Researchers Close in on an Epigenetic Mechanism for Carcinogenesis

Epigenetic lesions, and not just genetic mutations, can cause brain cancer, a keynote speaker reported at the 21^st Annual Scientific Meeting of the Society for Neuro-Oncology (SNO), held November 17–20 in Scottsdale, Arizona.

“We think of cancer as a genetic disease, with mutations turning genes on or off-but that’s only half of the story,” said Bradley Bernstein, MD, PhD, of the Massachusetts General Hospital Cancer Center and Harvard Medical School in Boston. “A critical goal for our field is to understand in detail how epigenetic lesions drive cancer.”

Epigenetic lesions involving DNA modifiers or altered genome folding can silence or activate genes, even in the absence of gene sequence mutations. One important implication of that observation is that precision-medicine genetic and genomic tools will miss epigenetic lesions that may drive tumor growth or drug resistance.

According to the presentation, the human genome has some 6 billion base-pairs in the diploid human genome. That includes 20,000 protein-coding genes-but they account for only about 1% of the human genome. “The other 99% is very poorly understood at this time,” Dr. Bernstein said.

That 99% is interspersed with a million or more regulatory elements, which are crucial for performing tissue- and cell-specific tasks because every non-malignant cell contains the same DNA sequences. “Cells need ways to switch these genes on and off in very specific ways under different circumstances,” Dr. Bernstein said, likening their operation to the dial of a rheostat.

Chromatin folding is one way cells manage their transcriptional states. Active chromatin allows expression of genes while repressive chromatin arrangements turn loci off. Another is partitioning of the genome into isolated loops or islands that put regulatory loci in proximity to appropriate target protein-encoding genes, while separating them from inappropriate targets.

“There’s some kind of homeostatic balance that allows switching from active to inactive states under the right conditions,” he explained. As cells differentiate and commit to a phenotype, epigenetic resistance increases, to avoid expression of inappropriate genes for that cell type.

In cancer, this homeostatic chromatin balance might tip toward being too restrictive (silencing tumor suppressors, for example), or toward being too permissive, allowing gene expression at inappropriate times. Such “spurious activity” can also facilitate tumorigenesis, Dr. Bernstein believes. “I think this has a lot to do with brain cancer.”

Like gene mutations, epigenetic lesions that disrupt chromatin regulation of gene expression probably occur in both driver and passenger forms-many lesions won’t contribute to tumor formation while a critical few will promote carcinogenesis.

Epigenetic lesions can result from gene mutations, but environmental exposures can also cause epigenetic aberrations in the absence of changes to DNA sequences, Dr. Bernstein emphasized-implying that non-genotoxic environmental exposures might nevertheless be carcinogenic under the right circumstances.

“IDH-mutant gliomas exhibit DNA hypermethylation,” Dr. Bernstein said, illustrating one example of how this can work. The IDH metabolic enzyme-mutated in sarcomas, glioma, and leukemia-generates 2-HG, which inhibits demethylases. “So we know that a universal feature of IDH glioblastomas is hypermethylation.”

It’s long been believed that hypermethylation silences tumor suppressor genes, but surprisingly few examples of that happening actually been identified. “They don’t explain the entirety,” he said. “So we started thinking about alternative explanations.”

CTCF binding is disrupted in IDH1-mutant gliomas, and it’s known that CTCF binding is blocked by epigenetic DNA methylation. Mapping the epigenetics of CTCF binding sites in IDH-mutant and wild-type cells, the team found more than 600 CTCF sites lose their CTCF binding in IDH-mutant cells. “There was nice binding in the wild-type that was lost in the mutants,” he said.

That pivotal observation might just point the way to understanding one epigenetic mechanism of tumorigenesis. “When the CTCF dimer binds to the CTCF site, a chromatin loop from 100 kilobases to half a megabase or larger is separated out from the rest of the genome, and becomes functionally partitioned,” Dr. Bernstein explained. “It can activate genes that are in the same loop, but cannot affect genes in other loops.”

In IDH-mutant gliomas, many of those partitioning insulators are compromised. “We can see a signature suggestive that you’re losing insulation, genome-wide,” he said. “All of a sudden a gene in one loop begins correlating with genes in neighboring loops, suggesting genome-wide disruption of gene insulation.”

One illustrative case involves the receptor tyrosine kinase–encoding platelet-derived growth factor receptor alpha (PDGFRA) gene.

CTCF insulators are situated between an enhancer gene and PDGFRA. In IDH–wild-type tumors, there is no interaction between the enhancer and PDGFRA on different chromatin loops, whereas in IDH-mutant gliomas, insulator sites are methylated, and CTCF binding at that site is lost, Dr. Bernstein noted. “In IDH–mutant cells, all of a sudden, we see very strong interaction between the enhancer and the neighboring-loop PDGFRA. PDGFRA is not amplified-its copy number is normal. But it’s strongly upregulated.”

As a proof-of-principle experiment, Dr. Bernstein’s team used CRISPR-Cas9 to knock out CTCF insulators situated between an enhancer gene and the receptor. “PDGFRA expression was doubled,” he reported. “And all we’ve done is remove a couple of bases from the insulator site.”

Cells began to grow twice as fast. “I think we have a new cancer-causing mechanism,” Dr. Bernstein concluded. “An epigenetic mechanism for oncogene activation. You simply remove the insulator and there’s an epigenetic way to switch on the gene.”