Investigation of the role of the 2',3'-epoxidation pathway in the bioactivation and genotoxicity of dietary allylbenzene analogs.

The genotoxic potential of naturally occurring allylbenzene analogs, including safrole, eugenol, estragole, and others, has been examined in many studies over the past 30 years. It has been established that these compounds are subject to biotransformation in the liver, which can lead to the formation of reactive electrophilic intermediates. The major route of bioactivation is via hydroxylation of the 1' carbon atom of the allylic side chain. We have synthesized 2',3'- (allylic) epoxide derivatives of allylbenzene, estragole eugenol and safrole, and have used them to characterize the genotoxic potential of epoxidation at the allylic double bond for allylbenzene and its naturally occurring analogs. In order to assert that this pathway has the potential for genotoxicity, it is necessary to demonstrate (1) that epoxide metabolites of these compounds are capable of forming covalent adducts with DNA bases; and (2) that these epoxide metabolites are actually formed in vivo. We have demonstrated that allylic epoxides derived from allylbenzene and estragole are capable of forming covalent adducts with all four deoxyribonucleotides in vitro and, in the case of deoxyguanosine, form at least four different adducts. We also deduce, from evidence obtained using the isolated perfused rat liver, that formation of potentially genotoxic 2',3' epoxide metabolites occurs readily in vivo, but that these metabolites are rapidly further metabolized to less toxic dihydrodiol or glutathione conjugates. We conclude that 2',3' epoxide metabolites of allylbenzene analogs are formed in vivo and that these epoxides are sufficiently reactive to facilely form covalent bonds with DNA bases. Epoxide formation at the allylic double bond represents, therefore, a potentially genotoxic bioactivation pathway for allylbenzene analogs. However, comparison of the relative kinetics of epoxide metabolism and epoxide formation suggests that a wide margin of protection from DNA covalent adduct formation exists in the rat liver, thus preventing genotoxicity resulting from this pathway to any significant degree. In this regard, we have also observed that the general rate of epoxide hydrolysis is much greater in human liver than in rat liver. We therefore suggest that while the epoxidation pathway poses a potential genotoxic threat to humans, no actual genotoxicity occurs as a result of this metabolic pathway.

Co-purification of microsomal epoxide hydrolase with the warfarin-sensitive vitamin K1 oxide reductase of the vitamin K cycle.

Vitamin K1 oxide reductase activity has been partially purified from rat liver microsomes. A three-step procedure produced a preparation in which warfarin-sensitive vitamin K1 oxide reductase activity was 118-fold enriched over the activity in intact rat liver microsomes. A major component of the multi-protein mixture was identified as a 50 kDa protein that strongly cross-reacts with antiserum prepared against homogeneous rat liver microsomal epoxide hydrolase. The reductase preparation also had a high level or epoxide hydrolase activity against two xenobiotic epoxide substrates. The K(m) values for hydrolysis by the reductase preparation were similar to those for homogeneous microsomal epoxide hydrolase itself, and the specific hydrolase activities of the reductase preparation were 25-35% of the specific activities measured for the homogeneous hydrolase preparation. Antibodies prepared against homogeneous microsomal epoxide hydrolase inhibited up to 80% of reductase activity of the reductase preparation. Homogeneous microsomal epoxide hydrolase had no vitamin K1 oxide reductase activity. This evidence suggests that microsomal epoxide hydrolase, or a protein that is very similar to it, is a major functional component of a multi-protein complex that is responsible for vitamin K1 oxide reduction in rat liver microsomes.

Isolation of liver nuclei that retain functional trans-membrane transport.

We have developed a method for the rapid isolation of hepatocyte nuclei, which employs gentle homogenization and centrifugation conditions, and involves minimal processing time. The purified nuclei were morphologically unaltered when observed by light and electron microscopy. No significant contamination from cytoplasm or mitochondria was detected when assessed by marker enzymes. Membrane transport function, measured as ATP-dependent calcium uptake, was intact. This isolation method was devised to be applicable to studies that involve measurement of uptake and active transport of a variety of substances by the cell nucleus.

Compartmentation of glutathione: implications for the study of toxicity and disease.

The fact that glutathione (GSH) plays many roles in biological protective mechanisms and critical physiological functions has been recognized for decades. Conjugates, disulfides, and other glutathione-derived products also have been studied as biomarkers of the chemical natures or specific identities of key metabolites of toxic agents and such studies have been crucial in the delineation of the nature of the interactions of proximal toxicants with target biomolecules. Despite the extensive evidence implicating the depletion and/or oxidation of glutathione in a wide variety of human and experimental toxicities, critical examination of such studies frequently reveals that injury is not simply related to glutathione status. GSH is compartmentalized at several levels and this compartmentation appears to exert considerable influence on the relationships between glutathione depletion or oxidation and the onset of injury. Although compartmentation is usually viewed from the perspective of different intracellular pools, the significance of extracellular glutathione in functionally important pools is gaining recognition. As the factors affecting the interactions of intracellular pools with extracellular pools are delineated, studies in humans can be designed and interpreted with greater precision and utility.

Covalent binding to DNA in vitro of 2',3'-oxides derived from allylbenzene analogs.

Epoxidation at the allylic side chain is a major metabolic pathway for allylbenzene and its naturally occurring analogs safrole, estragole, and eugenol. We demonstrate herein that the epoxide metabolites of allylbenzene, estragole, and safrole can form covalent adducts with DNA in vitro, binding primarily to guanine, but also to the other three DNA bases. Epoxide hydrolases can prevent the binding of allylbenzene 2',3'-oxide to DNA in vitro. Four distinct adducts were detected by analytical TLC after the reaction of 2'-deoxyguanosine with allylbenzene 2',3'-oxide. One unstable adduct was formed rapidly, but gradually disappeared, whereas the other three adducts were formed more slowly but persisted. The major persistent adduct, which was isolated by preparative chromatography, was examined by MS and NMR. The structure of this adduct is 3'-N1-deoxyguanosyl-(2'-hydroxypropylbenzene). In addition, a generally applicable paradigm for the identification of deoxyguanosine or guanosine adducts by 13C and 1H NMR spectroscopy is presented.

Development of an in situ toxicity assay system using recombinant baculoviruses.

A new method for experimentally analyzing the role of enzymes involved in metabolizing mutagenic, carcinogenic, or cytotoxic chemicals is described. Spodoptera fugiperda (SF-21) cells infected with recombinant baculoviruses are used for high level expression of one or more cloned enzymes. The ability of these enzymes to prevent or enhance the toxicity of drugs and xenobiotics is then measured in situ. Initial parameters for the system were developed and optimized using baculoviruses engineered for expression of the mouse soluble epoxide hydrolase (msEH, EC 3.3.2.3) or the rat cytochrome P4501A1. SF-21 cells expressing msEH were resistant to trans-stilbene oxide toxicity as well as several other toxic epoxides including: cis-stilbene oxide, 1,2,7,8-diepoxyoctane, allylbenzene oxide, and estragole oxide. The msEH markedly reduced DNA and protein adduct formation in SF-21 cells exposed to [3H]allylbenzene oxide or [3H]estragole oxide. On the other hand, 9,10-epoxyoctadecanoic acid and methyl 9,10-epoxyoctadecanoate were toxic only to cells expressing sEH, suggesting that the corresponding fatty acid diols were cytotoxic. This was confirmed by showing that chemically synthesized diols of these fatty acid epoxides were toxic to control SF-21 cells at the same concentration as were the epoxides to cells expressing sEH. A recombinant baculovirus containing a chimeric cDNA formed between the rat P4501A1 and the yeast NADPH-P450 reductase was also constructed and expressed in this system. A model compound, naphthalene, was toxic to SF-21 infected with the rat P4501A1/reductase chimeric co-infecting SF-21 cells with either a human or a rat microsomal EH virus along with P4501A1/reductase virus. These results demonstrate the usefulness of this new system for experimentally analyzing the role of enzymes hypothesized to metabolize endogenous and exogenous chemicals of human health concern.

Metabolism of allylbenzene 2',3'-oxide and estragole 2',3'-oxide in the isolated perfused rat liver.

The metabolism of allylbenzene 2',3'-oxide, estragole 2',3'-oxide, allylbenzene and estragole was studied in the isolated perfused rat liver. Formation of dihydrodiol and glutathione conjugate metabolites was detected for both epoxides and the presence of dihydrodiol metabolites after perfusion of allylbenzene or estragole indicated the formation of allylic epoxide intermediates in the intact liver. A comparison of elimination kinetics for parent compounds and epoxides indicated that epoxides were relatively rapidly detoxified and probably do not accumulate on formation in vivo. Acute toxicity of epoxides, measured as the release of alanine aminotransferase activity into the perfusate, or genetic toxicity, determined as covalent binding of radiolabeled epoxide to DNA, were not observed. It was concluded that both epoxide hydrolases and glutathione S-transferases can effectively detoxify the allylic epoxides derived from either allylbenzene or estragole and effectively prevent cellular or genetic toxicity of these reactive intermediates. Epoxide hydrolases appear to play the major role in the detoxication of these epoxides in vivo.

32P-postlabeling analysis of adducts formed between DNA and safrole 2',3'-epoxide: absence of adduct formation in vivo.

We have used the 32P-postlabeling technique to examine the binding of safrole 2',3'-oxide to DNA. At least 8 covalent adducts are formed when calf thymus DNA is incubated with this oxygenated metabolite of safrole in vitro. However, no corresponding adducts are formed with liver DNA when whole animals are exposed to safrole 2',3'-oxide, or safrole itself. Although safrole 2',3'-oxide is readily formed in vivo, and is sufficiently reactive to covalently bind to DNA, it is probably not a factor in the in vivo genotoxicity of safrole. We also demonstrate that adducts with similar mobility to the major safrole 2',3'-oxide-DNA adduct are formed in vitro between safrole 2',3'-oxide and deoxyguanosine, and also between its chemical analogs allylbenzene 2',3'-oxide or estragole 2',3'-oxide and DNA.

Detoxication of the 2',3'-epoxide metabolites of allylbenzene and estragole. Conjugation with glutathione.

The enzymatic detoxication in vitro of the 2',3'-epoxide derivatives of allylbenzene and estragole was examined, and the relative rates of enzymatic glutathione conjugation and epoxide hydrolysis were compared with those for styrene 1',2'-oxide. HPLC was used to determine the amounts of dihydrodiol and glutathione conjugate metabolites formed by cell extracts from several sources. Although some differences among species were observed, in general, the rates of epoxide inactivation by both pathways are similar. We conclude that one explanation for the apparent lack of genotoxicity of these allylic epoxides in vivo may be their rapid metabolic inactivation by both glutathione S-transferases and epoxide hydrolases, which occur to approximately equal degrees in vitro.

Depletion of a discrete nuclear glutathione pool by oxidative stress, but not by buthionine sulfoximine. Correlation with enhanced alkylating agent cytotoxicity to human melanoma cells in vitro.

The existence of a distinct pool of glutathione in the nucleus of cultured human melanoma cells was demonstrated. Melanoma cell nuclei contained 13-35 pmol of glutathione/10(6) nuclei, or approximately 0.4-1.3% of the total cellular glutathione. This nuclear glutathione pool resisted depletion by buthionine sulfoximine, an agent that inhibits glutathione synthesis, but was rapidly and reversibly depleted by subtoxic concentrations of Adriamycin plus carmustine, two agents that promote oxidation of glutathione without permitting its regeneration through enzymatic reduction of glutathione disulfide. The ability of Adriamycin plus carmustine to deplete this small but significant pool of glutathione in the cell nucleus may explain why these agents potentiate the cytotoxic effects of the DNA-alkylating agent melphalan to a much higher degree than does buthionine sulfoximine at concentrations that are equipotent in depleting cytosolic glutathione.

Detection of human lung cytochromes P450 that are immunochemically related to cytochrome P450IIE1 and cytochrome P450IIIA.

We have used monoclonal antibodies that were prepared against and specifically recognize human hepatic cytochromes P450 as probes for solid phase radioimmunoassay and Western immunoblotting to directly demonstrate the presence in human lung microsomes of cytochromes P450 immunochemically related to human liver cytochromes P450IIE1 (CYP2E1) and P450IIIA (CYP3A). The detected levels of these cytochromes are much lower than levels in human liver microsomes, but similar to the levels seen in microsomes from untreated baboon lung. Proteins immunochemically related to two other constitutive hepatic cytochromes P450, cytochrome P450IIC8 (CYP2C8) and cytochrome P450IIC9 (CYP2C9), were not detectable in lung microsomes.

Hydrolysis of the 2',3'-allylic epoxides of allylbenzene, estragole, eugenol, and safrole by both microsomal and cytosolic epoxide hydrolases.

2',3'-Allylic epoxide derivatives of allylbenzene and its analogs estragole, eugenol, and safrole were synthesized, and their enzymatic conversion to dihydrodiols by cytosolic and microsomal epoxide hydrolases was examined. All four epoxides were good substrates for both epoxide hydrolases, with Michaelis constants in the low micromolar range. Two putatively selective inhibitors of cytosolic and microsomal epoxide hydrolases, trichloropropylene oxide and nordihydroguaiaretic acid, were used to inhibit the hydrolysis of these allylic epoxides. Minimal selectivity toward either hydrolase was seen with either inhibitor, suggesting that the "selectivity" of these inhibitors is highly substrate-dependent. The susceptibilities of these epoxides to rapid hydrolysis by both epoxide hydrolases may explain their low genotoxic potencies in vivo.

An investigation of the formation of cytotoxic, protein-reactive and stable metabolites from carbamazepine in vitro.

The formation of chemically reactive metabolites from carbamazepine (CBZ) in the presence of mouse and human liver microsomes has been investigated using cytotoxicity and irreversible binding of radiolabelled compound as quantitative end-points. For comparison, the formation of the stable CBZ-10,11-epoxide (CBZ-10,11-E) has been measured. The formation of the cytotoxic, protein-reactive and stable metabolites of CBZ was increased by induction of the cytochrome P450 enzymes by phenobarbitone and reduced by co-incubation in vitro with ketoconazole (10-250 microM), suggesting that the formation of these metabolites is cytochrome P450 dependent. All human livers tested (N = 6) bioactivated CBZ to a protein-reactive metabolite, the mean covalent binding increasing from 0.08 +/- 0.01% (without NADPH) to 0.27 +/- 0.09% (with NADPH; P less than or equal to 0.05). The formation of the chemically reactive metabolites was reduced by a subphysiological concentration of reduced glutathione (GSH) (500 microM), while ascorbic acid (100 microM) had no effect. Neither compound affected the formation of CBZ-10,11-E. Microsomal epoxide hydrolase (mEH), but not cytosolic epoxide hydrolase, caused a concentration-dependent inhibition of cytotoxicity reaching a maximum of 60% at 100 U of mEH. Covalent binding was also reduced by 60% by 100 U mEH. The separated T- and B-lymphocytes showed no difference in sensitivity when incubated with CBZ and mouse microsomes. The study demonstrates that the balance between activation of CBZ by the cytochrome P450 enzymes to a chemically reactive arene oxide metabolite and its detoxification by mEH and GSH may contribute to individual susceptibility to CBZ idiosyncratic toxicity.

Direct measurement of melphalan conjugation with glutathione: studies with human melanoma cells and mammalian liver.

The rate of formation of the major glutathione conjugate of the antitumor alkylating agent melphalan can be directly measured by high pressure liquid chromatography. Rates of melphalan-glutathione conjugate formation were determined in the presence and in the absence of human melanoma cell homogenates, or cell fractions from various tissue sources, and the relative contributions of enzyme-catalyzed and nonenzymatic conjugate formation to the overall rates of conjugation were determined. Significant rates of conjugation were observed in the absence of any enzyme-containing cell fractions. These rates were not increased by the addition of melanoma cell homogenates, animal liver microsomes or human liver cytosol or microsomes, even though these preparations all enhanced the rate of conjugation of 1-chloro-2,4-dinitrobenzene. Animal liver cytosol contains enzymes that provided a significant contribution to the overall rate of melphalan conjugate formation. We conclude that although liver cytosol contains enzymes that significantly enhance the rate of glutathione conjugation with melphalan, in the case of the tumor cells studied, cellular glutathione S-transferase-catalyzed activity appears to be, at best, a very minor determinant of the overall rate of melphalan-glutathione conjugate formation.

Sensitization of human melanoma cells to melphalan cytotoxicity by adriamycin and carmustine.

Exposure of cultured human melanoma cells from three different cell lines to Adriamycin and carmustine at non-cytotoxic (micromolar) concentrations results in a rapid, reversible depletion of cellular glutathione; maximal depletion is achieved within 1 h, and glutathione levels recover within 2-3 h. Glutathione depletion is accompanied by an enhancement of the cytotoxic effects of the alkylating agent melphalan, which ranges from 15- to 55-fold. These results suggest that the combination of Adriamycin and carmustine may provide a rational drug combination for the rapid depletion of glutathione from malignant melanoma, thereby sensitizing these tumor cells to alkylating agent cytotoxicity.

Immunochemical analysis of a cytochrome P-450IA1 homologue in human lung microsomes.

The monoclonal antibody MAb 1-7-1, which specifically binds to cytochromes P-450IA1 and P-450IA2 in 3-methylcholanthrene-induced rat liver microsomes, was used to identify a cytochrome P-450IA1 homologue in human lung microsomes. Although MAb 1-7-1 had similar affinity constants for human and rat microsomes, the amount bound to human lung microsomes was severalfold lower than that bound to microsomes from untreated rat or rabbit lung and much lower than the amount bound to 3-methylcholanthrene-induced rat lung or liver microsomes. The amount bound to untreated baboon lung microsomes was similar to that bound to human lung microsomes. Three cytochrome P-450IA1-catalyzed activities, 7-ethoxyresorufin O-deethylase, 7-ethoxycoumarin, O-deethylase, and aryl hydrocarbon hydroxylase, were measurable in human lung microsomes, but the cytochrome P-450IA2-dependent activity acetanilide 4-hydroxylase was not. MAb 1-7-1 inhibited, and its binding correlated strongly with, 7-ethoxyresorufin O-deethylase activity (r = 0.92, p less than 0.01) in human lung microsomes. 7-Ethoxyresorufin O-deethylase activities in human lung were similar to those measured in untreated baboon lung but considerably lower than those present in untreated rabbit lung, untreated or 3-methylcholanthrene-induced rat lung and liver, or human liver. We conclude that MAb 1-7-1 recognizes a cytochrome P-450IA1 homologue in human lung and that no cytochrome P-450IA2 homologue is detected. Cytochrome P-450IA1 is expressed in human lung at relatively low levels, similar to those observed in untreated primate (baboon) lung. The majority of the 19 human lung samples examined do not exhibit a permanent polycyclic aromatic hydrocarbon-induced state with respect to this isozyme.

Inhibition of catalase and epoxide hydrolase by the renal cystogen 2-amino-4,5-diphenylthiazole and its metabolites.

Subchronic feeding of 2-amino-4,5-diphenylthiazole (DPT) to rats results in the development of renal cysts and has been used as a model system to study polycystic kidney disease. Because previous studies revealed changes in renal enzymes following DPT administration, a possible direct effect of DPT and its phenolic metabolites on catalase and a related enzyme, epoxide hydrolase, was examined. Experiments with three in vitro systems (suspensions of rabbit renal tubules, rat kidney homogenates, and commercially obtained bovine liver catalase) revealed direct inhibition of catalase activity by the diphenolic metabolite (diOH- DPT: 2-amino-4,5di(4'-hydroxyphenyl)-thiazole), the known renal cystogen nordihydroquaiaretic acid (NDGA) 2-amino-4(4'-hydroxyphenyl),5-phenyl-thiazole (4OH-DPT), and the known catalase inhibitor 3-amino-1,2,4-triazole; DPT did not inhibit catalase activity. Following oral administration to rats of the DPT congeners, 4OH-DPT caused the greatest decrease in both renal catalase and cytosolic epoxide hydrolase activities and the shortest time to onset of cystic lesions. In vitro, mouse liver cytosolic epoxide hydrolase activity was substantially inhibited by 4OH-DPT and dioH-DPT, and NDGA, but not by 2-amino-4-phenyl,5-(4'-hydroxyphenyl)-thiazole (5OH-DPT) or DPT itself. Microsomal epoxide hydrolase (mEH) activity was inhibited by 4OH-DPT, unaffected by DPT or dioH-DPT, and stimulated 2-fold by 5OH-DPT. Finally, mEH activity was substantially higher in samples of normal human kidney than in samples of kidney derived from a patient with autosomal recessive polycystic kidney disease; no differences were observed in cEH activity in these samples. Although the role of altered catalase and epoxide hydrolase activities in cystogenesis is unknown, DPT-induced cyst formation is associated with loss of these enzyme activities in kidney tissue. To our knowledge, this is the first report of an in vivo diminution of cytosolic epoxide hydrolase activity by xenobiotics.

Production of monospecific antiserum to a cytosolic epoxide hydrolase from human liver.

A method for the purification to apparent homogeneity of cytosolic trans-stilbene oxide hydrolase from human liver is presented. The method employed ion exchange and gel filtration chromatography. From 50 g of human liver, 4.9 mg of homogenous enzyme protein was obtained. Although the enzyme had lost much of its catalytic activity during purification, it was nevertheless suitable for the preparation of antibodies to the enzyme. Only one immunogenic species was present in the antigen preparation, but some antibodies that were cross-reactive to sites on catalase were present in the antiserum. These catalase-specific antibodies were removed by immunoaffinity chromatography, and an IgG fraction that is monospecific to the cytosolic epoxide hydrolase was obtained. The usefulness of antibodies to this enzyme in immunoblotting experiments, following either sodium dodecyl sulfate-polyacrylamide gel electrophoresis or isoelectric focussing, as well as in enzyme-linked immunosorbent assays, is demonstrated.