[CANCER RESEARCH54, 5059-5063, October 1, 19941 Perspectives in Cancer Research Microsatellite Instability: Marker of a Mutator Phenotype in Cancer' Lawrence A. Loeb2 The JosephGoustein Memorial Cancer ResearchLaboratory. Department ofPa:hology SM.30, University of Washington.SchoolofMedicine. Seattle. Washington,98195 greater than in normal cells; i.e., cancer cells must exhibit or have exhibited a mutator phenotype (1, 8, 12). The concept that oncogenesis involves a mutator phenotype has been considered by several investigators (1, 8, 12). An alternative hypothesis to account for the multiple mutations in tumors can be formulated based on mutation-driven clonal repopulation. Let us assume that the first mutation leads to a strong proliferative advan tage. If the mutant cell expands to 10'°progeny within the tumor, then a second mutation occurring at a normal frequency of 10@10 would occur with a high probability in cells already harboring the first mutation. Among the early mutations would have to be ones that confer immortality or at least extend cellular life span. If this and each subsequent mutation results in a similar clonal proliferation, multiple mutations could accumulate in each tumor cell in the absence of an increase in the mutation rate. However, this hypothesis is less attrac tive than that of a mutator phenotype, because it requires that each mutation, even a single mutation in one of two recessive alleles, causes a profound growth advantage. Despite this limitation, the fact that tumors are predominantly clonal in origin indicates that many mutations result in cellular proliferation. Moreover, clonal evolution and a mutator phenotype are not exclusive; both may be required to account for the frequency of mutations in cancers. Multiple Mutations in Cancers Until recently, the hypothesis that cancer cells exhibit a mutator phenotype received little experimental support (1). Most comparisons between normal and cancer cells used rodent cells in which there is less DNA repair than in human cells, or between nontumorigenic and tumorigenic cell lines, both of which are immortal and already likely to contain multiple mutations. Experiments on the fidelity of DNA synthesis using the SV4O replication complex did not reveal differ ences between normal cells and malignant cells (13) and thus failed to support the hypothesis of a mutator phenotype. However, experiments on the fidelity of DNA synthesis are insensitive; they can only detect errors in DNA synthesis which occur more frequently than l0_6, which is four orders of magnitude greater than the background mu tation rates exhibited by somatic cells. In contrast to these negative studies, the recent demonstration of microsatellite instability in dif ferent human tumors has provided strong evidence for a mutator phenotype (14—16).Each microsatellite contains multiple repetitive nucleotide sequences that are relatively constant in normal cells but vary in length in certain tumors. Even though microsatellite variations might not affect the phenotype of the cell, they are by definition mutations. Microsatellite instability in some of these cancers (17) occurs coordinately with mutations in genes that are homologous to those involved in mismatch repair in bacteria (18) and yeast (19). The implication is that mutations in these mismatch repair genes decrease the capacity of that system to correct errors made during DNA replication, particularly errors in repetitive nucleotide sequences. Mu tations in the mismatch repair genes could be an early event in the carcinogenic process, since these mutations have been demonstrated in diploid colon cancer cells (16). Microsatellite repeats are likely to be hot spots for mutagenesis (20), and mutations within these se Introduction There is increasing evidence that multiple mutations are present in many human tumors. Molecular techniques are progressively making it feasible to dissect the eukaryotic genome, from chromosomes, down to genes and nucleotide sequences, and eventually even to three dimensional structures. With each deeper level of exploration, more and more mutations are being documented in cancer cells. The emerg ing concept is that genomes of cancer cells are unstable, and this instability results in a cascade of mutations some of which enable cancer cells to bypass the host regulatory processes. Cancer as a Mutator Phenotype Based on the high frequencies of chromosomal abnormalities and mutations in human cancers, I offered the hypothesis that cancer is manifested by a mutator phenotype (1). This hypothesis was based on the argument that the spontaneous mutation rate in normal cells is insufficient to account for the high frequency of mutations in human cancer cells (1, 2). Moreover, the frequency of mutations in cancer cells may be even higher than that which is detected. Current methods are biased towards detecting large rearrangements; deletions, frame shifts, base substitutions, and small rearrangements are likely to be missed. In fact, we lack methods to efficiently identify random single base substitutions in the nucleotide sequence of genes. The prediction is that more and more mutations will be found in cancer cells as we clone and sequence genes from tumors and then compare their nude otide sequences with those in adjacent normal tissues. The multiple chromosomal changes currently detected in cancer cells may be the tip of the iceberg. There may be a much larger number of single nude otide substitutions, frameshifts, small deletions, and insertions. A feature of cancer may be the continual accumulation of mutations within individual cells. In this Perspective, I will focus on point mutations and mutations occurring in repetitive nucleotide sequences. Is the rate of spontaneous or background mutations in normal cells sufficient to account for the many mutations in human cancers? A compendium of studies of somatic mutation rates of human cells in culture using hypoxanthine-guanine phosphoribosyltransferase (3) and adenine phosphoribosyltransferase (4) suggests a background mutation rate of approximately 1.4 X 10 ‘° mutations/base pair/cell generation (1). This rate is similar to that deduced by Chu et a!. (5), based on the presence of electrophoretically distinct protein variants at unselected loci in cultured human lymphoblastoid cells. Considering that cancers arise in one or a few cells, we have estimated that the background mutation rate in normal cells can account for only two or three mutations in each tumor and not the much larger number of mutations that are reported (8—1), or the even greater numbers that are likely to be found as methods for detection become more sensitive. At some time during the life of a tumor, the mutation rate must be Received 6/21/94; accepted 8/4/94. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. I This work was funded by Grant OIG CA 39903 from the National Cancer Institute. 2 To whom requests for reprints should be addressed. 5059 Research. on December 9, 2021. © 1994 American Association for Cancer cancerres.aacrjournals.org Downloaded from