Mutagenesis Advance Access originally published online on March 22, 2005
Mutagenesis 2005 20(2):81-92; doi:10.1093/mutage/gei017
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
COMMENTARY |
The emperor wears no clothes in the field of carcinogen risk assessment: ignored concepts in cancer risk assessment
National Food Safety Toxicology Center, Department of Pediatrics and Human Development, Michigan State University, East Lansing, MI 48824, USA
 |
Abstract |
|---|
Top
Abstract
Introduction: ‘Cancer is...The following is a position paper challenging the paradigm that ‘carcinogen = mutagen’, and that the current rodent bioassay to predict risks to human cancers is relevant and useful. Specifically, we review current observations concerning carcinogenesis that might lead to another approach for assessing the identification of human carcinogenic hazards and the risk assessment that chemicals might pose. We give a brief review of the multistage and multimechanism process of cancer in a tissue that involves not only genotoxic but also epigenetic events, and the importance of stem and progenitor cells in the development of cancer. We focus on the often ignored ‘epigenetic’ effects of carcinogens and the role of cell communication systems in epigenetically altering gene expression that leads to an imbalance of cell proliferation, differentiation and apoptosis in a tissue that can contribute to the cancer process. To draw attention to the fact that the current paradigm and policy to test toxic chemicals is often misleading and incorrect, we discuss how oxidative stress, in spite of the DNA damaging data, most probably contributes to cancer at the epigenetic level. Additionally, we briefly review how this mutagenic concept has greatly diverted attention away from doing research on the lower molecular weight, non-genotoxic, polycyclic aromatic hydrocarbons (PAHs), and how these low molecular weight PAHs are etiologically more relevant to the disease potential of environmental mixtures such as cigarette smoke.
|
Introduction: ‘Cancer is not a disease of a single cell’—limitations of a reductionalist's view |
|---|
‘...The cancer problem is not merely a cell problem, it is a problem of cell interaction, not only within tissues, but also with distal cells in other tissues. But in stressing the whole organism, we must also remember that the integration of normal cells with the welfare of the whole organisms is brought about entirely by molecular messages acting on molecular receptors’ (1
). The goal of this rather iconoclastic challenge is to draw attention to the current paradigm and policy to test toxic chemicals that are correlated with cancer. Specifically, most still believe that cancer-causing chemicals are linked to DNA damage and mutations and that the current rodent bioassay to predict hazards and assess risks to human cancers is relevant and useful. While it could be argued that this ‘carcinogen = mutagen’ paradigm has long been challenged, one needs only to examine the huge number of papers that are still published where this paradigm still drives the use of in vitro ‘genotoxicity’ assays that are misinterpreted, and where both government regulatory agencies and pharmaceutical/industrial labs still require the use of these questionable assays. Specifically, we wish to review current observations concerning carcinogenesis that might lead to another approach for assessing the human carcinogenic risk that chemicals might pose.
Along with Potter's insight and the old adage, ‘The whole is always greater than the sum of its parts’, any best approach for assessing the human carcinogenic risk after chemical exposure will be confronted by no easy solutions. To begin with, the paradigm, ‘carcinogens as mutagens’ (2), illustrates the point we wish to make, namely, that carcinogenesis is ‘more’ than mutagenesis! What would that ‘more’ be? Clearly, when a cell (an adult stem cell; a progenitor cell or a terminally differentiated cell) is exposed to any chemical, there is a possibility of DNA damage and mutations in either or both nuclear and mitochondrial DNA. In addition, cell death by either necrosis or apoptosis could be the result. Finally, epigenetic alterations could ensue which, then, could alter the cell's commitment to cell division, cell differentiation, apoptosis, adaptive responses of differentiated cells or senescence.
In the recent past, the term, ‘epigenetic mechanism’, was a vague, non-universally accepted term. However, molecular demonstrations have now shown that genes in cancer cells can be abnormally expressed or that certain toxic/carcinogenic chemicals, without DNA-damaging capacity or mutagenic potential, can alter the patterns of gene expression by modifications of methylation and acetylation of DNA and histones (3). Almost without reference to the introduction of DNA micro-array technology, which measures altered gene expression, much attention has been unconsciously focused on the role of epigenetic changes occurring during the carcinogenic process. It is as though a kind of risk assessment ‘schizophrenia’ occurs in that the combination of using so-called in vitro genotoxicity assays, rodent bioassays, detection of DNA lesions and mutations in cells of tumors and the monitoring of altered gene expression, using powerful DNA micro-array technologies, co-exist without critical examinations of what is being measured. What does it all mean and what does it do to the prevailing paradigm shaping the risk assessment of potentially cancer-causing chemicals?
At the same time that all these cancer-associated chemicals presumably inducing lesions in DNA of various cell types in tissues of exposed animals, they induce intracellular signaling and alter gap junctional intercellular communication (GJIC) (4). The broadest definition of an ‘epigenetic’ change induced in a cell is that which alters the expression of the information of the genome at the transcriptional, translational or post-translational levels. It could occur as a heritable transcriptional change in proliferating cells or as a change in the expression of a stem cell to terminally differentiate or apoptose. However, altered gene expression is usually preceded by changes at the cell signaling level that governs transcriptional changes, as well as alterations in cell–cell signaling within tissues. While epigenetic alterations are sometimes defined as ‘inherited’ non-mutagenic alterations found in the expression of genes, it must be recalled that in vivo, chemicals affect the few stem cells, the progenitor and terminally differentiated cells. Stem cells induced to differentiate or apoptose by toxic chemicals, rather than proliferate, do so by epigenetic mechanisms when the cells do not proliferate. Moreover, when a terminally differentiated cell responds to a potentially toxic chemical, it might express ‘stress’ genes. In addition, proliferating cells in Go phase can be induced to alter its gene expression transiently, to express genes for initiating cell cycle entrance but which return to the original transcription state after proliferation. Therefore, we chose to include the broader definition of epigenetic changes, which can occur in both proliferating and non-proliferating cells. The fact is that many chemicals can contribute to the carcinogenic process without inducing mutations or cell necrotic death. DDT, phenobarbital, saccharin, peroxisome proliferators, etc. are all experimentally known to contribute to the carcinogenic process in rodents.
Therefore, the real challenge is: ‘How can one assess the epigenetic potential of a chemical's contribution to human cancer when it must be measured either in vitro or in experimental animals?’ No 2D in vitro assay, using either normal, primary rodent or human cells, or any immortalized normal or cancer cell line, can mimic the in vivo, human in situ condition of complex interactions between stem cells, proliferating progenitor cells and terminally differentiated cells within a tissue and between tissues (i.e. stromal–epithelial interactions) (5). Moreover, historic evidence has demonstrated the limitations of using experimental animals (6) and epidemiological studies for a number of obvious reasons.
The fundamental issues raised by this challenge seem to include: (i) carcinogenesis in animals and human beings is a multistage, multimechanism process (i.e. the initiation/promotion/progress model) (7); (ii) while experimental animals and human beings are fundamentally and biologically alike, in principle, they differ in many ways that could affect any one or all those stages; (iii) in both animals and human beings, individual genetic-, developmental stage-, gender-, dietary-, life style- and environmental-factors can, and do, influence each of these stages; (iv) since the experimental animal and human being consist of a hierarchical/cybernetic organization of multicell types, each interacting with cells within and between tissues (8), use of cell-free, molecular targets, and 2D cell cultures, using either ‘pure’ animal or human primary or immortalized/tumor cells, will never mimic the in vivo situation (9). This is not to suggest that the use of molecular/biochemical or 2D cell cultures of primary or immortalized rodent or human cells will not generate a much needed mechanistic understanding of the carcinogenic process, as has experimental rodent carcinogenesis studies, particularly using the initiation/promotion/progress models, as well as transgenic and specific gene knock-outs/knock-ins; (v) delineation of mechanisms of each stage of carcinogenesis will be needed, particularly the potential of thresholds for epigenetic agents; (vi) the study of the in vitro mechanisms of carcinogenesis that might be identical to the carcinogenesis process in human beings; and (vii) the role of mixtures of carcinogenic factors (both endogenous and exogenous) that could influence any of the three steps of carcinogenesis to be additive, synergistic or antagonistic. As wonderful as these approaches have been and will continue to be, more will be needed.
The irony of this analysis is that, while we argue that current in vitro assays to measure mutagenesis, cytotoxicity or even ‘epigenetic’ potential of chemicals are not adequate for accurate extrapolation to humans, the use of current animal bioassays will also fall far short of adequacy. Since space will not permit an attempt for rationalizing the use of 3D, human adult stem cells (9) to attempt to breach this impossible task, the bottom line is that the use of human stem cells is where we believe that this field must aim for.
|
Limitations of animal bioassays |
|---|
While not wedded to using animal bioassays for the pre-testing of new chemicals for their potential risk in contributing to human carcinogenesis, one has to remember that today's bioassay protocol assumes that the induction of cancer in rodents, after an exposure to sublethal concentrations for the lifetime of the animal, generates a meaningful risk extrapolation information to human beings. However, such assumptions are overly simplistic. Since carcinogenesis is a multistage, multimechanism process, one would have to either assume that (i) the high dose-exposure of the single chemical being tested can induce all the mechanisms by itself; or (ii) the high dose of the chemical upsets the normal homeostatic control of the physiological state of the animal, which, in turn, affects one aspect of the multistage process.
Given that the cells of a tumor appear to be monoclonally derived from a single cell (10), and in spite of their heterogeneity and genomic instability, and given that ‘initiators’ seem to induce a stable, irreversible event, both mutagenesis (a irreversible change in the qualitative or quantitative nature of the genome) and a stable ‘epigenetic’ mechanism (an altered expression of a gene) can contribute to the initiation phase of carcinogenesis. Short of directly testing changes in DNA sequences after exposure to an agent that is or could be an initiator, the problem is that the use of short-term in vitro assays to detect phenotypic changes, which are used as surrogates for mutations, will always generate too many false positives (11,12). Even measuring mutations in the cells of tumors created in rodents after exposure to a carcinogenic agent could be interpreted incorrectly if the agent acted to select a spontaneously generated mutation in an oncogene or tumor suppressor gene (13).
Agents that can promote tumors appear to act ‘epigenetically’, i.e. the process of promotion can be interrupted; appear to have threshold levels of action; and can be ameliorated by antipromoters (14). In addition to being species-, gender- and organ-specific, including developmental stage-specificity, promoting agents can act to select, clonally, the initiated cell by mitogenesis and/or blockage of apoptosis. Irritation or cell death or cell removal by surgery, can also act as a promoting condition of the initiated cell (14).
Those agents, that, after the terminal stages of carcinogenesis, impart invasive, metastatic and angiogenic properties (progression phase of carcinogenesis), seem to do so in a stable manner. Both irreversible mutagenic and stable epigenetic events could contribute to this phase.
|
Blinded by the ‘carcinogen as mutagen’ paradigm, epigenetic mechanisms have been ignored! |
|---|
As these reviewers look back at this complex problem, we see where Hanahan and Weinberg's comment is extremely relevant to the challenge: ‘those researching the cancer problem will be practicing a dramatically different type of science than we have experienced over the past 25 years. Surely much of this change will be apparent at the technical level. But ultimately, the more fundamental change will be conceptual’ (15
). We think the major paradigm that has paralyzed any reasonable solution to understand the tens of thousands of published cancer studies is the paradigm ‘carcinogens as mutagens’ (2). It is the result of the classic ‘lamp post effect’. We look for the lost keys under the lamp post because that is where there is light. Based on the fact that (i) mutations (gene and chromosomal) are found in cancers in animals exposed to chemicals (ii) these chemicals are sometimes tested positive in the accepted and easy to run, but highly misleading, in vitro assays to detect ‘mutations’ (using phenotypic markers as surrogates for changes in DNA sequences) or (iii) predispositions to cancers do occur in humans that inherit germ-line mutations, it was assumed that mutagenesis was the ‘end all, be all’ for carcinogenesis. It goes without much examination that, by not using animal bioassays in an initiation/promotion/progression protocol, and by not testing chemicals at the extreme limits of their toxic tolerance, which is unrealistic for real-life human exposures, no realistic mechanistic understanding can be derived for use in any risk assessment model. (Here we assume the best cancer risk assessment model must include mechanistic data.) The high-dose protocol used to test potential chemical carcinogens, and misclassification of the presumed mechanism of carcinogen action of the chemical used, based on in vitro genotoxicity assays, which, themselves, are chock-full of limitations and artifacts, cannot lead to any concise interpretation of the underlying mechanism of action. Viewing the use of insensitive in vitro assays used to understand the equivalent insensitive rodent bioassays can be likened to the blind leading the blind. |
Promoters as inducers of oxidative stress do not make them ‘genotoxic’ |
|---|
Even when the promotion process appeared to involve a mitogenic or inhibited apoptotic process to expand the initiated cell (assumed to be induced by a chemical mutagenic/carcinogen, to ‘save’ the paradigm), investigators claimed that mutations needed to be fixed during the promotion process. While this hypothesis is, in part, true, it does not explain the mechanisms of agents that stimulate mitogenesis or inhibit apoptosis, both biological processes involved in the promotion phase (16
). As Weinstein has pointed out, ‘...extensive proliferation occurs during normal fetal and embryonic development, as well as in the continuous renewal in the adult of the entire skin, gastrointestinal epithelium and bone marrow... Thus, excessive cell proliferation per se is probably not the exclusive causative factor in human breast cancer’ (17). Mutagens usually induce necrosis and stop cell proliferation. After his studies on mutations in human tissues, Thilly (13) concluded that ‘direct measurements of kinds and numbers of point mutations have found no clear relationship with exposure to environmental agents, save for sunlight in the skin’. He goes on to offer an alternative explanation of those environmental chemicals, classified as ‘mutagens’ from short-term tests for genotoxicity, wherein they select pre-existing cells with mutations rather than induce oncomutations. This supports the hypothesis previously suggesting that most chemical ‘carcinogens’, tested in animal bioassays and in in vitro genotoxicity assays, are epigenetic agents which are tumor promoters, not ‘epigenetic’ carcinogens (14). Spontaneously initiated cells, induced by either errors in repair of endogenous or exogenous agents or by errors in replication of normal DNA templates, are probably those that are promoted by the so-called ‘carcinogenic mutagens’. In order that it might not be missed, we suggest that chemicals, associated with carcinogenesis, might indeed induce DNA damage at high concentrations in certain kinds of cells, which probably die or will never divide (‘dead cells do not give rise to cancers’), and which might induce DNA lesions in mitochondrial DNA, are probably ‘epigenetic’ in nature, not mutagenic or genotoxic. If these concentrations are needed in vitro to induce DNA lesions and the recovery of phenotypic changes called mutations, the same concentrations in vivo would probably induce acute tissue damage, if not the death of the organism. To demonstrate that electrophiles of metabolized chemicals can interact with naked DNA or finding DNA lesions in extracted DNA from animal organs containing cells, most of which are probably not the target cells for cancer, do not constitute rigorous proof of the chemical's mechanism of action as an animal carcinogen. In other words, it would be highly unlikely that these so-called genotoxic chemical carcinogens do contribute to carcinogenesis as the mutagens of the mutations found in the oncogenes or tumor suppressor genes of the tumor found in the chemically treated animal. The latter statement should justify the claim that this review should be rather ‘iconoclastic’ and a ‘paradigm-buster’.
Cha et al. (18) also reported that N-nitroso-N-methylurea-induced rat mammary tumors arise from cells with pre-existing oncogenic Hras1 gene mutations. While this and other similar studies have been swept under the regulatory rug, it also demonstrates that a chemical, interpreted as a mutagen through in vitro assays and shown to be a carcinogen in a bioassay, was assumed to induce mutations found in oncogenes in the tumors of the chemically exposed animal. The chemical was most likely an epigenetic agent that promoted a pre-existing spontaneously initiated cell. Even several in vitro transformation assay studies, which again seem to be ignored, have provided data that challenge the prevailing paradigm that a chemical, which induces rodent transformation in vitro, must be a mutagen. Brookes et al. (19) and Mass and Austin (20) showed that 7,12-dimethylbenz[a]anthracene (DMBA), the ‘quintessential’ chemical mutagenic/carcinogen, did not mutate the Ki-ras and Ha-ras oncogene of the DMBA-transformed cells.
Apparently, many chronic diseases affiliated with oxidative stress, such as cancer, are not always a consequence of tissue necrosis, DNA mutations, or protein damage but rather, owing to an altered gene expression through epigenetic mechanisms (16,21). Another example that oxidative reactions do not contribute primarily to the genotoxic, initiating phase of cancer, are the results of rodents treated with organic and hydrogen peroxides in the two–stage cancer model systems, in which these compounds exhibited tumor promoting and not initiating activity (22–25), indicating that these oxidants are not mutagens but rather epigenetic toxicants. In the past 15 years, considerable research in oxidative stress has shifted from understanding how oxidations lead to macromolecular damage, to comprehending how reactive oxygen species (ROS) reversibly control the expression of genes at non-cytotoxic doses (26). In this respect, at least 127 genes and signal transducing proteins have been reported to be sensitive to reductive and oxidative (redox) states in the cell (26).
Although many intracellular signaling pathways are known to be redox-sensitive, the most studied signal transduction factors are mitogen-activated protein kinases (MAPK) and nuclear factor-kB (NF-kB) (26–28). These two pathways either directly or indirectly transduce most redox responses (26). MAPK is not only activated by ROS (29) but actually requires the presence of endogenously produced H2O2 (30). This is one of several studies demonstrating that endogenous growth factors (extracellular ligands) generate ROS, which are then required downstream in intracellular signaling to successfully transmit their signals to the nucleus (31). As mentioned above, the successful transmission of an extracellular signal from the membrane to the nucleus through intracellular signaling pathways in solid tissue cell types is also dependent upon intercellular signals through gap junctions (16,21). Not surprisingly, ROS have also been demonstrated to reversibly inhibit GJIC at non-cytotoxic levels (32). If gap junctions were not closed, then the H2O2 generated by extracellular ligands could escape through gap junctions into neighboring cells, thereby potentially diluting to a subthreshold level that would be insufficient for MAPK-dependent activation of transcription factors. These examples demonstrate how extra-, intra- and inter-cellular signaling pathways might interact to coordinate the epigenetic expression of genes in response to ROS.
Antioxidants have also been demonstrated to serve as subcellular messengers for normal cell function (26). For example, a major H2O2-scavenging pathway is the two-electron reduction of H2O2 catalyzed by glutathione peroxidase, which clearly serves as a protective role against peroxide-dependent oxidative injury. However, depletion of intracellular pools of glutathione (GSH), by inhibiting the rate-limiting step of its biosynthesis, paradoxically reverses the biological effect of H2O2 in several systems. For example, inhibition of GJIC (32), induction of c-jun (33) and activation of NF-kB (34) by H2O2 was completely reversed when the cellular systems were depleted of GSH, which indicates that these signaling pathways not only required H2O2 but also GSH. Inhibition of GJIC and the induction of early-response genes are hallmarks of tumor promotion and in the results just described, a reduction in the natural antioxidant GSH could also potentially protect a cell from proliferative responses to extracellular ligands. Apparently, assessment of the true risk that oxidative stress poses to human health will need to move beyond the genotoxic, DNA-damaging paradigm, and incorporate an understanding of how oxidative reactions contribute to the epigenetic expression of genes.
|
Polycyclic aromatic hydrocarbons: the mutagenic concept falls short of a comprehensive understanding of carcinogenic potential |
|---|
In many cases, it has been argued that tumor-promoting chemicals can induce oxidative stress. Ergo, by definition by some, they are ‘mutagens’, yet when tested in in vitro assays, they prove not to be DNA damaging agents or mutagens (TCDD, TPA, DDT, phenobarbital, peroxisome proliferators, etc.). An excellent example is pentachlorophenol, which has been shown to induce oxidative stress (35
), to be a tumor promoter and not an initiator in mouse liver (36), and can inhibit GJIC at non-mutagenic, non-cytotoxic conditions, both in vitro and in vivo (37). Another example is chemicals from cigarette smoke, long been thought to be primarily a carcinogenic mutagen, because of the identification of high molecular weight (five- to seven-ringed), mutagenic polycyclic aromatic hydrocarbons (PAHs) such as benzo[a]pyrene. Yet, the most abundant PAHs in cigarette smoke are the low molecular weight fractions, particularly the three- and four-ringed PAHs in which the methylated anthracenes and phenanthrenes are 62 times higher than benzo[a]pyrene and benzo[e]pyrene (Table I) (38).
|
Considering that the fraction of cigarette smoke containing the three- and four-ringed PAHs is highly co-carcinogenic when applied to the skin of mice treated with benzo[a]pyrene (Figure 1) (39
), and that cigarette smoke is a strong promoter and a weak complete carcinogen (40–43), suggest that this fraction could significantly contribute to cancer. Ten to fifteen years after giving up smoking, the ex-smoker faces the same low risk of developing cancer of the upper digestive tract, the lung, the pancreas and the urinary tract as the non-smoker (Figure 2) (44). This fact strongly suggests that cigarette smoke contributes to the non-genotoxic and reversible phases of cancer. Contribution of cigarette smoke to the non-genotoxic phase of cancer is particularly important. Early work on the carcinogenicity of cigarette smoke condensates strongly indicated that the neutral fractions, which contained primarily PAHs, were the most carcinogenic fractions, but that the concentrations of the most prominent complete carcinogens, i.e. benzo[a]pyrene, was far too low to account, by themselves, for the carcinogenic activity of the condensates (40,45). Furthermore, the much less studied area of co-carcinogenesis is a closer fit to the extended exposure of human smoking than the complete carcinogenic nature of selected PAHs from cigarette smoke condensates (40). Therefore, understanding the biological effects of these three- and four-ringed PAHs, which are prevalent in cigarette smoke and possess co-carcinogenic activity, on cell signaling pathways relevant to the epigenetic, non-genotoxic phase of cancer is important. These three- and four-ringed PAHs are not only the most prevalent PAHs in cigarette smoke but also in many environmental systems and food products.
|
|
To date, we have successfully demonstrated that many PAHs inhibit GJIC in pluripotent mammalian epithelial cells (46
–49). The most significant results indicated that three- and four-ring PAHs containing bay regions or bay-like regions are far more active in inhibiting GJIC than the isomers that do not have these structural motifs. We would like to note that the term ‘bay region’ refers to the pocket formed by the stereically hindered region created by an angular benzo ring and, similarly, the term ‘bay-like’ region is used to describe the angular pocket formed at the top of the benzene ring by a methyl- or chloro-group (Figure 3). These bay regions are not in reference to those formed by the well-known DNA reactive diolepoxides but rather refer to the unmetabolized parent structure. Our focus was on the epigenetic effects of PAHs, which does not require the metabolic activation of a PAH to a chemically more reactive electrophilic compound. The epigenetic effects of PAHs, in particular, are poorly understood and the toxic risks of unmetabolized PAHs need to be reassessed at this non-genotoxic level.
|
More specifically, we demonstrated that the 1- and 9-methyl- or 1- and 9-chloroanthracene or 1,9-dimethylanthracene (1-meA, 9-meA, or 1-Cl-A, 9-Cl-A or DMA), which form bay-like regions inhibit GJIC as well as phenanthrene that contains a bay region, whereas the 2-methyl or 2-chloroanthracene (2-meA, 2-Cl-A), which have linear configurations, do not inhibit GJIC (50
). The same exact pattern of GJIC inhibition with respect to dose and time that was observed with the chlorinated-anthracene isomers compared with the non-chlorinated PAH isomers suggests that the structure is more important than the chemical reactivity of these compounds, considering that the chlorine substitution converts these PAHs from a neutral to a more nucleophilic compound. The methyl versus chloroanthracenes is chemically very different but structurally quite similar, suggesting a common type of receptor. Moreover, multiple bay-like regions do not significantly increase the potency of the compound to inhibit GJIC when compared with a compound that contains a single bay-like region. For example, 9-methylanthracene and 9,10-dimethylanthracene possesses multiple bay-like regions, while 1-methylanthracene possesses only one bay-like region; yet, these compounds all had similar dose–response curves. Inhibition of GJIC by these PAHs was also a reversible process, which is consistent with the reversible nature of tumor promotion in vivo. Inhibition occurred in a short time period for all the PAHs, indicating that the gap junctions are being modified at the post-translational level.
Although inhibition of GJIC may contribute to the mitogenic events of a promoter by removing an initiated cell from growth suppression, other epigenetic events such as the actual activation of a mitogenic-signaling pathway are also required. We have published results showing that GJIC-active (inhibitory) PAHs activate MAPK signaling pathways, while the GJIC-inactive PAHs do not induce MAPK (50). The kinetic results also indicate that GJIC activity was affected before MAPK induction (a difference of 5 versus 15 min). These results are also consistent with the hypothesis that a quiescent cell must first be removed from growth suppression through inhibition of GJIC and at the same time or before the onset of mitogenic events.
The significances of these results are: (i) the three- and four-ringed PAHs, which have been determined to be non-mutagenic, are biologically active by altering cell signaling that favors proliferation; (ii) ignoring these molecular signaling events greatly underestimates the potential risk of these compounds to human health, particularly cancer; (iii) they are more consistent with the in vivo results of rodent systems exposed to tobacco smoke, indicating that combusted tobacco mixtures are strong tumor promoters but very weak initiators and complete carcinogens; (iv) they are more consistent with the reversible nature of tobacco smoke as indicated by the drastic decrease in cancer risk after cessation of smoking; (v) they offer novel molecular targets for the design of new chemopreventative and therapeutic strategies. Traditionally, the risk of these lower molecular weight PAHs to human health have been ignored precisely because of their non-genotoxic properties, and is a classic case of how the ‘carcinogens = mutagens’ paradigm has greatly contributed to the underestimation of the true risk that environmental and food borne compounds pose to human health.
|
The need for a mechanistic, biological cancer-risk assessment model |
|---|
One major implication of knowing the mechanism of action of a potential chemical (Is it a mutagen, cytotoxicant or epigenetic toxicant?) is for epidemiological studies and risk analyses. Ignoring for the moment the contentious debate on ‘chemical mutagenic–carcinogenic initiators (e.g. Do they exist?; Are there threshold levels of exposure?; Do they cause the mutations in oncogenes and tumor suppressor genes in tumors of animals exposed to the chemical’, etc.?), cytotoxicants can lead to cancers by their ability to induce compensatory hyperplasia, an indirect tumor promoting stimuli for surviving initiated cells. More importantly, it is the measured opinion of these commentators that most chemicals associated with tumors in exposed animals or human populations are epigenetic in character, by selectively cloning out of the target tissue pre-existing initiated cells. As tumor promoters, these chemicals do work at threshold levels (51
) and must be given in a regular, sustained and chronic fashion in the absence of antitumor promoters. Irregular exposures, exposures at subthreshold levels, or exposures that are interrupted (as stopping cigarette smoking) or are simultaneously present with antitumor promoters, will modify the predicted outcomes. Even knowing the mechanism of action, but without understanding the characteristics of the epigenetic agents, might lead to erroneous predictions of risk. E. Franco et al. (98) have implied that, in spite of its successful track record, epidemiology as a discipline has become the focus of considerable controversy concerning its usefulness and limitations to identify cancer-causing exposures. This is because few cancer epidemiological studies incorporate the multistage, multimechanism concept of carcinogenesis in the design, execution and interpretation of their studies. The studies that have done so (52,53), are yet to incorporate the newer ‘ignored hallmarks’ of cancer, such as the stem cell as the ‘target cells’ of cancer and the complex set of interactions to which Potter refers (54). Even the term, ‘molecular epidemiology’ seems to miss the point that Potter makes in his quote. The DNA sequence alone will never predict with certainty the risk of a cancer in an individual in a population exposed to a chemical. |
‘Good news’–‘bad news’ as the conundrum of chemical carcinogenesis |
|---|
To these commentators, the ‘emperor wears no clothes in the field of chemical carcinogenesis’ when it comes to the selective view of using chemical carcinogen data to support one's assessment of the dangers for a particular chemical's ability to ‘cause’ cancer (55
). It has been noted for some time that many chemicals, while being associated with the induction of cancers in animal bioassays, can, at the same time, in the same animal, protect against other cancers or demonstrate the limitations of the animal bioassays as adequate predictors of potential human carcinogenesis (56). Even more striking is the observation that the same chemical, in this case, phenobarbital, a valuable human anti-epileptic drug, can promote liver cancers in rodents if given after the initiation of a post-weaned rat, but it suppresses liver tumors if given to an initiated pre-weaned rat (57). In addition, the tumors of these differentially treated rats were, themselves, different histologically. This suggests that the target cells for the initiation stage of the pre- and post-treated rats were, themselves, different.
Phenobarbital appears to act as a promoter when it inhibits gap junction intercellular communication (58). The liver tumors of the pre-weaned, initiated rat were ‘embryonic-like’ (basophilic), suggesting that they might have originated from liver stem cells, which do not seem to have expressed connexin genes or functional gap junctions, similar to other adult stem cells (54). Therefore, these pre-weaned rat tumors would not be promoted by phenobarbital, since they have no gap junctions. Agents that cause mitogenesis of stem cells probably stimulated the promotion of these basophilic tumors. It seems to be a fact that tumors can be characterized by a lack of functional heterologous or homologous GJIC (54). In fact, the classic HeLa and MCF-7 cancer cell lines are cells that do not express their connexin genes (59,60), whereas many other cancer cells do have expressed connexins that are rendered non-functional because of some expressed oncogene.
Since phenobarbital was used as the example, it is interesting to note that, had regulations been in effect during the drug development and safety evaluation of this drug, it might never have been given human use approval because it is one of the classic promoters for rodent liver tumors. Phenobarbital can ‘induce’ liver tumors in non-‘initiated’ animals at high doses and with chronic treatment. After decades of human use, there has been no epidemiological evidence of liver or head/oral cancers in human beings.
|
Stem cells and cell communication as ignored hallmarks of cancer |
|---|
Guided by the insight of V.R.Potter in the aforementioned quote, it should be clear that carcinogenesis occurs in a complex in vivo environment where the single ‘target’ cell exists in a tissue where there are cell–matrix interactions, stromal–epithelial interactions (5
) through soluble extracellular signals between a few adult stem cells, their finite-life, progenitor daughter cells, and their lineage-N-stage, terminally differentiated cells and any invasive, inflammatory-related cells. Consequently, any cell-free, in vitro reductionalistic approach to study mechanisms of carcinogenesis must carefully re-integrate components before any extrapolation to the in vivo human situation. First, and foremost, resolution must be reached on the classic problem, ‘Is the "target" cell for initiation of the carcinogenic process a stem cell (61) or any differentiated cell that can "de-differentiate"?’ (62). Second, the understanding of the complex integration of extra-, intra- and gap junctional intercellular communication in the target organ (Figure 4) must be accomplished before we can truly find a biologically based risk assessment from non-human systems to individual human beings.
|
While the problem is yet to be resolved, the list of cancer ‘hallmarks’ (11
) includes the phenotype of ‘immortalization’ as an early characteristic of all cancer cells. This implies a resistance to apoptosis and blockage of terminal differentiation, which are, also, ‘hallmarks’ of cancer cells. Since stem cells are usually defined as cells that are naturally ‘immortal’ until they are induced to terminally differentiate or to ‘mortalize’, it is observed that one explanation to support the stem cell theory of cancer is by ‘initiating’ an adult, immortal stem cell wherein the initiated stem cell stays immortal, rather than being induced to become ‘immortal’, a prediction of the ‘de-differentiation’ hypothesis. Recent observations have shown that immortal, normal human breast epithelial stem cells can be prevented from terminally differentiating by SV40 large-T transfection as well as preventing neoplastic transformation (61). The discovery of ‘cancer stem’ cells (63), while not rigorously excluding the ‘de-differentiation’ hypothesis, is more easily explained using Ockham's razor, by the stem cell hypothesis.
Recently, the well-studied human carcinogen, benzene, which most, including the authors of this study, have thought was a mutagen (directly or via its metabolites) has been shown to differentially induce the cell death of human bone marrow CD34 hematopoietic progenitor cells (64). Yet, Reddy et al. (99) had already shown that benzene-treated mice did not exhibit any DNA adduct formation in the bone marrow. In fact, in the previous study, apoptosis was observed, suggesting a benzene-induced signal transduction-altered induction of the apoptosis signaling-gene system.
|
The best approach to be used |
|---|
Assuming the aforementioned concepts that have been used to criticize our past and current approach to evaluate the risk to cancer after chemical exposure suggests the ‘standard animal’ bioassay, and along the current short-term tests for ‘mutagenicity’, are, in large part, misinterpreted, we would recommend that their use in the future be discontinued. A recent series of invited papers has focused on assessing the current protocols used to determine the carcinogenic risk of chemicals (65
). In the interim period before the development of a more human-relevant assay, any chemical that might need testing should be tested as having ‘epigenetic’ potential in vitro or in vivo. As a ‘carcinogen’, it should be tested as a tumor promoter. This approach might be done using either (i) the classic initiation/promotion models in rodent systems (possibly using only ionizing radiation as initiators of internal organs or UV light for skin initiation, followed by the test chemical) or (ii) the connexin32-knockout mouse (66,67). In the latter, the KO Cx32 is already a ‘constitutive promoter’ of the mouse liver. Therefore, by testing new chemicals as a potential ‘initiator’, one does not need to follow the initiation with a known promoter (68). Testing the new chemical as a promoter in this mouse after it has been initiated with ionizing radiation should not enhance the tumor frequency over that of the initiated-only animals. However, it should be noted that even this system is not free from potential misinterpretation, in that these animals are highly susceptible to chemical carcinogenesis. If chemicals are not mutagenic, then how can these chemicals increase the initiation process? To be consistent with the hypothesis being put forward here, the chemical initiator that increases liver tumors in the Cx32 KO mouse might affect, epigenetically, pre-existing mutated cells (i.e. ‘oval’ cells) in such a manner as to allow them to escape tumor suppression in a liver tissue environment without Cx32, a tumor suppressor gene. Moreover, if most of the carcinogens act as promoters, testing them as liver promoters in this mouse should be ‘negative’, whereas in the wild-type mouse, it would be positive after the liver was exposed to some ‘initiator’.
Ultimately, we recommend that a new emphasis be placed on the development and validation of several normal human stem cell 3D in vitro assays (human lung, liver, breast, prostate, kidney, brain, hematopoeitic, etc.) to test for cytotoxic and epigenetic endpoints (altered cell proliferation, differentiation, apoptosis, methylation changes, cell–cell communication) at non-cytotoxic levels (3,69). These should be used to identify if any threshold levels of change are seen for these endpoints at non-cytotoxic levels.
|
Three dimension ‘organoid’ and differential sensitivity of 2D and 3D cultures |
|---|
Cells in vivo exist in a dynamic, interactive 3D milieu. Jacks and Weinberg have stated: ‘The notion that cellular transformation and tumor progression involve the cooperative effects of proliferative signaling pathways and antiapoptotic pathways has been well studied in standard monolayer culture and in some in vivo models. However, the 3D culture system has the distinct advantage that it takes into account physiologically relevant interactions while being amenable to facile manipulation and biochemical analyses’ (9
). Use of 2D in vitro assays to measure mutagenesis, cytotoxicity (necrosis and apoptosis) and epigenetic alteration of gene expression are, clearly, not perfect surrogates to replicate the in vivo situation. Moreover, the 2D in vitro assay uses proliferating cells usually (either primary or immortalized cells). In addition, these culture systems are usually of one cell type, albeit, even here, the cells are not necessarily genotypically or phenotypically homogenous. To make matters even more complex when trying to extrapolate results from in vitro to in vivo assays, many of these toxicity assays are performed on cells in two dimensions when in log phase, a situation usually not seen in vivo except during massive tissue growth or hyperplasia during wound healing.
In the last decade, more innovative techniques (70–75) are being used to reconstruct some of the 3D dynamics seen in vivo, such as organotypic cultures (76), co-culture of several cell types (77), formation of 3D ‘organoids’ (61), insert wells to detect soluble factors that might identify stromal–epithelial communication (78), and embryonic and adult stem cells for testing toxicants (69,79). In all tissues, there exist three basic cell types, the few adult stem cells, the majority committed progenitor or transit cells with a finite life span and the terminally differentiated cells. The ‘niche’ in which the adult stem cell resides controls its behavior, in addition to feedback signals from the terminally differentiated daughter cells (80), in addition to environmental factors, such as the oxygen tension and calcium levels (81), and other nutrients. Most in vitro assays used to detect the toxicity of chemicals are done at log phase and high oxygen levels, in media with high calcium levels and complex growth factors, such as serum. All these factors create conditions in the tested cells that do not mimic the physiological state of the potential target cells in vivo. Many studies have shown that the same cells tested in 2D log phase conditions, confluent 2D conditions or 3D systems do not yield the same toxic result (82). No one has performed such an in vitro study with a proper mixture of human stem cells, their differentiated progenitor daughter cells and terminally differentiated progeny. This ought to be the ultimate goal for future risk assessment after chemical (or radiation) exposure from the in vitro to in vivo human situation. Such a study can be the closest experimental risk assessment that can be performed in humans. Even in this case, the limitations of individual genetic background, and other complex interacting factors will never be mimicked for extrapolation to a particular individual. It might be the best for which one can strive.
|
Stromal–epithelial interactions as one example of extracellular communication linked to GJIC: hormones, growth factors, cytokines, extracellular matrix modulation of gap junctions |
|---|
With the use of co-culture conditions and the development of new growth conditions for epithelial cell cultures, clear evidence of stromal–epithelial interactions have been observed, which influences the biological control of cell behavior (83
). In addition, in the case of stem cells that have the potential to proliferate either symmetrically or asymmetrically, identifying those factors that could control which manner of cell division a stem cell will take is crucial. Recently, endothelial cells, co-cultured with neural stem cells were shown to stimulate the self-renewal of cells and expand neurogenesis (78). Even the long-term effects of radiation to induce a persistent increase in the plasma levels of pro-inflammatory cytokines owing to an imbalance of CD4 T cells (84) again illustrates the interaction of different cell types on each other in the complex carcinogenic process.
Modulating cell–cell communication through GJIC plays an essential role in the development of most epithelial and fibroblasts progenitor cells in vivo (85). Endogenous factors that can modulate (increase or decrease GJIC) include growth factors (86), various cytokines and hormones (87), as well as extracellular matrices (88). Modulation of GJIC between coupled cells can influence cell proliferation, cell differentiation and apoptosis, as well as synchronize electronic and metabolic functions (16,89). The complex coordination of extracellular matrix, growth factors and nutrients in the medium can influence the expression and function of the gap junction proteins, as well as the state of differentiation of cells.
|
Redox disturbances of homeostatic control of cell proliferation/differentiation/apoptosis |
|---|
Inhibition of GJIC and apoptosis, altered differentiation, and stimulation of cell proliferation of these cells in the 3D cultures of normal human epithelial cells should be the closest we can get to mimicking the in vitro condition of the in vivo human situation. Any chemical testing positive at a given non-cytotoxic dose could be tested in a rodent system as a tumor promoter of initiated breast, liver, skin systems during a validation and species comparison phase. If a chemical is tested rigorously as being negative for any of the cytotoxic and epigenetic endpoints, even at the highest tolerated level, non-cytotoxic levels will not prove the chemical is ‘safe’. However, it might be the best we would ever do.
When a chemical enters the human body, its ultimate fate on its potential biological and health effects will depend on a number of factors (genetic, gender, developmental state, target organ, cell type, cell cycle status, destination, en