Carcinogenesis, Vol. 22, No. 12, 1903-1930,
December 2001
© 2001 Oxford University Press
HISTORICAL REVIEW |
Synergistic mechanisms in carcinogenesis by polycyclic aromatic hydrocarbons and by tobacco smoke: a bio-historical perspective with updates
Department of Molecular and Cell Biology and Virus Laboratory, Life Sciences Addition, University of California, Berkeley, CA 94720-3200, USA
Email: hrubin{at}uclink4.berkeley.edu
Abstract
B[a]P (benzo[a]pyrene) has been used as a prototype carcinogenic PAH since its isolation from coal tar in the 1930's. One of its diol epoxides, BPDE-2, is considered its ultimate carcinogen on the basis of its binding to DNA, mutagenicity and extreme pulmonary carcinogenicity in newborn mice. However, BPDE-1 has a similar binding to DNA and mutagenicity but it is not carcinogenic. In addition, BPDE-2 is a weak carcinogen relative to B[a]P when repeatedly applied to mouse skin, the conventional assay site. Its carcinogenicity is increased when applied once as an initiator followed repeatedly by a promoter. This indicates a major role for promotion in carcinogenesis by PAHs. Promotion itself is a 2-stage process, the second of which is selective propagation of the initiated cells. Persistent hyperplasia underlies selection by promoters. The non-carcinogenicity of BPDE-1 has yet to be resolved. PAHs have long been considered the main carcinogens of cigarette smoke but their concentration in the condensate is far too low to account by themselves for the production of skin tumors. The phenolic fraction does however have strong promotional activity when repeatedly applied to initiated mouse skin. Several constituents of cigarette smoke are co-carcinogenic when applied simultaneously with repeated applications of PAHs. Catechol is co-carcinogenic at concentrations found in the condensate. Since cigarette smoking involves protracted exposure to all the smoke constituents, co-carcinogenesis simulates its effects. Both procedures, however, indicate a major role for selection in carcinogenesis by cigarette smoke. That selection may operate on endogenous mutations as well as those induced by PAHs. There are indications that the nicotine-derived NNK which is a specific pulmonary carcinogen in animals contributes to smoking-induced lung cancer in man. Lung adenoma development by inhalation has been induced in mice by the gas phase of cigarette smoke. The role of selection has not been evaluated in either of these cases.
Abbreviations: 4-
-PDD, phorbol-12,13-didecanoate with the 4-hydroxyl in the position; B[a]P, benzo[a]pyrene; B[e]P, benzo[e]pyrene; BPDE, B[a]P 7,8-dihydrodiol-9,10-epoxide; BPDE-1, also called the syn or cis isomer and diol epoxide II; BPDE-2, also called the anti or trans isomer and diol epoxide I; DBA, dibenz[a,h]anthracene; DB[a,c]A, dibenz[a,c]anthracene; DBP, dibenzo[a,i]pyrene; DMBA, 7,12-dimethylbenz[a]anthracene; HeLa, human cervical cancer cells; MCA, 3-methylcholanthrene; MMS, methyl methane-sulfonate; NNK, nicotine-derived nitrosamino ketone; PAHs, polycyclic aromatic hydrocarbons; PDD, phorbol-12,13-didecanoate; RPA, phorbol-12-retinoate-13-acetate (originally called PRA but later and here called RPA); TPA, 12-O-tetradecanoylphorbol-13-acetate, also known as phorbol 12-myristate-13-acetate (also called PMA).
Modern research on chemical carcinogenesis began with the isolation of benzo[a]pyrene (B[a]P) from coal tar in 1930, and the demonstration that B[a]P and several synthetic polycyclic aromatic hydrocarbons (PAHs) induced tumors upon repeated painting on mouse skin. Many fruitful years of biological and chemical experimentation followed but a new era was marked by the finding that some PAH metabolites, namely their vicinal bay region diol epoxides, formed covalent adducts with DNA. It was generally agreed that the resultant mutations in DNA were the essential initiating steps in carcinogenesis and that one of the diol epoxide enantiomers was the ultimate carcinogen. The most carefully studied ultimate carcinogen was B[a]P diol epoxide-2 (BPDE-2), one of four enantiomeric diol epoxides of B[a]P, which lived up to expectations by being far more effective than its parent hydrocarbon in the induction of lung tumors in newborn mice upon intraperitoneal injection. However, BPDE-2 was far less effective than B[a]P in producing epidermal tumors on repeated topical application to mouse skin despite the demonstration that it bound to DNA as well after its application, as did BPDE-2, which was formed metabolically after applying B[a]P. Although the tumorigenic activity of a single topical application of BPDE-2 was enhanced by following it with repeated treatment of the skin with the promoting agent 12-O-tetradecanoyl-phorbol-13-acetate (TPA) in an initiation–promotion sequence, it was still less effective than promotion after B[a]P initiation. Furthermore, the enantiomer BPDE-1, which had a similar overall binding capacity to epidermal DNA as BPDE-2, was far less active as either a complete carcinogen or an initiator. These and other results suggested that other metabolites of B[a]P may have participated in skin tumor induction, perhaps in a manner synergistic to that of DNA adduct formation.
The foregoing inconsistencies occasioned the present re-examination of a wide range of results in the biology of carcinogenesis by PAHs, including early pathogenesis of PAH-induced skin tumors, localization and persistence of PAHs in skin, early chromosomal changes, transformation of cells in culture, cytotoxicity and mutagenicity. Detailed consideration is given to the role of the two-stage initiation–promotion sequence (later evolved into an additional stage of promotion) which exhibited synergistic interactions between the stages, and first raised the possibility that some metabolites of B[a]P other than the diol epoxides act as promoters which introduce a large element of selection in the carcinogenic process. Promotion by chemicals other than PAHs in coal tar is also considered, based on findings with cigarette smoke condensate which exhibits strong promotional activity in fractions that contain no PAHs. It is also likely that PAHs and their metabolites have indirect mechanisms of carcinogenesis, exemplified by the production of reactive oxygen and nitrogen species, that are complementary to the mutations induced by the diol epoxides. Several possible directions for future research are indicated with emphasis on the disposition of the PAHs and their metabolites, which could help to clarify the mechanisms involved in skin carcinogenesis by the PAHs.
Most studies of carcinogenesis by tobacco smoke have used mouse skin as the target organ. The concentrations of PAHs in tobacco smoke are far too low to be carcinogenic in skin by themselves, and require the presence of promoters and/or co-carcinogens to induce tumors. The presence of these cofactors in effective concentrations in cigarette smoke condensate indicates that selection of mutated cells makes a significant contribution to carcinogenesis by tobacco smoke. Recently, a reproducible method for producing lung tumors in mice by inhalation of tobacco smoke has been developed. The results indicate that volatile pulmonary carcinogens are present in the gas phase of tobacco smoke.
There is currently sharp disagreement about the relative roles in human pulmonary carcinogenesis of the direct induction of mutations by PAHs in tobacco smoke versus the selection of endogenous mutations. The disagreement is based on differing interpretations of base changes in codons of the p53 gene of human lung cancers. It is unlikely that the issue can be fully resolved without taking into account the presence in tobacco smoke of (i) grossly subcarcinogenic concentrations of active PAHs, (ii) PAHs that are initiators but have little or no activity as complete carcinogens, (iii) effective concentrations of promoters and co-carcinogens for selective clonal expansion, (iv) N-nitrosamines which selectively induce lung tumors in rodents regardless of the route of administration, and (v) gas phase components that induce lung tumors in susceptible mice by inhalation according to a particular regime. The situation is reminiscent of the finding in carcinogenesis of rabbit skin that unrefined coal tar induces tumors that appear earlier, enlarge much faster and are more likely to develop carcinomatous characteristics than those induced by concentrations of pure PAHs far higher than their concentration in the coal tar. The overall results indicate that many diverse substances in coal tar and in tobacco smoke interact via synergistic mechanisms in producing tumors. Selection of endogenous and smoke-induced mutations is likely to play a significant role in human lung carcinogenesis.
Binding of PAH metabolites to purine residues in DNA
The polycyclic aromatic hydrocarbons (PAHs) were the first pure compounds of known composition that were shown to cause cancer experimentally (1). This was definitively achieved with dibenz[a,h]anthracene (DBA) (Figure 1E
) and its 3-methyl derivative, which had been synthesized and tested for tumor production in mouse epidermis because they had a fluorescence spectrum similar to the dominant spectrum associated with carcinogenic fractions of coal tar (1). Hieger then purified an alcohol extract of coal tar that was highly carcinogenic and exhibited the correct fluorescence (2,3). This product was identical with synthetic B[a]P (Figure 1A), a previously unknown PAH, which was also highly carcinogenic for mouse skin (2). Soon thereafter, many more PAHs were synthesized, a number of which were highly carcinogenic (3). The Kennaway group sought endogenous biochemical pathways for generation of carcinogenic hydrocarbons and the only endogenous compounds that appeared to be chemically related were the recently discovered steroids (4). The idea that a steroid-like hormonal action was responsible for the carcinogenic action of the PAHs gained credence with the finding that the steroidal ovarian hormone estrone induced mammary cancer in male mice. It was decided, however, that conversion of PAHs to steroids was unlikely to occur in vivo (5) and the matter was largely dropped for many years. There was a renewed interest when careful examination showed a remarkable steric resemblance between the carcinogenic PAHs, most of which contain four to five condensed aromatic nuclei, and the steroids (6). The authors' analysis indicated a direct increase in carcinogenicity as the PAHs become sterically more similar to steroids (6). Interest in this relationship, however, virtually disappeared from sight when it was found that a small fraction of several radioactively labeled PAHs applied to mouse skin was firmly bound to DNA (7). The time course of the binding suggested that metabolism of the compounds was necessary to enable their reaction with DNA. Although there was also binding to RNA and protein, only the binding to DNA was correlated with the carcinogenicity of the PAHs for mouse skin as determined many years earlier (8). The binding to DNA lent weight to the widely held view that the essential step in carcinogenesis is a mutation or series of mutations.
|
It was then confirmed that the binding of PAHs represents the true extent of metabolic reaction between the hydrocarbons and DNA (9). However, the significance of the correlation between the extent of binding to DNA of mouse skin and carcinogenicity at that site was called into question by the observation of two instances in which the binding of a PAH to DNA was not associated with carcinogenicity. The binding of the non-carcinogenic compound DB[a,c]A (Figure 1F
) to DNA of mouse skin was greater than that of its carcinogenic isomer DBA (9). And the binding of 7,12-dimethylbenz[a]anthracene (DMBA) to DNA of a strain of hairless mice, in which it produced no tumors, was identical to that in normal mice, in which carcinogenesis does occur. The authors suggested several alternative explanations for the discrepancy between DNA binding and carcinogenesis, including the possibility that the binding is necessary for carcinogenesis but some additional reaction or environmental condition modified by the carcinogen and its metabolites must contribute to eventual tumor formation. This introduced the possibility that some degree of selective growth is required before mutational change can be expressed in tumor formation.
A major clue about the PAH metabolites that bound covalently to DNA came from the report that B[a]P-7,8-diol, a metabolite of B[a]P, bound to DNA extracted from Syrian hamster cells to a 10-fold greater extent than did B[a]P itself (10). The precise identification of the metabolites bound to DNA of intact mammalian cells was revealed by experiments in which radioactively labeled B[a]P was incubated with primary cultures of Syrian hamster cells (11). The elution profiles of DNA hydrolysates from these cells had peaks that were coincident with those of hydrolyzed DNA obtained from incubation of the cells with B[a]P-7,8-dihydrodiol and of DNA from hamster cells incubated with B[a]P-7,8-dihydrodiol-9,10-epoxide. The conclusion was that the coincident peaks from the four treatments were B[a]P-7,8-dihydrodiol-9,10-epoxide-deoxyribosides. A second product was not further investigated, nor was a peak from incubation of the cells with B[a]P-9,10-dihydrodiol. The results indicated that B[a]P is metabolized in cells to the 7,8-dihydrodiol which is a precursor to the 7,8-dihydrodiol-9,10-epoxide (BPDE) (Figure 1B and C) that reacts with cellular DNA. Preliminary experiments indicated that analogous metabolic activation mechanisms applied to other PAHs. The B[a]P diol epoxides bind extensively to guanine residues of DNA (12) (Figure 1D) and the diol epoxides of 7,12-dimethylbenz[a]anthracene (DMBA) and of DB[a,l]P bind mainly to adenine residues of DNA (13,14).
Carcinogenesis and mutagenesis by PAHs and their metabolites
The great bulk of studies in the 1930s on the carcinogenic activity of PAHs was done by repeated painting on the skin of mice to produce papillomas and carcinomas (15). Some auxiliary studies, such as the persistence of the carcinogens in tissue were mainly done by single subcutaneous inoculation to produce sarcomas. In the 1940's it was discovered that although a single painting of a low dose of the carcinogenic PAHs on mouse skin produced few if any tumors, subsequent repeated painting with croton oil, which by itself appeared to be non-carcinogenic, resulted in efficient tumor production (16). This method was used extensively by Berenblum and Shubik (17–19) to develop the two-stage concept of carcinogenesis. The first or initiating stage brought about by a PAH consists of a specific and irreversible conversion of normal cells to latent tumor cells that lie dormant until stimulated by croton oil in the promoting phase to become visible tumors (20). The majority of tumors resulting from the initiation–promotion procedure were benign papillomas. The promoting agent croton oil itself was initially thought to be non-tumorigenic in repeated painting even if followed by a single initiating dose of a carcinogenic PAH. Subsequent work to be considered later has changed this point of view and resulted in a more complex picture of promotion. The active ingredient of croton oil is TPA. Many of the papillomas in the initiation–promotion scheme eventually regress.
Since the repeated application of the carcinogenic PAHs alone results in production of tumors, many of which are malignant carcinomas (20), they are considered complete carcinogens (15). Complete carcinogens carry out both initiating and promoting functions. When used in both initiating and promoting stages, they yield a higher incidence of carcinomas than does initiation with a PAH and promotion by TPA (20). It was suggested that malignant tumor formation results from two or more carcinogen-induced mutations and that the role of promotion is to enlarge the size of the target cell population available for the second mutation (21). Papillomas induced by repeated carcinogen applications arise from significantly more cells than those induced by the carcinogen-promoter sequence (22). Since many of the papillomas in the initiation–promotion sequence regress when TPA applications cease, it was suggested that TPA allows the expression of the neoplastic phenotype in the initiated cells by a selective (23) or epigenetic mechanism (22). A selective role of TPA in neoplastic development was unequivocally demonstrated in spontaneous transformation of an established line of mouse fibroblasts in which the TPA promotional activity was effective only when continuously applied for at least 4 weeks to cells maintained under strong regulatory conditions (24).
Early testing of PAHs for mutagenic capacity in Drosophila, mice and microorganisms failed to yield consistent evidence of mutagenesis (25). Later results indicated that the active forms of the PAHs are generated by mammalian metabolism, in particular by the TPNH-dependent microsomal enzymes of liver (cited in ref. 26). Mixing of the PAHs with liver homogenates yielded products that were potent frameshift mutagens in certain mutants of Salmonella typhimurium (26).
Bacteria do not duplicate mammalian metabolism in activating carcinogens, so they can be used to test the mutagenic capacity of diverse pure compounds of the type that are formed during the metabolism of the PAHs in mammalian cells. Likewise, the V79 line of Chinese hamster cells does not have the capacity to metabolically activate PAHs, so it can also be used to test various PAH metabolites for mutagenic activity (27). The two mutagenesis test systems were used in combination with carcinogenic capacity in repeated application of B[a]P and its metabolites to mouse skin for 60 weeks to determine the relation of mutagenesis to carcinogenesis for each compound. The results, adapted from the extensive work of Conney and colleagues (28–31) are summarized in Table I. B[a]P, B[a]P-7,8-dihydrodiol and 2-HOBP are equally strong complete carcinogens on mouse epidermis but have no significant mutagenic power in either V79 hamster cells or S.typhimurium. B[a]P-7,8-oxide is a less active carcinogen with slight mutagenic power in S.typhimurium and none in V79 cells. B[a]P-7,8-diol-9,10-epoxide-2 (Figure 1C) (BPDE-2, also called the anti or trans isomer and diol-epoxide I, is a vicinal, bay region diol epoxide) has slight carcinogenic power on mouse epidermis and high mutagenic capacity in both test systems. Its isomer BPDE-1 (also called the syn or cis isomer and diol-epoxide II) (Figure 1B) produced no tumors even at the highest dose tested, but is highly mutagenic. B[a]P-4,5-oxide and 11-HOBP have slight carcinogenic activity and little or no mutagenic activity. The rest of the metabolites tested for both activities are non-carcinogenic and are at best slightly mutagenic in S.typhimurium. Thus there is a striking disjunction between strong epidermal carcinogenicity of B[a]P and its metabolites on the one hand and strong mutagenicity on the other. The strong carcinogenicity of B[a]P-7,8-dihydrodiol, which is the precursor to the BPDEs, combined with the binding of the latter to DNA and their mutagenicity is the rationale for considering the B[a]P-7,8-dihydrodiol as the proximate carcinogen (32). The binding of BPDEs to DNA and their mutagenic potency qualify them as ultimate carcinogens, but their weak or non-carcinogenic capacity for the standard test as complete carcinogens on mouse epidermis has yet to be accounted for, as does the strong carcinogenesis by 2-HOBP, which is not a known precursor to the BPDEs, nor is it a known metabolite of B[a]P (33).
|
The capacity of B[a]P dihydrodiols to initiate tumor formation was tested in a single application followed by repeated treatment with the promoter TPA for 52 weeks (34). B[a]P-7,8-dihydrodiol was almost as active as initiator as B[a]P itself when measured by the number of mice that developed tumors, but significantly less active when measured by the number of tumors per mouse. However, in another study, B[a]P-7,8-dihydrodiol was as potent an initiator as B[a]P itself (35). The 4,5- and 9,10-dihydrodiols did produce tumors but were less active than the 7,8-dihydrodiol (34). When enantiomers of B[a]P-7,8-dihydrodiol were separately tested, the (–) enantiomer was more active and the (+) enantiomer was considerably less active than B[a]P (28). One of the diol epoxides had 20–30% the initiating activity of B[a]P and the other had only 1% the activity of B[a]P (36). The 9,10- and 11,12-oxides had 2 and 10% the tumor-initiating activity of B[a]P. Hence, both diol epoxides were somewhat more active as tumor initiators than as complete carcinogens, but considerably less active than B[a]P itself in both activities. The only metabolite found to be more potent as a tumor initiator than B[a]P was the (–) enantiomer of trans 7,8-dihydrodiol, which recommended it as a proximate carcinogen requiring further metabolism to an ultimate carcinogen, presumably the diol epoxides.
In a later study, one of four enantiomers of BPDE, designated (+)-BPDE-2, was ~60% as active a tumor initiator as B[a]P (37). The (–)-BPDE-2 and both enantiomers of BPDE-1 had little or no tumor initiating activity. The data suggested that (+)-BPDE-2 is the ultimate carcinogen of B[a]P. The tumor initiating activity of the racemic mixture (±)BPDE was somewhat higher than its activity as a complete carcinogen (29,30), which suggests a deficiency of promoting activity in repeated application as a complete carcinogen.
The diols of 3-methylcholanthrene (MCA) and DMBA capable of conversion to bay region vicinal diol epoxides (MCA-9,10-diol and DMBA-3,4-diol) were not as active in tumor initiation as their parent hydrocarbons when measured in terms of the total number of tumors produced (38). The vicinal diol epoxides of DMBA applied at 100 nmol were less active as tumor initiators than one-tenth that concentration of DMBA itself, especially when evaluated as tumors per mouse rather than percent of mice with tumors (39). Surprisingly, the K-region 5,6-diol of DMBA, which cannot be converted directly into a vicinal diol epoxide, was more active in producing tumors than the 3,4-diol. No satisfactory explanation for the tumor-initiating activity of the DMBA-5,6-diol could be offered in view of the generally accepted importance of the bay region vicinal diol epoxide in DMBA initiation and mutagenesis (40). Indeed, the disagreement in tumor-initiating activities of DMBA derivatives between different laboratories is problematic (38,40). If the initiating activity is thought to be mediated exclusively through the DMBA diols and their epoxides, there is an apparent anomaly in the very small amount of those derivatives present in tissues after application of the parent hydrocarbon in view of the fact that the application of large amounts of the diols or their epoxides are less effective as tumor initiators than the parent hydrocarbon. A similar picture was observed for the very highly carcinogenic dibenzo[a,l]pyrene and its fjord-region diol and diol epoxides (41). This can be only partly explained by the claim that PAHs have an alternate mechanism of activation in the form of radical cations (42) which is, in any case, offset by evidence against such a claim (43).
The derivatives of a number of other hydrocarbons have been tested in the initiation–promotion sequence. While the dihydrodiol precursors of bay region diol epoxides are as active as the parent hydrocarbons in most cases, several other derivatives also exhibit significant activity, especially for the weak carcinogens chrysene and benz[a]anthracene (44). The dihydrodiol epoxides were generally less active than the parent hydrocarbons, leading to the suggestion that `the tumor initiation requires more from a hydrocarbon than the generation of a reactive dihydrodiol epoxide metabolite' (44).
Cytotoxicity of B[a]P and its metabolites
The designation of BPDE-2 as the ultimate carcinogen for B[a]P leads to the expectation that it would be at least as carcinogenic as the parent hydrocarbon, which it is obviously not on mouse skin, either as a complete carcinogen or as an initiator. One conceivable explanation for this failure that has received little attention is that BPDEs are extremely cytotoxic and destroy the cells in which they are most concentrated. These are presumably rapidly dividing stem cells found in the hair follicles (45–47).
BPDE-1 is at least 60-fold more cytotoxic than B[a]P-4,5-oxide in V79 Chinese hamster cells (48). Although BPDE-1 was highly mutagenic at very low concentrations, there was a sharp reduction in mutations at concentrations higher than 1.0 nmol. In contrast, mutagenesis by B[a]P-4,5-oxide rose continuously far beyond concentrations of the diol epoxide that reduced BPDE-1-induced mutations per surviving cell to less than the control level. Further testing showed that BPDE-2 was even more cytotoxic to V79 cells than BPDE-1 and almost 100-fold more cytotoxic than the moderate carcinogen H4-7,8-epoxide (49). Non-carcinogenic H4-9,10-epoxide was almost as cytotoxic as BPDE-2. In another study (50), BPDE-2 was also highly cytotoxic at low concentrations to V79 cells, although BPDE-1 was much less cytotoxic (Table II). The parental hydrocarbon BP displayed no cytotoxicity, nor did most of the phenolic derivatives, of which only one fairly representative member is shown. There were discrepancies between the two laboratories about the relative cytotoxicities of BPDE-1 and B[a]P-4,5-oxide, which were attributed to differences in methodology and specific V79 clones used as target cells, but there was agreement on the extreme cytotoxicity of BPDE-2 (49). It is noteworthy that both BPDEs are highly toxic to newborn mice upon intraperitoneal injection (51). It has since been shown that the toxicity of dibenzo[a,l]pyrene, the most carcinogenic PAH known, interferes with its carcinogenicity at higher doses (52). Therefore the toxicity of BPDEs should be taken into account as a possible explanation for their low carcinogenicity for mouse skin. Particular attention should be paid to their effect on the stem cells of the hair follicles. However, the reduction in tumor production by BPDE-2 with dilution (Table IB) argues against this interpretation.
|
Binding of diol epoxides to DNA
If, as widely believed, metabolic alteration of B[a]P is necessary for carcinogenesis and BPDE-2 is the ultimate carcinogen, then one might expect the application of the pure compound to be much more active in tumor production than the parent hydrocarbons. The fact that it was much less active on mouse skin calls for an explanation. BPDEs are much more reactive than other B[a]P derivatives and much less stable, with a half-life in phosphate buffered saline of at most a few minutes (49). It therefore seemed perfectly reasonable to attribute their low carcinogenicity to their inability to reach their target DNA when applied directly to skin (29).
The first breach in this reasoning came with the report that the amount of BPDE-2 combined with epidermal DNA after application of the pure derivative was the same as that bound after application of equal amounts of B[a]P or B[a]P-7,8-dihydrodiol (53). It was then suggested that metabolites derived from B[a]P and B[a]P-7,8-dihydrodiol other than BPDE-2 may be involved in skin tumor initiation, or that in vivo metabolically generated diol epoxide reacts more specifically with DNA than does topically applied BPDE-2 (53).
There remained the question as to whether the diol epoxides were actually bound to the DNA of basal cells of the epidermis, which were assumed to be the target cells for initiation. It was found that both BPDE diastereomers bound equally to DNA of the basal layer, as well as to RNA and protein (54). This indicated a lack of correlation between extent of binding and carcinogenicity since BPDE-2 is a far more active initiator than BPDE-1. There is the additional problem that although the (+) enantiomer of BPDE-2 is ~66-fold more potent an initiator than the (–) form, the rate of formation and disappearance of individual adducts was similar over a 72 h period (55). The possibility was suggested that there might be a small subpopulation of stem cells in which the specific binding of the two enantiomers of BPDE-2 is correlated with their tumorigenicity, but there has been no substantiation of this hypothesis.
In later studies it was found that BPDE-2 penetrates rapidly through the epidermis after its topical application and that the bulk of it is rapidly removed (56). The relatively low tumorigenicity of BPDE-2 for mouse skin might partly be accounted for by its rapid passage through the epidermis. However, the levels of total BPDE-2 present in the epidermis after application of BPDE-2 or B[a]P-dihydrodiol were 50–100-fold greater than after application of B[a]P. Despite the differences in total BPDE-2 in the epidermis after application of the three compounds, the extent of formation of BPDE-2 adducts to DNA was about the same and therefore once again not correlated with their tumorigenicity in skin. It was suggested that the `disposition' of the metabolites when applied externally may differ quantitatively—and perhaps, qualitatively—from the disposition of metabolites generated intracellularly from B[a]P (56). It is noteworthy in this regard that BPDE-2 is bound 7–8-fold more efficiently to DNA than BPDE-1 when the parental hydrocarbon is topically applied to mouse skin (57), in contrast to the similar binding of the diol epoxides when the derivatives themselves are applied (54–56). A precise account of the distribution of diol epoxide adducts at specific nucleotide positions of DNA would help to resolve the relationship between binding and carcinogenesis.
The relation between carcinogenicity and DNA binding has also been studied with DMBA and its bay region diol epoxides (DMBADEs). Both syn and anti DMBADEs initiate papillomas in mouse skin when followed by repeated application of TPA, but are less active than DMBA (39). The parent hydrocarbon produces ~6-fold more papillomas per mouse than a 10-fold higher concentration of either the syn or anti DMBADE, although the sum of the binding of the diol epoxides to DNA when they are applied at those high concentrations is the same as that of the derivatives of DMBA when the latter is applied at one-tenth the concentration. It was suggested that the lower efficiency of DNA binding by the topically applied diol epoxides is the result of their high reactivity toward nucleophiles, plus the need to pass through the stratum corneum and granular spinous layers of the epidermis before reaching the target cells in the basal layer (39). However, mouse epidermis consists of only one or in some places two cell layers (58,59), unlike the multilayered epidermis of humans. Measurement shows that large amounts of topically applied diol epoxides reach the basal cells (56). It was also contended that extrapolation of the `tumor-initiating potential' for a given DMBA adduct from the maximum number of tumors initiated by the same adduct produced by the DMBADEs at the doses used (10 nmol for DMBA, 100 nmol for the DMBADEs), one could fully account for all the tumors formed by DMBA and by the DMBADEs in terms of the measured DNA adducts (39). However, there is no evidence that extrapolation of the low tumor yield from applied DMBADEs to the high tumor yield from DMBA application is warranted, since there is no indication of a linear relation between major DNA adducts and tumor development. All that can be said is that the number of papillomas formed by low doses of DMBA is much higher than 10-fold as much of the DMBADEs when the opposite relation would be expected if the latter accounted for the full tumorigenic potential of DMBA in mouse skin. This discrepancy remains despite the fact that tumor initiating activity, as used in this case, is a more efficient method for production of tumors by diol epoxides than is the method of complete carcinogenesis (28,30,35–37) and therefore presents the best case for considering the diol epoxides as the ultimate carcinogens of PAHs in mouse skin.
Tumor production in newborn mice
Unlike the equivocal findings with mouse skin, a pivotal role for the diol epoxides was strongly supported by a series of articles on tumorigenesis induced by three weekly intraperitoneal injections into newborn mice of step-wise increases of B[a]P or its metabolites (Table III). The diol epoxides were highly toxic to mice even at 1/50th the dose of B[a]P, but BPDE-2 produced about the same incidence of pulmonary adenomas as the 50-fold higher but non-toxic dose of B[a]P (51). (B[a]P-7,8-dihydrodiol produced even more pulmonary adenomas per mouse than B[a]P at the same high dose and also produced a high incidence of malignant lymphoma, a lesion not produced by the other compounds.) BPDE-1 was so toxic in these experiments that it left too few survivors for accurate evaluation of its carcinogenicity. BPDE-2 and B[a]P-7,8-dihydrodiol were, respectively, about 40- and 15-fold more active than B[a]P in causing pulmonary adenomas in newborn mice (60). The BPDE-1 and B[a]P tetraol products of BPDE-2 produced no more adenomas than the control level of 0.13 per mouse. When BPDE was separated into its two optical enantiomers and tested at extremely low doses (7 and 14 nmol) in newborn mice, only BPDE-2 produced significant increases in pulmonary adenomas over the control value, and B[a]P was ineffective at these doses (61). The extremely powerful capacity of the (+)-BPDE-2 enantiomer to induce pulmonary tumors in newborn mice when compared with its parental hydrocarbon supports its role as an ultimate carcinogen in the newborn mouse, but raises questions about the ultimate carcinogen designation of BPDE-2 in skin, where it is less carcinogenic than B[a]P.
|
It should be noted that an average of ~10% of the newborn controls spontaneously develop pulmonary adenomas. A large increase in pulmonary tumor incidence can be induced in newborn mice with only a single low subcutaneous dose of B[a]P, DMBA or 3-MCA within 24 h of birth (62). A single subcutaneous inoculation into 8-week-old mice of a 20–30-fold greater amount of DMBA produced only one-fifth the number of pulmonary tumors as the lower dose in the newborns. The high spontaneous incidence of pulmonary adenomas in the newborns and the susceptibility to low doses of BPDE-2 suggests that only a single step is required for more tumors to develop, but it also emphasizes that tumor formation in the adult involves more variables.
Pre-neoplastic effects of PAHs in cultures of mouse epidermal cells
Mouse basal epidermal cells or keratinocytes multiply continuously in cell culture if the concentration of calcium in the medium is reduced to 0.1 mM or less (63). At higher concentrations of calcium, proliferation ceases and the vast majority of the cells terminally differentiate. A few cells continue to proliferate and form colonies in the high calcium. The proportion of colony-forming cells increases with the time the cells are kept in low calcium before switching to high calcium. A distinction of malignant epidermal cells is that they continue to proliferate when switched to high calcium. One day treatment of normal epidermal cultures with DMBA in low calcium, followed by several weeks of continued incubation in low calcium results in a 4–10-fold increase in the number of colonies formed after switching to high calcium (63). However, the altered cells do not exhibit the full range of characteristics of the malignant cells, which include growth in agar and tumor production in mice. There is a correlation between the initiating potency of skin carcinogenesis in vivo and capacity to induce resistance to terminal differentiation (64). The results suggested that the carcinogens induce a pre-neoplastic state in the epidermal cells.
BPDE-2 was ~5-fold more effective than B[a]P in the induction of transformed foci in the epidermal cells and this difference correlated with the quantity of DNA adducts formed (65). The apparent transforming capacity of BPDE-2 for the cultured epidermal cells is in marked distinction to its poor tumorigenic capacity on mouse skin. A similar distinction from mouse skin is the high capacity of BPDE-2 to produce pulmonary adenomas in newborn mice. There are other biological features of carcinogenesis in these two BPDE-2 sensitive systems that distinguish it from carcinogenesis in mouse skin. One is the fact that some keratinocytes resistant to terminal differentiation occur in control cultures, as do some pulmonary adenomas in control newborn mice, while spontaneous skin papillomas are extremely rare in controls. A second difference is that there is no requirement for an extended promotional treatment in the two systems in contrast to that in the skin. A third difference is the plateau in binding of BPDE-2 when it is applied in high doses to mouse skin in contrast to the absence of such a limit when it is applied to the cultured epidermal cells (65). Information about the basis of the difference in DNA binding in the two cases could be revealing about the difference in biological response.
In vitro malignant transformation of rodent cells by PAHs
Exposure of primary or secondary cultures of Syrian hamster embryo cells to PAHs led to their morphological transformation associated with a capacity to induce tumors when inoculated into adult hamsters (66). Small numbers of cells were seeded for cloning and left for 8 days in the presence of PAH. Up to 25% of the clones that developed had a transformed morphology. Mass cultures treated for 8 days with carcinogenic PAHs and cultured in the absence of carcinogen for several months produced sarcomas in hamsters. The high percentages of transformed clones may be deceptive because it was later found that continued incubation after removal of the carcinogenic PAH from the low cell density procedure resulted in a marked reduction in the number of morphologically transformed colonies, indicating that most were only phenotypically transformed (67).
Before the discovery that the vicinal bay region diol epoxides were the major PAH derivatives that bound covalently to DNA (11), much attention was focused on the K-region of the PAHs as the likely active site for carcinogenesis (68). This was based on the false premise that the parent hydrocarbons were biologically active as such and that their metabolism was a purely detoxifying process (3). When it was found that many compounds did not display the carcinogenicity expected on the basis of the reactive K-region, the idea that chemically unaltered PAHs were the carcinogens was undermined.
In 1950, however, it was suggested that the carcinogenicity of PAHs was mediated through metabolically formed epoxides (69). All the PAHs examined formed epoxides at the K-region, which focused attention on the possibility that the K-region epoxides are the ultimately reactive forms in vivo. Synthetic K-region epoxides reacted with nucleic acids and proteins (70). They were highly mutagenic in the Chinese hamster V79 cell line without affecting the modal diploid chromosome number (71) and produced frameshift mutations in S.typhimurium (72). Interest in the K-region epoxides had earlier been tempered by their low to moderate activity as complete carcinogens in either mice or rats when inoculated subcutaneously or applied topically to the skin (73,74) and moderate activity as initiators on mouse skin (73). The findings of their binding to DNA and mutagenicity warranted examination of the in vitro transforming activity of the K-region epoxides for cells in culture. They proved to be more active in transforming Syrian hamster embryo cells than the hydrocarbons or corresponding K-region phenols (75). Oddly enough, the K-region epoxide of BA was a more efficient transforming agent than the K-region epoxides of DBA and MCA, although the parent hydrocarbon of BA is a much weaker carcinogen than DBA or MCA. Similar effects were produced in a line of cells derived from mouse prostate although the K-region epoxides of MCA were more active in these cells than those of BA and DBA (76). The results in the mouse cells were extended to the K-region epoxides of other PAHs (77). In both sets, cells from the morphologically transformed foci induced anaplastic soft tissue sarcomas in isologous mice.
Subsequently it was found that the hydrocarbon–deoxynucleoside products isolated from mouse embryo cells treated with PAHs differed from those of cell-free DNA treated with K-region epoxides of PAHs (78). It was then apparent that the major metabolites responsible for the covalent modification of DNA were not K-region epoxides and this path of research reached a dead end (3). About the same time it was shown that the chromatographic profiles of DNA digests from hamster cells treated with B[a]P matched those of DNA reacted with the bay region 7,8-diol-9,10-epoxide of B[a]P which suggested that the 7,8-diol-9,10-epoxide of B[a]P was the major DNA-bound adduct formed from B[a]P metabolism (11). The bay region diols of 7-methyl BA, DMBA and B[a]P were more potent in the transforming of mouse fibroblasts than the parental hydrocarbons, and much more potent than K-region diols (79). The K-region epoxides ranged from weakly (7-methyl BA and B[a]P) to moderately (DMBA) transforming but always less so than either the bay region diols or the parental hydrocarbons. When the transforming capacity of B[a]P and its metabolites was tested on normal embryo Syrian hamster cells, the bay region trans 7,8-diol was more active than the other non-K-region 4,5- or 9,10-diols and B[a]P (80). BPDE-2 was more active than BPDE-1, which was slightly more active than the K-region epoxide of B[a]P.
In summary, the results in cell culture, along with the high mutagenicity and DNA binding of BPDE-2, supported its status as the ultimate carcinogen of B[a]P. The results agreed in part with the capacity of BPDE-2 to produce both the lung adenomas in newborn mice and the differentiation resistant state of mouse keratinocytes in culture. But they differed from the findings on mouse skin in vivo where BPDE-2 is a poorer carcinogen and initiator than B[a]P or B[a]P-7,8-diol, and BPDE-1 is virtually non-carcinogenic in either mouse skin or newborn mice. When combined with the moderate tumor-initiating activity in mice of K-region epoxides (73), which could not be detected as adducts to DNA when the parental hydrocarbons were administered, the results call for a fuller analysis of the mechanisms of transformation and carcinogenesis by PAHs. To carry out such an analysis, it is important to consider the relevant biological aspects of PAH-induced skin tumor development that have to be accounted for by any proposed mechanisms. Many of these were established in the pre-molecular era and are rarely acknowledged in present day discussions.
Requirement for long-term repeated application of PAHs for skin carcinogenesis
Yamagiwa and Ichikawa (81) had succeeded in producing cancer with coal tar because they applied it (on rabbit ears) more persistently over a longer period of time than had their predecessors. The interest in coal tar arose from the high incidence of skin cancer among workers in the coal gasification industry which had coal tar as a by-product. It was natural therefore that application to the skin of experimental animals—more particularly mice because of their convenience—should be the method of choice for testing components of coal tar for carcinogenicity, as well as for related synthetic chemicals. A consistent finding, first with B[a]P from coal tar and synthetic DBA, then with other synthetic PAHs such as DMBA and MCA, was that they had to be applied repeatedly to mouse skin on a weekly or more frequent basis for months before papillomas appeared, and even longer for carcinomas to appear. For example, Hieger (82) found that painting mouse skin bi-weekly for 8 weeks with 0.3% of either DBA or B[a]P yielded no tumors, but painting for 8 weeks with DBA followed by 8 weeks of B[a]P was tumorigenic. A single subcutaneous inoculation of B[a]P in a variety of vehicles rarely produced sarcomas if disappearance of the particular combination was complete in 3 or 4 months but was highly tumorigenic in other combinations which remained detectable by fluorescence in ultraviolet light for more than 5 or 6 months (83).
Tumors were common at the site of subcutaneous inoculation of B[a]P where it persisted for a long time, but no tumors appeared at the site of intraperitoneal inoculation where the rate of disappearance was rapid (84). Rabbits, which were very susceptible to skin carcinogenesis by topical application of DMBA, were highly resistant to subcutaneous inoculation, unlike mice which were susceptible by both routes (85). The resistance of rabbits to subcutaneous inoculation was correlated with the much faster disappearance of DMBA from the subcutaneous tissue of rabbits than of mice.
In one of the first examples of PAH persistence, tumors induced by a lard solution of DBA contained an appreciable amount of the carcinogen 6–8 months after its subcutaneous inoculation (86). Another example of extreme persistence was seen after subcutaneous inoculation of DB[a]P in tricaprylin which produced 98% sarcomas in 24 weeks (87). No metabolites of DBP could be detected in the feces collected for a month after inoculation and the unaltered carcinogen could be found in the tumors months later. The low solubility of DBP was thought to account for the slowness of its action as compared with MCA, and the persistence at the inoculation site was believed to explain the equivalent tumor yield to MCA at one-third the dose of MCA.
The dynamics of disappearance of PAHs could be studied when radioactive labeling of the carcinogens was introduced. The results with three PAHs after subcutaneous inoculation in mice and with two of them after painting on skin are shown in Table IV. DBA disappeared from the subcutaneous site most slowly of the three with a half-life of ~12 weeks, followed by MCA with a half-life of 3.5 weeks and B[a]P with an initial half-life of 1.75 weeks (88,89). DBA had about the same half-life after skin painting as it did in subcutaneous inoculation, but B[a]P disappeared much more rapidly in skin painting with an initial half-life of 1.7 days and 4.3 days after the second day. Excretion was studied only with B[a]P, which appeared entirely in the feces as metabolites (89). The ratio of the neutral fraction (which consists mainly of unchanged hydrocarbon plus the diols) to the fraction bound to macromolecules remained high for subcutaneously inoculated B[a]P up to 8 weeks and decreased sharply at 12 weeks. That shift occurred within days when B[a]P was applied to skin. The neutral to bound ratio was much higher with DBA applied to skin and only showed a significant decrease at 20–21 days. Obviously the metabolism of B[a]P is much faster than that of DBA, with the latter persisting largely unchanged for extended periods of time as previously indicated using the method of fluorescence under ultraviolet light (86).
|
Localization of PAHs in tissues and cells
Striking features of PAHs are their low chemical reactivity and high lipophilicity. The PAHs which are created by treating organic material at extremely high temperatures are not chemically reactive but are altered under physiological conditions by enzyme metabolism. They are also very insoluble in water as indicated by the logarithm of their octanol:water coefficients as shown for B[a]P in Table V (90). Even most of the B[a]P metabolites, which have added polar groups to the parental hydrocarbon, are >104-fold more soluble in octanol than in water. BPDE has a considerably lower lipophilicity than the other metabolites, but it is still >103-fold more soluble in octanol than in water. It is only when conjugated with glucuronic acid or glutathione that the B[a]P derivatives become soluble in water and are excreted (91,92). Only a very small fraction of labeled B[a]P appears bound to macromolecules (89) including DNA (53,65).
|
Localization was studied mainly by observing fluorescence. A large fraction of MCA and B[a]P applied once to the skin is found in sebaceous glands and keratin (93,94). Many epidermal cells are killed by the carcinogenic PAHs, but resistant cells appear in a hyperplastic response to repeated application and the PAH can be seen in the cytoplasm, but not in the nucleus. B[a]P fluorescence was used as a method for detecting `masked' lipid not seen with conventional stains for lipid like osmic acid (95). B[a]P was associated in these early studies with nuclear membranes, regions rich in mitochondria and regions with an ultrastructure of `oriented proteins and lipids', i.e. presumably membranes (95). Later work in cell culture as well as epidermis using both fluorescence and autoradiography of radioactively labeled PAHs reported the bulk of the PAHs in lysosomes (96). None of these studies detected PAHs within the nucleus, but this is understandable in light of the small fraction bound to DNA (53,65) and the insensitivity of the techniques used in localization (97). Obviously, improved methods for isolating DNA and detecting radioactively labeled adducts of diol epoxides demonstrated that the diol epoxides were bound to nuclear DNA, but the localization studies emphasize that the great bulk of PAHs and their derivatives were associated with extranuclear structures and could be contributing to the carcinogenic process at those locations. One possible role of the unaltered hydrocarbon is to act as a reservoir for long-term production of diol epoxides.
Tissue response to PAH application on mouse epidermis
Application of 0.6% MCA in benzene three times a week to the backs of mice results in the first appearance of small wart-like swellings at 2 weeks, papillomas beginning at 4 weeks and increasing in number to 12 weeks and carcinomas beginning at 12 weeks and involving 100% of the mice at 24 weeks (58). Microscopically, after a single application there is widespread death of keratinocytes in the central area of direct application in the first few days with cellular and nuclear swelling accompanied later by cell proliferation in the peripheral areas (98). This response includes the hair follicles which help to repopulate the damaged areas. There are multiple layers of epidermal cells in the marginal areas at 3 days with further multilayering at 6 days. When lower concentrations of either MCA or DMBA are applied only once, there is less destruction in the central area but there is cell swelling at 2 days (Figures 2A and B). There is a mild hyperplasia at 6 days (Figure 2C), which decreases somewhat by 30 days but treated epidermis is still clearly distinguishable from controls (Figure 2D) (59). With multiple applications (six times a week) of the lower doses of MCA and DMBA, there is an extremely high degree of hyperplasia at 30 days, irregularity of shape and variability of size of both cells and nuclei, and abnormality of chromatin localization (Figure 2E) (59,98). There are also pronounced irregularities in the arrangement of various cell layers and a high degree of mitotic activity. After 12 weeks of MCA painting three times a week, a diverse reaction can be seen in a broad, low power field consisting of hyperplastic epidermis, regenerating hair follicles, benign and pre-cancerous papillomas and an invasive carcinoma (Figure 2H). The full reaction described above, including its persistence, is considered specific for carcinogenic PAHs (59,98,99). The shorter the latent period for tumor production, the smaller is the dose required to provoke the specific hyperplasia (100). This indicates that the carcinogenic potency of a PAH, which takes months for its full expression is anticipated by the nature of the early tissue reaction.
|
Early chromosomal responses to PAHs
The abnormal distribution of chromatin described above (98) was the first direct indication that the genetic material of the cell might be involved in the carcinogenic process. At the time there were no adequate techniques for characterizing individual chromosomes in tissues but squashes were made of epidermal cells at various times after multiple MCA applications were begun and the chromosomes were stained (101). Beginning at 3 days, an approximate doubling in size of chromosomes could be seen in ~8% of the cells, with an indication that this reflected an increase in chromosome strands rather than a mere swelling. The proportion of cells with this alteration remained about the same for 2 months and then rose to 20–30% at 72 days. In contrast, only 0–2% of the cells in papillomas had chromosome abnormalities, indicating they were selected against in development of the tumors. However, the later developing carcinomas had much higher percentages (40–76%) of chromosomally abnormal cells than any in the hyperplastic epidermal cells. It is noteworthy that subsequent improvements in chromosome analysis of papillomas after an initiation–promotion sequence confirmed that the great majority of them had a diploid karyotype that only changed to aneuploidy during the development of the carcinomas (102).
A quarter of a century after the initial relatively crude observations, improvement in cytogenetic techniques warranted further investigation of early chromosomal changes in cells treated with PAHs. Chinese hamster cells with their easily identified set of 22 metacentric chromosomes were treated with DMBA for 24 h (103). Single chromatid and chromosome breaks were seen in 23% of the cells at 12 h, with the proportion rising to 55% at 24 h (Table VI). The rise increased to >90% at 24 h after removal of DMBA with a gradual decrease in the next 72 h. Rejoining of the broken ends of chromosomes started after removal of DMBA at 24 h and was seen in 77% of the cells in the next 24 h, with a subsequent decline paralleling that of the chromosome breaks. At the fifth passage after treatment, the majority of DMBA-transformed cells and control cells were normal diploid, but the frequency of cells with structurally new chromosomes, presumably arising from rejoining between different broken chromosomes, was ~4-fold higher in the transformed than in the control cells.
|
Chromosome breaks were also seen in bone marrow cells of Chinese hamsters after subcutaneous or intraperitoneal inoculation of DMBA (104). The percentage of DMBA-treated cells with breaks in the hamster was generally lower than those found in cell culture, indicating greater stability of the genetic apparatus in the normal environment of the body than in cell culture. This point was reinforced by the low percentage of 0.4% breaks in the control marrow cells in the hamster in contrast to the 8% with breaks among the control cells in culture. Interchromatid exchanges occurred frequently, often involving interchanges between chromosomes. Multiple chromosome breaks were commonly seen in cells both in vitro and in vivo, but such cells were probably selected against in the long-term. Interchromosomal exchanges, of course, allow for gene rearrangements that might affect gene regulation and increase the likelihood of transformation.
There are a number of possible mechanisms by which PAHs could produce chromosome breakage. One is, of course, the binding of the bulky diol epoxides to DNA. A second involves the induction of a prooxidant state in PAH-treated cells (105). Evidence for a role of reactive oxygen species in chromosome breakage was found in human leukocyte cultures treated with DMBA (106). DMBA induced about a 3-fold increase in chromosome breaks and these could be reduced by as much as 64% by the presence of antioxidants during DMBA treatment. The antioxidant effect is corroborated by the observation that peroxidation of unsaturated lipids increases in mouse skin ~3-fold by 20 days after a single application of DMBA (107).
The predominant localization of PAHs in lysosomes provides another opportunity for chromosome aberrations (96). Lysosome membranes are extremely labile to peroxidative damage as indicated by the release of bound enzymes (108). Lysosomal DNase, which has two active sites and can cut both strands of DNA, could be among the enzymes leaked and thereby initiate chromosome breaks (109). It has in fact been shown that cathepsin D protease is released to the cytosol under stressful conditions (110) and nucleases might be similarly released. Other forms of stress increase the expression of an endonuclease associated with DNA breakage (111) which might be another source of chromosome damage by PAHs.
Initiation and promotion of carcinogenesis: a synergistic interaction
The carcinogenic PAHs are known as complete carcinogens because they can induce tumors in the skin without the intervention of other treatments. But as emphasized above, tumor production requires repeated application of the PAHs over extended periods of time for maximum effect. That raises the question of whether there is more than one stage in the tumorigenic process. In the early 1940s, that question was answered positively in studies of skin carcinogenesis in rabbits and mice. Rabbit ears were painted repeatedly with coal tar for several months resulting in the appearance of some papillomas (112). The tarring was then stopped and over the next 8 months many of the papillomas disappeared. Some of them reappeared in <1 month when tarring was resumed. After several rounds of starting and stopping the tarring with disappearance of many of the papillomas, a non-tumorigenic irritant, turpentine, was repeatedly applied and was found to restore some of the tumors and produce new ones. In another experiment a single round of tarring produced one papilloma and was followed by making a large punch hole in the ear (`discing') to create a prolonged period of wound healing with active proliferation of epidermal cells (113). Numerous tumors appeared on the healing surface of the wound of the tarred ears, but none appeared on healing surfaces of a punch hole on the untreated control ears.
The terms initiation and promotion were introduced in an article describing the separate effects of coal tar, B[a]P and MCA on rabbit skin (114). Tar produced tumors much more quickly than did large amounts of B[a]P, which was considered the main carcinogen in tar, though present there in very small amounts. The tar-induced tumors also enlarged much faster than the B[a]P-induced tumors and were more likely to develop carcinomatous characteristics. The tumor-inducing capacity of B[a]P could be greatly enhanced by the wound healing process. MCA was a more active inducer of tumors than B[a]P, but less so than tar. The strong action of tar was attributed to the hyperemic and inflammatory effects produced in the dermis, which enhanced the growth of cells initiated by the carcinogens. The non-specific growth stimulation of `discing' quickly brought out the neoplastic properties of cells which had been permanently altered but not producing tumors despite months of B[a]P treatment. If B[a]P were dissolved in mineral oil instead of benzene, tumor production was further delayed for months (although demonstrable by `discing') and this was correlated with a diminished hyperplastic response to the B[a]P in mineral oil. It was remarked (114) that the inhibitory effect of caloric restriction on the occurrence of spontaneous mammary tumors in mice (115) was probably an effect on their proliferation, not their initiation. The same might be said for the reported inhibition by caloric restriction of skin tumor development induced in mice by B[a]P dissolved in benzene (115). The great enhancement of tumor development by coal tar despite the minute presence of the carcinogens that were known at that time was an important signal that the carcinogenic capacity of complex mixtures could not be judged merely by the known carcinogens present.
The same concepts of initiation and promotion developed independently from studies of skin carcinogenesis in mice, although the terms originally used to describe the processes were different (15). A standard method was developed in which a single subcarcinogenic painting with a PAH was applied to mouse skin as an initiating agent to be followed by repeated applications of croton oil for promotion (18). The application of croton oil could be delayed for as long as 43 weeks after the PAH treatment without loss of tumor production (116). Croton oil itself was very weakly tumorigenic, but applied after the PAH resulted in rapid appearance of many papillomas. In contrast, application of croton oil before the PAH was ineffective at tumor promotion. The initiation step was induced quickly and irreversibly, while promotion required repeated application of croton oil until tumors appeared. Initiation had the characteristics of mutation or chromosomal alteration, while promotion allowed the neoplastic expression of the altered state. Since the combined effect of initiator and promoter is far greater than additive of separate applications, it can be considered synergistic (117).
The active promoting ingredient in croton oil proved to be TPA (118). This compound is a diester of the alcohol phorbol, with one short- and one long-chain fatty acid. Maximum promoting action of the class of phorbol diesters is achieved when the long-chain fatty acid has 14 carbon atoms. TPA can substitute for croton oil in its promotional activities, which are associated with the induction and maintenance of epidermal hyperplasia through the entire period of repeated applications. It was reported that not all agents which induce hyperplasia act as promoters, but that hyperplasia is an essential component of promotion (119). Croton oil, and presumably TPA, does not promote skin tumor development in rabbits, guinea pigs and rats. As noted earlier, wound healing is an effective promoter in rabbits (113), but it was inactive in mice (119) unless preceded by enough carcinogen treatment to be almost adequate for high tumor yield (114). However, extending the area of wounding in repeated abrasion of conventionally initiated mouse skin resulted in ample production of tumors, though not as many as induced by TPA (120). The regenerative hyperplasia for effective promotion must be intense enough to produce substantial epidermal thickening since the mild hyperplasia induced by plucking hair or repeatedly rubbing the skin does not act as a promoter. Acetic acid, which produced a low degree of hyperplasia that is not increased in multiple applications, acts as a weak promoter (121).
Although `membrane active' agents such as sodium lauryl sulfate and Tween 60 are sometimes classified as weak promoters in mice (122), others have found them to offer a wide range of promotion (59). The degree of promotion of these agents was correlated with the degree of hyperplasia they produced in mouse skin. A single application of Tween 60 produced intense cell multiplication and high degree hyperplasia which commenced immediately upon application. Repeated application of Tween 60 maintained the hyperplasia unchanged (Figure 2F). It caused neither the cellular nor nuclear atypia, nor disturbances in cell–cell organization characteristic of the application of carcinogenic PAHs. It was concluded that tumor promotion is comparable to intense, continuous, simple reparative cell multiplication (59). If a single initiating dose of MCA was followed 30 days later by 30 days of repetitive treatment with Tween 60 promoter, there was a pronounced irregular hyperplasia consisting mainly of basal cells with small spindle shaped nuclei at right angles to the dermo–epidermal junction (Figure 2G). The pretreatment with MCA resulted in abnormalities in distribution and arrangement of cells by the promoter (compare Figure 2G with F); promotional treatment of initiated skin also produced much more hyperplasia and irregularity of the epidermis than a single initiating dose of MCA alone (compare Figure 2G with D). The irregular hyperplasia induced in initiated epidermis presumably allows selection of rogue clones and the development of discrete neoplastic lesions.
Twice weekly treatment of uninitiated mouse skin with the strong promoter TPA produced a reactional hyperplasia during the first week characterized by cell damage, edema and acute inflammation in both epidermis and dermis (123). Unlike skin treated with the weak promoter Tween 60, there was nuclear pyknosis, cytoplasmic vacuolization and epidermal degeneration concomitant with epidermal growth (124). This picture changed gradually during the following 3 weeks leading to epidermal hyperplasia with mild chronic inflammation of the dermis and hyperplasia of the hair follicles (123). After TPA treatment stopped, the hyperplasia regressed abruptly during the first 2 weeks. Within 2–4 months after ceasing TPA, the epidermis became thinner than normal, as did the rate of proliferation. It was suggested that the treatment selected for cells particularly sensitive to the stimulating effect of TPA (123) which, in skin pretreated with a carcinogenic PAH, presumably would favor the growth of initiated cells.
From the early days of croton oil use in studies of promotion, occasional occurrence of tumors in mice receiving croton oil alone raised the question of whether it was a weak carcinogen (15). The same question arose again with TPA, especially in the Oslo strain of hairless mice (125). Repetitive painting of these mice twice a week with 17 nmol TPA led to tumors in 28% of the mice in 39 weeks, and similar painting with 10 nmol TPA gave 9% tumors in 55 weeks. Smaller yields of tumors were produced with fewer TPA paintings. An argument against the weak carcinogen status of TPA was that it was merely promoting cells that were spontaneously initiated by endogenous mutations. This argument was widely accepted because TPA is non-mutagenic in hamster cells (126) and does not appear to require metabolic activation for its promoting effect in mouse skin (127). Most human cancers are, however, thought to arise from selection of endogenous mutations (128), so it may be a semantic problem to make a qualitative distinction between carcinogens and strong promoters like TPA. All one can say is that TPA is a weak carcinogen in comparison to some PAHs, and that its activity in the initiation–promotion sequence far outweighs its action alone, or as the sum of the independent actions of TPA and initiators. Indeed, it has been estimated that it requires >10-fold as much PAH acting as a complete carcinogen in repeated applications to produce as many papillomas as PAH plus TPA in an initiation–promotion sequence (R.Boutwell as quoted in ref. 129).
The initiation phase of two-stage carcinogenesis in mouse skin can be replaced by activated V-ras genes of the mouse sarcoma virus (130). Virus was introduced by scarification of the skin and twice weekly painting with TPA was begun either immediately or 4 months later. Papillomas were produced in almost all of the mice treated with the activated viral genes and TPA, but none developed in those receiving only the former. The latent period for papilloma appearance was considerably shorter than for those initiated with DMBA. This was thought to reflect the higher levels of expression of the mutated ras genes driven by viral promoters than are observed in chemically-induced tumors associated with activated ras genes. It had previously been shown that the mere establishment of an activated ras gene is not sufficient for transformation of cultured cells and that cooperation with other transfected genetic elements, such as the early region of adenovirus, is needed to provide complementing biochemical activities (131). Since tumor development in the skin is more difficult to induce than transformation of cells in culture, the requirements of more than one event for transformation indicates that several events are required for papilloma formation.
Two stages of promotion
It was originally thought by some that promotion by TPA was a single process related to the strong and persistent hyperplasia induced by repeated application of the compound after initiation by a carcinogen. This view began to change as a result of the following experiments. Application of croton oil to initiated skin for only 6 weeks resulted in few tumors but extension of the croton oil treatment to 14 weeks completed the promotional process with the appearance of many papillomas (132). Substitution of turpentine for croton oil in the last 8 weeks of the promotional period also resulted in many papillomas, but exclusive use of turpentine during the full 14 week period produced none. The incidence of carcinomas, although later than that of papillomas, followed the same general pattern in response to the three agents. The combined treatment of limited croton oil followed by turpentine was synergistic in the sense that the tumor response was much greater than that engendered by either agent alone. This synergism suggested that there was a qualitative difference between the croton oil and turpentine treatments.
A similar conclusion arose from the substitution of wound healing for turpentine after croton oil treatment of initiated mice (132). Wound healing alone in the initiated mice yielded no papillomas, but many of the mice treated with croton oil for 5 weeks after initiation developed one or more papillomas along the line of an isolated wound. The three-stage sequence of events was described as (i) initiation to a pre-cancer state carried out by a single small dose of a carcinogen which, in the case of PAHs, could carry out all three stages by itself if it were given repeatedly or in large doses, and probably involves an irreversible change in DNA; (ii) conversion, the first part of promotion induced by croton oil which converts the initiated cell into a dormant tumor cell; (iii) propagation, the second part of promotion, induced by turpentine or wound healing, which is dependent on cellular proliferation (132). There is overlap in these stages in the sense that carcinogenic PAHs can induce all three and croton oil can induce the latter two. In fact, croton oil may be able to induce all three since it does produce some tumors in normal skin and turpentine does promote some tumors in initiated skin without croton oil if it is continued beyond 14 weeks. The possibility should be considered that there is a variety of interactions, cellular events and processes at the molecular and biochemical levels that are driven to different extents by all these treatments, which gives the appearance of synergy when favored by the particular experimental setup.
Croton oil is a complex mixture of substances, and when TPA was shown to be the active ingredient, experiments on promotion were refined and extended while eliminating extraneous effects due to other components. Turpentine also is not a pure chemical, so better defined substances were tested for their ability to carry out the second stage of progression. Mezerein is a diterpene similar to TPA which brings about most of the biochemical changes and the hyperplasia seen with TPA but is >1/50th as active as a promoter (133). Treatment of DMBA-initiated mouse skins with TPA for 20 weeks yielded a large number of papillomas, but treatment for 2 weeks yielded none at 20 weeks (133). If a high dose of mezerein was continuously applied after 2 weeks of TPA, almost as many papillomas appeared by 20 weeks as in continuous treatment with TPA. Mezerein by itself was only a very weak promoter. Hence, mezerein was acting chiefly as a propagator or a second stage promoter.
The weak promoting activity of mezerein by itself complicates the stage-wise analysis of promotion, so a second stage promoter or proliferegen devoid of full promotion activity was sought by substituting a retinoic acid residue for the long fatty acid chain of TPA in synthesizing phorbol-12-retinoate-13-acetate (or RPA) (134). RPA provokes epidermal hyperplasia as effectively as an equimolar dose of TPA, but seemed in early experiments to have no promoting activity whatsoever. Its hyperplastic activity was partially inhibited by the cyclooxygenase inhibitor indomethacin, and the inhibition was reversed by prostaglandin E2. It is a far more effective second stage promoter than mezerein when applied twice weekly after 2 weeks of TPA treatment of DMBA-initiated mouse skin.
The stability of the first converting stage of promotion was tested by one or two applications of TPA to initiated mouse skin followed by twice weekly applications of RPA beginning 1 or 2 weeks after TPA treatment (135). RPA retained some promotional capacity even when its repeated application was delayed up to 8 weeks after the brief TPA treatment. The decay of first stage promotion was later found to have a half-life of 8–10 weeks (136). The relative stability of conversion, the first stage of promotion, raised the question whether it could be established even before the initiation step was taken. To avoid complications from short-term TPA effects, such as hyperplasia and inflammation, which disappear within a few weeks of ceasing TPA application, initiation by DMBA was delayed for 2–6 weeks after TPA application. Application of TPA at 2 and 6 weeks before DMBA which was followed by long-term RPA application resulted in production of papillomas to almost the same extent as the conventional TPA application after initiation (136). It should be noted that in this set of experiments, RPA itself had some limited ability to induce the first stage of promotion without TPA. This suggests a degree of overlap between the two stages of promotion, just as there appeared to be between initiation and promotion by croton oil or TPA because of their weak carcinogenic activity (117).
Chromosomal effects produced by promoters
Although the strong promoter TPA does not interact directly with DNA and is not mutagenic (126), a single application does cause DNA strand breaks in human leukocytes (137). The DNA damage in the leukocytes was apparently related to a sharp increase in reactive O2 production that occurred shortly after the addition of TPA, since an inhibitor of reactive O2 production prevented DNA damage. Enzymes that remove reactive O2 species had a pronounced inhibitory effect on the TPA-induced DNA damage.
Numerical and structural aberrations of chromosomes were produced by TPA treatment of mouse keratinocyte cultures (138). There was an increase of gaps and chromatid breaks accompanied by intra- and inter-chromosomal exchanges. Aberrations were visible within 24 h (one cell cycle) and increased with longer and multiple exposure. They persisted for several days after removal of TPA and further cultivation in fresh medium. Brief treatment with TPA produced double minute chromosomes which are the cytogenetic equivalent of gene amplification (139). The `non-convertogenic' tumor promoter RPA and non-promoting phorbol esters caused no substantial chromosome aberrations. Treatments which inhibit tumor induction by TPA in initiated skin also inhibited the chromosome aberrations, indicating a possible causal relationship between them and the process of conversion, or first stage promotion.
Further indication of a primary role of chromosome aberrations in conversion came from experiments with the well-known alkylating and clastogenic agent methyl methane-sulfonate (MMS) (140). MMS induced chromosome breaks and gaps in epidermal cells when topically applied to mouse skin but did not exhibit any initiating capacity. It did prove to be a rather powerful agent of conversion or stage I tumor promotion. In contrast to TPA though, MMS is a weak inducer of DNA synthesis, which is necessary for conversion (141), so it had to be accompanied by RPA in order to exert a converting effect comparable with that of TPA (140). The re