Subscribe to eTOC

In Unraveling a Novel Mechanism Behind Cancer Cells, Researchers Report a Potential New Approach to Blocking Gliomas



In IDH1 wild-type cells (top), a functional insulator (red) defines a topological boundary between the oncogene PDGFRA and a nearby constitutive enhancer (green), which usually drives expression of the essential housekeeping gene FIP1L1. Following a mutation in the IDH1 gene (bottom), glioma cells gain methylation marks on the DNA at the insulator sequence, which inactivates the insulator (dashed box). This allows for a novel tumor-specific interaction between the enhancer and PDGFRA, activating the oncogene and driving tumor growth.

A research team has uncovered a novel mechanism for the growth of cancer cells, which could present a potential new target for preventing gliomas.

Scientists at Massachusetts General Hospital and the Broad Institute have identified a novel biological mechanism that could explain what drives the expansion of some cancer cells, in particular brain tumors. The finding, reported in the January 7 issue of Nature, could provide novel treatments that block this cascade and prevent the formation of gliomas.

Bradley Bernstein, MD, PhD, a member of the Broad Institute and a professor of pathology at Massachusetts General Hospital, and his colleagues discovered that a mutation in isocitrate dehydrogenase-1 (IDH1) triggers a series of events that disrupt the very complex organization of the genome, changing its structure and compromising the integrity of the intricate boundaries that organize genes into specific neighborhoods in the 10,000 or so loops that fold into DNA. These loops normally insulate the genetic information within these folds.


DR. BRADLEY BERNSTEIN (LEFT) AND DR. WILLIAM FLAVAHAN were part of the team that identified a novel mechanism for causing the growth of gliomas.

The researchers observed, however, that the DNA's folding mechanism was failing as a result of the IDH mutations in the tumor cells. The neighborhoods seemed to have lost their boundaries, or insulation, and the result is that a known growth factor and oncogene called PDGFRA is being regulated by an on-switch from another neighborhood.

This finding builds on an important discovery reported in 2008 in Science by a team of scientists at Johns Hopkins University. The investigators identified a novel glioma-associated mutation in IDH1 in an exome-sequencing study of glioblastoma (GBM).

IDH1/2 mutations are found in 5- to 10-percent of GBM and in more than 75 percent of grade II and III gliomas (the tumors that eventually develop into GBM). The IDH gene, which is involved in cellular metabolism, had never been linked to cancer. It was unclear how these mutations were contributing to brain tumor development. A number of laboratories set out to solve this puzzle.

The IDH1 gene mutations are somatic (tumor-specific), occur only in cells that become cancerous, and are not inherited. The mutation causes a single amino acid to change in the IDH1 enzyme; and the mutant form no longer does what it had been intended to do. Instead, it converts IDH to 2-hydroxyglutarate. Normally, there is virtually no 2-hydroxyglutarate in the brain.

The hypothesis suggested by the recent Nature study is that this metabolite interferes with the ability of the cell to put the brakes on DNA methylation, which in turn damages the insulation of DNA's discrete folds, changes the DNA landscape and regulation of other genes important for cell proliferation, and sets the stage for tumor growth.


Dr. Bernstein, post-doctoral fellow William Flavahan, PhD, and their colleagues at Mass General conducted their study using glioma specimens from patients and on patient-derived tumor cells in assays. They also studied brain tissue and data from cancer genome projects, including the Cancer Genome Atlas, to determine why IDH mutations are commonly seen in gliomas.

They knew that production of 2-hydroxyglutarate blocks the removal of methyl groups that are attached to DNA. These methyl groups are the on-off switches on the cellular brakes. Dr. Bernstein and his colleagues hypothesized that this over-abundance of methyl groups might contribute to cancer by disabling a critical regulatory protein, called CTCF.

They studied expression data from normal brain tissue samples, as well as 230 IDH mutant and 56 wild-type lower-grade gliomas, and they were able to show that there was far more crosstalk between genes from neighborhoods outside of their insulated zones in the IDH-mutated gliomas. They reported that the disrupted boundaries exhibited higher DNA methylation and lower CTCF binding in the IDH mutant compared with wild-type tumors. CTCF binding holds the structure of the loops together.

The new metabolite, 2-hydroxyglutarate, was blocking removal of the methyl tags, leading to an overload of the methyl groups.

PDGFRA, an oncogene, was switched on and began taking orders from FIP1L1, a gene that processes messenger RNAs needed to make proteins. The “always on” regulatory sequence that normally controls FIP1L1 was now controlling PDGFRA in IDH tumors.

Dr. Bernstein said that they are “not sure how the FIP1L1 gene fits into the cancer-causing mechanism. Maybe it's just the ‘donor’ of the aberrant regulatory sequence that turns on PDGFRA when the insulator is disrupted,” he said.

These steps form a “powerful cancer-promoting combination,” Dr. Bernstein said. This is the first evidence that an oncogene can be turned on by changes in how the genome folds, he added.

These changes were not seen in gliomas without IDH mutations, Dr. Bernstein said. He believes that this series of events may be a general cancer-causing mechanism.

“This is a totally new mechanism for causing cancer,” he said, “and we think it will hold true not just in brain tumors, but in other forms of cancer.”

The scientists then set out to prove the mechanism by delivering a first-generation chemotherapy drug (5-azacytidine) that blocks DNA methylation to the IDH-mutant laboratory models. They were hoping that it would prevent the toxic cascade that leads to the proliferation of cancer cells. And it worked.

According to Dr. Bernstein, the medicine partially restored insulation function and downregulated PDGFRA. The loops in the DNA were reformed and the methyl tags were diminished. The oncogene was no longer activated.

The scientists also used CRISPR-mediated editing to disrupt the insulator sequence directly (called the CTCF motif) and found that this also upregulated PDGFRA and increased cancer cell proliferation.

Neuro-oncologists are excited about this work, Dr. Bernstein said, but there is more work to do. “We need to get more clinical samples from patients and study them to test whether medicines that block DNA methylation would work as a treatment and be safe.”


Donald W. Parsons, MD, PhD, co-director of the Neuro-Oncology Program and the Cancer Genetics and Genomics Program at Baylor College of Medicine, who was part of the Johns Hopkins team that reported the hotspot mutations in IDH1 and IDH2 in 2008, calls the new finding “very provocative.”

“The recent results from Dr. Bernstein and his colleagues have served to shed some light on one intriguing potential mechanism by which IDH mutations might link to this abnormal methylation (previously demonstrated by other investigators) and what this hypermethylation might be doing,” he said. “This disruption of methylation leads to an altered topography, or specific organization of the DNA, potentially causing a loss of insulation and resulting in the abnormal activity of PDGFRA, a known cancer gene, and potentially other cancer genes.”

“We knew that IDH mutations were important in gliomas, but it was difficult to understand how it contributes to the growth of the tumor,” said Patrick Y. Wen, MD, FAAN, director of the Center for Neuro-Oncology at the Dana Farber/Brigham and Women's Cancer Center and a professor of neurology at Harvard Medical School. “This is a partial explanation of how it leads to tumor growth and provides possible new ways to treat cancer. It is likely that there may be other as yet unknown mechanisms, given the complexity of epigenetics and how little we know about this field.”

“There are a lot of hypotheses about the role of IDH in oncogenes and tumor development,” said Linda M. Liau, MD, PhD, a professor and vice chair of neurosurgery at University of California, Los Angeles (UCLA) and director of the UCLA Brain Tumor Program.

“This new finding shows that hypermethylation alters the three-dimensional structure of DNA and activates an oncogene. It's a very interesting and novel mechanism.

“Further studies will be needed to see how this hypothesis plays out in tumor models and in the clinical setting,” she said. “It has been challenging to maintain the IDH mutation in vivo because the mutations tend to occur in slow growing lower-grade gliomas. There are no good animal models of human low-grade gliomas with IDH mutations yet, but hopefully this finding will lead us to a better understanding of how to develop such models.

“By the time a patient develops a high-grade glioblastoma, this process of gliomagenesis and malignant transformation has already occurred,” she said. “How do you reverse that? Will we be able to prevent a low-grade glioma from becoming high-grade? This finding should certainly lead to further studies to test whether blocking hypermethylation will reverse this progression. It would be great to translate this to patients someday.”


•. Flavahan WA, Dier Y, Liau BB, et al. Insulator dysfunction and oncogene activation in IDH mutant gliomas Nature 2016; 529(7584):110–114; Epub 2015 Dec. 23.
    •. Parsons DW, Jones S, Zhang X, et al. An integrated genomic analysis of human glioblastoma multiforme Science 2008; 321(5897): 1807–1812.
    •. Yan H, Parsons DW, Jin G, et al. IDH1 and IDH2 mutations in gliomas N Engl J Med 2009;360(8):765–773.