A Biologist's View of Human Cancer
F. Anders    Hämatol. Bluttransf. Vol32

The work of the author's research group was supported by the Justus-Liebig-Universität Giessen, by the Deutsche Forschungsgemeinschaft, by the Bundesminister für Forschung und Technologie, and by the Umweltbundesamt

This biennial lecture reflects the generosity of Dr. Mildred Scheel, whose life was dedicated to the fight against cancer. I met Mildred Scheel personally on the occasion of several conferences on human cancer and remember her with gratitude. It is an honor to have been invited to present the 1988 lecture. The ultimate purpose of all who study cancer biology falls within the general goal of the efforts of Dr. Scheel: to analyze the biological factors that are involved in tumor development for the purpose of preventing cancer. At times the analytical work of many scientists of Mildred Scheel's generation appeared to meet certain opposition when they have seen printed in large letters "cancer is not inherited" and "genes that determine cancer do not exist." Such statements came from well-meaning people intent on calming the fears of families that have had cancer in their ancestry. We all arc involved in the fight against cancer, the physicians, epidemiologists, biochemists, immunologists, virologists; everybody in his place. I am a zoologist, trained as a geneticist who views human beings as products of nature with all their potentials, limitations, and inadequacies arising from their animal background. A. Oncogenes in Phylogeny Neoplasia is not limited to human beings, or to mammals, but develops in all taxonomic groups of recent Eumetazoa and even in multicellular plants. Neoplasia was also found in Jurassic Sauria and in other fossils including humans. Neoplasia, therefore, was not created by human civilization, but is inherent in the multicellular organization of life [1].It is, therefore, not surprising that the genes coding for human cancer are distributed throughout the animal kingdom (Fig. 1, [2 -10]). The most venerable oncogene seems to be the ras oncogene, which probably has evolved together with the heterotrophic organization of the carly Eucaryotes. This supposition does not exclude the idea that certain sequences of ras (and other oncogenes) might have been evolved before thc heterotrophs in the history of life. Actually ras is distributed as a normal genomic constituent from yeast [11 ], where one obviously cannot recognize a cancerous state, through all groups of the animal kingdom studied up to humans and is possibly involved in the development of human tumors such as bladder carcinoma, melanoma, neuroblastoma, fibrosarcoma, lung sarcoma, lung carcinoma, and acute myeloid leukemia (for review see [12, 13]). Its early appearance in the history of life suggests fundamental functions for our life. Its product is a GTP-binding protein which probably activates phospholipase C that generates the internal promoter diazylglycerol for kinase C, thus signalizing cell proliferation [14-16].

Fig. I. Attempted outline of the evolution of oncogene systems in the animal kingdom (com piled from [2-10]). See text

As one moves up the evolutionary scale to the multicellular organization of the living beings, i. e., to the Metazoa, the ,src oncogene appears in the parazoic sponges and is, thereafter, traceable through the Eumetazoa up to humans [2, 17, 18]. We have not identified cancer in sponges, but ,src was found highly active in the sponges which, because of the autonomy of their cells, can be considered to grow as independently as tumors. In Coelenterata such as sea anemone both ,src activity and abnormal growth comparable to teratomas of higher species have been observed. High activity of src measured as activity of its product, the pp60c-src kinase, was detected in the nervous cell systems of all groups of animals tested. Its activity is also high in animal and human melanoma [19, 20], the cells of which are probably all derived from the neural crest cell-system. The ,src oncogene is possibly, like src, involved in the transmission of proliferation signals which, on this evolutionary level, possibly include the phosphoinositide phosphoinositol turnover [15]. It serves probably in intercellular communication for coordination of growth and function of the Metazoa, perhaps through gap junctions. As we go up to the Bilateria the Metazoa branch out to the Protostomia and Deuterostomia, This period must have been evolutionarily very active and successful. A large variety of taxonomic groups containing a large packet of oncogenes has been evolved. In addition to ras and src, the following have been identified: (a) abl, fes, neu, erB, which belong to the src family and exhibit tyrosine kinase function, (b) myc and myb, which are assumed to fulfill regulatory functions of gene expression in the nucleus, (d) raf coding for a serine/threonine kinase, and (e) bcl2, isolated from human B-cell lymphoma. Since the viral oncogenes which mostly have been used as probes originate from higher vertebrates (i. e., Deuterostomia), one can conclude that the respective cellular genes must have been already present in the last common ancestor of both Protostomia and Deuterostomia. The clear hybridization signals always found with abl and myb lead to the presumption that they evolved still carlier in the history of life as can be shown by present data (see arrows in Fig. 1). Nothing is known about the tumorigenic function of these oncogenes in the tumors observed in invertebrates. Little is known about these functions in human tumors [12]. abl, myb. fes. bcl2 present in Drosophila. Limulus. etc., organisms which have no blood in the sense of the blood of mammals, are possibly involved in human hematopoietic malignancies; but no convincing data from human biopsy specimens or fresh cells from a variety of human leukemias und lymphomas are available showing that these early oncogenes are crucial in human neoplasia [12]. The appearance of the sis oncogene, which codes for the platelet-derived growth factor (PDG F) in the Chordata, represented by Amphioxus and lamprey in the outline of the phylogenetic tree, might be critical for the evolution of the closed blood circulation apparatus that exposes the blood to pressure. Up to the teleosts this oncogene is represented by only one copy. Later on, moving from lower Tetrapoda to Mammalia, a second sis copy occurs. In humans PDGF is coded by two distinct but related genes, namely the PDGF-A gene and the PDGF-B gene, the latter one being known as human c-sis, which is less homologous to the teleost c-sis than the PDGF-A gene [6]. Although human c-sis is apparently inactive in most human cells, it is supposed that both PDGF A and B (and their receptors) are involved in general rcgulatory processes, cell proliferation, and tumor formation [12], The yes oncogene occurs in the animal kingdom together with the appearance of the Gnathostomata, which are represented in our studies by sharks. This gene is a member of the src family which is highly homologous to src itself. This poses the question of gene duplication in evolution. Another example, the single sis copy of the teleosts that corresponds to the human PDGF A became duplicated (probably), as mentioned above. One could extend this question asking whether the large src family including the already mentioned ab/,.fes. neu. erbB, yyes and the not yet mentioned fgr, ros, and mos could have been evolved by gene duplication. The idea that oncogene families might have been evolved by gene duplication contributes to the general concept of evolution by gene duplication proposed by Ohno [21] almost 20 years ago. At the evolutionary level of Vertebrates, fgr, a member of the src family, fos, a nember of the myc/myb family, and erbA, a partial homolog of the receptors of thyroid hormone, estrogen, progesterone, glucocorticoid hormone of humans, and the human X-factor, appear together in the teleosts. Since erb A of the fish shows strong homologies to the viral gene, one could assume that it has evolved earlier in the history of life than the present data indicate. It seems not to be involved in neoplastic transformation but in tumor promotion, perhaps supporting erb B, which appears to be involved in transformation [22]. It is notable that, based on our earlier genetic and histogenetic experiments, not only have gene patterns favorable to neoplasia been observed in teleost species but also genes which limit the action of these genes to certain cell types [23]. This is an important point to consider in human neoplasia [3]. It appears that nature's way of keeping the oncogenes from their transforming capacity as soon as they became too dangerous for the increasing complexity of life has been to establish a new category of genes, namely the oncogene-specific regulatory genes [24], today sometimes called anti-oncogenes or oncostatic genes. Finally, ros. a member of the .src family, possibly involved in cell proliferation and tumor promotion through the internal promoter diazyglycerol [14-16], appears to be specific to the Tetrapoda, and mos, related to the src family and also to raf, appears to be specific to Mammalia [4, 5, 25]. Nothing is known, at least to my knowledge, about the specificity of these genes to the organization of the Tetrapoda and Mammalia, respectively. mos is probably involved in human acute myelogeneous leukemia [12]. In conclusion it appears that, in parallel with the advancement of the animal kingdom, particular oncogenes were subject to their own evolution and that, furthermore, the systems of the oncogenes corresponding to this advancement increased in number, several of them probably by gene duplication. From yeast to mammals we found an increase from 1 to 17 (see Fig. 1, right). This increase might reflect the increase of complexity required for advancement in the anima] evolution but might in addition reflect an increase of sensitivity to any endogenous and exogenous impairment of the systems. Therefore, our phylogenetic view might reflect some rough observations on the tumor incidence in the animal kingdom which so far have never been studied seriously. Although both oncogenes and cancer have been observed in all systematic categories of the Eumetazoa, it appears that mammals are more afflicted with cancer than any other group of animals.

B.Low and High Susceptibility to Neoplasia

Neoplasia occurs infrequently in the natural populations of Eumetazoa, and induction of cancer by initiating carcinogens and tumor promoters is difficult to achieve [26]. This phenomenon was studied in detail in the Central American teleost genus Xlphophorus [26-29] and in East Asiatic mice [30]. Natural selection in Mendelian populations will not favor one population or race and discriminate against the other but will always work against susceptibility to cancer in all populations and races. However, certain nontaxonomically defined groups of animals are highly susceptible to spontaneously developing, carcinogen-initiated, and promoter-stimulated neoplasms (Table 1). These groups consist mainly of animals of hybrid origin, such as naturally occurring or experimentally produced interspecific, interracial, and interpopulational hybrids as well as laboratory and domesticated animals which actually are also hybrids, i. e., homozygous combinations of chromosomes of different populational or racial provenance. These animals share their high susceptibility to neoplasia with humans [26, 31]. While we do not have data on the relationship between hybridization and cancer in human beings comparable to the data on animals, it is interesting to speculate whether the many facts on tumor incidence in humans that do not agree with the concept of the primacy of environmental factors in carcinogenesis can be explained by interpopulational and interracia1 matings in our ancestry. Certainly interpopulational and interracial mating may have occurred at any time in any place. Because of the high and increasing mobility of modern humans as compared with other species, one should expect high heterogeneity. Various estimates based on enzyme variation showed that heterogeneity in humans is comparable to that of domestic animals such as cats, but is about six times higher than that of wild macaques, about ten times higher than that observed in the large wild mammals such as elk, moose, polar bear, and elephant seal, and about twice as great as that of most feral rodents studied so far [32 34]. Based on these

Table I. Animals that exhibit a high tumor incidence (for references see [26, 31])

data and on the assumption that tumor incidence in general is related to interpopulational and interracial matings, one could explain why humans have a high incidence of neoplasia comparable to that of the domestic animals. Furthermore, there are some data on chromosomal heteromorphisms in human populations that might be useful for estimates of heterogeneity within and among different populations. According to such estimates it appears that, for instance, Japanese populations exhibit a low degree of Q- and C-band chromo some heteromorphisms. whereas Americans have a much higher degree of this heteromorphism, with blacks having more prominent heteromorphisms than whites [35, 36]. One is tempted to assume that this heteromorphism refects the differences in the degree of heterogeneity among the Japanese and white and black United States populations. In this context it is notable that the ratio of prostatic cancer in Japanese, United States whites, and United States blacks is reported as 1: 10: 30 and that the black citizens in San Francisco have double the risk of developing neoplasia as compared with their Japanese fellow citizens [37, 38]. We cannot explain these facts by environmental factors or racial differences. The high susceptibility to neoplasia in domestic or hybrid animals, respectively, could show us how to approach the problem. Of course, it is very difficult to study the heterogeneity of a recent human population of a city or country in terms of biological measures. However, new methods such as the determination of restriction fragment length polymorphisms available today could be helpful in revealing the possible relationship between genetic heterogeneity and tumor incidence in modern human populations.

C. Cancer in Xiphophorus as a Model for Cancer in Humans

Human biology is unique, but is not so unique in its fundamentals as to make studies on animal models irrelevant for an explanation of human diseases including cancer. Although mice and rats are the classical laboratory animals used in experimental cancer research, several genera of small teleost fish serve increasingly as models in new cancer research programs [39]. One of these genera is Xiphophorus (Fig. 2; for portraits of different phenotypes see [2, 3, 22, 23, 29, 31]), the animal model from Central America that we have used in our laboratories for 30 years [24,40]. Neoplasia appears to develop only very exceptionally in the wild populations of xiphophorine fish. In spite of the fact that thousands of individuals of many wild populations that are isolated from each other have been collected by several investigators and myself, no tumor has been detected. In the progeny of the wild populations that have been inbred in the laboratory for about 8o- 100 generations, no tumor has occurred spontaneously and almost

Fig. 2. Female and male of the "spotted dorsal' platyfish, Xiphophorus maculaIus, from Rio Jamapa (Mexico)

Table 2. Oncogenes in Xiphophorus

no tumor could be induced even with the strongest mutagens-carcinogens such as X-rays and N-methyl-N-nitrosourea (MNU). This fact requires special clarification since most of the oncogenes that are known to transform the cells and to drive the tumors are present in the fish (Table 2). If, however, interpopulational and interspecific crossings are performed, depending on the genotype, the progeny spontaneously or following treatment with initiating carcinogens (Xrays, MNU, ethylnitrosourea, diethylnitrosamine, 2-amino-3-methylimidazo(4,5-f)quinoline, etc.) and/or tumor promoters (12-0-tetradecanoylphorbol-13acetate = TPA, 5-azacytidine, phenobarbital, cyclamate, testosterone, nortestos terone, methyltestosterone, trenbolone, ethinylestradiol, cAMP, biphenyl, butylhydroxy toluene, deoxycholic acid, thioacetamine, bis(2-ethylhexyl)-phthalate, betel nut extract, etc.) develops neoplasia (data in [41]). Neoplasms originate from all neurogenic, epithelial, and mesenchymal tissues (Table 3). The suitability of the model is, except for research on mammalian-specific tumors such as breast cancer, lung cancer, etc., beyond question and its efficiency is more economic and time-saving than that of the laboratory mammals. Agents that induce neoplasia in certain high-risk genotypes of the fish hybrids, might, in principle, also affect certain high-risk human individuals.

Table 3. Neoplasms in xiphophorine hybrid fish induced by physical and chemical agents (i) or spontaneously developed (s)

D. Classification of Tumor Etiology in Xiphophorus and Humans

The neoplasms of Xiphophorus can be classified 1. Mating conditioned: accessory oncogenes are introduced into, and/or regulatory genes for the oncogenes are eliminated from, the germ line by replacement of chromosomes carrying the respective genes or lacking them, and vice versa. 2. Mendelian inherited: regulatory genes for oncogenes are impaired, lost, or dislocated in the germ line by mutation. 3. Mutagen-carcinogen conditioned: regulatory genes for oncogenes are impaired, lost, or dislocated in a somatic cell by mutation. 4. Nutrient and endocrine conditioned: resting stem cells are pushed to differentiate by tumor promoters (the genetic preconditions according to a, b, and care fulfilled by earlier events). 5. Virus conditioned: accessory oncogenes are introduced (so far not convincingly shown in the fish). The same classification can be applied to human cancer comprising a small group of (a) "familial"; (b) "hereditary" neoplasms in which genetic factors are supposed to be involved, e. g., retinoblastoma, meningioma, melanoma; (c) a large group of "carcinogen-dependent" neoplasms, e, g., lung cancer; (d) a large group of "endocrine-dependent" and "'digestion-related" neoplasms, e. g., breast, prostatic, colon cancer; and, finally, (c) a group of viral-conditioned neoplasms, e. g., leukemia, genital tumors. In Xiphophorus derived from a wild population neoplasia develops in general only if different protocols for the induction of tumors are combined by the experimenter, for instance, (a) the elimination of regulatory genes by selective matings, (b) the induction of germ line mutations, and (c) the induction of somatic mutations, etc. The particular events that alone do not lead to neoplasia, summate, and appear as a multistep process that goes beyond the generations and, finally, reaches the last step that leads to neoplasia in a certain individual. The experimenter must detect the sequence of the different steps, and it is easy to see that the last step that completes the multistep process determines the etiological type of neoplasia. This was shown for Xiphophorus but might be helpful to explain the different types of tumor etiology in humans in which both the ancestry of an individual and the individual itself are involved. In the following paragraphs we shall try to approach the biological basis of spontaneously developing, carcinogenmutagen induced, and promoter-dependent neoplasms.

E. Tumors Appearing and Disappearing in the Succeeding Generations

Human tumors such as a certain colon cancer that afflicts individuals 15-20 years sooner than generally may appear ..spontaneously" in a family in one generation and may disappear in the succeeding generation. This is demonstrated by means of a cartoon (Fig. 3, upper part) adapted from Lynch and his colleagues [50]. We cannot explain this phenomenon. The Xiphophorus model (Fig. 3, lower part) provided the opportunity to study a similar appearance through the fish generations. Crossings of a spotted platyfish (A) with a nonspotted swordtail (B) result in F 1 hybrids (C) that develop enhanced spot expression and sometimes benign melanoma instead of the spots. Backcrossings of the F 1 hybrids with the swordtail as the recurrent parent result in BC1 offspring (D, E, F), 50% of which exhibit neither spots nor melanomas (F) while 25% develop benign melanoma (D) and 25% develop malignant melanoma (E). Further backcrossings of the fish (not shown in Fig. 3) carrying benign melanoma with the swordtail result in a BC2 that exhibits the same segregation as the BC1. As opposed to the crossing procedure that gave rise to the melanoma, backcrossings of the melanoma-bearing hybrids (E), with the platyfish as the recur rent parent (A), result in an alleviation of the melanoma in the offspring (C*), which in the following BC generation grow into healthy fish (A *). In conclusion, malignant melanoma of the BC animal (E) originates from the spots of the preceding platyfish generations (A) and is reduced to spots again in succeeding generations (A *). The formal parallelism in the occurrence of neoplasia in the human family and in the experimental model is striking. In our search for causes of human cancer there might be some value in realizing the types of factors that can be passed from the fish parents to the fish offspring to influence the occurrence of cancer. The experiment with the model suggests that certain human cancers may be expected to occur in individuals because of a combination of factors from both parents that by themselves did not cause cancer in either parent. More data are required in order to compare more stringently human familiar cancer with mating-conditioned neoplasia in the model.

F. Oncogene Expression in the Tumors

The appearance of tumors in both human and model brings about the question for the oncogenes expressed in human and xiphophorine neoplasms. Data available for melanoma indicate an elevated expression of both the human and thc xiphophorine src, erbB, sis, ras, and myc ([2,6,7,18-20,40,44,45] personal communication, u. Rodeck). Measurements concerning the significance of the xiphophorine src oncogene (X-src) for the development of melanoma and other kinds of neoplasia in the fish (Table 4) showed that the activity of its product, the pp60x-src kinase, may be elevated in the tumors up to 50 times over that of the controls [46]. Furthermore, the phosphoinositide phosphoinositol turnover , which is supposed to be linked to the .X-src activity [14-16], was found up to more than ten times elevated over that of the controls (Table 5). This finding is important because the turnover may serve as a measure for the activation of phospholipase C, which generates the internal promoter diacylglycerol.

Fig.3. Appearance and disappearance of neoplasia in succeeding generations (cartoon adapted from [50]). See text


Table 4. Elevation ofpp60x-src kinase activity in tumors and brain of Xiphophorus hybrids. (Data from [46])

A tremendous amount of work on oncogene expression and its possible secondary processes in the tumors and in tumor-derived cell lines of experimental mammals and of humans [12] has been performed in the expectation of finding a particular tumor type-specific initial gene and the initial event of the formation of a particular neoplasm. While we were never able to identify what one could term a "liver cancer gene" or a "melanoma gene", others have thought they did. Our own studies on the Xiphophorus model showed only a relationship of a number of regulatory genes of a number of tissue-specific developmental genes which in total we called "tumor gene-complex" (Tu complex); but we interpreted this as an association rather than a true genetic entity, and we assigned the different kinds of neoplasms such as those listed in Tables 3 and 4 to the same Tu complex. The nature of the causality of neoplasia remained unclear.

G. An Approach to the Study of the Genetic and Molecular Basis of Neoplasia

The genes underlying neoplasia in Xiphophorus were most successfully studied in the generations developing the "spontaneously occurring" mating-conditioned tumors, and it appears to be in the nature of things that those laboratories working presently on the small group of familial and hereditary human tumors approached the fundamentals of neoplasia at least as closely as those working on the large groups of carcinogen- and promoter-dependent tumors. Our approach in the model is described by means of Fig. 4, which refers to the same fish as indicated in Fig. 3 by the same capital letters (for the mutants see later). Based on breakpoint data the genes responsible for melanoma inheritance are located terminally in one Giemsa band of the X chromosome [51] and represent a complex consisting of (a) the pterinophore locus (Ptr) which is responsible for pterinophore differentiation, (b ) the compartment-specific dorsal fin locus (Dr, impaired to Dl,) which restricts both pterinophore and macromelanophore differentiation to the dorsal part of the body, ( c) the region in which a viral erb Brelated oncogene (erb B*, an oncogene related to the receptor of the human epidermal growth factor, EGF, x-egfr) is located, (d) the melanophore locus (Mel), which appears to be under control of DI and erbB*, and (e) the arbitrarily symbolized "tumor gene" (Tu), which appears as a Mendelian factor but might possibly be composed of both erb B* and Mel [22, 52]. Oncogenes in addition to the xiphophorine erb B* (x-erb B*) could not be detected in the X chromosome. Based on our present knowledge, the respective region of the X chromosome of the platyfish, the "Tu complex," can be roughly mapped as follows (commas represent breaking points observed): X ….Ptr, Df, erbB*, Mel-Tu At least about 20 linked genes are involved in the regulation of the Tu complex, but there are also several nonlinked regulatory genes, e. g., the Diff gene, which, if present in the homozygous state, restrains the transformed pigment cells from proliferation by terminal differentiation [53]. The swordtail (B) has neither evolved a comparable Tu complex nor the linked and nonlinked regulatory genes. Since platyfish and swordtails have a rather high number of chromosomes (11 = 48) and since clear-cut chromosomal conditions concerning their origin were required, the experimental animals, besides the purebreds (A, B), were taken from the F 1 (C), which contains one platyfish and one swordtail genome, and from high backcross generations comprising BC8 up to BC22 (F, E), the genome of which virtually consists of swordtail chromosomes except for the Tu complex containing X chromosome selected from the platyfish by the crossings. The phenotypic overexpression of the Tu complex thus depends mainly on the crossing-conditioned replacement of platyfish autosomes carrying regulatory genes such as the differentiation gene D iff: by swordtail autosomes lacking such genes. More information about the Tu complex comes from studies on the restriction length polymorphism of the oncogene

Fig.4. Appearance of mating-conditioned development of melanoma after crossings of X. macu/alus x X. he//eri (platyfish x swordtail; A x B) and backcrossings of the F 1 hybrid (C) with x. he//eri. F and E represent the backcross generation (BCn). E1 and E2 represent deletions. The fish indicated by the capital letters correspond to those indicated in Fig. 3 by the same letters. Note that the 4. 9-k b Eco R 1 Southern fragment is inherited along with the tumor gene-complex. Ptr, pterinophore locus; Df, impaired dorsal fin-specific regulatory gene; erbB*, xiphophorine copy of an oncogene related to the viral erb B; Me/-Tu, melanophore locus containing the potential for tumor formation. Diff: a nonlinked differentiation gene;-, chromosomes of x. maculatus; … , chromosomes or X, helleri. See text

derived from platyfish and from swordtail. Some of the xiphophorine oncogenes (x-oncs) listed in Table 2 show restriction fragment length polymorphism (RFLP) the patterns of which have been differently evolved in the wild fish of different provenance [6-8,22, 40]. For instance, the patterns of the lengths of the restriction fragments of x-sis are specific to each of the different species, but show no RFLP within each of the species; actually these species show a monomorphism of the restriction fragment lengths of x-sis. In contrast, the patterns of the lengths of restriction fragments of x-erbA and x-erbB are species nonspecific, but are specific to the different races and populations of the species. The lengths of certain fragments of x-erbB are even different in females and males of the same population. We used the RFLP phenomenon as an indicator for the Mendelian inheritance of the x-oncs through the purebred and hybrid generations. If a certain oncogene fragment is independently inherited from the inheritance of spot or melanoma formation, then one can conclude that the respective oncogene is not "critical" for the first step of melanoma formation. This is not to say that such an oncogene is not involved in melanoma formation at all; as already mentioned, x-src, x-sis, x-ras, x-myc are expressed in the melanomas and are certainly involved in tumor growth or tumor progression, but they are not involved in the first step leading to melanoma because they are contributed by the swordtail to the hybrid genome whereas the appearance of the spots and the melanomas is contributed by the X chromosome of the platyfish. Furthermore, since 47 chromosomes of the malignant melanoma bearing backcross hybrids are contributed by the swordtail and only 1, name]y the Tu complex carrying X chromosome, is contributed by the platyfish, one can assume that most of the oncogenes in the genome of the tumorous backcross animals are contributed by the swordtail genome. Actually, the only x-onc detected so far on the platyfish chromosome carrying the Tu complex is the x-erb B*. This oncogene is represented in Fig. 4 by a 4.9-kb Eco Rl Southern restriction fragment which is inherited along with spot and/or melanoma development (A, C, E) and is lacking in the melanoma-free swordtail (B) and the melanoma-free BC hybrid (F). The other EcoRl fragments that a]so indicate erb B sequences could not be assigned to the X-chromosomal locus where the inheritance of the melanomas comes from. Additional information about the correlation between the inheritance of melanoma formation and the inheritance of the x-erb B*-representing 4.9-kb Southern fragment comes from two mutants of the type E BC hybrids. Both types (Fig.4, El and El) have lost the locus Me/-Tu, i.e., the capability to develop melanoma, but only one type (El) has also lost x-erb B* as is shown by the ]ack of the 4.9-kb fragment. This result indicates that (a) x-erb B* is located between Dfand Me/-Tu and (b) information crucial for melanoma formation depends on Me/- Tu, which codes for the differentiation of certain pigment cells. This is, however, not to say that there are no links in the chain of events leading to the very beginning of melanoma formation that precede the function of Me/-Tu. As was already mentioned, pp60x-src kinase activity and inositol lipid turnover activity was found enormously elevated in the melanomas. This is true for all kinds of tumors so far studied and for all types of tumor etiology (Tables 4, 5). Unexpectedly, these activities were also found elevated in the healthy tissues of the fish carrying mating-conditioned and Mendelian-inherited melanomas. figure 4 (upper part) shows the rounded data measured in the brain of the melanomatous BC hybrids type E in comparison to those of types A, B, C, F. The results suggest that the genes controlling pp60x-src and the inositol lipid turnover are expressed not on]y in the melanoma tissues but also in the healthy tissues of the tumorous individuals, independently of whether they are involved in neoplasia or not [46, 49]. Possibly this phenomenon corresponds to the often-occurring multiple tumors in combinations such as melanoma, neuroblastoma, rhabdomyosarcoma, and retinoblastoma in the BC segregants, sometimes even in a particular anima]. Multiple tumors and cancer family syndromes have been reported also in humans [54]. The working group of Lampert [55], for instance, studied a family which, despite a healthy ancestry, developed neuroblastoma, ganglioneuroma, and other neurogenic tumors running through two generations. Lynch and his colleagues [ 56] reported the pedigree of a family afflicted with cancer on breast, urinary b]adder, brain, colon, cervix, endometrium, pancreas, prostate, skin, stomach, and uterus. We cannot explain this phenomenon, but the model shows us the possibility of an approach to the study of some of its molecular and biochemica] fundamentals. It appears that the measurements of pp60x-src kinase activity and inositol incorporation into phosphoinositides in the brain of the deletion mutants of the fish which are incapable of developing melanoma (Fig. 4, right) open new possibilities for intervention in key signals critica] to the endogenous induction of

Fig.5. Incorporation of [³H] inositol in phosphatidylinositol in brain extracts, plotted against activity of pp60x-src .Capital letters correspond to the fish indicated by the same letters in Figs. 3 and 4; E3 ist not shown. Eg corresponds to the promoter-sensitive fish shown in Fig. 9 (on right) Note the high correlation of both parameters. Data from [49]. See text

neoplastic transformation in the animal model and possibly in humans. Both pp60x-src kinase activity and inositol lipid turnover activity are highly elevated in the brain of those insusceptible deletion animals that have lost the Mel- Tu locus but have retained the x-erbB* oncogene (Fig. 4, E1). In contrast, the deletion animals having lost the x-erb B* together with the Mel- Tu locus (E2) exhibit no elevation. This result suggests that the molecular and biochemical machinery supposedly involved in melanoma formation may be running for genetic reasons, without forming melanoma. Our results, moreover, suggest that there may be a particular type of activation of x-.src and the inositol phospholipid system that is a marker for predisposition to cancer and could be used for the determination of pro-neoplasia conditions in cancer risk studies. Support for this suggestion comes from the excellent correlation existing between pp60x-src kinase activity and the [³H]inositol incorporation into phosphatidylinositol (Fig. 5). One more suggestion arises if one compares the different results obtained with the E1 and E2 BC hybrids. Because of the backcross procedure applied to the animals most of the genes involved in melanoma formation are contributed to the hybrids by the swordtail genome. In the deletion hybrid E2 lacking x-erbB* they appear to rest in low activity, indicating that, in order to become involved in melanoma formation, they require a signal for the change from a resting to activated state. The results obtained with the deletion hybrid El show that this signal is transmitted from that region of the Tu complex containing platyfish chromosome where x-erb B* is located and where the inheritance of the melanoma is determined. In conclusion, based on the possibility of distinguishing between genes originating from platyfish and swordtail in the genome or certain hybrids, we round that development and growth or melanoma is mainly run by a set or genes that requires a signal for its activation which, due to the onset or the crossing experiments with the mutants, is transmitted from an x-erb B*-containing chromosome locus. This locus, however, is probably deregulated by the crossing-conditioned replacement of platyfish chromosomes carrying regulatory genes for the Tu complex (i. e., probably x-erb B*) by swordtail chromosomes lacking them. The 4.9-kb Eco R1 restriction fragment was cloned, subcloned, and sequenced. It contains exon c and d or the kinase domain and shows high homology to the respective sequences or the human epidermal growth factor receptor (HEGFR) gene and to the viral erbB (for complete data see [22, 40]). Hybridization or this xiphophorine fragment against genomic xiphophorine DNA revealed the presence or highly homologous sequences located on the Y-chromosome (6.7 kb; see later), on the Z-chromosome, and on an autosome present in all individuals. Another species, Xiphophorus variatus, which was studied for comparison, also exhibited an homologous fragment which is inherited along with tumor susceptibility. Each of the x-erb B* copies corresponding to these homologous fragments from different chromosomes is also part of a Tu complex [40]. Hybrids carrying these Tu complexes, however, require treatment with carcinogens as a precondition for melanoma development.

H. Carcinogen-Dependent Neoplasia

The remainder or my review or human cancer is devoted to the large groups or mutagen-carcinogen conditioned (somatic mutation conditioned) and nutrient and endocrine conditioned (promoter conditioned) neoplasms. Both types or etiology comprise probably more than 90% of all tumors. A large body or consistent and contradictory observations on their causation are available. Lung tumors or humans probably offer the most convincing observations on the involvement or exogeneously induced somatic mutations in the initiation or the tumor. They appear not to be influenced by many environmental factors, and there is no evidence that hormonal or nutritional factors are involved in their causation. The simple interpretation or the induction or a somatic mutation by a physical or chemical carcinogen, however, does not explain the different susceptibility or the different individuals that are exposed to the carcinogen. There must exist hereditary factors that enable most of the individuals to escape lung cancer while others become victims. We cannot explain this observation. Recently Newman and her colleagues [57] reported on breast cancer in an extended family (Fig. 6). A complex segregation analysis indicated that susceptibility to breast cancer in the family can be explained by autosomal inheritance or a defective regulatory gene while the appearance or the tumor requires a somatic mutation in a target cell. This example shows that steps toward breast cancer had already occurred unnoticed in the preceding generation; the somatic mutation represents only the last step that completes the chain or events leading to cancer. The Xiphophorus model provided more details for the study or the complex situation in the somatic mutation-dependent tumors. In mutagenesis studies [52] we detected nontumorous hybrid genotypes which, following treatment with directly acting carcinogens (X-rays, MNU), develop after a latent period or 8-12 months foci or transformed pigment cells that grow out to compact melanomas (Fig. 7). The smallest cell clones to which these melanomas could be traced consisted or eight cells indicating that there were three cell divisions between a somatic mutation event and the occurrence or the transformed pigment cells [23]. The incidence or these tumors depends on the dosage or the treatment, and may reach up to 100%.

Fig. 6. Pedigree of a family at high risk of breast cancer, adapted from(57). See text

Fig.7. Mutagen-carcinogen-sensitive fish developing MNU-induced melanoma. Note the closely circumscribed growth reminiscent of the somatic mutation-conditioned unicellular origin of the tumor

These observations led to the assumption that the Tu complex of the treated hybrids is under control of only one regulatory gene which, following treatment, is impaired in a particular pigment cell. Assuming the total of the pigment cell precursors that are competent for neoplastic transformation is 10 high 6 (this is the average number in the pigment cell sys tem of young fish), and the induced mutation rate is 10 high-6, then the tumor incidence is 1 ( on average 100% of the treated animals will develop one tumor). If, however, the Tu complex is under the control of two regulatory genes, the rate of simultaneous mutations of both of these regulatory genes in 1 cell is 10 high-12, and the tumor incidence is 10 high-6. This calculation shows that it is difficult to succeed in inducing somatic mutationconditioned neoplasms if the Tu complex is controlled by more than one regulatory gene. This calculation also suggests that the insusceptibility of the animals of the purebred wild populations is based on a polygenic system of regulatory genes directed against cancer. Support for the assumption that the Tu complex of these animals is controlled by only one regulatory gene comes from germinal mutation-conditioned melanoma which occurred in the same genotype. As a consequence of the inheritance of the mutation through the germ line, the Tu complex becomes active in the developing progeny as soon as the pigment cell precursors become competent for neoplastic transformation. This process starts in the embryo and continues in all areas of the developing fish where the pigment cell precursors become competent, thus building a lethal "whole body melanoma," which reflects the genuine effect of the Tu complex on the pigment cell system. It should be emphasized that the tumorous growths that appear on germinal inherited melanoma (and other hereditary neoplasms), i. e., both the mating-conditioned and the germ line mutation-conditioned melanomas, are not due to the occurrence of somatic mutations during development, because, in contrast to the somatic mutation-conditioned tumors, the transformed cells always occur simultaneously in large areas of the body and show permanent transformation and relapse after complete removal. To study the molecular and biochemical background of the somatic mutationconditioned melanomas we modified the experiment that led to mating-conditioned spontaneously occurring melanomas (see Fig. 8 and compare with Fig. 4). The Tu complex containing platyfish chromosome was replaced by another which, instead of the mutated dorsal fin

Fig.8. Crossing procedure for the production of mutagen-carcinogen-sensitive backcross hybrids. Differences to the scheme shown in Fig. 3 are the replacement of the mutated Dr' by the nonmutated body side-specific regulatory gene Bs that suppresses melanoma formation. and the replacement of the 4.9-kb EcoR1 fragment indicating x-erbB * by a 6.7-kb fragment indicating the same x-erbB * gene. See text

specific regulatory gene Dl', contains the nonmutated body side-specific regulatory gene Bs; in addition, the x-erb B* oncogene represented by the 4.9-kb fragment was replaced by a translocated Y-chromosomal copy that is represented by a 6.7-kb fragment. The other genetic conditions are the same as those described in Fig. 4. Melanoma development is suppressed by Bs in all purebred and hybrid animals carrying the Tu complex. All BC hybrids carrying the Tu complex including x-erbB* (they can be recognized by their pterinophore-specific reddish coloration coded by Ptr) are susceptible to melanoma (and other neoplasms) and may develop melanoma after treatment with physica] or chemical carcinogens. Susceptibility to neoplasia or sensitivity to carcinogens, respectively, is inherited in a Mendelian fashion, but the tumors are, as a consequence of a somatic mutation of Bs to Bs', nonhereditary and show no relapse after complete removal. In contrast to the mating-conditioned spontaneous melanoma developing BC hybrids the carcinogen-sensitive HC hybrids show no elevation of pp60x-src activity as well as no elevation of inositol lipid turnover in the brain. Elevations of these functions are only detected in the neoplasm.

I. Nutrient- and Endocrine-Conditioned Neoplasia

Evidence for nutritional as well as endogenous and exogenous hormonal influences on human cancer has been accumulating over the past 20 years [58]. The agents exerting these influences, of ten called ""promoters" or ""cocarcinogens,,' are by no means mutagenic carcinogens, i. e., '"initiators," but appear as agents affecting the course of differentation and the rate of proliferation of cells that have already undergone the genetic key event underlying neoplasia irrespective of whether they are tumor precursor cells or definite tumor cells; the changes in cell differentiation and cell proliferation appear as the last step in the chain of events resulting in cancer . Many data on this subject come from epidemiological studies [59, 60]. It has been found that breast and colon cancer, which represent a high percentage of total neoplasias in humans, are highly correlated to animal tat intake in a large number of countries, and it has been proposed that low animal fat intake is responsible for a low incidence of these neoplasms, while high animal fat intake is responsible for a high. incidence. The order of countries begins (low fat intake, low tumor rate) with Thailand, the Philippines, Japan, Taiwan, continues to Czechoslovakia, Austria, France, Switzerland, Poland, the Netherlands, and Finland, and ends with the United States, Canada, Denmark, and New Zealand (high fat intake, high tumor rate). A more critical view, however, indicates that the tumor incidence of the Dutch is twice as high as that of the Finns, though both have the same fat intake. The same is true, if we compare the Swiss (high tumor incidence) with the Poles (low tumor incidence, but same fat intake). The Danes have an extremely high animal fat intake and an extremely high incidence of breast cancer. If one compares, however , the population of Copenhagen with that of the rural Denmark one finds that fat intake in Copenhagen is much lower than in rural Denmark while urban Danes have a higher tumor incidence than rural Danes. This is not to say that fat intake will have no influence on the incidence of breast and colon cancer; however, our critical view of the data makes clear that fat intake alone cannot explain the differences in tumor incidence in different countries. There could be genetic factors involved in such a way that countries showing a high tumor incidence not only have a high fat intake but also contain a high percentage of individuals that are highly sensitive to the tumor-promoting effect of the fat. These genetic factors may also be related to an effect on normal body growth as has been reported in mouse studies [61, 62]. Thus, these genes might interact with a multitude of other nutritional factors, such as simple caloric intake, quantity and quality of protein ingested, as well as drugs that influence the general condition of an individual. Our own studies concentrated first on the construction of strains of Xiphophorus that are highly sensitive to tumor promoters. Figure 9 shows the development of such a strain based upon the same genotypes and crossing procedures as were used for the production of BC hybrids that develop melanoma spontaneously (see fig. 4). The only difference is that the genome of the animals contains a homozygous autosomal gene, ""golden" (gig), by which pigment cell differentiation is delayed in the stage of stem melanoblasts. Thus, the BC hybrids corresponding to those developing malignant melanoma spontaneously are incapable of developing a neoplasm. Chemical agents, such as cyclic AMP , corticotropin, a large variety of steroid hormones including testosterone, trenbolone [41 ], as well as general environmental changes, such as decrease in temperature and increase in salinity of the water in the tank, promote after a latent period of only 4 weeks (latent period of the carcinogen-dependent melanomas is 8 -12 months; see preceding paragraph) almost simultaneously the differentiation of large amounts of the noncompetent cells to the competent ones, which subsequently give rise to the melanoma exactly at that place at the body of the fish where they are expected to grow according to the basic crossing experiment (compare Figs. 4, 9).

Fig. 9. Crossing procedure for the production of promoter-sensitive backcross hybrids. The only difference to the scheme shown in Fig. 3 is the presence of the homozygous gene "golden", g/g, by which pigment cell differentiation is delayed. See text

The very short latent period and the very fast growth of the occurring melanomas as compared with that of the carcinogen-induced tumors is remarkable, but is in line with the enhanced pp60x-src kinase activity and the enhanced phosphoinositide turnover found in the healthy tissues. It appears that, corresponding to the deletion mutant El (see Fig.4), the molecular and biochemical machinery leading to neoplasia is running in the susceptible but still tumorfree fish and becomes immediately effective as the competent cells become available for promotion of cell differentiation. The latter results, again, indicate that both the enhanced activity of the xiphophorine src oncogene and the enhanced phosphoinositide turnover are intimately linked with the inheritance of x-erb B*, which is presumably involved in the key signal preceding melanoma formation in Xiphophorus. They furthermore show again that it should be possible to screen for sensitivity and insensitivity to tumor promoters.

J. Future Goals

In this lecture I have tried to explain some observations on human cancer from the view of a biologist working with a fish model. Of course, what I have presented is not altogether new. Nevertheless, what can we learn from the fish? First of all we should make informed decisions to control the chemical and physical carcinogens and promoters we receive today from our polluted environment. However, we should keep in mind that cancer not only depends on the agents but also on the genes that have been part of our evolution since life began. These genes have experienced mutation, duplication, selection, and genetic drift, and are controlled by oncostatic genes that keep a tight rein on them. To learn how these regulatory genes keep the oncogenes in check should be an challenging but fulfilling task in the future of cancer research.

Thanks are due to Professor Steward Sell, Houston (Texas) and to Prolessor Avril Woodhead, Brookhaven Natl. Lab. (New York), for discussions and critical reading of the manuscript.


1. Kaiser HE (1981) Neoplasms -comparative pathology of growth in animals, plants, and man Williams and Wilkins, Baltimore
2. Anders F (1983) The biology of an oncogene, based upon studies on neoplasia. In. Neth R, Gallo RC, Greaves MF, Moore MAS, Winkler K (eds) Modern trends in human leukemia V. Springer, Berlin Heidel berg New York, pp 186-206 (Haematology and blood transfusion, vol 28.)
3. Anders F, Schartl M, Barnekow A, Schmidt CR, Lüke W, Jaellel-Dess G, Anders A ( 1985) The genes that carcinogens act upon In: Neth R, Gallo CR, Greaves MF, Janka G (eds) Modern trends in human leukemia VI. Springer, Berlin Heidelberg New York, pp 228-252 (Haematology and blood transfusion, vol 29.)
4. Burk O (1987) Konservierung voll homologen Sequenzen der kernproteinkodierten Onkogene v-fos, v-myc und v-myb im Tierreich. Thesis, University of Giessen
5. Kaiser P ( 1988) Restriktionsanalytische Untersuchungen der Konservierung von homologen Sequenzen ZU den Onkogencn v-erb A, v-crb B and c-ncu im Tierreich. Thesis, University of Giessen
6. Schleenbccker U ( 1988) Untersuchungen zur Struktur und Funktion Voll homologen Sequenzen des Wachstumsfaktors und des Wachstumsfaktorrezeptors bei Xiphophorus (Teleostei. Poeciliidae). Thesis, University of Giessen
7. Zechel C (1988) Untersuchungen zur Struktur und Funktion v-crb A- und v-crb B-homologer Sequenzen im XiphophorusTumorsystem. Thesis, University of Giessen
8. Schmidt D (1988) Restriktionsanalytische Untersuchungen an genomischer DNA Voll Xiphophorus. Thesis, University 0f Giessen
9. Gröger H (1987) Isolierung und Charakterisierung von c-myc-spezifischen Klonell aus einer lambda EMBL 4 Genbank von Xiphophorus maculatus ( DrLi/Sr ) . Thesis, University 0f Giessen
10. Pfütz M (1988) Sequenzierung c-crb Aspezifischer Sequenzen aus dem Genom voll Xiphophorus. Thesis, University 0f Giessen
11. DeFeo-Jones D, Scolnik E, Koller R, Dhar R (1983). ras-Related gene sequences identified and isolated from Saccharomyccs cerevisiae. Nature 306.707709 12. Pimentel E (1986) Oncogenes. CRC, Boca Ratoll
13. Reddy EP, Skalka AM, Currall T (1988) The oncogenc handbook. Elscvicr, New York
14. Berridge MJ (1987) Insosit01 triphosphatc and diacylglyccr01. TWO interacting second messengers. Allnu Rev Biochem 56.159-193
15 Berridge MJ (1987) Inosit011ipids and cell prolifcratioll Biochim Biophys Acta 907.33-45
16. Berridge MJ, Irvine RF (1984) Inositol triphosphate, a novel second messenger in cellular signal transduction. Nature 312.315-321
17. Schartl M, Barnekow A (1982) The expression in eukaryotes of a tyrosine kinase which is reactive with pp 60v-srr antibodies. Differentiation 23.109-113
18. Barnekow A, Schartl M ( 1983) Cellular src gene product detected in the freshwater spongc Spongilla lacruslis. Mol Cell Biol 4:1179-1181
19. Albino AP, Strange R L, Oliff AL, Furth ME, Old L.J (1984) Transforming ras genes from human melanoma: a manifestation of tumor heterogeneity. Nature 308.69-72
20. Barnckow A, Paul E, Schartl M (1987) Expression of the c-sre protooncogene in human skin tumors. Cancer Rcs 47.235-240
21. Ohno S (1970) Evolution by gene duplication Springer, Berlin Hcidelberg New York
22. Zechel C, Schleenbecker U, Anders A, Anders F (1988) v-erbB related sequences in Xiphophorus that map to melanoma determining Mendelian loci and overexpress in a melanoma cell line Oncogene 3.605-617
23. Anders A, Anders F (1978) Etiology of neoplasia as studied in the platyfish swordtail systeru. Eiochiru Eiophys Acta 516:61-95
24. Anders F ( 1967) Tumor formation in platyflsh swordtail hybrids as a problcru of gene regulation. Experientia 23.1-10
25 Vielkind J, Dippel E (1984) Oncogcnc-related sequences in xiphophorine fish prone to hereditary melanoma formation. Can J Genet Cytol 26: 607 -614
26. Anders F, Schartl M, Scholl E (1981) Evaluation of environmental and hereditary factors in carcinogenesis, based on studies in Xiphophorus. In. Dawe CD, Harshbarger JC, Kondo S, Sugimura T, Takayama S (eds) Phyletic approaches to cancer. Japan Scientific Societies, Tokyo, pp 289-309
27. Gordon M (1947) Speciation in fishes. Adv Genet 1.95-132
28. Kallman KD (1975) The platyfish, Xipho rus maculatus. Handb Genet 4: 81 -132
29. Anders F, Schartl M, Earnekow A, Anders A ( 1984) Xiphophorus as a in vivo model for studies on normal and defective control of oncogenes. Adv Cancer Res 42.191-275
30. Moriwaki K, Shiroishi T, Miyashita N, Kondo N, Imai H (1980) Intersubspecies hybrid of mice as a tool for studying genetic governing tumor development Gann Monogr 25.165 176
31. Anders F (1981) Erb- und Umweltfaktoren im Ursachengefüge des neoplastischen Wachstums nach Studien an Xiphophorus Verhandl Ges Dtsch Naturforscher Ärzte 22. 106-119
32. O'Erien S (1980) The extent and character of biochemical genetic variation in the domestic cat. J Hered 71.2-8
33. Fuerst PA, Chakraborty R, Nei M (1977) Statistical studies on protein polymorphism in natural populations I. Distribution ofsinglc locus hetcrozygosity. Genetics 86.455-483
34. Schull W J (1979) Genetic structure of human populations. J Toxicol Environ Health 5 17-25
35. Lubs IIA, Kimberling WJ, Hecht F, Patil SR, Brown J, Gerald P, Summit RL (1977) Racial differences in the frequency of Q and G chromosomal heteromorphisms. Nature 268.631-632
36. Yamada K, Hasegawa T ( 1978) Types and frequencies of Q-variant chromosomes in a Japanese population. Hum Genet 44. 89-98
37. Higginson J (1969) Present trends in cancer epidemiology. Proc 8th Canadian Cancer Conf. Pergamon, Toronto, pp 4075
38. I Iigginson J ( 1988) Changing concepts in cancer prevention. limitations and implications for future research in environmental carcinogcnesis. Cancer Res 48. 1383 1389
39. Iloover K (1984) Use of small fish species in carcinogenicity testing. NCI Monogr 65
40. Zechel C, Schleenbecker U, Anders A, Anders F (1988) Regulation of gene expression in the Gordon-Kosswig mclanoma system. In. Schröder H (ed) Genetics and mutagenesis of fish. (in press)
41. Montesano R, Earth H, Vainio H, Wilborn J, Yamasaki H (1986) Long-term and short -term assays for carcinogenesis. IARC scientifIc publications no 83. IARC, Lyon, pp 103-126
42. Mawdesley- Thomas LE (1975) Neoplasia in fish. In. Tibelin WE, Migaki G (eds) The pathology of fishes. University of Wisconsin Press, pp 805-870
43. Dahme E, Weiss E ( 1988) Grundriß der speziellen pathologischen Anatomie der Haustiere, 4th edn. Enke, Stuttgart
44. Mäueler W (1988) Untersuchungen zur tumorspezifischcn Genexpression bei Xiphophorus (Teleostei. Poeciliidae). 1. Enzyme des Intermediärstoffwechsels, 2. Expression von Proto-Onkogenen. Thesis, University of Giessen
45. Krekeler G (1988) Isolierung und Charakterisierung v-myb-homologer Sequenzen aus einer Genbank von Xiphophorus (Pisces. Poeciliidae). Thesis, University of Giessen
46. Schartl M, Schmidt CR, Anders A, Eamekow A ( 1985) Elevated expression of the cellular src gene of differing etiologies in Xiphophorus. lnt J Cancer 36: 199207
47. Gronau T (1987) Untersuchungen zur Organisation, Aktivität und Wirkung des zellulären Onkogens c-src iru Xiphophorus- Tumorsystem. Thesis, University of Giessen
48. Pröfrock A (1988) Untersuchungen zuru Phosphatidylinosit- Tumovcr an ausgewählten Xiphophorus-Genotypen. Thesis, University of Gicssen
49. Smith AD, Gronau T, Pröfrock A, Zcchcl C, Bird IM, Lane PA, Earnekow A, Anders A, Anders F ( 1988) EG F receptor gene, inositol lipid turnover and C-src activity in key processes preceding melanoma in Xiphophorus. In. Lynch HT, Fusaro RM (eds) Hereditary Malignant Melanoma, CRC, Boca Raton (in press)
50. lynch HT, Lynch JF, Fusaro RM (1985) Clinical importance of familial cancer, In' Müller H, Weber W, Kuttapa T (eds) Familial cancer, Karger, Basel, pp 6-12
51. Ahu,ja MR, Lepper K, Anders F (1979) Sex chromosome aberrations involving loss and translocation of tumor-indueing loci in Xiphuphurus, Experientia 35'28-29
52. Anders A, Anders F, Klinke K (1973) Regulation of gene expression in the Gordon-Kosswig melanoma system, In' Schröder H (ed) Genetics and mutagenesis of fish, Springer, Berlin Heidelberg New York, pp 33-63
53., Anders A, Dess G, Nishimura S, Kersten H (1985) A molecular approach to the study of malignancy and benignancy in melanoma of Xiphuphuru,s, In' Bagnara J Klaus SN, Paul E, Schartl M (eds) Pigment cell 1985 University of Tokyo Press, pp 315-324
54. Müller HJ Weber W, Kuttapa T (1985) Familial cancer, Karger, Basel
55. Rudolph B, 1Harbott J Lampert F (1988) Fragile sites and neuroblastoma' Fragile site at one p 13,1 and other points on lymphocyte chromosomes from patients and family members, Caneer Genet Cytogenet 31'83-94
56. Lyneh HT, Lynch JF (1985) Genetics and eolorectal cancer, In' Müller HJ Weber W, Kuttapa T (eds) Familial cancer, Karger, Easel
57. Newman E, Austin MA, Lee M, King MC (1988) Inheritance of human breast cancer' Evidence for autosomal dominant transmission in high-risk families, Proe Natl Aead Sci USA 85'3044-3048
58. Cohen LA (1987) Diet and cancer Sci Am 257 (5)42-48
59. Hecker E, Fusenig NE, Kunz W, Marks F, Thielmann HW ( 1982) Cocarcinogenesis and biological effects of tumor promoters, Raven, New York
60. Carroll KK (1975) Experimental evidence of dietary factors and hormone-dependent cancers Cancer Res 35: 3374-3383
61 Wynder EL (1975) The epidemiology of large bowel cancer, Cancer Res 35'3388-3394
62. Heston WE (1957) Effects of genes located on chromosomes III, V, VII, IX, and XIV on the occurrence of pulmonary tumors in the mouse, Proc Intern Genetics Symposia, 1956 Cytologica [Suppl]' pp 219-224