Stress- and glucocorticoid-induced priming of neuroinflammatory responses: Potential mechanisms of stress-induced vulnerability to drugs of abuse

Abstract

Stress and stress-induced glucocorticoids (GCs) sensitize drug abuse behavior as well as the neuroinflamatory response to a subsequent pro-inflammatory challenge. Stress also predisposes or sensitizes individuals to develop substance abuse. There is an emerging evidence that glia and glia-derived neuroinflammatory mediators play key roles in the development of drug abuse. Drugs of abuse such as opioids, psychostimulants, and alcohol induce neuroinflammatory mediators such as pro-inflammatory cytokines (e.g. interleukin (IL)-1β), which modulate drug reward, dependence, and tolerance as well as analgesic properties. Drugs of abuse may directly activate microglial and astroglial cells via ligation of Toll-like receptors (TLRs), which mediate the innate immune response to pathogens as well as xenobiotic agents (e.g. drugs of abuse). The present review focuses on understanding the immunologic mechanism(s) whereby stress primes or sensitizes the neuroinflammatory response to drugs of abuse and explores whether stress- and GC-induced sensitization of neuroimmune processes predisposes individuals to drug abuse liability and the role of neuroinflammatory mediators in the development of drug addiction.

1. Introduction

Our charge was to discuss whether neuroimmune processes might play a role in the potentiation of drug reward and drug seeking produced by stressors. Stress and stress-induced glucocorticoids (GCs) sensitize drug abuse behavior as well as the neuroinflammatory response to subsequent pro-inflammatory challenges. The central notion to be explored here is that experience of life stressors may induce a vulnerable phenotype characterized, in part, by a neuroimmune microenvironment sensitized to pro-inflammatory stimuli. We will develop the thesis that drugs of abuse (e.g. methamphetamine, morphine and alcohol) may function, in part, as pro-inflammatory stimuli or danger signals to induce neuroinflammatory responses. Therefore, upon exposure to drugs of abuse, a neuroinflammatory cascade ensues, which facilitates drug reward, dependence, and tolerance, as well as modulation of drug-induced analgesia. A central question to be explored here is whether exposure to stressors sensitizes the neuroinflammatory response to drugs of abuse, thereby potentiating the drug-related phenomena noted above. We will present converging evidence that stress-induced sensitization of neuroinflammatory processes may be a key mechanism(s) by which stress predisposes individuals to developing drug abuse. Understanding the central mechanisms by which stress sensitizes (“primes”) the neuroinflammatory response to drugs of abuse may be the key to understand how stress can promote the development of addiction, and as such is the focus of the present review. Clearly, there is an extensive literature that elaborates upon the effects of stress on peripheral immune function. However, the scope of the present review will be restricted to stress-induced modulation of central nervous system (CNS) innate immunity and its potential role in the development of drug addiction.

2. CNS innate immunity, inflammation and microglia

CNS innate immunity and its primary immune effector cell, microglia, are the key immunologic substrates for understanding how stress and GCs potentiate neuroinflammatory responses to drugs of abuse. Therefore, a brief orientation on microglia will be provided.

Innate immunity is the first line of defense against infection. Within the CNS, microglia, as part of the myeloid lineage of monouclear phagocytes, constitute the predominant innate immune cell in the brain parenchyma and serve many functions including immunosurveillance of the brain microenvironment for pathogen invasion, cellular debris, apoptotic cells, and alterations in neuroal phenotype (Ransohoff and Cardona, 2010). It is important to note that other mononuclear phagocytes including meningeal, choroid plexus, and perivascular macrophages, which reside outside the brain parenchyma, are also myeloid in origin. These macrophage subtypes also serve a critical role in the brain's innate immune response (Schiltz and Sawchenko, 2003) and may potentially contribute to the processes under discussion, but will not be further addressed here. In the normal healthy CNS, microglia send out processes that sample the local environment at a rate of several times per second (Nimmerjahn et al., 2005) and have been termed “surveillant” microglia (Ransohoff and Cardona, 2010). If a microorganism or danger signal (see below) is encountered, the cell undergoes rapid morphological and functional changes that include the synthesis and secretion of inflammatory mediators including pro-inflammatory cytokines (PICs; e.g. interleukin-1beta, IL-1β), chemokines, nitric oxide (NO), prostaglandins (PG), and reactive oxygen species (ROS). This response induces neuroinflamation. Microglial function is complex, and microglia cannot be regarded as being in only inactive or activated states. It is common to consider whether these cells are activated “classically” (via Toll-like receptors and interferon (IFN)-gamma) or “alternatively” (via IL-4 or IL-13), each of which produces cells with quite different properties. However, recent views (Mosser and Edwards, 2008) suggest that macrophages and microglia can enter a spectrum of activation states, thereby producing varying blends of pro- and anti-inflammatory products depending on how they are further stimulated. Of particular relevance to the present focus, these cells can enter a state that can be called “sensitized” or “primed” (Perry, 2004). Here, macrophages undergo a morphological transformation from “ramified” to “activated”, which is characterized by an enlargement of the soma, a shortening of processes and cell surface up-regulation of myeloid markers (e.g. major histocompatibility complex II; MHCII). Though “activated”, these “sensitized” macrophages do not produce inflammatory or anti-inflammatory products but, if further stimulated, produce exaggerated levels of inflammatory products.

A note on our use of the term inflammation is likely in order. Neuroinflammation involves a number of processes such as infiltration of the CNS by peripheral immune cells, including neutrophils, monocytes, dendritic cells, and various sub-classes of T-cells. However, these processes are generally under the control of inter-cellular signaling molecules including PICs, chemokines, etc. It is common to refer to an increase in these signaling molecules as inflammatory, and that is what is intended here.

At least 50 cell surface antigens distinguish mononuclear phagocytes from other cell types (Ransohoff and Cardona, 2010). Of these, Toll-like receptors (TLRs) are of particular relevance here. First, microglia express many of the TLRs characterized in rodents (Carpentier et al., 2008) (13 in all). Second, stress and GCs up-regulate expression of TLRs (Frank et al., 2010). Lastly, drugs of abuse signal through TLRs (see below for discussion). Briefly, cells of the innate immune system recognize microbial products via a relatively small number (roughly 30) of germ line-encoded receptors that recognize general molecular patterns that characterize different classes of pathogens (Pathogen Associated Molecular Patterns, PAMPs; Janeway and Medzhitov, 2002). For this reason, receptors that recognize PAMPs have been called pattern recognition receptors (PRRs). Of these PRRs, TLRs have been the most extensively characterized (Barton and Kagan, 2009). TLRs are a family of highly conserved membrane or cytosolic (dependent on TLR subtype) proteins that transduce signals through a family of cytosolic Toll adapter proteins that link to downstream signaling cascades (Barton and Kagan, 2009). The ligation of many of the TLR family members ultimately activates NF-κB as well as other transcription factors (Kawai and Akira, 2007), and NF-κB activation is viewed as key to the inflammatory effects of TLR ligation (Salminen et al., 2008). TLR2 and TLR4 have been the most intensively studied PRRs within the CNS, and are densely expressed on microglial cells (Aravalli et al., 2007). They are sometimes reported to be present on astrocytes and sometimes to be absent, and are typically reported to be absent on neurons (Lehnardt et al., 2003). TLR4 recognizes the lipopolysaccharide (LPS) motif that is present in the cell membrane of all Gramegative bacteria, while TLR2 recognizes lipoteichoic acid that characterizes Gram-positive bacteria (Kawai and Akira, 2007). Interestingly, TLR signaling, in particular TLR2 and TLR4, has been co-opted by endogenous signaling molecules which are thought to alert microglia to a variety of internal conditions such as cellular stress, damage or death (Kawai and Akira, 2010). The PRRs on innate immune cells have typically been thought to function to allow discrimination of “self” from “non-self”. However, because endogenous molecules can activate TLR signaling, this view has shifted in recent years to one in which TLRs discriminate “danger” from “non-danger” (Bianchi, 2007). These endogenous molecules are capable of acting like PAMPs, activating innate immune cells via ligation of TLRs and producing inflammatory responses. In 2006, Joost Oppenheim coined the term “alarmins” to represent these substances. Putative alarmins include high mobility group box 1 (HMGB1), S100 proteins, heat shock proteins (HSPs), and hyaluronan (Bianchi, 2007), with HMGB1 having received the most attention to date.

To be developed in detail below, TLR signaling may be co-opted by drugs of abuse, which raises the interesting notion that drugs of abuse act as xenobiotic agents (i.e. a substance which is found in an organism, but which is not normally produced or expected to be present in it) and may be sensed as danger signals, thereby eliciting a pro-inflammatory response.

3. Stress and stress-induced GCs sensitize neuroinflammatory responses

3.1. Stress

Exposure to stressors and pro-inflammatory agents induces a similar profile of pro-inflammatory “cytokine signatures” in the CNS (Maier, 2003). Stressors can induce two very different types of inflammatory responses in the brain. The first is a rapid, short duration (several hours) increase in inflammatory mediators (Nguyen et al., 1998). The second is a slower developing, longer lasting (days) sensitization (or “priming”) of neuroinflammatory responses to subsequently occurring infectious/pathogenic stimuli or stressors (i.e. delayed challenge) (Johnson et al., 2003). These two inflammatory sequelae of stress are very likely produced by different mechanisms (see below). The focus here is on stressor-induced inflammatory priming to a subsequent challenge (i.e. drugs of abuse) that is delayed in time relative to the initial stressor. This is the focus because inflammatory priming may have especially important implications for understanding how stress predisposes individuals to developing addiction to drugs of abuse. Importantly, the mechanism(s) of stress-induced inflammatory priming is almost entirely unknown.

The acute neuroinflammatory responses to stressors are induced rapidly and are short-lived. Brain levels of pro-inflammatory cytokines are already elevated immediately after a moderate duration (∼2 h) of stress (O'Connor et al., 2003) and persist for 4–6 h. Indeed, a considerable literature shows that several different kinds of stressors induce pro-inflammatory cytokines in the CNS (Anisman, 2009; Hennessy et al., 2009; Minami et al., 1991; Shintani et al., 1995). These cytokines are very likely microglial in origin. The mechanism(s) by which stressors activate microglia and lead to the production of inflammatory molecules is not well understood, but involves CNS noradrenergic processes (Johnson et al., 2005). Importantly, it is quite clear that GCs are not involved in the initiation of this immediate short-term inflammatory response to stress as adrenalectomy potentiates the response (Nguyen et al., 1998) and GC injection before the stressor inhibits it (Nguyen et al., 2000). That is, stressors induce inflammation despite elevations in GCs, not because of the elevations. These same points can be made with regard to the immediate neuroinflammatory response to immune activators such as LPS (Goujon et al., 1996).

While stress-induced brain cytokine increases are transient and short-lived, the neuroimmune microenvironment displays a gradual increase in glial activation markers post-stress, which may underlie the sensitized neuroinflammatory response that occurs to a subsequent pro-inflammatory challenge, such as drugs of abuse. For example, our laboratory has shown that the microglial activation marker MHCII is up-regulated, and the neuronal glycoprotein CD200 (which constitutively constrains microglial activation, see below) is down-regulated 24 h post-stress (Frank et al., 2007). The CNS parenchyma shows very low levels of basal MHCII (Ransohoff and Perry, 2009) as well as high-level expression of CD200 (Barclay et al., 2002), indicative of the quiescent immunophenotype of the normal brain microenvironment. Therefore, up-regulation and down-regulation of MHCII and CD200, respectively, indicates a stress-induced shift in the activation state of the neuroimmune microenvironment. CD200 functions to inhibit microglia and other CNS macrophages through the CD200 receptor expressed predominately on brain macrophages/microglia (Barclay et al., 2002; Koning et al., 2009), thereby constraining the onset of pro-inflammatory processes. It should be noted that pro-inflammatory cytokines have returned to basal levels by 24 h after stress. However, we have demonstrated that stress potentiates the neuroinflammatory response to a pro-inflammatory challenge (peripheral LPS) occurring 1–4 days post-stress (Frank et al., 2007; Johnson et al., 2003). Further, we found that hippocampal microglia rapidly isolated 24 h post-stress exhibited up-regulated MHCII expression and potentiated pro-inflammatory responses to LPS in vitro, indicating that the prior stressor had sensitized microglia to a later pro-inflammatory stimulus (Frank et al., 2007).

Taken together, the studies conducted in our laboratory demonstrate that exposure to an acute stressor shifts the neuroimmune microenvironment toward a microglial activation state that predisposes the CNS to a heightened pro-inflammatory response if exposed to a subsequent pro-inflammatory challenge. However, delayed stress-induced activation of the neuroimmune microenvironment does not include tonic induction of increased levels of pro-inflammatory cytokines; rather, increased pro-inflammatory responses are not observed until a subsequent challenge occurs. Of note, GC levels have returned to basal levels 24 h post-stress. In fact, we have found that serum GC levels rise rapidly during the initial stress and return to baseline within 2–6 h after the stressor used in the above studies (Fleshner et al., 1995). Consistent with the results summarized above, other studies have also shown that exposure to stressors increases the immune activation state of the CNS. For example, Nair has shown that chronic restraint stress increases microglia proliferation as well as microglia activation (Nair and Bonneau, 2006). Further, blockade of the GC receptor with the GC receptor antagonist RU486 abolished stress-induced microglia activation, and exogenous GC administration mimicked the effect of stress on microglia proliferation. Likewise, a regimen of chronic stressors administered for 9 days resulted in microglia activation (as reflected by increased MHCII expression) and potentiated the neuroinflammatory response (IL-1β, tumor necrosis factor (TNF)α, nitric oxide synthase (NOS), inducible (i)NOS and p38 MAPK) to intra-hippocampal LPS (Espinosa-Oliva et al., 2011). Similar to our findings (Frank et al., 2007), chronic stress did not by itself tonically increase pro-inflammatory cytokine levels, although the activation state of microglia had been shifted toward a preferential pro-inflammatory immunophenotype. Consistent with the results of Nair and Bonneau, (2006), blockade of the GC receptor with the GC receptor antagonist RU486 abolished stress-induced potentiation of the microglia activation (reflected by MHCII) and the pro-inflammatory response to LPS. Similarly, Munhoz subjected rats to a 14-day chronic mild stress regimen (Munhoz et al., 2006). Twenty-four hours after the last stressor, LPS was administered peripherally and, 2 h later, various measures of inflammation were examined in brain. Prior chronic stress potentiated the LPS-induced hippocampal and prefrontal cortex increase in IL-1β, TNFα, and iNOS mRNA, as well as NF-κB activity. Importantly, this potentiation was shown to be GC mediated (i.e. blocked by RU486). In a similar study, de Pablos exposed mice to a 9-day regimen of chronic variable stress, with LPS injected intracortically 24 h after the last restraint. The stressor regimen facilitated LPS-induced microglial activation (MHCII expression) and cytokine expression; such effects were blocked by daily RU486 (de Pablos et al., 2006). However, in these chronic stress experiments, the number of days of stress was not manipulated, so it is possible that a single session of stress (followed by a 24 h delay before LPS) would have produced the same outcomes. It is important to note that although our laboratory uses a single session of acute stress (1.5 h of unpredictable shock) to sensitize the neuroimmune microenvironment, the results obtained from our laboratory (summarized above) are remarkably similar to the results obtained by other laboratories using chronic stressors.

3.2. Glucocorticoids

There has been tremendous interest in the effects of stress and GCs on inflammation since the pioneering reports by Selye (1950) that “stress” produces thymic involution and subsequent findings that “stress” seemed to suppress various aspects of immune function (Selye, 1955). The subsequent discoveries of the GC hormones, and their role in mediating many of the effects of stressors, led naturally to the view that GCs are immunosuppressive in general and anti-inflammatory in particular (Boumpas et al., 1993). An enormous literature has examined the effects of GCs on inflammatory responses. The general findings have shown that elevated levels of GCs suppress a variety of functions (cytokine, chemokine, NO, PG, ROS production) of innate immune cells as well as inflammatory responses to infection and injury. Although a variety of mechanisms have been implicated, it is agreed that GCs inhibit both NF-κB and AP-1 mediated gene transcription (Smoak and Cidlowski, 2004). Multiple mechanisms are involved. For example, activated glucocorticoid receptors (GRs) induce IkBalpha expression, preventing NF-κB translocation to the nucleus (Aljada et al., 1999) and also block NF-κB binding activity via interactions with the NF-κB p65 subunit (McKay and Cidlowski, 2000). Of course, these transcription factors, particularly NF-κB, are the key regulators of inflammatory genes (Solt and May, 2008).

It should be noted that a categorically anti-inflammatory effect for GCs is somewhat counter-intuitive, and does not fit with other GC actions. It is unclear why it would be adaptive for an innate immune function to be reduced during fight/flight emergencies, as these are periods of enhanced probabilities of infection. There is little reproductive value to an injured animal that escapes a predator only to die of infection. Because the idea that immune function would be reduced during an acute stressor is so obviously non-intuitive, several explanations have been offered to support why this might be rational (Munck et al., 1984). However, the basic idea that GCs are always, canonically, immunosuppressive has rarely been challenged.

Over the years, a number of papers have reported either no effect or even pro-inflammatory effects of GCs (Sapolsky, 1999), but these have largely been ignored. However, this issue has recently been re-examined in a review by Sorrells and Sapolsky (2007), with the authors concluding that GCs are indeed sometimes either without effect or strongly pro-inflammatory. Their review has elicited a great deal of controversy (e.g. see comment, Miller, 2007) because the conclusion that GCs are anti-inflammatory is strongly entrenched, and the issue is of great importance both theoretically and clinically. Clearly, if GCs and stress are not always anti-inflammatory, then the critical issues are (a) the conditions that determine when GCs are anti- and when pro-inflammatory, and (b) the mechanisms involved. Sorrels and Sapolsky concluded that pro-inflammatory effects have been most often observed when neuroinflammatory responses, rather than peripheral innate immune responses have been examined, with dose, duration of GC increase, and timing being important variables.

We recently conducted an investigation into whether the timing of GC administration relative to a pro-inflammatory challenge was an important factor in GC-induced sensitization of neuroinflammatory responses (Frank et al., 2010). A single dose of GC was administered 24 h before, 2 h before, and 1 h after an ip injection of LPS. Of note, the dose of GCs used results in blood levels of GCs similar to those observed during and after a severe acute stressor (Fleshner et al., 1995). Several mediators of neuroinflammation were measured 4 h after LPS. GCs given 1 h after LPS reduced the neuroinflammatory response. However, GCs given either 2 h or 24 h before LPS potentiated the neuroinflammatory response to LPS. The potentiation at 2 h occurred despite the fact that blood levels of GCs were still elevated by the injection at this time. Importantly, exogenous GCs increased central expression of TLR2 and potentiated the LPS induction of TLR4 24 h post-GC administration. These effects of GCs on TLR2 and TLR4 are consistent with a small literature showing that GCs up-regulate TLRs on various peripheral cell types (Galon et al., 2002; Hermoso et al., 2004; Rozkova et al., 2006; Sakai et al., 2004; Shibata et al., 2009).

To further determine whether microglia mediated, at least in part, the GC-induced sensitization effects observed in vivo, rats were administered GCs in vivo and 24 h later hippocampal microglia were isolated and exposed to varying concentrations of LPS in vitro. In vivo GC treatment potentiated the PIC (IL-1β, TNFα) response to LPS ex vivo in a dose dependent manner. This experiment provided compelling evidence that GCs sensitize microglia to a pro-inflammatory challenge. A recent study also showed that chronic administration of GCs potentiated LPS-induced pro-inflammatory signaling in the CNS (Munhoz et al., 2010). In this study, exogenous GCs were administered to elevate GC to levels observed with mild, moderate, and major stressors. Prior exposure to GCs potentiated the central PIC response to LPS as well as intracellular signaling cascades (NF-κB, MAP kinases).

As a summary, these experiments show that stress and stress-induced GCs sensitize CNS innate immune responses to pro-inflammatory stimuli. In light of evidence that drugs of abuse exhibit pro-inflammatory properties and activate TLRs (see below), stress- and GC-induced sensitization of neuroinflammatory responses to drugs of abuse may be involved in driving the neuroadaptations underlying drug abuse liability. It is important to note that stress- and GC-induced sensitization of the neuroinflammatory response to drugs of abuse has not been investigated.

4. Drugs of abuse and neuroinflammation

An emerging body of evidence suggests that activated glia and glia-derived neuroinflammatory mediators play an important role in drug reward, dependence, tolerance and analgesic properties (Hutchinson et al., 2007).

4.1. Morphine

Several lines of evidence indicate that drugs of abuse activate glial cells. Song and Zhao found that chronic morphine activated glia in both the spinal cord and brain (Song and Zhao, 2001). Interestingly, inhibition of glial activation facilitated the analgesic efficacy of morphine. Similarly, chronic morphine activated glia and up-regulated the expression of pro-inflammatory cytokines (i.e. IL-1β, IL-6, and TNFα) (Raghavendra et al., 2004). Inhibition of pro-inflammatory cytokines reversed the development of morphine tolerance and withdrawal-induced hyperalgesia. Consistent with these findings, glial inhibition potentiated morphine analgesia and attenuated morphine tolerance (Mika et al., 2009). Studies from our laboratory have also demonstrated that chronic morphine induces pro-inflammatory cytokines. In addition, blockade of morphine-induced IL-1β signaling potentiated morphine analgesia and prevented the development of hyperalgesia and morphine tolerance (Johnston et al., 2004). Further, we found that chronic morphine as well as methadone increased the expression of TLR4 (see below for discussion of TLRs) and increased the expression of numerous pro-inflammatory cytokines and chemokines (Hutchinson et al., 2008). Inhibition of morphine-induced pro-inflammatory cytokines also increased morphine analgesia and attenuated the induction of morphine tolerance (Hutchinson et al., 2008). Importantly, we have also shown that morphine and other clinically relevant opioids (oxycodone and methadone) function as TLR4 agonists using TLR4 transfected human embryonic kidney (HEK-293) cells (Hutchinson et al., 2007). Moreover, a small molecule glial inhibitor (Ibudilast) that readily crosses the blood–brain barrier attenuated morphine-induced pro-inflammatory cytokine and chemokine expression as well as glial activation markers (Hutchinson et al., 2009), and was later found by our group to be a TLR4 antagonist as well (Hutchinson et al., 2010b). Ibudilast also reduced morphine withdrawal behaviors while potentiating morphine's analgesic properties (Hutchinson et al., 2009) Importantly, in line with such observations, we demonstrated that TLR4 knock out mice exhibit enhanced morphine analgesia and attenuated development of morphine tolerance (Hutchinson et al., 2010b).

An intriguing, and potentially clinically relevant, extension of the work with morphine arises when morphine metabolites are considered. From our work with in silico (computer) modeling of opioid interactions with the TLR4–MD2 co-receptor pair (Hutchinson et al., 2010a), we discovered that one of the two morphine metabolites had strong attraction to the same binding pocket on MD2 as does morphine. This dissociation of metabolite characteristics was very interesting as the metabolite that had no attraction to TLR4–MD2 was morphine-6-glucuronide (M6G), which binds opioid receptors avidly and produces potent analgesia (Abbott and Palmour, 1988). In contrast, M3G has been quite a mysterious metabolite, as it is a very long-lived metabolite known to fail to bind classical opioid receptors and to produce potent pain enhancement by as-yet-to-be-discovered mechanisms (Bartlett et al., 1994). This profile of long-lived presence plus potent pain enhancement has gained focus on M3G as an attenuator of the pain-relieving effects of its parent compound, morphine, and an important contributor to the development of morphine tolerance (Mazoit et al., 2007). What our group has discovered is that the strong attraction of M3G to the key TLR4–MD2 binding pocket is paralleled by: (a) activation of TLR4 receptor signaling, blocked by LPS-RS (LPS derived from Rhodobacter sphaeroides), a classic TLR4 antagonist, (b) stimulation by M3G of IL-1β production in microglial cells, and (c) amplification of pain which is blocked by the TLR4 antagonist (+)-naloxone, the microglial inhibitor minocycline, and IL-1 receptor antagonist (Lewis et al., 2010). Taken together, these data suggest that metabolites of drugs of abuse bear careful consideration, in addition to the parent molecules, for exploring how glial activation may contribute to the effects observed.

Opioid-induced glial activation also plays a role in morphine dependence and withdrawal behaviors (Hutchinson et al., 2007) as well as opioid reward (Bland et al., 2009; Hutchinson et al., 2008). Intriguingly, TLR4 has been implicated in both the behavioral and neurochemical sequelae of morphine reward (Northcutt et al., 2010). Similarly, Narita found that inhibition of glial activation decreased the rewarding properties of morphine (Narita et al., 2006). They also documented that astrocyte conditioned media enhanced morphine reward. Although there are a variety of ways in which glial activation and the consequent release of inflammatory mediators could play a role in drug-related processes, it may be especially important that IL-1β can down-regulate the glutamate transporter GLT-1 (Tilleux and Hermans, 2007), thereby increasing glutamatergic neurotransmission. If this were to occur in the ventral tegmental area (VTA) or nucleus accumbens (NAcc), the dopaminergic neurotransmission that is critical to drug reward would be potentiated. Indeed, Narita et al. (2006) found down-regulation of glial glutamate transporters to be implicated in the glial-induced modulation of drug reward, and we have shown that the glial inhibitor, Ibudilast and the TLR4 antagonist (+)-naloxone each attenuate morphine-induced dopamine efflux in the nucleus accumbens (Bland et al., 2009; Northcutt et al., 2010).

4.2. Methamphetamine (Meth)

A considerable number of reports show that Meth activates microglia (Bowyer and Ali, 2006; Buchanan et al., 2010; Escubedo et al., 1998; Fantegrossi et al., 2008; Goncalves et al., 2010; Guilarte et al., 2003; Ladenheim et al., 2000; LaVoie et al., 2004; Pubill et al., 2003; Sekine et al., 2008; Sharma and Kiyatkin, 2009; Sriram et al., 2006; Theodore et al., 2006; Thomas et al., 2004, 2008; Thomas and Kuhn, 2005). Of note, a study in humans demonstrated that microglial activation was elevated in several brain regions including the striatum in long-term Meth users (Sekine et al., 2008). Strikingly, microglia activation was observed at least 2 years after Meth abstinence. Further, Meth induces an array of neuroinflammatory effects including PIC, chemokines, and oxidative stress (Cadet and Krasnova, 2009; Krasnova and Cadet, 2009). Two initial reports demonstrated that acute Meth treatment increased hypothalamic IL-1β (Yamaguchi et al., 1991a,b), which reflects the hyperthermic effects of Meth (Krasnova and Cadet, 2009). Several groups have extended this finding to show that Meth induces PICs across several brain regions including brain reward pathways, and Ladenheim et al. implicated IL-6 in Meth-induced astrogliosis, oxidative stress, and neurotoxicity in several brain regions including the striatum, frontal cortex, and hippocampus (Ladenheim et al., 2000). Here, the effects of Meth were found to be ameliorated in IL-6 KO animals. Additional studies have demonstrated that Meth indeed induces PICs in CNS. Acute Meth increased striatal and frontal cortex expression of TNF (Flora et al., 2002). Likewise, acute Meth increased TNFα and IL-1β in striatum and potentiated cytokine response to HIV TAT (Flora et al., 2003), as well as the chemokine expression to the HIV Tat protein (Theodore et al., 2006). Moreover, prior Meth administration potentiated the pro-inflammatory cytokine response to LPS across several brain regions including striatum, hypothalamus, hippocampus, cortex, and cerebellum, and prior Meth also potentiated the microglial activation induced by LPS (Buchanan et al., 2010). Consistent with a key role for microglia, acute Meth-induced increases in the expression of TNF, TNFr1, TNFr2, IL-1α, and IL-6 were blocked by pretreatment with the microglia inhibitor minocycline (Sriram et al., 2006). Similarly, indomethacin (a non-steroidal anti-inflammatory) pretreatment can prevent the neuroinflammatory effects of Meth (Goncalves et al., 2010). In rhesus monkeys, chronic Meth increased interferon α/β in frontal lobe (Coutinho et al., 2008) and striatal expression of TNF and NF-κB (Lai et al., 2009). Consistent with several studies cited above, repeated Meth treatment increased TNFα throughout the brain (Nakajima et al., 2004). Interestingly, Meth-induced behavioral hyperactivity and sensitization was potentiated in TNFα KO animals. Furthermore, pretreatment with exogenous TNFα reduced Meth-induced locomotor activity, the rewarding effects of Meth, and several indicators of neurotoxicity. This study suggests that the TNFα response to Meth plays a neuroprotective role.

Few studies have examined whether stress modulates the neurotoxic effects of Meth. A regimen of chronic unpredictable stress potentiated the hyperthermic response to Meth (Matuszewich and Yamamoto, 2004). Prior chronic stress was also found to potentiate the hyperthermic response to Meth, which was blocked by treating animals with the glucocorticoid synthesis inhibitor, metyrapone, during chronic stress exposure (Tata and Yamamoto, 2008). Lastly, treatment with the glial inhibitor, Ibudilast (a drug used extensively in our laboratory to modulate the pro-inflammatory effects of morphine; see above), attenuated stress-induced reinstatement of extinguished Meth-maintained responding (Beardsley et al., 2010).

A considerable literature has demonstrated that Meth is neurotoxic, depending on dose and time course of administration (Krasnova and Cadet, 2009). It should be noted that evidence of Meth acting through PRRs such as TLRs is lacking. Therefore, the above evidence suggests that the neuroinflammatory effects of Meth may be secondary to the neurotoxic effects of Meth. In the case of Meth exposure, neurotoxicity may result in the release of inflammatory factors (i.e. HMGB1 etc.) resulting in the activation of microglia and release of pro-inflammatory mediators.

4.3. Alcohol

Chronic exposure to alcohol in vivo activates microglia and astrocytes (Fernandez-Lizarbe et al., 2009) and increases brain levels of inflammatory mediators such as IL-1 (Qin et al., 2008) and activates inflammatory transcription factors such as NF-κB (Valles et al., 2004). Indeed, microglial activation, astrogliosis, and increased expression of PICs have been observed in postmortem examination of alcoholic brains (He and Crews, 2008). The duration of exposure to alcohol that is required is unknown, but as few as three daily doses of alcohol (4 g/kg i.g. in mice, Fernandez-Lizarbe et al., 2009) have been reported to induce microglial activation in cortex. In vitro, even a brief exposure to alcohol at physiological concentrations (50 nM) induces inflammatory processes in astrocytes (Blanco et al., 2005) and microglia (Fernandez-Lizarbe et al., 2009).

Although there are a variety of mechanisms by which alcohol could induce neuroinflammation, a small but consistent literature suggests action at TLR4. As noted above, alcohol rapidly activates microglia in vitro, up-regulating microglial activation markers and leading to NF-κB activation and the production of PICs. Fernandez-Lizarbe et al. (2009) found all of these to be absent in microglia taken from TLR4 deficient mice. Importantly, TLR4 knockout (Fernandez-Lizarbe et al., 2009) and knockdown with siRNAs (Alfonso-Loeches et al., 2010) also blocked alcohol-induced neuroinflammatory responses in vivo.

It is worth noting that the induction of TLR4 signaling by alcohol (and other drugs) does not imply that there is a classical ligand–receptor interaction between alcohol and TLR4. There are a number of other mechanisms by which TLR4 signaling could be produced. For example, it has been suggested (Blanco et al., 2008) that alcohol interacts with lipid rafts in such a way that TLR4 is recruited, triggering its endocytosis and interaction with co-signaling molecules such as MD2, and the consequent induction of downstream signaling. An understanding of the mechanism (s) by with alcohol and other drugs lead to TLR signaling awaits clarification.

5. Stress and GC sensitization of drug abuse behavior

Stress and GCs facilitate and sensitize the behavioral response to drugs of abuse including potentiation of drug reward, locomotor activity, self-administration, and relapse (de Jong and de Kloet, 2004; Piazza and Le Moal, 1997). This facilitation of drug reward processes by stress has been the subject of numerous reviews (Briand and Blendy, 2010; Koob, 2009; Sinha, 2008) and need not be reviewed here. Importantly, stress and GCs activate brain reward pathways, such as dopamine signaling in the nucleus accumbens, in a manner similar to drugs of abuse (Marinelli and Piazza, 2002). Further, drugs of abuse activate the hypothalamo-pituitary–adrenal (HPA) axis (Armario, 2010). Our laboratory has also found that stress as well as GCs sensitizes the rewarding properties of morphine, and that just as with stress-induced potentiation of later inflammatory responses to immune activation, adrenalectomy blocks stress-induced potentiation of drug reward (Der-Avakian et al., 2005).

6. Clinical implications

Briefly, a considerable literature has documented the high degree of comorbidity between substance abuse and psychiatric disorders (Merikangas and Kalaydjian, 2007). Particularly noteworthy is the high degree of comorbidity between major depression and alcohol abuse (Davis et al., 2008). This relationship between major depression and alcohol abuse is of particular relevance in the context of the present review given that stress (Anisman, 2009), GCs (Pariante and Miller, 2001), and inflammatory processes (Dantzer et al., 2008) have been implicated in the etiology of major depression. Of note, pro-inflammatory cytokines administered to humans induce depression-like symptoms (Pollak and Yirmiya, 2002). Interestingly, comorbid depression and alcohol abuse result in a more extreme clinical phenotype characterized by earlier onset of major depression, longer duration of illness, increased number of depressive episodes, increased comorbid psychiatric disorders, more suicide attempts, and a greater family history of suicide (Davis et al., 2008). This raises the intriguing possibility that the pro-inflammatory effects of alcohol may potentiate the ongoing pro-inflammatory processes in major depression, thereby exacerbating the clinical course of major depression.

7. Conclusion

The evidence summarized to this point indicates that (1) stressors, possibly mediated by the GC rise that they produce, sensitize microglial pro-inflammatory responses to subsequent immune challenge, likely via up-regulation of TLRs; (2) a number of drugs of abuse activate microglia, likely via action at TLRs, with this aspect of drug action being involved in the mediation of several drug-related behaviors; and (3) stressors potentiate a variety of drug-related behaviors, with GCs playing a role. From the pattern of these findings it is natural to suggest that stressors potentiate drug-related behaviors, in part, because the drugs of abuse act at TLRs on microglia that have been up-regulated by prior stressor exposure, thereby leading to an exaggerated neuroinflammatory response to the drugs. This neuroinflammatory response then potentiates well-known drug pathways such as the VTA–NAcc dopamine system.

At the core of this model, we are proposing a new framework for understanding part of how stressors modulate reactions to drugs of abuse, in which drugs of abuse, viewed as xenobiotic agents, function as pro-inflammatory stimuli much like a microbial pathogen, which the organism “sees” as both foreign and dangerous and thereby mounts a central immune response.