primer structure (the 13-mer template and 9-mer primer described above with the primer extended to include A across from the adduct) suggests that the size and shape of the (syn-adducted-G)⅐(anti-A) pair very much resembles that of a normal T⅐A pair. The 100-to 1000-fold preference for base pairs having the approximate size and shape of a Watson-Crick base pair could also cause the enzyme to select syn-adducted G paired with anti-A over pairing with the correct C. The observed red shifts (41) increase (more favorable base stacking) in the same order as the present V max /K m values for pol misincorporation of purines at trans adducts.
Even though an NMR structure could not be determined (38) when the trans S dG adduct above is replaced by a trans R dG adduct, it is likely that the pol preference for forming (adducted-G)⅐(purine) mispairs with the BaP trans R adduct can be explained by a mechanism similar to that described above for the trans S adduct. UV spectra of oligonucleotide duplexes containing the trans R dG adduct mispaired with purines show significant red shifts of the long wavelength absorbance (41), consistent with base stacking and syn glycosidic rotation, and are indicative of a structure analogous to that observed with the trans S dG adduct.
Structural analysis of either a cis R or cis S BaP DE-2 dG adduct placed in the same position as the trans adducts just described, either before or after a correct or incorrect incorporation of nucleotide has yet to be reported. In fully duplex DNA, the guanine base bearing either a cis R (43) or a cis S (44) BaP adduct is displaced out of the DNA helix as is its complementary C, and neither NMR nor UV studies with purine-purine mispairs of cis adducts are available. Thus no structural information exists that might explain preferential purine incorporation versus C incorporation at the present cis adducts; however, as in the case of the trans adducts, the observed preference for purine misincorporation presumably could also result from anti to syn rotation of the adducted dG and Topal and Fresco (33) base pairing. 8-Oxo-dG forms stable G⅐A pairs in DNA with the G rotated anti to syn (45), resulting in G to T transversions (A misincorporation) with pol␤ (46). In mammalian cells, G 3 T transversions also commonly occur at N-(deoxyguanosin-8-yl)-acetylaminofluorene adducts depending on the DNA sequence (47) and are attributed to the adducted dG adopting a syn conformation (48). pol, however, mostly incorporates the correct C across from these two types of C-8 G adducts (16,49), suggestive that with these types of adducts the active site of pol forces the G to rotate back to the anti conformation. It is interesting that this occurs with these C-8 dG adducts but not with the present BaP DE-2 N 2 -dG adducts.
Effect of Sequence Context on Misincorporation-The most pronounced sequence effect seen in this study is the sharply lower set of V max /K m values for misincorporation observed with sequence context IV(G) (Table III) compared with sequence context III(G) (Table II), seen with all adducts except for trans S BaP DE-2. For the other adducts in sequence context IV(G), the decrease in the rate of purine misincorporation is more than an order of magnitude with the cis R adduct, about an order of magnitude with the trans R adduct and 3-to 5-fold with the cis S adduct. The V max /K m values for C incorporation decrease by small amounts with two of the adduct isomers and rise with the other two. The higher rate of C incorporation in sequence context IV(G) compared with context III(G) could arise from misalignment using the next 5Ј-template G in place of the adducted G.
With sequence context IV(G) the V max /K m for misincorporation of A at the trans S adduct is the largest for any nucleotideadduct combination and is 18-fold greater than for A misincorporation at the trans R adduct. This observation is of special interest in light of the fact that the trans S is the predominant adduct formed on DNA (50) by the highly carcinogenic (ϩ)-BaP DE-2 isomer whereas trans R is the predominant adduct formed by the weakly or non-carcinogenic (Ϫ)-BaP DE-2 isomer (26).
A significant finding of this study is that the sequence context effect on the efficiency of base misincorporation depends on the specific adduct. The present two sequences differ in both the 3Ј-and 5Ј-nearest neighbors to the adducted G. The 3Јnearest neighbor is A in sequence context III(G) and T in sequence context IV(G). Specific stacking interaction between an adduct and the preceding (3Ј) base pair is suggested by NMR analysis (38) of a partial template⅐primer duplex in which the 3Ј-nearest neighbor to the adducted G is C (see above). Small differences in stacking geometry between the adduct and different 3Ј-nearest neighbor bases and their complements could account for the observed differences in misincorporation frequency between templates III(G) and IV(G). Specific effects of the 5Ј-nearest neighbor are also possible, such as template slippage leading to enhanced correct insertion of C with template IV(G) as mentioned above or changes in stacking depending on the identity of the 5Ј-base. Our observed dependence of misincorporation on sequence suggests that it will be fruitful to focus bypass studies of misincorporation on specific proto-oncogene and tumor suppressor sequences.
Mutational Spectra of BaP DE-2 dG Adducts in Mammalian DNA-It is of considerable interest to establish what role pol plays in determining the mutational spectrum that has been observed for PAHs in mammalian cells. The preponderant products of in vitro reaction of BaP diol epoxides with dG in DNA are trans ring-opened adducts formed at the exocyclic nitrogen of dG. Upon reaction of (Ϫ)-BaP DE-2 with calf thymus DNA in vitro, trans R and cis S dG adducts are formed in the ratio of ϳ6:1 (50). At both types of adduct, pol strongly favored misincorporation of purines, with both G and A being misincorporated about equally and 5 to 20 times the rate for T misincorporation. Once purine misincorporation occurred, such mispairs were extended somewhat better than the correctly paired primers for three out of the four adducts studied in sequence III(G). However, the extension of these mispairs was still quite inefficient (V max /K m generally about 10-to 20-fold smaller than V max /K m for purine incorporation; Tables II and IV), suggesting that if purine misincorporation by pol plays a role in mutagenesis induced by BaP DE adducts, an additional polymerase or polymerases are likely to be required for full extension of the damaged DNA to give an observed mutagenic event. Precedent for such a mechanism is provided by the observation that pol and pol act sequentially to bypass abasic sites as well as the (6-4) T-T photoproduct (51).
The relative lack of T misincorporation by pol contrasts with limited data (52) on the mutational spectrum of (Ϫ)-BaP DE-2 in the HPRT gene in Chinese hamster V-79 cells, in which G 3 C, G 3 T, and G 3 A mutations (misincorporation of G, A, and T, respectively) occurred to similar extents (1.5: 1.0:0.8). Far more information is available about the spectrum of mutations at G produced by (ϩ)-BaP DE-2 in mammalian cells. In the HPRT gene in V-79 cells, the relative proportions of the three possible base substitutions at G are virtually independent of dose (53). At three different doses of (ϩ)-BaP DE-2, G 3 T transversions accounted for 67-75%, G 3 C transversions for 18 -21%, and G 3 A transitions for 7-11% of the total mutations (53). The use of repair-deficient Chinese hamster V-H1 cells (54) resulted in little change in the relative mutational frequencies at two different doses: 60 -63% of G mutations were G 3 T transversions, 20 -23% were G 3 C transversions, and 15-19% were G 3 A transitions. These frequencies of misincorporation are consistent with our pol results described below.
The spectrum of mutations produced in the ras proto-oncogene was determined (55) in mice whose skins were painted with the parent hydrocarbon, BaP. When forming bay-region diol epoxides from the (ϩ)-and (Ϫ)-enantiomers of BaP 7,8dihydrodiol, liver microsomes from 3-methylcholanthrenetreated rats produce predominantly (ϩ)-BaP DE-2 and much less of the other three diol epoxide isomers (56). Despite the different cell types and the overlay of BaP metabolism in the skin experiments, the mutational pattern is quite similar for BaP on mouse skin and (ϩ)-BaP DE-2 in hamster cells in culture. In mouse skin, the proportion of G 3 T transversions is slightly higher (83%), and G 3 C transversions (13%) and G 3 A transitions (4%) are correspondingly lower than what was seen in the cultured cell studies.
Misincorporation by pol at (ϩ)-BaP DE-2 adducts fits the above mutational spectra well. In vitro reaction of (ϩ)-BaP DE-2 with calf thymus DNA gives trans S and cis R adducts in the ratio of ϳ40:1 (50). Therefore, it is reasonable to assume that mutations arising at G in mammalian cells are likely to arise largely from the trans S adducts formed in multiple sites