GENOME-WIDE MAPPING OF DNA MODIFICATIONS IN ADDICTION
Although most studies to date have taken a candidate gene approach to identify genes regulated by DNA methylation in addiction models (Table 8.1), it is essential to gather an unbiased, genome-wide view of such regulation. We and other groups have used chromatin immunoprecipitation (ChIP) followed by promoter arrays and more recently ChIP sequencing (seq) to map drug-induced changes in several histone modifications in specific brain regions (Renthal et al., 2009; Zhou, Yuan, Mash, & Goldman, 2011), and similar approaches are now a high priority for DNA methylation.
Given the role of TET1 in NAc in cocaine action as already described, we mapped cocaine-induced 5hmC alterations in this brain region. By use of a chemical biotin labeling approach (Song et al., 2011), 5hmC-enriched DNA fragments were purified from mouse NAc and sequenced (Feng et al., 2015). In total, we recognized more than 20,000 differential regions, the majority of which occurred in gene body and intergenic regions. To better understand the potential function of 5hmC regulation in intergenic regions, we focused on 5hmC dynamics at putative enhancer regions. Enhancers are regulatory elements that can exist long distances from transcription start sites. We generated ChIP-seq maps for two histone marks, H3K4me1 and H3K27ac, both of which have been used to identify enhancers (Creyghton et al., 2010). We then used a combinatorial approach to define distinct chromatin states (Ernst et al., 2011) at nonpromoter regions based on altered binding of these histone marks plus 5hmC. We observed dynamic regulation of chromatin states at putative enhancers in response to cocaine (Feng et al., 2015). Moreover, we detected cocaine regulation of 5hmC around exon boundaries, sites correlated positively with alternative splicing changes detected by RNA-seq. Overlay of RNA-seq with 5hmC data in coding regions further revealed that increased levels of 5hmC in gene bodies is associated positively with increased steady state transcription of that gene or its greater inducibility in response to a subsequent cocaine challenge (Feng et al., 2015). The genes that displayed this regulation are highly enriched in addiction-related gene categories. Moreover, we observed that, at least at some loci, the cocaine-induced changes in 5hmC persist at least 30 days after cocaine administration (Feng et al., 2015). It would be interesting to map 5hmC at this later time point as well as in cocaine self-administration models and to compare such patterns to those of 5mC. Studies of drug regulation of DNA methylation in other brain reward regions are also needed.
Another recent study demonstrated dynamic changes in DNA methylation in the NAc during incubation of cocaine craving in rats (Massart et al., 2015). By use of methyl-DNA immunoprecipitation—which is mostly selective for 5mC—followed by microarray analysis, the authors examined methylation changes at gene promoters of all coding genes and across the entire gene length of a custom panel of 47 candidate genes previously implicated in drug addiction. They identified broad and time-dependent alterations in DNA methylation in this brain region after cocaine withdrawal and cue- induced cocaine seeking. Interestingly, patterns of DNA methylation varied dramatically between 1 day and 30 days of withdrawal and were rapidly reversed within 1 h of cue- induced reinstatement (Massart et al., 2015). These findings demonstrate the highly dynamic regulation of the DNA methylome in brain and raise the possibility that certain alterations might offer prognostic value if detectable in peripheral tissues.
Studies of the DNA methylome in alcoholism are also beginning to appear. In an investigation on frontal cortex of human alcoholics by use of NimbleGen Human DNA Methylation promoter arrays, numerous differential methylation loci were observed in both novel and known target genes that are either hypo- or hypermethylated in alcoholism (Manzardo, Henkhaus, & Butler, 2012). Another microarray-based genome-wide DNA methylation analysis, this one with Illumina chips, surveyed the DNA methylome in lymphocytes from ~60 alcohol-dependent patients with a similar number of controls (R. Zhang et al., 2013). The study found 1710 CpG sites that were differentially methylated between alcoholics and controls, with the majority hypomethylated in patients. The identified genes were enriched in several interesting categories such as stress, immune response, and signal transduction. Other Illumina chip-based studies also found DNA methylation differences in the blood or saliva of alcoholics versus control subjects (Harlaar et al., 2014; Philibert et al., 2014; H. Zhang et al., 2013). In an investigation of 18 monozygotic twin pairs discordant for alcohol use disorders, genome-wide DNA methylation array screening and Sequenom EpiTYPER validation of the same peripheral blood DNA samples identified ~20 differentially methylated regions associated with alcohol (Ruggeri et al., 2015). It would be important to validate these novel changes in methylation in additional cohorts and to determine whether similar changes occur within relevant brain reward regions.
Even though the vast majority of studies to date have focused on DNA methylation associated with coding genes, recent work suggests that addiction can also regulate DNA methylation at normally silenced repetitive elements. By applying a novel transcriptome analysis approach, namely, weighted gene coexpression network analysis (Zhang & Horvath, 2005), Ponomarev et al. (2012) studied transcriptional alterations associated with alcohol use disorders in postmortem human brain. The authors identified previously unrecognized epigenetic determinants of gene coexpression relationships and discovered novel markers of chromatin modifications in the amygdala and superior frontal cortex of alcoholics (Ponomarev et al., 2012). Higher expression levels of endogenous retroviruses in alcoholics were then confirmed to be associated with DNA hypomethyl- ation, suggesting a critical role of DNA methylation in alcohol addiction. Consistent with this, chronic cocaine administration has been shown to trigger the loss of H3K9me3 at repetitive genomic sequences in mouse NAc and hence increase their expression (Maze et al., 2011). It would be interesting to follow up these observations with measures of DNA methylation at these loci after cocaine exposure. Although further work is needed, studies to date suggest that the control of repetitive elements by DNA methylation and related repressive chromatin mechanisms represents a novel form of epigenetic regulation in the addicted brain.