Reward-related memory systems
Unlike learned fear responses, the ability to associate predictive environmental cues with reward-related experiences is regulated by the mesolimbic dopamine system. This type of learning can be formally tested by measuring behavioral responses to sensory cues that predict natural rewards (eg, sugar, water). As a result of pairing with rewards, cues themselves begin to elicit anticipatory learned approach behaviors. Dopamine neurons located in the ventral tegmental area (VTA) undergo dynamic alterations in firing rate during the acquisition of these cue-reward associations (Schultz, Dayan, & Montague, 1997), and this process is associated with synaptic plasticity at glutamatergic synapses onto dopamine neurons (Stuber et al., 2008). Furthermore, the activity of dopamine neurons is both necessary and sufficient for learned reward responses (Di Ciano, Cardinal, Cowell, Little, & Everitt, 2001; Tsai et al., 2009). Recently, this form of motivated learning was also shown to depend on DNA methylation in the VTA (Day et al., 2013). Associative cue-reward learning induced a selective upregulation of the immediate early genes Fos and Egrl in the VTA, and immunohistochemistry using a dopamine cell marker indicated that this increase occurred specifically in dopamine neurons. Moreover, the degree of these changes was correlated with memory acquisition and was associated with altered DNA methylation patterns at these genes, suggesting a potential link between DNA methylation and reward memory formation. Direct infusion of the small molecule DNMT inhibitor RG108 in the VTA before learning produced selective impairment of cue-evoked conditioned responses, without altering reward consumption or baseline behavioral responses. Critically, DNMT inhibition in the VTA after memory acquisition did not impair previously learned behaviors, suggesting that DNA methyla- tion in the VTA is required for the encoding of associative reward memories, but not long-term memory storage or retrieval (Day et al., 2013).
Experience with drugs of abuse also exerts potent control over brain reward circuits and is capable of generating robust memories that can drive addicted individuals to relapse. DNA methylation seems to have a critical role in this process as well (Anier, Malinovskaja, Aonurm-Helm, Zharkovsky, & Kalda, 2010; Feng et al., 2015; LaPlant et al., 2010; Massart et al., 2015). DNMT3a expression is dynamically modulated in the nucleus accumbens (NAc; a key reward structure) after passive or active cocaine administration in animal models, and this change endures even after 28 days of drug withdrawal (LaPlant et al., 2010). Similarly, cocaine experience alters DNA methylation patterns at key genes that have been shown to regulate drug-related behavioral and synaptic responses, such as Fosb (Anier et al., 2010). Unlike results in other learning models, blockade of DNMT activity or Cre-mediated DNMT3a excision within the NAc accelerates conditioned place preference memory, whereas herpes simplex virus-mediated overexpression of DNMT3a disrupts cocaine reward memory (LaPlant et al., 2010).
Similarly, expression of Tetl, one of the three enzymes responsible for hydroxylation of methylcytosine, is downregulated in the NAc after cocaine experience (Feng et al., 2015). Loss of Tetl results in increased cocaine memory, whereas Tetl overexpression impairs cocaine place preference. Exposure to cocaine also induces increases in 5hmC levels at numerous genes that are upregulated by cocaine, and knockout of Tetl causes increased 5hmC levels at the same loci. This is perhaps counter-intuitive given that Tetl has been shown to regulate the conversion of 5mC to 5hmC (Guo, Su, Zhong, Ming, &
Song, 2011b; Kaas et al., 2013; Tahiliani et al., 2009). However, given that Tetl may also be involved in the generation of further oxidative species 5-formylcytosine (5fC) and 5-carboxylcytosine, it is possible that 5hmC accumulation after Tetl loss is the result of the lack of this continued oxidation (Raiber et al., 2015) or potentially through other noncatalytic roles ofTet1 in the regulation of gene expression (Tsai et al., 2014). Nevertheless, it is clear that DNA methylation and hydroxymethylation play a crucial role in drug-related behavioral memory, indicating that selective targeting of DNA methylation processes may be a potential avenue to targeted addiction therapeutics.