Epigenomic and Noncoding RNA Regulation in Addictive Processes

Abstract

The phenotypic effects of drugs of abuse are partially mediated by transcriptional and epigenetic regulatory mechanisms. This chapter will provide a brief overview of substance abuse and then focus on the roles of three epigenetic regulatory mechanisms in addictive processes: histone modifications, DNA modifications, and noncoding RNAs. This chapter will conclude with a focus on three other important areas: (1) the potential for long-lasting epigenetic effects due to drugs of abuse, (2) obstacles and opportunities in this scientific area as they pertain to addiction biology, and (3) the potential for translating epigenomic and noncoding RNA discoveries into improvements in human health and the treatment of substance use disorders.

1 Introduction

1.1 The Environment and Epigenomic Regulation

With certain exceptions, an individual’s genome is believed to be more or less identical in every cell. However, the epigenomes of different cell types within an individual appear to differ significantly from one another (Hawkins et al. 2010). This is consistent with the distinct phenotypes, functions, and gene expression profiles of particular cell types. There have been a number of reviews indicating that our epigenomes may be sensitive to “environmental” influences which can be broadly defined to include diet, toxins, stressors, and psychosocial influences (Jirtle and Skinner 2007; Zhang and Meaney 2010; Caldji et al. 2011). It has been hypothesized that environmental exposures may lead to changes in signaling in specific cell or tissue types. These changes may in turn impact epigenetic regulation of gene expression, ultimately leading to transient or long-lasting changes in gene expression and cellular or organismal phenotypes. In some cases, these changes appear to be passed on through mitosis or even to subsequent generations (Youngson and Whitelaw 2008).

1.2 Epigenetic Regulation in the Nervous System

Epigenetic regulation has been shown to be important in neurogenesis, neural fate specification, neuronal development, behavioral plasticity, synaptic plasticity, circadian regulation, and learning and memory (Day and Sweatt 2011; Haggarty and Tsai 2011; Ma et al. 2010; Zocchi and Sassone-Corsi 2010; Bellet and Sassone-Corsi 2010; Maze and Nestler 2011; Nelson and Monteggia 2011; Ma et al. 2010; Dulac 2010; Namihira et al. 2008). Additionally, misregulation of epigenetic processes has been implicated in a number of human disorders including neurodevelopmental disorders (e.g., Rett and Prader-Willi syndromes) and psychiatric disorders (e.g., schizophrenia, depression) (Horsthemke and Wagstaff 2008; Graff and Mansuy 2009; Moretti and Zoghbi 2006; Tsankova et al. 2007; Pidsley and Mill 2011; Renthal and Nestler 2009a). Epigenetic regulation has also been implicated in response to early childhood abuse associated with suicide completion (McGowan et al. 2009). For the remainder of this chapter, I will focus on what is known about the role of epigenetic regulation in substance use and abuse.

2 Substance Abuse

2.1 Substance Use and Abuse

Our brains are inherently plastic, possessing connections and signaling processes that can change in response to distinct environmental exposures. Brains exposed to drugs of abuse on a continuing basis develop changes in particular neuronal regions, including those involved in the reward system. In addicted individuals, these brain changes can lead the individual to seek out drugs of abuse despite serious negative consequences. The definition of drug addiction has evolved over the years. According to Drs. Koob and Volkow: “Drug addiction is a chronically relapsing disorder that has been characterized by (1) compulsion to seek and take the drug, (2) loss of control in limiting intake, and (3) emergence of a negative emotional state (e.g. dysphoria, anxiety, irritability) reflecting a motivational withdrawal syndrome when access to drug is prevented” (Koob and Volkow 2010; American Psychiatric Association 2000; Koob and Le 1997). Common addictive substances include nicotine, alcohol, caffeine, cocaine, methamphetamine, opioids, certain prescription medications, inhalants, and cannabis. Food and sex are sometimes referred to as “natural rewards” and can exhibit effects similar to those caused by drugs of abuse (Olsen 2011; Avena et al. 2008). There are also compulsive “behavioral addictions” such as gambling or internet addiction that have some of the hallmarks of drug addiction (Grant et al. 2010; Ambermoon et al. 2011).

In broad strokes, exposure to addictive substances, such as alcohol, opiates, cannabinoids, and nicotine, leads to increased levels of dopamine within the mesolimbic dopamine system (Sulzer 2011; Justinova et al. 2009). For example, cocaine inhibits dopamine reuptake such that more dopamine remains at the synapse (Newman and Kulkarni 2002; Fleckenstein et al. 2007). Continued drug exposure ultimately leads to adaptations in the strength of circuit connections between different brain regions including the nucleus accumbens (NAc). These changes in circuit strength are mediated in part by alterations in signaling by the neurotransmitters dopamine and glutamate in specific types of neurons with concomitant gene expression changes (Luscher and Malenka 2011; Kalivas et al. 2009; Gardner 2011). A strengthening of the reward connections leads to changes such that an individual craves the drug of abuse more, even if substance use leads to adverse consequences. A weakening of the inhibitory influence of the prefrontal cortex (PFC) on the reward circuitry can also decrease the ability of the individual to resist substance use (Van den Oever et al. 2010).

Not all individuals exposed to drugs of abuse become addicted. Based on heritability measurements, this complex disease appears to have an important genetic component with some individuals particularly susceptible to addiction, while others are resistant to it (Johnson et al. 1996; Kendler et al. 1999; Uhl et al. 2008; Buckland 2008). Individuals that are particularly impulsive or have an enhanced propensity for risk taking are more likely to explore the use of drugs, and this impulsivity phenotype may have a genetic component (Perry and Carroll 2008; Dalley et al. 2011). Environmental influences are important in the development of substance abuse disorder (SUD); access to drugs of abuse, early life adversity, poverty, or exposure to drugs during critical periods such as adolescence can all influence the potential of individuals to develop substance abuse disorder (Caspi et al. 2005; Buka et al. 2003; Hill et al. 2005).

The overall economic cost of substance abuse in the USA has been estimated to be greater than $600 billion per year (Table 7.1). For illicit drugs alone, the cost is estimated to be $193 billion, while for alcohol and tobacco, the costs are estimated to be $235 billion and $193 billion, respectively. In addition to the serious economic cost to society from SUD, the consequences of addiction are extraordinarily destructive to the addicted individuals and their families. Behavioral therapies can be used to improve outcomes in substance abusers, and in some cases, therapy in concert with medication can improve outcomes (Carroll and Onken 2005). Despite ongoing efforts to develop safe and effective medications for the treatment of SUDs, only limited success has been achieved. Currently, approved medications exist to aid in smoking cessation as well as to treat opiate and alcohol dependence, including therapies for the initiation of and maintenance of abstinence (e.g., nicotine replacement therapy, buprenorphine, varenicline, naltrexone), to alleviate symptoms of withdrawal (e.g., varenicline, diazepam), and to prevent relapse (e.g., naltrexone, bupropion). However, with that said, no approved medications exist for the treatment cocaine, methamphetamine, or cannabis addiction even though efforts to develop these are ongoing (Montoya and Vocci 2008; McCann 2008). Unfortunately in the absence of good treatment options for the various SUD-related indications, addicted individuals will continue to struggle with this devastating disease.

Table 7.1 The estimated yearly economic cost of licit and illicit abused substances

In order to advance the development of effective medications and therapies to treat SUDs, it is critical that we understand the molecular and neurobiological mechanisms that lead to the development of addiction. Since exposure to drugs of abuse leads to long-lasting brain changes, it is not surprising that molecular studies exploring changes in gene expression in the nervous system have been very illuminating. For the rest of this chapter, I will focus on the role of transcriptional regulation in substance abuse with a focus on epigenomic regulation.

2.2 Transcriptional Regulation and Substance Abuse

Because exposure to drugs of abuse can lead to long-lasting brain changes including changes in neurotransmission in specific brain circuits, it has been proposed that alterations in gene expression via transcriptional regulation play a significant role (Nestler et al. 2001). The role of transcriptional changes in response to drugs of abuse has been well studied and reviewed (Nestler 2008; Nestler and Malenka 2004). Several transcription factors have been shown to play a role in addictive processes. For example, the cAMP-response element-binding protein (CREB), which has a well-established role in learning and memory, can alter drug abuse behaviors (Briand and Blendy 2010; Carlezon et al. 2005). In many ways addiction is an example of learning and memory gone seriously awry, so the identification of CREB is perhaps not so surprising.

Probably the most well-characterized transcription factor involved in drug addiction is the delta-FosB protein. Delta-FosB can be induced in the nucleus accumbens by opiates, cocaine, nicotine, other drugs of abuse, and natural rewards (such as sucrose and sex), and the targets of delta-FosB (e.g., GluR2, dynorphin, Cdk5) are congruent with the signaling molecules previously implicated in addictive processes (Nestler 2008). It has been proposed that significant delta-FosB induction can lead to “excessive sensitization of the nucleus accumbens circuitry” and ultimately lead to compulsive drug taking (Nestler 2008).

Transcription factors bind to and can alter the properties of chromatin, and conversely chromatin state may impact transcription factor binding and function (Birney 2011; Adrian et al. 2010; Koche et al. 2011). Although the role of transcription factors in mediating long-term changes to gene expression in the brain has been fairly well studied, the role of epigenetic regulation in addictive processes had not been investigated to a significant extent until recently.

2.3 Epigenetic Regulatory Mechanisms in Substance Abuse

Of the major epigenetic regulatory mechanisms, histone posttranslational modifications including histone acetylation, histone methylation, and histone phosphorylation have been the best studied in the area of substance abuse. For DNA modifications, DNA methylation has been recently investigated, while the role of the recently discovered DNA hydroxymethylation has not been investigated to date. The roles of ATP-dependent chromatin remodeling and nucleosome position have not been well characterized in substance abuse as yet. Noncoding RNAs can be important regulators of gene expression. While microRNAs have been investigated with respect to substance abuse, other noncoding RNA types including lincRNAs (long intergenic noncoding RNAs) have not been well characterized to date.

3 Histone Modifications and Addictive Processes

More than 100 distinct posttranslational histone modifications have been identified and more are likely to be discovered in the future. While some modifications are associated with active chromatin and others are associated with silenced chromatin, the function of the majority of these modifications is currently unknown (Campos and Reinberg 2009). It is also unclear whether histone modifications at particular chromatin regions cause chromatin structural changes or are simply a consequence of these changes (Henikoff and Shilatifard 2011). The enzymes that are responsible for the deposition and removal of posttranslational histone modifications, such as histone acetyltransferases (HATs), histone deacetylases (HDACs), histone methylases, and histone demethylases, are sometimes referred to as the “writers” and “erasers” of the histone code. Additionally, there are proteins that bind to histone posttranslational modifications, which are important for functions relevant to the modification. These molecules are sometimes referred to as “readers” of the histone code. Some of these “readers” contain protein domains such as PHD, Tudor, or WD40 that bind to methylated lysines or arginines, bromodomains that can bind to acetylated lysines, or 14-3-3 domains which can bind to phosphorylated residues (Gardner et al. 2011; Musselman and Kutateladze 2009; Sanchez and Zhou 2011; Kim et al. 2006; Bonasio et al. 2010). There has been limited work studying histone modification “readers” in the nervous system, although the little information that exists is tantalizing. For example, the JARID1C protein is involved in X-linked mental retardation and in addition to being an H3K4 demethylase (a “writer”); it also appears to bind to H3K9me3 residues (Jensen et al. 2005; Iwase et al. 2007).

One of the most informative assays for detecting histone modifications is chromatin immunoprecipitation (ChIP) in which an antibody specific for a particular histone modification is used to immunoprecipitate cross-linked chromatin. The DNA regions associated with the histone modification can then be analyzed using either microarray analysis (ChIP-chip) or high-throughput sequence analysis (ChIP-seq) (Park 2009). Many genome-wide histone modification datasets for a diversity of cell or tissue types have been generated, enabling one to look at similarities and differences in histone modifications across cell types (Bernstein et al. 2010; Myers et al. 2011; Ernst et al. 2011). These datasets are accessible through a variety of web links including http://www.roadmapepigenomics.org/data and http://www.ncbi.nlm.nih.gov/epigenomics. In addition, a few cell types have been mapped genome-wide for up to 24 different histone modifications to reveal the extent to which histone posttranslational modifications co-occur with one another and with other genomic features as well as to identify chromatin states associated with particular sets of modifications (Heintzman et al. 2009; Hawkins et al. 2010; Ernst and Kellis 2010). Unfortunately, ChIP-quality affinity reagents do not exist for many histone modifications, and so their genomic pattern and function remain mysterious.

Histone acetylation has been the most well-studied histone modification in the nervous system. Histone acetyl marks are covalently attached to histone tails via HATs which include a variety of structurally distinct enzymes including the well-characterized CREB-binding protein (CBP) which has important functions in the nervous system (Dekker and Haisma 2009; Hallam and Bourtchouladze 2006). The HDAC enzymes that can remove these modifications comprise three classes: Class I, Class II, and Class III (sirtuins) (Thiagalingam et al. 2003). Histone acetylation tends to be associated with actively expressed genes, while deacetylated regions tend to be associated with gene silencing (Thiagalingam et al. 2003).

In the nervous system, histone modifications are known to have important functions (Miller 2011; Akbarian and Huang 2009; Tsankova et al. 2007; Bredy et al. 2010; Morris et al. 2010; Haggarty and Tsai 2011). For example, disruption of the HAT enzyme CBP leads to memory defects (Korzus et al. 2004). Mutations in CBP are associated with Rubinstein-Taybi syndrome, which has an intellectual disability phenotype (Petrij et al. 1995). Histone deacetylases have been implicated in depression (HDAC5), regulation of dendritic spine density and memory formation (HDAC2), negative regulation of long-term memory formation (HDAC3), cognition (the Class III HDAC SIRT1), and synaptic transmission (Tsankova et al. 2006, 2007; Guan et al. 2009; McQuown et al. 2011; Gao et al. 2010; Morris et al. 2010).

Chromatin modifications, such as histone acetylation, mediate some of the neuronal and behavioral changes induced by cocaine. As can be seen in Table 7.2, histone acetylation has been the best studied class of histone modifications for addictive processes, although no doubt much more remains to be learned about its functions. Histone acetylation is particularly interesting from a translational point of view since certain medications that inhibit HDAC activity are clinically approved for treating seizure disorders and particular types of cancer (Sharma et al. 2010). The effects of these inhibitors on nervous system function and drug-taking behaviors are discussed in more detail in the Therapeutics section.

Table 7.2 Selected epigenetic changes associated with exposure to drugs of abuse

The role of histone acetylation and deacetylation in addictive processes in rodents has been recently reviewed; therefore, I will focus on a few of the key studies related to cocaine responses (Renthal and Nestler 2009b; McQuown and Wood 2010; Laplant and Nestler 2011; Wong et al. 2011). This will be followed by descriptions of some recent work on the role of the Class III HDACs (sirtuins) and then studies on tolerance to benzyl alcohol in a Drosophila model of inhalant exposure which helps delineate some of the detailed molecular events that may be occurring. I will then touch on several other histone modifications of particular interest including histone dimethylation, histone trimethylation, and histone phosphorylation.

3.1 Histone Acetylation and Cocaine Responses

Acute, but not chronic, cocaine exposure is known to induce expression of the mRNA that encodes the cFOS transcription factor. Using ChIP Dr. Eric Nestler and colleagues found that acute, but not chronic, cocaine exposure led to increased acetylation of histone H4 at the cFOS gene promoter, but had no significant effect on histone H3 acetylation (Kumar et al. 2005). Conversely chronic, but not acute, cocaine exposure induces the BDNF and Cdk5 genes. Chronic cocaine exposure led to increased acetylation of histone H3 but had no significant effect on histone H4 acetylation. These data suggest that a chromatin state (acetylation of histones H4 or H3 near the promoters of particular genes) may in part indicate which genes are modulated in response to acute or chronic cocaine administration. Taking these experiments one step further, the investigators showed that administration of the HDAC inhibitor trichostatin A to rodents prior to cocaine administration enhanced the reward response to cocaine, while overexpression of the HDAC4 gene in the striatum using herpes simplex virus decreased the reward response to cocaine.

In a second paper further exploring the roles of HDACs in drug responses, several HDACs were found to be expressed in the NAc, with HDAC3 and HDAC5 having the highest expression levels (Renthal et al. 2007). Viral overexpression of HDAC5 in the NAc led to a reduction in the rewarding properties of cocaine using a conditioned place preference assay, while HDAC5 knockout animals were found to have the converse phenotype, with increased preference for the cocaine-paired chamber. HDAC5 was required specifically in the NAc to regulate this behavioral response, since viral expression of HDAC5 in the NAc of HDAC5 knockout animals reduced the preference for the cocaine-paired chamber. Interestingly, the investigators also looked at the role of HDAC5 in chronic stress using a social defeat behavioral assay. While acute stress had no effect on HDAC5 levels in the NAc, chronic stress was associated with reduced levels of the HDAC5 mRNA, and HDAC5 knockouts exhibited hypersensitivity to chronic stress. HDAC5 modulation of behavioral responses to cocaine reward and chronic stress responses is quite significant given the important role of stress in drug abuse relapse.

3.2 Histone Acetylation, Sirtuins, and Cocaine Responses

In experiments described in the previous section, Dr. Renthal and coworkers characterized genome-wide levels of histone acetylation from the nucleus accumbens (NAc) brain region of rodents treated chronically with cocaine. These studies revealed that many genes previously known to be upregulated by cocaine exposure also have increased acetylation of histone H3 and H4. The investigators then looked genome-wide to identify the binding sites of the cocaine-induced transcription factors delta-FosB and CREB in the NAc of cocaine-treated animals (Renthal et al. 2009a). Cross comparison of these datasets identified many genes that had not previously been implicated in response to cocaine, including the Sirtuin genes (Sirt1 and Sirt2) which function as NAD-dependent histone deacetylases (Vaquero et al. 2007). While these genes function in many biological processes, including circadian and metabolic regulation and aging, their role in the nervous system is not well understood (Haigis and Sinclair 2010; Herranz and Serrano 2010). The investigators used pharmacological activators and inhibitors of sirtuins to look at their function in cocaine responses. Interestingly, systemic pharmacological activation of sirtuins dramatically enhanced the rewarding effect of cocaine, while inhibition of sirtuins had the opposite effect. Pharmacological modulation of sirtuin function may be a fruitful future avenue to explore in the development of therapeutic agents to treat cocaine addiction.

3.3 Histone Modifications and Inhalant Exposure in Drosophila

Dr. Nigel Atkinson and coworkers exploited the genetically powerful fruit fly model system to investigate the molecular basis of inhalant tolerance. It has been observed that adult flies become tolerant to sedation by organic solvents (which sometimes are abused as inhalants) and this reduced sensitivity to inhalant requires increased expression of the slowpoke potassium channel gene which in turn alters neuronal function (Wang et al. 2007a).

In order to investigate the molecular mechanism behind this observation, Dr. Atkinson and coworkers found that a single exposure to inhalant led to epigenetic changes in regulatory regions of the slowpoke gene leading to altered expression of the slowpoke gene and reduced sensitivity (tolerance) to additional inhalant exposure. Specifically, they observed that the pattern of acetylation of histone H4 was altered across the slowpoke gene, which likely led to a more open localized chromatin structure and subsequent increased expression of the slowpoke gene. Exposure of the animals to a pharmacological inhibitor of histone deacetylases, the class of enzymes responsible for the histone H4 acetylation, also led to similar epigenetic and gene expression change as well as tolerance of the animals to inhalant.

Interestingly, Dr. Atkinson and colleagues found DNA elements within the slowpoke promoter that could be bound by the CREB transcription factor. A number of labs have shown that the CREB transcription factor is important in the responses of organisms to illicit substances, as well as in other neuroplastic processes such as learning and memory (Han et al. 2007). Using a genetic manipulation to “turn off” CREB, the researchers found that the epigenetic and expression changes to slowpoke gene and the development of behavioral tolerance no longer occurred. These results indicated that the CREB transcription factor is required for these processes.

Dr. Atkinson and coworkers also found that sedation with benzyl alcohol leads to increased expression of positively acting CREB isoforms and reduced expression of negatively acting CREB isoforms (including dCREB2). Specifically the dCREB2 isoform shows increased occupancy at the slowpoke promoter immediately after benzyl alcohol sedation in a chromatin immunoprecipitation assay. Animals with a knockout in dCREB2 no longer have increased benzyl alcohol-induced slowpoke gene expression and also no longer develop tolerance to this organic solvent.

Overall, this work clearly shows that exposure to an organic solvent can alter future sensitivity to the solvent via epigenetic regulatory mechanisms. It also provides insight into the precise mechanisms by which exposure to an inhalant can lead to epigenetic and expression level changes for a single gene, resulting in altered neuronal function and altered behavioral responses of an animal to future inhalant exposure. Although this work was performed using model inhalant, similar mechanisms may well be utilized for responses to other drugs of abuse.

3.4 Histone Dimethylation (H3K9me2) and Cocaine Responses

Histone dimethylation (H3K9me2) is normally associated with gene silencing (Wen et al. 2009; Barski et al. 2007). In an article published in Science, Dr. Maze and colleagues observed that histone methylation levels are reduced in the nucleus accumbens (NAc) of rodents (Maze et al. 2010). To further explore this phenomenon, the researchers investigated the gene expression levels of the histone dimethyltransferases and demethylases that regulate this chromatin modification and found that levels of the G9a and GLP histone dimethyltransferases are downregulated upon cocaine administration. The investigators then looked at the effect of G9a manipulation in the NAc on the behavioral effects of cocaine and found that overexpression of G9a decreased the rewarding properties of cocaine, while knockdown of G9a increased the rewarding properties of cocaine. The researchers showed that these behavioral changes were correlated with concomitant changes in G9a levels and global histone dimethylation. Looking upstream of G9a, repeated cocaine exposure increased the levels of the transcription factor delta-FosB, leading to decreased G9a levels. Looking downstream, many of the genomic targets of histone dimethylation are known to play roles in the regulation of dendritic plasticity, and in fact dendritic spine density was shown to be altered by G9a levels. Overall Dr. Nestler and colleagues have elucidated an elegant multistep molecular pathway in which repeated cocaine exposure leads to delta-FosB activation, downregulation of G9a, and reduction in global histone dimethylation levels. Histone dimethylation is normally associated with gene silencing, so decreased histone dimethylation likely leads to increased expression of genes that regulate dendritic plasticity. This change in gene expression leads to increased dendritic spine density and ultimately increased behavioral preference for cocaine. Thus, small molecules that target the activities of histone demethylases or histone dimethyltransferases could have potential efficacy as therapeutic agents for treating cocaine addiction.

3.5 Histone Trimethylation (H3K9me3) and Cocaine Effects on Heterochromatin

Histone H3 Lysine 9 trimethylation (H3K9me3) is associated with silencing of heterochromatic regions of the genome (Schotta et al. 2004; Yamada et al. 2005). Work by Dr. Maze and colleagues found that cocaine exposure results in changes in H3K9me3 levels in the NAc but not in brain regions such as the caudate putamen or medial prefrontal cortex (Maze et al. 2011). When ChIP-seq assays were performed on the NAc from cocaine-treated animals, thousands of repetitive elements (e.g., LINEs, SINEs) were associated with increased H3K9me3 binding, while thousands of other sites had decreased binding. Overall, “repeated cocaine decreases H3K9me3 binding and un-silences several specific retrotransposons (e.g. LINE-1)” in the NAc (Maze et al. 2011).

3.6 Histone Trimethylation (H3K4me3) Changes Associated with Cocaine and Alcohol Exposure

Histone trimethylation (H3K4me3) tends to be associated with gene promoters (Guenther et al. 2007; Bernstein et al. 2005). Work by Drs. Goldman and Mash and coworkers have shown that histone trimethylation is important in chronic exposure to cocaine and to alcohol (Zhou et al. 2011a). Genome-wide ChIP-seq for H3K4me3 was performed on postmortem hippocampal samples from chronic cocaine addicts, alcoholics, and controls. The investigators found that chronic cocaine use was associated with H3K4me3 changes at >1,100 gene promoters, while chronic alcohol use was associated with changes at >700 promoters. The authors indicate that “there was significant overlap of the changes between the cocaine and alcohol exposure groups” (Zhou et al. 2011a). Interestingly, these H3K4me3 changes did not correlate well with gene expression changes measured in parallel, suggesting that perhaps additional chromatin or transcriptional regulation is important in mediating gene expression changes in response to cocaine and alcohol exposure.

3.7 Histone H3S10 Phosphorylation and Cocaine Responses

Phosphorylation of H3S10 has previously been shown to be important in chromatin condensation and transcriptional activation (Nowak and Corces 2004; Crosio et al. 2003). The DARPP-32 protein (dopamine-regulated and cyclic-AMP-regulated phosphoprotein) has been a well-characterized role in responses to cocaine and other drugs of abuse (Svenningsson et al. 2005). Interestingly, Stipanovich and colleagues have identified a regulatory cascade whereby DARPP-32 leads to altered Histone H3 phosphorylation (Stipanovich et al. 2008). Exposure to drugs of abuse via dopamine 1 receptor (D1R) regulation leads to the accumulation of DARPP-32 in the nuclei of D1R-expressing striatal neurons. This nuclear accumulation appears to be regulated by phosphorylation of Ser-97 of DARPP-32, such that when Ser-97 is unphosphorylated, DARPP-32 is primarily nuclear. The dephosphorylation of Ser-97 is mediated by protein phosphatase 2A. The researchers then looked at the effects on histone phosphorylation and found that cocaine exposure led to increased levels of H3S10 phosphorylation. This pathway reveals a mechanism by which drugs of abuse, via dopamine signaling, can influence chromatin and presumably impact gene expression.

4 DNA Modifications and Addictive Processes

Methylation of cytosine nucleotides (mC) has long been thought to be the only covalent mammalian DNA modification and is often referred to as the “fifth base” (Lister and Ecker 2009; Miranda and Jones 2007). DNA methyltransferases (DNMTs) are the enzymes that methylate DNA; this methylation frequently occurs at cytosines that occur as dinucleotides followed by guanine (CpG) (Turek-Plewa and Jagodzinski 2005). DNA methylation appears to be a dynamic rather than a static process, especially in the nervous system (Guo et al. 2011; Ma et al. 2009). The enzymes responsible for active DNA demethylation have been difficult to pin down despite significant efforts, and a number of candidate enzymes and mechanisms have emerged (Wu and Zhang 2010; Ooi and Bestor 2008). There have been recent discoveries addressing DNA methylation reversal mechanisms which involve conversion of mC to an intermediate molecular form which can then be excised by thymine DNA-glycosylase (He et al. 2011; Cortellino et al. 2011). DNA methylation has been traditionally associated with gene silencing, but the context of DNA methylation (methylation in CpG islands, CpG island shores, in promoters, or in gene bodies) appears to be important in mediating the functional effects of mC (Ndlovu et al. 2011). Interestingly, genome-wide single-base resolution DNA methylation maps reveal that human embryonic stem cells and reprogrammed induced pluripotent stem cells contain high levels of DNA methylation in a non-CpG context, although the function of non-CpG methylation is not clear (Lister et al. 2009, 2011). As more whole methylome datasets are generated for different cell/tissue types and this information is compared to histone modification, gene expression, and other data, the precise role of mC and the cross talk between DNA methylation and other regulatory mechanisms will become more clear.

Additional covalent DNA modifications were known to occur in other organisms such as plants and bacteriophage (Vanyushin 2006; Fleischman et al. 1976). However, in 2009 a “sixth base” was discovered in mammalian cells: hydromethylcytosine (hmC) (Kriaucionis and Heintz 2009; Tahiliani et al. 2009). Hydroxymethylcytosine was discovered in Purkinje cells in the cerebellum, and an independent paper showed that the TET1 enzyme can convert mC to hmC. Since then, researchers have been working to discover putative functions for hmC and have identified likely roles for hmC in transcriptional regulation and regulation of pluripotency (Pastor et al. 2011b; Ndlovu et al. 2011; Ficz et al. 2011; Wu and Zhang 2011; Stroud et al. 2011; Wossidlo et al. 2011; Szulwach et al. 2011). A very exciting publication indicates that TET proteins can catalyze the in vitro formation of 5-carboxylcytosine (caC) and 5-formylcytosine (fC) from mC (Ito et al. 2011). These new DNA modifications may be intermediates in a TET-mediated DNA demethylation pathway or perhaps could have unexpected regulatory properties of their own. Only time will tell whether these or other novel DNA modifications will be discovered in the genomes of nervous system cells.

As described earlier, specific proteins can bind to different histone modifications and may play an important role in mediating their effects. Similarly for DNA modifications, proteins in the MBD and Kaiso families have been shown to bind to mC, while a recent report indicates that the Uhrf1 protein can bind to hmC (Bogdanovic and Veenstra 2009; Frauer et al. 2011). Methyl CpG-binding protein-2 (MeCP2), a member of the MBD family discussed below, can bind to mC and presumably plays a role in mediating the effects of DNA methylation, perhaps by recruiting additional proteins or protein complexes to a particular region of chromatin.

There are a number of useful assays to determine DNA methylation status including MethylC-seq, which provides genome-wide DNA methylation information at single-base resolution (Lister et al. 2009; Harris et al. 2010). Unfortunately, bisulfite-based sequencing strategies do not distinguish between mC and hmC, but the development of new assays to detect and distinguish between hmC and mC should help us to elucidate the function of hmC in the nervous system (Pastor et al. 2011b). In fact a recently developed Tet-based bisulfite sequencing protocol (TAB-Seq) in combination with traditional bisulfite sequencing enables base resolution mapping of hmC and confirms widespread distribution of 5hmC in embryonic stem cells (Yu et al. 2012).

There is now a substantial body of work showing that DNA methylation has multiple roles in the nervous system, including a significant role in memory formation (Day and Sweatt 2010; Feng and Fan 2009). Mutations in the DNA methyltransferase DNMT1 lead to neurodegeneration (Klein et al. 2011), while the GABAergic neurons of some human schizophrenics show increased DNMT levels (Klein et al. 2011; Zhubi et al. 2009; Mill et al. 2008). In addition, the methyl CpG-binding protein-2 (MeCP2) is a transcriptional regulator originally identified as a protein that can bind to mC (Lewis et al. 1992). Work by Dr. Huda Zogbhi and colleagues have indicated that defects in MeCP2 function are associated with the neurodevelopmental disorder Rett syndrome (Amir et al. 1999).

As can be seen in Table 7.2, there have been a number of studies investigating the effects of drugs of abuse on DNA methylation. I will discuss the role of DNA methylation with respect to cocaine and nicotine exposure as well as the role of the mC-binding protein MeCP2 in addictive processes.

4.1 DNA Methylation and Cocaine Responses

Work by Dr. Laplant and coworkers in the Nestler group shows that both chronic cocaine exposure and chronic social defeat stress can lead to changes in the expression of the DNMT3a DNA methyltransferase gene. DNA methylation was found to be required for the cocaine-induced formation of thin dendritic spines in the NAc. To functionally test the role of DNMT3a in cocaine reward, DNMT3a was conditionally knocked down in the NAc, and the treated animals preferred the cocaine-paired chamber in a CPP assay. In the converse experiment, DNMT3a was overexpressed in the NAc using a herpes simplex virus vector, and the preference of the animal for the cocaine-paired chamber was reduced. NAc injection of the DNA methylation inhibitor RG108 led to increased cocaine CPP, while administration of methionine, which promotes DNA methylation, led to a decrease in cocaine CPP. The authors conclude from these pharmacological and genetic manipulations of DNMT3a that “increased Dnmt3a expression in NAc negatively regulates cocaine reward, whereas decreased Dnmt3a enhances cocaine reward” (Laplant et al. 2010).

4.2 DNA Methylation and Nicotine Responses

DNA methylation appears to also play a role in mediating responses to nicotine. Mice injected with nicotine had decreased levels of DNMT1 in the frontal cortex and the hippocampus, but had normal DNMT1 levels in the GABAergic neurons of the striatum (Satta et al. 2008a). Pharmacological experiments were used to show that nicotinic acetylcholine receptor (nAChR) function was required to achieve the change in DNMT levels in the frontal cortex. Further experiments revealed that nicotine exposure led to upregulation of GAD67 (glutamic acid decarboxylase 67) protein in the frontal cortex, but not in the striatum, and that this upregulation was associated decreased CpG methylation in the GAD67 promoter. Overall this study identifies a potential mechanism of action by which nicotine could mediate neuronal gene expression changes via DNA methylation. This study also suggests that therapeutic agents that modulate DNA methylation changes in the appropriate brain regions could be of potential use in treating nicotine addiction or perhaps schizophrenia.

4.3 A Role for MeCP2, a Methyl-C-Binding Protein, in Substance Abuse

MeCP2 is known to play a role in the neurodevelopmental disorder Rett syndrome; however, the neurobiological functions of MeCP2 are not completely understood and are under active investigation. MeCP2 can bind to methylated cytosine residues and presumably can recruit additional proteins or protein complexes to a particular region of chromatin to regulate transcription or other molecular processes. Some studies have suggested that MeCP2 regulates alternative splicing, can bind promoters that are not methylated, or may regulate neuronal genome function through histone acetylation in a more global fashion (Yasui et al. 2007; Young et al. 2005; Skene et al. 2010). Neuronal activity is known to stimulate CaMKII phosphorylation of MeCP2 Serine 421 to modulate “dendritic patterning, spine morphogenesis, and the activity-dependent induction of Bdnf transcription” (Zhou et al. 2006; Chen et al. 2003). Functional work indicates that disruption of MeCP2 specifically in GABA-releasing neurons leads to behavioral phenotypes, including compulsive grooming, reminiscent of phenotypes characteristic of Rett syndrome patients (Chao et al. 2010). MeCP2 is not just required during a specific developmental window but is required in adult animals for proper brain function (McGraw et al. 2011). Recent studies suggest that MeCP2 loss stimulates L1 transposition (Muotri et al. 2010).

Several publications point to a role for MeCP2 in regulation of responses to drugs of abuse. Work by Dr. West and colleagues has shown that MeCP2 knockdown in the NAc leads to increased preference of mice for amphetamine using a CPP assay, while animals no longer had preference for the amphetamine-paired chamber when MeCP2 was overexpressed in the NAc (Deng et al. 2010). Interestingly, in a strain of MeCP2 mutant mice, the authors found an almost twofold increase in GABAergic synapses in the NAc as compared to control animals, indicating that MeCP2 is required for the developmental wiring of this brain structure. Furthermore, MeCP2 mutant mice did not show normal amphetamine-induced changes in dendritic spine density of medium spiny neurons, nor did they show normal amphetamine-induced changes in immediate early gene expression in the striatum, both of which correlate with impaired amphetamine-induced behavioral changes in this strain.

In related work Dr. Kenny and colleagues showed that lentiviral knockdown of MeCP2 in the striatum leads to decreased cocaine consumption (Im et al. 2010). Interestingly, MeCP2 was found to repress expression of the miR-212 microRNA involved in regulation of cocaine-taking (discussed in more detail in the microRNA section of this chapter) (Im et al. 2010; Hollander et al. 2010). Furthermore, miR-212 can repress MeCP2 expression, while cross talk between miR-212 and MeCP2 regulates the impact of cocaine on brain-derived neurotrophic factor (Bdnf) levels in the striatum. Interestingly, Dr. Sadri-Vakili and colleagues found that cocaine self-administration was associated with decreased MeCP2 binding to brain-derived neurotrophic factor (Bdnf) promoter IV in the medial prefrontal cortex (mPFC) (Sadri-Vakili et al. 2010). The medial prefrontal cortex (mPFC) is one of several brain regions that has dopaminergic inputs from the ventral tegmental area (Le et al. 2005). Based on this study and the two publications above, it appears that MeCP2 has an important role in addictive processes and may have special functions in distinct brain regions.

Work by Teresa Reyes and colleagues has shown that in an animal model of diet-induced obesity, animals on a chronic high-fat diet were found to have epigenetic changes in the mu opioid receptor (MOR) promoter in reward areas of the brain (VTA, NAc, PFC). These epigenetic changes included increased H3K9me2, decreased H3ac, increased DNA methylation, and increased MeCP2 binding to the MOR promoter (Vucetic et al. 2011). Related work revealed that high-fat diet influenced dopamine reuptake transporter (DAT) gene expression in the VTA, NAc, and PFC (Vucetic et al. 2010). DNA methylation changes were also observed in DAT and other reward-related genes in the hypothalamus, NAc, and PFC. These molecular changes associated with exposure to high-fat diet suggest that epigenetic regulation in response to diet and to drugs of abuse could have significant overlap and raises the possibility of whether or not specific dietary regimens, in combination with other therapeutic interventions, could influence addictive processes.

4.4 MeCP2 and Epigenomic Regulation of Genomic Structure

As mentioned earlier, MeCP2 was found to play a role in regulation of transposition of L1 retrotransposons (Muotri et al. 2010). Dr. Muotri and colleagues found that there is increased L1 retrotransposition in neural precursor cells derived from patients with MeCP2 mutations. Additionally in a mouse strain designed to detect L1 transposition events using a fluorescent reporter, there were higher numbers of GFP-positive cells in brain tissue from MeCP2 knockout animals as compared to wild type. The results from this paper suggest a possible scenario in which L1 retrotransposons may be methylated and bound by MeCP2, inhibiting retrotransposition. In the absence of MeCP2, retrotransposition becomes more frequent. This study raises the intriguing possibility that our somatic genomes may be much more diverse than we previously expected and that epigenomic regulation may play an important role in regulating somatic genomic diversity.

Could drugs of abuse impact somatic genomic structure via epigenomic regulation? Work by Dr. Maze and colleagues described earlier showed that cocaine exposure leads to changes in Histone H3 Lysine 9 trimethylation (H3K9me3) in the NAc but not two other brain regions (Maze et al. 2011). H3K9me3 is associated with heterochromatin silencing (Schotta et al. 2004; Yamada et al. 2005). The authors indicate that overall “repeated cocaine decreases H3K9me3 binding and unsilences several specific retrotransposons (e.g. LINE-1)” in the NAc (Maze et al. 2011).

LINE-1 retrotransposition during neurodevelopment and neurogenesis could contribute to genomic diversity within the somatic genomes of neurons (Singer et al. 2010). Taken together, the effects of MeCP2 mutation on LINE-1 retrotransposition and the H3K9me3 work suggest the testable hypothesis that repeated cocaine administration could lead to unsilencing and retrotransposition of LINE-1 elements in the NAc leading, in some cases, to permanent gene disruption or long-lasting alterations in gene expression (Muotri et al. 2010). Obviously animal and postmortem brain studies of cocaine-exposed individuals would be needed to assess whether cocaine exposure induces LINE-1 retrotransposition in the NAc and, if so, what the functional consequences of retrotransposition might be.

5 Noncoding RNAs and Addictive Processes

Many large and small noncoding RNAs (ncRNAs) have been identified in recent years, and some of these have significant regulatory functions. In particular, small ncRNAs (approximately 20–30 nucleotides) play key roles in gene transcription and translation. For example, Piwi-interacting RNAs (piRNAs) are involved in transposon silencing, and small interfering RNAs (siRNAs) are involved in regulating mRNA levels and chromatin formation (in plants and yeast) and in antiviral responses (in animals), while microRNAs (miRNAs) can simultaneously regulate the mRNA levels and translational efficacy for tens to hundreds of genes (Kaikkonen et al. 2011; Hannon et al. 2006; Czech and Hannon 2011; Marques and Carthew 2007). In the nervous system, a recent paper shows that piRNAs and the piRNA-associated protein MIWI are found in the hippocampus and may play a role in the morphogenesis of spines (Lee et al. 2011). miRNAs have been shown to play diverse roles in processes including neural development, survival, and degeneration, synaptic plasticity, dendritic spine morphology, and memory formation (Olde Loohuis et al. 2011; Saba and Schratt 2010; Davis et al. 2008; Schratt 2009; Schaefer et al. 2007; Schratt et al. 2006; St. Laurent et al. 2009; pp. 81–88; Lin et al. 2011). In mammalian cells miRNAs have important cytoplasmic functions, although as yet they have not been shown as yet to have epigenetic regulatory function (Khraiwesh et al. 2010). Some miRNAs have a profound effect on addictive behaviors and for this reason have been included in this chapter.

In addition to small RNAs, some of the longer ncRNA species include enhancer RNAs (eRNAs) and large intergenic noncoding RNAs (lincRNAs). eRNAs are associated with enhancer regions of the genome that likely play a role in transcriptional regulation (Wang et al. 2011). Recent work utilizing a specific chromatin signature identified more than 1,000 mammalian lincRNAs, and some lincRNAs have been shown to be involved in regulation of cellular differentiation (Guttman et al. 2009; Guttman et al. 2011). The lincRNA HOTAIR seems to be able to impact chromatin remodeling by binding to multiple enzymes which are able to modify histones, while the lincRNA HOTTIP seems to be able to “transmit information from higher order chromosomal looping into chromatin modifications” in order to regulate gene expression (Tsai et al. 2010; Wang et al. 2011). Some lincRNAs form complexes with proteins such as polycomb repressive complex 2 (PRC2), and the development of assays such as RIP-seq has allowed the identification of RNAs associated with PRC2 (Zhao et al. 2010). Assays enabling the identification of genomic regions associated with lincRNAs will be an important future tool needed to help uncover regulatory cross talk that may occur between lincRNAs and other epigenetic regulatory mechanisms. Some lincRNAs are expressed in brain and may play a role in the specification of glial and neuronal fates (Mercer et al. 2008, 2010). For example, the RCNR2 RNA has been show to play a role in specification of retinal cell fate (Rapicavoli et al. 2010). Future research is likely to uncover additional roles for lincRNAs in the nervous system.

As shown in Table 7.2, investigations into the role of noncoding RNA regulation in addictive processes have been quite limited to date, and only one study has investigated any role for lincRNAs. This section of this chapter will focus on an emerging role for lincRNAs in heroin abuse followed by a more complete story on microRNA regulation in cocaine-taking behavior.

5.1 lincRNAs and Heroin Use

Drs. Bannon and Lipovich and colleagues found that 23 lincRNAs were represented on microarrays being used to characterize gene expression changes in postmortem NAc tissue from heroin and non-heroin users (Michelhaugh et al. 2011). These investigators found that five of these lincRNAs (MIAT, MEG3, NEAT1, NEAT2, and EMX2OS) were expressed in the NAc and further that all five were upregulated in the heroin users as compared to non-heroin users. Thus, a potential lincRNA function may be to regulate widespread gene expression, and this regulation may be disrupted in the NAc of drug-abusing individuals. Further exploration of the role of lincRNAs in neuroplastic and addictive processes is an important area of investigation for the future.

5.2 miRNAs and Cocaine

The role of miRNAs in addictive processes has recently been reviewed (Li and van der Vaart 2011; Pietrzykowski 2010). As indicated in Table 7.2, roles have been described for microRNA regulation of the dopamine 1 receptor involved in nicotine dependence and for microRNA regulation of response to alcohol exposure (Huang and Li 2009; Pietrzykowski et al. 2008). There have been several projects that have successfully identified miRNAs involved in cocaine responses (Eipper-Mains et al. 2011; Chandrasekar and Dreyer 2011). Another study shows that rodents with an Argonaute 2 (Ago2) protein deficiency in the dopamine receptor 2 expressing neurons have altered intravenous cocaine self-administration in mice (Schaefer et al. 2010). Ago2 is known to play an important role in miRNA biogenesis and function, supporting a role for miRNAs in cocaine reinforcement (O’Carroll et al. 2007).

A major discovery in this area has been made regarding miRNA regulation of cocaine-taking behavior. Dr. Paul Kenny and coworkers have identified a 21-nucleotide microRNA, miR-212, that is found at higher levels in the dorsal striatal brain region of animals that self-administer cocaine (Hollander et al. 2010). In rats with extended access to cocaine, reduction of miR-212 levels in the striatum leads to increased cocaine intake, while overexpression of miR-212 leads to decreased cocaine intake. Further molecular experiments revealed that miR-212 achieves its effects via simultaneous reduction in expression of several mRNAs encoding regulatory proteins that impinge upon the Raf1 protein kinase signaling pathway (including the SPRED1 repressor of Raf1). Overall these gene expression changes lead to increased levels of Raf1 protein kinase activity, increased expression of the CREB regulatory protein TORC, and ultimately increased activity of the transcription factor CREB. In a separate publication, miR-212 was found to regulate and be regulated by MeCP2 (see MeCP2 section of this chapter for details) (Im et al. 2010).

The identification of novel miRNA regulatory pathways that control cocaine intake, as well as responses to other drugs of abuse, could reveal new and unexpected targets for the development of potential therapeutic agents to treat addiction. Two key areas for future research in the miRNA arena are the identification and characterization of the target mRNAs regulated by these miRNAs as well as studies investigating whether or not these and/or other miRNAs are associated with addictive behaviors in human populations.

6 The Perdurance of Epigenomic Changes

While some chromatin changes are transient and occur as a normal part of transcriptional regulation, others are more long lasting and could be particularly important in the case of terminally differentiated post-mitotic neurons (Miller et al. 2010). There is also evidence for mitotically heritable chromatin changes that may impact progeny cells (Ng and Gurdon 2008). In some cases, epigenomic changes may even be meiotically heritable and affect the next generation (Youngson and Whitelaw 2008). The occurrence and perdurance of some types of epigenetic changes is likely to be influenced by factors including the nature of any environmental exposures involved, the cell type involved, and whether or not that cell type is exposed during a particular developmental window. There is increasing evidence that certain environmental exposures during critical developmental periods such as prenatal development, adolescence, or periods of germline maturation are associated with disease consequences. This concept has been captured in the developmental origins of health and disease (DOHaD) hypothesis which “proposes that during critical periods of prenatal and postnatal mammalian development, nutrition and other environmental stimuli influence developmental pathways and thereby induce permanent changes in metabolism and chronic disease susceptibility” (Waterland and Michels 2007). In the following two sections, we will discuss the limited literature concerning the stability of epigenomic changes associated with exposure to drugs of abuse during particular developmental periods.

6.1 Somatic Effects of Drugs of Abuse

The stability of epigenomic changes is a particularly difficult problem to investigate in the brain. For animal studies, individuals need to be sacrificed, and thus, only a single time point can be generated per animal. Surprisingly few studies have been performed looking at how long epigenomic changes can last within the nervous system. Work by Dr. Courtney Miller and colleagues indicates that fear conditioning-associated DNA methylation changes in the Calcineurin gene can last for at least 30 days in the dorsal medial PFC (Miller et al. 2010). Researchers in the Nestler laboratory found H3K9me2 changes in the nucleus accumbens that last for at least 28 days using a social defeat behavioral paradigm (Wilkinson et al. 2009). Investigations into the stability and dynamics of somatic epigenomic changes in response to drugs of abuse and potential drug abuse therapeutics will be important for the future.

6.2 Multigenerational and Transgenerational Effects of Drugs of Abuse

There is evidence that exposure to certain chemical toxins, social environments, or nutrient levels can, in some cases, lead to specific phenotypes in subsequent generations. These phenotypes can be transmitted without an apparent DNA change through multiple generations even when these generations have not been exposed to the inducing factor (Youngson and Whitelaw 2008; Skinner et al. 2010; Champagne 2008). The phenotypic consequences can in part be dependent upon when the exposure occurred. When an individual encounters an environmental exposure such as a drug of abuse, the exposure could potentially impact the individual, any fetuses present, and any germ cells or gametes present. Effects on progeny derived from the exposed parent can be referred to as multigenerational phenotypic effects, while transgenerational effects usually refer to phenotypes observed in progeny that were not exposed in utero or derived from exposed germ cells (Skinner et al. 2010).

For example, there is evidence that caloric restriction can lead to impaired glucose tolerance in subsequent generations (Zambrano et al. 2005). Exposure to certain chemical toxins or social environments can also impact phenotypes across generations (Skinner et al. 2008; Champagne and Meaney 2007). Several groups of researchers have shown that early life stress or adversity can lead to a number of important phenotypic effects including altered transcription factor binding to and histone acetylation of the glucocorticoid receptor (which plays a critical role in stress responses), altered DNA methylation of the BDNF gene in the adult prefrontal cortex, altered serotonin signaling in the dorsal raphe nucleus, depressive-like phenotypes, and changes to energy metabolism and feeding behavior (Weaver et al. 2004; Roth et al. 2009; Franklin et al. 2010; Dietz et al. 2011; Pankevich et al. 2009; Weiss et al. 2011). The mechanism for phenotypic transmission in all of these cases has not been fully elucidated, although the potential involvement of epigenetic processes is an important avenue of investigation.

There are several lines of evidence suggesting that it would be worthwhile to explore whether or not exposure to drugs of abuse could lead to multigenerational or transgenerational effects. Human epidemiological studies by Dr. Marcus Pembrey and colleagues indicate that preadolescent paternal smoking is associated with greater body mass indices in sons (at 9 years of age), with no significant effects observed in daughters (Pembrey et al. 2006). There is also preliminary evidence that paternal cocaine exposure in mice could lead to phenotypic consequences in the progeny. In this study, mice were exposed to cocaine via inhalation in an animal model of crack cocaine use (He et al. 2006). Cocaine-exposed males were mated to unexposed females and the progeny characterized morphologically and behaviorally. The progeny had reduced head diameters, perhaps reflecting reduced cerebral volume, and also had impaired spatial working memory and attention. Interestingly, the investigators did not observe significant DNA breaks in the sperm of cocaine-exposed fathers, but did observe “reduced levels of expression of Dnmt1, together with increased levels of expression of Dnmt-3a, in the germ cell-rich seminiferous tubular tissue” of cocaine-exposed males which may mean that cocaine exposure could impact progeny phenotypes through DNA methylation changes in the sperm (He et al. 2006). Related work indicates that cocaine exposure of pregnant female mice can lead to altered DNA methylation patterns in the hippocampus of progeny (Novikova et al. 2008).

In 1972, an article published in Science indicated that maternal morphine exposure prior to fertilization was associated with a decrease in body weight in the progeny (Friedler and Cochin 1972; Friedler 1978). Further studies indicated that male adolescent morphine exposure could impact endocrine responses in offspring (Cicero et al. 1991). More recent work by Dr. Elizabeth Byrnes and colleagues indicates that maternal morphine exposure prior to conception can lead to phenotypic effects on the progeny (Byrnes et al. 2011; Byrnes 2005; Johnson et al. 2011). Female rats were exposed to multiple doses of morphine during adolescent development, drug exposure was halted for at least 20 days, and the animals were mated to males not exposed to morphine (Byrnes et al. 2011). The adult female, but not male, progeny from morphine-exposed mothers had decreased anxiety as measured by the open-field assay. However, adult male, but not female, offspring of morphine-exposed mothers had increased sensitivity to morphine. The potential phenotypic consequences of adolescent or young adult morphine exposure on the next generation are particularly significant given the recent sharp increase in prescription opioid use among adolescents (Sung et al. 2005).

Genomic imprinting is an “epigenetically regulated process that causes genes to be expressed in a parental-origin-specific manner rather than from both chromosome homologues” (Ferguson-Smith 2011). In at least one case, the parental origin of a single genetic variant has been associated with disease protection and disease risk, depending upon whether the variant came from the mother or father (Kong et al. 2009). Imprinting could play a very interesting role in nervous system function and is known to be important in certain neurodevelopmental disorders such as Prader-Willi syndrome (Allen et al. 1995; Gregg et al. 2010; Gurrieri and Accadia 2009). There has been some work suggesting a role for imprinting in alcohol dependence, and it has been reported that the imprinting control region H19-IGF2 may play a role in specification of dopaminergic precursor neurons (Strauch and Baur 2005; Freed et al. 2008). However, overall, little is known about the role of parental imprinting in drug abuse resilience or susceptibility.

It is clear from this section that additional research, including extremely well-controlled studies, need to be carried out in order to prove unequivocally whether or not any drugs of abuse have authentic transgenerational effects. If substance use/abuse were conclusively shown to lead to deleterious transgenerational phenotypic effects, this new knowledge would have significant public health implications and would likely influence the development of future drug prevention programs. If exposure to drugs of abuse were shown to lead to transgenerational phenotypes, then the mechanism for such transmission would need to be elucidated. There have been a number of mechanisms proposed for transmission of transgenerational effects including (1) viral, microbiome, or prion transmission, (2) neurobehavioral or societal transmission, and (3) altered epigenomic states of germ cells or gametes via altered parental imprinting, or other epigenetic effects (Youngson and Whitelaw 2008).

7 Challenges and Opportunities in Epigenomics and Addiction Research

As research proceeds in the area of transcriptional and epigenomic regulation in human disease, there are a number of scientific challenges and opportunities that present themselves. These include investigations into less well-studied chromatin features, renewable affinity reagents, addressing cell-type heterogeneity in the nervous system, epigenomic maps of brain tissues or cell types, epigenome-wide association studies (EWAS), and manipulation of epigenetic changes to understand function and mechanism. These challenges and opportunities are discussed in the following section with an emphasis on their impact on neuroscience and addiction research.

7.1 Underexplored Areas: From Novel Modifications to Higher Order Chromatin Structure

There are a number of research areas that have the potential to be quite exciting but have received very limited attention with respect to neuroplasticity and drugs of abuse. Several of these are discussed below including new DNA, histone, and RNA modifications; histone variants; RNA editing; ATP-dependent chromatin remodeling; and higher order chromatin structure.

Histone posttranslational modifications and DNA modifications have both been shown to be important in regulation of gene function, and it is likely that our catalog of these modifications is not complete. Histone variants are known to be important in a number of biological processes including regulation of transcription; however, little is known about whether they have any special roles in post-mitotic neurons (Talbert and Henikoff 2010). RNAs have a surprising number of posttranscriptional modifications (e.g., 6-methyladenine, 1-methyladenine), but the functional roles of these modifications have not been carefully explored in the nervous system (Cantara et al. 2011; He 2010).

Some mRNAs are modified through the process of A-I RNA editing in which an adenosine in the RNA is converted to inosine by adenosine deaminases (ADARs). The inosine can be translated as a guanosine by the ribosome which can result in the presence of an amino acid in the protein product that was not encoded by the original DNA sequence (Mattick and Mehler 2008). RNA editing may serve to increase the diversity of proteins that can be produced, but it could also enable neurons to modify their properties in response to particular environmental changes. The extent to which RNA editing occurs in noncoding RNAs is poorly characterized but could impact their regulatory functions. mRNAs that have been shown to be edited include the serotonin biosynthetic enzyme TPH2 and the serotonin 2C receptor mRNA (Grohmann et al. 2010; Iwamoto et al. 2009; Dracheva et al. 2008). Little work has been done looking at the role of RNA editing in addictive processes, although Dr. Stella Dracheva and colleagues have found higher serotonin 2C editing in the prefrontal cortex associated with rats that exhibit high locomotor response to novelty (Dracheva et al. 2009).

The ATP-dependent chromatin remodeling proteins, such as members of the SNF2, ISWI, or CHD families, are able to “disrupt nucleosome DNA contacts, move nucleosomes along DNA, and remove or exchange nucleosomes” (Hargreaves and Crabtree 2011; Gkikopoulos et al. 2011). These nucleosome changes regulate access to genomic DNA which can have consequences in terms of gene expression. Some of these chromatin remodeling proteins have been shown to function in neural development and differentiation (Yoo and Crabtree 2009; Brown et al. 2007; Pirotte et al. 2010). When neurons exit from mitosis, there is a switch in the subunit composition of the BAF chromatin remodeling complexes that appears to be important in regulating dendritic morphogenesis, and this switch is regulated by the microRNAs miR-9* and miR-124 (Yoo et al. 2009). For neuroplasticity and substance abuse, investigations into chromatin remodeling has been limited; however, it has been reported that increased levels of the ATP-dependent chromatin remodeling protein Brg1 are found at the Cdk5 promoter in response to chronic cocaine exposure (Kumar et al. 2005).

Higher order chromatin structure within the nucleus may play an extremely important role in regulation of gene expression and in mediating other cellular functions (Eskiw et al. 2010; Li and Reinberg 2011). Technologies have recently been developed that enable the characterization of higher order chromatin (e.g., Hi-C, ChIA-PET) (Lieberman-Aiden et al. 2009; Fullwood et al. 2009; Espinoza and Ren 2011). Studies investigating the role of higher order chromatin structure in the nervous system, or in response to neuroplastic changes or drugs of abuse, are a very interesting area for future investigation.

7.2 Renewable Affinity Reagents for Epigenomic Research

Chromatin immunoprecipitation (ChIP) assays can provide extremely valuable information concerning the chromatin landscape of particular cells and tissues. These assays have only become possible due to the development of very low-cost and very high-throughput sequencing (Zhang and Pugh 2011; Park 2009). The continued reduction in the cost of sequencing DNA will improve the ability of researchers to apply this technique to important biological processes and disease studies. However, one of the key needs for successful ChIP assays is a high-quality antibody or other affinity reagent that binds specifically to the target of interest enabling chromatin enrichment of DNA regions in the vicinity of that particular epitope. Unfortunately some commercially available antibodies do not have appropriate specificity or are not useful for ChIP assays (Egelhofer et al. 2011). Efforts have been made to begin validating commercially available antibodies through Western blot, dot blot, or ChIP-seq analyses (http://compbio.med.harvard.edu/antibodies/). Even if an antibody is found to be useful for ChIP assays, the supplies are finite unless the antibody is monoclonal. Thus, there is a great need to develop a renewable resource of ChIP-grade antibodies (such as monoclonal antibodies) or affinity reagents (using recombinant affinity technologies) so that the scientific community has an unlimited supply of these reagents. A ChIP affinity reagent resource would allow researchers to compare ChIP experiments performed in different labs using identical antibodies, which is not always possible when the ChIP assays are performed with polyclonal antibodies. The development of a panel of one or more renewable ChIP-grade affinity reagents for each posttranslational histone modification, DNA modification, and ultimately each transcriptional regulatory protein would be an extremely valuable resource for the scientific community as a whole.

7.3 Addressing Cell-Type Heterogeneity in the Nervous System

One of the major barriers impeding epigenetic studies in the nervous system, as well as other organ systems, is cellular heterogeneity. The mixture of neurons, glia, microglia, and cardiovascular tissue in different brain regions may mask or confound epigenomic changes that may be taking place. One strategy to address this problem includes the sorting of labeled nuclei from specific brain regions or other clinically relevant tissue to enrich for cell types of interest while preserving the relevant epigenomic information (Cheung et al. 2010). Genetically tractable systems have been used to label ribosomes from specific cell types for purification and molecular identification of cell-specific mRNAs (Heiman et al. 2008; Doyle et al. 2008). In recent years, our ability to epigenomically characterize smaller and smaller numbers of cells has improved significantly, and in the future it might even be possible to assay the epigenomes of single cells (Gu et al. 2010; Goren et al. 2010; Geng et al. 2011; Cipriany et al. 2010).

7.4 Epigenomic Maps of Brain Cell/Tissue Types

Scientific consortia such as the NIH Roadmap Epigenomics program are generating comprehensive maps of chromatin from a wide variety of “normal” human cell and tissue types (Bernstein et al. 2010). These maps typically include DNA methylation information, ChIP assays for several highly informative histone modifications (H3K4me1, H3K4me3, H3K9me3, H3K27me3, H3K36me3, H3K9ac, or H3K27ac), chromatin accessibility information using the DNAse1 hypersensitivity assay, and gene expression information. At the moment there are 65 epigenomic maps of cells that have all of these data types, while partial datasets are available for around 180 additional cell or tissue types. For brain researchers there are currently partial datasets for fetal brain from six time points between 85 and 142 days, as well as postmortem adult brain from anterior caudate, cingulate gyrus, hippocampus, inferior temporal lobe, midfrontal lobe, and substantia nigra (http://www.roadmapepigenomics.org/). Other epigenomic datasets for human and model organisms cells/tissues can be found at the NCBI Epigenomics gateway (http://www.ncbi.nlm.nih.gov/epigenomics) or produced by the ENCODE consortium (http://www.genome.gov/10005107).

One goal for the future would be to develop a comprehensive atlas of chromatin maps for a wider variety of brain regions and brain-specific cell types for both human and mouse. It will be important to link these epigenomic features with other molecular phenotypes such as mRNA and ncRNA expression, transcription factor binding sites, and higher order chromatin structure information. It will also be important to link molecular phenotypes to other cellular phenotypes such as morphology, connections with other neuronal or support cells, and the electrophysiological properties of the cells. These maps would be an important aid to researchers studying neuropsychiatric, neurodevelopmental, and cognitive disorders and may also yield neuronal cell type-specific targets for developing small molecular probes and therapeutic compounds.

For drug abuse researchers, the systematic generation of an “addiction epigenomics” data resource cataloging molecular phenotypes for drug abuse relevant brain regions with and without exposure to different drugs of abuse would provide an invaluable resource. Researchers would be able to mine this data to identify novel candidate loci to test for their involvement in addictive processes. They would also be able to compare profiles of molecular phenotypes responses for different drugs of abuse to begin to identify loci and networks common to addictive processes in general as well as those that might be unique for a particular drug of abuse.

Genome-wide datasets for DNA methylation, histone modifications, chromatin accessibility, ncRNAs, and transcription factor binding sites can be harnessed to interpret data from genome-wide association studies (GWAS) for various diseases. Of particular interest, recent investigations reveal that disease SNPs identified by GWAS frequently fall in regions of accessible chromatin or in enhancer elements of cell types relevant to the disease (Maurano et al. 2012; Pennisi 2011; Ernst et al. 2011). Overall these studies “can facilitate the interpretation of GWAS data sets by predicting specific cell types and regulators related to specific diseases and phenotypes” (Ernst et al. 2011). Application of this strategy to polymorphisms associated with addictive behaviors could help to shed light on the function of some of these SNPs, particularly those in noncoding genomic regions, help flesh out the gene networks involved, as well as to confirm or generate new hypotheses concerning the brain regions and cell types involved in addiction to particular drugs of abuse.

7.5 Epigenome-Wide Association Studies (EWAS) for Drug Abuse Research

Although GWAS have been valuable in identifying unexpected genes and loci involved in particular human diseases, some diseases have a significant environmental component. If certain epigenomic states are indeed influenced by environmental exposures, then EWAS, which look at the epigenomic states of disease-relevant tissues in a case/control design, could be of great value in identifying loci involved in particular environmental exposures (Rakyan et al. 2011). Identification of genes and loci using EWAS approach could point the way to new therapeutic targets to treat disease. As scientists begin to perform EWAS using readily accessible human tissues, it will be interesting to see how the genes and loci identified compare and contrast with those identified in GWAS. At this time, EWAS for an addiction phenotype would be difficult to implement since this would necessitate the ability to monitor brain epigenomic regulation either through in vivo imaging techniques or through the use of an accessible surrogate tissue type that accurately reflects epigenomic changes that occur within the relevant brain regions.

7.6 Pharmacological and Molecular Manipulation of Epigenetic Changes to Understand Function

As correlative hypotheses are generated, it becomes essential to determine the functional role of a particular epigenetic or ncRNA regulatory pathway on a phenotype. Small-molecule probes that activate or inhibit specific epigenetic regulators provide an invaluable resource for testing the function of specific regulation for substance abuse phenotypes. As interest in epigenomic and ncRNA regulation unfolds, more small-molecule modulators are being made available to the scientific community. The NIH Molecular Libraries Program has several projects to identify epigenetic modulators underway (http://mli.nih.gov/mli/) as does the Structural Genomics Consortium (http://www.thesgc.org/chemical_probes/epigenetics/) (Austin et al. 2004). These small-molecule modulators could be used as probes to confirm whether or not a given pathway is functionally important and should be further investigated. These small molecules could also serve as the foundation for developing therapeutic agents. As small-molecule reagents become more readily available, it will become easier to determine which epigenetic regulatory pathways have the most impact on drug abuse phenotypes and might be useful to target therapeutically.

Molecular genetic approaches can also be useful for investigating the function of epigenomic and noncoding RNA regulation. Unfortunately most small molecules and molecular genetic manipulations impact epigenetic function on a global level. Efforts have been made to manipulate chromatin states in a locus-specific manner, typically by using fusion proteins to target epigenetic modifying enzymes to particular DNA loci (Hansen et al. 2008). These techniques will need further development to enable robust locus-specific manipulation of chromatin states in the future.

8 Translating Epigenomic Discoveries into Improvements in Human Health

Although a deeper understanding of the biological mechanisms of drug abuse is of great significance, this understanding has the potential to be translated into improvements in human health. Most substance abuse studies to date have investigated epigenomic regulation in the brain regions of rodents since the level of drug exposure can be readily quantitated and tissues from the exposed brains are accessible to the investigator. However, it will be important to pursue epigenomic studies on postmortem brain samples from substance users and abusers to begin to determine the extent to which the elegant discoveries in rodents are recapitulated in humans. In addition to these types of studies, some of the fundamental discoveries that have been made in addiction epigenomics could have future impact on addiction diagnosis, prevention, and therapy.

8.1 Future Substance Use Disorder Diagnostics?

Epigenetic changes have been identified that could serve as potentially useful cancer biomarkers or diagnostics. For example, promoter methylation of a panel of genes may be useful for early detection of colorectal cancer, a DNA methylation phenotype has been used to identify a glioma subgroup, DNA methylation of the promoter of the MGMT DNA repair enzyme can be used as a biomarker to predict glioblastoma chemotherapy outcome, and promoter methylation of the IGFBP-3 growth factor binding protein may predict cisplatin chemotherapy outcome in non-small lung cancer (Lind et al. 2011; Noushmehr et al. 2010; Weller et al. 2010; Ibanez et al. 2010). In the realm of brain disorders, DNA methylation information can be useful for predicting the efficacy of treatment of Fragile X using an mGluR inhibitor, suggesting the potential for epigenomically informed personalized medicine (Jacquemont et al. 2011).

A significant barrier to developing diagnostic tools for substance abuse based on epigenomic changes is our current inability to assess the epigenomic state of tissue types within the human brain. Unlike genomic studies which can readily be carried out using blood samples, different cell types within the brain express different suites of genes and are thus expected to have epigenomes that differ from one another (Doyle et al. 2008). Thus, to study epigenomic dysregulation in disease, one would ideally investigate the cell or tissue type of most relevant to the disease. In the case of substance abuse, epigenomic studies would thus focus on postmortem human or animal brain tissue. There has been speculation that epigenomic changes in human samples such as specific blood cell types, olfactory neurons, or other more accessible tissues could serve as a surrogate for epigenomic changes in particular brain regions, but to date there has been little compelling evidence that surrogate tissues are of significant utility in studying epigenomic processes in psychiatric diseases.

The ability to image epigenomic processes or changes within the nervous system in a noninvasive manner would be a major technological advance that could help bring epigenomic studies of substance abuse and other psychiatric diseases into living humans. One could imagine using this technology in future clinical settings for diagnosis, monitoring of disease progression, or monitoring of therapeutic efficacy. To date, only limited efforts have been made to image epigenomic processes in vivo. As a good first step, Dr. Joanna Fowler and colleagues have generated reagents to visualize the levels of histone deacetylases in vivo using positron emission tomography (PET) (Hooker et al. 2010; Reid et al. 2009). These and related strategies could eventually be used to image gross changes in the levels and/or activity of epigenetic modifying enzymes relevant to substance abuse and other diseases in vivo.

Measurement of epigenomic regulatory changes in brain using in vivo imaging techniques, or perhaps through assay of more accessible tissues that serve as a surrogate for the brain, might one day be used to help predict susceptibility to substance use disorder, to diagnose disease progression, or perhaps to provide biomarkers that accurately reflect levels and duration of chronic drug use. Future development of substance use disorder diagnostics will require us to more fully understand what epigenomic changes truly mean with respect to addictive processes as well as how long these changes persist.

8.2 Preventing Substance Use Disorder?

As mentioned earlier, there is evidence that exposure to certain chemical toxins, social environments, or nutrient levels can occasionally lead to organismal phenotypes in subsequent generations. Whether or not this phenomenon is also true for any drugs of abuse remains unclear. However, if particular drugs of abuse were shown to have phenotypic effects on subsequent generations, then this scientific information could be used to strengthen public health messages documenting the known adverse health consequences of drug abuse for public dissemination to facilitate scientifically informed choices on the use of licit and illicit drugs.

8.3 New Therapeutics for Substance Use Disorder?

Epigenetic changes are fundamentally more plastic than genetic changes and thus appear to be more amenable to therapeutic intervention (Haberland et al. 2009). Epigenetic therapeutics have shown great potential in cancer and other diseases. For example, DNA methyltransferase inhibitors have been approved by the FDA to treat myelodysplastic syndromes and may be useful for treating certain leukemias (Sharma et al. 2010). There are also HDAC inhibitors such as SAHA that have been approved to treat T-cell lymphoma (Sharma et al. 2010). Other HDAC inhibitors have been previously approved to treat urea cycle disorders, while the HDAC inhibitor valproic acid has been used to treat seizures, migraines, and bipolar disorder (Mack 2006; Bialer and Yagen 2007). HDAC inhibitors have also shown very promising effects in certain animal models of neurodegeneration, depression, and cognitive disorders (Fischer et al. 2010).

There is interest in testing FDA approved compounds for efficacy in treatment of a wider variety of diseases. As one example, clinicians have been investigating whether HDAC inhibitors can be used to activate latent HIV within the genome making the cells susceptible to antiretroviral therapy (Margolis 2011). If successful for all the tissue reservoirs that contain the latent virus, this strategy could point the way to a possible cure for HIV/AIDS. In another very exciting example, the HDAC inhibitor SAHA was used to successfully treat a patient with seizure disorder likely due to a genetic mutation, suggesting that in some cases epigenetic therapies may have the potential to treat genetic diseases (Almeida et al. 2007).

Scientists are also developing new compounds that impact epigenetic targets other than HDACs and DNMTs, such as histone methyltransferases and histone demethylases (Grant 2009; Hamada et al. 2010; Fiskus et al. 2009). There have even been efforts to target proteins that bind to histone modifications. For example, a molecule that can inhibit the BRD4 protein, which can bind to acetylated lysines, has potential for treating acute myeloid leukemia (AML) (Zuber et al. 2011).

The effects of small-molecule modulators of HDACs and other epigenetic regulatory enzymes suggest an important role for histone posttranslational regulation in the nervous system (Haggarty and Tsai 2011). For example, Class I HDAC inhibitors have been shown to ameliorate cognitive defects in an Alzheimer’s rodent model system, while environmental enrichment and HDAC inhibitors have been shown to enable “the recovery of impaired learning and lost long-term memories after animals had developed severe neurodegeneration and synaptic loss” (Kilgore et al. 2010; Fischer et al. 2007).

Histone acetylation is particularly interesting from a translational point of view since, as described above, certain medications based on inhibition of HDAC activity are clinically approved for treating seizure disorders and particular types of cancer (Sharma et al. 2010). Administration of nonspecific inhibitors of HDACs has yielded mixed results with respect to responses to drugs of abuse (Table 7.2). In some cases these inhibitors led to an increase in the rewarding properties of cocaine or an increase in cocaine intake (Kumar et al. 2005; Renthal et al. 2007; Sun et al. 2008a; Wang et al. 2010a). In other cases, these compounds have led to decreased cocaine intake (Romieu et al. 2008). The precise timing of HDAC inhibitor administration may play a crucial role in determining the effects of the compound. For example, in work by Dr. Marcelo Wood and colleagues, the HDAC inhibitor sodium butyrate was found to “facilitate extinction and prevent reinstatement of drug-induced behavioral changes” (Malvaez et al. 2010). In aggregate, these studies suggest that epigenetic therapies should be further explored as a potential treatment for addictive disorders.

As described earlier, despite ongoing efforts to develop safe and effective medications for the treatment of substance use disorders (SUDs), only limited success has been achieved, and no approved medications exist for the treatment of cocaine, methamphetamine, or cannabis addiction even though efforts are ongoing. It is possible that future epigenetic therapies could serve to complement current gaps in the treatment of individuals who are addicted to these drugs of abuse. In this chapter, several possible new avenues of inquiry for possible therapeutic intervention are indicated, including the development of isoform-specific HDAC inhibitors, sirtuin modulators, H3K9me2 demethylase inhibitors, DNA methylation inhibitors, and MeCP2 modulators. In addition, targeting of miRNAs or components of the pathways they regulate (Raf1 protein kinase, SPRED1, TORC, and CREB) could be of therapeutic value.