In solid tissues, cells are normally contact-inhibited; therefore there is very little cell division.  Promotion must involve a mechanism which overcomes contact inhibition.  It has been hypothesized that the inhibition of gap junctional intercellular communication might be the unifying cellular mechanism for any and all kinds of tumor promotion.  The function of tumor promotion would appear to increase the target size of the cells containing the first genetic hit, assuming that a cell must accrue at least two, possibly more hits (Potter, 1981a; Trosko and Chang, 1980; Trosko and Chang, 1985).  If we assume that the probability of activating an oncogene and losing a tumor suppressor gene in a single cell is the product of two rare independent events, then the probability would be the product of those rare probabilities.  If a cell with one hit is clonally amplified, then the longer the promotion period, the greater the number of such cells.  Now the probability of a second hit would involve the multiplication of the product of the independent probabilities and the number of cells.  This has led to the prediction of the “I-P-I”model of carcinogenesis (Potter, 1981a; Potter, 1984).  Experimental evidence seems to be consistent with the hypothesis (Hennings et al., 1983; Reddy and Fialkow, 1987; Scherer et al., 1984).

Progression is assumed to involve the irreversible alteration of one of the promoter-dependent initiated cells to become an autonomous malignant cell.  Agents which appear to facilitate this transition seem to be mutagenic (Hennings et al., 1983; Jaffe et al., 1987; Reddy and Fialkow, 1987; Taguchi et al., 1984).

Assumptions related to the oncogene/tumor suppressor gene must also be integrated in this analysis.  In general, it is assumed that oncogenes need to be present and activated for a cell to become a tumor.  Tumor suppressor genes are assumed to be inactivated or lost in order that a cell with an activated oncogene becomes a tumor cell.  Because of many observations which suggest that this assumption might be too simple (e.g., two kinds of activated oncogenes can transform some cells without the known loss of any tumor suppressor genes; one activated oncogene can transform some cells without the known loss of any tumor suppressor genes (Land et al., 1983; New bold and Overell, 1983; Spandidos and Wilkie, 1984)), the Yin-Yang model of oncogene/tumor suppressor genes has been hypothesized.  The basic element of this hypothesis is that it is not so much a matter of which or how many oncogenes or tumor suppressor genes have been activated or lost in a given cell, but rather what is the net effect in any given type of cell of the two types of genes on regulating contact-inhibition or gap junctional intercellular communication.

Assumptions related to the mechanisms of action of agents which might act as initiators, promoters and Progressors and which might cause activation of oncogenes and de-activation of tumor suppressor genes need to be reviewed.  In general mutagens, which act primarily by deleting genes and being clastogenic rather than by being a “point” mutagen (e.g., ionizing radiation) (Morgan et al., 1990), would be expected to be a better agent at de-activating tumor-suppressor genes than as an activator of oncogenes or as a tumor promoter.  However, it is not as clear cut as that, since an agent, such as ionizing radiation, can cause chromosome breaks leading to some stable translocations.  If by chance the rearrangement of chromosomes placed an un-mutated, but suppressed oncogene, next to a piece of DNA which now stably activated this gene, then ionizing radiation could affect oncogenes, as well as tumor suppressor genes, controlling cell growth.  In addition, ionizing radiation, at doses which could kill a significant number of cells in a tissue, could act as an indirect promoter by causing any surviving initiated cell to go into compensatory hyperplasia.  It would seem that this would be described as a threshold-like effect of ionizing radiation in tissues.  In general, it would seem that ionizing radiation would be better Progressors, than either an initiator or promoter, even though it has the potential to act at all three stages.

Point mutagens can activate oncogenes, cause cell killing at high doses, and inactivate tumor suppressor genes.  In effect, they can act as initiators, indirect promoters by causing cell death-induced compensatory hyperplasia, and Progressors.  Ultraviolet light might be considered a good example of a skin cancer initiator and indirect promoter (Trosko, 1981).  It might be poor Progressors for the reason that little ultraviolet light would not be able to penetrate many cells to induced damage to the DNA.  Chemicals also might be effective as initiators and both direct promoters or indirect promoters by their cytotoxic effects, at least in targeted tissues.  Many chemical mutagens might not be good Progressors by virtue of the fact that if they need to be metabolized, many initiated cells seem to be resistant to the very chemical that might have initiated it because they no longer have the same drug metabolizing systems (Morgan et al., 1990).

The first assumption to be examined is that only the stem cells are the target cells for the carcinogenic process.  In other words, those cells which are near or terminally-differentiated will not give rise to cancer.  As a corollary, it assumed that the stem cell pool in different organs varies with not only age, but the organ itself due to a variety of biological reasons.  In order to illustrate this point, it will be further assumed that there are three major types of organ systems: “closed,” “open” and “mixed.” “Closed organ” systems refer to organs, such as the liver or the kidney, where the stem cells are tightly controlled by the differentiated daughter cells in order to maintain a constant volume.  On the other hand, in “open organ” systems, such as the skin and lining of the G.I.tract, the stem cells are constantly dividing because the differentiated cells are constantly being lost.  In organs, such as the lung or testis, while the volume size is maintained, stem cells are constantly dividing to replace those being lost.  This classification system might have relevance to those risk factors interacting with the carcinogen, such as radiation, which might affect cell proliferation.  They should affect “closed organ” systems more than “open organ” systems, since without a source of mitogenic stimulation (either endogenous growth factors or hormones or exogenous chemicals or cell death and removal), the initiated stem cell would be held in check by the negative growth regulators of the differentiated cells.

Going back to the assumption that the stem cell pool might vary in certain tissues, it would be predicted that the stem cell pool might decrease in “closed organ” systems with age/developmental/pregnancy status.  The number of stem cells in “open organ” systems might be expected to remain relatively constant.  On the other hand, the stem cell pool might be expected to decrease with age in closed systems.  This might explain, in large part, “age at the time of exposure” effects, as has been noted with the relationship between breast cancer and pregnancy status.  Pregnancy would be expected to deplete the stem cell pool to produce differentiated milk-producing cells.  Therefore, there would be fewer target cells for the initiation of the carcinogenic process with earlier and more frequent pregnancies.  Women, who accrue initiated stem cells in their breast tissues with age but do not get pregnant, have the chance of having these initiated stem cells promoted every month by the hormones associated with the menstrual cycle.

Biological observations related to the role of stem cells in carcinogenesis include the fact that dividing cells seem to be more susceptible to the cytotoxic and mutagenic action of carcinogen initiators (Rossman and Klein, 1988).  In addition, cell division seems to be imperative for the promoting phase of carcinogenesis (Bout well, 1974).  While recently much attention has been given to the role of mitogenesis in carcinogenesis (Ames and Gold, 1990a), it should be pointed out that in the cases where apparent mitogenesis of cells in a given organ (or the lack thereof) has not been correlated with carcinogenesis, it must be shown that the initiated stem cell behaves identically to the normal stem cell.  If initiated stem cells respond to a mitogen differently than the normal stem cell, one cannot simply use mitogenesis of tissues as a “biomarker” (Ledda-Columbano et al., 1989).  Evidence for this might be from the observation that Phenobarbital treatment of initiated rat livers does not lead to general enlargement or overall proliferation of hepatocytes, but does lead to the clonal expansion of the initiated cell into a large focus of cells.  In other words, Phenobarbital appears to stimulate, selectively, the initiated cell to proliferate (Kaufman et al., 1988).

An additional assumption related to the carcinogenic process is that the process involves multi-steps and multi-mechanisms.  In brief, the initiation, promotion, progression concept, while recently criticized (Ames and Gold, 1990a; Ames, 1991), is applicable to most carcinogenic processes.  The one exception might be the formation of teratomas.  Even here, this cancer could be explained by a non-initiation/promotion/progression hypothesis without having to resort to the rejection of this major concept.  First, one has to realize that “initiation,” and “progression” is operational concepts, derived from whole animal studies, not from experiments involving cells in culture or molecular events in the test tube.  No implied mechanisms are intrinsic to these terms.  While the observations leading to these concepts were derived from experimental rodent studies and while, in real life, neither rodents or human beings are subjected to those conditions that allow one to clearly observe distinet events taking place in the experimental conditions, it is further assumed the concept is applicable to the real life human situation, In the experimental condition, it appears to be a fact that initiation is the result of an irreversible event.  Since, by definition, mutagenesis is an irreversible event (while some types of mutations can be back mutated, for all practical purposes, in this situation it would be highly unlikely).  In addition, true mutagens, such as UV light, appear to be good initiators (Fry et al., 1982).  On the other hand, if a critical gene is stably altered by an epigenetic event (e.g., a gene is stably turned on or off by, let’s say, an ethylating or de-ethylating event), then it would be possible for an initiating event not to involve a mutagenic event.  In the case of teratomas, if a critical oncogene is de-repressed in the developing embryo, the cell might grow as a tumor with the microenvironment providing the promoting stimulus.  When the microenvironment changes due to the growth, some of the cell could terminally differentiate giving rise to many of the disorganized tissues found in the teratomas.

From whole animal studies, it appears that the promotion phase is an interrupt able process (Bout well, 1974; Pitot et al., 1981).  While the underlying mechanism is still unknown, it is assumed that if initiation involves the irreversible alterations of a single stem cell, then promotion is that process which allows the colnal expansion of that cell.  Several additional assumptions have to be made.  First, the gene (s) altered in the initiated stem cell must prevent terminal differentiation of that stem cell.  If the initiated cells are stem cells which cannot terminally differentiate, then any mitogenic stimuli (i.e., promotion) would allow these cells to accumulate as “non-terminally” differentiated cells in a tissue, whereas the normal stem cells, after proliferation, would terminally differentiate.  The more these undifferentiating cells are forced to proliferate, the larger this focus of cells becomes in the tissue.  In the skin, as normal cells differentiate, they eventually die and are removed.  Only the undifferentiated initiated cells would remain and accumulate as a papilloma.  In the closed organ, such as the liver, the assumption is that the stem cell, perhaps the oval cell (Fausto, 1990; Sell, 1990; Sirica et al., 1990), will selectively proliferate after a mitogenic or promoting stimuli accumulating as a focus of incompletely terminally differentiated hepatocytes.  A great deal of experimental evidence, which led to the “oncogeny as partially blocked ontogeny” hypothesis (Potter, 1981b) and in vitrol in vivo studies (Miller et al., 1987; Yuspa and Morgan, 1981; Yuspa et al., 1985), seems to be consistent with these assumptions.  While it is generally assumed that promotion involves mitogenesis, some have argued that not all mitogens are promoters and not all promoters are mitogens.  However, one has to keep this clearly in mind and to be critical of the so-called exceptions.  In the first place, if the tumor, consisting of billions of clonally derived cells results from some “promoting” treatment, then it should be obvious that, while all cells might not have divided during the promoting treatment, the initiated cells did.  For how else could a single initiated cell become a tumor of billions of cells? In the second place. It has to be shown that a “non-promoting” mitogen actually can sustain mitogenesis for a long period of time and can actuallyu promote the initiated cell.  For if it can only trigger mitogenesis for a short period of time or that it preferentially acts as a mitogen only for the normal stem cell and not the initiated cell, then these superficial observations do not constitute evidence against promoters as mitogens.

The ultimate understanding of how a multicellular organism, such as the buman being, maintains its ability to regulate cell growth, proper differentiation, adaptive responses and wound-healing will necessitate adhering to the principle that an organism is “greater than the sum of its parts” (Eagle, 1965).  This holistic concept, together with the hierarchical (Brody, 1973) and cybernetic (Potter, 1974) concepts, has been utilized to explain how the regulation of ions and molecules is achieved at the molecular/biochemical levels, at the cellular level and at the organ and system levels.  The idea that there exists a cybernetic-like control system in multicellular organisms to regulate cell functions within and between tissues was created by Claude Bernard (1878). W.B. Cannon coined the term, homeostasis, to describe this regulatory process (1929).  In addition, Weiss and Kavanan (1957) postulated the existence of growth regulators of cell proliferation, while Mazia (1961) suggested that “removal of a block” was necessary for cell division, Later, Potter (1983) integrated both the positive and negative factors into a scheme whereby, locally and systematically, they could regulate growth.

The explanation of how a constant “milieu interieur” is maintained was provided by the conceptualization of positive and negative factors, produced by cells of one type/tissue which affects distal cells of another type/tissue.  It predated the discovery of most of the actual molecular entities involved in homeostasis.  Eventually, this concept of homeostasis by a cybernetic process was fleshed out with the discovery of various hormones, biologically active peptides, lymphokines and neurotransmitters.  Note, that oncogenes code for these kinds of molecules (and their receptors).  The bilolgical process by which cells influence control over each other in the multicellular organism is referred to as intercellular communication.  In general there are three forms of intercellular communication; extracellular communication would refer to how cells communicate with distal cells via the secretion of a molecular signals (e.g., hormone, peptide, growth regulator, neurotransmitter); intracellular communication would refer to those transmembrane-signalling elements triggered by the extracellular signals; and Intercellular communication would refer to the transfer of ions and small regulatory molecules from contiguous cells through a protein channel on the cell membrane, called the gap junction (Loewenstein, 1990).

This interconnected communication network allows a multicellular organism to control its fundamental cellular, tissue, organ and system functions.  For example, a hormone secreted by the hypothalamus (extracellular) could trigger intracellular biochemical changes in distal cells of another organ, which in turn, could either increase or decrease intercellular communication in that tissue which has a receptor for that hormone.  By increasing intercellular communication, these hormone triggered cells might now differentiate or if already differentiated, produce a substance unique to that hormone signal.  Alternatively, if the biochemical changes occurring in the targeted cells cause a decrease in gap junction-mediated intercellular communication, then the cells have an opportunity to divide.

The key to this homeostatic process is an understanding of the fundamental role the gap junction plays in multicellular organisms.  This channel-like structure, made of proteins which are highly conserved through evolution, is found in almost all cells of the solid tissues of metazoans (Revel, 1988).  These gap junction channels convey ions and small molecular weight molecules from one cell to all others which are connected by these channels.  In effect, all coupled cells are in equilibrium for these ions and small molecules.  If any one cell has a momentary increase or decrease in these substances, these will soon equilibrate via these channels.  These channels allow a single cell to be either a “sink” or “point source” to the other coupled cells for these ions and small molecules (Sheridan, 1987).  Therefore, if critical a level of a regulatory molecule or ion is needed before a cell can proliferate of differentiate modulation of the gap junction function or number might be needed.  For example, if a series of cells are coupled and only some of them receive a growth factor molecule, the intracellular changes induced by the extracellular signal would be shared by all the coupled cells, such that no one cell would have enough to start the cell division process if the gap junctions remained open (Lowenstein, 1967).  On the other hand, it has been postulated that if the gap junction function is down regulated by the growth factor, these critical intracellular signals would reach a critical mass and trigger the cell division process.  This appears to be the case, since it has been shown that various growth factors and hormones can down regulate gap junction function (Aylsworth et al., 1989; Maldonado et al., 1988).

Gap junction communication has also been implicated in the regulation of normal development and differentiation (Lo, 1985; Schultz, 1985).  The fertilized egg and early blastula cells do not have gap junctions (Lo and Gilula, 1974).  As the early embryo starts to develop and differentiate, gap junctional communication starts to appear (Lo and Gilula, 1974).  Even stem or progenitor cells in various human tissues (e.g., breast, kidney) do not appear to have gap junctions (Chang et al., 1987; Chang et al., 1990).  If developing embryos are treated with antibodies or anti-sense RNA directed to gap junctions. Abnormal development ensues (Bevilacqua et al., 1989; Warner et al., 1984).  Several chemicals, such as dilation, retinoid and alcohol, known to be teretogens but not mutagens, have been shown to down regulate gap junctions (Trosko and Chang, 1988a).

At this point, the evidence, while circumstantial, implicates the up or down regulation of gap junctional communication in triggering quiescent cells to proliferate or proliferating cells to contact-inhibit; undifferentiated cells to differentiate; or differentiated cells to adaptively respond to environmental changes (Trosko and Chang, 1984).  This up or down regulation of gap junctional communication can be adaptive or maladaptive.  To date, a large number of chemicals, both natural and human-made, have been shown to modulate gap junctional intercellular communication in a variety of mammalian, including human, in vitro systems (Trosko and Chang; 1988a).  Some of the chemicals are natural and normal physiological chemicals, such as epidermal growth factor and hormones (Kihara et al., 1990; Larsen, 1983; Madhukar et al., 1989; Maldonado et al., 1988).  Others are nutrients, food additives, cigarette tar condensates, drugs, pesticides, pollutants, biological toxins and natural plant products (Trosko and Chang. 1988a).  These chemicals have been shown to down regulate gap junctional intercellular communication reversibly, at non-cytotoxic levels, in a dose dependent fashion in both rodent and human in vitro cell systems.

Modulation of the critical cellular function can be at the transcriptional (expression of a gene).  Translational (production of a gene-related protein) or at the posttranslational (modification of a synthesized protein) levels.  Intracellular changes induced by endogenous or exogenous extra cellular communication molecules can modify levels of calcium, H+ ions, as well as the activation of enzymes which posttranslational modify proteins, increase cyclic AMP and generate free radicals.  All of the induced intracellular changes have been associated with the regulation of gap junctions (Saez et al., 1990).  In other words, physiological changes can bring about cellular changes, such as cell differentiation, cell division or adaptive cell responses, by triggering intracellular biochemical changes which modify gap junctions.

To bring this concept of intercellular communication directly to the focus of the review, it should be noted that two of the earliest observations characterizing cancer cells were that they lost growth control and they could not normally terminally differentiate.  One would predict, assuming that the gap junction played a major role in these two fundamental processes, that cancer cells would have some dysfunction in gap junctional intercellular communication.  Lowenstein and his co-workers (Kanno, 1985; Lowenstein, 1966) observed very early in the study of gap junctions that many cancer cells did, indeed, have abnormal gap junction function.  Later, assitional evidence showed that tumor metastasis correlated with decreased gap junction function (lijima et al., 1969; Nicolson et al .,1988; Ren et al., 1990).  Some of the apparent discrepancies have been explained by the demonstration of selective communication, that is, some cancer cells could communicate with themselves but not with normal cells (Yamasaki et al., 1987).

Linking gap junctions with the initiation / promotion / progression model of carcinogenesis, the demonstration that tumor promoting chemicals and physical conditions could down regulate gap junctions is critical (Murray and Fitzgerald, 1979; Yancey et al., 1979; Yotti et al., 1979).  These chemicals are “mitogenic” for the initiated cells.  By blocking cell-cell communication, they allow the initiated stem or progenitor cell to escape the suppressing influence of the surrounding communication normal cells.  Chemical tumor promoters, such as the phorbol esters, have been shown to decrease the gap junctions on cells in vitro, as well as in vivo (Beer et al., 1988; Janssen-Timmen et al., 1986; Kalimi and Sirsat, 1984; Sugie et al., 1987; Yancey et al., 1982). Even partial keratectomy, a tumor promoting stimulus (Arguis, 1985). Has been associated with the down regulation of gap junctional communication (Dermietzel et al., 1987).  On the other hand, anti-tumor promoting chemicals, the retinoid, up regulate gap junctions in phenotypic ally transformed cells to reverse them to a normal phenotype (Mehta et al., 1986; Mehta et al., 1989; Mehta and Lowenstein, 1991).  Of some relevance to this analysis is the observation that these tumor promoting chemicals appear to have threshold limits, below which they are unable either to down regulate gap junctions (Trosko and Chang, 1985), and promote the growth of tumors in vivo (Deml and Oesterle, 1987; Goldsworthy et al., 1984; Pitot et al. 1987; Verma and Bout well. 1986).

To link gap junctions to oncogenes and tumor suppressor genes, there have been a series of recent reports showing that cells expressing certain oncogenes (e.g.Ras, Src,Neu,Mos,Raf) had reduced gap junctional intercellular communication (Atkinson et al., 1981; Atkinson et al., 1986; Azarnia and Loewenstein, 1984; Azarnia and Lowenstein, 1987; Azarnia et al., 1988; Chang et al., 1985; Dotto et al., 1989; EI-Doulyy et al., 1988; EI-Fouly et al., 1989a; EI-Fouly et al., 1989b; Kalimi et al., 1990).  In addition, non-communicating cancer cells (HeLa), when fused to form a non-tumorigenic hybrid with a normal human fibroblast cell, were shown to have functional gap junctional communication (Kalimi et al., 1990).

Consequently, “inactivation” of tumor suppressor genes must occur in order for the effect of an activated oncogene be felt.  Alternatively, activation of more than one oncogene might overcome the suppressing effect of the normal tumor suppressor genes.

With scores of oncogenes having been discovered and several tumor suppressor genes having been identified, one wonders, especially after seeing no “universal” pattern related to molecular, biochemical, or cellular structure of function within or between tumor types, if a unifying hypothesis will ever explain these observations.  To date, proto-oncogenes and their tumor-related counter-parts, the oncogenes, seem to code for growth factor or growth factor receptor molecules, molecules found in the cytoplasm (transmembrane signally elements needed for cell growth much as coded by the Ras oncogene) or  the nucleus (transcription elements needed to regulate gene expression such as Myc)(Weinberg, 1985).  The tumor suppressor genes, again, represent a completely different set of genes (Harris, 1988; Klein, 1987; Knudson, 1985; Lee et al., 1991; Sager, 1986).

While some tumor types (e.g., breast cancer) have been associated with a high frequency of particular oncogenes (Slamon et al., 1987) [or loss of specific tumor suppressor genes] (Malkin et al., 1990)), the association has never been shown to be universal.  In other words, there seems to be several ways or combinations of oncogenes and tumor suppresson genes by which a given type of cell can be converted to a cancer cell.

Earlier concepts had suggested that one needed at least two oncogenes to make a normal cell a cancer cell (e.g., a Myc-type to “immortalize” a cell and a Rastype to transform it to a malignant phenotype).  Then it was shown one could transform some cells with only one oncogene (Spandidos and Wilkie, 1984).  Later, to make things more complicated, the loss of tumor suppressor gene function was needed (Harris, 1988; Klein, 1987; Knudson, 1985; Lee et al., 1991; Sager, 1986).  In addition, others seemed to indicate that while, genetically only “two-hits” seemed to be needed for cancer-prone syndromes such as retinoblastoma (Moolgavkar and Knudson, 1981), many more “hits” were needed to transform human colon and bladder cells (Marx, 1989; Vogelstein et al., 1988).  In addition, others seemed to indicate that while, genetically only “two-hits” seemed to be needed for cancer-prone syndromes such as retinoblastoma (Moolgavkar and Knudson, 1981), many more “hits” were needed to transform human colon and bladder cells (Marx, 1989; Vogelstein et al., 1988).

On one hand, these observations might simply mean that the number of changes needed to transform some cells is different than for others, it could also mean the manner in which these studies were done demand different interpretations.  For example, in the case of whether one or two oncogenes are necessary to transform a normal cell, one might ask if in the case o;f one oncogene being sufficient whether those cells already had highly expressed “proto-oncogenes” or exogenous factors in the cell culture system which interacted with the assed experimental oncogene.  Secondly, in the case of the multiple oncogenes needed to transform human lung epithelial or bladder cells, one needs to know if the culture conditions forced the need for multiple oncogene changes which might not be needed in vivo.  Alternatively, the target cells used in these experiments might not be the “target” cells in vivo.

One additional way of conceptualizing this complex set of observations is to view, not the individual specific molecular/biochemical nature of the oncogene/tumor suppressor gene products, but their cellular net effect and its consequence on an important cellular function; namely, the maintenance of homeostasis.

The “Yin/Yang” model of oncogenes/tumor suppressor genes might be a way to explain the perplexing array of experimental observations on oncogenes/tumor suppressor genes.  Moreover, as will be seen in the next section, independent of the specific oncogene/tumor suppressor gene interaction in any given tumor, if the consequent of each type of these interactions is the loss of homeostasis, then the unifying concept for all these combinations might be the answer to this questions: “What is the cellular basis for tissur homeostasis?”.

As is illustrated in Figure 4, a cell has normal growth control, and it has a specific balance dictated by the particular proto-onogene/tumor suppressor genes expressed in that particular tissur (e.g., breast, lung, liver).  However, if that balance is upset by the proto-oncogenes being activated such that they can negate the normal tumor suppressor genes in that particular tissue micro-environment, the the cell would have a growth advantage.  Alternatively, in another cell, this (these) activated oncogene(s) might still be controlled by the normal tumor suppressor genes in that micro-environment.  If the tumor suppressor genes function is lost or suppressed by the micro-environment, then again the cell would have a growth advantage.

Since certain oncogenes might be expressed in only certain tissues or since the physiological state of different differentiated cells expressing the same oncogene is different, it would not be surprising to find that in one cell the activated oncogene is associated with transformation, yet in the other it is not.  The point is oncogenes and tumor suppressor genes do not act in a vacuum.  Their interaction is in the context of a complex interacting set of that interaction is not only on that cell, but on a community of interacting cells.

Integrating this concept with the “nature and nurture” concept, one finds the Li-Frameni syndrome as an example where an inherited gene, which predisposes the individual to certain types of cancer, appears to be a tumor suppressor gene (Malkin et al., 1990).  In other words, at conception, these individuals inherit one mutated tumor suppressor gene.  During development and life, if it acquires an activated oncogene in a stem cell of an organ in which the mutation or loss (deactivation) of the second tumor suppressor gene has occurred, then cancer would develop.  Why tumors do not occur in other organs of this syndrome even though the inherited mutant tumor suppressor gene is found in all the cells is not known.

In addition, new evidence has identified specific genes, which if mutated, over expressed, abnormally expressed or lost, can influence the carcinogenic process.  These genes, the proto-oncogenes and tumor suppressor genes play critical roles in normal cells by regulating cell proliferation or differentiation (Sager, 1986; Weinberg, 1985).  When altered, these two fundamental processes, which are needed for the survival of all multicellular organisms, contribute to the general characteristics of all cancer cells, namely, these cannot control their growth, nor can they terminally differentiate under normal conditions.  Returning to the xeroderma example, by inheriting mutated genes at conception, the xeroderma individual cannot repair its DNA if exposed to sunlight.  If the cell are not exposed to sunlight, the DNA will not be damaged (ignore for the moment, damage done to these cells by UV-mimicking chemicals).  If the DNA is damaged, mutations will occur in many genes, including the proto-oncogenes [Nature and Nurture].  By inheriting a mutation in this case, more mutations can occur in the body after exposure to the environmental agent, sunlight.  Some of these mutations then can start the cancer process (Glover et al., 1979; Maher and McCormick, 1976).  That is, these mutations, if they occur in the proper oncogene in a stem cell, could prevent the terminal differentiation of the stem cell when it is stimulated to divide and differentiate.  If the amount of UV-induced DNA damage is severe in the terminally-differentiated cells, these cells would die since the XP cells do not repair their DNA very efficiently.  This cell death of the differentiated cell population would force the surviving initiated stem cell to proliferate, but not to terminally differentiate (Trosko, 1981).  This clonal expansion of the initiated cell would be the result of a cell death-type of promotion process.  As these initiated cells accumulate on the skin surface, additional exposure could induce another mutation in one of the initiated cells, thereby providing a second “hit” which might be needed to complete the carcinogenic process in this case.  In other words, xeroderma pigmentosum might represent an “initiation-promotion-progression” or “complete carcinogen” syndrome.

Skin cancer can, of course, be induced in non-xeroderma pigmentosum individuals.  When normal individuals get exposed to massive amounts of ultraviolet light, such that their DNA repair capacities are unable to repair all the damage, some of the damage would then act as substrates for mutations as in the case of xeroderma pigmentosum.  Either repeated exposures to large amounts of sunlight to induce cell killing to act as promoting and progression agents or exposure to non-cytotoxic mitogenic agents would ultimately bring about the skin cancer.

This xeroderma pigmentosum syndrome can further illustrate the “nature and nurture” model when one examines the internal tumors of these individuals.  While all the cells of these XP individuals are deficient in DNA repair, and while there have been reports of internal tumors found in these individuals (Kraemer, 1980), the numbers are small in comparison to the skin tumors.  Some would argue that is because they do not live long enough to develop internal tumors.  While that is probably true, it points out that the environmental exposure to chemical mutagens is far lower than exposure to the physical mutagen, sunlight.  Chemical mutagen exposure would probably never exceed levels which would induce massive cell killing as exposure to sunlight would do.  Therefore, internal initiated cells in the XP individual would have to be exposure to a high enough level, in a sustained, chronic manner, of noncytotoxic mitogenic stimuli to act as a promoter.

Recent molecular findings on the retinoblastoma, Wilms’ and Li-Frameni syndromes might suggest that these could represent “initiator-prone” syndromes (Malkin et al., 1990), if one assumes that the inheritance of one mutated gene through the germ line as the “initiating” event.  On the other hand, Downs and neurofibromatosis syndromes, which have not been sh;own to have defects in DNA repair (Trosko et al., 1985; Yotti et al., 1980), might be conceptualized as “promoter-prone” syndromes (Trosko et al., 1985).  In general, there are genes which could enhance or prevent environmental mutagen-induced DNA damage (e.g.,albinism or dark pigmented individuals, respectively) and genes which stimulate or reduce cell proliferation in certain tissues (e.g. genetic imbalanced growth regulators and hormones).

While there have been several major observations prompting support for the stem cell theory, one of the relevant observations to this analysis is the finding of Nakano and Ts’o (1991).  They demonstrated, in an in vitro hamster transformation system, that a small subpopulation of less differentiated and contact-insensitive cells was more susceptible to neoplastic transformation than populations of cells depleted of these “presumptive” stem-like cells.  In brief, not all cells could be transformed when exposed to various “carcinogens”.  Moreover, this subpopulation of transformable cells decreased with the developmental age of the animal.  These observations not only relate to the stem cell theory but also to cancer risk assessment.  If the same phenomenon exists in the human body, then, assuming not all cells are equal targets for carcinogenesis, the risk associated with exposure to carcinogenesis would be a function of that population of cells.  In addition, if that stem cell population in various organs is differentially modulated by genetic, developmental or environmental factors, the n the risk at the time of exposure would have to take this into account. (More will be said on this later.)

Evidence has been accruing recently suggesting that these stem cells do exist in human tissue.  Chang et al., (1987; 1990) have isolated contact insensitive cells from normal human kidney and breast tissue.  Several important features related to these cells provide linkages to additional hypotheses concerning the carcinogenic process.  The first is that these cells, while normal, appear to be immortal, i.e., they do not have the so-called “Hayflict” lifespan (Hayflict, 1965).  The second is that they do not have functional gap-junctional intercellular communication which is normally found in the differentiated normal cells.  The former is of some philosophical as well as scientific importance.  The latter will be discussed later.

It is conventionally assumed that a normal cell is mortal while a cancer cell is immortal.  It is understandable, from an experimental biologist’s perspective since most normal human cells in tissue culture appeared to “senesce”, while most cancer cell could be grown indefinitely.  However, it now appears that the culture conditions in which normal tissue was placed did not favor the growth of stem cells.  Therefore, the senescence of the cells was probably due to the terminal differentiation of the cells and the loss of the stem cell pool.  If the stem cell is fundamentally immortal when it is naturally induced to terminally differentiate, it then becomes “mortal”.  Consequently, although recent molecular oncology studies continue to view the early process of carcinogenesis as one needing to “immortalize” a normal.  “Mortal” cell (Land et al., 1983; New bold and Overell, 1983), it has been hypothesized that the process is just the opposite (Trosko and Chang, 1989a).  This would also fit the stem cell theory very well.  If the stem cell is immortal, and if the first step in the carcinogenic process blocks the cell’s ability to terminally differentiate or become “Mortal”, the cell would still be able to proliferate and self-renew the error that is the inability to terminally differentiate.

The concepts of initiation, promotion and progression were derived from animal experiments to help explain phenomenologically distinct processes occurring during carcinogenesis (Bout well, 1974).  Operationally, on the whole animal level, initiation refers to the stable or apparent irreversible conversion of a normal stem or progenitor cell (see later discussion of the de-differentiation hypothesis) to a “premalignant” cell (Trosko et al., 1990b).  This premalignant or initiated cell has, by definition, several features; namely, it has the inability to terminally differentiate (Trosko et al., 1988); it has the potential to proliferate and maintain its inability to terminally differentiate; and lastly, it has the potential to acquire all the other genetic/epigenetic changes needed to become a neoplastic, invasive and meta-static cell.  Promotion, on the other hand, is that phase of the multi-stage process which enhances the frequency and earlier appearance of tumors in the initiated animal (Bout well, 1974).  Progression refers to the conversion of one of the promoted initiated cells to the neoplastic cell.

Although the mechanisms underlying each of these three phases are not vet known, some hypotheses to explain the apparent irreversible, initiation phase, the potentially interruptible or reversible promotion phase and the irreversible progression phase have been proposed.  Since mutagenesis is an irreversible event, and since known mutagens, such as ionizing and ultraviolet radiations, have been shown to be initiators in several initiation-promotion animal model systems (Trosko and Chang, 1985), it would be consistent to hypothesize that initiation is based on a mutagenic mechanism.  Many chemicals, considered to be mutagens based on interpretations of a wide range of cell-free and in vitro/in vivo data, are also speculated to act as initiators because of their presumed mutagenic action.  However, as will be discussed later, just because an animal can have some of its cells promoted after it has been exposed to a known or suspected mutagen, and just because either DNA lesions or DNA damage are found in tissues of the exposed animal, it does not prove that the mutation ultimately found in the tumor cell was induced by the agent that was used to induce the irreversible or initiation event.  In addition, conceivably, an agent, which could induce a stable epigenetic change, could be thought to be an initiating agent.  In short, at present we can only speculate that initiation, induced either by a mutation or stable epigenetic event, is that process which prevents a stem or progenitor cell from terminally differentiating but not from dividing.  This must be the case, for if an affected cell has the ability to terminally differentiate under normal conditions; it could not give raise to a cancer (see later discussion).

Promotion has been postulated to be that process by which a single initiated cell is selectively amplified by one of several mitogenic stimuli.

Here again, the underlying mechanisms are not known.  However, several lines of evidence have been used to hypothesize that promoting agents and conditions are mitogens, not mutagens (Ames and Gold, 1990b; Cohen and Ellwein, 1990; Trosko and Chang, 1988a).  Many natural chemicals, such as phorbol esters and palyiotoxins, endogenous factor, such as hormones and growth factors, nutrients, such as unsaturated fatty acids, pollutants, such as polybrominate biphenyls, pesticides, such as DDT, dieldrin and aldrin, cigarette condensates, and drugs, such as Phenobarbital, have been demonstrated to act as promoters in in vivo model systems (Trosko and Chang, 1988a).  The weight of the other experimental results with these chemicals is that they are not initiators in in vivo animal systems, and do not induce DNA damage or mutations in systems which simulate the in vivo conditions.  Other promoting conditions include normal cell growth (as might be evidenced by initiating young animals prior to their normal growth), cell removal (surgery or wounding), cell killing, and solid state object-induced hyperplasia (Trosko et al., 1990d).  In the animal systems, the mitogenic stimulation must be sustained (Sisskin et al., 1982).  In other words, even if an animal is initiated and exposed to mitogenic stimuli for a short time or for irregular times or at concentrations not sufficient to cause proliferation of the initiated cell (Bout well, 1974), then tumors will not appear at earlier times or higher frequencies.

While the implied cellular basis of promotion is mitogenesis, it does not necessarily mean that all promoting agents will stimulate cell division in all cells exposed to the agent.  By definition, if a tumor is clonally derived from a single cell during the carcinogenic process, then a promoter needs only to stimulate the initiated cell.  This has some important implications to those who wish to identify promoters from non-promoters.  If an agent can be shown to stimulate cell division in an animal tissue, it might have the potential of being a promoter in humans if it can be shown that it also can stimulate human cells.  Also, while a chemical could stimulate normal cells to proliferate, there is no guarantee that an initiated cell would also be stimulated to divide.  Conversely, if a chemical does not stimulate normal tissues to divide, it does not necessarily means it would not cause an initiated cell to proliferate.  In the case where a promoting agent causes both the normal and initiated cell to divide or to stimulate only the initiated cell to divide, the net effect is that the initiated cells will selectively accumulate in the tissue, since the normal progenitor cells will terminally differentiate, whereas the initiated cell will not.  These terminally differentiated cells would eventually be replaced, whereas the initiated cell will have the potential to further proliferate, thereby increasing their numbers.  At this point, a single initiated cell has increased its numbers by sustained mitogenic stimuli, such that a large mass of these undifferentiated cells occupy a tissue.

During that clonal expansion period whereby cells containing a stable alteration in a gene which affects terminal differentiation (This could be a number of genes; possibly a different gene for each tissue of different embryonic origin.), additional mutations or stable epigenetic changes could occur, allowing the initiated cell to acquire all the phenotypic alterations needed to become promoter independent (Trosko et al., 1983a).  Clearly, it at least two or more rare events are needed for complete neoplastic conversion, the probability that these two or more independent events occurring in one initiated cell of the body would be very small.  On the other hand, if one increases that number by promoting the single initiated cell, the larger that mass of initiated cells become, the chances for one of those cells containing one genetic “hit” to acquire the second or additional “hit” increase (Potter, 1981a; Trosko and Chang, 1980; Trosko and Chang, 1985).  Agents, which are thought to be effective at this stage, have been speculated to be able to induce stable genetic or epigenetic events (Hennings et al., 1983; jaffe et al., 1987; Reddy and Fialkow, 1987; Taguchi et al., 1984).  Once this has occurred the cell no longer would need an exogenous stimulator of cell division or of a blocker of cell suppression.

As a summary of the multi-stage multi-mechanism concept of carcinogenesis, three distinct steps have been identified in experimental animal systems.  In reality, for the human being, we are never exposed in unique sequence to those physical and chemical agents and conditions which are known to be initiators, promoters and Progressors.  Humans are constantly exposed to a wide and changing mixture of initiators, promoters and Progressors.  Rarely are we exposed to physical or chemical agents at cytotoxic levels, except possibly the ultraviolet rays of sunlight, alcohol and therapeutic radiation or chemical treatments.  We are all exposed to our growth factors at discreet periods of our lives, to chemicals which could interact additively, synergistically or antagonistically with each other (Aylsworth et al., 1989; Hirose et al., 1991; Warngard et al., 1987), as well as to the different and changing physiological endogenous chemicals of our bodies, due to genetic, sex developmental, life style and dietary factors.

Following up on cell death due to lethal gene or chromosome mutations, it has to be said that not all agents which cause cell death have to damage DNA or cause mutations.  A cell can be killed by many non-specific, ubiquitous means (i.e., membrane disruption causing ionic imbalances) or by very specific means (i.e., inhibition of vital enzymes).  While it should be obvious that dead cells do not give rise to cancers, cell death could contribute to the carcinogenic process by inducing surviving stem or progenitor cells to proliferate. (More will be said on this point when the tumor promotion process is discussed.)  It should be noted that if cell proliferation is an important step in the carcinogenic process, then one must recognize that in a solid tissue interactions between cells could influence the cytotoxicity of either ionizing radiations (Kavanagh et al., 1988; Kwok and Sutherland, 1991) or chemicals (Tofilon et al., 1984).  This then raises the possibility of thresholds for agents affecting the cytotoxicity-related step in carcinogenesis.  In any event, when a cell in a tissue dies, depending on the number and kind, stem or progenitor cells, which were quiescent up to this point, are stimulated either by the loss of contact-inhibition (a process whereby two or more contiguous cells prevent each other from dividing) or by some released mitogenic chemicals from the dead cell or by the loss of some negative growth regulator from the dead cell (Trosko et al., 1990c).  Genes in these surviving cells are modulated (expressed and/or repressed) to convert this non-dividing cell to one which can divide to replace the dead cells and to repair the tissue (compensatory hyperplasia).

This process of altering gene expression can be triggered by non-cytotoxic means, as well as the cytotoxic one just mentioned.  There are many natural and human-made, endogenous and exogenous chemicals which can alter gene expression.  This phenomenon is referred to as an epigenetic change.  That is, the alteration of the expression of a gene, at the transcriptional, translational or posttranslational levels, is an epigenetic event.  The genetic information has not changed as happens in a mutation, but rather that a gene’s information is now either expressed or repressed in a way such that it now influences the cell.  Chemicals, such as growth factors, hormones, neurotransmitters or drugs, such as retinol A, and natural or environmental pollutants, such as phorbol esters, polybrominated biphenyls or dieldrin, can alter gene expression without mutating cells or killing them (Trosko and Chang, 1988a; Trosko and Chang, 1989b; Rrosko et al., 1990a).  These chemicals are, therefore, referred to as epigenetic agents.  There are some interesting features related to how they trigger gene expression.  They usually have threshold levels by which cells respond to them, whether they are receptor or non-receptor dependent.  In principal, epigenetic events are reversible, whereas mutations, in practical terms of a cell in the body, are not.  Obviously, cell death is an irreversible event.  However, there are epigenetic events which can be very stable, such as that which occurs during development.  As a totipotent stem cell is committed to become a pluripotent stem cell for a given organ, genes needed for the generation of cells and functions of another organ are stably repressed.  On another level of biology, agents which act epigenetically can have some irreversible consequences.  For example, if chemicals, such as thalidomide or retinoid, are given to a developing embryo at a critical period, irreversible organogenic events happen, leading to teratogenesis.  These critical developmental events must occur at the appropriate times.  If an epigenetic chemical blocks the event, even though its molecular actions are reversible on the cell level, they are irreversible on the organism level.

To summarize this section, agents, physical and chemical, this can influence the multistage mechanism of carcinogenesis, do so by either mutating cells (gene or chromosomal), killing them or altering gene expression, epigenetically.  A mutagen can act by multiple mechanisms, and it can kill cells.  It can indirectly act to stimulate altered gene expression in the surviving cells.  In other words, a mutagen, depending on the dose or concentration, can have all three actions.  Cytotoxicants do not have to be mutagenic; however, by virtue of killing cells, they are automatically indirect stimulators of altered gene expression in the surviving tissue.  Lastly, epigenetic agents can alter gene expression without mutating or killing cells. ( Although all chemicals can kill cells at some concentration .)

To begin the process of trying to understand the carcinogenic process, introduction to several current hypotheses will be necessary; namely, the ‘initiation promotion-proression’ hypothesis (Pitot et al., 1981); the stem cell theory or the ‘disease of differentiation’ theory[“ontogeny as partially blocked ontogeny”] (Markert, 1968; Pierce, 1974; Potter,1988; Trosko and Chang, 1989a); the nature and nurture theory (Trosko and Chang,1978b); oncogenes and tumor suppressor gene theory (Bishop,1983; Sager, 1986); and the blocked intercellular communication hypothesis (Trosko et al., 1983a).  To some, to muse on the theoretical nature of carcinogenesis will get nowhere.  One only needs either to test the suspected cancer-causing agent in an animal system, short term assay designed presumably to test for a suspected mechanism in carcinogenesis or do epidemiological analyses on humans exposed to the agent.  There are all kinds of inherent problems with all of these approaches, as has been demonstrated with recent challenges to the current procedures to test for “cancer-causing” chemicals (Trosko, 1988; Trosko, 1989).  The basic nature of the problem seen by some is that the animal bioassay procedure, by not being designed with the multi-stage nature of carcinogenesis in mind, cannot provide information as to how the chemical could contribute to the multi-stage nature of carcinogenesis (i.e., Is the chemical a mutagen/initiator or is it a nitrogen/promoter?) (Ames and Gold, 1990b; Cohen and Ellwein, 1990).  On the other hand, because of so many fundamental artifacts and limitations of short-term tests (most assumed only to measure mutations (Trosko, 1988; Trosko,1989)), extrapolation of these results (positive or negative) from an artificial in vitro single cell system to an interacting multicellular rodent, let alone a human being, is truly stretching our current scientific knowledge.

To defend the necessity of trying to understand the mechanistic/theoretical bases of carcinogenesis, it seems that those who claim one needs only empirical animal and short-term tests fail to mention that these tests themselves are based on many implicit assumptions and hypotheses concerning the nature of carcinogenesis and “carcinogens”.  Therefore, in order for an epidemiologist to design a study to check if chemical x or radiation y, shown to induce a positive result in either a short term assay for mutagenesis or animal bioassay for tumor genesis, an understanding of how that particular agent contributes to the multi-stage, multimechanism carcinogenic process will be necessary for the reasons to be shown later.  In the same sense, any agent found by an epidemiologist to be associated with human cancers cannot automatically be assumed to be a mutagen/initiator of carcinogenesis as is implied with the concept, “carcinogen as mutagen” (Ames et al., 1973).

At this point, we have introduced several terms which need to be defined; namely, carcinogen; mutagen; initiator; and promoter.  In the process of defining these terms, the terms of mitogen, cytotoxicant, epigenetic and progressor will b=have to be defined.  Based on the fact that carcinogenesis ins a multi-stage, multi-mechanism process, the implied definition of a “carcinogen” is that it is an agent which “causes” cancer by interacting with a normal cell, transforming it through all the stages, via all of the different mechanisms, and altering all the required genetic factors needed to escape biological defense mechanisms against cancer.  The word, “carcinogen”, itself, creates the wrong impression in the minds of lay persons, risk assessors and scientists alike.  In other words, “carcinogen” does not reflect our current understanding of the multi-stage nature of carcinogenesis.  If, in fact, an agent is truly associated with the ultimate appearance of a cancer, the real question is “Which step in the multi-stage process of cancer did it influence and by what mechanism did it act?”  Possibly, we would do well if the word was banished from the language (Trosko et al., 1983b).

Now when an agent influences the carcinogenic process, it interacts with the organism/human ultimately at the cell level (after various routes fo entry and survival/chemical transformation via pharmokinetic action).  At this point, the biological consequence of the agent’s interaction at the molecular/biochemical level could be: mutagenesis; cell death; or epigenetic alteration (altered gene expression).

Knowledge of the types of mutations caused by oxygen free radicals in cells could define the contribution of mutagenesis by oxygen free radicals to carcinogenesis.  Efforts are required to document mutations produced in shuttle vectors that infect cells and more importantly in specific chromosomal genes.  The comparison of the types of mutations produced by oxygen free radicals in the test tube and the types of mutations found in genes associated with malignancy will go a long way in defining the role of oxygen free radicals in he induction of human cancer.

Acknowledgments

This work has been supported by grants from the National Cancer Institute: CA39903 (LAL), and CA08855 (TMR); by a Damon Runyon-Walter Winch ell Cancer Fund Grant: DRG 049 (KCC) and by a training grant from the National Institutes of Health: GM07266 (DF)