7 December 2011. Heritability is no longer just about DNA. A large body of research implicates epigenetic alterations—mitotically heritable changes that control gene activity without altering the DNA sequence—in a number of major illnesses, including schizophrenia. The role of epigenetics in disease was the subject of a 15 November 2011 Presidential Special Lecture at the 41st annual Society for Neuroscience Meeting by Andrew Feinberg of Johns Hopkins University School of Medicine in Baltimore, Maryland. Feinberg’s talk, entitled “The Epigenetic Basis of Common Human Disease,” detailed a role for alterations in DNA methylation in cancer, and discussed interesting issues that may be relevant to schizophrenia research.
Epigenetics is the study of non-sequence, heritable information in DNA and chromatin. A major epigenetic mechanism is DNA methylation, the transfer of a methyl group to a cytosine residue at the dinucleotide sequence CpG (adjacent cytosine and guanine nucleotides on the same DNA strand connected by a phosphodiester bond) that generally leads to gene repression. In general, CpG sites are heavily methylated, although clusters of CpG sites (termed CpG islands) at promoter regions tend to be less methylated. Substantial evidence points to a role for epigenetic alterations in psychiatric illness (see SRF Current Hypothesis paper by Dennis Grayson, as well as an interview with Art Petronis), and studies suggest that epigenetics may, at least in part, account for the high discordance for schizophrenia among identical twins. For example, a recent twin study has demonstrated altered DNA methylation patterns associated with schizophrenia (see SRF related news story).
In his talk, Feinberg pointed out that heart and brain tissue are very different, despite carrying the same DNA sequence, and that such differences may be mediated by epigenetic mechanisms. In an attempt to elucidate the role of DNA methylation in these differences, Feinberg determined regions of the genome that are differentially methylated across tissue types. Using a novel technique they dubbed CHARM, which stands for “comprehensive high-throughput arrays for relative methylation” (Irizarry et al., 2008), Feinberg’s group identified over 16,000 regions in the genome that contained DNA methylation differences among brain, liver, and spleen tissue (Irizarry et al., 2009). This novel technique is notable for its agnostic approach regarding the location of genes and CpG content, and thus the ability to explore DNA methylation in regions beyond the usually studied CpG islands (which are generally enriched on arrays owing to their high CpG content). In fact, the majority of tissue-specific methylation was not found in CpG islands, but in adjacent regions (within 2 kb), termed “CpG island shores” (Irizarry et al., 2009). These data suggest that the popular CpG island-centered approach for hunting epigenetic changes in disease may need to be reconsidered (Jones and Baylin, 2007), a point that may also be applicable to schizophrenia methylomic studies. For example, in a 2011 study by Dempster and colleagues (Dempster et al., 2011), the top psychosis-associated methylation site identified was not located within a CpG island.
Feinberg also discussed his work examining colon cancer-related changes in DNA methylation using CHARM. Specifically, he detailed his finding that many of the regions that exhibit altered methylation in cancer (cancer-specific differentially methylated regions, or cDMRs) also show methylation differences among tissue types (Irizarry et al., 2009). Additionally, these same cDMRs are differentially methylated during stem cell reprogramming of induced pluripotent stem cells (Doi et al., 2009), suggesting that the sequences that govern the differentiation of disparate tissues are also altered in cancer. Moreover, the cDMRs are not just specific to colon cancer: the same regions are also differentially methylated in a variety of cancers including those of the breast, lung, and thyroid, which is suggestive of a general disruption of the cancer methylome (Hansen et al., 2011). Especially of interest, in contrast to the low variability in methylation levels observed in normal tissue, these regions exhibit high levels of stochastic variation in methylation levels within each cancer type (Hansen et al., 2011). The most variable cDMRs are located within genes implicated in tumor heterogeneity and progression. If epigenetic variation plays a role in cancer, could it also be involved in schizophrenia? Ongoing studies by Feinberg’s and other labs examining methylation variability in schizophrenia may answer that question.—Allison A. Curley.