Chromosomal Instability and Tumors Promoted by DNA Hypomethylation Amir Eden, 1 Franc ¸ois Gaudet, 1,3 Alpana Waghmare, 1,2 Rudolf Jaenisch 1,2 * Human tumors often display changes in DNA methylation, including both genome-wide hypomethylation and site-specific hyper- methylation (1, 2). In mice, DNA hypo- methylation is sufficient to induce T cell lym- phomas with consistent gain of chromosome 15 (3), indicating that genome-wide hypomethy- lation plays a causal role in cancer. To explore further the link between DNA hypomethylation and chromosomal instability, we studied the effect of DNA hypomethylation on tumor-prone mice carrying mutations in both the Neurofibromatosis 1 (Nf1) and p53 tumor suppressor genes. The Nf1 and p53 genes are closely linked on mouse chromosome 11, and double heterozygotes carrying the mutations on the same chromosome (NPcis) develop soft tis- sue sarcomas, which show simultaneous loss of heterozygosity (LOH) of Nf1 and p53 (4 ). When we induced genomic hypomethylation in the Nf1 +/– p53 +/– (NPcis) mice by introducing a hypomorphic allele of DNA methyltransferase 1 (Dnmt1 Chip/– )(3), the mice developed sarcomas at an earlier age compared with NPcis littermates with normal levels of DNA methylation (Dnmt1 Chip/+ ) (Fig. 1A). To determine whether hypomethylation promotes the initial LOH re- quired for tumor development in the NPcis mice, we compared the rates of LOH in methylated and hypomethylated primary embryonic fibroblasts. We developed an assay to score for Nf1 –/– p53 –/– cells within a population of heterozygous cells (i.e., cells that have undergone LOH) (fig. S1). We then used this assay in a fluctuation analysis (5) to calculate the rate of LOH in hypomethyl- ated versus methylated cells. This analysis re- vealed a significant increase in LOH rate in hy- pomethylated cells (2.2-fold; P = 0.01) (Fig. 1B), consistent with the hypothesis that hypo- methylation promotes tumor development in NPcis mice by increasing the rate of LOH. To investigate the chromosomal mechanism leading to LOH in the hypomethylated cells and whether specific chromosomal regions are in- volved, we analyzed LOH along chromosome 11 in single colonies representing independent LOH events (Fig. 1C). In methylated (Dnmt1 Chip/+ ) cells, LOH affecting the whole chromosome (in- cluding at position 1.5 cM from the centromere) occurred in 45% of analyzed events. The remain- ing 55% showed heterozygosity at 1.5 cM but were homozygous for markers at 20 cM or 39 cM and were, therefore, the result of mitotic recombi- nation or of loss of the distal portion of the chro- mosome. Interestingly, the frequency of LOH events affecting the whole chromosome (including at 1.5 cM) was significantly higher in hypomethy- lated cells (77% compared with 45% in methylat- ed cells) (Fig. 1C), suggesting that the increase in LOH rate in hypomethylated cells is the result of a specific effect of hypomethylation on the stability of the centromeric or pericentric regions. A link between hypomethylation and the stability of whole chromosome arms is also found in the human Immunodeficiency–Centromeric Instabili- ty–Facial Anomalies (ICF) syndrome, in hypom- ethylation-induced T cell lymphomas in mice (3) and in human hepatocellular and prostate carcino- mas (6, 7 ). These results suggest that DNA hypo- methylation promotes cancer through effects on chromosomal stability. Further character- ization of the relations between DNA meth- ylation, chromatin composition, and chroma- tin structure may allow a better understanding how DNA hypomethylation affects chromo- some structure and integrity. References and Notes 1. A. P. Feinberg et al., Cancer Res. 48, 1149 (1998). 2. P. A. Jones, S. B. Baylin, Nature Rev. Genet. 3, 415 (2002). 3. F. Gaudet et al., Science 300, 455 (2003). 4. K. Cichowski et al., Science 286, 2172 (1999). 5. W. S. Kendal, P. Frost, Cancer Res. 48, 1060 (1988). 6. N. Wong et al., Am. J. Pathol. 159, 465 (2001). 7. W. A. Schulz et al., Genes Chromosomes Cancer 35, 58 (2002). 8. We thank T. Jacks and K. Reilly for NPcis mice and V. Carey for advice on biostatistics. Supported by an EMBO fellowship ALTF 43-1999 (A.E.), by Boehringer Ingelheim (A.E.), and by an NIH grant CA87869 (R.J.). Supporting Online Material www.sciencemag.org/cgi/content/full/300/5618/455/ DC1 Fig. S1 1 Whitehead Institute for Biomedical Research, 2 Depart- ment of Biology, Massachusetts Institute of Technology, Nine Cambridge Center, Cambridge, MA 02142, USA. 3 Department of Biology, Ludwig Maximilians University, Munich and Max Delbru ¨ck Center, Berlin, Germany. *To whom correspondence should be addressed. E- mail: jaenisch@wi.mit.edu Fig. 1. ( A) Survival curve of meth- ylated [ Dnmt Chip/+ (Chip/+), n = 42] and hypomethylated [Dnmt1 Chip/– (Chip/–), n = 44] Nf1 +/– p53 +/– cis (NPcis) mice. Survival curve for each genotype is plotted twice: One curve in- cludes all mice (open symbols), and the second curve represents only mice that developed soft tis- sue sarcomas (solid symbols Chip/ +), n = 33; Chip/–, n = 22). Mice of both genotypes typically devel- oped one predominant, fast- growing sarcoma. Eighteen Chip/– mice developed T cell lymphomas [as described in ( 3)], including five mice that developed both a sarco- ma and a lymphoma. ( B) In- creased rate of LOH events in hy- pomethylated mouse embryonic fibroblasts (MEFs), as determined by fluctuation analysis ( 5) (fig. S1). Graph shows the average rate in four independent experiments ( SD). The statistical significance was determined by Student’s t test to be P = 0.01. ( C) LOH anal- ysis. Cells in this study were de- rived from C57/B6  129/svJae F 1 mice. The polymorphic markers indicated were used to examine heterozygosity along mouse chromosome 11 in cell colonies homozygous for Nf1 and p53 derived during fluctuation analysis. Only colonies from independent parallel cultures were considered independent events. We analyzed colonies obtained from three different fluctuation analyses ( n = number of independent colonies). Nf1 and p53 mutant alleles (marked with “x”) resided on the B6 chromosome, and LOH always resulted in the loss of the 129 allele. Results are shown as the percentage of colonies with only the B6 allele at the marked position. BREVIA www.sciencemag.org SCIENCE VOL 300 18 APRIL 2003 455 View publication stats View publication stats