Although most colorectal cancers are chromosomally unstable, the basis for this instability has not been defined. To determine whether genes shown to cause chromosomal instability in model systems were mutated in colorectal cancers, we identified their human homologues and determined their sequence in a panel of colorectal cancers. We found 19 somatic mutations in five genes representing three distinct instability pathways. Seven mutations were found in MRE11, whose product is involved in double-strand break repair. Four mutations were found among hZw10, hZwilch/FLJ10036, and hRod/KNTC, whose products bind to one another in a complex that localizes to kinetochores and controls chromosome segregation. Eight mutations were found in Ding, a previously uncharacterized gene with sequence similarity to the Saccharomyces cerevisiae Pds1, whose product is essential for proper chromosome disjunction. This analysis buttresses the evidence that chromosomal instability has a genetic basis and provides clues to the mechanistic basis of instability in cancers.
A very large fraction of cancers consists of cells with an abnormal chromosomal content, called aneuploidy (1) . Aneuploidy is often associated with chromosomal instability (CIN), a condition in which cancer cells gain and lose whole chromosomes or large parts thereof at elevated rates compared with normal cells (2) . The molecular basis of CIN has remained mysterious. Many mechanisms have been postulated to be responsible for CIN (3 , 4) . Like other phenotypes characteristic of cancer, it is possible that mutations in genes that control chromosome stability are responsible for CIN. However, only a small number of human cancers with mutations in genes known to cause experimental forms of CIN have been identified. These genes include hBUB1, ATM, ATR, BRCA1, and BRCA2, each of which is very infrequently mutated in nonfamilial cancers (5, 6, 7, 8) . Increased copy numbers of aurora2/STK15 and PLK1 have been reported to occur in a higher fraction of cancers (9, 10, 11, 12) , but no definitive mutations of these genes have been identified, and their contribution to CIN remains conjectural. 5
In contrast, a large number of genes have been identified that trigger CIN when mutated in Saccharomyces cerevisiae (13 , 14) . These genes are involved in a variety of cellular pathways including chromosome condensation, sister-chromatid cohesion, kinetochore structure and function, microtubule formation, and cell cycle control. Similarly, several genes can cause CIN in Drosophila melanogaster when genetically altered (15 , 16) . Based on these observations, we wondered whether the previous failures to detect mutations in potential CIN genes in human cancers were simply due to the fact that there are a large number of such genes, only a small number of which have been analyzed.
Materials and Methods
Yeast genes that can cause an instability phenotype were identified in the Saccharomyces cerevisiae genome. 6 The corresponding protein sequences were used to search for human homologues in the Celera draft human genome sequence. In addition, several genes were selected by close homology to identified human genes and Celera hCTs or by membership to the same Panther protein family. All exons and adjacent intronic sequence of these genes were extracted from the Celera draft human genome sequence.
PCR and Sequencing.
Primers for PCR amplification and sequencing were designed using the Primer 3 program 7 and were synthesized by MWG (High Point, NC) or IDT (Coralville, IA). PCR amplification and sequencing were performed on tumor DNA from 24 early-passage cell lines as described previously (17) using a 384 capillary automated sequencing apparatus (Spectrumedix, State College, PA). Of the 1351 exons extracted, 1282 were successfully analyzed in an average of 23 tumor samples. Sequences of all primers used for PCR amplification and sequencing are available in Supplementary Table 1 ⇓ . For the five genes identified in the initial screen, coding exons were analyzed in tumor DNA from an additional 168 early-passage aneuploid colorectal cancer cell lines passaged in vitro or as xenografts in nude mice
Analysis of Mutations.
Sequence traces were assembled and analyzed to identify potential genomic alterations using the Mutation Explorer software package (SoftGenetics, State College, PA).
Results and Discussion
We compiled a list of more than 1000 genes by computational identification of human homologues of “instability” genes of yeast and D. melanogaster. From this list, 100 candidate genes were selected based on the strength of the phenotypes observed in yeast or D. melanogaster and the extent of similarity to the human homologue (Table 1 ⇓ ). The complete sequence of these genes was then determined in a panel of colorectal cancers.
Public and private genomic databases were used to extract the 1351 exons that encode the 100 candidate genes, and 5022 primers were designed for PCR amplification and sequencing (Supplementary Table 1 ⇓ ). Using these primers, each exon was then individually amplified and sequenced from DNA of 24 colorectal cancers. This analysis revealed the presence of 373 variations not present in current human genomic databases. To find out whether these variations were somatic (i.e., tumor specific), we determined whether any of them were present in DNA from matching normal tissues of the patients in whom the mutations were originally detected. We thereby discovered somatic mutations in five genes. All five genes were then analyzed for mutations in a larger panel of tumors including 168 additional colorectal cancers. Altogether, more than 10 Mb of DNA was sequenced, allowing us to identify 19 somatic mutations distributed among three classes of genes (Table 2 ⇓ ; Fig. 1 ⇓ ).
Eight somatic mutations in seven different CIN cancers were found in the MRE11 gene (Table 2) ⇓ . Four mutations generated premature translational stop signals within exons 7, 8, or 13 (Fig. 1) ⇓ , whereas the others were missense mutations; all but one of these were heterozygous. MRE11 is known to be involved in DNA double-strand break repair and participates in exonuclease and endonuclease activities (18) . Hypomorphic mutations in MRE11 have been described in two families with ataxiatelangiectasia-like disorder (19) . The cellular features resulting from these mutations are similar to those seen in ataxiatelangiectasia as well as in patients with the Nijmegen breakage syndrome and include hypersensitivity to ionizing radiation, radioresistant DNA synthesis, and abrogation of ATM-dependent events. The mutations listed in Table 1 ⇓ represent the first report of MRE11 mutations in human cancers. MRE11 is known to form complexes with the hRad50 and NBS1 proteins, whose role in genomic stability is well documented. Recently published homozygous knockouts of MRE11 in vertebrate cells showed that this gene is essential for cell proliferation and the maintenance of a normal chromosomal content (20 , 21) .
hZw10, hZwilch, and hRod.
We found four somatic mutations in the hZw10, hZwilch/FLJ10036, and hRod/KNTC1 genes (Table 2) ⇓ . The mutation in hRod was a homozygous missense change (Fig. 1) ⇓ , whereas the heterozygous mutation in Zwilch affected a splice acceptor site predicted to result in a premature truncation of the Zwilch RNA in its fifth exon. The other two mutations were heterozygous missense changes in hZw10 (Fig. 1) ⇓ . These three genes act in the same pathway and cause a severe CIN phenotype when mutated in D. melanogaster. Both Zw10 and Rod (rough deal) mutations disrupt the accuracy of chromosome segregation in D. melanogaster and promote aberrant anaphases that lead to aneuploidy (22, 23, 24) . Rod, Zwilch, and Zw10 proteins each localize to the kinetochore in an identical manner and function in the spindle checkpoint (25 , 26) . Recent experiments show that the human homologues of these three genes (hRod, hZwilch, and hZw10) are physically associated and function together in a large, evolutionarily conserved complex (26 , 27) . Interestingly, hRod and hZw10 do not have obvious homologues in budding yeast.
The Ding gene was found to be somatically mutated in eight CIN cancers. These tumors did not overlap with those containing mutations in MRE11, hRod, hZw10, or hZwilch. One mutation affected the splice donor site two bases downstream of the last codon in exon 7. Another mutation resulted in a stop codon within exon 14 and the other six mutations were missense changes (Fig. 1 ⇓ ; Table 1 ⇓ ). Ding is a previously uncharacterized gene (KIAA0853) that was selected for sequence analysis because its COOH terminus is homologous (20% identity, 46% similarity) with the yeast protein Pds1. Pds1 is an anaphase inhibitor in budding yeast and plays a critical role in the control of anaphase (28 , 29) . Its human ortholog, hSecurin is essential for the proper function and processing of the separin protease, for separin-dependent cleavage of the cohesion subunit Scc1, and for maintaining chromosome stability (30) .
These data provide novel evidence to support the hypothesis that CIN has a genetic basis. However, analysis of mutations in tumors is complicated by the fact that mutations can arise either as functional alterations driving neoplasia or as nonfunctional “passenger” changes. Two independent lines of evidence suggested that the alterations we observed were functional rather than accidental. First, the distribution of mutations was strikingly nonrandom: seven mutations in MRE11; four in the hRod/hZw10/hZwilch cluster; eight in Ding; and none in 95 other genes. The prevalence of mutations in the coding regions of the five mutated genes was significantly higher than the prevalence of nonfunctional alterations found in the colorectal cancer genome (17) .
It has been unclear whether mutations in one or a few “master” CIN genes is responsible for most CIN cancers or whether CIN is a more “democratic” process, with mutations occurring in dozens or hundreds of different genes, each accounting for only a small portion of CIN cancers. Our sequencing analysis identified somatic mutations in cancer CIN genes that together account for ∼10% of CIN cancers and supports the democratic model. Our data also substantiate previous ideas about the mechanisms that may contribute to CIN in human cancers (3) . In particular, they suggest that defects in proteins that are involved in double-strand break repair, kinetochore function, and chromatid segregation are likely to contribute to aneuploidy. Future studies to identify mutations in other genes that function in these three functional pathways, in both colorectal and other human cancers, should be informative.
Grant support: The Virginia and D.K. Ludwig Fund for Cancer Research, The Benjamin Baker Scholarship Fund, The Clayton Fund, and NIH Grants CA 43460, CA 57345, and CA 62924.
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.
Note: Z. Wang, J. Cummins, and D. Shen contributed equally to this work. D. Cahill is currently at Massachusetts General Hospital, Department of Surgery, Boston, MA; P. Jallepalli is currently at Molecular Biology Program, Memorial Sloan-Kettering Cancer Center, New York, NY; G. Traverso is currently at Trinity College, Cambridge, United Kingdom; and M. Awad is currently at Department of Pediatrics, Johns Hopkins University, Baltimore, MD. Supplementary data for this article can be found at Cancer Research Online (http://cancerres.aacrjournals.org).
Requests for reprints: Christoph Lengauer, Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, CRB, Room 585, 1650 Orleans Street, Baltimore, MD 21231. Phone: (410) 955-8878; Fax: (410) 955-0548; E-mail:
↵5 While this paper was under review, it was shown that hCDC4 is frequently mutated in aneuploid colorectal cancers, and that its inactivation causes CIN. (Rajagopalan et al. Nature 2004;428:77–81.)
- Received February 18, 2004.
- Accepted February 27, 2004.
- ©2004 American Association for Cancer Research.