Chemotherapeutic agents that are toxic through hypomethylation of DNA can cause cancers to develop in other tissues, according to an article in the April 18th issue of Science. The two studies conducted with mice were designed to test whether changes in gene expression caused by changes in methylation of DNA could be carcinogenic in the absence of genetic mutation.

According to the authors, the new findings are important because some chemotherapeutic drugs in clinical trial are cytotoxic through hypomethylation. Previous studies with mice had shown that drug-induced hypomethylation is protective against cancer in the intestinal tract. In the current study, the researchers found that hypomethylation can cause cancer elsewhere in the body.

"You have to know how you interfere with these cancer mechanisms," said Rudolf Jaenisch, Ph.D., the leader of the research team. "At the moment we have two totally opposite results when we look in two different tissues. In the intestines, hypomethylation protects against cancer, and in the thymus it enhances cancer."

Recent research into the cellular mechanisms underlying carcinogenesis has focused on changes in gene expression as well as gene sequence. Methylation of DNA, which can activate or terminate gene expression, is one process that has been under study. In the current study, investigators studied mice that were genetically engineered to show hypomethylation in their body cells.

"It was a shot in the dark to see whether we could generate animals that could actually survive and at the same time be hypomethylated," said Francois Gaudet, Ph.D., who developed the experimental mouse strain and was lead author of the article.

Gaudet's mice expressed only about 10 percent of normal levels of methyltransferase, the enzyme that maintains normal levels of methylation. About 80 percent of the mice developed an aggressive form of T-cell leukemia. Furthermore, all of the cancer cells were the same, suggesting that the disease originated from a single cell. The gene whose expression was modified (c-myc, which has been associated with tumor formation) was found to be overly active in the leukemic cells.

Gaudet said. "The induction of hypomethylation might treat cancer in some tissues but may induce cancer in other tissues. It is important therefore to test the effect of hypomethylation on cancer in different tissues."

In an extension of Gaudet's work, Amir Eden, Ph.D., found that mice engineered to develop tumors did so at an earlier age when hypomethylation was induced. Eden's mice had mutations in tumor suppressor genes; when hypomethylation was induced, they developed soft-tissue sarcomas earlier than mice with a similar mutation but normal methylation levels. In these mice, Eden could show that hypomethylation-induced chromosomal instability caused an increase in tumor formation.

The data offer "real food for thought about the role of losing methylation in cancer," noted Professor Stephen Baylin, who was not associated with the research. However, he cautioned that the findings in mice may not be valid in humans. For instance, the bioengineered mice had more reduction in methylation than that seen in most human tumors. In addition, the engineered mice demonstrated hypomethylation from embryonic development onward, unlike the situation in tumors that develop in adult people.

Jaenisch had similar thoughts. "It is very important to assess whether findings made in animal models apply to patients."

The investigators plan to extend the work by looking at other tumors, such as those of the pancreas, breast, and liver, changing methylation levels and observing the results. "I think it will reveal metabolic differences between tissues," Jaenisch said, "and may lead to a deeper understanding why reducing methylation levels seems to harm some cancers and benefit others."