Journal of the National Cancer Institute, Feb. 8, 2012.

One reason progress in epigenetic therapy for cancer has been so slow and painstaking is that the available drugs are old and imperfect.  For example, the heavily-used azacitidine was first synthesized in the 60s.  But a new generation of epigenetic drugs appears to be on the way. 

They’re badly needed, because epigenetic therapy to date has shown little effect on solid tumors.  (A recent lung cancer trial is the only exception.)  There is even doubt that these drugs work by an epigenetic mechanism—that is, by altering chromatin (the packaged form of DNA) to change gene expression.  “There is still a major issue about lack of really true epigenome targeting,” said Robert Brown, Ph.D., chair of translational oncology at Imperial College London, at the November 2011 Molecular Targets and Cancer Therapeutics meeting in San Francisco.  Researchers have also had difficulty identifying the subset of patients who could benefit from treatment. For all these reasons, the field of epigenetic therapy, while continuing to refine the use of the older drugs, is eager to try new ones.   Current epigenetic drugs, said Brown, “we hope are just the tip of the iceberg,” said Brown.

Fortunately, completely novel classes of epigenetic drugs should soon become available for testing in patients.  In the last few years, researchers have reported genetic alterations in tumors for several enzymes that have epigenetic functions.  These enzymes—histone methyltransferases and histone demethylases–place and remove methyl groups from histones, proteins around which DNA is coiled.  At least three biotech companies and several major pharmaceutical companies are now targeting these enzymes in cancer.  Unlike histone deacetlyases (HDACs), these drug targets have the imprimatur of genetics to show their importance in cancer, and drugs now in development have the potential to target individuals with specific genetic defects in their tumors.

Genetics Validates Epigenetics

 Histones are important cogs in the machinery of DNA transcription and gene expression.  An octamer consisting of two copies each of four core histones, with their associated DNA, forms the nucleosome, the repeating subunit of chromatin.  Each histone has a flexible amino terminus, or “tail,” that projects from the nucleosome.  Histone tails are the sites of a variety of post-translational modifications, including phosphorylation, ubiquitination, acetylation—and methylation.  Until 2004, histone methylation was thought to be stable and irreversible.  That year the first histone demethylase, LSD1, was reported, and more than two dozen others have since been identified.  “Many of these methyltransferases [and] demethylases are just critical,” said epigenetics researcher Yang Shi, Ph.D., of Harvard University, who first identified LSD1 as a demethylase.  “You cannot not have them in the right shape and form; otherwise… the cells just don’t do well.” Researchers now view histone methylation as a dynamic process, a balance between methylation and demethylation caused by their respective enzymes acting in concert. Imbalances between histone methylation and demethylation, Shi said, often lead to disease.  

It makes sense that histone methylation status is important in cancer, because it probably influences gene expression.  Unlike DNA methylation, which always shuts down gene expression, histone methylation is more complicated.  It can be associated with chromatin either in its open or condensed state, which leads to gene expression and gene silencing, respectively.   This depends on site and degree of methylation.   Lysine residues, especially on histone H3, are the ones most often methylated.

Do the enzymes that methylate and demethylate histones drive cancer?  Recent genetic discoveries suggest that the answer is yes.  In 2009 mutations in UTX, the enzyme that demethylates lysine 27 on histone H3 (H3K27 in the standard shorthand), were reported in a variety of human tumors.  The following year researchers reported mutations in EZH2, the enzyme that methylates the same residue, in 7% of follicular lymphomas and 22% of diffuse large B cell lymphomas.  Also in 2010, mutations in JARID1C, a H3K4 demethylase, and SETD2, a H3K36 methyltransferase, were each reported in 3% of clear cell renal cell carcinomas. 

Besides gene mutations, other abnormalities implicate histone-modifying enzymes. The histone methyltransferase DOT1L, responsible for methylating H3K79, is associated with several proteins that fuse with the MLL (mixed lineage leukemia) gene in leukemias and is required for their maintenance.  The histone methyltransferase SETDB1 is frequently amplified in malignant melanoma and, in 2011, researchers reported that it accelerates melanoma formation in animal models.   And MMSET, which methylates H3K36, is frequently overexpressed in multiple myelomas that have the t (14; 4) translocation.   

Finally, the metabolic enzymes IDH1 and IDH2 are mutated in over 70% of low- and medium-grade gliomas and 15-20% of adult acute leukemias. These mutant enzymes generate an “oncometabolite” that inhibits histone demethylases, among other enzymes with epigenetic functions.  (See JNCI July 7, 2010, p. 926-928.)

Enzymes that methylate and demethylate histones have become hot targets.  “There’s a lot of genetic evidence for their role in cancer,” said Kristian Helin, Ph.D., director of the Biotech Innovation and Research Centre at the University of Copenhagen.  “I think they’re absolutely valid targets for the development of new drugs.”

A Prolific Oncogene

At least three biotech companies—Epizyme, Inc. and Constellation Pharmaceuticals, both in Cambridge, Massachusetts, and EpiTherapeutics in Copenhagen—are heavily invested in this approach.  So are pharmaceutical companies like GlaxoSmithKline, Astra-Zeneca and Novartis.  “Almost every big pharmaceutical company has a robust program in epigenetics,” says Shi.  “It’s quite a change in the past few years.”

The histone methyltransferase EZH2 is probably the most popular target, with drug development starting well before the discovery of the lymphoma mutations.  In 2002 pathologist Arul Chinnaiyan, at the University of Michigan in Ann Arbor, reported frequent EZH2 overexpression in prostate cancer, especially in aggressive tumors.  Similar reports in breast cancer, bladder cancer, melanoma and other carcinomas followed, revealing EZH2 as one of the most overexpressed epigenetic regulators in cancer.  Genetics ramped up interest even further when a group from the British Columbia Cancer Agency in Canada reported the EZH2 mutations in lymphomas in 2010.   This paper and others “tremendously increased the interest in EZH2, because that sort of credentialed EZH2 as an oncogenic factor with a genetic basis, at least in certain cases,” said Chinnaiyan.  “In terms of cancer evolution, this is something that could be selected for, and be a driver in certain tumor types.”  UTX mutations also validate EZH2 as a driver in cancer, because by interfering with demethylation of H3K27, they achieve the same result as EZH2 activating mutations.

A brief controversy flared when the British Columbia group characterized the lymphoma EZH2 mutations as inactivating mutations, suggesting that EZH2 was a tumor suppressor gene, not an oncogene.  Researchers at Epizyme then reported that the mutations were activating, but only when it came to adding methyl groups to H3K27 on top of existing methyl groups. Since the mutation is only found in one copy of the gene, tumors contain both wild-type and mutant enzyme, and it turns out they cooperate to pile on methyl groups until the residue is trimethylated—the modification most strongly associated with a repressive chromatin state.  (This presumably inhibits the expression of tumor suppressor genes.)  “This type of catalytic coupling between a wild-type and a disease associated mutant is unprecedented as a mechanism of pathogenesis in human disease,” said Robert Copeland, Ph.D., Epizyme’s chief scientific officer, at the 2011 annual meeting of the American Association for Cancer Research (AACR). The British Columbia group now agrees that mutated EZH2 is an oncogene in lymphoma.  (Inactivating EZH2 mutations in myelodysplastic syndrome have been reported, and that paradox remains unresolved.)

 Epizyme, at the AACR meeting, reported selective EZH2 inhibitors with single-nanomolar potency in in vitro assays.  These compounds “show dramatic selectivity for killing cells, lymphoma cells, that contain the mutation,” said Copeland.  Epizyme is partnering with Eisai in Japan to develop these compounds for selectively treating lymphoma patients whose tumors bear EZH2 mutations.  Other companies, less visible, are also doing this.  “EZH2 is certainly the target of a number of drug development programs in the industry,” said Chinnaiyan. 

Can EZH2 be inhibited without unacceptable side effects?  It’s a subunit of the Polycomb repressive complex 2 (PRC2), which has important functions in normal biology, including differentiation, maintenance of cell identity and proliferation, and stem cell plasticity. A therapeutic window in adults, said Chinnaiyan, is plausible. “The thought is, in a differentiated organism, the toxicities would be minimized,” he said. “But it’s still early days.”

Undisclosed Targets, Uncertain Biology

Other histone methyltransferases and demethylases are probably targets, although company disclosures at this early stage are scant.  Epizyme is targeting DOT1L, presumably because of its involvement in many leukemias.  LSD1, which demethylates mono- and bimethylated H3K4, may be a target at some companies.  (The group of Robert Casero, Ph.D., at Johns Hopkins showed in 2009 that LSD1 inhibitors, in combination with azacitidine, could block colon cancer tumor growth in mice.)   And Agios Therapeutics in Cambridge, Massachusetts is developing small molecule inhibitors of mutant IDH1 and IDH2 as a way to indirectly target histone and DNA demethylases, although it’s hard to know exactly which downstream effects are most important in driving tumor growth.  “It’s probably a combination of effects at the different levels, at the histone and the DNA level,” said Agios chief scientific officer Scott Biller, Ph.D.

 Because these enzymes are so new, companies are working hard to understand the biology while simultaneously rushing drug candidates through preclinical pipelines. In 2006 Kristian Helin at the University of Copenhagen showed that GASC1, one of many enzymes containing the so-called Jumonji C catalytic domain, was a histone demethylase with a H3K9 substrate, and that inhibitors could slow cancer cell growth. According to Helin, EpiTherapeutics is working on GASC1, among other enzymes, as a target in cancer.  Excessive levels of GASC1, Helin speculates, can lead to genomic instability and cancer, or oncogene expression.  But he stresses that this is mostly guesswork.  “It’s not like we have for Ras [oncogene] 30 or 40 years of science behind this.  This is a relatively new field.”

Researchers like Helin are careful not to claim too much knowledge at this point.  What’s known, from studies of the role of histone methyltransferases and demethylases in embryonic development and in embryonic stem cells, is that “they participate in the regulation of the [cell] differentiation programs,” Helin said. “Those enzymes are clearly involved in regulating the processes, the cell fate decisions [and] the specification of cells.” Since cancer can be viewed as a disease of faulty differentiation, it makes sense that these enzymes are involved.  But there is no proof, Helin pointed out, that the methylation “marks” on histones that are deposited and removed by these enzymes actively influence gene expression, although he thinks they probably do.  “In reality, none of those marks have been shown to be important for anything,” he said.  “But the enzymes which put on the marks, they have.”  The mutations, amplifications and translocations involving those enzymes are ample evidence that the enzymes drive cancer, in Helin’s view. 

So the genetic evidence for the importance of histone-modifying enzymes exists, but epigenetic evidence is lacking.  Since these enzymes methylate and demethylate other proteins besides histones, and can even have nonenzymatic roles in biology, researchers can’t yet know whether the histone modifications themselves are the key causal factor in the changes that lead to cancer. The enzymes could be modifying other classes of proteins, or acting as scaffolds in unknown protein complexes, Helin said. 

Not knowing exactly how these enzymes are working in cancer makes it hard to know whether histone modifications themselves determine cell fates, or whether they’re a consequence of the underlying nucleosome patterning.  “This is still an open question,” said Helin.  But even if histone-modifying  enzymes function only to maintain transcription programs rather than initiate them, that “doesn’t make them less important,” Helin said, “because the genetic evidence suggests that they’re important.”  Nor, he stressed, does it lessen their potential as drug targets in cancer.  Clinical researchers should soon have drugs to test the idea that restoring or removing a simple chemical modification on DNA’s packing material can help defeat cancer.

Dr. Shi is a cofounder of Constellation, has an equity interest in the company, and is on its scientific advisory board (SAB).  Dr. Helin is a cofounder of EpiTherapeutics, is a consultant to the company, and is on its SAB and board of directors.  Dr. Chinnaiyan is on Constellation’s SAB, is an advisor to SmithKlineBeecham and to Ventana/Roche, which has licensed University of Michigan intellectual property originating in Dr. Chinnaiyan’s lab.