Converging lines of evidence suggest that cocaine commandeers traditional reward- related learning and memory pathways to instill pathologically persistent memories, which encode the association between stimuli in the drug use environment and the rewarding effects of the drug. When triggered, these maladaptive memories can recall the pleasurable effects of the drug, stimulate cocaine craving and precipitate relapse. However, the cellular and molecular adaptations that enable the maintenance of these memories during abstinence remain enigmatic.
As outlined in Chapter 2, comparatively stable epigenetic modifications, such as DNA methylation, may function as a conserved means of perpetuating memory in the face of rapid transcriptional and proteomic turnover and degradation. Learning-induced modifications of DNA methylation have been implicated in the maintenance of contextual fear memories and may therefore also underpin the maintenance of cocaine-related memories. To investigate this possibility, we first established a novel next-generation sequencing technique (MBD Ultra-Seq) to probe genome-wide region- and cell type- specific changes in DNA methylation in individual animals (outlined in Chapter 3). I then applied this technique to identify genome-wide changes in DNA methylation in mice following chronic cocaine self-administration and passive (yoked) cocaine exposure, after 1 or 21 days of forced abstinence (Chapter 4). Modifications of DNA methylation that regulate the maintenance of cocaine-related memories must arise from learned cocaine- seeking and therefore be unique to the cocaine self-administration paradigm, in addition to being persistent, in order to be congruent with the enduring nature of memory. Overall, I identified 29 genomic regions that became persistently differentially methylated during cocaine self-administration and 28 regions that became selectively differentially methylated during abstinence, all of which may contribute to the maintenance of cocaine- related memories.
Functionally, persistent learning-induced changes in DNA methylation are thought to produce enduring modifications of gene expression, thereby altering the physiology of activated neurons and perpetuating memory. However, as posited in Chapter 2, experience-dependent variations in DNA methylation might also represent a form of genomic metaplasticity that is transcriptionally quiescent during memory storage and instead primes the transcriptional response upon subsequent neuronal or memory re-activation. As preliminary in vivo evidence of this hypothesis, I examined how the relationship between altered DNA methylation and the transcription of overlapped or proximal genes is regulated in response to memory reactivation (Chapter 4). In some cases cocaine self-administration induced changes in DNA methylation that had lasting transcriptional consequences, while in others a relationship between altered DNA methylation and the transcription of proximal genes was only evident following the explicit reactivation of cocaine-associated memories. This is the first evidence to suggest that the reactivation state of a memory may govern the relationship between learning-induced changes in DNA methylation and transcription.
Taken together, these data constitute the first in vivo neuron-specific genome-wide profile of variations in DNA methylation associated with learned cocaine seeking and not simple drug exposure, where the former is more relevant to the development and persistence of addiction. Moreover, we demonstrate that the relationship between learning-induced modifications of DNA methylation and transcription is complex and mediated not only by the genomic location of DNA methylation, but also by the reactivation state of the relevant memory.