The constitutional chromosomal deletions in several dominant neoplastic diseases supported the two-hit hypothesis, but such examples (Table 5.3) were relatively unusual, and the biochemical evidence, like that exemplified by the esterase D changes noted in hereditary retinoblastoma, was exceptional. More recent techniques have allowed more detailed analysis of genetic material to the point of and including the isolation and structural characterization of the genes involved in many dominantly inherited diseases with high incidences of neoplasia (Table 5.4).
As we have already noted (Table 5.2), by the two-hit theory, both alleles of the gene in- volved in a specific neoplasm, whether of hereditary or spontaneous origin, will have lost the function of the gene in question. In the case of a germline inherited neoplastic condition, all cells of the organism will exhibit heterozygosity of the gene locus, one allele being functional and the other not. Thus, the neoplasm will have lost this heterozygous state to become homozygous for the defective gene. Below it is shown that loss of heterozygosity can also be determined in dis- eases of somatic cell origin since, by the two-hit theory, they would also be expected to exhibit a homozygous genotype, as noted in Table 5.2.
The mutated gene in dominantly inherited neoplastic disease may be the result of point mutations, small or larger deletions, or nonsense mutations leading to a truncation of the protein. Mutation in the second allele, especially in dominantly inherited neoplasms, will be the result of chromosomal abnormalities such as translocations, major deletions, recombination, etc. (Lasko et al., 1991). It is also possible, however, that loss of function of the second allele may be far more subtle, such as point mutations or minor deletions. When such changes occur, examination of the karyotype may not be informative, since no chromosomal aberration can be detected. However, one may use a molecular analysis, termed restriction fragment length polymorphism (RFLP) analysis, which involves the use of restriction enzymes that cleave DNA strands at spe-
Table 5.4 Structure and Function of Genes of Dominantly Inherited Neoplastic Diseases
cific base sequences. Since numerous such restriction enzymes, each acting on a specific se- quence of DNA extending from four to seven or more bases, are known (Roberts and Macelis,1992), one may survey a variety of sequences in the DNA for changes. Such an RFLP analysis is diagrammed in Figure 5.5 between normal and tumor DNA, using two examples of DNA struc- ture. In one case the normal DNA of a gene is cleaved in three places by a single restriction enzyme. This leads to three possible fragments of different sizes. In the second instance the dia- gram involves a segment of DNA containing numerous repeated sequences. In each instance shown, each horizontal line represents a segment of one allele of the gene investigated. In the case where tandem repeats are present and the restriction enzyme cuts adjacent to the repeats as well as somewhere else in the genome, only two fragments are obtained. If there is a deletion of one of the copies or of the segment with a restriction sequence or if there is a mutation of the restriction sequence, then digestion of the DNA in each case results in only a single band. Elec- trophoresis of the bands and their reaction with the probe, a complementary sequence to a por- tion of the DNA, as shown in the figure, results in a pattern shown in the lower portion of the
Figure 5.5 Restriction fragment length polymorphism (RFLP) method for the detection of allelic loss in somatic cells. Minor variation in base sequences or repetitive sequences may occur without alteration of the proteins encoded. Such alterations induce variations in recognition sites for restriction enzymes which cleave DNA at specific sequences. Following cleavage the DNA is electrophoresed, bands (molecules) sep- arated, blotted to a filter and hybridized with a radiolabeled DNA probe which recognizes a sequence in- volved in the restriction reaction. In the Southern blot hybridization depicted, the normal genomic DNA on the left contains a polymorphism recognized by the probe as two bands of different molecular sizes, each of which corresponds to one of the two allelic copies. If one of the alleles has been deleted as a result of restriction digestion, then one band will be missing allowing molecular confirmation of the deletion event. (Modified from Viallet and Minna, 1990, with permission of the authors and publisher.)
figure. The small band expected in the digestion of normal DNA without repeats would have migrated off the gel and not be seen. As noted, in both instances one copy of the DNA is lost in the neoplasm as a result of genetic alteration, thus demonstrating the loss of heterozygosity in the tumor DNA. Table 5.5 lists dominantly inherited neoplastic diseases in which a loss of het- erozygosity has been demonstrated. Examination of the table reveals that a number of regions of loss of heterozygosity do not necessarily indicate the chromosomal localization of the gene for the disease (Table 5.6), and in some instances several different chromosomal sites with loss of heterozygosity may be noted in the same genetic condition—e.g., MEN II, familial breast can- cer. Furthermore, in this example, the chromosomal localization of the gene primarily affected in MEN II does not exhibit loss of heterozygosity (Table 5.6). This conforms to the nature of the gene involved, the ret proto-oncogene, which was first isolated from DNA originating from a human T-cell lymphoma (Takahashi et al., 1985). As discussed previously (Chapter 4), alter- ations in proto-oncogenes are dominant in their effects, and thus mutation in a single allele of the ret gene might be expected to result in neoplastic disease. In Table 5.5, there are also several instances in which the loss of heterozygosity has been used to aid in the localization and ulti- mate isolation and characterization of the gene affected in the disease under study—e.g., von Hippel–Lindau syndrome, familial melanoma, tuberous sclerosis.
Through the use of the two techniques described above, cytogenetic analysis and RFLP analysis, the chromosomal localization of a number of autosomal dominant conditions predis- posing to specific neoplastic disease has been elucidated (Table 5.6). With localization of the genes to specific chromosomal regions coupled with a knowledge of other genes and genetic markers in the region of the locus being investigated (Figure 5.4), several of these genes have been isolated and characterized through the use of a variety of molecular biological and molecu- lar genetic techniques. A reasonably up-to-date listing of genes that are mutated in specific dom- inantly inherited neoplastic disease and whose sequence structure has been elucidated are listed
in Table 5.6. While it is beyond the scope of this text to depict the structure and sequences of the genes listed in Tables 5.4 and 5.6, examples of a single, relatively large gene, that for retinoblas- toma (over 200 kb pairs), and a relatively small gene (about 50 kb pairs), that for the von Hip- pel–Lindau syndrome, are depicted in Figure 5.6. Thus, the degree of complexity of the genes is not a factor in whether they served as the basis for dominantly inherited neoplastic disease. All of the genes listed in Table 5.6 with the exception of that for MEN IIA and B conform to the two-hit theory of Knudson, in that both alleles of the gene must be mutated in order that one or more neoplasms arise. As noted earlier, the principal exception to this to date is MEN IIA and B, in which only one of the two alleles need be mutated for neoplasia to arise.
At this point the student should appreciate that the evidence in support of the two-hit the- ory is quite substantial for a number of dominantly inherited neoplastic conditions. However, other mechanisms have been described in which the neoplastic phenotype may result from muta- tions in only one of the alleles of the gene in question. Dominant negative mutations of some genes noted in Table 5.6 have been described and are the result of specific alterations in the tumor suppressor gene such that the mutant protein may disrupt the function of the products of the normal allele through the formation of protein complexes between mutant and normal pro- teins or through abnormal interactions with DNA sequences that are the target of the normal gene. This phenomenon has now been described for the WT1 (Reddy et al., 1995) and the p53 genes (Srivastava et al., 1993).
Another mechanism resulting in the loss of function of both alleles in neoplastic tissues may be seen when the gene in question is imprinted in one or the other parent (Rainier and Fein- berg, 1994). Gene imprinting occurs when the allele of a gene inherited from one parent is re- pressed and not expressed in cells of the offspring. A diagram of such a phenomenon is seen in Figure 5.7 in relation to the imprinting of a tumor suppressor gene. Imprinting of the gene is presumed to involve methylation of the DNA bases, resulting in repression and nonexpression of
Figure 5.6 A. Diagram of the RB-1 gene, which consists of 27 exons spread over a length of 200 kb. (From Kloss et al., 1991, with permission of the authors and publisher.) B. Structure of the VHL tumor suppressor gene. The boxes show the exons with the coding region dark and the 3′ untranslated region unshaded. The numbers correspond to the nucleotide numbers in the cDNA, and the codon numbers are in parenthesis. (Reproduced from Foster et al., 1994, with permission of the authors and publisher.)
Figure 5.7 Parental imprinting of the insulin growth factor-2 (IGF2) gene, which is potentially involved in Wilms tumor development. With imprinting of one gene, only mutation or loss of the other gene is required to eliminate its function. (Reproduced from Sapienza, 1990, with permission of the author and publisher.)
the gene from that parent (Razin and Cedar, 1994). The methylation is inherited epigenetically from one or the other parent, thus resulting in the expression of only a single, nonmethylated gene from the other parent. While as yet there is no clear and distinct example that has been proven to follow the scheme seen in Figure 5.7, there is evidence of the imprinting of tumor suppressor genes or other related genes affecting or affected by the expression of a tumor suppressor gene. In Table 5.7 are listed some examples of dominantly inherited neoplasia for which there is some evidence for the imprinting (repression) of genes, both known and un- known. New mutations giving rise to bilateral retinoblastoma exhibit a marked preference for the paternal allele of the RB gene, suggesting as one mechanism that imprinting early in embry- onic life affects chromosomal susceptibility to mutation (Zhu et al., 1989). Similarly, in MEN II, all new mutations were of paternal origin (Carlson et al., 1994). Alleles in the neoplasms of familial and sporadic cases of embryonal rhabdomyosarcoma are of paternal origin (Scrable et al., 1989). In contrast, in familial glomus tumors it is the maternal allele that appears to be imprinted (repressed), although the exact nature of the gene is as yet unknown (van der Mey et al., 1989).
A potential role for genomic imprinting in carcinogenesis is best exhibited by findings of the Beckwith-Wiedemann syndrome. In addition to various somatic manifestations including or- ganomegaly (kidney, liver, and adrenal), hemihypertrophy, and gigantism, neoplasms arise in these individuals, including Wilms tumor and embryonal rhabdomyosarcoma (cf. Tycko, 1994). Although there is not substantial evidence for imprinting of the WT1 gene, two genes found very close to WT1, H19 and IGF2, are imprinted from different parents. H19 is expressed from the maternal allele and IGF2 from the paternal allele. Rainier demonstrated that more than two- thirds of Wilms tumors exhibiting no loss of heterozygosity in the region of the WT1 gene ex-
hibited biallelic expression of one or both of these genes. Since IGF2 codes for a growth factor, one possible mechanism contributing to the development of this neoplasm is the loss of the reg- ulation of imprinting of this growth factor gene as well as H19, a function for which is as yet unknown.
The two-hit theory of Knudson has gained substantial evidence in favor of its functioning as a major mechanism in carcinogenesis. However, it is clear from this brief discussion that other genetic and epigenetic mechanisms are involved in the development of malignant neopla- sia. A number of these mechanisms are the subject of later chapters in the text.