p53: a tumor suppressor hiding in plain sight

摘要

The hunt for tumor suppressor genes was a high-risk endeavor in the 1980s. Somatic cell hybridization studies provided early evidence of the existence of tumor suppressor genes by showing that tumorigenic growth was a recessive trait. Hybrid cell fusions of malignant cells with non-malignant cells could suppress tumorigenic growth, and microcell mediated-transfer of specific human chromosomes could suppress the tumorigenic growth of human cancer cell lines in immunodeficient mice (; ; ; ). Alfred Knudson’s insightful analyses of retinoblastoma incidence connected the concept of tumor suppressor genes with human disease pathogenesis (, ). What later became known as the ‘two-hit hypothesis’ proposed that patients with familial retinoblastoma inherited the first ‘hit’, an inactivating mutation in one allele of a tumor suppressor gene that predisposed them to the disease. A somatic mutation, or second ‘hit’, in the remaining allele would result in complete loss of function of the gene and contribute to tumorigenesis. In sporadic tumors, both hits were hypothesized to occur within the same somatic cell. Due to the much lower likelihood of two somatic mutations, sporadic retinoblastomas were typically unilateral and arose at a later age. Cytogenetic abnormalities and submicroscopic deletions of chromosome 13q14 revealed the chromosomal location of the putative tumor suppressor gene in retinoblastoma (; ; ; , ; ). These studies later led to the positional cloning of the RB1 gene, thus validating the two-hit hypothesis (; ; ). Mapping the chromosome 17p tumor suppressor locus In the late 1980s, our group mapped regions of chromosomal loss in colorectal cancer to identify the locations of tumor suppressor genes. The highest frequency loss involved the short arm of chromosome 17 (17p), which occurred in 75% of colorectal carcinomas (; ; ; ). Sporadic colorectal cancers posed a significant challenge compared to hereditary cancer predisposition syndromes like retinoblastoma where small constitutional deletions narrowed down the target area for analysis. To more precisely localize the candidate tumor suppressor, we performed Southern blots with a panel of 20 different polymorphic markers on 17p to evaluate loss of heterozygosity (LOH) in 58 paired samples of colorectal carcinoma and matched normal colorectal tissues. This analysis identified a minimal common region of deletion shared among all tumors in which any LOH was observed. This region encompassed approximately half of 17p (). With today’s genomic maps, we can estimate that the common region of deletion spanned 12.5 megabase pairs of DNA and contained ~577 genes, including 480 protein-coding genes. Relative to today, genomic maps in 1988 were extremely sparse and much of the genome could be considered uncharted territory in terms of the density and identity of genes. Identifying the tumor suppressor gene within this area was a daunting prospect, and we did not consider it likely when we started this project in the mid-80’s that the gene could actually be identified within a time-frame consistent with a pre-doctoral thesis. Remember that at the time (1985), oncogenes were already known but tumor suppressor genes were mythical beasts, predicted to exist but not yet sighted. p53 : oncogene or tumor suppressor gene p53 was identified in 1979 by Arnie Levine’s group, as well as four other groups, as a host protein bound to large T antigen in cells infected with the transforming virus SV40 (; ; ; ; ), and by another group as a transformation-related antigen in chemically induced mouse sarcomas (). Multiple lines of research initially supported the hypothesis that p53 was an oncogene, including association of the p53 protein with viral transforming proteins of SV40 or adenovirus in infected cells, and elevated expression of p53 protein in transformed cells and human tumor cell lines. The p53 cDNA was cloned by several groups, often from tumor cell lines with robust p53 protein expression, and expression of p53 cDNA could cooperate with other oncogenes to transform primary mouse cells and increase tumorigenic growth of established tumor cells. While multiple lines of evidence supported the widespread view that p53 was an oncogene, there were some observations that had been interpreted as suggesting a more complex story (; ; ; ; ). One explanation for these complexities was that p53 was actually a tumor suppressor gene rather than an oncogene. This interpretation was largely based on the fact that the normal p53 gene from mice inhibited cell growth. The investigators were appropriately cautious about this interpretation; we now know that many oncogenes, such as BCL2 and IDH1, when overexpressed, can inhibit growth (). Testing the two-hit hypothesis We began our studies of p53 in 1987, assuming that it was an oncogene, for the reasons described above. In particular, we initially did not think that it was the tumor suppressor gene target of the chromosome 17p delet