In considering the effects of drugs of abuse on glial cells, and on glial contributions to mechanisms of addiction, it is important to appreciate to what degree relevant classes of receptors and transporters are expressed on these cells within the brain reward circuitry. Because the relationship between astrocytes and neurons are the most thoroughly studied in this context, astrocytes will be the predominant glial cell type considered in this section. It has for some time been appreciated that receptors for many transmitter systems are expressed on astrocytes, including glutamate, γ-aminobutyric acid (GABA), serotonin, opioids, purines, and others (Porter and McCarthy 1997; Abbracchio et al. 2006; Shigetomi et al. 2012) . More recently, receptor expression has also been demonstrated for endocannabinoid CB1 receptors, and stimulation of astroglial CB1 receptors by neuronal endocannabinoid release leads to synaptic potentiation via astroglial-derived glutamate transmission (Navarrete and Araque 2008, 2010) . Of particular relevance to this chapter, recent evidence has accumulated showing expression of multiple types of DA receptors in a range of glial cell types in vivo, particularly astrocytes. For example, DA receptors of the subtypes D1, D3–5 (Miyazaki et al. 2004) and D2 (Khan et al. 2001; Shao et al. 2013) have been reported.

Astrocytes also express a wide range of plasma membrane transporters for amino acids, nucleosides, glucose, monocarboxylates, GABA, monoamines, and more (Morgello et al. 1995; Inazu et al. 2003; Pierre and Pellerin 2005; Nagasawa et al. 2007; Fuente-Martin et al. 2012; Kittel-Schneider et al. 2012) . Although astroglial expression of DA transporters has been found in some cases (Takeda et al. 2002; Karakaya et al. 2007), but not others (Dahlin et al. 2007; Kittel-Schneider et al. 2012) , functional uptake of both DA and serotonin in astrocytes has been reported (Kimelberg and Katz 1985; Hirst et al. 1998; Inazu et al. 1999) . Of relevance to this chapter, expression of glutamate transporters on astrocytes has been demonstrated to be of particular importance for protection against excitotoxicity and in maintaining fidelity in synaptic communication. Greater than 90 % of glutamate uptake occurs via high affinity glutamate transporters expressed on astrocytes (Anderson and Swanson 2000; Danbolt 2001) . A deficit in astroglial glutamate uptake has been widely reported following exposure to multiple drugs of abuse, and is discussed in greater detail below.

Functionality of astroglial G protein-linked receptors has also been well described (Bradley and Challiss 2012) . Considerable evidence both in vitro and in vivo indicates that neurotransmitters released from neurons (as well as neuromodulators and hormones) can stimulate astroglial receptors, leading to a Ca²⁺ signal which triggers release of gliotransmitters and in turn evoked Ca²⁺ signals in neurons (Bezzi and Volterra 2001; Agulhon et al. 2008; Lu et al. 2009; Zorec et al. 2012) , as well as local and long-distance Ca²⁺ waves that are propagated through networks of astrocytes via gap junctions (Goldberg et al. 2010) . In vivo, sensory stimulation to a mouse whisker leads to an astroglial Ca²⁺ response that is dependent upon metabotropic glutamate receptors (mGluRs) (Wang et al. 2006; Halassa and Haydon 2010) .

Signaling from neurons to astrocytes also leads to reciprocal modulation of synaptic and neuronal function by astrocytes themselves (Lu et al. 2009) . This bidirectional nature of communication between neurons and astrocytes and its role in shaping synaptic function has led to the concept of the tripartite synapse (Araque et al. 1999) , which is further supported by electron microscopy and reconstruction studies illustrating the intimate and intricate physical relationship between neurons and astrocytes (Genoud et al. 2006; Witcher et al. 2007; Belanger et al. 2011) .

Because changes in synaptic strength and signaling within the reward neurocircuitry have been heavily implicated in the cellular pathophysiology of addiction , in particular to cocaine (Stuber et al. 2010; Luscher and Malenka 2011) , the morphological and functional relationships between neurons and astrocytes indicate the potential for involvement of astrocytes in mechanisms of drug dependence. Accordingly, the following sections will in more detail discuss evidence for drug-dependent changes in glial cell biology in both human addicts and animal models of addiction, with consideration for how these changes relate to changes in synaptic function.

Human brain imaging studies have revealed a great deal of information about neural differences between drug-dependent individuals and healthy, abstinent controls, particularly within regions of the PFC (Goldstein and Volkow 2011) . Brain matter is largely characterized as white and gray matter; white matter is primarily comprised of glial cells and myelinated axons, in contrast to neuronal cell bodies, which are the major constituent of gray matter (Fowler et al. 2007) . Many studies have documented deficits in both volume of both white and gray matter, as well as in metabolic activity in the PFC, of human drug abusers (for review see (Fowler et al. 2007)) .

For example, a number of studies have revealed significant differences in white matter content between drug addicts and healthy controls (Schlaepfer et al. 2006) . Further, while frontal and temporal white matter volume in humans continues to increase until the mid-40s, compared to normal control subjects, cocaine dependent subjects do not exhibit an age-dependent increase in white matter volume (Bartzokis et al. 2002) . Relatedly, deficits in frontal white matter integrity also appear with chronic drug abuse (Lim et al. 2002, 2008) . Integrity is a measure revealed by techniques including diffusion tensor imaging, which quantifies directional diffusion of molecules along a fiber tract, providing an indicator of fiber structure and connectivity (Le Bihan et al. 2001) . White matter integrity across the brain is significantly correlated with treatment outcome for cocaine dependence (Xu et al. 2010) ; relatedly, frontal lobe white matter integrity at the start of treatment is a predictor of alcohol relapse 6 months later (Sorg et al. 2012) . Collectively, these studies indicate that structural deficits in white matter, composed largely of glia , correlate with cortical structural deficits in addicted individuals.

Other studies have reported decreased gray matter measured in a variety of brain regions within cocaine-dependent subjects (Bartzokis et al. 2000; Franklin et al. 2002; Sim et al. 2007) . Prolonged cocaine abuse has been postulated to represent an accelerated aging process with respect to reduction in gray matter volume (Bartzokis et al. 2000; Ersche et al. 2013) . Indeed, the rate of loss of cortical and subcortical gray matter volume in cocaine-dependent subjects is almost twice that of healthy age-matched subjects (Ersche et al. 2013) and accelerated age-dependent loss of gray and white matter has also been reported in chronic alcoholics (Pfefferbaum et al. 1992) .

What is the implication of a prefrontal white matter deficit? Deficient prefrontal cognitive control is a hallmark feature of addiction (Goldstein and Volkow 2002; Kalivas and Volkow 2005; George and Koob 2010) . Impairments of this nature are correlated with increased measures of impulsivity and loss of executive control; for example, decreased frontal and parietal cortex white matter integrity of cocaine-dependent subjects correlates with impaired performs on the Iowa Gambling Task, suggesting that white matter integrity may well be functionally related to functional impairments in decision making (Lane et al. 2010) . However, it is difficult using human studies to dissociate correlation with causation in the relationship between white matter content and integrity with impaired executive function. Furthermore, in all human imaging studies, changes identified between drug-dependent and healthy controls cannot be differentiated between drug-related changes and pre-existing difference which might serve a correlative or causative relationship. As such, preclinical animal models serve a valuable function to allow comparison between naïve baseline and changes which develop with drug experience. Moreover, human studies are confounded by complexity of single drug versus polydrug abusers, length and severity of drug abuse, and other environmental factors and individual differences. These confounds underscore the importance of animal models in understanding the cellular mechanisms responsible for addiction-related behaviors.

In a more limited scope of studies, these principles have been recapitulated in preclinical animal models. For example, diffusion tensor imaging of chronic cocaine treated rats has revealed a significant decrease in white matter integrity of the corpus callosum compared to saline controls; a decrease in myelin-associated protein in the same region was also observed (Narayana et al. 2009) . In a separate study, rats trained to self-administer cocaine demonstrated impairment in a working memory task, which was correlated with decreased counts of both neurons and oligodendrocytes in the PFC (George et al. 2008) . Moreover, drug seeking by rats trained to self-administer cocaine who choose drug seeking despite an adverse consequence (foot shock) was reversed by stimulation of prefrontal cortical activity (Chen et al. 2013) , providing an analog of the hypofrontality observed in human addicts. Future studies will be necessary to further ascertain the relationship between drug abuse, white matter content, and compulsive drug seeking. If then, chronic drug use leads to deficiencies in white and/or gray matter, it becomes important to understand the cellular mechanism(s) by which this may occur, and whether targeting these mechanisms may represent a means of treatment intervention for addiction.

In addition to cellular adaptations within astrocytes, drug-dependent changes in oligodendrocytes and microglia have been reported. Withdrawal from chronic cocaine administration is associated with decreased expression of white matter proteins in the NAc including myelin basic protein, proteolipid protein (PLP), myelin oligodendrocyte glycoprotein, and myelin associated glycoprotein, which is prevented by treatment with the β-lactam antibiotic ceftriaxone (Kovalevich et al. 2012) . As mentioned above, ceftriaxone is a known inducer of GLT-1 expression (Rothstein et al. 2005) , and is effective at preventing reinstatement of cocaine (Sari et al. 2009; Knackstedt et al. 2010; Fischer et al. 2013) . Interestingly, postmortem gene expression profiling in NAc tissue from human addicts has revealed a decrease in mRNA coding for myelin-associated proteins, including myelin basic protein and PLP1 (Albertson et al. 2004; Bannon et al. 2005) . In a separate study, a decrease in PLP1 was confirmed by in situ hybridization in the ventral caudate, putamen, and internal capsule of postmortem tissue from human cocaine abusers (Kristiansen et al. 2009) .

As stated in the preceding section, it is well documented that activation of glial cells occurs in response to opioid exposure. However, in contrast to the analgesic effects of opioids that are mediated through opioid receptors, activation of glial cells appears to occur via toll-like receptors (TLRs) and mediate nonanalgesic effects of opioids (Hutchinson et al. 2011) . TLRs are microbe receptors expressed on immune cells, named for homology to the Drosophila Toll protein (Anderson et al. 1985; Lemaitre et al. 1996) . The signaling of TLRs is a critical component in activating an immune response (Hanke and Kielian 2011) . Within the CNS, TLRs are expressed predominantly on microglia, but are found on activated astrocytes as well, and on oligodendrocytes and neurons to a lesser degree (Hanke and Kielian 2011) . TLR4 serves as the receptor for lipopolysaccharide, which has allowed for elucidation of a complex downstream signaling cascade leading to increased transcriptional regulation by nuclear factor kappa-light-chain-enhancer of activated B cells (Kawai and Akira 2006) . Activation of the TLR4 receptor in response to opioid exposure in vivo opposes the analgesic effects and is responsible for opioid-induced hyperalgesia, or hypersensitivity to pain that often occurs after long term opioid use and abuse and in fact oppose the analgesic effects mediated through opioid receptor (Ossipov et al. 2005; Buchanan et al. 2010) . Effects of opioid versus TLR actions can be dissociated using (−) and (+) stereoisomers of opioid agonists and antagonists; while TLRs are nonstereoselective, opioid receptors only bind to (−)-isomers. Genetic deletion or pharmacological inhibition of TLR4 receptors with (+)-naloxone impair opioid-induced CPP and reduce opioid self-administration (Hutchinson et al. 2012) .

The study described above illustrates the potential for astrocyte-derived D-serine as pharmacotherapy for substance abuse disorder. D-serine is one of a number of examples of therapies based in glial cell biology, and adaptions in neuron-glial signaling (Cooper et al. 2012) . For example, as stated previously, at least three compounds known to induce expression of GLT-1 are also under investigation as addiction therapies. N-acetylcysteine, ceftriaxone, and propentofylline are all compounds of separate classes which upregulate GLT-1, and suppress reinstatement to cocaine (Baker et al. 2003; Sari et al. 2009; Knackstedt et al. 2010; Reissner et al. 2014) . Human clinical trials indicate safety and efficacy of N-acetylcysteine against drug use for nicotine, cocaine, and cannabis (LaRowe et al. 2007, 2013; Knackstedt et al. 2009; Gray et al. 2012; Berk et al. 2013) . Further, the restored expression of GLT-1 is necessary in order for propentofylline to effectively block reinstatement to cocaine (Reissner et al. 2014) . Thus, while the field of gliobiology in drug abuse is in a relatively early stage, credible evidence already exists to support the hypothesis that targeting glial physiology affects neuronal cell biology and drug craving.

There is no question that both acute and chronic exposure to drugs of abuse leads to changes in astrocytes, oligodendrocytes , and microglia, within regions heavily implicated in reward neurocircuitry. However, our understanding of the effects of drug-dependent changes in glial structure and activity is at an early stage. Important questions remain to be addressed, such as: what are the relationships between changes in expression of myelin-related genes, deficits in white matter volume and integrity, and compulsive drug seeking? What are the relationships between glial inflammation and changes in white matter content, or between drug-dependent changes in astrocyte physiology and neuronal synaptic communication? These questions will be central toward future progress in understanding and treating addiction as a brain disease.