Although the descriptive and morphological characteristics of the stages of carcinogenesis are critical to our initial understanding of the pathogenesis of neoplasia, a complete knowledge of the molecular mechanisms of carcinogenesis may be necessary to control the disease through rational therapy, earlier diagnosis, and reasonable methods of prevention. However, our under- standing of the molecular mechanisms of carcinogenesis is incomplete. On the other hand, there has been an explosion of knowledge in this area during the past decade. From some of this knowledge, it has now become possible to establish some reasonably valid mechanisms for the stages of initiation and promotion. Some of these mechanisms for these two stages are given in Table 7.5.
While the morphological and biological characterization of the stage of initiation has been somewhat limited, mechanistic studies of this stage have been more extensive, especially in rela- tion to the metabolic activation of chemical carcinogens and the structure of their DNA adducts. As indicated earlier, however, the molecular mechanisms of this stage must conform to the ob- servable biological characteristics of this stage. At least three processes are important in initia- tion, including metabolism, DNA repair, and cell proliferation. Perturbation of any of these pathways has an impact on initiation. Just as initiated cells are difficult to distinguish morpho- logically and phenotypically from their normal counterparts, the molecular alterations responsi- ble for initiation may be equally subtle. As already indicated, initiating agents or their metabolites are mutagenic to DNA (see above). Thus, carcinogenic agents that are capable of initiating cells when administered at doses that do not induce neoplasia (incomplete carcinogen- esis) initiate cells in experimental models of multistage carcinogenesis (Boutwell, 1964; Dragan et al., 1994a). Furthermore, such subcarcinogenic doses of initiating agents may induce substan- tial DNA alkylation (Pegg and Perry, 1981; Brambilla et al., 1983; Ward, 1987) but do not ap- pear to induce karyotypic alterations.
More than three decades ago, mutational lesions were classified into two general groups, microlesions and macrolesions. While our knowledge of mutational events has increased since then, the classification is still generally applicable (Table 7.6). The stage of initiation appears to result from the formation of microlesions in DNA, while macrolesions are characteristic of the
Table 7.6 Classification of Mutations
Microlesions Base-pair substitutions (transitions, transversions) Frameshifts
Gene amplification (duplications) Chromosomal rearrangement, clastogenesis
Modified from Banks, 1971.
stage of progression. By definition (Chapter 9), direct induction of major chromosomal alter- ations in a cell can immediately induce the stage of progression directly, thereby bypassing the stages of initiation and promotion, provided that the altered cell remains viable and capable of replication. Sargent et al. (1989) demonstrated normal karyotypes of cells from altered hepatic foci in the stage of promotion in the rat, and Aldaz et al. (1987) showed that epidermal papillo- mas in the mouse that occurred early during promotion were diploid in character. A number of investigations have demonstrated specific point mutations in genes that are compatible with those induced in vitro by the adducts resulting from treatment with carcinogenic chemicals (cf. Anderson et al., 1992). The potential genetic targets for initiating agents have now been eluci- dated to some extent (Chapter 6). Individual variability, species differences, and organotropism of the stage of initiation are a balance of carcinogen metabolism, cell proliferation, and DNA repair.
The activation of proto-oncogenes and cellular oncogenes by specific base mutations, small deletions, and frameshift mutations results from DNA synthesis in the presence of DNA damage, including the presence of adducts. Methods for determining such alterations in speci- mens consisting of only a few hundred or a thousand cells have been available only during the last decade (cf. Komminoth and Long, 1993; Alard et al., 1993), permitting the analysis of muta- tions in specific genes potentially involved in the neoplastic transformation of DNA from very small samples.
The ras genes code for guanosine triphosphatases, which function as molecular switches for signal transduction pathways involved in the control of growth, differentiation, and other cel- lular functions (Hall, 1994). Table 7.7 lists a number of examples in rodent tissues of specific mutations in two of the ras genes, the Ha-ras proto-oncogene and the Ki-ras cellular oncogene. With the exception of mouse skin, the frequency of such mutations in preneoplastic lesions in experimental animals in the stage of promotion is about 20% to 60%. In multistage carcinogene- sis in mouse skin, the frequency increases to nearly 100% (Bailleul et al., 1989). In general, the mutations noted are those which theoretically could result from DNA adducts formed by the particular carcinogen.
Interestingly, spontaneously occurring neoplasms in mice also exhibit a significant inci- dence of point mutations in the ras proto- and cellular oncogenes (Rumsby et al., 1991; Can- drian et al., 1991), but neoplasms in corresponding tissues in other species do not necessarily exhibit activating mutations in proto- or cellular oncogenes (Tokusashi et al., 1994; Kakiuchi et al., 1993; Schaeffer et al., 1990). In addition, mutated ras genes have been described in normal- appearing mouse skin after DMBA or urethane application (Nelson et al., 1992). On the other hand, Cha et al. (1994) have reported that a very high percentage of untreated rats contained detectable levels of Ha-ras mutations in normal mammary tissue. Thus, mutations seen in neo- plasms in untreated animals may result from the selective proliferation of cells containing preex- isting mutations.
While several classes of genes appear appropriate as targets for DNA-damaging carcino- gens, the actual role of proto- and cellular oncogene mutations in establishing carcinogenesis is not entirely clear. In the earliest preneoplastic lesions studied (Table 7.7), only about one-third exhibit mutations in the ras gene family, but it is quite possible that other proto- and cellular oncogenes may be targets. Evidence that tumor suppressor genes may be targets for the initiation of early malignant development comes largely from studies of genetically inherited neoplasia. In these rare hereditary cancers, one of the alleles of a tumor suppressor gene contains a germline mutation in all cells of the organism (Paraskeva and Williams, 1992; Knudson, 1993), thus, in theory, initiating all cells in the organism. However, the exact gene(s) critical for the initiation of a specific cell type after administration of a specific carcinogen has not yet been defined in any instance. Furthermore, as noted from Chapter 5, other genes may be involved in the stage of
initiation, although it is likely that alterations in one or a very few genes constitute the “rate- limiting step” in the initiation of any individual cell.
Boutwell (1974) was the first to propose that promoting agents induce their effects through their ability to alter gene expression. During the past decade, our understanding of mechanisms in- volving the alteration of gene expression by environmental agents has increased exponentially (Morley and Thomas, 1991; Rosenthal, 1994). The regulation of genetic information is mediated through recognition of the environmental effector—hormone, promoting agent, drug, etc.—and its specific molecular interaction with either a surface or a cytosolic receptor. Several types of receptors exist in cells (Mayer, 1994; Pawson, 1993; Strader et al., 1994) and are depicted in Figure 7.6. Plasma membrane receptors may or may not possess a tyrosine protein kinase do- main on their intracellular region (Figure 3.25B), while others have multiple transmembrane do- mains with the intracellular signal transduced through G proteins and cyclic nucleotides (Figure 3.25A) (Mayer, 1994; Collins et al., 1992). The other general type of receptor mechanism in- volves a cytosolic receptor that interacts with the ligand (usually lipid-soluble) that has diffused through the plasma membrane (Figure 3.24). The ligand-receptor complex then travels to the nucleus before interacting directly with specific DNA sequences known as response elements (Figure 7.6).
The figure shows in a highly simplified manner the cascade effect of receptor-ligand and protein kinase interactions, resulting in changes in transcription as well as cell replication within the nucleus. As shown in the figure, interaction of transmembrane receptors containing a ty- rosine protein kinase domain involves initially their dimerization, induced by ligand interaction. This activates the protein kinase domain of the receptor, causing autophosphorylation. This, in turn, attracts a cytoplasmic complex, Grb2-Sos, to the plasma membrane (Aronheim et al., 1994; Egan et al., 1993). Sos is a member of a family of regulatory proteins termed guanine-nucleotide exchange factors (GEFs) (Feig, 1994). Sos association with the G protein Ras stimulates, along with other protein interactions, the exchange of GDP with GTP on the Ras α subunit. The GTP- Ras, in turn, interacts with B-raf, a cytoplasmic serine-threonine protein kinase, with subsequent activation of its catalytic activity and initiation of a kinase cascade, ultimately resulting in the phosphorylation and activation of transcription factors including Jun, Fos, Myc, CREB, and ulti- mately E2F and Rb, the tumor suppressor gene (Lewis et al., 1998; Janknecht et al., 1995; Rous- sel, 1998). A number of other transcription factors are also activated by other pathways involving phospholipase C, phosphatidylinositol kinase, and protein kinase C (Vojtek and Der,1998; Takuwa and Takuwa, 1996).
As indicated in Figure 7.5, the critical, rate-limiting step in this process is the mediation of the signal through the G-protein family. The G proteins are targeted to the plasma membrane through lipid moieties, both isoprenoid and fatty acyl, covalently linked to the carboxyl terminal region of the protein (Yamane and Fung, 1993). In this way, the initiation of the signal by the ligand-receptor interaction can be physically related to the rate-limiting G-protein activation step. The activation cycle of the G-protein family is seen in Figure 7.7. As seen from the figure, activation is accompanied by GTP binding to the α subunit of the G protein, such binding being dramatically stimulated by GEF proteins such as Sos. Activation also involves dissociation of the α from the β and γ subunits, allowing the α subunit to interact with and activate downstream members of the pathway, a protein kinase in the case of the growth factor-related pathway (Fig- ure 7.6) or with other membrane molecules such as adenyl cyclase in the case of multiple trans- membrane domain receptors (Figure 7.6). The activated G protein has an extremely short half- life because of the action of RGS (regulator of G-protein signaling) proteins, which stimulate
Figure 7.6 Diagram of principal mechanisms of intracellular signal transduction initiated either within the cytosol or at the plasma membrane. Lipid-soluble molecules such as steroid hormones interact directly with a cytoplasmic receptor after passing through the plasma membrane, this receptor being in association with heat shock protein 90 (HSP 90). Following association and dissociation with different protein species, the ligand receptor complex enters the nucleus and interacts with DNA in association with specific DNA sequences known as hormone response elements (HRE). Receptors for many polypeptides hormones (Chapter 3) and growth factors (Chapter 16) have a single transmembrane domain and associate or dimer- ize after interaction with its ligand. This association activates the tyrosine kinase activity of the cytoplasmic component of the receptor, phosphorylating both itself and other proteins. In the figure, an intermediate protein, Sos, complexes with the G protein, Ras, which interacts with a protein kinase, B-raf, activating its kinase activity and initiating the kinase cascade shown in the figure, ultimately ending up with activation of transcription factors, in this case the E2F class. The G protein–linked receptors have multiple transmem- brane domains, and interaction with the ligand results in the activation of a membrane-associated G protein, which in turn activates adenylcyclase (AC) to produce increased levels of cyclic AMP. This small molecule, termed a second messenger, activates the cyclic AMP–dependent protein kinase (PKA), which in turn phosphorylates several other substrates, ultimately including the cyclic AMP response element (CRE)–binding protein (CREB). This protein interacts with its DNA response element, CRE, resulting in enhanced transcription. MAPK, mitogen activated protein kinase; MEK, MAPK kinase; RSK, ribosomal S6 kinase. p91ARF and E2F are specific transcription factors.
Figure 7.7 The guanine nucleotide cycle for heterotrimeric G proteins. The inactive G protein on the left is a GDP-bound heterotrimer. Ligand-bound activated receptors, either directly or indirectly, catalyze the exchange of GDP for GTP, leading to association of the α subunit from the βγ dimer (right). The separated G-protein subunits are thus activated for signaling. The G protein becomes an inactive heterotrimer again following action of the intrinsic GTPase activity of the α subunit returning it to the GDP bound form. RGS proteins can bind to activated G-protein α subunits and greatly accelerate the GTP-hydrolysis reaction. Pi,
phosphate. (Adapted from Koelle, 1997, with permission of author and publisher.)
GTP hydrolysis to GDP with subsequent reassociation of the G protein in its inactive state (Koelle, 1997). Some receptors having a single transmembrane domain do not possess tyrosine protein kinase activity but require interaction with a cytoplasmic protein kinase. The exact mechanism of this interaction, which does not involve protein phosphorylation, is not clear at the present (Silvennoinen et al., 1997). The kinases involved have a direct interaction with the cyto- plasmic domain of the receptor and are termed Jak or janus (two-faced) kinases. These in turn may cause phosphorylation of the tyrosines in the cytoplasmic region of the receptor with subse- quent activation via the pathway seen in Figure 7.6. In some cells, especially those of the im- mune system, the Jak kinases may cause phosphorylation of specific transcription factors known as STATs (Chapter 19; Silvennoinen et al., 1997).
The multiple transmembrane domain receptors are in direct association with G proteins and, on activation of the receptor by interaction with a ligand, may activate a kinase termed a G- protein receptor kinase (GRK) or another effector such as adenyl cyclase (AC, Figure 7.6) (Böhm et al., 1997; Rasenick et al., 1995). While the α subunits of G proteins are major factors in this signaling pathway, the β and γ subunits also appear to be involved (Pitcher et al., 1992).
In addition to the plasma membrane receptors, gene expression can be regulated through the interaction of cytoplasmic receptors with their ligands as previously discussed in Chapter 3 (Figures 3.23 and 3.24). Just as with membrane receptors, the pathways of the cytoplasmic re- ceptors involve multiple interactions with proteins, phosphorylation, and ultimate alteration of transcription through factor interaction with DNA (Weigel, 1996; Pratt and Toft, 1997). In all of these pathways, in addition to alteration of transcription and gene expression, enhancement or inhibition of cell replication may also be an end point achieved through transcriptional modula- tion of the cell cycle (Chapter 9).
Many promoting agents exert their effects on gene expression through perturbation of one of the signal transduction pathways, as indicated in Figure 7.6. One may, in general, classify receptor mechanisms into three broad classes—steroid, tyrosine kinase, and G protein–linked. The majority of the more commonly studied promoting agents exert their actions by mediation of one or more of the receptor pathways indicated in Figure 7.6. In Table 7.8 are listed some of
Adapted from Pitot and Dragan, 1996, with permission of the publisher. Further references may be found in the text. Cyclosporine as a promoter of murine lymphoid neoplasms has been described by Hattori et al. (1988).
the most studied promoting agents known or postulated to be effectors in signal transduction pathways. While protein kinase C (PKC) is not itself one of the three types of receptors noted in Figure 7.6, it is a mediator of the signal transduction pathways of both the tyrosine kinase and G protein–linked transduction pathways. TPA interacts directly with membrane-bound PKC, dis- placing the normal activator, diacylglycerol, and serving to maintain the kinase in its active and soluble form (Ashendel, 1985). The continual activation of this kinase then stimulates further transduction pathways by phosphorylation of specific proteins (Stabel and Parker, 1991). TCDD acts in the steroid pathway via a specific receptor, the Ah receptor, the ligand-receptor complex together with other proteins, ultimately altering the transcriptional rate of genes possessing spe- cific regulatory sequences (HRE). In a similar manner, sex steroids, some synthetic antioxidants, and peroxisome proliferators interact with specific soluble receptors and altered gene expression by presumed mechanisms similar to that of TCDD (Chapter 3, Figures 3.23 and 3.24). While in some instances the actual receptor is still not defined, those for polypeptide hormones and growth factors consist of either the tyrosine kinase or G protein–linked types depending on the structure of the polypeptide. The “receptors” for okadaic acid and cyclosporine have been re- ported to be protein phosphatase 2A and cyclophilin-A respectively (Fujiki and Suganuma,1993). These proteins, like PKC, are involved in phosphorylation mechanisms of the tyrosine kinase and G protein–linked pathways, although specific sites and mechanisms have not been completely clarified at this time. Thus, the action of promoting agents in altering gene expres- sion may be mediated through specific receptors. This hypothesis provides a reasonable expla- nation for the tissue specificity demonstrated by many promoting agents. The receptor-ligand concept of promoting agent action is based on the dose-response relationships involving phar-macological agents. The basic assumptions of such interactions argue that the effect of the agent is directly proportional to the number of receptors occupied by the ligand. The intrinsic activities of the chemical and the signal transduction pathways available in the tissue are important factors in the type and degree of response observed (Saltiel, 1995).
Indirect Pathways of Tumor Promotion. In addition to the specific action of tumor pro- moters on gene expression mediated directly via specific ligand (promoting agent)-receptor in- teractions, another mechanism of tumor promotion has been identified that involves the indirect action of promoting agents in inducing the formation of reactive oxygen radicals (Chapter 3), which in turn involve signal transduction pathways specifically activated by such “oxidative stress.” Cells may also be subjected to other types of stress involving changes in temperature, pressure, substrate availability, etc., but the roles of these mechanisms in tumor promotion, if any, have not been clarified. On the other hand, reactive oxygen radicals, whether developing internally within the cell from the indirect action of chemical agents or directly from the action of chemical and physical agents, alter a number of signal transduction pathways. One mecha- nism for such alteration appears to be a change in the redox state of several of the protein species themselves (Cimino et al., 1997). Cerutti and colleagues (Cerutti, 1985) were among the first to argue that reactive oxygen species played a major role in tumor promotion. Since that time, it has become apparent that many extracellular ligands generate and/or require reactive oxygen radicals and their derived species in the transmission of some of their signals to the nucleus (Lander, 1997). Such pathways include those involving growth factors and hormones, ion trans- port, transcription factors, and apoptosis (Lander, 1997). In particular, the oxidation state of tran- scription factors is of paramount importance in the effects of free radicals, especially of oxygen metabolism, and redox potential in governing signal transduction pathways. In Figure 7.8 may be seen several examples of the effect of reactive oxygen species (ROS) in regulating the func- tion of several transcription factors (cf. Cimino et al., 1997). Notable in the figure are zinc-finger transcription factors such as the AP-1 family of transcription factors, which are involved in a variety of signal transduction pathways including that from TPA, ultraviolet-A, TCDD, and ion- izing radiation (cf. Dalton et al., 1999). Another transcription factor noted in the figure (B) is NF-κB, which plays a central role in the regulation of many genes involved in cellular defense mechanisms, pathogen defenses, immunological responses, and expression of cytokines and cell adhesion molecules (Dalton et al., 1999). As shown in the figure, NF-κB consists of two sub- units, p50 and p65, which are restricted to the cytoplasm because of their association with an inhibitory subunit, I-κB. NF-κB translocates to the nucleus following phosphorylation of I-κB which frees NF-κB from the complex. This allows NF-κB to interact with specific sequences in DNA (cf. Cimino et al., 1997). Such interaction with DNA is markedly altered by oxidation of the proteins, as is the phosphorylation of I-κB. As noted in C of the figure, heat-shock proteins (HSF) may also be altered in their function as chaperones and in the ability of the cell to cope with external stress by alteration in the ROS (cf. Cimeno et al., 1997). There is also substantial evidence that ROS may alter the function of a variety of kinases involved in signal transduction including protein kinase C (Konishi et al., 1997) and the mitogen-activated protein kinase (MAPK) family (Guyton et al., 1996). Even G proteins themselves, such as members of the ras family, play a significant role in the signal transduction effects resulting from ROS, sometimes themselves being important targets for redox changes (cf. Finkel, 1998).
Promoting agents, which themselves are responsible for the direct induction of reactive oxygen molecules (e.g., growth factors, phorbol esters), induce the production of ROS in a vari- ety of cell types. Some examples of this are seen in Table 7.9. The table was restricted to non- phagocytic cells and is modified from a similar table by Gamaley and Klyubin (1999). Thus, the specific signal transduction pathway resulting from the specific interaction of ligand with recep- tor is complemented by the mostly indirect production of ROS, which, in turn, may enhance the
Figure 7.8 Models of the mediation by ROS in the regulation of transcription factors. A. A zinc-finger transcription factor (Zn) and its alteration by ROS preventing its interaction with DNA in the oxidized state. B. NF-κB (p65-p50) and I-κB association in the cytoplasm indicating areas of change induced by ROS prior to entrance into the nucleus. C. Heat-shock proteins (HSF) and their interaction with each other and other proteins as chaperones (not shown) indicating the potential action of ROS. (Adapted from Cimino et al., 1997, with permission of the authors and publisher.)
induced alteration of genetic expression by the promoting agent. The importance of the induc- tion of ROS by promoting agents is perhaps best noted by the effect of antioxidants and related chemical species capable of significantly inhibiting the promoting action of many promoting agents (Chapter 8). In addition, since ROS are capable of altering the structure of DNA in a variety of ways (Chapter 3), this effect may also play a role in the action of promoting agents. However, that such is unlikely is seen by the reversible nature of the action of promoting agents (see above), in contrast to the irreversibility of the induction of mutations by ROS. Furthermore, as the student may recall from Table 3.6, extensive endogenous DNA damage resulting from ROS and other factors is occurring all of the time. Repair of such damage is largely the function of the mismatch repair system also discussed in Chapter 3. Although theoretically ROS may initiate cells, there has been little if any evidence that such occurs (Glauert and Clark, 1989; Denda et al., 1991; Takaba et al., 1997). It is more likely that structural alterations induced by ROS during the action of promoting agents may play a major role in the transition of cells from the stage of promotion to that of progression (Chapter 9).