The Neuroimmune-Glutamate Hypothesis of Addiction

Abstract

Addictive drugs impact corticostriatal glutamate signaling and have immunomodulatory effects which may underlie drug-associated behaviors during different phases of the addiction cycle. Here we hypothesize that glutamate dyshomeostasis induced by addictive drug use and withdrawal is heavily orchestrated by the neuroimmune system. We systematically define how drug-induced pathologies within the nucleus accumbens (NA) glutamate tripartite synapse are tightly regulated by neuroimmune signaling. Targets within the neuroimmune system represent a novel approach that can be leveraged for clinical studies with immunomodulatory therapeutics to reverse neurobiological changes induced by addictive drugs, and thus meaningfully reduce negative clinical outcomes relevant to substance use disorders (SUDs). We outline a novel hypothesis that control of a newly defined neuroimmune-glutamate circuit and inflammasome is heavily dependent upon the type of addictive drug as well as on phase of the addiction cycle. We further provide translational evidence underscoring the tenet that neuroimmunomodulation by addictive drugs functions according to an opponent process, and we outline predictions of our opponent process hypothesis when applied to relevant polysubstance use patterns in people who use drugs. This framework could be strategically leveraged in the experimental design of clinical studies of novel SUD therapeutics.

1. Introduction

The field of addiction neuroscience has indicated that addictive drugs dysregulate innate neural systems, pathologically altering normal function and driving addiction behaviors¹. Early studies on the neurobiology of addiction largely focused on dopaminergic innervation from the ventral tegmental area (VTA) to the nucleus accumbens (NA), which encapsulates a key circuit within the mesocorticolimbic reward pathway². It is thought that this pathway underlies use of all addictive drugs, with links between intensity of dopamine release and reinforcing effects of the drug itself³. Learned behaviors require dopamine neurotransmission⁴, and literature suggests that the development of addiction requires a transition from goal directed behavior early in drug use to habit formation which occurs after longer term use of the drug due to loss of control⁵. As such, addictive drugs shift dopaminergic signaling which leads to an addictive state⁶ characterized by a spiraling cycle of chronic use, withdrawal, and relapse⁷. Evolution of the field has led to an integrative approach to defining the neurobiology underlying addictive behaviors, which has layered complexity onto this initial circuit whereby there is intense focus on how addictive drugs “hijack”⁸ neural circuits both within microcircuitry of key brain areas (e.g., the VTA, NA, prefrontal cortex (PFC), amygdala, among others) and between key brain areas in the mesocorticolimbic reward pathway. Indeed, with the advancement of neuroscience techniques, cell- and circuit-specific contributions to addiction have become increasingly more granular.

While the dopamine hypothesis of addiction has translational validity and has been heavily supported by clinical^9,10 and preclinical^11,12 work, the field has come to specifically define neurobiological mechanisms that uniquely occur in different phases of the addiction cycle¹³ and contribute to addiction vulnerability (Figure 1). Dopamine neurotransmission was stated to be the underlying neurobiological mechanism of acute drug exposure¹⁴. However, recent advancements in our understanding of neurocircuits have highlighted a large role of dopamine signaling in opioid withdrawal¹⁵ as well as the complexities of VTA dopaminergic neurotransmission¹⁶ with VTA dopamine being modulatory of glutamate plasticity during reinstatement of cocaine seeking after extinction from self-administration¹⁷, co-release of glutamate and dopamine from midbrain dopamine neurons^18,19, and neurophysiological changes in the dopamine system occurring outside of acute drug exposure^20–22. Thus, a dichotomous definition of dopamine’s role in addiction as being attributed to acute effects while glutamate is involved in relapse after protracted drug-free periods is likely overly simplistic²³. Regardless, in the late 2000s there was a shift in focus of the preclinical addiction neuroscience field to glutamate homeostasis as a pathological dysregulation within the mesocorticolimbic reward pathway which drives relapse to addictive drugs following a period of abstinence from drug use (e.g., see work from several groups evaluating drug-induced changes in the glutamate system^24–37). This became a prevalent working theory¹⁴ and the basis for an entire area of research which we describe below (for a more thorough review of NA glutamate and addiction, see³⁸).

Figure 1. The Addiction Cycle.

Following initial acquisition of drug use, maintenance of chronic use involves a cyclic process in which chronic drug use is followed by withdrawal, craving, and relapse to chronic drug use. Preclinical (non-human) animal models have been developed to test different phases of the addiction cycle utilizing operant tasks of self-administration, extinction, drug seeking, and reacquisition of self-administration. Withdrawal severity following drug exposures and use can be modeled utilizing biological (thermal hyperalgesia, body weight changes) and cognitive/behavioral measures (cognitive outcome measures such as memory tasks, somatic signs, open field testing for anxiety-like behavior). Different brain regions within the reward pathway have been implicated in different addiction cycle phases, including the basal ganglia, insula, and prefrontal cortex (PFC) which are depicted in the specific addiction cycle phase for which evidence suggests they are involved.

2. Glutamate Dyshomeostasis Driving Relapse After Drug Abstinence: A Working Mechanistic Theory

Glutamatergic signaling has been heavily studied for its role in addiction processes^39–43. The glutamate homeostasis hypothesis of addiction was originally proposed in 1995 based on findings from the benzodiazepine and ethanol literature to explain tolerance and dependence (termed a “kindling” process by which withdrawal symptoms sensitize with repeated withdrawal cycles)⁴¹ but was subsequently shifted away from withdrawal symptomatology per se to compulsive relapse, with neural circuits within the mesocorticolimbic pathway woven into the tapestry of the hypothesis. The revised glutamate homeostasis hypothesis was based on a distillation of work in the field in 2009, indicating that protracted time from chronic drug exposure uniquely dysregulates this neurotransmitter system in the mesocorticolimbic pathway, and these neurobiological shifts underlie maladaptive drug seeking behavior during relapse¹⁴ (see BOX 1 for definitions of frequently employed procedures to measure phases of the addiction cycle in preclinical studies). The theory is based on the idea that there is an impaired ability to inhibit drug seeking behavior due to an individual’s inability to update behaviors with changing environmental contingencies, driven by glutamate dyshomeostasis within the NA³⁸. Specifically, after long-term drug use, glutamatergic afferents from the PFC become dysregulated whereby there is a shift towards hypofrontality of cortical circuits^44,45 during withdrawal characterized by reduced cortical control over limbic regions⁴⁶. This results in reduced tonic levels of glutamate within the NA leading to a reduction in glutamate autoreceptor function including metabotropic glutamate 2/3 receptors (mGluR2/3)⁴⁷ and in essence reducing the “brake” on these glutamate terminals. It is important to note, however, that basal glutamate is increased after ethanol exposure⁴⁰, indicating potential unique glutamate signaling consequences induced by ethanol as compared to other addictive drugs. During drug seeking (in preclinical nonhuman animal models, reinstatement of drug seeking is typically utilized as the “relapse” test in the absence of drug delivery⁴⁸; see BOX 1), phasic glutamate is released from the prelimbic (PL) subregion of the PFC into the NA, activating post-synaptic ionotropic glutamate receptors including GluA1 (calcium permeable) AMPA³⁰ and GluN2B-containing NMDA⁴⁹ receptors on GABAergic medium spiny neurons (MSNs).

BOX 1: Preclinical (Non-Human) Modeling of the Addiction Cycle.

Preclinical non-human animal models have been used to identify neurobehavioral underpinnings of addictive drug use, withdrawal, and relapse. Here we define models that have been established and used to derive a significant amount of data supporting glutamate and neuroimmune dysregulations in addiction. We group them by phase of the addiction cycle and identify which behavior(s) they are used to characterize. We have omitted non-contingent drug exposure paradigms (e.g., osmotic minipump, locomotor sensitization, conditioned place preference) as these methods are used to model other aspects of drug exposure.

Self-Administration

Drug self-administration allows for volitional intake of addictive drugs by the animal, which models self-titration of drug dose as well as aspects of behavior associated with drug taking. Self-administration models often incorporate either an operant task, which involves a learned response on an operandum (lever press or nose poke) according to a schedule of reinforcement resulting in drug infusions via a drug pump and intravenous catheter or oral administration via a dipper well or lixit device. Self-administration can also be used to describe drinking from a bottle in the homecage, most commonly implemented for ethanol studies). Drug self-administration can be used to evaluate acquisition, maintenance, and escalation of intake after acquisition.

Withdrawal

Withdrawal procedures can either measure somatic or physiological signs of withdrawal. Somatic signs of withdrawal are often characterized via individual ratings of recorded behaviors following drug exposure using a standardized scale of various signs including escape attempts, jumping behaviors, defecation, writhing, among others. These signs can be summated for an overall withdrawal severity score or individual signs can be reported and grouped according to whether they represent anxiety-like behaviors, changes in body temperature (piloerection or teeth chattering), gastrointestinal (defecation, diarrhea), or hyperactivity (grooming, locomotion). Physiological characterization of withdrawal has been demonstrated using measures of hyperalgesia (paw withdrawal latency), body temperature (via thermometer), cognitive outcomes (memory tasks, reversal tasks), body weight changes, and open field testing (time spent in the center of the field indicates less anxiety-like behavior and lower levels of withdrawal).

Craving

Preclinical modeling of craving typically incorporates self-administration procedures followed by a protracted drug-free period. The incubation of craving model incorporates a forced abstinence period in which animals no longer have access to drug self-administration. Following a defined abstinence period (e.g., 0, 30, or 60 days), animals are returned to the operant chamber and allowed to respond on an operandum for cues previously paired with drug delivery (typically a cue light and/or a tone is delivered contingently upon the lever press, but drug delivery is not provided). Studies using this procedure often equate number of responses with craving level for the drug.

Relapse

Models of relapse can include either reinstatement of lever pressing following self-administration after an extinction period (often ~two weeks of extinction is employed, where lever pressing no longer leads to programmed consequences or delivery of the cue previously paired with the drug but not the delivery of the drug itself) or reacquisition of drug self-administration following a period of saline substitution. Reinstatement paradigms do not incorporate resumption of drug taking but rather can include either operandum response-contingent cue delivery during the reinstatement test, whereby the cue previously paired with drug delivery during self-administration is again presented upon an emitted response after an extinction period, or they can involve a drug priming (injection of the drug previously self-administered) or contextual renewal whereby an animal is placed back into the context previously associated with drug taking. The operant response rate during the reinstatement test is defined as the “relapse” outcome measure. During reacquisition, rats undergo typical drug self-administration. Following a period of acquisition, rats undergo a saline substitution period whereby the drug is replaced with saline infusions upon an emitted operant response. Once responding is stably below a defined threshold, drug reacquisition is introduced whereby responses again yield drug delivery. This model is often employed using intravenous drug self-administration but can also be utilized with drinking paradigms where water is used during the substitution phase.

A significant body of literature indicates that after withdrawal from self-administered addictive drugs, clearance of synaptically released glutamate is significantly reduced due to a downregulation of the glutamate transporter GLT-1^26,50–54, leading to overflow of glutamate into the extrasynaptic space⁵⁴ and this further dysregulates post-synaptic plasticity during drug seeking^55,56. Interestingly, MSN plasticity during drug seeking (specifically to cues previously paired with drug infusions during self-administration) is transient, with its intensity being positively correlated with drug seeking behavior and decreasing as behavior decreases (measured as increases in spine head diameter and termed transient synaptic plasticity, or t-SP^55–57). Although most studies on t-SP have evaluated this in the absence of drug delivery, one study found that when cocaine was delivered upon lever pressing during a reinstatement test, t-SP reduced back to pre-reinstatement levels⁵⁸. These data together with the numerous studies during a drug-free cued reinstatement test (see ref³⁸ for an extensive review of the literature) indicate a bidirectional effect of drug seeking versus taking on post-synaptic plasticity within the NA. These studies initially focused on a tripartite synapse within the NA, including the presynaptic glutamate afferents from the PFC, post-synaptic MSNs, and astrocytes given their role in clearing ~90% of extracellular glutamate from the synaptic cleft⁵⁹ via GLT-1. These initial findings led to discoveries that while under drug naïve conditions, astrocytes ensheathe synapses to tightly regulate extracellular glutamate, however, these cells retract from synapses during withdrawal from addictive drugs⁶⁰ and this dysregulation underlies drug seeking. NA astrocytic perisynaptic processes appear to reengage with synapses on a similar timeline, in line with drug seeking behavior⁶⁰. Further, astrocytic proximity to synapses appears to be regulated by the β-lactam antibiotic ceftriaxone, as retraction of processes from the synapse after abstinence from cocaine was reversed by treatment with ceftriaxone⁶⁰. In line with our initial findings regarding t-SP in drug seeking, we further found that NA GLT-1 transiently upregulates⁶¹ as a compensatory mechanism of astrocytes during initial drug seeking, which decreases as drug seeking discontinues. Together, these results indicate that during drug-free periods following drug use, synapses rest in a potentiated state⁵⁶ with reduced basal glutamate tone⁶², which may be due to dysregulation in astrocytic communication with and support of neurons. We recently found that fractalkine (FKN) levels within the NA are increased immediately following nicotine self-administration⁶³, which may indicate that neurons are secreting this cytokine as a distress signal^64,65 to glial cells due to dysregulated basal activity after drug use. Given high levels of extracellular glutamate during drug seeking⁶⁶ along with reductions in efficient clearing of glutamate from the synapse due to reductions in GLT-1 function and proximity to synapses on astrocytes, it is possible that neurons are in distress and possibly vulnerable to excitotoxicity. While some studies directly evaluated excitotoxicity mechanisms induced by addictive drugs^67–71, there has been little evaluation of how excessive glutamate due to drug-induced dyshomeostasis within the NA may lead to excitotoxicity, consequent neuroimmune responses, and how this may contribute to drug seeking behavior.

While the above defined NA glutamate tripartite microcircuit has been extensively studied, recent advancements in our field have identified a more complex microcircuit within the NA playing a significant role in addiction neurobiology which also includes the extracellular matrix (ECM)⁷² and microglia^73–75. Here we propose a pentapartite addiction circuit (Figure 2), including presynaptic and postsynaptic neuronal components, astrocytes, microglia, and the ECM. While the concept of a pentapartite synapse has been previously described⁷⁶, this was outside of the NA addiction microcircuit and was indeed more broadly applied to the central nervous system (CNS). Here we incorporate the concept of a pentapartite synapse specifically in the NA, expanding beyond the well-established components of the tri-partite synapse (the PFC-derived presynaptic glutamate terminal, the NA post-synaptic GABAergic MSN, and the astrocytic processes³⁸) to include microglia and the ECM. We further describe how each of the established components of the tri-partite synapse are regulated by the neuroimmune system and incorporate microglia as brain immune cells that in a drug-naïve system are always surveilling as sensors of dysregulated synaptic activity⁷⁷. We describe how addictive drugs may disrupt this process and thus dysregulate the pentapartite synapse within the NA. We evaluate evidence for alterations in neuron-glia (including astrocytes and microglia) crosstalk as an integral part of an overarching neuroimmune-glutamate addiction circuit, leveraging established mechanisms outside of the addiction field and integrating this with the initial findings within neuroimmune addiction neuroscience. Given that some concepts below provide foundations for mechanistic interactions between the neuroimmune system and the NA addiction microcircuit, we provide a table to delineate which findings are within the context of addictive drugs and demonstrate causal relationships of the neuroimmune system with addiction-related behaviors (Table 1). Next, we describe how the neuroimmune-glutamate circuit shifts with the addiction cycle akin to an opponent process^78,79, dependent upon changes in environmental contingencies (i.e., whether an individual is consuming drugs, in withdrawal, or seeking drugs). Moving more granular, we define specific neuroimmune mechanisms within an addiction inflammasome that are shifted during drug use and that drive drug seeking after withdrawal and argue that these effects are likely drug-dependent. Finally, we integrate our mechanistic model into a translational framework based on clinical evidence of immune dysregulations in substance use disorders (SUDs) as clinically defined. We further highlight what clinical evidence currently exists in establishing efficacy of immunotherapeutics for SUDs and outline future strategies for the design of clinical studies to test novel compounds based on whether drugs are actively in use or whether an individual is in withdrawal.

Figure 2. The Hypothesized Neuroimmune-Glutamate Pentapartite Synapse.

Components of the pentapartite synapse within the nucleus accumbens (NA), including presynaptic glutamate afferents from the prefrontal cortex, post-synaptic GABAergic medium spiny neurons (MSNs), perisynaptic astrocytic processes which ensheathe synapses, surveilling microglia, and the extracellular matrix, require communication with each other through neurotransmitter release and cytokine secretion. Neuroimmune signaling tightly regulates glutamatergic plasticity, which plays a critical role in driving motivated behavior during drug use and seeking. Addictive drugs impact the pentapartite synapse differentially depending upon stage of the addiction cycle as well as the specific addictive drug. NIC=nicotine; OPI=opioids; METH=methamphetamine; COC=cocaine; EtOH=ethanol; CBD=cannabinoids.

Table 1.

Table of causal in vivo studies demonstrating neuroimmune modulation of reward signaling and drug-associated behaviors in drug exposure and addiction models.

Primary Question and Outcome(s)	Participants/Subjects	Neuroimmune Manipulation	Drug Exposure/Addiction Model	Main Finding	Ref.
Does disrupting microglial activity alter nicotine withdrawal-associated changes in microglial morphology, proinflammatory cytokines, and anxiety-like behaviors?	Male B6/129 F1 mice	Microglia depletion via CSF1R inhibitor PLX5622 in chow during chronic nicotine treatment	Chronic exposure via osmotic minipump and nicotine withdrawal elicited pump removal	Nicotine withdrawal increases relative abundance of pro-inflammatory markers and anxiety-like behaviors, and microglial depletion during nicotine withdrawal attenuates increases and associated anxiety-like behaviors.	Adeluyi et al., 2019 (108)
Does VTA IL-13 alter oral (Exp 1) or intravenous (Exp 2) nicotine intake?	Exp 1: Male IL-13Rα1 knockout mice Exp 2: Male Wistar rats	Exp 1: Transgenic knockout of IL-13 receptor alpha 1 vs wild-type littermates Exp 2: Bilateral intra-VTA injections of either vehicle or IL-13 at 15 or 45 ng	Exp 1: Oral nicotine self-administration (0.16 mg/mL, free base) Exp 2: Nicotine IVSA (0.03 mg/kg/infusion) versus food-maintained responding	IL-13 acts in a neuromodulator-like fashion to lower the firing rate of midbrain dopaminergic neurons, thereby decreasing the excitatory effects of nicotine and its associated rewarding behaviors measured via nicotine intake.	Liu et al., 2025 (187)
Do microglia and p38 signaling inhibit the development, maintenance, and expression of conditioned morphine reward?	Male Sprague-Dawley rats	Bilateral intra-NA injections of 10 ug/kg minocycline (microglia inhibitor) or SB203580 (p38 MAPK inhibitor) 30-min before morphine injections during conditioning	Conditioned place preference/place conditioning to systemic morphine (7.5 mg/kg morphine for 5 days), preference tests 1 and 6 days later	p38 inhibition in the NA impairs microglial participation in the acquisition and maintenance of morphine CPP.	Zhang et al., 2012 (131)
Does G-CSF alter the conditioned rewarding effects of cocaine (Exp 1) and dopamine modulation (Exp 2)?	Female C57BL/6J mice	i.p. injections of either saline or 50 ug/kg dose of G-CSF	Exp 1: Place conditioning for i.p. cocaine (3.75 mg/kg, 7.5 mg/kg, or 15 mg/kg) Exp 2: NA slices treated with 1, 3, 10 uM cocaine	G-CSF treatment shifted the CPP dose effect curve for cocaine where female rats treated with G-CSF displayed CPP for cocaine at lower doses than saline controls. Further, G-CSF alters cocaine effects on evoked dopamine release in an estrous cycle dependent manner, where G-CSF effects are potentiated when female rats were in the estrus phase at time of sacrifice.	Siemsen et al., 2021 (130)
Do microglia regulate accumulation of calcium permeable AMPA receptors in the NA?	Adult male C57BL6/J mice Cx3cr1::CreER; RC::LSL-tdTomato mice	Microglia depletion via CSF1R inhibitor PLX5622 in chow during cocaine withdrawal period	Cocaine place conditioning after cocaine withdrawal	Microglia are necessary for synaptic adaptations in the NA during cocaine withdrawal. Microglia depletion during cocaine withdrawal prevents dendritic spines changes and CP-AMPAR accumulation in the NA shell and prevents cocaine induced hyperlocomotion.	Testen et al., 2025 (109)
Does microglial TNFα modulate cocaine induced neuroplasticity measured via AMPA/NMDA ratios and behavioral locomotor sensitization	Male CX3CR1::Cre+; TNFαflox/flox and littermate controls CX3CR1:Cre− or TNFα+/flox mice	Comparison of transgenic TNFα knockout to wild-type littermates	Non-contingent cocaine exposure induced locomotor sensitization	Mice that had TNFα specifically deleted from microglia (versus full knockout or astrocytic deletion) displayed higher cocaine sensitization after a prolonged drug-free period. TLR4 agonism via MPLA administration attenuates locomotor sensitization and reduces AMPA/NMDA ratios following a cocaine-free period.	Lewitus et al., 2016 (102)
Does modulation of NF-κB regulate behavioral sensitization of cocaine?	Male C57BL/6 mice	Bilateral intra-NA injections of HSV-GFP, - IKKdn, - IKKca	Exp 1: Cocaine place conditioning (0, 5, 10 mg/kg) Exp 2: place conditioning sensitization paradigm. 4 days of 20 mg/kg cocaine, 4 days of rest (no drug, no injections), trained for cocaine place conditioning	NF-κB activation is a key regulator of the structural properties of MSNs in the NA during chronic cocaine exposure. Inhibition of NF-κB blocks the rewarding effects of cocaine and the ability of previous cocaine exposure to increase an animal’s preference for cocaine.	Russo et al., 2009 (105)

Note: Although listed studies often included multiple experiments, only experiments that reported causal neuroimmune manipulations were described in the table. Acronyms not defined in the text: Exp = Experiment; MPLA = Monophosphoryl lipid A; HSV = Herpes Simplex Virus; i.p. = Intraperitoneal; dn = dominant-negative; ca = constitutively active.

3. Moving Towards a Neuroimmune-Glutamate Addiction Circuit

As outlined above, glutamatergic dysregulations induced by addictive drugs are well established and have set a foundation upon which therapeutics have been tested as drug relapse mitigation medications^51,80–83. A key component of the theory, as noted above, is that during drug seeking behavior there is unchecked glutamate release into the extrasynaptic space that cannot be adequately transported into astrocytes for metabolism⁸⁴ or released as a gliotransmitter to regulate synaptic networks⁸⁵, which leads to post-synaptic dysregulations that engage motor subcircuits^86,87 (e.g., the sensorimotor cortex, substantia nigra, ventrolateral thalamus, dorsal striatum⁸⁸, and ventral pallidum^89–92) to drive continued drug seeking. Here, it is important to note that excess glutamate can lead to excitotoxicity and neuronal degeneration⁹³ induced by prolonged glutamate exposure leading to excessive ion influx and neuronal cell death⁹⁴. Although there is evidence in the literature that addictive drugs can induce neurotoxicity^69,71,95 and that oxidative stress (which can induce inflammation^96,97) may play a role in responses to addictive drugs^98,99, immune consequences of excessive glutamate release have not been a sufficiently studied component of the glutamate dyshomeostasis hypothesis of addiction. It is further important to note that immune and glutamate signaling are intimately linked, with key components implicated in the original “glutamate homeostasis hypothesis of addiction”¹⁴ being regulated by the neuroimmune system (e.g., GLT-1 transcription¹⁰⁰, the cystine-glutamate antiporter (system xCT)¹⁰¹, AMPA receptor internalization¹⁰² and membrane-insertion of calcium (Ca²⁺)-permeable AMPA receptors¹⁰³, GluN2B and post-synaptic plasticity within the basal ganglia¹⁰⁴, and NA dendritic spine density¹⁰⁵). Evidence implicates neuroimmune disruptions during addiction-related behaviors across drug classes^61,106–113, and thus it is biologically feasible that neuroimmune signaling orchestrates glutamate dyshomeostasis induced by addictive drugs within the mesocorticolimbic reward pathway. Below we first detail how neurons and glial cells communicate with each other to maintain glutamate homeostasis. Next, we outline how the established components of the NA glutamate synapse listed above are intricately connected to the neuroimmune system and establish a hypothesis whereby the neuroimmune-glutamate addiction circuit regulates glutamate dyshomeostasis. Finally, in this section we discuss that neuroimmune-glutamate interactions can be bidirectional, with neuroimmune signaling being a consequence of glutamate dysregulations¹¹⁴.

3.1. Neuron-Glia Communication is Critical for Glutamate Homeostatic Regulation.

As shown in Figure 2, neurons and astrocytes within the NA comprise two functional components of the pentapartite synapse and extensive evidence indicates that while astrocytes play a critical role in regulating extracellular glutamate via clearance through GLT-1¹¹⁵, this function becomes disrupted following withdrawal from chronic drug use^{26,50–52,61,116–119}. While the glutamate homeostasis hypothesis describes the role of neurons and astrocytes in drug seeking, there remain unknown mechanisms by which they communicate with each other, or how this communication may become disrupted following drug use. Indeed, astrocytes have been an established component of the original hypothesized synapse. However, there is also an emerging role of microglia being defined in which they may regulate synaptic plasticity¹²⁰ via phagocytosis of astrocytic processes¹²¹ within the NA addiction circuit¹⁰⁹. Microglia are the resident immune cells of the brain and are highly plastic and dynamic in their structural morphology¹²², with microglial processes rapidly moving through the parenchyma¹²³ as well as targeted migration¹²⁴ to respond to pathogens and also influence synaptic transmission. Microglia also regulate astrocyte processes via pruning to regulate neuronal activity, whereby pruning results in decreased glutamate uptake, increased extrasynaptic glutamate tone on NMDA receptors¹²¹, and increased neuronal activity. Morphological properties of microglial cells align with their cellular function^125–127 with surveilling microglia often characterized by numerous thin and elongated processes. Microglia can structurally transition into a “reactive” state¹²⁸, characterized by an enlarged soma and a simplified or reduced process field to defend the CNS from environmental insults¹²⁹ and injury. Microglial structural or functional responses within the NA have been implicated in addiction processes including nicotine withdrawal¹⁰⁸, ethanol exposure¹³⁰, morphine place preference¹³¹, and cocaine seeking¹⁰⁹. However, there is a rather large gap in our understanding of how these immune cells are altered to drive glutamate dyshomeostasis and drug seeking.

The pentapartite synapse⁷⁶ requires communication between its different components for normal function^65,132. Glial cells including microglia and astrocytes secrete cytokines which are peptide signaling molecules that act through receptors. For example, during an immune response, tumor necrosis factor alpha (TNFα) is cleaved by TNFα converting enzyme, or TACE, which is also known as A Disintegrin and Metalloproteinase 17 (ADAM17)¹³³ and is then secreted by microglia and binds to TNF receptors located on other cells such as astrocytes and neurons¹³⁴. Microglial structural changes also result in secretion of cytokines as a form of communication. For example, reductions in microglia branching increase release of the inflammatory cytokines such as interleukin 1β (IL-1β)¹³⁵. Cytokines enhance glutamate release from astrocytes¹³⁶ and microglia¹³⁷, but also inhibit neuronal glutamate plasticity¹³⁸. Chemokines are messengers of communication between neurons and glia, exerting potent chemotactic and immunoinflammatory activities¹³⁹. Cytokines such as TNFα and chemokines such as FKN (also known as CX3C chemokine ligand 1, or CX3CL1⁶⁴; predominantly expressed on neurons in the NA¹⁴⁰) are critical messengers for immune cell-neuron communication¹³², as microglia synthesize and secrete cytokines in response to infections or tissue damage. As well, neurons secrete FKN in response to stress as a “help me” signal to microglia^65,140,141, as its receptor, CX3CR1, is predominantly expressed on microglia¹⁴⁰ and binds FKN. Although these are just some examples by which neurons communicate with glial cells for normal function, the addiction field is just beginning to discover how chronic drug use and/or withdrawal interrupts these signaling axes to drive continued use, withdrawal severity, and relapse.

3.2. A Central Role of NA GLT-1 and xCT in Glutamate Dyshomeostasis in Addiction: Regulation by the Neuroimmune System.

As mentioned above, GLT-1 and xCT within the NA have been extensively implicated in addiction processes^{26,51,52,118,142} as they are heavily expressed on astrocytes and regulate synaptic glutamate and oxidative stress, respectively. Less is known, however, regarding how addictive drugs interact with the immune system regulating transcription of these proteins. This regulation of these transporters (e.g., regulation of GLT-1 through the nuclear factor-kappa B or NF-ĸB pathway¹⁰⁰) may play an important role in how they regulate the NA glutamate synapse during drug seeking behavior.

The cystine-glutamate antiporter xCT is the main functional subunit of system xc-, and this antiporter plays a large role in maintaining the redox state¹⁴³. This protein is predominantly localized on glial cells¹⁴⁴ and is a dedicated sodium-independent transporter of cystine and exchanges extracellular cystine for intracellular glutamate in a 1:1 stoichiometry¹⁴⁵. xCT is critical for enhancing glutathione export from non-neuronal cells, which exerts neuroprotection¹⁴⁶. xCT expression is regulated by nuclear factor erythroid 2-related factor 2 (Nrf2), which is a transcription factor that regulates cellular defenses against toxic and oxidative insults in a variety of cell types^147,148, including astrocytes, where it is highly stable¹⁴⁹, and neurons where it is less stable¹⁵⁰. While Nrf2 is mainly studied for its role in oxidative stress, it also plays important roles in innate and adaptive immunity through immune surveillance and immune response to inflammatory diseases¹⁵¹. Importantly, xCT is involved in immune responses, and inhibition of xCT gene expression can reduce glutathione production in tumor cells and ultimately lead to cell death¹⁵². Preclinical rodent studies have demonstrated reductions in xCT function and expression following cocaine^51,153, and have suggested that restoration of xCT function may be a viable therapeutic avenue for treatment of SUDs.

Therapeutics that restore xCT function, such as the β-lactam antibiotic ceftriaxone or the antioxidant N-Acetylcysteine (NAC), have both been shown to reduce drug seeking as well as ethanol intake¹⁵⁴ in rodent models³⁸, and these compounds also reduce immune system activity. Specifically, evidence shows that ceftriaxone reduces neuroinflammation-induced glial cell activation¹⁵⁵ and cytokine production¹⁵⁶. NAC acts on xCT by catalyzing a 1:1 release of astrocytic glutamate for extracellular cystine, which provides more than 50% of the basal extrasynaptic glutamate in the NA¹⁵⁷. In addition, NAC reduces cytokine production, replenishes glutathione, and suppresses microglial activation^158–160. It has also been well established that NAC activates the Nrf2 pathway^161–163, and thus this may be a biologically viable mechanism by which it restores xCT in the NA after drug self-administration. Despite these clear relationships, preclinical studies on both drug taking and drug seeking under abstinence conditions have not extensively evaluated the ability of ceftriaxone and NAC to reverse dysregulated glutamate signaling by impacting the Nfr2 pathway, and if this may contribute to the ability of these compounds to restore xCT and reduce drug seeking behavior. Further, it is difficult to disentangle glutamate versus immune effects of NAC, as we have shown that manipulating GLT-1 by holding down its expression during reinstated drug seeking enhances microglial activation⁶¹. Given that neuroimmune and glutamate systems are intertwined, it is critical that we determine how they interact in SUDs, and tailor treatment development around these relationships.

GLT-1 has become a primary target of interest for addiction therapeutics⁵⁰, given a vast amount of preclinical literature implicating dysfunction of this transporter protein after self-administration of a number of drugs including cocaine²⁶, nicotine^{53,61,118,142}, ethanol^164,165, among others. Despite this heavy focus, it remained unclear how NAC, which acts on xCT as described above, could restore expression of GLT-1 within the NA. Thus, our prior study evaluated how the NF-ĸB pathway, which regulates transcription of GLT-1¹⁰⁰, may be involved in the ability of NAC to reduce drug seeking and restore GLT-1⁶¹. Through use of viral-mediated gene transfer of constitutively active or dominant negative forms of I kappa kinase (IKK) which regulates activity of the NF-ĸB pathway, we determined that inhibition of this pathway is required for NAC to reduce nicotine seeking behavior and restore GLT-1 protein in the NA. We further found that while NAC reduced nicotine seeking, it did so through GLT-1-independent mechanisms as we were able to block reinstatement of nicotine seeking by inhibiting IKK, which blocked NAC-induced restoration of GLT-1. Others have also shown that overexpression of GLT-1 is not sufficient to inhibit drug seeking¹⁶⁶. Together, these studies indicate that beyond GLT-1 as a primary target of interest for SUD therapeutics⁵⁰, neuroimmune signaling may be a critical regulator of glutamate dyshomeostasis that could be a viable therapeutic target for reducing drug seeking behavior.

3.3. AMPA and NMDA Receptors: Evidence for Alteration by Addictive Drugs and Regulation by the Immune System.

AMPA and NMDA receptors play large roles in synaptic plasticity within the NA during drug seeking behavior. Indeed, there is a hypothesis that the brain reward pathway returns to a developmental-type state after extensive exposure to addictive drugs whereby subunits of these glutamate receptors switch from stable to more plastic subtypes¹⁶⁷. It is thought that these switches of receptor subunits in glutamate receptors from stable to Ca²⁺-permeable subtypes are part of a generation of silent synapses underlying drug-associated memories¹⁶⁸. Silent synapses (originally termed “ineffective synapses”¹⁶⁹) are glutamatergic synapses that have been found in nearly every location in the brain tested¹⁷⁰. They contain NMDA receptors but either lack AMPA receptors or contain highly labile AMPA receptors (thus they are AMPA-silent). These synapses are immature connections that emerge during early development as early neuronal networks that eventually develop into fully functional neural circuits¹⁷⁰. Functionally, these synapses lack excitatory post-synaptic currents (EPSCs) at the resting membrane potential but generate them upon depolarization. These synapses are evident in early development but have also been found in the adult rodent brain after exposure to addictive drugs^33,34,171. There is evidence that AMPA receptor composition shifts from early to late withdrawal timepoints¹⁷², concurrent with a timeline associated with “incubated craving” of drugs³⁰ (measured as increased lever pressing to cues previously associated with the drug during self-administration). Specifically, AMPA receptor composition becomes GluA1-enriched (GluA2-lacking) across an extended withdrawal period after cocaine, typically evaluated around 45 days has been tested out to 70 days¹⁷³ (although this has not been evaluated in comparison to a measure of withdrawal severity and is thought more to contribute to seeking the drug after protracted withdrawal rather than the withdrawal experience itself). This receptor subtype is Ca²⁺-permeable and is more prevalent during early stages of synaptic plasticity, which is then replaced by GluA2-lacking AMPA receptors at later maintenance stages¹⁷⁴. Interestingly, surface expression of AMPA receptors is regulated by the cytokine TNFα, which is released from glia¹⁷⁵, and TNFα binds to its receptor on neurons (TNFR1) to increase expression of GluA2-lacking AMPA receptors on neurons which results in increased excitatory synaptic strength. Further, these results were specific to TNFα and no other cytokines¹⁷⁶, indicating that the cocaine withdrawal-associated insertion of GluA2-lacking AMPA receptors within the NA may be specifically regulated by this proinflammatory cytokine¹⁷⁶. In further support of immunoregulation of Ca²⁺-permeable AMPA receptor insertion within the NA synapse, one study demonstrated that depletion of microglia resulted in blockade of accumulated Ca²⁺-permeable AMPA receptors¹⁷⁷. Thus, it is biologically feasible that insertion of these AMPA receptors during cocaine exposure and withdrawal may be mediated by the neuroimmune system.

NMDA receptor subunit composition has also been shown to shift after exposure to addictive drugs. Silent synapses as described above involve insertion of both Ca²⁺-permeable AMPA receptors and NMDA receptors which contain GluN2B subunits³³. GluN2B-containing NMDA receptors are critical for the induction of long-term depression (LTD)¹⁷⁸ as they are predominantly expressed extrasynaptically (with GluN2A-containing NMDA receptors being expressed synaptically¹⁷⁹). Elevated GluN2B protein has been detected in the NA following a 2-week extinction period after self-administration of both heroin⁴⁹ and nicotine⁵³, indicating that this specific mechanistic shift is conserved between addictive drugs from different drug classes. Although NMDA receptor trafficking has been shown to be regulated by a large number of mechanisms¹⁸⁰ including neuronal activity-dependent targeting¹⁸¹, there is some evidence that GluN2B expression is regulated by the immune system. For example, one study found reduction in GluN2B-containing NMDA receptors in the striatum of mice lacking IL-17A¹⁰⁴, which is a cytokine that acts as neuromodulator of synaptic transmission and plasticity¹⁸². While there is less evidence of immune regulation of NMDA receptors, it is important to note that neuroimmune signaling regulates synaptic plasticity^138,183 and thus may indirectly impact NMDA receptor function. In the section below, we describe immune regulation of synaptic plasticity within the NA.

3.4. MSN Plasticity is Regulated by Immune Signaling.

As mentioned in the previous section, synaptic plasticity is tightly regulated by immune signaling¹⁸⁴. Immune molecular signals are heavily involved in circuit remodeling, as cytokines secreted by microglia (e.g., P2Y12¹⁸⁵) or neurons (e.g., FKN^64,186, IL-17A¹⁰⁴ or IL-13¹⁸⁷) and microglia processes interact with neurons to orchestrate synaptic activity^64,188. Further, disruptions in neuron-microglia communication can disrupt synaptic plasticity, underscoring the importance of neuroimmune regulation of neuronal plasticity. A key player in cytokine regulation of MSN plasticity in the striatum is TNFα. Evidence indicates that TNFα tightly regulates AMPA/NMDA ratios of direct pathway MSNs (those expressing dopamine 1 (D1) receptors and heavily implicated in driving drug seeking behavior¹⁸⁹), as blocking TNFα normalized reductions in synaptic strength induced in a mouse model of Huntington’s Disease¹⁹⁰. TNFα has also been studied in the context of cocaine sensitization, whereby repeated cocaine exposure increased activation of microglia and production of TNFα as a result of decreasing glutamate synaptic strength in the NA¹⁰². Importantly, this result was found immediately after cocaine exposure, however, after a cocaine-free period, reactivation of microglia with a toll-like receptor 4 (TLR4) agonist increased TNFα and decreased synaptic strength, resulting in reduced cocaine sensitization. These results indicate that reversal of abstinence-induced increases in NA MSN plasticity by a neuroimmune therapeutic may have beneficial effects on cocaine-related behaviors.

As mentioned in the introduction, dendritic spines on MSNs undergo significant morphological changes after drug self-administration, abstinence, and during drug seeking behaviors¹⁹¹. Specifically, dendritic spine head diameter and density appear to change in response to drug exposure^24,192 as well as during drug seeking after a period of abstinence from self-administration⁵⁷ (in all referenced examples here, rats were exposed to cocaine although NA spine dynamics have been evaluated after other addictive drugs as well^49,193,194). Although dynamics and components of the actin cytoskeleton¹⁹⁵ have been heavily studied as the mechanism by which spines are structurally changing in response to cocaine^24,196,197 and has been postulated to be a potential therapeutic target for SUDs¹⁹⁸, regulation of these dynamic structural changes in MSNs remains unclear. Although unstudied in the context of addictive drugs, there is evidence in the literature that organization of the actin cytoskeleton is regulated by TNFα¹⁹⁹ by transiently increasing polymerized F-actin²⁰⁰. Thus, increased TNFα during abstinence from addictive drug use and during drug seeking (as we⁶¹ and others¹⁰⁸ have previously found after nicotine) may induce actin cytoskeleton remodeling, allowing for structural changes in spine dynamics during drug seeking behavior.

3.5. Beyond the Tripartite Synapse: Immune Cell-Glutamate Protein Interactions in Addiction

A large focus of the glutamate dyshomeostasis hypothesis of addiction has been on the tripartite synapse, with glutamate protein expression and function being characterized on presynaptic and postsynaptic neurons as well as on astrocytes within the NA. However, in many studies, protein expression has been characterized by subfractionation to evaluate membrane-bound protein^52,53,56,61 or biotinylation^201,202 to evaluate cell surface expression without evaluation of cell-type specificity. While astrocytes are part of the neurovascular unit along with microglia²⁰³ and have been considered in the glutamate dyshomeostasis hypothesis²⁵, the focus has largely been on astrocyte expression of glutamate transporters²⁰⁴ and their proximity to synapses⁶⁰ in clearing extracellular glutamate as well as providing non-vesicular glutamate tone on mGluR2/3²⁰⁵ receptors through xCT²⁰⁶. In some cases, evaluations of GLT-1 contributions to drug seeking behavior were conducted without cell-type specificity. For example, vivo morpholinos have been created to manipulate expression of glutamatergic proteins within the NA synapse. Vivo morpholinos are oligomers with enhanced cell-permeability that are commercially available and either splice or bind to mRNA to block protein translation to suppress protein expression^207,208. These vivo morpholinos have been developed to suppress GLT-1²⁰⁹, and studies have shown that blocking the ability of NAC to restore GLT-1 abolishes its efficacy in reducing drug seeking behavior (both for cocaine⁵² and nicotine⁶¹). However, these vivo morpholinos were not designed to only evaluate astrocyte-specific GLT-1, and indeed, holding down GLT-1 expression with these vivo morpholinos in the NA resulted in increased microglial activation as measured via CD40 expression⁶¹. Importantly, GLT-1 is inducible on microglia and brain-infiltrated macrophages²¹⁰ (as is xCT²¹¹) and is also expressed on neurons²¹², making it difficult to disentangle contributions of GLT-1 on each cell subtype in the synapse to drug seeking behavior as currently established in the literature. Thus, it is possible that restoration of microglial GLT-1 may be an important mechanism for reducing drug seeking behavior that is as-of-yet unknown. Given recent focus on microglia to the NA addiction tripartite synapse²¹³, and specifically evidence that they may prune astrocytes to increase drug seeking behavior¹⁰⁹, it is possible that microglial dysregulations in the NA play an outsized role in driving drug seeking after abstinence.

Immune cells, including microglia and macrophages, also express ionotropic and metabotropic glutamate receptors to respond to neuronal activity²¹⁴. For example, microglia express NMDA receptors, and activation of these receptors results in proinflammatory responses such as microglial proliferation, activation and release of pro-inflammatory messengers²¹⁵ which may link aberrant glutamate overflow during drug seeking to chronic inflammation via activation of microglia-expressed NMDA receptors. Further, microglia express AMPA receptors, the activation of which increases production of TNFα²¹⁶ thus these receptors serve as a negative feedback mechanism in regulating microglial TNFα²¹⁷. Our prior findings support this possibility in the context of drug seeking behavior, as we previously showed that lever pressing for cues previously paired with nicotine in a reinstatement test increased the secreted form of TNFα⁶¹ and MSN plasticity driven by increased AMPA receptor current peaks⁵³. We have also shown increased GluA1 expression in the NA following extinction training from self-administration (after a drug-free period)⁵³, although it is not clear whether this was from neurons or glial cells.

Glial cells also express mGluRs, which are G-protein coupled receptors that modulate synaptic transmission and neuronal excitability. In the context of the NA tripartite synapse, mGluR2/3 receptors have been shown to play an important role in the dysregulation of presynaptic glutamate transmission^218,219 from cortical regions (including the PL) into the NA, as their function to decrease glutamate neurotransmission via activation from glutamate released from xCT is disrupted by addictive drugs such as cocaine²⁰⁵. However, this assumes that the mGluR2/3 receptors being modulated by pharmacological agents, such as antagonists to inhibit their function and agonists to enhance it²²⁰, are localized on presynaptic glutamate neurons¹⁴ which is premised on prior findings that mGluR2/3 receptors induce LTD through a presynaptic mechanism²²¹. In addition to mGluR2/3, post-synaptically expressed mGluR5 receptors have been heavily implicated in driving drug seeking behavior²⁰¹. It is thought that blockade of mGluR5 on post-synaptic MSNs inhibits post-synaptic plasticity due to glutamate overflow during drug seeking, and this mechanism underlies aberrant drug seeking behavior³⁸ which could be therapeutically targeted²²². While presynaptic mGluR2/3 and postsynaptic mGluR5 receptors have been shown to contribute to glutamate dyshomeostasis, it is important to note that reactive astrocytes express both mGluR2/3 and mGluR5 receptors during injury (in this case, spontaneous seizures²²³). Further, activation of microglia-expressed mGluR5 can have protective effects²²⁴, as it reduces microglia activation induced by lipopolysaccharide (LPS), a bacterial toxin frequently utilized to induce immune responses²²⁴. Pharmacological activation of mGluR5 via the selective antagonist (RS)-2-chloro-5-hydroxyphenylglycine (CHPG) also reduced TNFα production and microglia-induced neurotoxicity following LPS exposure, as well as reduced microglia-associated inflammatory responses and improved recovery following spinal cord injury (although it should be noted that CHPG also activates mGluR1)²²⁵. While mGluR5 inhibition may result in reductions in drug seeking behavior^222,226, it was shown that pharmacological blockade of mGluR5 with the selective antagonist 3-[(2-methyl-1,3-thiazol-4-yl)ethynyl]-pyridine (MTEP) did not reduce increased extracellular glutamate during cue-induced cocaine seeking²²⁷. Thus, NA glutamate overflow induced by drug seeking may render the synapse vulnerable to microglia-induced inflammatory responses, and blockade of mGluR5 without cell-type specificity in this context may result in concurrent increases in microglia reactivity and subsequent neurotoxicity.

Another major system that is known to regulate neuroimmune responses, synaptic functioning and glial activity is the endocannabinoid system (ECS; for a thorough review of ECS-glutamate interactions, please see²²⁸). Within the CNS, the ECS modulates neuronal transmission through activation of cannabinoid type 1 receptor (CB1) that are primarily localized on presynaptic GABAergic and glutamatergic terminals²²⁹. Specific to glutamatergic transmission, release of glutamate from the presynaptic membrane will bind to mGluR1/5 located on the postsynaptic membrane, leading to the synthesis of endogenous ligands N-arachidonoyl ethanolamine (anandamide, AEA) and 2-arachidonoyl glycerol (2-AG). Endocannabinoids act as retrograde transmitters, which will migrate back to the presynaptic membrane, bind to CB1 receptors located on the presynaptic membrane, resulting in suppression of glutamatergic neurotransmitter release probability (for an extensive review on endocannabinoid signaling see^230,231). As mentioned above, mGluR5 receptors have been heavily implicated in drug-related behaviors^201,232,233. Importantly, mGluR5 activation by presynaptically released glutamate can also result in release of endocannabinoids²³⁴ and subsequent retrograde activation of CB1 receptors located on glutamatergic terminals from the PFC to the NA²³⁵. Activation of CB1 receptors results in reductions in synaptic vesicle release as CB1 receptors are Gi-coupled receptors. Therefore, mGluR5-CB1 receptor signaling is another mechanism by which presynaptic glutamate release is regulated in the pentapartite synapse.

CB1 receptors can also impact glutamatergic transmission through astroglial functions. Endocannabinoids 2-AG and AEA not only engage in retrograde signaling from post and presynaptic membranes, but also through astrocytes^236,237. CB1 receptors are located on glial cells including astrocytes²³⁸, which may modulate neuronal activity. Specifically, endocannabinoids 2-AG and AEA once synthesized, will travel retrogradely to astrocytes, bind to CB1 receptors and this results in an increase in cytosolic Ca²⁺²³⁸. This is an important distinction between CB1 receptor functioning between neurons and astrocytes; activation of presynaptic CB1 receptors will inhibit glutamatergic transmission²³⁹, while activation of astrocytic CB1 receptors results in increased glutamatergic exocytosis²³⁸. Importantly, animal models of drug addiction have implicated a major role in astrocytic CB1 receptor-mediated synaptic glutamate transmission, specifically within the NA. For example, Zhang et al. (2021) showed that astrocytic CB1 receptor activation led to greater vesicular transport and release of glutamate in astrocytes within the NA, preventing relapse to cocaine seeking in rats. This CB1 receptor-mediated glutamatergic exocytosis was specific to AEA signaling, suggesting an important role for AEA and astrocytic CB1 receptor-mediated glutamatergic transmission²⁴⁰.

Cannabinoids, including cannabis sativa, Δ⁹ tetrahydrocannabinoid and cannabidiol (CBD) have also been shown to induce immunomodulatory effects. These chemicals and compounds activate CB1 and CB2 receptors that are located on neurons, microglia, and other immune cells. THC and CBD modulate immune responses^241,242, induce anti-inflammatory processes^243,244, decrease pro-inflammatory cytokines such as TNFα, IL-1β, and IL-6 and suppress microglial reactivity²⁴². While untested, the ability of CB receptor activation to regulate TNFα may have indirect effects on GLT-1 transcription following use of CBDs. Further, Downer et al. (2011) showed that the synthetic cannabinoid WIN55,212–2 can regulate TLR3 and TLR4 and inhibit proinflammatory signaling²⁴⁵. Given that TLR4 activation subsequently activates the NF-ĸB pathway²⁴⁶, this may also be an indirect way in which the ECS may regulate GLT-1 transcription. Previous research also indicates that reactive microglia express CB2 receptors^247,248 which are thought to regulate microglial cell function during an immune response^249,250. Taken together, these studies demonstrate that various addictive drugs modulate the immune system and glutamate signaling, underscoring the need to expand the glutamate addiction circuit to a pentapartite synapse.

3.6. The ECM Regulates Immune Signals and is Implicated in Glutamate Dyshomeostasis.

A dynamic component of the pentapartite synapse (Figure 2) is the ECM, which is comprised of macromolecules which provide structural and biochemical support to various cells. The ECM regulates immune cell mobility and function, and the immune system also repairs the ECM when it is damaged²⁵¹. The ECM is remodeled by matrix metalloproteinases (MMPs), which are zinc-dependent endopeptidases that degrade and remodel the ECM and thus play critical roles in neurophysiological processes including acute and chronic inflammation. MMPs can play both pro- and anti-inflammatory roles in inflammatory responses and can also have intracellular roles for regulating immune responses²⁵². Within the context of addictive drug use, prior studies have identified specific MMPs (MMP-2 and −9) as being critical in remodeling the ECM during reinstated drug seeking (specifically, cocaine, heroin, and nicotine) to allow for t-SP to occur⁷². MMPs not only remodel the ECM but also regulate pro-inflammatory cytokines including TNFα²⁵³ and IL-1β^254,255 and allow these signaling molecules to regulate cytokine signaling. MMPs can also facilitate long-term potentiation (LTP), as MMP-9 knockout mice demonstrate impaired LTP in the hippocampus which can be reversed with application of recombinant MMP-9²⁵⁶. Thus, while less studied than other components of the glutamate circuit, MMPs play an outsized role in neuroinflammatory processes which could have significant implications for the neuroimmune-glutamate pentapartite circuit as proposed here.

3.7. Neuroimmune-Glutamate Crosstalk is Bidirectional

Thus far, we have described how the neuroimmune system orchestrates dysregulations in glutamate signaling that have been heavily studied in the context of addictive drugs. However, glutamate signaling can also be immunomodulatory, and thus the bidirectional nature of neuron-glia crosstalk may play an important role in driving drug use, withdrawal, and relapse processes. While outside of the addiction field, studies have demonstrated that glutamate instructs microglial responses during pathology (e.g., neurodegenerative disorders) via activation mGluR type III which are localized on the microglial cells²⁵⁷. Also, activation of kainate glutamate receptors can lead to morphological and chemotactic functional changes in microglial cells²⁵⁸, demonstrating how glutamate can functionally and structurally modulate microglia. Although one study has shown that microglia prune astrocytes after abstinence from cocaine self-administration¹⁰⁹, it is not clear if this microglial pruning activity is a direct result of glutamate signaling. Mechanistic studies are needed to parse out the specific signaling mechanisms driving this effect, and whether glutamate signaling is regulating microglial function after cocaine abstinence, or whether microglial signaling (e.g., through cytokine secretion and binding to receptors on astrocytes) is inducing loss of astrocytic peripheral processes. While these studies are needed, they remain difficult to conduct as it has been challenging for the preclinical addiction field to specifically modulate microglia because self-administration studies largely utilize rats (although some labs can conduct self-administration in mice^259,260). One current technique involves ablating microglial cells via colony stimulating factor 1 receptor-1 (CSFR-1) inhibitors to bidirectionally evaluate microglia (before and after repopulation²⁶¹); however, this is typically done in chow²⁶² resulting in global microglia depletion rather than region-specific depletion, and repopulated microglia are naïve and thus may not be functionally identical to depleted microglia^263,264. This technique has been used to evaluate microglial contributions to nicotine withdrawal²⁶⁵. One paper established Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) as infecting microglia in the spinal cord using a CD68 promoter²⁶⁶, however, we utilized this virus and found exclusive neuronal expression when directly administered into the NA (unpublished data). Recently, one paper established the CX3CR1-Cre rat strain²⁶⁷ which would allow for cre-dependent DREADD expression in microglia. Importantly, this strain has yet to be validated in a self-administration model (although we have begun to do this with nicotine self-administration²⁶⁸). To date, no studies have evaluated if glutamate overflow during reinstated drug seeking (e.g., with cocaine²⁶⁹, heroin, or nicotine⁵³) after protracted withdrawal with extinction training results in morphological or functional changes in microglial cells, demonstrating a large gap in the field. Together, studies are needed to determine causal relationships between glutamate, immune signaling, drug use, and withdrawal.

4. An Addiction Inflammasome?

Above we describe the ways in which the numerous components of the established tripartite synapse as described in the original 2009 “glutamate homeostasis hypothesis of addiction”¹⁴ is likely regulated by the neuroimmune system (although the neuroimmune system is also bidirectionally regulated by glutamate signaling¹¹⁴). A confluence of evidence suggests that addictive drugs may disrupt components of the inflammasome²⁷⁰ located in microglia²⁷¹ including specific molecules along the pathway in the NA, which is part of the innate immune system that is responsible for immune activation and cell death²⁷². Here we first define the inflammasome, where we review evidence of inflammasome function from other brain regions and outside of the addiction field. We then review literature implicating inflammasome modulation by addictive drugs in regions outside of the NA and hypothesize that exposure to addictive drugs may alter the activity of an inflammasome within the NA, and this could contribute to glutamate dyshomeostasis. Because addictive drugs range in their primary mechanism of actions, and patients who prefer different drugs have distinct clinical presentations, it is feasible that these different drugs have different actions on the inflammasome. Specifically, drugs that have been established to activate immune signaling such as alcohol, benzodiazepines, opioids, and stimulants, may enhance or prime the immune system whereas drugs like to nicotine with immunosuppressive properties²⁷³ may dampen inflammasome responsiveness to glutamate excitotoxicity.

The inflammasome is a dynamic entity²⁷⁴ that is comprised of a multiprotein complex in the cytosol of immune cells that regulates activation of inflammatory caspases (Figure 3). Since its discovery in 2002, numerous inflammasomes have been identified, with each having a unique pattern-recognition receptor (PRR) that responds to pathogen-associated molecular patterns (PAMPS; detected by the Nod-like receptor (NLR) family of cytosolic proteins²⁷⁵) or endogenous signals which indicate insult or injury in the cytosol of the cell. Activation of PRRs results in a signaling cascade including recruitment of an adaptor protein which has two death domain fold proteins²⁷⁶, including a pyrin domain and a caspase recruitment domain (CARD). These domains allow the adaptor protein to bridge the upstream sensor molecule that activates caspase-1, which leads to proximity-induced autoprocessing into an enzymatically active form that cleaves downstream targets. Downstream responses include the release of IL-1β, IL-18, and pyroptosis. IL-1β is a pro-inflammatory cytokine produced by activated macrophages and monocytes and is an inactive cytoplasmic precursor which is cleaved to an active form by IL-1β-converting enzyme, or caspase-1^277,278.

Figure 3. An NA Addiction Inflammasome.

Addictive drugs can activate or inhibit the inflammasome within microglia through activating or inhibiting cytokines which interact with inflammasome signaling (specifically, IL-1β or TNFα), or through activation or inhibition of TLRs. Drugs such as ethanol, cocaine, methamphetamine, and opioids induce activation pathways, whereas nicotine inhibits them. Activation of these receptors leads to modulation of the NF-ĸB pathway and consequent changes in gene transcription of targets such as pro IL-1β or NLRP3, which are important signaling mechanisms in the inflammasome. This leads to priming and activation of the inflammasome, activating caspase-1 and increased cytokine release, pyroptosis, and consequent neuroinflammation. TLR=toll like receptor; IL-1β=interleukin-1beta; NF-ĸB=nuclear factor kappa B; TNFα=tumor necrosis factor alpha; NLRP3=NOD-like receptor 3; NIC=nicotine; OPI=opioids; METH=methamphetamine; COC=cocaine; EtOH=ethanol; CBD=cannabinoids.

While still largely unstudied, accumulating evidence indicates that exposure to addictive drugs may alter the activity of the NLRP₃ inflammasome. Activation of the NLRP₃ inflammasome first requires binding to the PRRs. Interestingly, addictive drugs such as cocaine, methamphetamine, synthetic opioids, and ethanol all bind and activate TLR4. TLR4 is a well-known PRR that activates proinflammatory signaling cascades, making it likely that these drugs may also bind to other PRRs to elicit NLRP₃ activation. Despite a small number of studies on specific inflammasome mechanisms in addiction, some studies have evaluated immune or cytokine activation after drug administration in brain regions outside of the NA. There is evidence that IL-1β is elevated in the hippocampus and PFC after repeated exposure to cocaine²⁷⁹. Further, evidence indicates that methamphetamine exposure triggers activation of the microglial inflammasome in culture²⁸⁰, thus raising the possibility that exposure to methamphetamine may lead to similar effects in vivo. Natural opioids, such as morphine, have been shown to have increased expression of mRNA of NLRP3 related proteins, caspase-1, and IL-1β, and naloxone, an opioid receptor antagonist, has been shown to inhibit IL-1β secretion and inflammasome activation²⁸¹. Synthetic opioids, such as fentanyl, have not been as thoroughly studied in terms of inflammasome activation; however, it has been shown that fentanyl produces larger pro-inflammatory effects as compared to morphine within the dorsal raphe nucleus (DRN), and that this difference may underlie how the two opioids produce different tolerance and hyperalgesia profiles²⁸². Interestingly, we have shown that IL-1β is decreased in the NA immediately following self-administration of nicotine⁶³, underscoring that inflammasomes may be sensitive to different addictive drugs and have different actions depending on the brain region affected and drug under investigation. Further supporting this possibility, cannabinoids bind CB receptors including CB2²⁸³. CB2 receptors are located on reactive microglia in an activity-dependent manner²⁵⁰, and activation of these receptors have been shown to reduce IL-1β²⁸⁴, IL-6, and increase NLRP3 clearance during an immune challenge (in this study²⁸⁵, LPS was administered). Thus, we hypothesize that chronic use of cannabinoids would inhibit activation of the NA inflammasome, akin to nicotine (see Figure 3). If chronic drug exposure and/or withdrawal results in inflammasome activation, this could be linked to dysregulations in glutamate signaling. Specifically, chronic IL-1β can induce synaptic downscaling of excitatory synapses²⁸⁶. This is in line with the directionality of effects with cocaine exposure as evidence indicates increased IL-1β²⁷⁹ but decreased NA synaptic plasticity when cocaine is present in the system (in this case, cocaine was reintroduced after a cocaine-free period in a self-administration paradigm⁵⁸). In the case of nicotine and inflammasome inhibition, glutamate dysregulations could promote learning and memory in reward-motivated behavior since low tonic levels of IL-1β are needed to facilitate long term potentiation (LTP)²⁸⁷.

There are reports that caspases, which mediate apoptosis, can also regulate synaptic plasticity and are involved in the weakening and elimination of synapses. For example, caspase 1 can inhibit AMPA-mediated LTP in the hippocampus, and inhibiting caspase 1 reverses this effect²⁸⁸. Further, IL-1β also inhibits LTP^289,290, demonstrating potential important regulation of synaptic plasticity by proteins in this pathway. Interestingly, IL-1β can also increase GLT-1 via post-transcriptional modifications in culture²⁹¹, indicating that elevation of this cytokine by drug exposure (e.g., cocaine or nicotine) may result in increased GLT-1 when the drug is present, which is in contrast to protracted drug-free conditions when GLT-1 levels are reduced⁵⁰. While there have been no links of these proteins to glutamate signaling disruptions in the NA induced by addictive drug use, it remains possible that there could be important relationships between glutamate and a NA inflammasome that could underlie drug use and seeking. Indeed, elevated IL-1β by drug self-administration could indicate increases in neuronal apoptosis within the NA, which could have profound impacts on the tripartite synapse. While untested, this possibility may be specific to the NA subcircuit and not the PFC, as one study conducted on post-mortem human tissue from individuals who used cocaine demonstrated no evidence of apoptosis in this cortical brain region²⁹². Notably, there are no reports to date indicating that an NA inflammasome is activated during abstinence from drug use or during drug seeking under drug-free conditions, raising the possibility that this mechanism may be specific to contexts in which the drug is present, although this hypothesis needs to be tested. There is evidence, however, that targets within the inflammasome such as IL-1β are upregulated in mice after morphine withdrawal in the hippocampus, and this was associated with jumping behavior during withdrawal (although IL-1β was decreased in the medial PFC, demonstrating potential brain region specificity of the increased expression)²⁹³. Thus, inflammasomes within specific brain regions may play a role in driving withdrawal severity from opioids, although additional studies are needed. While no studies to date have evaluated the role of NA IL-1β signaling in reinstated drug seeking behavior, systemic administration of the TLR4 antagonist naltrexone decreased cocaine-primed reinstatement in rats. Further, intra-NA shell administration of LPS reduced cocaine seeking behavior in this same study²⁹⁴. An additional study also demonstrated that intra-VTA LPS and IL-1β antagonism reduced cocaine seeking behavior²⁹⁵. Together, there is biological feasibility that addictive drug use and withdrawal modulate an NA inflammasome, and that this plays an important role in drug-induced NA glutamate dyshomeostasis although studies are needed to confirm these relationships.

5. An Opponent Process Theory of the Neuroimmune-Glutamate Addiction Circuit: Specificity to Drug Class

Above we describe mechanistic interactions between neurons and glial cells, with immune signaling orchestrating glutamatergic signaling and disruptions of these processes by addictive drugs. It is important to note that the “glutamate homeostasis hypothesis” as defined in 2009¹⁴ was based on a large number of studies that focused on cocaine, which is a psychostimulant that has a very different clinical and mechanistic profile from other addictive drugs such as opioids, nicotine, and ethanol. For example, as we have previously outlined²⁹⁶, withdrawal is a critical aspect of drugs that is often cited as a primary reason for relapse to drug use^297,298. However, withdrawal from cocaine is a different clinical experience than from opioids, characterized by several psychological symptoms²⁹⁹ as well as sleep and energy disruptions³⁰⁰. In contrast, opioid withdrawal is often characterized by an acute period involving muscle aches, nausea, diarrhea, yawning, vomiting, pupil dilation, etc., as well as protracted period that lasts for weeks involving sleep and mood disruptions as well as cravings that often requires medication management³⁰¹. In addition to these clinical differences, there are mechanistic and pharmacological differences between addictive drugs that may differentially impact immune interactions with glutamate signaling. For example, cocaine blocks the dopamine reuptake transporter (DAT) to increase dopamine in the synapse, whereas opioids are agonists at μ opioid receptors (MORs), ĸ opioid receptors (KORs), and delta opioid receptors (DORs). Activation of MORs results in inhibition of GABAergic tone on VTA dopamine neurotransmission¹⁵ and opioids can also directly bind and activate TLR4 on microglial cells³⁰². Importantly, different addictive drugs have different primary mechanisms of action and thus can influence neuroimmune signaling through different mechanisms. For example, cocaine is a dopamine transporter (DAT) inhibitor, it also activates TLR4 on microglia to initiate immune responses³⁰³. Nicotine is an agonist at nicotinic acetylcholine receptors (nAChRs)^304,305, which are pentameric ligand-gated ion channels^306,307 that are expressed on numerous cell types within the brain and body including immune cells³⁰⁸. nAChRs can exist in both heteromeric and homomeric configurations, and specifically, α7 nAChRs are expressed on microglia and their activation results in reductions in pro-inflammatory signaling³⁰⁹. Nicotine can also inhibit TLR4³¹⁰. As mentioned above, THC binds to CB1 and CB2 receptors and differentially modulates neuronal^193,311 and astrocyte signaling to reduce immune responses^312,313. In the NA glutamate system, there are established mechanistic differences between psychostimulants and opioids, with THC being more akin to opioids than psychostimulants¹⁹³. For example, extinction from heroin decreases AMPA/NMDA ratios in the NA⁴⁹, and in contrast, extinction from cocaine⁵⁶ and nicotine^53,61 increases this functional glutamate plasticity measure. These results collectively support the tenet that different addictive drugs alter the neuroimmune-glutamate synapse, and that the directionality of effects on the pentapartite synapse is likely drug-dependent (Figure 2).

While a large majority of studies on NA glutamate homeostasis disruptions in addiction have focused on a protracted timepoint following drug self-administration and a seeking test (under drug free conditions) in rodent models, there has been little focus on potentially important nuances in glutamate-immune function that may exist depending upon whether drug is present versus if an individual is in withdrawal or seeking the drug while in a drug-free state. Here we describe how different addictive drugs can either increase or suppress neuroimmune responding within the NA when they are biologically active immediately following exposure and/or use. Next, we describe how withdrawal from addictive drugs may exert the opposite effects on the neuroimmune system within the NA, and how this may underlie craving and relapse to drug taking after withdrawal (Figure 4). We make a case that there may be an opponent process within the neuroimmune-glutamate NA system depending on stage of the addiction cycle, with an “a” process reflecting neuroimmune signaling condition when drug is present, and conversely, a “b” process reflective of neuroimmune signaling during withdrawal.

Figure 4. The Opponent-Process Theory of the Neuroimmune-Glutamate Circuit.

In a drug naïve system, innate and adaptive immune processes function at setpoint, whereby these systems are functioning normally. Neuroimmune shifts in the positive (increased immune system reactivity) or negative (suppression of immune system reactivity) results in consequent changes in glutamate plasticity within the mesocorticolimbic reward pathway. (a) When drugs that inhibit the pro-inflammatory activity of immune system (nicotine) is present, immune responses are inhibited (a process), and thus reactivity of immune system is below setpoint. Conversely, when an individual enters withdrawal, the immune system rebounds above setpoint, resulting in increased pro-inflammatory signaling (b process). The shift between a and b processes become greater as drug use becomes more chronic (indicated by black, orange, and red arrows). (b) When drugs such as cocaine and methamphetamine are present, immune system responses (including pro-inflammatory responses) increase from setpoint (a process), and conversely, when drugs are metabolized and an individual enters a withdrawal state, immune responses reduce below setpoint (b process). With repeated drug use, these changes from setpoint become significantly greater (indicated by black, orange, and red arrows).

Exposure to addictive drugs can exert immunomodulatory effects on the brain reward pathway, and evidence indicates that the directionality of immune responses may be drug-dependent when drug is present (a process). For example, experimenter-delivered methamphetamine increases pro-inflammatory cytokines and microglial reactivity markers in the PFC, VTA, and NA shell³¹⁴, whereas nicotine suppresses a specific neuromodulatory cytokine, IL-13, in dopamine neurons in the VTA which increases dopamine neurotransmission and contributes to nicotine reward¹⁰⁸. A recent study also demonstrated increased TNFα after nicotine self-administration in male mice⁶¹ which contradicts our findings in female rats, but we recently published that nicotine’s effects on the neuroimmune system in the brain reward pathway is likely sex-specific²⁷³. Additional studies have shown that cocaine upregulates the pleiotropic cytokine granulocyte colony-stimulating factor (G-CSF) in the NA³¹⁵, which is associated with increased dopamine release at terminals³¹⁶.

Contrary to conditions in which drug is present, evidence indicates that withdrawal may exert opposite effects on the neuroimmune system (b process). As mentioned above, methamphetamine exposure increases cytokines (including IL-1β, indicating activation of the inflammasome²⁸⁰), however, withdrawal from methamphetamine resulted in some cytokine suppression (IL-1β, IL-9, and IL-15) in humans undergoing acute methamphetamine withdrawal³¹⁷. These studies support that the inflammasome may be differentially activated and suppressed depending on whether methamphetamine is present or whether an individual is in methamphetamine withdrawal. Evidence also indicates that abstinence from cocaine increases FKN in plasma³¹⁸ and decreased pro-inflammatory cytokines (IL-17A, transforming growth factor alpha, and macrophage inflammatory protein 1α [MIP-1α]³¹⁹) yet in individuals with cocaine use disorder (CUD), cocaine increases pro-inflammatory cytokines³²⁰. These findings align with an opponent process of neuroimmune system with cocaine, and also within the context of our preclinical nicotine NA data as we show that FKN is increased yet other pro-inflammatory markers are suppressed when nicotine is present after self-administration⁶³. Thus, here we cluster drugs which activate the immune system when they are present yet result in suppression of immune markers during abstinence/withdrawal (cocaine, methamphetamine; Figure 2) from those this suppress the immune system when they are present yet result in increased immune system reactivity during abstinence or withdrawal (nicotine, CBDs; Figure 2). Notably, opioids appear to increase immune responding during both use and withdrawal, which we discuss in detail below. Further, studies evaluating immune markers during withdrawal from nicotine exposure¹⁰⁸ and after 2 weeks of extinction from self-administration⁶¹ indicate increased immune system activation. It is thus important to differentiate withdrawal severity outcomes from extinction and reinstatement after self-administration, as these are two different experimental outcomes intended to model different aspects of the addiction cycle (Figure 1; also see BOX 1 for a description of different paradigms which are intended to model different phases of the addiction cycle). Nicotine withdrawal has been measured in the preclinical rodent literature and can be characterized by a number of outcome measures including paw withdrawal latency during a radiant heat stimulation test (as a measure of thermal hyperalgesia) and somatic signs³²¹. Interestingly, one study found that CX3CR1, the receptor for FKN (or CX3CL1), was increased on microglia in the spinal cord following 4 days of nicotine withdrawal (peaking at day 4 of withdrawal and decreasing in subsequent withdrawal days)³²², and this was associated with paw withdrawal latency. Although this was evaluated in the spinal cord, we also showed an increase in the CX3CL1-CX3CR1 axis immediately after nicotine self-administration in the NA⁶³, indicating that this neuronal “help me” signal may be increased after both nicotine use and acute withdrawal, and could indicate enduring disruptions in neuronal health that underlie both use and withdrawal severity. Additional studies have demonstrated elevated TNFα and IL-1β mRNA in the NA after 48 h of nicotine withdrawal¹⁰⁸, although withdrawal severity was not measured and it is not clear how these cytokines relate to behavioral expression of nicotine withdrawal. Together, these results indicate that withdrawal from nicotine may increase NA immune signaling.

As mentioned above, studies often integrate two weeks of extinction training from drug self-administration, which is a drug-free period prior to testing drug seeking behavior under drug-free conditions. This model has been heavily utilized to define glutamate dyshomeostasis after self-administration of a number of drugs³⁸. Indeed, we have previously shown that after two weeks of extinction training from nicotine self-administration, NA MSNs rest in a potentiated state as compared to saline controls⁵³. We have further shown that this basal elevation in synaptic plasticity (as measured via increased AMPA/NMDA ratios) increases even further from this baseline, coinciding with elevated drug seeking behavior (measured as lever presses to previously nicotine-paired discrete cues). We expanded upon this initial finding by defining astrocytic dysfunction (measured as reductions in glial fibrillary acidic protein, or GFAP) and immune consequences to elevated MSN plasticity and reductions in GLT-1 protein. Specifically, we found that holding down GLT-1 transcription during reinstatement (also commonly referred to in the literature as a “relapse” test) resulted in elevated microglial reactivity marker CD40 and increased TNFα secretion⁶¹. These data collectively indicate that nicotine seeking under nicotine-free conditions activates NA microglia and results in increased secretion of a pro-inflammatory cytokine (TNFα), which regulates glutamate plasticity¹⁷⁵ and is behavior-dependent. Thus, it is possible that increased neuroimmune signaling in the NA orchestrates dysregulated glutamate neurotransmission that drives drug seeking behavior although this needs to be tested with other addictive drugs.

Opioids have unique impacts on neuroimmune signaling, and a number of studies indicate that opioid exposure increases cytokine levels and immune cell reactivity³²³, including microglia activation³²⁴ in regions within the brain reward pathway³²⁴. Levels of elevated cytokines include C-X-C chemokine ligand (CXCL)1) (CXCL1)^325,326, IL-1β³²⁷, IL-6³²⁸ and TNFα³²⁹ and have been implicated in opioid tolerance³³⁰. Intriguingly, morphine has been shown to directly activate neuroinflammation, similar to endotoxin, by binding to the accessory protein myeloid differentiation protein 2 (MD-2) of TLR4. This then induces TLR4 oligomerization and downstream signaling which triggers proinflammatory responses. Together, these results suggest that a TLR4 antagonist such as TAK-242 (resatorvid), could be an effective strategy to reduce opioid-induced neuroinflammation³³¹. opioids induce a neuroinflammatory response through glial cell activation^332–334 and the release of proinflammatory cytokines, including IL-1β^{333,335–337}. Furthermore, this neuroinflammation has been shown to suppress the analgesic effect of opioids and contribute to the development of tolerance, dependence, reward signals, withdrawal effects, and hyperalgesia^{333,338–343}. With preclinical studies demonstrating TLR4 signaling involvement in opioid-related behaviors^{323,331,344–346}, exploration of novel pharmacological treatments for OUD by targeting glial activity remains a promising approach^347–351. More recently, it has been demonstrated that neuroimmune responses diverge across different brain regions via region-specific glia cell phenotypes and distinct cytokine profiles³⁵² including the PFC³⁵³, cerebellum³⁵⁴, and hippocampus^355,356. Importantly, the monoterpene compound α-pinene^357,358 whether administered during dependence induction or throughout withdrawal reduced pro-inflammatory cytokine levels in the rat hippocampus, highlighting the therapeutic potential of alpha-pinene for controlling neuroinflammation and subsequent opioid-related outcomes such as tolerance and withdrawal³⁵⁶.

Akin to nicotine, withdrawal from oxycodone is associated with increased cytokines in the brain’s reward pathway (in this example, the VTA³²⁴), and blockade of the cytokine IL-17A reduced oxycodone withdrawal-associated hyperalgesia³²⁴. Similarly, several other opioid withdrawal studies demonstrate increased proinflammatory cytokine levels that differ by brain region and sex^329,359,360 and suggests that an increase in inflammatory mediators in the brain during the withdrawal period is associated with the severity of withdrawal behavior^293,361. Thus, opioids may be unique from other addictive drugs in that they do not follow an opponent process with their effects on the immune system (although we hypothesize that they exacerbate neuroimmune signaling effects when used in combination with other drugs as depicted in Figure 5), which may be important for clinical study design as described below.

Figure 5. Testing the Theory: How the Accumbens Neuroimmune State May Drive Opioid and Nicotine Co-Use.

Opioid use and withdrawal increase accumbens neuroimmune signaling, however, nicotine functions according to an opponent process whereby use suppresses and withdrawal increases the accumbens neuroimmune state. We posit that opioid use increases neuroimmune signaling in the accumbens above setpoint, driving nicotine use to reverse this effect. As withdrawal from both opioids and nicotine begins, neuroimmune signaling then increases in a synergistic fashion above opioid use alone. This heightened neuroimmune state drives use of nicotine-containing products during opioid withdrawal, however, this leads to a suppression of neuroimmune signaling, which drives opioid craving and relapse to opioid use.

While little is known regarding how withdrawal from CBDs may influence the immune system, evidence indicates that their use suppresses immune function²⁴³. Long-term suppression of immune cell activity (e.g., microglia) through CB receptor activation by CBDs may lead to significant impacts on innate and adaptive immunity³⁶² that could rebound during withdrawal from these substances. Given the challenges with establishing preclinical (non-human animal) THC self-administration models²⁹⁶, only a few studies have begun to evaluate the impacts of volitional THC use and withdrawal on the NA glutamate system¹⁹³, and none have evaluated the NA immune system or its potential interactions with glutamate. Studies show that withdrawal from THC self-administration results in astrocytic retraction from synapses in the NA³⁶³, akin to cocaine⁶⁰, as well as a loss of large MSN spine heads³⁶⁴ and loss of NMDA receptor-dependent LTD³⁶⁵. These studies show that THC self-administration modulates components of the pentapartite synapse, and thus we hypothesize that CBDs may function according to an opponent process with suppression of the NA immune system during use and an increase in immune activity during withdrawal.

Here we describe neuroimmune system changes induced by drugs depending on whether the drug is present, if an animal is in withdrawal, or if an animal is actively seeking the drug under drug-free conditions. It appears that the drug-specific pharmacological effects may have differential impacts on the neuroimmune landscape within the NA, which may account for some of the drug-dependent differences in glutamate dyshomeostasis that have been discovered (e.g., t-SP is different following extinction from self-administration of heroin⁴⁹ versus cocaine⁵⁶ and nicotine⁵³, and there is evidence for elevated NA GluN2B after 2 weeks of extinction from heroin⁴⁹ and nicotine but NA GluN2B levels were decreased following 30 days of abstinence from cocaine self-administration³⁶⁶). This possibility is biologically feasible because as outlined in previous sections, glutamate signaling is heavily regulated by the neuroimmune system. Here we make the case that different addictive drugs exert meaningfully different effect profiles on the neuroimmune-glutamate circuit, which also differs depending on use or withdrawal state. Given these important differences, we argue below that clinical studies should be designed using this framework by evaluating immunotherapeutic drugs that either increase or decrease immune system activity depending on the primary addictive drug of use as well as whether individuals are in withdrawal or actively using addictive drugs.

6. Testing the Theory: Predictions for the Neuroimmune-Glutamate Circuit in Polysubstance Use

Thus far, we have identified changes in the neuroimmune landscape within the brain reward pathway induced by different addictive drugs and outlined important differences between drugs regarding directionality of effects on various targets within this system. While these preclinical findings are critical in our understanding of how addictive drugs impact the neuroimmune-glutamate circuit, clinical drug use patterns often involve multiple substances with varying use patterns (see our prior papers for more detail^296,367). Thus, the field should model epidemiologically prevalent patterns of polysubstance use to understand how exposure to multiple substances may alter the neuroimmune-glutamate circuit and drive further co-use. Indeed, a large majority of our knowledge regarding neuroimmune and glutamate mechanisms induced by addictive drugs is derived from single substance exposure paradigms, despite polysubstance use being the norm rather than the exception in people who use drugs³⁶⁷. In support of our hypothesis that polysubstance use uniquely alters neuroimmune-glutamate signaling, two groups have identified changes in glutamate signaling within the NA induced by polysubstance use which were not found in single substance use conditions using rat self-administration models^368–370. Thus, it is possible that polysubstance use uniquely alters our hypothesized circuit and this neurobiological shift drives continued polysubstance use. Using the opponent process framework outlined above, here we identified two highly prevalent patterns of polysubstance use, nicotine/opioid and fentanyl/methamphetamine, and provide two predictions of how polysubstance use may uniquely drive the neuroimmune-glutamate circuit to drive these use patterns. We note that opioids may not function via the opponent process framework given the persistence of neuroimmune activation during both use and withdrawal, but we hypothesize that the activated neuroimmune state induced by opioid use and withdrawal uniquely modulates the opponent process functioning of other drugs such as nicotine (described below and depicted in Figure 5).

Use of nicotine-containing products (including combustible cigarettes) is highly prevalent in opioid-using populations, with prevalence rates estimated to be over 90%^371,372. There are relationships between opioid withdrawal and cigarette smoking, as one study found that cigarette smoking during opioid detoxification was associated with higher opioid craving and lower detoxification completion rates³⁷³. Further, smoking cessation treatment outcomes are generally poor compared to cessation outcomes from individuals without comorbid OUD³⁷⁴ which may be due to more severe tobacco withdrawal³⁷⁵. Above we describe the immunosuppressive properties of nicotine when it is present, and conversely, increased neuroimmune signaling in the NA during nicotine withdrawal. We further outlined how opioids increase neuroimmune signaling in the reward pathway when they are present, and this elevated immune response is sustained throughout opioid withdrawal. Thus, the opponent process theory would predict that use of nicotine-containing products during opioid use may counteract the increase in immune response induced by opioids, thus allowing for return to setpoint (Figure 5). Given that both opioid and nicotine withdrawal lead to enhancement of neuroimmune signaling, it is possible that these conditions work synergistically to increase neuroimmune signaling beyond either drug alone. Thus, the enhanced immune response due to nicotine withdrawal during opioid withdrawal may prime the neuroimmune-glutamate circuit to drive craving for opioids and nicotine, thus driving relapse to both substances. Studies are needed to identify unique neuroimmune-glutamate circuit conditions during both nicotine/opioid use and withdrawal and determine if these mechanistic changes drive this prevalent polysubstance use pattern.

Fentanyl and methamphetamine have become prevalent co-used substances as they are frequently concurrently detected in the illicit opioid supply (termed “unintentional” co-use) and this has driven a “twin epidemic” of fentanyl and methamphetamine use crises^376,377. Intentional fentanyl and methamphetamine co-use is also a frequent pattern, as individuals report methamphetamine use during fentanyl withdrawal as a way to self-manage physical opioid withdrawal symptoms³⁷⁸ as well as prevent and/or reverse overdose³⁷⁹. As outlined above, both opioids (including fentanyl³²³) and methamphetamine enhance neuroimmune signaling when they are present, thus it is possible that these two immune-enhancing drugs act synergistically in the brain reward pathway to further enhance neuroimmune signaling when they are unintentionally co-used, although studies are needed to determine if this is the case. In contrast to when these drugs are present, methamphetamine withdrawal is associated reductions in pro-inflammatory markers³¹⁷ whereas withdrawal from opioids is associated with elevated immune signaling³⁶¹. Given that the illicit opioid supply is increasingly adulterated with methamphetamine, it is possible that some individuals are being unintentionally co-exposed despite intended opioid use thus altering the neuroimmune-glutamate circuit. The opponent process theory as outlined above would predict that if these two drugs synergistically elevate immune activation when they are concurrently used, withdrawal from the drug combination would result in a significant decrease in neuroimmune signaling, thus this could drive methamphetamine use. Together, studies are needed to identify how polysubstance use shifts the neuroimmune-glutamate circuit and how this may underlie prevalence of polysubstance use patterns among individuals who use drugs.

7. Found in Translation? Clinical Therapeutic Strategies to Treat Neuroimmune-Glutamate Dysregulations in SUD

Immunotherapeutics testing for SUD is a field in its infancy, with few clinical laboratory studies or trials having been conducted thus far to test efficacy of drugs that reverse drug-induced immune dysfunctions. Here, we make a case that specifically targeting immune system dysregulations based on an individual’s primary drug of use may yield positive clinical results with reductions in SUD-related outcomes including use and withdrawal severity. A case for utilizing immunotherapeutics for OUD is starting to be made³⁸⁰, although clinical studies are needed to identify potential efficacious therapeutics. We note that polysubstance use complicates the clinical treatment landscape and we identify strategies to guide clinical treatments for different SUDs based on drug class and whether an individual is undergoing withdrawal or actively using an addictive drug.

There are no therapeutics currently approved by the Food and Drug Administration (FDA) for CUD, although the preclinical addiction glutamate field has indicated that therapeutics which normalize glutamate dysfunction, such as NAC (which is also an antioxidant and can reduce inflammation³⁸¹), may be useful as novel treatments (e.g., see clinical trials showing some efficacy of NAC in reducing cocaine-related outcomes^81,382). As noted throughout this review, different addictive drugs influence the neuroimmune-glutamate circuit differently, and underscoring this is clinical evidence that NAC does not always show efficacy in reducing drug-related outcomes (e.g., for smoking³⁸³). Importantly, several positive clinical studies with NAC have required exclusion of individuals with complex medical histories and/or substance use trajectories⁸¹, demonstrating potential limited efficacy of NAC when polysubstance use is present. This is an especially important consideration in the current era of high polysubstance use prevalence^367,376. An emerging therapeutic under investigation is pentoxifylline (PTX), which is a methylxanthine derivative that has been tested as a potential CUD therapeutic. In the Cocaine Rapid Efficacy Screening Trial (CREST), PTX was the only compound to show some initial efficacy in reducing cocaine use (although this effect was not present after the first 4 weeks of the trial)³⁸⁴. As mentioned above, cocaine exposure increases immune markers including pro-inflammatory cytokines (TNFα¹⁰², IL-6, among others). Notably, PTX has been shown to reduce serum levels of TNFα, IL-8 and IL-6³⁸⁵, supporting the tenet that therapeutically targeting immune system activation when cocaine is actively being used may result in decreased cocaine use clinically. Other compounds that could be tested for cocaine use include IL-6 receptor inhibitors, which have been evaluated and FDA-approved for other conditions such as rheumatoid arthritis (e.g., sarilumab or Kevzara^®), as well as TNFα receptor inhibitors or monoclonal antibodies (e.g., infliximab, adalimumab, etanercept, golimumab, and certolizumab), IL-1β inhibitors (e.g., anakinra, canakinumab, and rilonacept), and/or TLR4 antagonists (e.g., resatorvid or TAK-242). One notable issue is that these compounds may not readily cross the blood-brain barrier (BBB) which is critical for medications to exert pharmacological effects and this could impact efficacy^386,387 in improving SUD-related outcomes, although studies have indicated weakening of BBB integrity by chronic drug use^265,388,389 which may result in greater ability of molecules to penetrate the BBB³⁹⁰. The BBB consists of both a dynamic biological interface of brain endothelial cells, smooth muscles, pericytes, astrocytes, and neurons that form the physical BBB, as well as a selective physicochemical barrier^391,392. Recent evidence has shown that addictive drugs alter tight junction formation and protein expression³⁸⁹ which may result in greater ability of molecules to penetrate the BBB such as medications or peripherally administered compounds that are usually blocked by the tight junctions to bypass the BBB. Conversely, increased BBB disruption could result in other substances or immune cells circulating in the blood having increased access to the brain parenchyma, creating risks of neurodegeneration and cognitive impairment^387,393. Thus, rather than protecting the brain, weakened barriers and systemic immunosuppression could result in a continuous, dysregulated entry of pro-inflammatory factors into the CNS, where immunosuppression could have broader and less predictable CNS consequences^394,395. Together, it should also be considered in clinical trials whether these compounds should be used as therapeutics depending on whether cocaine is being actively used or whether individuals are being treated during withdrawal, as blockade of immune markers during withdrawal may result in additional suppression of the immune system if cocaine is not present.

Preclinical evidence with methamphetamine indicates that it activates the neuroimmune system and can cause neuronal toxicity⁷¹ as well as activation of the inflammasome^69,396. However, these preclinical data are difficult to translate to clinical populations because additional constituents in methamphetamine supplies may contribute to cellular toxicity and complicate treatment outcomes. Methamphetamine use is a complex clinical issue because it can be adulterated with other constituents, making it impure³⁹⁷, and has recently been detected in the illicit opioid supply^376,398. Regardless, reversing immune system dysregulations induced by methamphetamine use and withdrawal may be a viable treatment avenue for methamphetamine use disorder (MUD). Akin to CUD, there are no FDA-approved medications for MUD, making this a critical area of focus for the field. As methamphetamine can increase IL-1β³⁹⁶, reducing pro-inflammatory markers that are elevated by methamphetamine may be a viable therapeutic strategy. In support, one clinical trial with 11 patients showed that treatment with the phosphodiesterase inhibitor ibudilast reduced both the subjective effects of methamphetamine³⁹⁹ and methamphetamine-induced inflammatory markers including soluble intercellular cell adhesion molecule-1 (sICAM-1) and soluble vascular cell adhesion molecule-1 (sVCAM-1)⁴⁰⁰, which are important in sustaining the integrity of the BBB and are expressed in inflammatory conditions⁴⁰¹. Additional studies are needed to determine if therapeutics that reverse methamphetamine-induced immune dysregulations during use and/or withdrawal may demonstrate clinically meaningful efficacy in reducing MUD-related outcomes.

Opioid therapeutics include medication-assisted therapies (MAT; e.g., methadone or buprenorphine⁴⁰²) as well as withdrawal management therapeutics such as clonidine and lofexidine⁴⁰³. While preclinical evidence indicates that pro-inflammatory neuroimmune signaling modulates opioid addiction processes including opioid reward, tolerance, and withdrawal symptomatology, little has been clinically evaluated with regard to immunotherapeutics for OUD⁴⁰⁴. Of the few tested, medications that suppress inflammatory markers (including cytokine and chemokine secretion) via the administration of glia modulators including Ibudilast^349,405,406, and minocycline (a tetracycline antibiotic)^407,408 have demonstrated some clinical efficacy. For example, ibudilast was well tolerated and showed therapeutic effects of reduced opioid withdrawal symptoms³⁴⁹, reductions in the development of opioid tolerance⁴⁰⁵, and attenuated positive subjective responses to oxycodone⁴⁰⁶. Medications have also been tested which selectively target individual cytokine signaling cascades rather than modulate glial activation. For example, one translational study demonstrated increased cytokine levels (specifically, CXCL1) in cerebral spinal fluid of cancer patients with documented opioid tolerance, an effect that was correlated with opioid intake and replicated in rats^326,409 (although the clinical comparison group was in patients who did not have a cancer diagnosis, making it difficult to decipher contributions of opioid tolerance versus cancer to the cytokine results). This study also found that a CXCL1-neutralizing antibody reduced opioid tolerance, raising the possibility that blocking this specific chemokine could be a viable treatment pathway for opioid tolerance in OUD. Although there are no well-established inhibitors that bind directly to CXCL1 itself, blocking its receptors (CXCR2/CXCR1) with a small-molecule inhibitor (e.g., with Reparixin⁴¹⁰, Sch527123⁴¹¹, or SB332235⁴¹²) would be an indirect way to determine if inhibition of this signaling pathway could be beneficial in the treatment of OUD-related outcomes. Given that CXCR2 antagonists that penetrate the BBB⁴¹³ have been developed, and there is evidence that these compounds can reduce inflammasome activation⁴¹⁴, development of compounds that target specific cytokine or chemokine signaling cascades for OUD therapeutics warrants further investigation. Evidence also indicates that inhibiting inflammatory targets in clinical populations with ongoing opioid use may be a viable treatment pathway. As preclinical evidence indicates that withdrawal from opioids is associated with increased inflammatory markers^324,356, clinical studies may need to test immunosuppressive therapeutics as withdrawal mitigation treatments.

Use of and/or exposure to nicotine-containing products, including combustible cigarettes and electronic nicotine delivery systems (ENDS) products, have known immune consequences in various tissue types including lungs⁴¹⁵, intestine⁴¹⁶, and brain^417,418. Further, cigarette smoking alters immunity by attenuating defensive immunity⁴¹⁹. Importantly, in vitro evidence shows that cigarette smoke can suppress TNFα in natural killer immune cells in the airway, as well as its downstream signaling pathway (NF-ĸB) that this cytokine activates⁴²⁰. There is clinical evidence that smoking status is immunomodulatory⁴²¹ as one study demonstrated that smoking status impairs immune functioning measured via positron emission tomography (PET) for the microglial marker translocator protein (TSPO; as compared to non-smokers). Together with our preclinical rat data indicating suppression of proinflammatory cytokines in the NA by nicotine self-administration⁶³, clinical studies should test compounds which can restore nicotine-induced disruptions in defensive immunity while individuals are actively using nicotine-containing products. One specific example that may yield promise is TNFα – we found this cytokine was the most suppressed by nicotine self-administration in our rat model⁶³. We also found that FKN was increased in the NA, indicating neuronal distress¹³². As both TNFα and FKN are cleaved by TACE, it is possible that nicotine induces specific dysfunction in TACE cleavage of FKN and TNFα, and that modifying TACE function may be a viable clinical pathway for treatment of nicotine use. While this remains untested both preclinically in relation to nicotine use, TACE inhibitors have been under investigation for inflammatory diseases⁴²² (e.g., TMI-005 (aprastat⁴²³) and BMS-561392 were tested in Phase II clinical trials⁴²⁴) and could thus be repurposed^425–428. We posit that this may be a viable path forward into clinical trials, however it is important to note that preclinical studies are first needed to determine if TACE inhibitors restore nicotine-induced immune dysregulations in the NA, as this may also further suppress TNFα and lead to increased nicotine use (although TACE inhibition increases IL-1β⁴²⁹, which we found is decreased after nicotine self-administration and thus inhibiting TACE may enhance immunity that is suppressed by nicotine by activating the inflammasome to allow for cleavage and activation of inflammatory caspases⁴³⁰). It is also important to note that there are sex differences in both immune function^431–434 and smoking-related outcomes. Indeed, the immune system plays a large role in sex-specific brain development, with endocrine-immune crosstalk playing an essential role in shaping sex differences in the brain⁴³⁵. Further, microglial function within the brain is sex-specific⁴³⁶ and regulated by steroid hormones^437,438 including progesterone and estrogen. Further, women have more difficulty maintaining abstinence⁴³⁹ and experience lower success rates with current smoking therapeutics⁴⁴⁰. Considering these sex differences, clinical studies are needed to determine whether our proposed opponent process framework applies uniformly across males and females. Together, these mechanistic interactions of nicotine and the immune system underscore the need for translational evaluations of neuroimmune therapeutics that may result in efficacious cessation of nicotine-containing products.

Throughout this section, we highlighted the few clinical studies that have evaluated therapeutics which have immunomodulatory properties as SUD treatments. We further outlined a strategy by which future clinical trials and human laboratory studies could implement experimental design strategies to test therapeutics depending on the drug type as well as whether an individual is in withdrawal or actively using addictive drugs. We note that this is a novel area of focus, given that there are few studies that have been conducted to evaluate immunotherapeutics in the context of SUDs. Additional clinical support for this approach could be found by evaluating large healthcare data sets. An example of this strategy found that bupropion prescription was associated with increased odds of remission from amphetamine use disorder in the Truven Marketscan database⁴⁴¹. A similar tactic could be used to determine whether prescription of any of the lead compounds identified here produced remission from SUD. Together, strategic and individualized targeting of the neuroimmune-glutamate addiction circuit could yield meaningful treatment outcomes for individuals with SUDs.

8. Conclusions

Here we lay out a novel hypothesis regarding how the neuroimmune system regulates each component of the glutamate homeostasis hypothesis of addiction as originally defined¹⁴. We further provide a framework for translational studies to evaluate novel immunotherapeutics as either drug use cessation or withdrawal mitigation agents, noting that this approach may need to be tailored for individuals with use disorders of different addictive drugs. A large majority of our hypotheses were based on a synthesis of in vitro or in vivo preclinical findings demonstrating that addictive drugs shift the immune system and the NA glutamate circuit, however, future research is needed to test each component of the addiction NA pentapartite synapse and inflammasome within the mesolimbic reward pathway as currently depicted in Figures 2 and 3. Future work must also systematically test this hypothesis as a function of drug use or withdrawal, sex differences, and depending upon the specific addictive drug under investigation. We posit that glutamate is heavily regulated by neuroimmune signaling and as such it is not possible to disentangle neuroimmune versus glutamate system impacts by addictive drugs or reversal by therapeutics. Thus, immunotherapeutics may exert downstream normalizations of glutamate dysregulations induced by addictive drug use, withdrawal, and relapse. Together, the NA neuroimmune-glutamate pentapartite circuit may be a viable target for future medications development testing for SUDs.

Highlights.

• Addictive drug use and withdrawal are regulated by a neuroimmune-glutamate pentapartite synapse within the nucleus accumbens.

• Drug use and withdrawal, with the exception of opioids, exert immunomodulatory effects on the brain reward pathway according to an opponent process.

• Addictive drug use and withdrawal activate an addiction inflammasome, which can lead to glutamate toxicity and neuronal death.

• Clinical therapeutic testing should leverage an opponent process framework to identify novel immunomodulators that could reverse drug-induced neuroimmune consequences and meaningfully reduce behaviors associated with substance use disorders.