p53 tumor suppressor gene: at the crossroads of molecular carcinogenesis, molecular epidemiology, and cancer risk assessment.

Carcinogenesis is a multistage process involving the inappropriate activation of normal cellular genes to become oncogenes, e.g., ras, and the inactivation of other cellular genes called tumor suppressor genes. p53 is the prototypic tumor suppressor gene that is well suited as a molecular link between the causes of cancer, i.e., carcinogenic chemical and physical agents and certain viruses, and the development of clinical cancer. The p53 tumor suppressor gene is mutated in the majority of human cancers. Genetic analysis of human cancer is providing clues to the etiology of these diverse tumors and to the functions of the p53 gene. Some of the mutations in the p53 gene reflect endogenous causes of cancer, whereas others are characteristic of carcinogens found in our environment. In geographic areas where hepatitis B virus and a dietary chemical carcinogen, aflatoxin B1, are risk factors of liver cancer, a molecular signature of the chemical carcinogen is found in the mutated p53 gene. A different molecular signature in the p53 gene is found in skin cancer caused by sunlight. Because mutations in the p53 gene can occur in precancerous lesions in the lung, breast, esophagus, and colon, molecular analysis of the p53 gene in exfoliated cells found in either body fluids or tissue biopsies may identify individuals at increased cancer risk. p53 mutations in tumors generally indicate a poorer prognosis. In summary, the recent history of p53 investigations is a paradigm in cancer research, illustrating both the convergence of previously parallel lines of basic, clinical, and epidemiologic investigation and the rapid translation of research findings from the laboratory to the clinic.

p53 tumor suppressor gene is the most striking example because it is mutated in about half of almost all cancer types arising from a wide spectrum of tissues. Other tumor suppressor genes important in human oncology such as APC, WT1, p16INK4, or NFl may have a more limited distribution (Table 1); given the variety of hereditary cancers and allelic deletions found in human cancers, additional tumor suppressor genes should be identified in the future, some of which may also have a conspicuous role in carcinogenesis.
Tumor suppressor genes are vulnerable sites for critical DNA damage because normally they function as physiological barriers against clonal expansion or genomic mutability and are able to hinder growth and metastasis of cells driven to uncontrolled proliferation by oncogenes.
Loss of tumor suppressor function can occur by damage to the genome through mutation, chromosomal rearrangement and nondisjunction, gene conversion, imprinting, or mitotic recombination. Tumor suppressor activity can also be neutralized by interaction with other cellular proteins or with viral oncoproteins. Comprehensive reviews of this rapidly advancing field of molecular carcinogenesis are available (4)(5)(6).
The p53 suppressor gene is the most prominent tumor suppressor gene because it is mutated in about half of human cancer cases (7,8). Although the retinoblastoma and APC tumor suppressor genes are most commonly inactivated by nonsense mutations that cause the protein to be truncated or unstable, about 80% of p53 mutations are missense mutations that change the identity of an amino acid. Changing amino acids in this way can alter the protein conformation and increase the stability of p53; it can also alter sequence-specific DNA binding and transcription factor activity of p53 (9). One explanation for the high frequency of p53 mutation is that the missense class of mutations can cause both a loss of tumor suppressor function and a gain of oncogenic function by changing the repertoire of genes whose expressions are controlled by this transcription factor (10,11). The central role of p53 in multistage carcinogenesis places it at the intellectual crossroads of molecular carcinogenesis, molecular epidemiology of human cancer, and cancer risk assessment. p53 participates in many cellular functions: cell cycle control, DNA repair, differentiation, genomic plasticity, and programmed cell death (8,10,12). p53 is one component of the DNA damage response pathway in mammalian cells ( Figure 1). Some of these normal cellular functions of p53 can be modulated and sometimes inhibited by interactions with either cellular proteins (e.g., mdm2) or oncoviral proteins (e.g., hepatitis B virus X protein) of certain DNA viruses. p53 is clearly a component in a biochemical pathway or pathways central to human carcinogenesis, and p53 mutations provide a selective advantage for clonal expansion of preneoplastic and neoplastic cells.
The mutation spectrum of p53 in human cancer can help identify particular carcinogens and define the biochemical mechanisms responsible for the genetic lesions in DNA that cause human cancer.  The frequency and type of p53 mutations can also act as a molecular dosimeter of carcinogen exposure and thereby provide information about the molecular epidemiology of human cancer risk. The p53 gene is wellsuited for this form of molecular archaeology. The majority of mutations in p53 are in the hydrophobic midregion of the protein ( Figure 2) (8). The function of the p53 protein as a transcription factor is exquisitely sensitive to conformational changes in this region that result from amino acid substitutions (13); p53 binding to other cellular and oncoviral proteins can easily be disrupted by mutations in these regions.
How can p53 mutation spectra lead to identification of the carcinogens that caused a particular tumor? Different carcinogens seem to cause different characteristic mutations. Exposure to one common carcinogen, ultraviolet light, is correlated with transition mutations at dipyrimidine sites (14); dietary aflatoxin B1 exposure is correlated with G:C to T:A transversions that lead to a serine substitution at residue 249 of p53 in hepatocellular carcinoma (15,16); and exposure to cigarette smoke is correlated with G:C to T:A transversions in lung carcinomas (17).
How these mutations arise can be further tested in the laboratory. For example, the predominant base changes in p53 found in lung cancers (G:C to T:A transversions) and skin carcinomas (C:G to T:A transitions) suggest that the causal lesion likely occurred on the nontranscribed strand, a finding that is consistent with the preferential repair after damage of the transcribed strand of active genes (18). Benzo[a]pyrene, a carcinogen in tobacco smoke, forms DNA adducts that are more slowly repaired when present on the nontranscribed strand than on the transcribed strand of the hypoxanthine (guanine) phosphoribosyltransferase gene (19), and ultraviolet light-induced cross-links of dipyrimidines in the nontranscribed DNA strand of the p53 gene also are more slowly repaired than in the transcribed strand  (20). Because DNA repair rates can be sequence dependent (21), the p53 mutation spectrum could be influenced by both the type and location of the promutagenic lesion. Transcription-repair coupling factors, the products of the mfd and XPD gene, have been recently identified and provide a mechanistic underpinning for strand-specific repair (22)(23)(24). The p53 protein binds to XPB and XPD DNA helicases in the TFIIH complex and modulates their function in nucleotide excision repair (25). Another example comes from areas of China and Mozambique where there is a high incidence of liver cancer. The high frequency of G:C to T:A transversions in human hepatocellular carcinomas in this region could be due to the high mutability of the third base of codon 249 by aflatoxin B1 or a selective growth advantage of hepatocyte clones carrying this specific p53 mutant in liver chronically infected with hepatitis B virus. Indeed, the third base of codon 249 in a human liver cell line exposed to aflatoxin B1 is preferentially mutated (26), and transfected 249se, p53 mutant enhances the growth rate of the p53 null hepatocellular carcinoma cell line, Hep3B (27). Other p53 codons show lower frequencies of G:C to T:A, G:C to A:T, and G:C to C:G mutations, which suggests that both preferential mutability and clonal selection are involved in human hepatocellular carcinogenesis. The p53 mutational spectra also can indicate that a particular cancer did not result from an environmental carcinogen but instead was caused by endogenous mutagenesis. The high frequency of C to T transitions at CpG dinucleotides in colon carcinomas (7) is consistent with mutagenesis by endogenous deamination mechanisms. A C to T transition would be generated by spontaneous deamination of 5-methylcytosine (28) or by enzymatic deamination of cytosine by DNA (cytosine-5)-methyl transferase when S-adenosylmethionine is in limiting concentration (or by both mechanisms) (29). Because oxygen radicals enhance the rate of deamination of deoxynucleotides (30,31), chronic inflammation and nitric oxide generated by nitric oxide synthases may explain why rats that inhale particulate materials, which cause inflammation but do not act directly on DNA, have a high incidence of lung cancer.
Mutations in p53 can also reveal that an individual has an increased susceptibility to cancer owing to inheritance of a germ-line mutation, a concept first proposed for the retinoblastoma (Rb) tumor suppressor gene (32). Germline p53 mutations are missense and occur frequently in the cancer-prone individuals with Li-Fraumeni syndrome (33). Laboratory animals with either a mutant p53 transgene or a deleted p53 gene, i.e., homozygous or heterozygous gene knockouts, also are particularly susceptible to cancer (34,35). These mutations in p53 are associated with instability in the rest of the genome (36). Such instability could generate multiple genetic alterations leading to cancer. Indeed, genomic instability (including gene amplification) increases in frequency in cells that lack a normal p53 gene (37,38). Furthermore, loss of the wild-type alleles of the p53 gene abrogates DNA damage-induced delay of the cell cycle in G1 (39). DNA repair of certain promutagenic lesions can proceed prior to DNA synthesis in S phase. Less time for repair would increase the frequency of mutations. Since p53 is an integral component in one pathway of programmed cell death (apoptosis) induced by DNAdamaging chemotherapeutic drugs or ionizing radiation (40,41), inactivation of p53 could increase both the pool of proliferating cells and the probability of their neoplastic transformation by inhibition of programmed cell death.
Such progress in the fields of molecular carcinogenesis and molecular epidemiology increases our ability to accurately assess cancer risk (Figure 3). Cancer risk assessment, a highly visible discipline in public health, has historically relied on classical epidemiology, from chronic exposure of rodents to potential carcinogens, and the mathematical modeling of these findings. The field has been forced to steer a prudent course of conservative risk assessment because of limited knowledge of the complex pathobiological processes during carcinogenesis; differences in the metabolism of carcinogens, different DNA repair capacities, variable genomic stability among animal species, and variation among individuals with inherited cancer predisposition have made definitive analysis of cancer risk almost impossible (5,42). Because vdq the scientific basis of risk assessment continues to be, and should continue to be, actively investigated (43). Many questions remain. Are the pathways of molecular carcinogenesis similar in rodents and humans? Because the time to develop cancer is generally shorter in rodents than in humans, could the apparent interspecies differences be due to the number of genetic and epigenetic events required for malignant progression or to the rate of transit between the events? Is the more frequent mutation of the ras protooncogenes in rodent cancer a reflection of a pathway that is parallel and equivalent to the p53 pathway in human carcinogenesis? Are the selective pressures for clonal expansion of preneoplastic and neoplastic cells in human carcinogenesis similar to those in animal models?
Investigations of the p53 tumor suppressor gene are an example of the recent progress in molecular aspects of cancer research. A better understanding of molecular carcinogenesis and molecular epidemiology will eventually decrease the qualitative and quantitative uncertainties associated with the current state of cancer risk assessment and improve public health decisions concerning cancer hazards. Indeed, determination of the type and number of mutations in p53 and other cancerrelated genes in tissues from healthy people may allow the identification of those at increased cancer risk and their consequent protection by preventive measures.