LONDON - Some cancers may result not from genetic mutations that put a gene out of action, but from a biochemical - and possibly reversible - block on the expression of the gene in question.

A new study suggests some colorectal cancers occur when methyl groups cluster on the genes that normally help to make a protein that repairs the genetic code. The effect is to deprive the cell of its toolbox for fixing errant DNA.

Sir Walter Bodmer, head of the Imperial Cancer Research Fund's Cancer and Immunogenetics Laboratory at the Institute of Molecular Medicine in Oxford, told BioWorld International: "This is part of a sequence of discoveries that will ultimately be very important in helping us to understand the cancer process. It has become clear that we need to understand which genes are switched off and which are not, and the nature of the switching process. It is particularly important because if one could reverse that process, this could provide the starting point for developing new therapies for some cancers."

The finding gives added impetus to work being carried out by a variety of groups to discover whether other genes in other common cancers, such as breast cancer, are not mutated but methylated.

Bodmer and his colleagues reported their work in the Aug. 31 Proceedings of the National Academy of Sciences in a paper titled: "Mechanisms of inactivation of mismatch repair genes in human colorectal cancer cell lines: The predominant role of hMLH1."

The starting point for the study by Bodmer and colleagues was the observation that hereditary nonpolyposis colorectal cancer (HNPCC) is caused by germline mutations in the DNA mismatch repair genes. There are several of these genes, and their products help to put right errors made by DNA polymerase when replicating DNA. Most families affected by HNPCC have mutations in the mismatch repair genes called hMLH1 and hMSH2. These tumors are known as "replication error defective."

Later work showed that 10 percent to 15 percent of sporadic colorectal cancers have the same kinds of genetic defect in the tumor tissue.

Bodmer's study aimed to find out what else was happening in the tumors of those people affected by sporadic colorectal cancer. Bodmer said, "Other researchers had shown that a key factor affecting the expression of a gene is whether or not the promoters controlling its expression are methylated. If they are, then this is a sign that the gene has been inactivated. We needed to know to what extent methylation could explain a change in gene expression in a cancer, because this could fill a considerable gap in the catalogue of changes that are involved in producing a cancer."

The researchers therefore decided to examine a group of 49 sporadic colorectal cancer cell lines. After determining which were replication error defective, they asked how many could be explained by mutations in the hMLH1 gene, how many by mutations in the hMSH2 gene, and how many by methylation.

"What we found," Bodmer said, "extending the work of others, was that if we took into account methylation and the known mutations, we could explain nearly every one. Some of the changes were mutations, some were methylations, and some were a combination of methylation and mutation. So you end up by knocking out the function of a gene in different ways, by methylation, by mutation or by a combination of the two."

In a few cases, they found methylation of the DNA in cell lines that were not replication error defective. Bodmer suggested this result may have arisen because each of the homologues of the relevant genes can become methylated independently. "We believe that where you see some methylation but not yet a replication error defect, then the first step has been taken in the process of giving a selective advantage to the cancer," he said.

No one knows what primary changes occur that make it possible to methylate a gene and thus switch it off. "A high proportion of genes have groups, known as CPG islands, which when they are methylated will prevent attachment of the proteins that initiate transcription," Bodmer said. "This is obviously part of the normal process of differentiation between cells, which somehow gets deranged in cancers. It's probably something to do with acetylation of histones. It will become important to understand what's going on in order to understand what's going on in the tumor."

Future goals for Bodmer and his team include understanding the nature of the methylation process better, and how it is controlled; and looking at the pattern of which genes are methylated and which are not in common human cancers.

"We want to look for methylation in situations where it has not been seen before," he said. "Ultimately our goal is to understand the way in which these changes give an advantage to a cancer, which we believe is by interfering with the pathways that lead to apoptosis."