Involvement of cytochrome P450, glutathione S-transferase, and epoxide hydrolase in the metabolism of aflatoxin B1 and relevance to risk of human liver cancer.

In recent years there has been considerable interest in the effect of variations in activities of xenobiotic-metabolizing enzymes on cancer incidence. This interest has accelerated with the development of methods for analyzing genetic polymorphisms. However, progress in epidemiology has been slow and the contributions of polymorphisms to risks from individual chemicals and mixtures are often controversial. A series of studies is presented to show the complexities encountered with a single chemical, aflatoxin B1 (AFB1). AFB1 is oxidized by human cytochrome P450 enzymes to several products. Only one of these, the 8,9-exo-epoxide, appears to be mutagenic and the others are detoxication products. P450 3A4, which can both activate and detoxicate AFB1, is found in the liver and the small intestine. In the small intestine, the first contact after oral exposure, epoxidation would not lead to liver cancer. The (nonenzymatic) half-life of the epoxide has been determined to be approximately 1 sec at 23 degrees C and neutral pH. Although the half-life is short, AFB1-8,9-exo-epoxide does react with DNA and glutathione S-transferase. Levels of these conjugates have been measured and combined with the rate of hydrolysis in a kinetic model to predict constants for binding of the epoxide with DNA and glutathione S-transferase. A role for epoxide hydrolase in alteration of AFB1 hepatocarcinogenesis has been proposed, although experimental evidence is lacking. Some inhibition of microsome-generated genotoxicity was observed with rat epoxide hydrolase; further information on the extent of contribution of this enzyme to AFB1 metabolism is not yet available.


Introduction
The role of enzymatic transformation in variation of levels of individual enzymes the activation and detoxication of chemi-that can occur (2). With experimental cal carcinogens has been recognized for animal systems, there is considerable evihalf a century (1). Studies on these dence that alterations in levels of some of enzymes have shown their multiplicity in these enzymes can have dramatic effects in many cases, as well as the large extent of influencing the incidence of cancer from chemical carcinogens, both in the classical initiation and promotion phases (3)(4)(5)(6)(7). This background in experimental models has led to hypotheses that variations in the activities of individual enzymes involved in xenobiotic transformation influence cancer incidence in humans (7,8). In the late 1980s it became possible to assign roles in the activation of many chemical carcinogens to individual P450 enzymes on the basis of in vitro results (8,9) (Table 1). The same approaches have been used with other enzymes such as GST and N-acetyltransferase. In some cases, the dominance of a particular enzyme (especially cytochrome P450) in the metabolism of a drug has made it possible to make in vivo evaluations of the contributions to the human metabolism of a carcinogen, especially if low levels of the carcinogen can be administered to humans (7,8,1 1).
The above information has led to studies involving what is often referred to as molecular epidemiology, particularly in the effort to associate cancer risks with levels of enzymes or with genetic polymorphisms in the enzymes. The difficulty in the approach may be exemplified by the results of efforts to relate levels of P450 2D6 with tobacco-induced lung cancer, where equivocal results have been obtained in different laboratories over the course of a decade (12)(13)(14)(15). Part of the difficulty in this situation may be a result of the myriad of potential carcinogens found in tobacco smoke and the lack of P450 2D6 to dominate in the activation of any of these (16)(17)(18).
Aflatoxin B1 (AFBI) is generally considered to play a major role in human liver cancer in some parts of the world (19,20), and much is now known about its mechanism of genotoxicity, which appears to be the result of the formation of a single initial DNA adduct (at the guanyl N7 atom) (21,22). We have considered some of the complexities of the metabolism of AFB1 and the relevance to efforts to implicate individual enzyme levels as factors in risk.
P450s can also detoxicate AFBI ( Figure   1). P450 3A4 forms AFQi, the 3a-hydroxylation product, which does not appear to be a good substrate for epoxidation (37). P450 1A2 forms AFMI (by 9a hydroxylation), which also seems to be a detoxication product (31,37). In animal studies, the induction of P450 1A2 and production of AFMI have been reported to account for the lower levels of AFB1induced hepatocellular cancer after administration of polycyclic hydrocarbons (40,41). Which of the human P450s form aflatoxin PI is not known.  (42,43). a-Naphthoflavone inhibits all activities of P450 1A2 (31); it also inhibits AFB, 3a-hydroxylation (to form AFQ1) by P450 3A4 but stimulates 8,9-epoxidation (31,37). The influence on the kinetic profiles is postulated to reflect an allosteric mechanism (44). fractions, the order of (enzymatic) GST activity is mouse > rat > human (25). The relative extent of GSH-AFB1 conjugate formation by some human GSTs is shown in Table 3.
Preliminary studies have indicated that the UV spectrum of AFB1-8,9-dihydrodiol is distinct from that of the epoxide ( Figure  3A). The fluorescence spectra are even more distinct, with the dihydrodiol having more than two orders of magnitude more fluorescence than the epoxide. Kinetics of hydrolysis were measured in a stopped-flow apparatus in a pH 7.0 buffer, with 9% (CH3)2CO present (final concentration).
The t1/2 was approximately 1 sec when either UV or fluorescence traces were measured ( Figure 3B).
Further studies indicated that the observed hydrolysis rate constant was rather constant between pH 4 and pH 9 but was increased at <pH 4, with a slope of the loglo observed versus pH having a slope of unity.
Interaction of AFB1-8,9-exo-epoxide with Glutathione &Transferase 3-3 and DNA In earlier studies the relative rates of reaction of GSTs with AFB1-8,9-epoxides had been estimated by quantitation of GSH-AFB1 with fixed concentrations of GSTs and time points (Table 3) (25). To examine the aspects of these interactions, we measured the extent of GS-AFB1 formation after mixing varying concentrations of AFB1-8,9-exo-epoxide and GST 3-3 in the presence of GSH (Figure 4).
To estimate constants for the reactions, we set up the equations   Table 4, along with the measured k,.   system containing P450 3A4 (plus all exo-epoxide, in dry acetone, was mixed with varying accessory components needed for oxidaamounts of GST 3-3 and 10 mM GSH in 50 p1 of 50 mM tion) and a suboptimal amount of GST. potassium phosphate buffer (pH 7.4) at 23°C. The final However, in other experiments we have concentration of AFB1-8,9-exo-epoxide was 4 (.), been able to increase the observed rate of 12 (*), or 24 (U) mM. After 15 sec, 20 ml of 2.0 M AFBI-8,9-exo-epoxide hydrolysis (from aqueous CH3CO2H was added and the protein was pre-0.64 to 0.78/sec) in the presence of 19 pM cipitated by centrifugation at 3xl03xgfor 10 min. rat EH. This result needs to be further Aliquots of the supernatant were analyzed for AFBevaluated.

8,9-dihydrodiol and GS-AFB, by HPLC as described
We also used another system in which a elsewhere (25,31). very low concentration of P450 3A4 (and accessory components) was used to activate Table 4. Reactions of AFB, 8,9- (31) in the presence of an NADPH-generating system, S. typhimurium TAl 535 containing plasmid pSK1001, and the indicated concentrations of purified rat (-) or human (A,) EH (the latter two samples were prepared from human liver samples, HL96 and HL1 05, of two different individuals). The response to heat-inactivated rat EH is also shown (o). The umu response was monitored by 0-galactosidase response and is expressed as described by Shimada et al. (32).

Small intestine
Liver Summary Further studies are needed to evaluate the role of EH in the metabolism of AFB1-8, 9-epoxide. The report of McGlynn et al. (49) is surprising in that the EH allelic variant was linked with higher levels of AFBI-albumin adduct, even though the major AFB1 protein adduct is thought to be derived from AFBI-8,9-dihydrodiol (51). The dihydrodiol results from the enzymatic or nonenzymatic hydrolysis of the epoxide. The possibility of direct reaction of protein with the epoxide cannot be ruled out at this time; however, the effect of the allelic variation on the catalytic activity of EH is not well established. The report of lower activity is the result of lower levels of expression in a transient system, not intrinsic catalytic activity (52).
The complexity of the enzyme systems involved in the metabolism of AFB1 points out the difficulties in doing molecular epidemiology studies, even when a single chemical carcinogen has been identified. The roles of at least two P450s in the activation process have been considered. There is suggestive evidence that human GSTs in the alpha, mu, and theta families may all have roles in the detoxication of the epoxide (25,(53)(54)(55) to consider the stereochemistry of the epoxide, which has been shown to be critical in genotoxicity. We anticipate that more careful analysis of the kinetics of the reactions under consideration here will provide insight into competition of detoxication enzymes with DNA ( Figure 2). However, even with this information there are considerable problems in the knowledge base underlying efforts in molecular epidemiology ( Figure  6). The P450s both activate and detoxicate AFB1, and the effect of inducing individual P450s is not easy to predict. Moreover, P450 3A4 is expressed in small intestine, the site of absorption of orally ingested AFB1. The extent of detoxication there is unknown. Further, even activation of AFB1 and DNA alkylation in the small intestine may be considered to be a detoxication process since the cells are sloughed rapidly and cancers of the small intestine are very rare. Other aspects regarding enzymes of activation and detoxication mentioned in Figure 6 have been discussed above. Other aspects not discussed here, but which may be involved, include whether the N7-guanyl adduct or its ringopened form is more mutagenic, the role of DNA repair, aflatoxin intake, the intraindividual variability of levels of the enzymes under consideration during the course of tumor initiation and development, and hepatitis B virus status.