While most of this chapter concerns itself with the carcinogenic action of specific chemicals, it is relatively unusual that an individual is exposed to a single carcinogenic agent. Despite this fact, relatively few detailed studies on mixtures of carcinogenic chemicals have been carried out experimentally. The most common environmental mixtures are those seen in tobacco smoke and other combustion products, including engine exhaust and air pollution (Mauderly, 1993). Inter- actions between chemicals in mixtures may be additive, multiplicative, or inhibitory (Mumtaz et al., 1993). In the examples given above, however, the exact chemical nature of components in tobacco smoke or air pollution is not always known, nor are their amounts determined. Thus, one may be forced to deal with a mixture as if it were a single entity or, if the constituents are known, to treat the effects of the mixture in some empirical way usually related to the most po- tent component of the mixture.
Studies on the carcinogenic action of defined mixtures of chemicals are usually done with a knowledge of the carcinogenic effect of the chemicals involved. Warshawsky et al. (1993) demonstrated that extremely low levels of benzo[a]pyrene, which yielded no skin tumors on re- peated application, resulted in a significant yield of neoplasms when added in the presence of five noncarcinogenic polycyclic aromatic hydrocarbons. In an earlier study, administration of two noncarcinogenic aminoazo dyes in the diet of the rat for a year resulted in the appearance of a variety of neoplasms (Neish et al., 1967). More recently, administration of three to five N- nitrosamines resulted in either an additive or synergistic carcinogenic effect of the combinations of the compounds given at low dose rates (Berger et al., 1987; Lijinsky et al., 1983). In contrast, administration of a mixture of 40 chemical carcinogens to rats for 2 years at 1/50 of the dose normally used to induce neoplasms in 50% of the animals and representing a wide variety of target sites resulted in significant tumor incidences only in the thyroid and liver (Takayama et al., 1989). In a more recent study, ingestion of a mixture of 20 pesticides given at “acceptable daily intake levels” was found to exert no effect on carcinogenesis in rat liver (Ito et al., 1994). While these are only a few examples, the toxicological study of complex mixtures, not only in the area of carcinogenesis, is a critical field in human health, as evidenced by disease resulting from tobacco smoke, engine exhaust, and other components of air pollution (Mauderly, 1993). One of the most important chemical mixtures associated with human neoplasia is that of diet.
Chemical Carcinogenesis by Diet
There is substantial evidence in the human to argue that many dietary components—including excess caloric intake (Osler, 1987; Lutz and Schlatter, 1992), excessive alcohol intake (IARC,
1987), and a variety of chemical contaminants of the diet including aflatoxin B1 (Figure 3.2) (Gorchev and Jelinek, 1985; Lutz and Schlatter, 1992)—are carcinogenic (Chapter 11). Some general and specific studies have supported these views (Jensen and Madsen, 1988; Habs and Schmähl, 1980; Miller et al., 1994), whereas others have been more controversial (Willett and MacMahon, 1984; Pariza, 1984). Evidence for dietary factors associated with cancer incidence
in animals is more substantial and serves to support much of the evidence relating environmental factors to increased cancer incidence in the human (Kritchevsky, 1988; Rogers et al., 1993).
Although a relative lack of “antioxidant micronutrients”—such as carotenoids, selenium, and the vitamins A, C, and E—has been implicated as a factor in the incidence of neoplastic development (Dorgan and Schatzkin, 1991), more studies are needed before the effectiveness of these agents in cancer prevention can be established unequivocally (Chapters 8 and 11). In con- trast, experimental evidence that the lack of available sources of methyl groups can actually in- duce liver cancer in rats is well documented (Mikol et al., 1983; Ghoshal and Farber, 1984). This observation may be closely related to the earlier studies by Farber (1963) on the induction of liver cancer in rats by the administration of ethionine, which, indirectly, may cause a lack of available methyl groups in this tissue.
METABOLISM OF CHEMICAL CARCINOGENS IN RELATION TO CARCINOGENESIS
Although the discovery that polycyclic hydrocarbons and other chemical compounds could pro- duce cancer gave hope that the complete understanding of the nature of neoplasia might be at hand, more than 60 years have elapsed since those findings appeared and we still seem to be a long way from such an understanding. The excretory metabolites of polycyclic hydrocarbons were found to be hydroxylated derivatives, which usually had little or no carcinogenic activity. Similarly, hydroxylation of the rings of the aromatic amine carcinogens, such as 2-acetylamino- fluorene (AAF) and 4-dimethylaminoazobenzene, usually resulted in a complete loss of activity. The enzymatic production of these more polar metabolites facilitated the further metabolism and excretion of the parent compounds.
The different classes of chemical carcinogens have no single common structural feature (Figures 3.1, 3.3, and 3.4). Thus, the complexity of the variety of chemicals capable of inducing cancer posed a striking dilemma in attempts to understand the mechanisms of action of these agents. The beginning of our present-day understanding of the solution to this dilemma was re- ported in 1947 by Elizabeth and James Miller, who first demonstrated that, during the process of hepatocarcinogenesis, azo dyes became covalently bound to proteins of the liver but not to pro- teins of the resulting neoplasms. Sorof and associates (1958) studied the proteins of liver that bind the dyes and demonstrated that azo dye–induced hepatomas did not contain the major dye- binding protein species, as evidenced by the technique of electrophoresis. Later studies with more highly differentiated hepatocellular carcinomas have shown the presence of the dye-bind- ing proteins, but even in such lesions there was little or no binding of the dye to proteins of the neoplasm even when the carcinogens inducing those neoplasms were fed to animals bearing the transplanted cancers (Sorof et al., 1966). The initial studies of the Millers led them to suggest that the binding of carcinogens to proteins might lead to the loss or deletion of critical proteins for growth control (Chapter 11). The reader should recall that, at the time this hypothesis was proposed (1947), the molecular concept of the gene was in its infancy if understood at all.
As an extension of this work, Elizabeth Miller (1951) demonstrated the covalent binding of benzo(a)pyrene or some of its metabolites to proteins in the skin of mice treated with the hydrocarbon. Later Abell and Heidelberger (1962) showed the same phenomenon with another carcinogenic polycyclic hydrocarbon, 3-methylcholanthrene. These findings strongly suggested that a critical step in the induction of cancer by chemicals was the covalent interaction of some form of the chemical with proteins and possibly other macromolecules as well. Since the parent compound in all cases studied was incapable of covalent binding directly with macromolecules, the logical conclusion was that the interaction of the chemical with the macromolecule was the result of the metabolic action of the cell. Although a number of studies in the 1950s (cf. Weis- burger and Weisburger, 1958) demonstrated that ring-hydroxylation was a major pathway in the metabolism of AAF, in 1960 the Millers and Cramer (Miller et al., 1960) reported that this com- pound was also metabolized by hydroxylation of the nitrogen of the acetylamino group. They isolated N-hydroxy-AAF from the urine of AAF-treated rats, and in subsequent investigations found this compound to be more carcinogenic that its parent, AAF. Furthermore, N-hydroxy- AAF also induced neoplasms that the parent compound was unable to induce, such as subcuta- neous sarcomas at the site of injection. In animals (such as the guinea pig) that convert little of the AAF to its N-hydroxy derivative in vivo, cancer of the liver was not produced by feeding the parent compound. These findings strongly supported the suggestion that, at least in the case of AAF, the parent compound was not the direct carcinogen but rather that certain metabolic deriv- atives were the active components in the induction of neoplasia. These studies paved the way to further investigations of the activation of carcinogens by means of their metabolism by cellular enzymes (cf. J. A. Miller, 1970).
Figure 3.5 depicts a number of metabolic reactions involved in the “activation” of chemi- cals to forms that are directly involved with the induction of cancer. One may divide such metabolic functions into two general classes (cf. Goldstein and Faletto, 1993). Those involved in phase I metabolism (Figure 3.5) occur within the intracellular membrane system known as the endoplasmic reticulum. These reactions involve metabolism by cytochrome P-450 enzymes and their reductase. Generally these metabolic reactions convert the substrate to a more polar
Figure 3.5 Structures of representative chemical carcinogens and their metabolic derivatives, the proxi- mate and ultimate carcinogenic forms resulting from the action of phase I metabolism of procarcinogens.
Figure 3.6 Structures of representative chemical carcinogens and their metabolic derivatives resulting from the action of phase II metabolism of procarcinogens.
compound through the insertion of molecular oxygen. Phase II metabolic reactions (Figure 3.6) involve mostly hydrolysis and conjugation and occur primarily, although not exclusively, in the extramembranous, cytosolic environment of the cell. This listing of xenobiotic metabolic reac- tions of the cell is not complete, a detailed consideration of these and other such reactions being beyond the scope of this text. The interested reader is referred to several pertinent references (Porter and Coon, 1991; Guengerich, 1992) for more complete coverage of this topic.
As noted in Figure 3.5, the N-hydroxylation of AAF can be followed by the esterification of the N-hydroxyl group, yielding a highly reactive compound capable of nonenzymatic reaction with nucleophilic sites on proteins and nucleic acids as well as comparable small molecules. The demonstration that the metabolism of AAF led to a highly reactive chemical prompted the Mill- ers to propose that chemical carcinogens are—or are converted by their metabolism into—elec- trophilic reactants (chemicals with electron-deficient sites) that exert their biological effects by covalent interaction with cellular macromolecules, the most critical target probably being DNA (cf. Miller, 1978). Furthermore, utilizing the metabolism of AAF as a model, the Millers pro- posed that chemical carcinogens requiring metabolism for their carcinogenic effect be termed procarcinogens, whereas their highly reactive metabolites, such as the N-hydroxy AAF esters,
were termed ultimate carcinogens. Metabolites intermediate between the procarcinogens and ul- timate carcinogens, where such existed, were termed proximate carcinogens. The ultimate form of the carcinogen, that is, the form that actually interacts with cellular constituents and probably causes the neoplastic transformation, is the final product shown in most of the pathways seen in Figure 3.5. In some instances, however, the structure of the ultimate form of certain carcinogenic chemicals is still not clear. In other cases there may be more than one ultimate carcinogenic metabolite.
After the demonstration by the Millers of the critical significance of electrophilic metabo- lites in chemical carcinogenesis, a number of such ultimate forms—especially those of aromatic amines such as benzidene, naphthylamine and 4-aminobiphenyl—were described. However, the carcinogenic polycyclic hydrocarbons still posed a problem. As early as 1950, Boyland had pro- posed the formation of epoxide intermediates in the metabolism of these chemicals. However, it was not until 1970 that Jerina and associates detected the formation of such an intermediate in a biological system. Other investigations showed that epoxides of polycyclic hydrocarbons could react with nucleic acids and proteins in the absence of any metabolizing system. Surprisingly, K-region epoxides of a number of carcinogenic polycyclic hydrocarbons were weaker carcino- gens than the parent hydrocarbons. Following this finding, scientific attention shifted to other reactive metabolites of these molecules. In 1974 Sims and associates proposed that a diol ep- oxide of benzo(a)pyrene was the ultimate form of this carcinogen. Subsequent studies by a num- ber of investigators (Yang et al., 1976; see reviews: Conney, 1982; Harvey, 1981; Lowe and Silverman, 1984) have demonstrated that the structure of this ultimate form is (+)anti- benzo(a)pyrene-7,8-dihydrodiol-9,10-epoxide. A number of the metabolic reactions that benzo(a)pyrene may undergo in vivo are seen in Figures 3.5 and 3.6. As mentioned earlier, the
4,5-(K-region) epoxide of benzo(a)pyrene is less carcinogenic than the parent compound and markedly less carcinogenic than the diol epoxide. The reader is referred to the cited reviews for further discussion of the implications of such metabolic pathways.
One of the interesting ramifications of these findings is the importance of oxidation of the carbons of the “bay region” of potentially carcinogenic polycyclic hydrocarbons. Figure 3.2 shows the bay regions of benz(a)anthracene and benzo(a)pyrene. Analogous bay regions may easily be identified in the other structures seen in Figure 3.1. Jerina, Conney, and associates (for example, Levin et al., 1978), as well as others (cf. Conney, 1982), have proposed that epoxida- tion of a dihydro, angular benzo ring that forms part of a bay region of a polycyclic hydrocarbon forms the most likely ultimate carcinogenic form of the hydrocarbon. The bay region is the ster- ically hindered region formed by the angular benzo ring. Although the bay-region concept has not been tested with all known carcinogenic polycyclic hydrocarbons, it appears to be generally applicable. In addition, Cavalieri and Rogan (1992) have proposed that radical cations of poly- cyclic aromatic hydrocarbons formed by oxidation of the parent compound via the cytochrome P-450 pathway are also important intermediates in the formation of ultimate carcinogenic me- tabolites of these chemicals.
Although conjugation reactions usually inactivate chemical carcinogens and cause their rapid urinary excretion because of their water solubility, an interesting exception to this has re- cently been shown. Both haloalkanes and haloalkenes react with glutathione as catalyzed by glu- tathione S-transferase. As an example, the glutathione-dependent bioactivation of ethylene dibromide is seen in Figure 3.6. The proximate carcinogen, glutathione S-ethylbromide, sponta- neously forms an episulfonium ion as the ultimate carcinogenic form. This highly reactive chemical alkylates DNA at the N7 position of guanine (Koga et al., 1986). Cysteine S-conjugates of several haloalkenes as well as the glutathione conjugates are nephrotoxic as well as mutagenic (cf. Monks et al., 1990). While such halogenated aliphatics may induce neoplasia in several or-gans, the kidney is the predominant target organ; but the actual mechanism of the carcinogenic effect is not yet clear despite the observations noted above (Monks et al., 1990).
In addition to the electrophilic intermediates previously discussed that make up many if not all of the structures of the ultimate forms of chemical carcinogens, substantial evidence has also indicated that free radical derivatives of chemical carcinogens may be produced both meta- bolically and nonenzymatically during their metabolism (cf. Nagata et al., 1982). Free radicals carry no charge but possess a single unpaired electron, which makes the radical extremely reac- tive. That such forms may be important in the induction of the neoplastic transformation by chemicals comes from two lines of evidence. Several different molecules that inhibit the forma- tion of free radicals, many of which are termed antioxidants, are capable of inhibiting the carci- nogenic action of many chemical carcinogens (Ito and Hirose, 1987; Simic, 1988). Although there is no question that free radical intermediates are sometimes formed during the metabolism of chemical carcinogens (Guengerich, 1991), the evidence suggests that relatively specific meta- bolic reactions of a number of chemical carcinogens may proceed through free radical interme- diates. Marnett (1981) has described the co-oxygenation of polyunsaturated fatty acids, especially arachidonic acid, with polycyclic aromatic hydrocarbons, leading to the formation of the diol epoxide (Figure 3.5). Such co-oxygenation occurs in the formation of prostaglandins, a series of hormones important in the normal homeostasis of the organism, from the polyunsatu- rated fatty acid arachidonic acid. The reaction involves the formation of a hydroperoxide by the incorporation of molecular oxygen (two atoms) at a double-bond position with the simultaneous formation of a ring peroxide. The peroxidase activity of the prostaglandin-synthesizing enzyme (synthetase) catalyzes the transfer of the free radical to another molecule with the formation of oxygen, as shown in Figure 3.7 to be the case with AAF. In the case of the 7,8-diol of benzo(a)pyrene (Figure 3.7), the peroxidase catalyzes the transfer of the free radical oxygen from the hydroperoxide to the hydrocarbon, forming an epoxide, which in this case is the ultimate form of the carcinogen. N-acetoxy AAF is also one of the ultimate forms of that carcinogen. This pathway of metabolic activation of carcinogens is not ubiquitous in all tissues but is more important in some than in others. Wise et al. (1984) demonstrated that the meta- bolic activation of 2-naphthylamine via the prostaglandin synthase reaction was quite marked in dog bladder but virtually nonexistent in the liver of this animal. This finding suggests that the carcinogenic action of 2-naphthylamine on bladder transitional epithelium involves prima- rily the co-oxygenation of the arylamine and polyunsaturated fatty acids in the prostaglandin synthetase reaction. Other chemical carcinogens, including nitrosamines (Bartsch et al., 1989) and nitro compounds (Conaway et al., 1991), may exhibit ultimate forms that are free radicals in nature.
Endogenous Free Radicals and Chemical Carcinogenesis
In addition to the formation of free radicals affecting the metabolism of exogenous procarcino- gens, endogenous free radicals are also formed both as by-products of carcinogen metabolism and during normal metabolic reactions. Molecular oxygen (O2) contains unpaired electrons and acts as the final pathway of electrons produced during metabolism. In such metabolic reactions oxygen is reduced by single electrons by processes depicted in Figure 3.8. The most reactive species that may be considered free radicals are the superoxide anion and the hydroxyl radical. Hydrogen peroxide is formed from the superoxide radical but itself is not a free radical. On the other hand, in the presence of trace amounts of metal ions, especially those of iron and copper, the hydroxyl free radical is formed from hydrogen peroxide (Bast et al., 1991). In addition to the reduction of oxygen in mitochondria, which is perhaps the largest source of these radicals, a process termed redox cycling, involving the one-electron reduction of a xenobiotic followed by
Figure 3.7 The metabolic activation of benzo(a)pyrene 7,8 diol and N-hydroxy 2-acetylaminofluorene during the peroxidation of arachidonic acid.
interaction with molecular oxygen and resulting in superoxide production and regeneration of the parent xenobiotic, also gives rise to reactive oxygen species (ROS) by a scheme noted in Figure 3.9. Free radicals of oxygen reduction, especially the hydroxyl radical, may abstract a hydrogen atom from lipids, leading to their peroxidation. Peroxidation of lipids having conju- gated double bonds may lead ultimately to the formation of a variety of aldehydes, which are also capable of reacting with macromolecules (Bast and Goris, 1989). Two toxic aldehydes formed by this process are malonaldehyde and 4-hydroxy-2-nonanal. These aldehydes are mu- tagenic (Chung et al., 1993) and at least malonaldehyde has been reported to be carcinogenic for mouse skin (Shamberger et al., 1974). Epoxyaldehydes of hydroxynonanal can react with deox- yribonucleosides to form etheno adducts as seen in Figure 3.10 (Chung et al., 1996). In addition, reactive oxygen species may be involved in the alteration of signal transduction pathways (Chapter 7), and as noted later in this chapter in discussing the effects of ionizing radiation on living cells (Toyokuni, 1999).
Figure 3.8 The univalent reduction of oxygen. (Adapted from Bast et al., 1991, with permission of authors and publisher.)
All of these studies taken together demonstrate that the majority of chemical carcinogens must be metabolized within the cell before they exert their carcinogenic effects. In this respect car- cinogenesis by some chemicals becomes a “lethal synthesis” analogous to that of the earlier studies by Peters (1957), who coined the term with reference to fluoroacetate. Furthermore, this finding explains how a substance that is not carcinogenic for one species may be carcinogenic
Figure 3.9 General scheme for redox cycling of xenobiotics (R) with generation of oxygen radicals. (Adapted from Ross, 1989, with permission of author and publisher.)
Figure 3.10 Formation of some etheno adducts from reactions of deoxyribonucleosides with epoxyalde- hydes. (Adapted from Chung et al., 1996, with permission of authors and publisher.)
for another, the result depending on the metabolic capacities present within the species itself. This becomes extremely important for carcinogen testing in whole animals (see below).
Not all chemical carcinogens require intracellular metabolism to become ultimate carcino- gens. Examples are the direct alkylating agents β-propiolacetone, nitrogen mustard, ethylene- imine, and bis(chloromethyl)ether (Figure 3.3), the latter having been shown to be carcinogenic for humans (Chapter 9). These chemicals have been termed direct-acting or alkylating carcino- gens because of the highly reactive nature of their native structure, which itself is the ultimate carcinogenic form of that molecule.