Metabolism and Activation of 7,8-Dihydrobenzo[a]pyrene during Prostaglandin Biosynthesis INTERMEDIACY OF A BAY-REGION EPOXIDE*

A Tween 20-solubilized preparation of prostaglandin endoperoxide synthase has been shown to metabolize 7,8-dihydrobenzo[a]pyrene (H2BP) to a form highly mu- tagenic to Salmonella typhimurium strain TA98. The arachidonic acid-dependent metabolism of HzBP by microsomal and purified prostaglandin endoperoxide synthase has been studied and the products identified. A spectral investigation of the metabolism indicated the bay-region double bond as the primary site of me- tabolism. Radiolabeled HzBP was synthesized and incubated with the enzyme preparations and the metab- olites were separated by reverse phase high perform-ance liquid chromatography and quantitated by liquid scintillation counting. Radioactive products were characterized by co-chromatography with chemically synthesized standards, W-visible spectra, and mass spec- trometry of acetate derivatives. The major polar products were determined to be trans- and cis-9,lO-dihy-droxy-7,8,9,10-tetrahydrobenzo[a]pyrene and 7,8,9,10-tetrahydrobenzo[a]pyrene-9-one in a ratio of 1:1.2:0.4. The inclusion of 5 m~ 3,3,3-trichloropropene-1,2-oxide, an epoxide hydrolase inhibitor, produced the same products but in a ratio of 1:2.3:1.2. Incubations with purified prostaglandin

glandin biosynthesis to 9,10-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene. (1). The widespread exposure of human populations to these agents is of great concern in that a number of PAH are known to be carcinogenic. In all cases studied, the parent hydrocarbons are inert, but are metabolized by the host organism to their active carcinogenic forms (2)(3)(4). As exposure to these agents is unavoidable, the detailed study of their metabolic activation is of great importance.
BP was one of the fiist identified carcinogens ( 5 ) , and as a result, is the most studied and best understood PAH. BP may be metabolized to a variety of oxygenated products, including phenols, quinones, dihydrodiols, and their conjugates (6)(7)(8). In addition, these primary metabolites may be oxygenated to produce secondary metabolites. The work of Borgen et al. (9) suggested that this secondary oxygenation might be the key to the activation of BP. Extensive work from several laboratories has established that this is indeed correct (10)(11)(12)(13)(14)(15). The pathway of BP activation is as shown in Equation 1 (10)(11)(12)(13)(14)(15).
This pathway has been elucidated using the microsomal cytochrome P-450-containing mixed-function oxidases for introduction of oxygen, and microsomal epoxide hydrolase for hydration of the arene oxide (12,(16)(17)(18)(19). These oxygenases are not, however, the only enzymes capable of catalyzing oxygenations of BP. Marnett et al. (20) demonstrated that BP may also be metabolized by hydroperoxide-dependent co-oxygenation, and that this occurs during prostaglandin biosynthesis. (Equation 2). Top of next page.
The products of BP eo-oxygenation were shown to be the 1,6-3,6-, and 6J2-quinones of BP (21). The formation of a reactive intermediate was demonstrated by metabolism-dependent covalent binding of BP to exogenous nucleic acid (22), but mutagenicity toward Salmonella typhirnuriurn strains TA98 and TAlOO was not detected (23). Further stud-PAH' are a class of ubiquitous environmental contaminants ies, however, showed that BP-7,8-diol could be activated by co-oxygenation to a form highly mutagenic to TA98 and TAlOO (23). The identification of the stable products of me-* This project was supported by Research Grant BC 244 from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.   (24,25) and the nucleic acid adducts formed (26) strongly suggested that the initial product of BP-7,8-diol cooxygenation was 7r,8t-dihydroxy-9t,lOt-epoxy-7,8,9,lO-tetrahydrobenzo[a]pyrene.

Dihydrobenzo[a]pyrene Metabolism
Since no one has ever isolated a bay-region diolepoxide from an in vitro incubation mixture, all of the evidence for their formation is indirect and is based upon the identification of stable derivatives formed by the attack of nucleophiles a t the benzylic epoxide carbon. An alternative approach to detecting unstable epoxide intermediates would be to determine the effect of epoxide-specific reagents on the profile of stable epoxide-derived products. An example of an epoxide-specific reagent is epoxide hydrolase which catalyzes the hydrolysis of arene oxides and other epoxides to trans-dihydrodiols (50,51). Unfortunately, microsomal epoxide hydrolase exhibits little or no catalytic activity towards dihydrodiolepoxides derived from PAH (30, 52). As a result, the relative distribution of diolepoxide solvolysis products is unchanged by the presence of epoxide hydrolase. Bay-region epoxides which lack hydroxyl groups in the tetrahydrobenzo ring do appear to be substrates for epoxide hydrolase so that the level of epoxide hydrolase activity should modulate the relative distribution of solvolysis products (29,46). HzBP has been shown by others to be metabolized by mixed-function oxidases to derivatives mutagenic to S. typhimurium TA98 (46) and I-LBP-epoxide is mutagenic (30). In addition, H4BP-epoxide undergoes spontaneous solvolysis to a mixture of products in which the cis-H,BP-diol is the major product and microsome-mediated hydrolysis to the trans-H4BP-diol (  generously provided by Professor Bruce Ames, University of California, Berkeley. 15-HPETE was prepared by the method of Funk et al. (33). All other chemicals and solvents were reagent grade.
Analytical Procedures-UV-visible spectra and spectrophotometric assays were performed on a Cary 210 spectrophotometer. NMR spectra were obtained using a Nicolet NTCFT-1180 spectrometer at. 300 MHz. Mass spectra were recorded on a Hewlett-Packard HP 5985 mass spectrometer at the Regional Mass Spectrometry Facility, located at Michigan State University, E. Lansing, MI. Liquid scintillation counting was carried out in a dioxane-based mixture on an Isocap 300. Quench correction was made using the sample channels ratio technique. Protein was measured by the method of Lowry et al. (34).
Synthetic Procedures-All synthetic procedures were carried out under subdued light, and reactions were run under argon. H2BP was prepared by a modification of published procedures (35,36). Trans-H,BP-diol was prepared as previously described (38), while cis-H,BPdiol was synthesized using a modified Prevost reaction (39). H4BP-9one was prepared from the H4BP-diols by an acid catalyzed dehydration (35). All compounds exhibited appropriate UV-visible spectra, and the acetate derivatives of the diols and the ketone were analyzed by proton NMR and direct probe mass spectrometry. The diols were additionally characterized by formation of an acetonide derivative (42) and by differential reactivity with potassium triacetylosmate (41,42). Details of the syntheses and characterizations may be found in the miniprint.
Enzyme Preprations-RSV were collected at a local slaughterhouse and stored at -80 "C. Microsomal and Tween 20-solubilized enzymes were prepared as previously described (43). Purified PES from RSV, exhibiting a single band on sodium dodecyl sulfate-polyacrylamide gel electrophoresis slab gels, was prepared by a procedure developed in this IaboratoryP Rat liver microsomes were prepared by the method of Oesch et al. (44).
Epoxide Hydrolase Assays-The epoxide hydrolase activities of various preparations were determined using the radiometric assay of Oesch et al. (44). [7-'H]Styrene oxide was added to a concentration of 2 m~ and incubated for 15 min at 37 "C. All values are corrected for control incubations consisting of styrene oxide and buffer without enzyme.
Spectrophotometric Assay of H&P Metabolism-Determinations were carried out in a stirred cuvette thermostatted at 37 "C. Mixtures containing 0.1 M KPO, (pH 7.8), enzyme, and H2BP were monitored at 370 nm for 3 min to establish the initial absorbance at 370 nm. 20:4 or 15-HPETE was added by syringe and the change in absorbance with time was recorded. The initial rate of the oxidation and the total consumption of HzBP were calculated using E = 3.79 X 10, M" cm -I . Incubations with solubilized enzyme or with purified PES also contained 0.1% Tween 20 (w/v).
Mutagenicity Experiments-Assays with S. typhimurium strain TA98 were carried out essentially as described previously (23,45). Overnight cultures in nutrient broth were used, however, without further concentration, and hemoglobin was not added to the incubations. HzBP

Dihydrobenzo[a]pyrene Metabolism
and 5 mM TCPO where indicated. These mixtures were preincubated at 37 "C for 3 min, whereupon 20:4 was added to a concentration of 100 p~. After a 10-min incubation at 37 "C with vigorous shaking, each mixture was extracted thrice with 2 ml of ethyl/acetate/acetone (4:l) each time. Extracts were taken to dryness using a rotary evaporator, and the residues were taken up in methanol for HPLC analysis. Preparative incubations for product characterization were identical with the above, except that HzBP was 20 p~. Extracts from 6-8 tubes were pooled and stripped at reduced pressure and the residue dissolved in methanol for HPLC separation.
Incubations with purified PES were carried out in 0.1 M KPO,, pH 7.8, containing 0.1% Tween 20 (w/v) and 1 p~ hematin. The incubation procedure was identical with the above, except that the metabolism was initiated by 50 p~ 15-HPETE rather than by 204.
HPLC Conditions" HPLC was performed using methanol/ water mixtures on C-18 columns. Analyses were performed using a Waters RCM-100 module containing a Radial Pak B cartridge. Metabolites and HzBP were separated by a linear gradient from 75:25 methanol/water to 955 at 2% rnin", then from 95:5 to straight methanol at 1% rnin". The flow rate was 1.5 ml rnin". Elution was monitored at 344 nm using a Varian Varichrom detector. Metabolites isolated by this gradient were purified further by isochratic elution using 75:25 methanol/water at 1.0 ml min-'. Acetate derivatives were purified by elution with 90:lO methanol/water at 1.0 ml rnin".

Metabolic
Activation of HzBP by Co-oxygenation-HzBP has no intrinsic mutagenic activity but it is activated to a highly mutagenic species by Tween 20-solubilized RSV preparations. The activation is dependent on RSV protein (Table   I), and this dependence is linear up to 400 pg of protein d-'.
Maximal activation requires the addition of either 20:4 or 15-HPETE, but a clear concentration dependence is not routinely obtained. This may be due to the extreme sensitivity of the system to lipid hydroperoxides. In the presence of 10 ~L M HzBP, as little as 0.5 p~ 15-HPETE produces a %fold increase in the reversion rate (data not shown). Even lipids endogenous to the enzyme preparation itself may become oxidized and thus serve to activate HzBP by co-oxygenation. The activation may be linked to prostaglandin biosynthesis in that the PES inhibitor indomethacin inhibits HZBP activation by up to 95% when 20:4 is added (Table I). The dependence of reversion rate on HZBP concentration is shown in Fig. 1. H2BP is activated to a potent mutagenic species, while the isomeric 9,10-dihydrobenzo[a]pyrene is not. This is in agreement with the results of Wood et al. (46), using a mixed-function oxidasedependent activating system for the two BP derivatives (30). The concentration dependence of HzBP activation is shifted to lower concentrations relative to the analogous curve for BP-7,8-diol activation (23). This is probably due both to the greater extent of metabolism of HzBP than of BP-7,8-diol (see below), and to the higher inherent mutagenic activity of the H,BP epoxide than of the diol epoxides (30).
Spectrophotometric Assay of H2BP Metabolism-Based on the established precedent for PAH activation, and on the   mutagenicity results with the isomeric dihydrobenzo[a]pyrenes, the bay-region double bond of HzBP appeared to be the most likely site for metabolism and activation. Spectrophotometric determinations demonstrated that this was indeed the major site of metabolism. HzBP contains an extended pyrene chromophore with a major absorbance peak at 370 nm in the mixed aqueous-detergent media used. Any saturation of the 9,lO double bond yields a pyrene chromophore, shifting the long wavelength maximum from 370 to 349 nm. Any metabolism altering the HzBP chromophore will cause a decrease in the absorbance at 370 nm, and if the pyrene chromophore remains, there will be a corresponding increase in the absorbance at 349 nm. The decrease in the 370 nm absorbance is shown in Fig. 2; scanning the mixture after metabolism had ceased showed the expected increase in absorbance at 349 nm (data not shown). The time course of HzBP metabolism was virtually identical with that of prostaglandin biosynthesis as measured by oxygen uptake. Similar

Dihydrobenzo[a]pyrene
results have been obtained for the co-oxygenation of BP-7,8diol (27). HzBP appears to be a better substrate for co-oxygenation than is BP-7,8-diol. Both the initial rate and the total metabolism of H2BP are approximately 2-fold higher than the values for BP-7,8-diol (data not shown). It has also been shown using this assay that the addition of 100 ~L M butylated hydroxytoluene to the cuvette at any point before or after the addition of 20:4 or 15-HPETE would immediately and totally abolish the consumption of HzBP. This further establishes the sensitivity of co-oxygenation to antioxidant inhibition. Epoxide Hydrolase Assays-RSV microsomes and solubilized preparations contain an active epoxide hydrolase. Using the radiometric assay of Oesch et al. (44) with labeled styrene oxide, as substrate. This enzyme appears similar to the microsomal epoxide hydrolase of liver (Table 11). The RSV activity is destroyed by boiling the preparation, and marked inhibition by TCPO is also seen. These results are also obtained using TABLE I1 Comparison of epoxide hydrolase activity in RLM a n d RSV were purified by isocratic HPLC as described under Experimental Procedures." a, trans-HIBP-diol. h, 9-acetoxy-HZBP.
rat liver microsomes. The specific and total activities are much higher in RLM than in RSV microsomes.
HPLC Profiles of H2BP Metabolism-Incubations were performed using [14C]H2BP with RSV microsomes in the presence and absence of the epoxide hydrolase inhibitor TCPO. Ethyl acetate-extractable material was subjected to reverse phase HPLC analysis. The column effluent was mon-

Dihydrobenzo[a]pyrene
Metabolism 11373 itored for absorbance at 345 nm, 30-s fractions were collected, and radioactivity quantitated by liquid scintillation counting. Three peaks of radioactivity, significantly more polar than HzBP, were found. These peaks were associated with the absorbances noted in Fig. 3, and co-chromatographed with authentic standards of the cis-and trans-H,BP-diols and H4BP-9-one. Total radioactivity in the three peaks was between 12 and 18% of the total eluted from the column. Unknown labeled species accounted for 2-776 of the total radioactivity; however, this label appeared dispersed throughout the chromatogram as no discrete peaks were found. The apparent enzymatic production of BP from HzBP comprised another 2.5% of the total radioactivity collected. The remainder was collected as H2BP. Overall recovery from the column was routinely 85-90% of radioactivity injected. Incubations with boiled microsomes showed less than 1% of total radioactivity eluting in the region of the three metabolite peaks, less than 0.7% as BP, and about 2% unknown species. Cooxygenation of HzBP by a purified PES preparation yielded the same three stable products. The purified preparation converted 40-45% of the total radioactivity to the diastereomeric diols and the ketone. Overall conversion of HzBP to more polar products was about 55%. Identification of Polar Products-Samples of each of the three metabolites were collected by HPLC and their UVvisible spectra were taken in methanol. All three exhibited spectra similar to the one shown in Fig. 4a. This spectrum is indicative of an H4BP chromophore and shows that all three metabolites still contain the intact pyrene chromophore but have lost the 9,lO-unsaturation. Each of the three samples was then acetylated, repurified by HPLC, and the spectra of the acetates were taken. The acetylation products of the two metabolites which co-chromatographed with the H4BP-diols exhibited UV-visible spectra indistinguishable from those of the metabolites themselves. The third metabolite when acetylated gave the UV-visible spectrum shown in Fig. 4b. This spectrum, shifted to longer wavelengths relative to the metabolite spectrum, is typical of a 7,8-H2BP chromophore. This is identical with the spectrum of the authentic 9-acetoxy-H2BP synthesized by the acetylation of H4BP-9-one.
The three acetylated metabolites were analyzed by direct probe mass spectrometry. Attempts at gas chromatographymass spectrometry were unsuccessful due to extensive decomposition of the compounds during gas chromatography. Identical molecular ions and fragmentation patterns were seen for the acetate derivatives from the diol peaks. The spectrum of the cis-H4BP diol diacetate is shown in Fig. 5a. The molecular ion is seen at m/e = 372.2. Major fragments are seen at m/e = 312.2 (loss of CH,COOH), 270.1 (base peak, loss of CH3COOH + CHsCO), 252.2 (loss of 2CH&OOH), and 239.1 (loss of 2CH3COOH + CH). The spectrum of the ketone derivative, seen in Fig. 5b, exhibits a molecular ion at m/e = 312.2 and shows a similar fragmentation pattern featuring peaks at m/e = 270.2 (base peak), 239.1, and 226.2. These mass spectra are consistent with H4BP-diol diacetates for the larger mass derivatives, and 9-acetoxy-7,8-HzBP for the smaller mass derivative. As expected, most fragmentation occurs leaving the pyrene nucleus intact, progressing through successive loss of acetate and then contraction of the benzoring (53). Product Distributions-Based on co-chromatography of the three stable metabolites with chemically synthesized standards, UV-visible spectra of the metabolites and their acetates, and mass spectra of the acetates, the metabolites of HlBP formed by co-oxygenation are the trans-H4BP-diol, cis-H4BP-diol, and H4BP-9-one. As varying amounts of HzBP were metabolized by different enzyme preparations, the ab-solute amounts of the three products cannot be compared directly. Rather, the ratios of the products have been compared from a number of incubations. The results using RSV microsomes (Table 111) are the ratios obtained from the product distributions in five experiments, a total of 12 separate incubations. The relative amounts of the three products were calculated for each separate incubation, based on the trans-H4BP-diol as 1.0. The ratios in Table I11 are the mean & S.D. of the 12 ratios. The purified PES ratios were calculated in the same manner from quadruplicate incubations in a single experiment.
A striking change in the ratio of transto cis-H4BP-diol may be seen when epoxide hydrolase is inhibited or is removed by purification. This increase in the cis-H4BP-diol and the accompanying increase in the amount of H.,BP-g-one are consistent with the metabolic scheme shown in Fig. 6. In this scheme, the initial product of HzBP co-oxygenation is the H4BP-epoxide. This unstable intermediate may be enzymatically hydrated by epoxide hydrolase to form the trans-H4BPdiol, or it may spontaneously hydrolyze and rearrange to form a mixture of three products. The changes in product ratios shown in Table I11 support the generation of the H,BPepoxide which is further transformed by both the enzymatic and the spontaneous pathways. When the enzymatic pathway is inhibited, the stable product profiie reflects the increased contribution of the spontaneous pathway. The changing stable product profiie resulting from epoxide hydrolase inhibition or removal, and the functional group specificity and stereospecificity of epoxide hydrolase (31,32) indicate that the unstable initial product of H2BP co-oxygenation is the H4BP-epoxide.

DISCUSSION
Co-oxygenation of BP-7,8-diol catalyzed by PES preparations has been shown to yield stable metabolites (24,25) and nucleoside adducts (26) consistent with the intermediacy of the 7r,8t-dihydroxy-9t,10t-epoxy-7,8,9,10-tetrahydrobenzo[a] pyrene. The production of a highly mutagenic product and the observed saturation of the 9,lO-double bond of HzBP both suggest that a related epoxide intermediate is formed by the co-oxygenation of HzBP. This intermediacy has been established in the present study.
We have shown that RSV microsomal preparations contain an active epoxide hydrolase for which the H4BP-epoxide is a substrate. This enzyme will specifically hydrolyze cyclic 1,2epoxides to their respective trans-diols (31,32). In the absence of this enzyme, the H4BP-epoxide will spontaneously hydrolyze and rearrange, forming primarily the cis-H4BP-diol and H4BP-9-one and a small amount of trans-H4BP-diol (28). If the H4BP-epoxide is formed in these incubations, then any inhibition of epoxide hydrolase should shift the product distribution toward the cis-H4BP-diol and the H4BP-9-one at the expense of the trans-H4BP-diol. This is indeed the case.
The ratios of cis-to trans-H4BP-diol formed in the presence and absence of TCPO are particularly important in reaching this conclusion. Equal amounts of the two diols are formed by the microsomal system, whereas 2.3 times more cis-than trans-H4BP-diol results when epoxide hydrolase is inhibited. Incubations with purified PES, which is devoid of epoxide hydrolase activity, produced the two diols in a ratio of 12.8, favoring the cis-H4BP-diol even more strongly than in the TCPO-inhibited microsomal incubations. These results compare favorably with the results of Waterfall and Sims (29), who identified the products of metabolism of HnBP and H4BPepoxide by the mixed-function oxidases in rat liver preparations. HzBP metabolism produced both diols, with the trans predominating. However, when the H4BP-epoxide itself was added, the ratio of the diols was 1:l. The results of their work and those of the current study underscore the key role of epoxide hydrolase in the interpretation of these stable product profiles. The common intermediate, the H4BP-epoxide is converted to the trans-H4BP-diol by epoxide hydrolase. If, however, the catalytic capacity of the hydrolase is exceeded, either by inactivation of the enzyme or by excess substrate challenge, then the stable product profile will reflect the growing contribution of the spontaneous hydrolysis and rearrangement to the total product profile. The higher specific activity of epoxide hydrolase in the rat liver allows very little H4BP epoxide to spontaneously hydrolyze, while the spontaneous pathway is far more pronounced in the RSV incubations of HnBP.
The relative amounts of H4BP-9-one formed under the different incubation conditions are less easily interpreted. Though some increase in the relative amount of H4BP-9-one is seen in the TCPO-inhibited microsomal incubations and in the purified PES incubations, the magnitude of the increase is less than might be expected, based on the reported extent of spontaneous rearrangement (28). The ketone has been noted to be unstable at basic pH, making quantitation difficult (28). Thus, the ketone may be decomposing during the course of the incubations. An alternative explanation may come from the results of Waterfall and Sims (29). In their studies with both HzBP and H4BP-epoxide in rat liver preparations they reported only the two diols as products. These results may indicate that the spontaneous hydrolysis and rearrangement of the H4BP epoxide in purely aqueous media and in the mixed aqueous-protein-lipid media may not produce identical product distributions.
In addition to H4BP-epoxide, the co-oxygenation product of H2BP, a nonoxygenated product is also formed in this system. Approximately 10% of the total metabolism of HzBP by RSV preparations results in the formation of BP. This co-oxidation product results from the dehydrogenation of H2BP. The dehydrogenation of dihydroarenes has been shown in various chemical systems to be a facile process (48) leading to the establishment of a fully aromatic system. The production of BP in this system is dependent on the intact RSV preparation and 20:4, and is inhibited by antioxidants. These similarities to the co-oxygenation reaction indicate that both BP and the H4BP epoxide are formed from HzBP by interaction with an oxidizing species generated during prostaglandin biosynthesis.
As has been noted, co-oxygenation of BP does not involve epoxidation (21) nor does it activate BP as a mutagen (23). Co-oxygenation, however, is capable of epoxidizing the isolated double bonds of BP-7,8-diol and H2BP to form their bay-region epoxides, the ultimate mutagenic and carcinogenic forms of these molecules. Additionally, work by Eling4 has shown that 3,4-dihydroxy-3,4-dihydrobenzo[a]anthracene and l,Z-dihydroxy-l,Z-dihydrochrysene are both co-oxygenated to bay-region diol epoxides by RSV preparations, indicating that co-oxygenation can be a general pathway for PAH activation.
The cytochrome P-450 dependent mixed-function oxygenases appear absolutely required for the initial oxygenation step in PAH activation, but the ultimate activation step may be catalyzed by either mixed-function oxidases or PES. Microsomal preparations from skin and lung, target tissues for BP carcinogenesis, have been found to catalyze epoxidation of BP-7,8-diol by both pathways in vitro (47). Further studies are underway to establish if both pathways function in vivo as well.