© 2004 MacMillan Magazines, Ltd.

Covalent histone modifications such as H3 lysine 9 acetylation and methylation create a secondary, self-reinforcing signal that regulates transcription. Histone deacetylases (HDAC) remove acetyl groups from lysine residues making way for methylation. HP1 recognizes MeK9 and also associates with histone methyltransferases (HMT), DNA methyltransferases (DNMT) and methyl-C binding proteins (MBD) which aid in spreading the silencing signal.

The genetic model of cancer – the idea that key mutations lead to unchecked cellular proliferation – has guided cancer research for decades. Thousands of papers report sequence alterations that disrupt, delete, or overexpress genes, leading to oncogenesis. Then, in 1983, Bert Vogelstein and Andrew Feinberg at Johns Hopkins University reported widespread loss of DNA methylation at cytosine-guanine (CpG) dinucleotides in tumor samples.1 This was the first evidence that eigenetic changes, which are heritable but outside of the genome sequence, might spur cancer.2 Many...


Epigenetic programming influences many cellular and organismal processes, especially during growth and development. Several initiatives now plan to draw an epigenetic map. "It seems obvious, but we still don't have a global view of the methylation pattern and other facets of epigenetic modification across the whole genome," Feinberg says.

Nevertheless, it's already clear that aberrations in the three main forms of epigenetic programming – DNA methylation, gene imprinting, and histone modifications – contribute to cancer. DNA methylation patterns are the best studied. Methylation is relatively rare in CpG islands (runs of successive C and G nucleotides) in the promoter regions of housekeeping genes. Hypermethylation of these islands can lead to aberrant silencing of tumor suppressor genes such as BRCA1, p16INK4a, and hMLH1.3

Manel Esteller, who heads the cancer epigenetics group at the Spanish National Cancer Center, says that hypermethylation also modulates newly discovered genes such as MGMT, a "key player for the sensitivity to chemotherapy" and silences the gene encoding E-cadherin, a cell-adhesion molecule. Downregulating E-cadherin promotes cancer invasion and metastasis. On the other hand, hypomethylation, either globally or at individual genes, can activate oncogenes.23 "Epigenetic changes occur from the beginning to the end of cancer development," Esteller says.

Tumor hypomethylation also seems to be associated with the chromosomal instability and translocations that underlie, among other malignancies, some ovarian and breast cancers, and that produce the chimeric oncoproteins that drive certain leukemias. For example, the fusion protein PML-RAR induces hypermethylation that silences tumor suppressor genes, thereby promoting leukemia.4 "The primary event is genetic, but all that follows is epigenetic," Reik remarks.

Rather than occurring in the tumor, the second epigenetic mechanism, genomic imprinting, silences one parental allele in the zygote or gamete. The offspring's somatic cells inherit the imprint. For instance, imprinting normally silences the maternal insulin-like growth factor II (IGF2) gene. Aberrant genomic imprinting, however, can allow excessive expression of some growth-promoting genes, such as IGF2, or silence certain tumor suppressor genes.3 Feinberg's group recently reported that loss of imprinting at IGF2 seems to contribute to colorectal neoplasia.5

Finally, covalent modifications on DNA-bound histones such as acetylation, methylation, and phosphorylation help regulate transcription and often remain stable during cell division. The mechanism that propagates this histone code may be "the most important open question in molecular genetics."2 Although less well studied than DNA methylation, a growing body of evidence suggests that histone modification contributes to cancer development and progression.

Esteller's group recently identified characteristic histone marks in several types of cancer cell lines and primary tumors.6 The cancer cells showed a deficit in the monoacetylated and trimethylated forms of histone H4. These changes appear early in the cancer's natural history and accumulated as the malignancy developed.



© 2004 MacMillan Magazines, Ltd.

Loss of imprinting (LOI, dark nuclei) can arise sporadically (a), or early in germline development as in Beck-with-Widemann syndrome (b) which causes overgrowth of the whole kidney due to a double dose of IGF2. A secondary, presumably genetic effect can lead to Wilms tumor formation.

Decades of scouring tumors for genetic signatures identified a few mutations that aid diagnosis and prognosis prediction, but only a few. The growing appreciation of epigenetics' contribution has fueled interest in its potential for diagnostics and prevention. Feinberg says he's excited about the prospect of preventing cancers by scanning for epigenetic changes and modifying a patient's risk profile. People with loss of IGF2 imprinting, for example, could modify their diets and undergo regular screening for colorectal cancer.

Several environmental toxins and dietary factors affect the methylation machinery. A group from Washington State University in Pullman recently reported on injecting pregnant rats with endocrine disruptors. Adult male offspring had reduced spermatogenic capacity that was passed down the male line as far as the F4 generation.7 The phenotype appeared to be related to DNA methylation. And dietary methionine has been shown to affect DNA methylation in mouse progeny affecting things like coat color.8 Similarly, Esteller notes that adequate consumption, neither excessive nor deficient, of methyl-donors such as S-adenosyl-methionine is important to prevent cancer.

Meanwhile, Berlin-based biotech Epigenomics is developing diagnostic tests that measure the extent and pattern of DNA methylation. The company found high levels of methylation on three markers that distinguish samples taken from men with prostate cancer who had early recurrences and those in whom the malignancy did not recur. In breast cancer, 86% of women with low methylation levels at the transcription factor PITX2 were metastasis free at 10 years. This compared to 67% in patients with high methylation levels. "Epigenetic analysis augments conventional diagnostic tests," says Reik, an advisor to Epigenomics. "There's no doubt that these will be important diagnostic tests; the extent of their impact needs to be fully explored."

And, some evidence exists that shows epigenetic manipulators to be clinically effective. For instance, azacitidine, used to treat myelodysplastic syndromes, acts in part by altering epigenetic patterns. Azacitidine failed to obtain FDA approval about 25 years ago as a conventional cytotoxic. But the recognition that it causes progressive DNA hypomethylation gave the drug a new lease on life and lead to FDA approval in May 2004. Azacitidine acts in part by reactivating previously silenced tumor suppressors and other genes, although other actions might contribute to its efficacy. For example, azacitidine can inhibit RNA synthesis and produces direct cytotoxicity.9 Additionally, histone deacetylase inhibitors, which have been in development since the late 1990s, could lead to another new generation of effective treatments. Changing the histone modification pattern could induce apoptosis, for example.

In a recent study, another drug called decitabine (5-aza-2-deoxycytidine) modified methylation of the CpG island in the E-cadherin promoter.10 Aberrant CpG hypermethylation in this region seems to reduce expression of E-cadherin, an adhesion molecule. In human cancer cells decitabine induced reexpression of E-cadherin, leading to reduced cell motility. In animal models, decitabine suppresses the growth of the primary tumor and inhibits metastases.

"We still need to know more about the clinical impact of affecting epigenetic mechanisms," Reik says. "However, the results with azacitidine in the clinic and with other drugs in animals are very promising." Feinberg agrees that "epigenetic modification could be a promising approach to treatment," but notes that an unselective approach to changing epigenetic methylation is counterintuitive. "This could switch on as many genes as it silences," he says. "We need to find a way to make much more selective changes." Then again, Feinberg notes, current cancer treatment with conventional agents is often unselective.

It'll be several years before epigenetics' full clinical implications become clear. In the meantime, understanding cancer's underpinnings will increasingly incorporate genetics and epigenetics. As George Klein from the Karolinska institute in Stockholm notes, "Changes in gene expression, due to modifications of chromatin structure, but not in the genes themselves, can play equally important roles in the malignant microevolution as genetic changes that affect the DNA sequence."

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!