Scientists who scoured the DNA of brain tumors, searching gene by gene for bad actors, were puzzled. The cancers had none of the mutations in growth-causing genes that are typical of other tumors, yet they grew quickly, with no brakes. The question was why — what had altered their genetic instructions to lead to runaway cell division?
The surprising answer, researchers reported Wednesday in the journal Nature, is that the three-dimensional structure and organization of the DNA had been disrupted. As a result, two genetic neighborhoods that are normally separated, as if they are two gated communities, were merged. The effect was to allow a powerful snippet of DNA from one neighborhood into the one next door, where it woke up a near-dormant growth gene. And the cells took off.
This is an entirely new way for cells to become cancerous, researchers said, and most likely is not unique to the brain cancers, low- and intermediate-grade gliomas, that were the subject of the new study. The study’s lead author, Dr. Bradley E. Bernstein, a member of the Broad Institute in Cambridge, Mass., and a pathology professor at the Massachusetts General Hospital, says he has already found a similar phenomenon in about a dozen other tumors. And, he says, the discovery suggests a potential treatment with an existing chemotherapy drug that restores the walls separating the DNA sections.
“It’s really exciting,” said Dr. Peter Dirks, a brain cancer expert at the Hospital for Sick Children in Toronto, adding that at least with gliomas, the suggested treatments could be tested in clinical trials very soon.
The finding was surprising, others said.
“What this tells me is that I know a lot less than I did before,” said Dr. Jeremy Rich, a brain cancer expert at the Cleveland Clinic.
Gliomas are the most common type of malignant brain tumor in adults — about 18,000 are diagnosed each year in the United States. The most aggressive gliomas, known as glioblastomas — the type of cancer that killed Senator Edward M. Kennedy — usually strike older people and have an average survival time of 18 months. The lower-grade ones tend to strike younger adults, and survival times can be much longer. Yet despite extensive study of gliomas, doctors have made little progress in treating them. Doctors usually remove the lower-grade tumors and treat patients with radiation and chemotherapy. But the tumors return, often as aggressive glioblastomas that resist cancer therapies.
“We are desperate in this disease,” Dr. Dirks said.
The researchers did, however, notice one puzzling feature in up to 80 percent of low- and moderate-grade gliomas — a common gene that seemed to have no particular relevance to cancer often was mutated. The gene, isocitrate dehydrogenase, or IDH, had long been considered humdrum, a so-called housekeeping gene that directs cells to make an enzyme used in energy production.
“It was really surprising,” Dr. Bernstein said. “Why would a metabolism gene cause cancer?”
But then again, why was it mutated so often?
“That frequency is just shockingly high,” said Dr. Ingo K. Mellinghoff, a brain cancer expert at Memorial Sloan Kettering Cancer Center. “It makes you think it must be important.” And, he said, the mutation is one of the very first changes in the cancer cells.
There was one intriguing clue. When the IDH gene was mutated, the cancer cells’ DNA became studded with chemical tags known as methyl groups. Dr. Bernstein and his colleagues focused on those tags, asking how they might be disrupting the three-dimensional structure of the DNA.
It has long been known that human DNA is tightly packaged in cells. If all the DNA in a human cell were spread out in a line, it would be six and a half feet long. Instead, it forms about 10,000 loops, like those of a shoelace. Each loop is an autonomous gated neighborhood. The walls between the loops are made by a protein called CTCF. The extra methyl tags, Dr. Bernstein and his colleagues found, remove the CTCF walls. Now, suddenly, neighborhoods merge.
Two merged loops in particular seemed important. One loop contains a gene, PGDFRA, that makes cells grow and is rarely turned on. But when the loop with that gene merges with an adjacent loop, the gene comes under the control of a different DNA segment that turns the gene on constantly. The cells starts to divide, and a cancer is underway.
Dr. Bernstein and his colleagues tested their hypothesis by growing glioma cells in petri dishes and adding a rarely used first-generation chemotherapy drug, 5-Azacytidine, that dissolves methyl groups. The loops re-formed in the DNA; the gates to each neighborhood swung shut. And the growth gene turned off because now it was properly controlled by the DNA segment that kept it mostly quiet.
The research suggests that treating gliomas early, as soon as they are detected, with a drug like 5-Azacytidine can restore the DNA loops, Dr. Bernstein said. The same method should apply to the growing list of other cancers with excessive numbers of methyl tags on their DNA, he added. They include liver cancers, sarcomas, colon cancers, bladder cancers and leukemia, Dr. Bernstein said.
What is needed now is a diagnostic test to detect the presence of methyl tags and thus the breaking of loops. And that, Dr. Bernstein said, is simple. And it should lead to a clinical trial to treat people early on, before the tumors get uncontrollable.
“I am biased, obviously,” Dr. Bernstein said. But, he added, “I am really optimistic about the potential of this information.”