• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Aromatic Amines in Experimental Cancer Research: Tissue-Specific Effects, an Old Problem and New Solutions

Posted on: Friday, 20 April 2007, 06:00 CDT

By Neumann, H G

Carcinogenic aromatic amines usually produce tumors in specific target tissue, such as 2-acetylaminofluorene (AAF) producing liver tumors in rats, in contrast to some other structurally related arylamines. A hypothesis is presented that explains the mode of action in this rat liver model. Genotoxic and nongenotoxic effects work together and make AAF a complete rat liver carcinogen. The cytotoxic, promoting effects are particularly important. N-Hydroxy- 2-aminofluorene and 2-nitrosofluorene, two metabolites of AAF, are able to uncouple the mitochondrial respiratory chain. They entertain a redox cycle that removes electrons from the respiratory chain and impairs ATP production. The dose-dependent opening of the mitochondrial permeability transition pore signals the viability of the cell. If the pore is opened to a certain extent, the cell is eliminated by apoptosis. As a consequence, oval cells proliferate, and as this process is overloaded, the liver transforms into a cirrhosis-like situation and thus provides the conditions under which initiated liver cells develop tumors. Such an interpretation is based on assumptions that have been debated for a long time. Some of these often forgotten developments are reviewed in support of the hypothesis, which allows a more comprehensive view of the complex in vivo situation at a time when in vitro models prevail.

Keywords Apoptose, Aromatic Amines, Liver Tumors, Mitochondria, Tissue Specificity

1. INTRODUCTION

In the last 50 years experimental cancer research made remarkable progress. Many hypotheses have been developed to explain the carcinogenic properties of chemicals, and most researchers agree that success in cancer therapy and prevention depends on the understanding of the multistage process at the molecular level. Experimental methods have changed over the years and influenced the main stream of interest. Looking back, it becomes clear that many contradicting statements were left unresolved. One of the unresolved questions regards the tissue-specific effects of carcinogens. Despite early suggestions to explain this phenomenon, it is quite clear that the answer depends on a fundamental understanding of the cancer process.

The history of dealing with tissue specificity corresponds with the history of cancer research, and the studies with aromatic amines acting on rat liver in particular are considered representative for essential steps and new concepts in experimental cancer research. In Table 1, milestones are compiled that mark the way to a more comprehensive understanding of the cancer process.

Progress was made in several stages reaching from pathology to molecular biology. It all began with the production of tumors in experimental animals, with pathology as the leading method. The authors of the first monograph about "Biochemistry of Tumors" (von Euler and Skarzynski, 1942) stated that "applying chemical carcinogens in research will open quite new insights into the pathogenesis of malignant cancer and into the biochemical problems associated with malignant growth." The study of structure-activity relationships dominated the second stage. Investigators tried to find the features common to carcinogens that could be responsible for tumor formation with uncountable experiments using myriads of partly newly synthesized chemicals. When it was detected that in many cases only reactive metabolites are active, interest turned to metabolic activation in the third stage. In the fourth stage the somatic mutation theory was widely accepted to explain the formation of tumors. Many correlations between DNA lesions and tumor formation were found, and research concentrated on the search for the "critical" reactions with DNA as the premutagenic lesion that leads to an irreversible step in the process. Presently molecular biological methods are expected to help to unravel the complex signals controlling growth, survival, and death of cells and organs in a world of cancer risk factors.

TABLE 1

Milestones on the way to new insights into the formation of tumors by carcinogenic aromatic amines

Studies in this laboratory tried to answer the simple but still unresolved question of organotropism. How can one explain why chemicals produce tumors mostly in specific tissues (Neumann, 1970, 1972)? Answers to this question are relevant practically for risk assessment of chemicals and fundamental theoretically for the understanding of cancer. We obtained new insights using as a tool 2- acetylaminofluorene (AAF), one of the most studied carcinogens, and the rat liver as a model tissue (Bitsch et al., 2000; Klhn et al., 2003)

2. RAT LIVER TUMORS FROM AAF

2.1. Aromatic Amines

Aromatic amines are known for their tissue-specific effects. Aniline workers in the dyestuff industry developed bladder tumors, which later were found to be caused by 2-naphthylamine and 4- aminobiphenyl present as side products and impurities of the process. 7ra/is-4-Dimethylaminostilbene is an extreme example for such tissue specificity. In animal experiments it produces selectively tumors of Zymbal's gland, a sebaceous gland in the external ear duct of rats. Stimulated by the work of Jensen and collaborators, who detected the hormone receptor for estrogens in target tissues (Jensen and Jacobson, 1960; Jensen et al., 1966), we tried to find a similar accumulation of the carcinogen in its target indicating a tissue-specific exposure. Results could only be expected with low doses, and extremely highly labeled test chemicals were required (Neumann, 1967). We selected three structurally related but biologically different-acting compounds for comparison: carcinogenic trans4-dimethylaminostilbene (trans-OAS), the isomeric but not carcinogenic c/i-4-dimethylaminostilbene (c/s-DAS), and 4dimethylaminobibenzyl (DABB), which is also not carcinogenic in the rat (Neumann, 1970). Distribution of radioactivity in various tissues aand binding of reactive metabolites to proteins, RNA, and DNA were used as a measure of the biologically active dose. This was always higher for frani-DAS than for cis-OAS and DABB. This was the expected result. However, it turned out that the biologically active dose was in most other tissues higher than in Zymbal's gland. It was highest always in liver, followed by kidney, lung, and glandular stomach (Baur and Neumann, 1980; Hilpert and Neumann, 1985). According to the existing paradigm, tumors should have developed predominantly in liver and kidney. We concluded that DNA lesions could not explain tissue specificity; they might be necessary but not sufficient.

After that, we turned the question around and asked: Why does irans-4-DAS not produce liver tumors? The closest answer at that time would have been to assume that the DNA lesions were unspecific, that is, that the critical reaction did not happen. We disproved this assumption, because tumors were found in livers of such animals if an initiating treatment with AAS was followed by partial hepatectomy or with typical liver tumor promoters, such as phenobarbital or DDT (Hilpert et al., 1983) (Table 2).

TABLE 2

Incidence of liver tumors combined from two initiation-promotion experiments with arylamines in new borne Wistar rats

For a new approach we compared three other, structurally related arylamines with different effects on rat liver: trans- 4acetylaminostilbene (AAS), which enters the same metabolic pathways as DAS, and 2-acetylaminophenanthrene (AAP, Scribner and Koponen 1979; Scribner et al., 1983), both of which are not carcinogenic in rat liver, and 2-acetylaminofluorene (AAF), a typical rat liver carcinogen. It turned out that all three carcinogens "initiate" liver cells, but only AAF has sufficient "promoting potential" (Figure 1). The question was then: What makes AAF a complete liver carcinogen with initiating and promoting properties? AAF acts on many parameters at the biochemical and the molecular level, but it is still difficult to correlate these different endpoints with the cancer process.

2.2. Nitrosofluorene, a Reactive Metabolite of AAF

Nitrosofluorene (NOF) has long been known as a metabolite of AAF, but interest declined considerably when it was found not to react directly with DNA (Kriek, personal communication). Nevertheless, Hecker et al. (1968) described the carcinogenic properties of NOF. Tumors were seen in 4 out of 10 animals after oral administration: 2 in the Zymbal's gland, 1 in the liver, 2 in the forestomach with metastases in the liver, and 1 adenocarcinoma in the intestine. There were only a few animals in this experiment, but the local tumors in forestomach and intestine indicate a possible role of NOF or an effective metabolite that despite of their chemical reactivity must be systemically available. yV-Hydroxy-AAF can be oxidized nonenzymatically to a nitroxyl radical that disproportionates into NOF and W-acetoxyN-2-AAF; that is, three potentially reactive compounds were formed. A simplified scheme of AAF metabolism is shown in Figure 2.

NOF adds to C=C double bonds like those in unsaturated fatty acids. A nitroxyl radical adduct is generated, whose formation correlates with mutagenicity in the Salmonella typhimurium test and can be demonstrated to occur in lipid membranes (Hampton et al., 1980).

If N-acetoxy-AAF or N-hydroxy-AAF is tested for mutagenicity in V79 cells in the presence of deacetylase inhibitors, the mutation rate i\s strongly reduced, which indicates that the deacetylated compounds are responsible for the biological effect. These are W- hydroxy-2-aminofluorene and NOF, which form a redox couple (Glatt and Oesch, 1985). NOF was significantly more toxic than the hydroxamic acid in AS52 cells (Bitsch et al., 1997) (Figure 4).

The acetamides need metabolic activation to become genotoxic, but their mutagenic potential appears to be rather similar (Figure 3). It appears now that NOF plays a much bigger role in the cancer process than expected so far. Moreover, analysis of hemoglobin adducts formed by the reaction of NOF with SH groups in hemoglobin demonstrates that AAF is metabolized in humans to NOF in substantial amounts. The analysis of hemoglobin adducts in blood samples is used to assess the environmental exposure of humans to AAF and 2- nitrofluorene, which is commonly present in automobile exhausts and seems to be a side product in most pyrolytical processes (Neumann etal., 1995, Zwirner-Baier and Neumann, 1999).

2-Nitrosofluorene (NOF) attracted new interest when we observed increased oxygen consumption in isolated perfused livers from AAF- treated rats (Ambs, dissertation; Neumann et al., 1994). This directed our attention to mitochondria as a possible target of AAF toxicity. It turned out that AAF uncouples the mitochondrial respiratory chain very effectively (Klhn et al., 1996; Klhn and Neumann, 1998). NOF modulates the opening of the permeability transition pore (Klhn and Neumann, 1997; Klhn et al., 1998), and Bcl2, an antiapoptotic component of the pore, may be overruled by proapoptotic Bax under oxidative stress (Figure 5).

FIG. 3. Mutation frequency in AS52 cells induced with and without metabolic activation. Genotoxic potency is comparable with metabolites from all three aromatic acetamides (Bitsch etal., 1997).

FIG. 4. Dose dependence of cell survival in AS52 cells, indicating that NOF is much more toxic than the hydroxamic acid N- hydroxy-acetylaminofluorene (Bitsch et al., 1997).

2.3. The Hypothesis

Further investigation of the uncoupling properties of AAF and our interpretation of the results lead to a new comprehensive hypothesis about the carcinogenic process induced in rat liver by AAF. According to this hypothesis, rat liver is stressed by the uptake of carcinogenic doses of AAF, but has the possibility to adapt to the situation up to a certain extent. If defense is exhausted, overstressed hepatocytes are eliminated by apoptosis and replaced by new cells, but replacement does not proceed normal. Oval cells proliferate, but the balance of differentiation into hepatocytes and bile duct cells is shifted toward more bile duct cells. The increased production of fibers alters the microstructure of liver, and microcirculation is disturbed in a cirrhosis-like situation. Only under these circumstances do foci consisting of altered cells begin to grow. These foci of enzyme-altered cells contain some irreversible alterations and are more resistant to the regulating signals from their immediate environment. They are considered to be initiated and promotable cells.

The following oberservations support this hypothesis: Numerous glutathione-5-transferase P-positive foci consisting of only a few cells appear from the onset of treatment throughout the liver. In mitochondria, oxidative phosphorylation by the respiratory chain is lowered. This is considered to be due to toxic metabolites of AAF that impair homeostasis in hepatocytes by oxidative stress in a more common sense; that is, the redox couple NOF/N-hydroxy-AF removes electrons from the respiratory chain, the membrane potential changes, and the transition pore opens. Important for the impact on the process is that opening of the permeability transition pore induces apoptosis. Cells are eliminated preferentially in the periportal area and are replaced. Beginning at the periportal area, fiber-producing cells spread around the liver substructures, leading to fibrosis, and only when a cirrhosis-like structure is seen do the foci begin to grow.

FIG. 5. Schematic representation of redox cycling of 2-NOF in mitochondria. Oxidative phosphorylation is partially uncoupled in the mitochondrial respiratory chain by altering the transmembrane electrochemical proton gradient (Neumann et al., 1997).

Two independent events are postulated to work together. Cells are initiated by an unknown, irreversible alteration and gain the potential to grow into a tumor. This depends on a situation resulting from toxic effects affecting the differentiation of stem cells. The hypothesis includes a number of assumptions about issues that were controversial in the past, and some of them are controversial even today:

* The metabolism of AAF.

* The two-stage model of carcinogenesis, which started the discussion about tumor initiation and promotion.

* The role of DNA lesions and the potential of oxygen to produce them.

* The role of toxicity.

* The role of proliferation of oval cells in particular.

* The concept of the resistant hepatocyte.

* The role of mitochondria.

* The role of apoptosis and necrosis.

These issues are next discussed. They have been selected primarily under the perspective to support our hypothesis.

3. THE PROCESS OF TUMOR FORMATION

3.1. Early Considerations

In the first monograph already mentioned in the introduction (von Euler and Skarzinsky, 1942) attention was given primarily to polycyclic aromatic hydrocarbons. 3,4-Benzo[a]pyrene was the only component of coal tar whose chemical structure was known at that time. Later a countless number of related chemicals was synthesized. The results of structureactivity studies, however, remained unsatisfactory, partly because metabolic activation was not sufficiently accounted for, but also because interest was limited to the genotoxic activity of the chemicals. Already in this early monograph it was stated that after higher doses and chronic treatment polyaromatic hydrocarbons (PAH)s induce hepatitis, which may progress to liver cirrhosis. These chemicals obviously were cytotoxic rather than growth-stimulating. The authors see two different modes of action, direct and unspecific toxicity leading to necroses followed by specific late effects responsible for malignant growth. The general effects on organs or the whole organism received little attention, with the exception of the immune system.

Cox et al. (1947) described the typical features after AAF feeding. Tumors were produced in many rat tissues, but predominantly in the liver. Cox et al. observed multiple nodular hepatocytes, clearly different from normal cells. Other nodules were described as bile duct-like structures and cysts separated from each other by connective tissue.

Wolf (1952) published the monograph "Chemical Induction of Cancer" 10 years after von Euler. At this time aromatic amines were brought into focus and the rat liver model was studied widely. In this context AAF became the most studied aromatic amine. Before that azo dyes were investigated, and these are cleaved metabolically into aromatic amines. Many of them were of practical interest, such as butter yellow (4-dimethylaminoazobenzene). One of the differences between PAHs and aromatic amines was that aromatic amines barely produced skin tumors after subcutaneous application but produced tumors in various tissues after oral administration. Wolf explained this difference with a much more complex process than the mere contact with the chemical delivered by circulating blood.

Wolf discussed in this monograph the mechanistic concepts of the time. He preferred the hypothesis of Haddow (1948), which dates back to 1938. Haddow detected the growthinhibitory effects of carcinogens and concluded that they inhibit growth by creating an unfavorable environment for the cell. This prompts adaptive processes leading to irreversible changes that give rise to a new cell line more resistant to the toxic environment. The concept of adaptation originates from studies about adaptive transformations in bacteria. In any case a causal relationship between growth inhibition and carcinogenic activity was postulated. Haddow supported this idea with the observation that trans-4-DAS, which was considered to be introduced into clinical trials as a tumor inhibitor is itself carcinogenic. It should be noted that nobody would have thought to test this chemical for carcinogenicity without the theoretical consideration that if carcinogens inhibit growth, then growth inhibitors should be carcinogenic.

The same year, Price et al. (1952) published data about early effects seen after 4 weeks of feeding 3'-methyl- 4dimethylaminoazobenzene to rats, showing a severe increase of bile duct-like cells in liver not seen with butter yellow. Tumors developed in these areas of cholangiofibrosis.

The effects of aminostilbene derivatives-originally developed as proliferation inhibitors-were first tested on tumor cells in vitro. Abnormal mitotic figures were observed, and Koller (1955) noted that tumor cells are selectively killed by fibrosis and are overgrown by histiocytes.

Mhlbock (1957) discussed chronic irritation and inflammation with hyperplasia as a possible cause of tumor formation. He used an interesting example to make that point. In certain mouse strains in cysts and tumors of the lower jaw hair was always found that had penetrated the paradental connective tissue and had caused inflammation. The author delineated the sequence: penetration of hair-inflammation-formation of cysts-tumor.

Walpole argued about the controversial two-stage hypothesis and tried to classify the multiplicity of carcinogens known in the meantime at the influential "Ciba Foundation Symposium, Mechanisms of Action" (1959). The process was divided into an initiation and a promotion phase. The controversy immediately began how to separate initiators and promoters. Some contradictions were explained by the notion that pure promoters could no\t be proven as such because in every organism initiated cells will exist, which should yield tumors after extensive promotion. Walpole expressed his opinion that chemicals exist that have both tumor-initiating and tumor-promoting properties, a view that has been repressed by many up to now, because interests were focused so much on the genotoxic effects. AAF is a good example, and the two properties can now be studied separately.

The two-stage model was originally a mouse skin model, but could readily be transferred to the rat liver. In both cases a certain latency period is necessary during which frequently hyperplastic alterations can be seen. Such alterations are found always in rat liver, irrespective of the liver carcinogen (Farber, 1956). One of the first initiation-promotion experiments was performed with AAF and the thyroid gland in rats. AAF alone does not produce thyroid tumors, but they develop if typical goitrogenic measures are taken afterward (Bielschowsky, 1944; Hall, 1948).

3.2. Approaches to the Cancer Problem

Among numerous approaches to explain the mode of action of carcinogens, two received more attention: the depletion theory and the search for minimal deviation tumors. Beginning with the observation that many carcinogens bind to proteins, it was an attractive assumption that certain enzyme systems involved in growth regulation could be irreversibly inactivated (Heidelberger, 1959; Kaplan, 1959; Custorand Sorof, 1984).

Another idea came from the observation that enzyme activities in experimental tumors were quite variable. Particularly enzymes of the glucose degradation seemed to be induced as part of adaptation. By comparing rapidly and slowly growing liver tumors it was expected to find enzyme patterns minimally deviating from normal tissue but common to all tumors and therefore characteristic for tumor development. The analysis of numerous transplanted tumors revealed that no tumors were alike (Morris, 1963, 1965).

In a surprisingly comprehensive book titled The Cancer Problem. Introduction Into General Pathology for Students, Physicians and Natural Scientists, K. H. Bauer (1963) wrote that as a target for the critical reaction a cellular regulator is postulated for cell division and the inner order of the cell. He envisaged that a heritable regulatory mechanism is impaired, but he also sees a connection between proliferation rate and cancer susceptibility. Later on he wrote, "Accordingly, the formation of tumors would result from a sudden transformation of body cells into tumor cells following a multiphasic precancerous tissue damage." With this statement Bauer accepts the somatic mutation theory, which may explain almost all symptoms of the process, but also reserves for precancerous tissue damage or preneoplasia an important place. Considering occupational cancers and the mimicking experimental approaches Bauer meant: "Not every precancer results in cancer, but almost every cancer has its precancer." He also said, "Irritations leading to tissue regeneration improve the chances for critical alterations in the cancer process, namely dedifferentiation with concurrent release of inhibition of cell division." According to this view, the specific event that makes a tumor cell occurs only after the tissue has already been damaged. It is worth remembering such views, since monocausal approaches are still very common.

4. SOME OF THE BASIC QUESTIONS

4.1. The Role of DNA

DNA was in the center of interest at the esteemed international symposium on "Physico-Chemical Mechanisms of Carcinogenesis" 1968 in Jerusalem (Bergmann and Pullman, 1969). Reactions of reactive metabolites with nucleic acids and their role in carcinogenesis were discussed in great detail. Charles Huggins concluded his contribution by saying, "Cancer is caused by a specific false genome" (Pataki and Huggins, 1969). Most contributions dealt with the search for either the critical metabolite, the critical reaction product, or the specific and critical DNA lesion. Countless correlations between chemical structure and carcinogenicity were described. Modifying factors were considered, but only in view of their capability to influence the biologically effective dose. Warwick (1969) said: "Many factors like sex, species, levels of protein intake, levels of vitamins and so on, which can affect carcinogenesis, in the case of 4-dimethylaminoazobenzene are probably operating simply on the basis that they modify the concentration or site of formation of the ultimate carcinogen." All the participants seemed to look for the specific event, expecting that there is a critical lock (or mutation) for which the right key has to be found.

However, Miller and Miller (1969) warned that with the knowledge about metabolic activation and the reaction products with DNA only a rough picture of the chemical properties of the carcinogen is obtained, and the critical biochemical events on the genetic as well as the epigenetic level of a cell exposed to an ultimate carcinogen are thereby not characterized.

Most agreed at that time that genotoxic properties of carcinogens are the cause for tumor initiation and this notion was particularly supported by the detection of point mutations that either activate oncogenes or inactivate tumor suppressor genes. Some, however, warned that not all available information went into the construction of mechanistic models. Rubin (1980) asked in an editorial: Are somatic mutations the most important cause of a malignant transformation? Even if a neoplastic transformation were initiated by a mutation, this would not describe the whole process. By the end of the 1970s carcinogenesis was considered to be the result of a DNA lesion produced by a genotoxic carcinogen in the chromosome of a somatic cell. If the reaction hits the right place, this starts the clonal expansion of a tumor cell.

Cairns (1981) criticized this simple explanation and expressed his opinion that most human tumors are not caused in this conventional way by one or more mutations, rather than by genetic transposition, in which whole DNA sections are rearranged. Subsequent promotion would be necessary, because either the critical mutation is recessive (and can be demonstrated only after deletion or somatic recombination), or a repressor gene is involved (whose expression can only be demonstrated if a supply of repressor molecules is exhausted).

Cairns mentioned several questions that cannot be answered by a point mutation: Why don't tumors develop more rapidly and why do they need such a long latency period? Why does an initiated (cancer) cell not immediately give rise to a clone? Why do xeroderma pigmentosum patients develop, whose DNA repair is inhibited only tumors in skin after ultraviolet (UV) irradiation and not in tumors in other tissues with genotoxic chemicals? According to Cairns, these questions could not be answered at that time because the whole process was not sufficiently understood. He considered local DNA alterations a possible cause for tumors in the human population, but estimated its contribution to be low.

Meermann et al. (1994) analyzed the reaction of numerous arylamines with DNA and concluded that the formation of nonacetylated C-8-adducts of deoxyguanosine correlates with their initiating effects, whereas the acetylated adducts are clastogenic, which correlates with their promoting properties. The formation of micronuclei in liver was used to demonstrate the clastogenic effects. This difference was explained by the fact that the adducts modify the DNA conformation and consequently DNA repair differently. This interpretation reflects the intention to make genotoxic alterations of DNA responsible for both initiating and promoting properties.

Another interesting issue regards the methylation of CpG sequences in DNA, which regulates DNA activity. The phenotype may be altered in this way without changing the base sequence. In the course of cell differentiation, genes can be shut off irreversibly. Several tumor suppressor genes are known to be inactivated by hypermethylation of their promoter region, which has been considered to be an epigenetic process (Sugimura and Ushijima, 2000).

4.2. Role of Oxygen

The role of reactive oxygen was increasingly discussed in the 1980s. One of the findings was that activated neutrophils are able to damage chromosomes of mammalian cells (Weitberg et al., 1983), and the DNA of neutrophils themselves was fragmented during phagocytosis (Birnboim, 1982; Lown et al., 1982). Mutagenic effects of hydroxy radicals generated in chronically irritated tissue have been thought to contribute to carcinogenesis. All tumor promoters release oxygen radicals from polymorphic leukocytes. However, the promoting effect does not necessarily depend on DNA damage; rather, it could affect membranes or other cell constituents. But indirect damage of DNA by tumor promoters is still considered an attractive possibility. DNA alterations could enable the initiated cell to express its malignant potential. Eriksson and Anderson (1992) described in detail the biochemistry of membranes in liver carcinogenicity.

Sies et al. (1983) emphasized the labile redox equilibrium during oxidative stress. Environmental chemicals are preferably reduced by NADPH in one electron transfer at one step. The resulting radicals reduce oxygen to Superoxide anion radicals, and a cyclic process may be initiated that produces activated oxygen continuously. Part of the defense mechanism is the redox system GSH/GSSG (reduced glutathione/oxidized glutathione), and it has been demonstrated that GSSG is indeed released and excreted during drug metabolism.

Richter (1988) proposed a role for mitochondrial DNA and its susceptibility to oxidative damage. According to this proposal, DNA fractions move from the mitochondria into the nuclear DNA, thereby transforming a cell into a tumor cell.

Most of the mechanistic considerations tried t\o correlate direct or indirect DNA damage to tumor formation and to assess the probability of tumor formation-i.e. the risk-from the extent of adducts or specific mutations. Such correlations may be found between DNA effects and tumor initiation but do not take into account the role of toxicity.

4.3. Oval Cells

Fibrosis and cirrhosis develop as oval cells proliferate. However, this association was controversial for a long time. Ogawa et al. (1974) do not use the term oval cells, but already describe the process clearly. With 3'-methyl-4-dimethylaminoazobenzene (0.06% in the feed for 2, 4, and 6 weeks), bile-duct cells proliferate in the periportal area and spread toward the pericentral area, substituting for eliminated hepatocytes. Sell and Salman (1984) saw quite clearly cells from a different cell population after feeding AAF (0.03%) with a choline-deficient diet, cells that Farber called oval cells. These cells looked different from normal parenchymal and mesenchymal liver cells and formed bile ducts. This was not in line with the view that oval cells originate by differentiation of hepatocytes. New cells are seen already 1 day after AAF treatment. They begin to proliferate, and periductular cells are clearly seen after 2-3 days and their number increases up to 2 weeks. By this time a mixed cell population has spread from the periportal area into the sinusoids of neighboring liver lobules. After 3 to 4 weeks the whole liver is occupied by untypical bile duct-like cells. Because proliferating oval cells form ducts, it was concluded that oval cells come from bile duct cells. This notion is supported by the observation that most of the later infiltrating cells look like duct-forming cells. However, this may be questioned because proliferating oval cells as well as bile duct-like cells express alpha-fetoprotein (AFP) and albumin, which is not the case with normal bile duct cells in the periportal area. Bile duct cells, oval cells and focal cells all express gamma-glutamyl-transpeptidase (GGT). This enzyme is therefore not a suitable marker to identify the origin of focal cells. Lastly, the data favor the view, that a stem cell in the periportal area differentiates into oval cells and bile duct-like cells. Interestingly, the authors consider the discussed cell lines as a nonmalignant adaptation that may not develop into a tumor.

Oval cells are not seen after partial hepatectomy despite the heavy regenerative proliferation, but they are clearly seen after AAF and azo dye treatment (Thorgeirsson et al., 1988). With AAF and ^sup 3^H-labeled thymidine, radioactivity was carried over from oval cells to basophilic hepatocytes, which argues in favor of oval cells being precurser cells of hepatocytes.

Beland et al. (1990) reported among others at a meeting on the bad correlation between DNA binding of labeled AAF and tumor formation in liver and bladder of mice. During the discussion of this meeting Trosko asked whether the oval cells in liver as tentative stem cells would be the target for the critical effects. M. Poirier cautiously answered that DNA adducts are much lower in oval cells than in hepatocytes, but binding could also be different between normal hepatocytes. However, with different methods it can be demonstrated that the oval cells from the periportal area are able to differentiate into two directions. With low AAF doses liver injury is repaired by oval cells. Most of them become small hepatocytes. High doses slow down the process, and newly formed hepatocytes are proposed to develop into foci (Paku et al., 2004).

Oval cells were observed in the periportal area when AAF (0.03%) was fed to Donryu rats for 3 to 6 weeks. Later rapidly growing hyperplastic foci developed in this area, and occupied most of the liver by 12 weeks. Histochemically the proliferating cells were deficient of glucose 6-phosphatase and canalicular ATPase. Enzyme alterations were also seen in the developing carcinoma. These authors did not believe that oval cell proliferation is a critical morphological indicator for developing liver tumors, because different carcinogens modify the proliferation differently; for instance diethylnitrosamine strongly, 3'-methyl-4- dimethylaminoazobenzene only minimal and AAF in between.

For Yoon et al. (2004) it is clear that in the recovering liver oval cells are stem cells for both hepatocytes and bile duct cells, which come from the periportal area and differentiate in both directions. Similarly, Lemmer et al. (2004) consider oval cells as precursors of hepatocellular as well as cholangiocellular tumors. They are interested in human liver tumors associated with the exposure to the mycotoxin Fumosine B1 in different parts of the world. Fumosine stimulates oval cell proliferation in animal experiments like AAF and this process may be supported by AAF, presumably due to the inhibition of hepatocyte regeneration.

The accumulating information favors oval cells to be the activated stem cells in the liver. But how are they activated? Arai et al. (2004) analyzed the expression of DNA in oval cells in mice after AAF feeding combined with partial hepatectomy by microarray in 2304-DNA clones from mouse liver. Sixty-nine genes were upregulated, among them 6 only in the complete model, but not if only partial hepatectomy was performed. The authors conclude that these genes are involved in the activation of oval cells and that there is a difference between cell substitution after partial hepatectomy and after a toxic insult.

When the expression of 3000 genes was analyzed by microarrays in persistent liver foci of AAF-treated Wistar rats, positive signals indicating active genes were obtained from 2000 transcripts, of which 8% occurred only in liver foci (Tellgreen et al., 2003). A lot more work seems to be required before the process can be described on the molecular biological level. For the time being it would be helpful to agree on the role of oval cells.

4.4. Liver Cirrhosis as a Precancerous Lesion

By the end of the 1960s the reaction products with DNA of activated metabolites were well studied. However, the critical biochemical events at the genetic as well as at the epigenetic level after exposure of a cell to an ultimate carcinogen were not characterized (Miller and Miller, 1969). Clinicians used to emphasize the general effects on the whole organism and made a difference between unspecific cell damage and specific late effects which ultimately lead to malignant growth. They considered liver cirrhosis as a precancerous lesion much more than theorists.

In the monograph "Cancer, Race, and Geography," Steiner described in 1954 the etiological implications of race and geographical differences for the 20 most frequent tumors. With regard to liver he saw in high rate areas four typical symptoms: Tumors grow in general but not exclusively in cirrhotic livers, they are usually hepatocellular, they occur preferentially in men, and most of these men are more than 50 years old. The author relates the quick decline of the patients with the concurrent effects of cirrhosis and tumor. In "Fundamental aspects of normal and malignant growth," Kirschbaum (1960) described the relationship between liver cirrhosis and liver tumor as well known and generally accepted.

Sutton (1962) studied liver toxicity of butter yellow in rats (0.06% in feed) and observed changes in both the bile duct system and the parenchyma. First, bile duct cells proliferate up to a "cirrhosis-like" situation and a few bile duct lesions. In the meantime, cell nuclei and cytoplasm change in the parenchyma cells, and finally hyperplastic cells appear. The author cites Firminger (1955), who studied butter yellow and AAF, saying that cholangiosarcoma may grow in areas of cholangiofibrosis, but these make only a small fraction of the total liver tumors. Because he saw some carcinoma in animals without cholangiofibrosis, he concluded that cholangiofibrosis may not be an essential prestage of liver tumors. We would now interpret these findings differently.

Williams (1980) admits that liver cirrhosis may play a role in humans, but disputes this for animal experiments, where in his opinion toxicity does not contribute significantly to the process.

Schaff et al. (1986) describe focal hyperplastic nodules consisting of cells comparable to adult hepatocytes and arranged in strings of two to three cell layers thick. The layers are separated by fibrous sheets in a cirrhosis-like manner. In 80% of patients with primary liver carcinoma they find fibrosis and cirrhosis. In 63% of the cases mixed micro- and macronodular cirrhoses are the predominant lesions. However, these authors dispute "nodular transformations," which may be associated with displasia, as a prestage of hepatocellular carcinoma.

The few examples show that a consistent morphological picture develops and that much more attention should be given to the role of fibrotic lesions, particularly since they seem to be involved as a prestage in other tissues like lung and kidney as well.

5. THE INITIATION-PROMOTION CONCEPT

5.1. The Two-Stage Model

Berenblum and Shubik (1947) produced skin tumors in mice and distinguished two separate stages. In a first step, target cells in the skin are exposed to the carcinogen and somatic cells are initiated. In the following step, the initiated skin area is treated with a promoting chemical, which itself was considered not to be carcinogenic. Boutwell (1964) extended the model to a threestep model and divided the promotion stage into a conversion and a propagation stage. In a three-step model, the steps are called initiation, promotion, and progression.

From the very beginning the multistage concept of skin tumor formation in mice was discussed controversially. The concept was supported, however, by studies that showed that similar stages were also seen in other tissues, particularly in the development of liver tumors in rats. Peraino et al.(1973) used AAF (0.02% in the feed) as an initiator and phenobarbital as a promoter. The tumor yield was significantly higher after promoter treatment than without (70% versus 20%). Although the initiated cells could not be seen histologically, the alterations must have been very resistant, since the initiated cells could be promoted after a 1-year intermission without treatment. Enzyme-altered foci occurred first, and changes in the enzyme pattern such as the expression of fetal glutathione S- transferase, normally absent in the adult, indicated that this was not due to a mutation rather than by controlled activation of gene expression. (Pilot and Sirika, 1980). Pilot (1980) already staled lhal somatic mutalion musl nol be the only and absolute prerequisite for neoplaslic lransformalion.

In a proliferating focus, only some of the altered cells (1-2%) persist and divide spontaneously. The key decision at that time is between remodeling and proliferation (Farber, 1990). Farber assumes that reactive metabolites produce hepatocytes of a new phenotype. This requires at leasl one round of proliferalion lhat is caused by toxic cell destruction. Farber calls the resisting cells accordingly a resistant phenotype. The changes in enzyme pattern regard a decrease in activating drug metabolism enzymes and an increase of inactivating enzymes. It should be noted that the "initiated" cells proliferate neither autonomously nor spontaneously. This only happens if a suitable environment for the expansion of the altered phenotype is provided by "promoting" effects. In a suitable environment proliferation of initiated cells is stimulated and that of normal hepatocytes is inhibited. Under the further presence of AAF only the resistant hepatocytes grow to nodules.

Such nodules seem to proliferate spontaneously. The labeling index increases from 4% after 2 months to 8% after 6 months, whereas it remains at 0.4% in the surrounding tissue. However, the nodules grow slowly because a growth fraction of 4% is paralleled by a cell loss of 3%. Furthermore, nodules within nodules were observed and it has been assumed that tumors arise from these nodules within nodules. The balance between cell loss and growth is disturbed when cells turn malignant. Farber and Rubin (1991) consider the loss of balance to be more important than stimulation of cell proliferation.

The multistage concept of tumor development sincerely influenced the further direction of research. Many experiments were based on the shortened version of the two-stage model, to demonstrate the idea of initiation and promotion. A whole new area began with the detection by Cramer et al. (1960) that W-hydroxylation of AAF represents the first step on the way to an ultimate carcinogen. The paradigm of somatic mutation was confirmed when identical DNA adducts could be demonstrated in vitro and in vivo. The dominating idea was that initiating chemicals irreversibly alter a cell and this could be envisaged only by mutation(s) because practically all known complete carcinogens were mutagenic.

However, it should be remembered that at the biochemical level toxic effects occur practically always, but carcinogenic effects only sometimes. The term toxic should be broadly defined in this context and is not restricted to acute toxic and lethal effects.

Since the biologically active dose is usually highest in liver as compared to other tissues, it has been frequently asked why so many genotoxic chemicals do not develop liver tumors. Liver responds in many ways when exposed to carcinogens, but the effects often do not correlate with carcinogenesis. Even irreversibility has been questioned. Typically specific functions are expressed that occur also in normal cells, such that differences are rather quantitative than qualitative.

Meanwhile, the expression "tumor promotion" is used more broadly for many different interactions that influence a multistage process. A common property of promoting chemicals seems to be that they stimulate directly or indirectly cell proliferation in the target tissue. Schulte-Hermann (1985, 1987) distinguishes two groups of liver tumor promoters: (1) cytotoxic chemicals like carbon tetrachloride and AAF, or nutrition-related situations, which select for resistant liver cells, and (2) nontoxic chemicals like phenobarbital that induce adaptive answers in overspecialized cells. Cells in preneoplastic foci are thought to grow more rapidly, because they are less sensitive to cytotoxic stress than normal liver cells. This corresponds to the concept of the resistant hepatocyte. Initiated cells are phenotypically altered cells that develop into foci. This interpretation requires that initiation is associated with alterations in a gene program for adaptive responses. Eventually, these cells escape the orderly integration into the liver lobe. The biochemical basis was not known at that time.

All prokaryotic cells are able to adapt to changes in cell physiology or the microenvironment to a certain extent. Many chromosomal alterations can be seen that seem not to be tumor specific; they may also occur in noncarcinogenic situations. Carcinogenicity could therefore be a special case of a natural phenomenon. Activation of oncogenes, altered chromosomes, and differentiation and redifferentiation could be epiphenomina (Nery, 1976).

Walpole's presentation at the Ciba Foundation Symposium prompted an interesting discussion. Alexander (S. Walpole, 1959) emphasized the dose dependence in such experiments and concluded that if a certain degree of tissue damage were required then a threshold value should exist: no tumors without tissue damage. In other words, a linear dose-response relationship of a complete carcinogen, which would be consistent with the mutation theory, would argue against a threshold.

Druckrey (1959) agreed with the two-stage model and used it to discriminate between carcinogens without threshold and promoters with threshold. Berenblum (1959) accepted this interpretation. In his own presentation, in which Druckrey reported about the dose- response relationship of dimethylaminoazobenzene (butter yellow) and frani-4-dimethylaminostilbene, he argued that his rats did not develop tumors with low doses, because the latency period was longer than life expectancy. He could not see any result that would support the existence of a threshold dose. In the discussion of his paper, his summation theory, that is, the proposal that all effects of the individual doses add up to a required total effect, was severely questioned.

5.2. Tumor Promotion

Some path-finding observations were made by Peraino et al. (1971) in initiation-promotion experiments. If phenobarbital was administered to rats simultaneously with AAF, tumor incidence was lowered. The tumor incidence, however, increased if phenobarbital was given after AAF pretreatment. In this case liver cirrhosis was usually observed and all animals developed liver tumors. Although phenobarbital alters the ultrastructure of liver, it does not produce tumors itself. The authors discuss several possible explanations: (1) Phenobarbital releases genotoxic metabolites from macromolecular adducts and increases the probability of genotoxic effects. (2) Phenobarbital stimulates the expression of molecular effects produced by AAF. (3) Phenobarbital increases the probability of cell transformation because DNA lesions are fixed more rapidly. (4) Phenobarbital inhibits immune suppression. They conclude that promoters act by at least two different and independent mechanisms. On the one hand, dormant tumor cells may be activated on the level of gene regulation; on the other hand, membranes may be altered, and these alterations could support the progression to a malignant tumor.

At the same time, Teebor and Becker (1971) applied AAF in an unusual way. They fed 0.06% AAF for 3 weeks to male Sprague-Dawley rats, followed by 1 week without treatment. After 3 such cycles, 90% of the animals had hyperplastic nodules in liver, most of which regressed within 6 posttreatment months. At the end of this time only 8% of the nodules remained. After four such cycles the result was quite different: 94% of the animals had nodules after 4 months and most of these nodules (70%) remained. After 14 months the tumor incidence in the first group was 4% and in the second group 61%. According to these authors, hyperplastic nodules develop as a heterogeneous population and the different regression in the two groups indicates that the additional treatment in the fourth treatment cycle changed the character of the nodules. They concluded that the tumor yield does not depend on the number of nodules but on an essential step on the way to malignant transformation. Tumors must develop from the persistent nodules.

Although liver nodules were observed by many authors, they were not accepted by many as tumor precursors. An early statement by Kitagawa and Sugano (1973) said: Before carcinoma develop in liver, various kinds of hyperplastic lesions are observed during and after treatment. Hepatocytes degenerate and are eliminated and these alterations are reflected also on the biochemical level.

Berenblum and Vlasta (1981) offer three theoretical explanations how promoters could affect DNA in an initiated cell: (1) by increasing cell division and consequently the mutation rate, (2) by inhibiting DNA repair, and (3) by affecting neighbouring regulator genes. The first two possibilities are probably not realized because promoters do not need to act at the time of initiation but can be effective a long time thereafter. With phorbol esters the mitogenic properties are discussed, but mitogens are not always good promoters. Although many observations would support the role of gene regulation, this cannot be definitely proved with the presently available molecular-biological methods

5.3. Tumor Promotion and Tissue Specificity

Hyperplasia or other histological al\terations are not necessarily required to induce skin tumors in mice with urethane (Berenblum, 1959). The limited effectiveness of urethane as an initiator in mouse skin contrasts the systemic effects after oral administration. When urethane was administered orally and the skin promoted with croton oil many tumors grew in the treated area. Similarly, AAF does not produce skin tumors after oral administration, but after promotion with croton oil. This is one of the first communications about tissue specificity of AAF, which obviously initiated cells in a "nontarget tissue," namely, skin (Ritchie and Saffiotti, 1955).

Paul and Hecker (1969) published data pointing in the same direction. After oral administration of PAK and AAF, skin tumors could be produced with local promotion. Wunderlich (1971) concluded that similar alterations are produced in all tissues by such chemicals but tumors arise only as a consequence of the promoter effect.

We have shown that with AAS doses sufficient to produce Zymbal's gland tumors in the rat (Hoffmann et al., 1993) neither foci nor tumors were found in liver or kidney even after 2 years. If, however, the animals were initiated with AAS and subsequently exposed to typical liver tumor promoters like phenobarbital or DDT, which stimulate cell proliferation in rat liver, tumors grew in this tissue (Hilpert et al., 1983). The same applies to kidney. If adaptive growth was induced in the remaining kidney after unilateral nephrectomy, or by treatment with nephrotoxic antibiotics, kidney tumors occurred (Hilpert et al., 1983; Hoffmann et al., 1993). These results clearly indicate that exposure to a genotoxic chemical presumably initiates cells in all tissues that can by promoted under suitable conditions. An experimental no-observed-adverse-effect level (NOAEL) therefore means little for the latent hazard.

One should keep in mind that AAF affects homeostasis in many ways. As an example.insulin controls liver metabolism and regeneration. Notably, AAF feeding reduces insulin binding to microsomes and a Golgi fraction early on (1 to 3 days) and consistently up to 85 days. This is considered to be due to a reduction of insulin receptors (Lev-Ran et al., 1984). AAF also modifies steroid hormone levels. Among others the zona fasciculata in the adrenals regresses and consequently the cortisol levels decrease (Lafaurie et al., 1980).

5.4. Promotion-Morphological Alterations

Hepatocytes are not arranged around the central vein but along a microcirculatory path, which follows the pressure gradient from the liver arteries and the portal vein to the liver venols. The zones 1, 2, and 3 in the liver venules are supplied with oxygen and nutrients in decreasing order. The maintenance vessels are central in the microcirculation and not in the periphery (Rappaport et al., 1954). The authors developed the concept of the structural unit in liver on the basis of microcirculation as it was observed in vivo. Surprisingly, the role of microcirculation in liver cirrhosis and its consequences for tumor development have been largely neglected so far. Considering the sensitive balance, it is quite conceivable that normal and resistant hepatocytes compete for nutrients and that disturbance of microcirculation contributes to a growth advantage for resistant cells.

Morphological alterations in rat liver have been studied since 1959. Shirasu et al. (1967) looked for a role of pituitary hormones for the promotion of liver tumors with labeled W-hydroxy-AAF and reported histological observations: Within a few days degenerative alterations and necrotic areas can be seen in rat liver. After 32 days fibroblast activity increases and bile duct cells proliferate in the periportal area. Eventually the electron microscope was used to analyze early alterations produced by AAF (0.05% in the feed). Glycogen depletion and amplification of the smooth endoplasmic reticulum and the Golgi-apparatus, combined with extra lysosomes, were detected. Later on mitochondria were smaller. Some of these features were also present in preneoplastic cells and in hepatoma of AAF-treated rats and a connection was established with carcinogenesis (Flaks, 1970). To test this proposal the experiments were repeated with 4-acetylaminofluorene, a noncarcinogenic isomer of AAF. No changes were found except glycogen depletion and an increase of the endoplasmatic reticulum. The fine structure of hepatocytes from 2- and 4acetylaminofluorene-treated rats was clearly different (Flaks, 1972).

At that time morphology was studied also with other carcinogens. Bannasch and Reis (1971) used N-nitrosomorpholin (NNM), which produces hepatoma and cholangioma in rat liver. Toxic doses (up to 50 mg/100 ml drinking water) increase the proliferation of mesenchymal cells and bile duct cells, and fibrosis, cirrhosis, and a focal cholangiofibrosis develop. In a first phase mesenchymal cells reproduce primarily in the periportal area, where the parenchymal lesions occur, as well as oval cells, which were classified as bile-duct epithelia because they arranged rapidly to small ducts. In a second phase proliferating mesenchymal cells produce collagenic fibers, bile duct cells mucous substances. This combination leads to a cholangiofibrosis similar to that seen after poisoning the liver with butter yellow. IfNNM is administered over a minimum period of time, the cholangiofibrosis persists and must be considered an irreversible liver lesion. Foci of cholangiofibrosis are often linked by bridges of connective tissue, which isolates islands of liver parenchyma and alters the cell-cell relationship. In a third stage cystic cholangiomas develop, which grow slowly, do not invade neighbouring tissue and do not metastasize. The authors interpret their results as follows: Parenchymal necroses are repaired by bile duct proliferation. This takes place similarly with many noncarcinogenic liver poisons as well as with many carcinogens. Bile duct epithelium begins to proliferate when cell-cell contacts to liver parenchyma get lost.

A different view follows from cytomorphological and morphometric analysis of alterations seen in the cytosol during NNM hepatocarcinogenesis. Bannasch et al. (1972) describe the pathological findings, which may occur simultaneously or in sequence, in the following way: (1) clear glycogen storage cells, (2) acidophilic glycogen storage cells, (3) lipid storage cells, (4) basophilic or eosinphilic glycogen depleted cells, (5) hepatoma cells, and (6) undefined X-cells.

The altered smooth endoplasmatic reticulum and the glycogen depletion are considered as a consequence of unspecific toxic effects that lead to necrosis followed by fibrosis, cirrhosis, and cholangiofibrosis but not to hepatoma. The argument of the authors is that the effects are at least partially reversible. They see it confirmed by the fact that the formation of hepatoma is preceded by a persistent hepato-cellular glucogenosis. As a major feature of cell transformation glycogen is steadily reduced, which happens in the stop experiment independent of further exposure to the carcinogen. Cell transformation is not an immediate consequence of intoxication, but a reaction of the cell to the primary toxic lesion. Since the morphological alterations often occur long after removal of the liver poison, it is rather envisaged as a conversion of cell metabolism independent of the carcinogen. Later this phenomenon is called adaptation.

In contrast to the opinion of Enzmann and Bannasch (1988), who related the elimination of early and severely altered foci after cessation of AAF exposure to unspecific toxic effects, Stenbck et al. (1994) propose that foci disappear due to remodeling-that is, they regain a normal phenotype due to homeostatic influences of the neighboring hepatocytes. In their final remarks the authors emphasized that the dose was low in these experiments and toxic effects could not have contributed. For a later paper, AAF doses were 0.126, 0.42, and 1.26 mmol/kg body weight. The number of GST-P- positive foci increased dramatically only after 12 weeks and only with the highest dose. The low dose was characterized as of low toxicity according to glutamine synthase-positive cells, the middle dose as moderately toxic, and the high dose as severely toxic. Proliferation was enhanced only with the high dose.

According to Ruch and Klaunig (1986), chemicals with promoting properties typically inhibit growth via control of intercellular communication. Genotoxic chemicals like AAF are ineffective against normal, that is, noninitiated cells in cell culture. They assume that communication is disturbed in initiated cells and that this effect makes a complete carcinogen from a genotoxic carcinogen. Farber (2000), in contrast to this proposal, points to the fact that liver focus cells have a particularly close relationship. If indeed communication through gap junctions or other cell-cell contacts were reduced, one should also expect a lower coherence between promoted cells.

5.5. The Role of Proliferation

Warwick (1969) was among the first who emphasized the role of cell division in carcinogenesis. 2-Methyl-4dimethylaminoazobenzene, a methyl derivative of butter yellow, which normally does not produce rat liver tumors, becomes effective when cell proliferation is stimulated, for instance by partial hepatectomy. The author proposed that a chemical produces liver tumors only if it produces a critical DNA lesion and in addition increases cell proliferation. He mentioned numerous carcinogens that fulfill these criteria and concluded that chemicals that are not hepatotoxic become effective only if for other reasons proliferation is increased.

Bannasch et al. (1974) feel their view confirmed: Regenerative hyperplasia should be irrelevant for the generation of nodules. Carcinogens induce an irreversible step at the very beginning of the process. Such primary events may differ betweendifferent hepatocarcinogens, but eventually enter a common path of cell evolution.

Schulte-Hermann et al. (1988) talked about apoptosis at the Fourth Sardinian International Meeting on Models and Mechanisms in Chemical Carcinogenesis. They explained the fact that enzyme- altered foci grow so slowly after a mitogenic stimulus, because apoptoses occur much more frequently than in normal tissue. They conclude that in the initiation-promotion model with NNM and phenobarbital, the latter inhibits apoptosis and thus prolongs the lifetime of cells in the focus.

We presented at this meeting results obtained with three structurally related aromatic amines (AAF, AAS, AAP) that generated liver tumors in Wistar rats only if cell proliferation was increased. AAF is hepatotoxic and produces tumors in this tissue. AAS and AAP are not hepatotoxic and do not produce liver tumors; however, both are genotoxic and are strong initiators in rat liver (Neumann, 1988) (Figure 1).

Farber (1995, 2000) cautioned against generalizing that every chemical that stimulates proliferation should be considered a tumor risk factor. The epithelium of the small intestine and psoriasis are examples. The lasting inhibition of proliferation should be responsible for cell transformation rather than its stimulation (Rubin, 1980). Any kind of physiological or pathological proliferation should loosen cell-cell communication through gap junctions, but it has not been proven that this is a particular property of tumor promoters. The essential feature is selectivity; proliferation without a selective effect does not promote (Farber, 2001).

Tight junctions control the exchange between neighboring cells of water, ions, and certain macromolecules. In the course of tumor development the polarity of cells is lost in this way. Signals involved are the transcription factor ZONAB, which regulates the expression of the protooncogene Erb-2 and epithelial proliferation. Small G-proteins of the Rab-family control specific membrane transport processes (Zahraoui, 2004).

5.6. The Role of Toxicity

Proliferation seems to contribute to tissue specific effects but it is still not clear what triggers it-what toxicity means in this context. As early as 1938 Haddow discussed a gradient of effects and proposed that cells closest to the source of the toxin would be killed, whereas more distant cells survive and cells even more distant remain unaffected. A tumor should therefore grow where the biologically active dose is highest, but where cells have a chance to adapt. Since then many studies have shown that in the target tissues cytotoxic effects are seen first. Clinical observations show heavy irritations in the bladder epithelium after 2-naphthylamine exposure and before tumors arise. Berenblum (1944) distinguished two kinds of cellular answers to wounding or irritation: an immediate irritation reaction and a later regenerative hyperplasia. The important initiating and irreversible cell transformation should therefore happen during the phase of preneoplastic hyperplasia. At that time it was assumed that the carcinogen reacts with cellular proteins and inactivates them. The role of DNA was not established.

Toxic effects play a major role in our hypothesis, and a more comprehensive picture can now be drawn. Exposure to toxic Chemicals induces counterreactions in the whole liver tissue and inhibits proliferation. In the case of AAF, cells become even resistant against the induction of apoptosis. Toxic doses may overcome such defense mechanisms and induce regenerative proliferation, which eventually causes a cirrhotic rearrangement of the liver lobule and thus provides a favorable environment for preferential growth of initiated foci. Hepatocytic nodules develop through clonal expansion. But only a few of these nodules develop into a tumor. It is therefore important to better understand and to explain how the carcinogens elicit their inhibitory properties.

Hoel et al. (1988) tried to assess the role of toxicity and evaluated the 2-year carcinogenicity tests for 99 chemicals. Of these, 53 tested positive but only 7 showed substance-specific tissue specific toxicity. Interpretation of the test results were frequently questioned and considered invalid because maximal tolerated doses (MTD) were used and toxic effects are restricted to these high doses and would not occur at lower doses.

Although preneoplastic lesions correlated well with tumor formation in this compilation, this was not the case for toxicity in male rats and marginally with female rats. However, toxicity correlated well with tumor formation in mice. The chances for liver tumors were four times higher when liver toxicity was observed than without toxicity. According to this compilation there exist liver toxicity without tumors and tumors without toxicity in both species (Hoel et al., 1988). It was also noted that in 25% of cases with toxicity body weight was reduced only with the MTD. The authors conclude that most chemicals produce tumors even if the exposure is not toxic in the target tissue, but they admit that altered proliferation might have escaped histopathological observation. It should however be noted that the selected chemicals were structurally quite unrelated and classical aromatic amines, like AAF, were not included. For these the correlation is expected to look different.

For Travis and Belefant (1992) the situation is rather simple: Cytotoxicity is not an essential feature of promotion. Any process that increases division of initiated cells also increases the probability that mistakes in the endogenous DNA are converted into mutations. Promoters are thus indirect mutagens. As a consequence, the authors deny reversibility of promotion. Although promoting effects as well as genetic alterations can be eliminated by apopto

Source: Critical Reviews in Toxicology

More News in this Category