phenanthrene−DNA Adducts in Mouse Epidermis ... - ACS Publications

586

Chem. Res. Toxicol. 1996, 9, 586-592

Benzo[c]phenanthrene-DNA Adducts in Mouse Epidermis in Relation to the Tumorigenicities of Four Configurationally Isomeric 3,4-Dihydrodiol 1,2-Epoxides Rajiv Agarwal,† Karen A. Canella,†,‡ Haruhiko Yagi,§ Donald M. Jerina,§ and Anthony Dipple*,† Chemistry of Carcinogenesis Laboratory, ABL-Basic Research Program, NCI-Frederick Cancer Research and Development Center, Frederick, Maryland 21702, and Laboratory of Bioorganic Chemistry, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892 Received August 22, 1995X 32P-Postlabeling assays were used to monitor the binding to epidermal DNA that resulted from the application of each of the four configurational isomers of benzo[c]phenanthrene 3,4dihydrodiol 1,2-epoxide to mouse skin in vivo. For three of these configurational isomers, there was a reasonable correlation between the relative level of binding to epidermal DNA and the known tumorigenic effects of these compounds. However, for the 4(S),3(R)-dihydrodiol 2(S),1(R)-epoxide, the tumorigenic response was considerably greater in relation to the level of DNA modification than was the case for the other isomers. This greater tumorigenic response was consistent with previous observations indicating that this isomer was more mutagenic, at equivalent levels of DNA modification, than the other two tumorigenic dihydrodiol epoxides. Additionally, the 4(S),3(R)-dihydrodiol 2(S),1(R)-epoxide reacts with DNA to generate predominantly (∼80%) adducts on the amino group of adenine residues. These findings might imply a greater intrinsic biological effect of such adenine adducts with respect to the other major adduct formed on the amino group of guanine residues.

Introduction It is known that polycyclic aromatic hydrocarbons require metabolic activation in order to express their carcinogenic potential and that the metabolites that have expressed the greatest tumorigenicity are bay-region or fjord-region dihydrodiol epoxides (1-3). Whereas the dihydrodiol epoxide diastereomer in which the benzylic hydroxyl group and the epoxide oxygen are trans to one another (frequently referred to as the DE-2 or anti diastereomer) is usually responsible for carcinogenic activity in the case of bay-region dihydrodiol epoxides, both diastereomers have been found to be tumorigenic in the case of the fjord-region dihydrodiol epoxides of benzo[c]phenanthrene (4, 5). When all four optically active benzo[c]phenanthrene dihydrodiol epoxides (BcPhDEs,1 see Figure 1) were subject to tests for tumorigenic activity on mouse skin, three of the four were found to be tumorigenic (6). The mutagenic activities of the four configurationally isomeric BcPhDEs have been examined in several systems. In Chinese hamster V 79 cells, the rank order of mutagenic activities closely paralleled the order for tumorigenicity, i.e., 4R,3S,2S,1R > 4S,3R,2S,1R > 4S,3R,2R,1S > 4R,3S,2R,1S, but this was not the case using Salmonella typhimurium strains for mutagenesis studies (7). In more recent studies using the pS189 shuttle vector system (8), mutagenicity was monitored under conditions designed to give equal extents of reac†

NCI-FCRDC. Present address: Laboratory of Comparative Carcinogenesis, NCI-Frederick Cancer Research and Development Center, Frederick, MD 21702. § National Institutes of Health. X Abstract published in Advance ACS Abstracts, March 1, 1996. 1 Abbreviation: BcPhDE, benzo[c]phenanthrene dihydrodiol epoxide. ‡

0893-228x/96/2709-0586$12.00/0

tion with DNA for each isomer (9). In these studies, the relative mutagenicities/DNA adduct of the four stereoisomers again did not match the relative tumorigenicities exhibited on mouse skin. These differences in relative biological activities suggested that (a) even though equal doses of dihydrodiol epoxide gave roughly similar extents of DNA modification in vitro (10), this must not hold in vivo, or (b) the intrinsic mutagenicities of these agents are unrelated to their tumorigenic properties. In order to determine the significance of the discrepancy between relative mutagenic activities toward the shuttle vector in vitro and relative tumorigenicity on mouse skin, experiments have been undertaken to monitor the extents of DNA adduct formation in vivo under conditions used in the earlier tumorigenicity study, utilizing the 32P-postlabeling protocol (11). To aid in the identification of adducts formed in mouse epidermis, 16 benzo[c]phenanthrene-deoxyribonucleoside 3′-phosphate markers (see Figure 1) were prepared previously (12). Here, the relative efficiencies of adduct detection by postlabeling for adducts from the four benzo[c]phenanthrene stereoisomers were determined using DNA modified by these agents in vitro. DNA adduct formation in mouse epidermis was then examined, and it was found that this varied considerably for the four isomeric dihydrodiol epoxides and that these data, in combination with the relative mutagenicities in the shuttle vector system, reasonably accommodate the observed tumorigenicities of the four optically active isomers.

Materials and Methods Materials. Caution: Some of the BcPhDE isomers used in this study have been shown to be tumorigenic in experimental animals, and they should be handled with appropriate precautions.

© 1996 American Chemical Society

Benzo[c]phenanthrene-DNA Adducts in Vivo

Chem. Res. Toxicol., Vol. 9, No. 3, 1996 587 The HPLC system used was a Hewlett Packard Model 1090 equipped with a diode-array detector. Radioactivity was measured using a Packard TRI-CARB 1500 liquid scintillation analyzer and the liquid scintillation cocktail Betafluor (National Diagnostics, Atlanta, GA). Modification of DNA with Dihydrodiol Epoxides. As described previously (15), solutions of DNA (1 mg/mL) in 0.1 M Tris-HCl (pH 7) were treated with 0.1 volume of an acetone solution of each of the four isomeric BcPhDEs (1 mg/mL) and incubated separately at 37 °C for 6 h. The mixture was then extracted three times with an equal volume of water-saturated 1-butanol, followed by three extractions with diethyl ether to remove tetraol hydrolysis products. The last traces of organic solvent were removed under a slow stream of nitrogen at room temperature. Aliquots of modified DNA were enzymatically digested to nucleosides using deoxyribonuclease I, snake venom phosphodiesterase [1 unit/mg DNA has been shown to be effective in fully digesting hydrocarbon adducted DNA (16)], and alkaline phosphomonoesterase (17). The hydrocarbon-nucleoside adducts were recovered on Sep-Pak cartridges as described earlier (12). These nucleoside adducts were then separated by HPLC, and the eluent was monitored at 255 nm using the conditions previously described (15). The adducts were quantified as optical density units at 255 nm. It is reasonable to assume that extinction coefficients for different adducts and for tetraol are all similar.

Figure 1. (A) Structures of the four configurationally isomeric dihydrodiol epoxides of benzo[c]phenanthrene. The isomers are distinguished in this article by reference to the configurations of the chiral centers listed in the sequence 4-, 3-, 2-, and 1-position. (B) Structures of the adducts derived from the R,S,S,R dihydrodiol epoxide isomer. Since the nucleic acid component can be derived from deoxyadenosine or deoxyguanosine, four adducts are generated from each dihydrodiol epoxide isomer. The structures of the adducts from the other three isomers bear the same relationship to their dihyrodiol epoxide structures as these do to the R,S,S,R structure. The four configurational isomers of BcPhDE were synthesized as described elsewhere (13, 14). Extinction coefficients (λmax, CH3CN) are 4.8 × 104 M-1 cm-1 (254 nm) and 4.4 × 104 M-1 cm-1 (260 nm) for the dihydrodiol epoxides in which the benzylic hydroxyl group and the epoxide oxygen are cis (DE-1) and trans (DE-2), respectively. Purity of the dihydrodiol epoxides was found to be g96% by trapping as the mercaptoethanol adducts (ref 13, supplementary material) followed by HPLC on a Sota C-18W column, 4.5 × 250 mm, eluted at 1.2 mL/min with a linear gradient of methanol in water that increased the methanol content from 40% to 60% over the first 20 min, followed by a linear gradient that increased the methanol to 90% over the next 15 min. Detection was at 255 nm. Elution times (min) are as follows. DE-1: trans-tetraol, 12.5; cis-tetraol, 18.5; mercaptoethanol adduct (from trans addition of the thiolate at C-1 of the epoxide), 19.4; DE-2: trans-tetraol, 12.5; mercaptoethanol adduct, 15.6. Enzymes and chemicals were purchased from the following sources: calf thymus DNA, nuclease P1, micrococcal nuclease, and apyrase (Sigma Chemical Company, St. Louis, MO), spleen phosphodiesterase (Boehringer Mannheim, Germany), T4 polynucleotide kinase (U.S. Biochemical Corp., Cleveland, OH), [γ-32P]ATP with an original specific activity of ∼3000 Ci/mmol (Amersham, Arlington Heights, IL), lithium formate (Johnson Matthey Catalog Co., Ward Hill, MA), aldehyde free formic acid (Fisher Scientific, Fair Lawn, NJ), PEI-cellulose TLC sheets (20 × 20 cm), Polygram CEL 300 PEI (Machery Nagel, Du¨ren, Germany), and C-18 Sep-Pak cartridges (Waters, Milford, MA).

DNA Adduct Formation in Mouse Skin. The dorsal skin of six week old female CD-1 mice, obtained from Charles River Breeding Laboratories, Inc. (Kingston, NY), was shaved 1 week prior to the topical application of dihydrodiol epoxide (10 nmol) in 200 µL of 5% dimethyl sulfoxide in acetone to groups of 15 mice for each dihydrodiol epoxide. A further group of 15 mice was treated with solvent only. Five animals from each of the five groups were killed by carbon dioxide inhalation at 4, 10, and 24 h. Treated areas of skin were excised, and the subcutaneous tissue was scraped away. The remaining skin was floated on trypsin/versene (Biofluids, Inc., Rockville, MD) for 3 h at room temperature. The epidermis was separated from the dermis by rolling it back in a single piece with a spatula (18, 19). The epidermal samples from the five mice in a given group were then pooled. DNA Isolation from Epidermis (20-22). The epidermal sheets in lysing buffer (12.5 mL) (10 mM Tris-HCl, 400 mM NaCl, 2 mM EDTA, pH 7.9) were disrupted at 4 °C with a Polytron at setting 6 for 45 s. This was followed by the addition of 0.125 mL of 20% sodium dodecyl sulfate and 12.5 mg of proteinase K (0.5 mL of 10 mM Tris-HCl, pH 8.0). The samples were put on a rocker and incubated overnight. Each sample was then extracted sequentially with 2 × 2 volumes of phenol/ chloroform/isoamyl alcohol, 2 × 2 volumes of chloroform/isoamyl alcohol, 2 × 2 volumes of chloroform, and 2 × 2 volumes of diethyl ether. Nucleic acids were precipitated by the addition of 0.1 volume of 2 M sodium acetate (pH 5.0) and 2.5 volumes of chilled ethanol followed by centrifugation for 20 min at 2250 rpm at 4 °C. The supernatant was discarded, and the pellets were washed with 70% ethanol and then absolute ethanol. The air-dried pellets were then resuspended in 10 mM Tris-HCl buffer (2.5 mL, pH 7.0) and incubated for 30 min at 37 °C with 17.5 µL of a 7.14 mg/mL solution of RNase I (Amersham Corp., Arlington Heights, IL). DNA was precipitated by the addition of sodium acetate and ethanol, spooled onto a glass rod, and washed with 70% ethanol and then with ethanol. The air-dried DNA was dissolved in 1.5 mL of sterile 10 mM Tris buffer on a rocking platform at 37 °C overnight. The 1.5 mL samples were added to 11.6 mL of a cesium chloride solution (290 g in 187.6 mL of 10 mM sodium phosphate, pH 7.1) along with 10 µL of a 10 mg/mL ethidium bromide solution. The samples were centrifuged in a Ti 70 rotor at 70 000 rpm for 14 h at 25 °C. The purple band containing the ethidium bromide intercalated with DNA was removed by syringe and extracted six times with an equal volume of water-saturated 1-butanol. The cesium chloride was removed by dialysis against 10 mM Tris-HCl buffer, pH 7.0 at 25 °C, to give a purified DNA solution. All

588 Chem. Res. Toxicol., Vol. 9, No. 3, 1996 volumes were brought to 1 mL, and 1:25 dilutions in 10 mM Tris-HCl buffer (pH 7.0) were made for quantitation by ultraviolet absorbance. 32P-Postlabeling Analyses of BcPhDE-Modified DNA. DNA samples (4 µg from the in vitro modified DNA) was incubated in duplicate with micrococcal nuclease (2 µg) and spleen phosphodiesterase (2 µg) in 10 µL of 100 mM sodium succinate, and 50 mM CaCl2, pH 6.0, at 37 °C for 3.5 h. Normal nucleotides were removed by incubation at 37 °C with nuclease P1 (4 µg) for 40 min (11). A 0.5 M solution of Tris base (1.92 µL) was added to the hydrolysates, and aliquots (equivalent to 7.13 ng of DNA) were 32P-postlabeled with 3 units of polynucleotide kinase and [γ-32P]ATP (100 µCi, ∼3000 Ci/mmol) at 37 °C for 30 min as described by Reddy and Randerath (11). Reactions were terminated by addition of apyrase (40 milliunits), and the total reaction mixture was spotted on a PEI-cellulose sheet for resolution by multidirectional TLC. In vivo DNA samples (2 µg) were digested similarly. The purified nucleoside 3′phosphate markers available from the previous studies were also postlabeled, as described earlier (12). Adducts were resolved using previously described TLC conditions (12), except the concentrations of the D1 and D5 buffers were changed from 0.1 M sodium phosphate to 1.0 M (pH 6.8). Thus, the 32P-labeled bisphosphate adducts were separated by thin layer chromatography using 1 M sodium phosphate (pH 6.8) for development onto a wick for D1, followed by 1.8 M lithium formate in 3.7 M urea (pH 3.24) for D3, 0.4 M sodium phosphate in 2.4 M TrisHCl that was 3.7 M in urea (pH 7.9) for D4, and finally a D5 development in the same buffer used for D1. Adduct spots were detected by autoradiography and identified by comparison with the authentic synthetic standards. For autoradiography, Kodak X0 Mat films were exposed at room temperature for 20 min for the in vitro experiments or for 2 days with an intensifying screen, for in vivo experiments. To detect very weak spots (adducts from the 4R,3S,2R,1S isomer in the in vivo experiment), exposure at -20 °C for 2 days with an intensifying screen was required. To quantify the radioactivity, appropriate spots were excised from TLC plates added to liquid scintillation fluid and evaluated in a liquid scintillation spectrometer. Appropriate blank areas of the chromatogram were also excised and counted, and these values were used to correct the above measurements. Before autoradiography, the four corners of the air-dried TLC plates were marked with a nonradioactive marker pen (Scienceware, Pequannak, NJ) for alignment after exposure and wrapped in Saran Wrap to prevent contamination of the cassette. Adduct yield is defined in this paper as the total radioactivity associated with known adducts. Adduct levels were calculated using the relative adduct labeling formula (11), and these values were usually converted into adducts/103 nucleotides for the in vitro experiments or adducts/106 nucleotides for the in vivo experiments.

Results The overall strategy adopted for these investigations was to modify calf thymus DNA in vitro with each of the four configurationally isomeric BcPhDEs at a relatively high level such that the adducts formed could be measured by their UV absorbance after separation by HPLC. These same samples were then analyzed by the postlabeling protocol, and by comparison of the findings on the same samples by the two different procedures, the relative efficiencies of adduct detection by the postlabeling protocol, for adducts from the four isomeric dihydrodiol epoxides, were evaluated. This latter information was then used to assess the relative binding of each isomeric dihydrodiol epoxide to DNA in mouse epidermis where the levels of binding were comparatively low and data could only be obtained by postlabeling. To establish appropriate procedures for postlabeling, the effect of nuclease P1 treatment on the recovery of 32P-labeled adducts was examined using native calf

Agarwal et al.

thymus DNA that had been modified in vitro by the 4S,3R,2R,1S BcPhDE isomer. With respect to total adduct recovery in the absence of nuclease P1, the presence of 0.5, 1.0, 1.5, and 2.0 µg of nuclease P1/µg of DNA led to adduct recoveries of 94%, 106%, 99%, and 73%. Thus, the recovery of labeled adducts was unaffected by the inclusion of nuclease P1 except perhaps when the nuclease was present at the highest level examined. On the basis of this observation, the nuclease P1 version of the postlabeling procedure (11) was employed in the following studies using 1 µg of nuclease P1 for every 1 µg of DNA analyzed. In view of the high levels of modification of the DNA samples derived from the in vitro study (this was necessitated by the need to undertake analyses by HPLC/UV methods), it was essential to establish that an excess of ATP was present in the labeling step of the postlabeling procedure. To this end, various amounts of DNA modified in vitro by the 4R,3S,2S,1R isomer, which is known to react more extensively than the other isomers with DNA (10), were subjected to postlabeling analyses, and the radioactivity in all the recovered adducts was monitored. Through the range 2.5, 5.0, 10.0, and 20.0 ng of DNA in the labeling reaction, total counts recovered in adducts increased linearly (0.9 × 106, 2.1 × 106, 4.3 × 106, and 8.8 × 106 cpm; correlation coefficient 0.99), indicating that ATP was in excess through this range of DNA sample size, and ∼7 ng aliquots were selected for use in further postlabeling analyses. DNA from mouse epidermis that had been exposed to 10 nmol of the R,S,S,R dihydrodiol epoxide isomer was used to establish the appropriate amounts of epidermal DNA for postlabeling with 100 µCi of [γ-32P]ATP. In this case, 0.5, 1.0, 2.0, and 4 µg aliquots of DNA gave a linear response (correlation coefficient 0.99) to postlabeling, with total cpm recovered in adducts ranging from 2.1 × 103, 6.8 × 103, 12.3 × 103, to 28.8 × 103 under these conditions. Aliquots of 2 µg of DNA were used, therefore, for subsequent analyses. Analyses of the four DNA samples modified by the four configurational isomers of BcPhDE in vitro showed that triplicate HPLC analyses of a given sample gave reproducible findings (i.e., differed by 20% of total adducts) and that the findings were essentially the same as those reported earlier (15). Duplicate postlabeling analyses, in which adducts were identified by comparison of chromatographic mobilities with those of the 16 postlabeled known marker adduct 3′-phosphates (12) run at the same time, similarly differed by less than 5%. In some cases, two adducts were not effectively resolved by the TLC procedures used, and quantification of each individual adduct was not possible (Figure 2) though total adduct formation was readily obtained. Comparison of the data for the in vitro DNA samples by the two analytical procedures indicated that, for a given dihydrodiol epoxide isomer, the adduct distributions, determined by either method, were similar. When all adducts were summed together, some variations were apparent, however, suggesting that postlabeling may vary somewhat with respect to the relative efficiencies with which adducts from the four isomeric dihydrodiol epoxides are detected (Figure 3). The axes are arranged in this figure so that the heights of the bars for the R,S,S,R isomer were similar for both determinations. However, as can be seen, the adducts from this isomer were detected by postlabeling with only 50% efficiency,

Benzo[c]phenanthrene-DNA Adducts in Vivo

Figure 2. Autoradiograms of TLC separations of 32P-labeled bisphosphate adducts from calf thymus DNA exposed in vitro to (a) the 4R,3S,2S,1R, (b) the 4S,3R,2S,1R, (c) the 4S,3R,2R,1S, or (d) the 4R,3S,2R,1S dihydrodiol epoxide isomers of benzo[c]phenanthrene. Individual bisphosphate adducts, identified by comparison with authentic standards, are labeled as deoxyadenosine (dA) or deoxyguanosine (dG) adducts, and the cis or trans opening of the epoxide is indicated by subscripts c or t, respectively.

Figure 3. Comparison of total adduct yields obtained by UV absorbance, after separation of nucleoside adducts by HPLC (black bars), with those obtained from 32P-postlabeling measurements of relative adduct labeling (11) (hatched bars) for calf thymus DNA modified by the four configurationally isomeric BcPhDEs in vitro. The findings are averages of duplicate postlabeling analyses and triplicate HPLC analyses. In the latter, adducts were quantified as OD units at 255 nm, expressed as fractions of the dihydrodiol epoxide dose bound to DNA, and then converted to adducts/103 nucleotides using 337 for the average molecular weight of a nucleotide.

and the efficiencies for the other isomers were somewhat lower: R,S,S,R, 50% > S,R,S,R ) R,S,R,S, 40% > S,R,R,S, 25%. Only the data for the S,R,R,S isomer seem substantially different from the others. The basis for variation in the relative efficiencies of adduct detection by postlabeling for adducts from different BcPhDE configurational isomers is not known, but similar variations have been reported by others (23). The chromatographic separations obtained for the postlabeling analyses of the epidermal DNA samples from mice exposed to 10 nmol of each BcPhDE stereo-

Chem. Res. Toxicol., Vol. 9, No. 3, 1996 589

isomer for 4 h are shown in Figure 4. Again, known markers were postlabeled and analyzed simultaneously so that, by comparison with these, specific adducts could be assigned to individual spots, as indicated in the figure. For the case of the 4S,3R,2R,1S dihydrodiol epoxide isomer, the cis and trans deoxyguanosine adducts overlapped one another, as did the cis and trans deoxyadenosine adducts, precluding assessment of individual adduct yields in this case. Similarly, the cis deoxyguanosine and cis deoxyadenosine adducts derived from the 4S,3R,2S,1R isomer were not fully separated. For each BcPhDE isomer, no major differences in distribution were seen at the three different time points and the product distribution for the well-separated adducts was similar to that found for DNA treated in vitro (Figure 5). This latter is most readily seen in the R,S,S,R BcPhDE data because binding was fairly substantial in this case. For comparison, in the in vitro modified DNA the trans deoxyadenosine product was present at about twice the level of the trans deoxyguanosine product (15). Although complete separation of all adducts was not achieved for the S,R,R,S and S,R,S,R dihydrodiol epoxides (see legend to Figure 5), total adduct formation for each dihydrodiol epoxide isomer at each time point was readily quantified by summing all the adduct spots. In Figure 6, the total binding data obtained experimentally for each dihydrodiol epoxide stereoisomer at each time point are shown by the horizontal lines in the columns, whereas the full column height represents these values after correction for the efficiencies derived from the studies with calf thymus DNA in vitro. It is clear that the extents of modification of the epidermal DNA by the four stereoisomers range from extensive for the 4R,3S,2S,1R isomer to barely detectable in the case of the 4R,3S,2R,1S isomer and that this contrasts sharply with the narrower range of binding observed for reactions in vitro (15), irrespective of whether the correction is made for the relative efficiencies of overall adduct detection by postlabeling or not. Figure 6 also illustrates the tumors/mouse reported by Levin et al. (6) for the four isomeric BcPhDEs under these conditions. The most striking feature of the comparison of binding and tumors/mouse is that the tumorigenicities of the 4R,3S,2S,1R and 4S,3R,2S,1R isomers are very similar, but the DNA binding of the latter is considerably less than that of the former at all time points. This observation suggests that the adducts formed by the 4S,3R,2S,1R isomer are inherently more tumorigenic than those of the other isomers.

Discussion The objective in these studies was to investigate the relationship between the tumorigenic responses to the four configurationally isomeric BcPhDEs in mouse skin and their levels of modification of epidermal DNA. Radiolabeled forms of the BcPhDEs were not available, and though some work has been done on the use of immunoassays to monitor BcPhDE-DNA adducts (24), we elected to use postlabeling methods because specific markers had already been prepared and characterized (12). However, as discussed recently by Baer-Dubowska et al. (25), a wide range of efficiencies of postlabeling from 3% to 60% have been reported in the literature, and therefore, we initially investigated the conditions for postlabeling and the efficiencies with which the BcPhDEDNA adducts could be detected.

590 Chem. Res. Toxicol., Vol. 9, No. 3, 1996

Agarwal et al.

Figure 4. Autoradiograms of TLC separations of 32P-labeled bisphosphate adducts from epidermal DNA of mice that had been exposed for 4 h to (a) solvent only, or (b) the 4R,3S,2S,1R, (c) the 4S,3R,2S,1R, (d) the 4S,3R,2R,1S, or (e) the 4R,3S,2R,1S dihydrodiol epoxide isomers of benzo[c]phenanthrene. Individual bisphosphate adducts, identified by comparison with authentic standards, are labeled as deoxyadenosine (dA) or deoxyguanosine (dG) adducts, and the cis or trans opening of the epoxide is indicated by subscripts c or t, respectively.

Figure 5. Distribution of BcPhDE-DNA adducts in mouse epidermal DNA at 4, 10, and 24 h after treatment with 10 µmol of one of the configurational isomers. *The dGuoc and dAdoc adducts were not cleanly resolved, and the entry for dAdoc represents the sum of these adducts. **The dGuo/dAdo cis pair and trans pair were not fully resolved, and the entries under dAdot and dGuot represent the sums of both cis and trans adducts in each case.

There are conflicting reports in the literature on the use of nuclease P1 in the detection of BcPhDE adducts by postlabeling. We had earlier failed to detect adducts from one BcPhDE isomer after digestion with nuclease P1 (12), and others had also reported low recoveries of adducts associated with use of this enzyme (26). Recently, Giles et al. (23) found that adduct yields could be either enhanced or reduced when nuclease P1 was used. These latter authors attributed this variation to experimental conditions, but this phenomenon is not yet clearly understood. However, under the conditions used in this study, nuclease P1 did not reduce adduct recovery. Additionally, for DNA modified at rather high levels by treatment with the dihydrodiol epoxide stereoisomers in vitro and for DNA modified at relatively low levels, from

Figure 6. Total DNA bound adducts in epidermal DNA exposed for various times to the four stereoisomeric dihydrodiol epoxides. The experimental findings are given as horizontal lines, and these values, after correction for the postlabeling efficiencies derived from Figure 3, are given as the full column heights. The tumors/mouse data were taken from the literature (6).

mouse skin in vivo, the conditions used here were shown to be consistent with maximal labeling of the adducts. DNA treated with the dihydrodiol epoxides in vitro was analyzed both by UV absorbance measurements after HPLC separation of nucleoside adducts and by scintillation counting after TLC separation of postlabeled nucleoside bisphosphates. No major differences in the relative efficiencies of detection of adducts from the same dihydrodiol epoxide isomer by the two methods were observed, although for two dibenz[a,j]anthracene adducts, others have reported a substantial difference (25). In the latter case, the authors also reported efficiencies of postlabeling of about 9%. This is also quite different from the 25-50% range found in the present studies (Figure 3). In general, the current study showed that the efficiencies for postlabeling adducts from different stereo-

Benzo[c]phenanthrene-DNA Adducts in Vivo

isomeric dihydrodiol epoxides were similar except in the case of adducts from the S,R,R,S isomer, for which adduct detection appeared less efficient than in the other cases. The most important aspect of the study of postlabeling efficiencies, however, is that all the adducts were shown to be detected with efficiencies that were comparable. Thus, it is reasonable to evaluate levels of DNA modification in mouse epidermal DNA using the postlabeling approach. On the basis of studies at three different doses, Levin et al. (6) have shown that the tumor initiating activities of the 4R,3S,2S,1R and 4S,3R,2S,1R BcPhDE isomers are more or less equal, whereas the 4R,3S,2R,1S isomer is inactive and the 4S,3R,2R,1S isomer displays an intermediate level of activity. The data for the 10 nmol dose, used in our binding experiments and reproduced in Figure 6, clearly parallel these overall findings, although the tumorigenicity of both the 4S,3R,2R,1S and the 4R,3S,2R,1S at this dose could not be shown to be statistically different from the control data. The remarkable feature of the experiments described here is that the two clearly similarly tumorigenic dihydrodiol epoxide isomers show such a large difference in the extents to which they have modified mouse epidermal DNA in vivo. Three different time points were examined, and this difference is apparent at each of these time points. This observation provides evidence that the adducts formed from the S,R,S,R isomer are intrinsically more tumorigenic than those from the other isomers. The adducts formed from the S,R,S,R isomer are almost entirely (∼80%) adducts with deoxyadenosine (15), so the above findings support previous suggestions, based initially on findings with the potent hydrocarbon carcinogen 7,12-dimethylbenz[a]anthracene (21), that deoxyadenosine adducts may, for some reason, be intrinsically more potent biologically than deoxyguanosine adducts. It is also of interest to relate the present findings to the mutagenic properties of these same stereoisomers. In our earlier studies of mutation induction in a shuttle vector replicating in human cells, point mutation frequency per adduct per plasmid was determined to be as follows: R,S,S,R, 0.4 × 10-4; S,R,S,R, 0.7 × 10-4; S,R,R,S, 0.3 × 10-4; and R,S,R,S, 1.0 × 10-4 (9). The most potent mutagen was the stereoisomer that is inactive as a tumorigen in mouse skin. In the light of the present work that shows that this particular stereoisomer does not bind substantially to the DNA of mouse epidermis, this apparent contradiction is resolved. Additionally, the greater biological effect per adduct seen in mouse skin for the S,R,S,R isomer is paralleled by a mutagenic activity per adduct that is also substantially greater for this particular stereoisomer than for the R,S,S,R isomer. In summary, the present findings, in conjunction with previous work on tumorigenicity, mutagenicity, and chemical reactions with DNA for the four configurationally isomeric BcPhDEs, present a reasonably coherent overall picture. Thus, all four isomers react to similar extents with DNA in vitro (10). However, after reaction with DNA in vitro, the mutagenic activities of the two stereoisomers, in which the epoxide oxygen and the benzylic hydroxyl group are cis, are substantially greater than those of the isomers in which these two groups are trans (9). In mouse epidermis in vivo, as shown here, one of the more mutagenic isomers does not substantially bind to DNA presumably because it is hydrolyzed before it reaches the DNA. The other more mutagenic isomer, S,R,S,R BcPhDE, binds to DNA of the epidermis at a

Chem. Res. Toxicol., Vol. 9, No. 3, 1996 591

much lower level than the R,S,S,R isomer yet produces a similar tumorigenic response, in concert with its greater mutagenicity. Previously, the high tumorigenic potency associated with hydrocarbons that are nonplanar, either because of a methyl substituent in a bay region or because of the presence of a fjord region, has been associated with the extensive reaction toward deoxyadenosine residues found with these compounds. The present findings indicate that a high biological activity per DNA adduct is associated with the S,R,S,R dihydrodiol epoxide structure, and it should be remembered that another difference between planar and nonplanar hydrocarbons is that an S,R,S,R dihydrodiol epoxide plays a much larger role in adduct formation from the parent hydrocarbons in the case of nonplanar hydrocarbons (17, 27).

Acknowledgment. We thank Dr. Jane M. Sayer for helpful discussions. Research sponsored in part by the National Cancer Institute, DHHS, under contract with ABL (R.A., K.A.C., A.D.). The contents of this publication do not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.

References (1) Harvey, R. G. (1991) Polycyclic Aromatic Hydrocarbons: Chemistry and Carcinogenicity, Cambridge University Press, Cambridge, England. (2) Jerina, D. M., Chadha, A., Cheh, A. M., Schurdak, M. E., Wood, A. W., and Sayer, J. M. (1990) Covalent bonding of bay-region diol epoxides to nucleic acids. Adv. Exp. Med. Biol. 283, 533553. (3) Dipple, A., Moschel, R. C., and Bigger, C. A. H. (1984) Polynuclear aromatic carcinogens. In Chemical Carcinogens (Searle, C. E., Ed.) pp 41-163, American Chemical Society Monograph 182, American Chemical Society, Washington, DC. (4) Levin, W., Wood, A. W., Chang, R. L., Ittah, Y., Croisy-Delcey, M., Yagi, H., Jerina, D. M., and Conney, A. H. (1980) Exceptionally high tumor-initiating activity of benzo[c]phenanthrene bay region diol epoxides on mouse skin. Cancer Res. 40, 3910-3914. (5) Jerina, D. M., Sayer, J. M., Yagi, H., Croisy-Delcey, M., Ittah, Y., Thakker, D. R., Wood, A. W., Chang, R. L., Levin, W., and Conney, A. H. (1982) Highly tumorigenic bay-region diol epoxides from the weak carcinogen benzo[c]phenanthrene. In Biological Reactive Intermediates IIA (Adv. Exp. Med. Biol.) (Snyder, R., Parke, D. V., Kocsis, J. J., Jollow, D. J., Gibson, C. G., and Witmer, C. M., Eds.) pp 501-523, Plenum Press, New York. (6) Levin, W., Chang, R. L., Wood, A., Thakker, D. R., Yagi, H., Jerina, D. M., and Conney, A. H. (1986) Tumorigenicity of optical isomers of the diastereomeric bay region 3,4-diol 1,2-epoxides of benzo[c]phenanthrene in murine tumor models. Cancer Res. 46, 2257-2261. (7) Wood, A. W., Chang, R. L., Levin, W., Thakker, D. R., Yagi, H., Sayer, J. M., Jerina, D. M., and Conney, A. H. (1984) Mutagenicity of the enantiomers of the diastereomeric bay-region benzo[c]phenanthrene 3,4-diol-1,2-epoxides in bacterial and mammalian cells. Cancer Res. 44, 2320-2324. (8) Seidman, M. M., Dixon, K., Razzaque, A., Zagursky, R. J., and Berman, M. L. (1985) A shuttle vector plasmid for studying carcinogen-induced point mutations in mammalian cells. Gene 38, 233-237. (9) Bigger, C. A. H., St. John, J., Yagi, H., Jerina, D. M., and Dipple, A. (1992) Mutagenic specificities of four stereoisomeric benzo[c]phenanthrene dihydrodiol epoxides. Proc. Natl. Acad. Sci. U.S.A. 89, 368-372. (10) Jerina, D. M., Sayer, J. M., Agarwal, S. K., Yagi, H., Levin, W., Wood, A. W., Conney, A. H., Pruess-Schwartz, D., Baird, W. M., Pigott, M. A., and Dipple, A. (1986) Reactivity and tumorigenicity of bay-region diol epoxides derived from polycyclic aromatic hydrocarbons. In Biological Reactive Intermediates III (Kocsis, J. J., Jollow, D. J., Witmer, C. M., Nelson, J. O., and Snyder, R., Eds.) pp 11-30, Plenum Press, New York. (11) Reddy, M. V., and Randerath, K. (1986) Nuclease P1-mediated enhancement of sensitivity of 32P-postlabeling test for structurally diverse DNA adducts. Carcinogenesis 7, 1543-1551.

592 Chem. Res. Toxicol., Vol. 9, No. 3, 1996 (12) Canella, K. A., Peltonen, K., Yagi, H., Jerina, D. M., and Dipple, A. (1992) Identification of individual benzo[c]phenanthrene dihydrodiol epoxide-DNA adducts by the 32P-postlabeling assay. Chem. Res. Toxicol. 5, 685-690. (13) Sayer, J. M., Yagi, H., Croisy-Delcey, M., and Jerina, D. M. (1981) Novel bay-region diol epoxides from benzo[c]phenanthrene. J. Am. Chem. Soc. 103, 4970-4972. (14) Yagi, H., Thakker, D. R., Ittah, Y., Croisy-Delcey, M., and Jerina, D. M. (1983) Synthesis and assignment of absolute configuration to the trans 3,4-dihydrodiols and 3,4-diol-1,2-epoxides of benzo[c]phenanthrene. Tetrahedron Lett. 24, 1349-1352. (15) Dipple, A., Pigott, M. A., Agarwal, S. K., Yagi, H., Sayer, J. M., and Jerina, D. M. (1987) Optically active benzo[c]phenanthrene diol epoxides bind extensively to adenine in DNA. Nature 327, 535-536. (16) Dipple, A., and Pigott, M. A. (1987) Resistance of 7,12-dimethylbenz[a]anthracene-deoxyadenosine adducts in DNA to hydrolysis by snake venom phosphodiesterase. Carcinogenesis 8, 491-493. (17) Sawicki, J. T., Moschel, R. C., and Dipple, A. (1983) Involvement of both syn- and anti-dihydrodiol-epoxides in the binding of 7,12dimethylbenz[a]anthracene to DNA in mouse embryo cell cultures. Cancer Res. 43, 3212-3218. (18) Pelling, J. C., and Slaga, T. J. (1982) Comparison of levels of benzo[a]pyrene diol epoxide diastereomers covalently bound in vivo to macromolecular components of the whole epidermis versus the basal cell layer. Carcinogenesis 3, 1135-1141. (19) Yuspa, S. H., and Harris, C. C. (1974) Altered differentiation of mouse epidermal cells treated with retinyl acetate in vitro. Exp. Cell Res. 86, 95-105. (20) Jeffrey, A. M., Weinstein, I. B., Jennette, K. W., Grzeskowiak, K., Nakanishi, K., Harvey, R. G., Autrup, H., and Harris, C. C. (1977) Structures of benzo[a]pyrene-nucleic acid adducts formed in human and bovine bronchial explants. Nature 269, 348-350.

Agarwal et al. (21) Bigger, C. A. H., Sawicki, J. T., Blake, D. M., Raymond, L. G., and Dipple, A. (1983) Products of binding of 7,12-dimethylbenz[a]anthracene to DNA in mouse skin. Cancer Res. 43, 5647-5651. (22) Pigott, M., and Dipple, A. (1988) 7,12-Dimethylbenz[a]anthraceneDNA adducts in mouse skin, dermis and epidermis. Cancer Lett. 40, 291-297. (23) Giles, A. S., Seidel, A., and Phillips, D. H. (1995) In vitro reaction with DNA of the fjord-region diol epoxides of benzo[g]chrysene and benzo[c]phenanthrene as studied by 32P-postlabeling. Chem. Res. Toxicol. 8, 591-599. (24) Butch, E. R., Lau, H. H. S., Shaw, K. L., Smolarek, T. A., Schmerold, I., Anderson, J. N., Baird, W. M., Yagi, H., and Jerina, D. M. (1992) High selectivity of polyclonal antibodies against DNA modified by diastereomeric benzo[c]phenanthrene 3,4-diol 1,2epoxides. Carcinogenesis 13, 895-899. (25) Baer-Dubowska, W., Nair, R. V., Cortez, C., Harvey, R. G., and DiGiovanni, J. (1995) Covalent DNA adducts formed in mouse epidermis from dibenz[a,j]anthracene: evidence for the formation of polar adducts. Chem. Res. Toxicol. 8, 292-301. (26) Carothers, A. M., Zhen, W., Mucha, J., Zhang, Y.-J., Santella, R. M., Grunberger, D., and Bohr, V. A. (1992) DNA strand-specific repair of (()-3R,4β-dihydroxy-1R,2R-epoxy-1,2,3,4-tetrahydrobenzo[c]phenanthrene adducts in the dihydrofolate reductase gene. Proc. Natl. Acad. Sci. U.S.A. 89, 11925-11929. (27) Pruess-Schwartz, D., Baird, W. M., Yagi, H., Jerina, D. M., Pigott, M. A., and Dipple, A. (1987) Stereochemical specificity in the metabolic activation of benzo[c]phenanthrene to metabolites that covalently bind to DNA in rodent embryo cell cultures. Cancer Res. 47, 4032-4037.

TX950148+