Opioid withdrawal: role in addiction and neural mechanisms

Abstract

Withdrawal from opioids involves a negative affective state that promotes maintenance of drug-seeking behavior and relapse. As such, understanding the neurobiological mechanisms underlying withdrawal from opioid drugs is critical as scientists and clinicians seek to develop new treatments and therapies. In this review, we focus on the neural systems known to mediate the affective and somatic signs and symptoms of opioid withdrawal, including the mesolimbic dopaminergic system, basolateral amygdala, extended amygdala, and brain and hormonal stress systems. Evidence from preclinical studies suggests that these systems are altered following opioid exposure and that these changes mediate behavioral signs of negative affect such as aversion and anxiety during withdrawal. Adaptations in these systems also parallel the behavioral and psychological features of opioid use disorder (OUD), highlighting the important role of withdrawal in the development of addictive behavior. Implications for relapse and treatment are discussed as well as promising avenues for future research, with the hope of promoting continued progress toward characterizing neural contributors to opioid withdrawal and compulsive opioid use.

Introduction

Opioid use disorder (OUD) is a chronic relapsing disorder characterized by various symptoms, including unsuccessful efforts to decrease use, craving, continued use despite negative consequences, and the development of withdrawal symptoms that may be relieved by taking more of the drug (American Psychiatric Association, 2013). More than 26 million people worldwide are affected by OUD, including more than 80,000 fatal overdoses in the USA in 2021 (Vos et al. 2017; Centers for Disease Control and Prevention, 2022). While gold-standard treatments such as methadone maintenance therapy are effective when individuals are in treatment programs, relapse remains a serious concern after the completion of these programs (Bell and Strang 2020; Magura and Rosenblum 2001), underscoring the need for better treatments that prevent relapse long-term.

Critically, the characteristics of OUD exist in a repetitive cycle in which individuals suffering from addiction experience negative emotion (e.g., aversion, anxiety) followed by intense craving and temptation to use the drug again, which has the potential to induce relapse (Koob and Volkow 2010). Relief of negative affect is also thought to contribute to the addictive cycle via negative reinforcement mechanisms (Koob 2013). Understanding the neurobiological mechanisms that produce negative emotional signs and symptoms of opioid withdrawal is therefore an important foundation for properly treating patients with OUD.

The aversive symptoms of withdrawal from drugs of abuse oppose the initial rewarding, euphoric effects of drug use (Solomon and Corbit 1974; Koob et al. 1989). For opioids, this may include somatic signs such as body aches and chills, gastrointestinal upset, hyperalgesia, and affective signs such as anxiety, irritability, and reduced motivation for natural rewards (Koob et al. 1989; Pergolizzi et al. 2020). Importantly, this subsequent aversive process relies on some of the same neural circuits that mediate the euphoric and reinforcing effects of opioids (Stinus et al. 1990; Vargas-Perez et al. 2009; Radke et al. 2011). Affective signs of withdrawal from opioids are observed after one or a few exposures in humans (Longnecker et al. 1973; Jones 1980) and animals (Schulteis et al. 1997; Parker and Joshi 1998; Harris and Gewirtz 2004; Rothwell et al. 2012) and intensify in severity with additional drug exposure (Kenny et al. 2006; Edwards et al. 2012; Williams et al. 2012). Thus, as the cycle of addiction continues, relief of withdrawal is an important factor driving continuation of drug use (Weiss et al. 2001; Hutcheson et al. 2001; Koob and Volkow 2010; Piper 2015; Kosten and Baxter 2019).

A critical distinction in the motivation to continue using drugs despite the aversive symptoms of withdrawal is between somatic vs. affective withdrawal signs, and contemporary literature argues that it is primarily the affective symptoms of withdrawal that contribute to addiction rather than somatic signs (Koob and Volkow 2010). However, symptoms traditionally considered to be entirely physical also induce negative affect and as such may serve as powerful motivators to continue opioid use, particularly in the immediate term when withdrawal begins. Because of their critical contribution to the cycle of addiction, this review will focus on the neural systems known to mediate the signs and symptoms of both affective and somatic signs of opioid withdrawal from a circuit perspective. A robust literature suggests that the functions of mesolimbic dopaminergic system, basolateral amygdala, extended amygdala (i.e., central nucleus of the amygdala, bed nucleus of the stria terminalis, and nucleus accumbens shell), and brain and hormonal stress systems are altered following opioid exposure (Table 1). These changes mediate behavioral signs of negative affect such as aversion and anxiety during withdrawal and parallel the development of drug-seeking behavior. Additionally, peripheral opioid receptors, the locus coeruleus, and the periaqueductal gray are important mediators of somatic signs of opioid withdrawal.

Table 1.

Brain regions and neurotransmitter systems involved in opioid withdrawal behaviors

Region/system	Role in withdrawal	Citation
Mesolimbic DA	Decreased DA neuron firing and DA release, enhancing impact of phasic signal	Pothos et al. (1991); Diana et al. (1995); Kaufling and Aston-Jones (2015); Fox et al. (2017); George et al. (2022)
	Negative affective signs (e.g., aversion, anhedonia, anxiety)	Stinus et al. (1990); Kenny et al. (2006); Chartoff et al. (2009); Radke et al. (2011); Radke and Gewirtz (2012)
BLA	Association of withdrawal-induced negative affect with cues	Frenois et al. (2002); Hellemans et al. (2006); Lucas et al. (2008); Lyons et al. (2013); Franco-Garcia et al. (2022)
	Negative affective signs (e.g., anxiety)	Harris et al., (2006); Deji et al. (2022)
	Enhanced incentive value of rewards during withdrawal	Wassum et al. (2016)
CeA	Negative affective signs (e.g., aversion, anxiety), e.g., through CRH signaling	Heinrichs et al. (1995); Watanabe et al., (2002b); Harris et al. (2006); McNally and Akil (2002); Criner et al. (2007); Cabral et al. (2009)
BNST	Negative affective signs (e.g., aversion, anxiety) and stress-related behaviors, e.g., through NE signaling	Aston-Jones et al. (1999); Delfs et al. (2000); Gracy et al. (2001); Frenois et al. (2002); Veinante et al. (2003); Nakagawa et al. (2005); Harris et al. (2006)
	Activation of stress systems	Aston-Jones et al. (1999); Fuentealba et al. (2000); Song et al. (2020)
Adrenal stress hormones	Increased release of cortisol and NE from adrenals	Fuertes et al. (2000); Houshyar et al. (2003); Almela et al. (2012); Navarro-Zaragoza et al. (2021)
	Negative affective signs (e.g., place aversion)	Garcia-Perez et al. (2012); Garcia-Perez and Milanes (2020); Solecki et al. (2019)
LC	Somatic withdrawal signs	Rasmussen et al. (1990); Maldonado and Koob (1993); Maldonado et al. (1992a, b)
PAG	Somatic withdrawal signs	Maldonado et al. (1992a, b); Chieng et al. (1995)
	Withdrawal-induced pain sensitivity and aversion	Stinus et al. (1990); Bagley et al. (2005); Bagley et al. (2011)

BLA basolateral amygdala, BNST bed nucleus of the stria terminalis, CeA central amygdala, CRH corticotropin releasing hormone, DA dopamine, LC locus coeruleus, NE norepinephrine, PAG periaqueductal gray

Mesolimbic dopamine system

The mesolimbic dopamine system consists primarily of dopamine (DA) neurons in the ventral tegmental area (VTA) that project to various limbic structures, including the amygdala and nucleus accumbens (NAc) (Kiyatkin 1995). Early studies demonstrated that animals robustly respond for intracranial stimulation of the VTA (Olds and Milner 1954; Crow 1972) and NAc (Phillips and Fibiger 1980; Cole et al. 2018) and since then an abundance of research has demonstrated the significance of the mesolimbic DA system in motivated behavior. The functions ascribed to the mesolimbic DA system are vast and varied, including roles in learning and plasticity, salience, value, effort, and choice (Wise 2004; Keiflin and Janak 2015; Berke 2018; Piantadosi et al. 2021). Reward prediction error theory, which is based on the observation that DA neuron firing rates are increased following unexpected presentation of a reward or reward-predictive cue (Schultz et al. 1993), has been particularly influential in the addiction field. Because all major drugs of abuse acutely increase extracellular DA levels in the NAc (Di Chiara and Imperato 1988), some have argued that inflated reward prediction errors in the presence of drugs enhance learning about drugs and drug-paired cues (Ahmed 2004; Redish 2004; Keiflin and Janak 2015). Sensitization of neural responses to drug-paired cues in turn elicits drug-seeking behavior, often at the expense of other available rewards and long after drug use has ceased (Robinson and Berridge 2001).

Because the VTA and NAc support the positive reinforcing effects of opioids (Koob and Volkow 2010), it is not surprising that affective signs of withdrawal arise from compensatory mechanisms in these same structures. Observations of increased cFos expression in the NAc shell during opiate withdrawal support this idea (Walters et al. 2000; Gracy et al. 2001). Further, infusion of an opioid receptor antagonist into the NAc of dependent animals induces conditioned place aversion (Stinus et al. 1990; Carlezon and Thomas 2009). A state of anxiety-like behavior can also be observed during spontaneous withdrawal from an intra-VTA infusion of morphine (Radke et al. 2011). In general, withdrawal from opioids is associated with decreases in mesolimbic DA neuron firing and DA release (Pothos et al. 1991; Diana et al. 1995; Fox et al. 2017; George et al. 2022). Reductions in DA signaling disrupt baseline reward sensitivity, as evidenced by increases in intracranial self-stimulation (ICSS) thresholds that occur during withdrawal (Kenny et al. 2006), and produce behavioral signs of withdrawal such as place aversion and increases in the startle reflex that can be reversed with infusions of DA receptor agonists into the NAc (Chartoff et al. 2009; Radke and Gewirtz 2012).

Intracellularly, the mu-opioid receptor (MOR) exerts an inhibitory influence on the membrane potential via coupling to the Gi intracellular signaling pathway, which inhibits adenylyl cyclase activity (Chakrabarti et al. 1995). As a compensatory mechanism, activity of adenylyl cyclase and other components of the cAMP signaling pathway are upregulated (Sharma et al. 1975; Chan and Lutfy 2016). This causes increased “rebound” firing in MOR-expressing VTA GABAergic neurons during the opioid-withdrawn state (Bonci and Williams 1997). Greater inhibition from these MOR-expressing GABAergic cells in the VTA therefore results in decreases in DA neuron activity during withdrawal (Diana et al. 1995; 1999). Projections from MOR-expressing neurons outside of the VTA, for example, in the medial habenula (Boulos et al. 2020) and rostromedial tegmental nucleus (Kaufling and Aston-Jones 2015; Bobzean et al. 2019), also contribute to modulations of DA neuron firing during withdrawal. Another potential regulator of DA release during withdrawal are the cholinergic interneurons of the striatum, which modulate DA release by inducing action potentials in the axons of DA neurons (Liu et al., 2022). Cholinergic interneurons express MORs (Ponterio et al. 2013; Mamaligas et al. 2016; Collins et al. 2019), and acetylcholine release is increased during precipitated withdrawal (Rada et al. 1991, 1996).

Decreases in baseline sustained (i.e., tonic) DA neuron activity that occur during opioid withdrawal (Kaufling and Aston-Jones 2015; George et al. 2022) may serve to make opioid-induced changes in phasic DA signaling more salient during withdrawal. In other words, withdrawal may be critical to the development of addiction because it enhances the “signal-to-noise” ratio of the DA response (Wanat et al. 2009; Zhang et al. 2012). This enhanced phasic DA signal that occurs after reexposure to drugs or drug-paired cues under withdrawal conditions may serve as a critical mechanism in the motivation to continue drug use.

In addition to reductions in DA levels, opioid withdrawal is characterized by plasticity in glutamatergic signaling in the NAc, originating from cortical, thalamic, or amygdalar inputs (Zhu et al. 2016; Hearing et al. 2018). In morphine withdrawal-exposed striatal cells, evidence shows increased phosphorylation of the AMPA receptor subunit GluR1 and increased AMPA receptor-mediated currents (Chartoff et al. 2003). Additionally, experimentally elevating GluR1 levels in the NAc enhances drug-induced CPA and AMPA receptor antagonist infusion in the NAc shell of morphine-dependent rats prevents naloxone-induced withdrawal CPA (Carlezon and Thomas 2009). Together, these findings suggest that heightened striatal glutamate is another important mediator of withdrawal-induced aversion (Russell et al. 2016).

Basolateral amygdala

The basolateral amygdala (BLA) signals the value of environmental stimuli, and as such has an important role in associative processes necessary for emotional learning (Wassum and Izquierdo 2015). Through its connections with areas such as the thalamus and sensory cortices, the BLA receives information about the external environment and guides behavior through its connections with the central amygdala (CeA), NAc, dorsal medial striatum, and bed nucleus of the stria terminalis (BNST) (Weller and Smith 1982; van Vulpen and Verwer 1989; McDonald 1998; Ambroggi et al. 2008; Vertes et al. 2015; Kim et al. 2017). The role of BLA in learning about aversive and anxiogenic stimuli is well documented (Maren 2003; O’Neill et al. 2018), so it is not surprising that conditioned opioid withdrawal memories are stored here (Frenois et al. 2002; Hellemans et al. 2006; Lucas et al. 2008; Lyons et al. 2013; Franco-García et al. 2022). It is important to note that this function of associating the negative emotional effects of withdrawal with previously neutral cues is likely the primary role of the BLA in opioid withdrawal, rather than mediating aversion itself. As evidence of this, BLA lesions prevent the acquisition of conditioned withdrawal behavior (i.e., suppression of a food-seeking response) in morphine-dependent rats but do not eliminate the acute expression of withdrawal signs (Schulteis et al. 2000). Precipitated withdrawal alone (without conditioning) increases cFos expression in a number of limbic brain regions, but not in the BLA (Frenois et al. 2002). Further, lesions of BLA reduced but did not eliminate acquisition of morphine withdrawal-induced CPA, suggesting that this manipulation does not prevent withdrawal-induced aversion (Watanabe et al. 2002b).

The BLA-dependent mechanisms underlying conditioned opioid withdrawal include a number of neurochemical systems and connections with cortical and hippocampal circuits. Reduced DA transmission in the BLA is correlated with the magnitude of withdrawal-induced place aversion (García-Pérez and Milanés 2020). BLA projections to the prelimbic cortex express cFos in response to withdrawal-induced CPA and optogenetically inhibiting these neurons eliminates expression of CPA (Song et al. 2019). Similarly, contextual fear conditioning is stronger when performed in a context previously paired with opioid withdrawal, and this effect involves projections from prelimbic cortex to BLA (Seno et al. 2022). Recruitment of BLA during opioid withdrawal also involves the corticotropin releasing hormone (CRH) system, as a CRH receptor 1 antagonist prevents acquisition of morphine CPA and withdrawal-induced increases in BDNF expression in this region (Martínez-Laorden et al. 2020).

Although studies using place conditioning have found that BLA’s role is primarily an associative one, other studies have implicated this structure in opioid withdrawal-induced defensive behaviors. Pharmacological inactivation of the BLA prevents elevations of the startle reflex, a behavioral index of anxiety-like behavior, during withdrawal from a single morphine exposure (Harris et al. 2006). Chemogenetic inhibition of BLA inputs and projections to the ventral hippocampus have also recently been shown to attenuate anxiogenic behaviors as measured on the elevated plus maze and open field test 2 weeks after cessation of chronic opioid exposure (Deji et al. 2022).

In addition to being required for forming memories of withdrawal, the BLA may mediate enhancements in the incentive value of rewards experienced during opioid withdrawal. Receipt of reward during opioid withdrawal enhances reward-seeking in rats (Hutcheson et al. 2001) and an elegant demonstration by the Wassum lab established that this effect is blocked by the inactivation of MORs in the BLA (Wassum et al. 2016). These BLA MORs appear to be necessary for recall of outcome-specific reward memories, and their recruitment by exogenous opioids likely plays an important role in promoting drug taking behaviors (Lichtenberg and Wassum 2017). Another intriguing study in mice exposed to chronic morphine and withdrawal suggests that this treatment enhances learning about fearful stimuli by enhancing GluA1-dependent plasticity in the BLA (Pennington et al. 2020). Thus, the role of withdrawal in shaping associated learning processes in the BLA appears to apply to both aversive and appetitive stimuli.

Central amygdala

The central amygdala (CeA) mediates the expression of behavioral output in a wide range of circumstances, both aversive and appetitive, and plays an important role in action selection and incentive motivation (Fadok et al. 2018; Warlow and Berridge 2021). This latter function of the CeA is particularly relevant to the maladaptive seeking responses that characterizes OUD. CeA plasticity induced by repeated cycle of opioid exposure and withdrawal may narrow and intensify behaviors directed toward drugs and drug-paired cues. In support of this idea, CeA has been shown to be necessary for the expression of conditioned and unconditioned negative emotional signs of opioid withdrawal.

Lesioning the CeA significantly reduces morphine withdrawal-induced CPA in rats (Watanabe et al. 2002b), suggesting that this nucleus is critical for the production of avoidance responses during withdrawal. The central importance of the CeA in withdrawal-induced aversion is also supported by observations of cFos expression and increases in CRE-mediated transcription in this region after morphine withdrawal (Stornetta et al. 1993; Gracy et al. 2001; Shaw-Lutchman et al. 2002; Frenois et al. 2002; Veinante et al. 2003; Baidoo et al. 2021) and the finding that CPA and suppression of operant responding for food can be precipitated following systemic morphine injection via intra-CeA infusion of a MOR antagonist (Heinrichs et al. 1995; Criner et al. 2007). Anxiety-like behavior, measured as startle potentiation, during acute opioid withdrawal is also prevented when the CeA is pharmacologically inhibited (Harris et al. 2006; Cabral et al. 2009).

A number of neurochemical systems in the CeA have been implicated in opioid withdrawal. Withdrawal-induced CPA is attenuated by intra-CeA infusion of glutamate receptor antagonists (Watanabe et al. 2002a; Ishida et al. 2008) and targeted deletion of NMDA receptors (Glass et al. 2008). The GluA1 subunit of the AMPA receptor is upregulated in the CeA during morphine withdrawal and suppression of GluA1 prevented elevated morphine-seeking behavior in withdrawn rats (Hou et al. 2015) (although this same treatment enhanced morphine withdrawal CPA, see Cai et al. 2020). Increased basal GABAergic transmission has also been demonstrated during opioid withdrawal, an effect that is mediated by activations in cAMP signaling (Bajo et al. 2014; Guo et al. 2019). Specifically, firing of CeA parvalbumin inhibitory interneurons, which regulate the activity of principal CeA neurons, is increased during withdrawal from morphine. Further, optogenetic inhibition of these neurons reduced expression of withdrawal-induced CPA, avoidance of the open arms on the elevated plus maze, and anhedonia in a saccharin preference test, whereas stimulation increased these affective withdrawal signs (Wang et al. 2016).

An important player in the CeA’s response to opioid withdrawal is the neuropeptide CRH. CRH receptor blockade in the CeA reduces CPA and lesion of CeA CRH-containing neurons attenuates expression of conditioned operant suppression of food seeking (Heinrichs et al. 1995; McNally and Akil 2002), suggesting an important role for CRH projections to and from this nucleus. In addition to attenuating negative affective signs of withdrawal, inhibition of the parvalbumin interneurons described above reduces upregulated levels of CRH mRNA in the CeA (Wang et al. 2016). There is also evidence to suggest that reciprocal connections between the VTA and CeA CRH neurons are required for opioid withdrawal-induced CPA. 6-Hydroxydopamine lesions of the CeA-projecting VTA DA neurons impair the formation of a place aversion during withdrawal (Xu et al. 2012). Further, CRH-containing neurons that project to the VTA are activated during withdrawal and participate in the formation of withdrawal-induced CPA (Jiang et al. 2021). Considering that stimulation of CRH neurons in the CeA has been shown to increase incentive motivation (Baumgartner et al. 2021), activation of this system during opioid withdrawal may serve as a neural substrate for the intense, maladaptive seeking behavior that characterizes addiction.

Bed nucleus of the stria terminalis

The role of the BNST, which is also part of the extended amygdala, in opioid withdrawal fits with its known role in mediating sustained defensive responses and stress-related behaviors (Lebow and Chen 2016). The BNST makes connections with brain regions already discussed, such as the amygdala, NAc, and VTA and projects to and influences neurons in the paraventricular nucleus of the hypothalamus that regulate the hypothalamic–pituitary–adrenal (HPA) axis (Song et al. 2020). Afferent projections from the locus coeruleus and other noradrenergic nuclei in the brainstem are also present here and play an important role in BNST activity during opioid withdrawal.

As evidence of the BNST’s involvement in opioid withdrawal, cFos is enhanced in this region during withdrawal and is correlated with the level of withdrawal-induced place aversion (Gracy et al. 2001; Frenois et al. 2002; Veinante et al. 2003; Nakagawa et al. 2005). Infusion of a MOR antagonist into the BNST also precipitates withdrawal, as measured by suppression of operant responding for food (Criner et al. 2007). Inactivation of BNST reduces the magnitude of withdrawal-potentiated startle during acute opioid withdrawal (Harris et al. 2006) and BNST lesions reduce withdrawal following chronic morphine exposure (Nakagawa et al. 2005). BNST neurons are more excitable following a history of opioid dependence (Francesconi et al. 2017), and this may result from changes in local inhibitory signaling. For example, systemic treatment with the GABA-B receptor agonist baclofen reduces morphine withdrawal-induced anxiety-like behavior and prevents decreases in BNST brain derived neurotrophic factor (BDNF) (Pedrón et al. 2016). Changes in BNST GABAergic signaling have also been observed during morphine withdrawal in mice, although the direction of the effects diverged in males and females (Luster et al. 2020). Activation of BNST during withdrawal appears to be mediated through projections from the CeA, as CeA lesions prevent withdrawal-induced increases in cFos immunoreactivity in the BNST (Nakagawa et al. 2005).

The BNST has the most significant density of noradrenergic inputs over any other brain region, and BNST norepinephrine (NE) signaling is important in mediating negative affect during opioid withdrawal. Extracellular levels of NE are increased in the ventral BNST after chronic morphine treatment (Fuentealba et al. 2000). Additionally, enhanced NE increases withdrawal-induced place aversion, whereas inhibiting NE activity blocks this behavior (Brownstein and Palkovits 1984; Aston-Jones et al. 1999; Delfs et al. 2000). The effects of NE are mediated by β-adrenergic receptors, as suggested by findings that withdrawal-induced cFos expression in the BNST is reduced by pretreatment with a β-adrenergic receptor antagonist and that intra-BNST β receptor antagonism reduces withdrawal-induced CPA (Aston-Jones et al. 1999). A major source of BNST NE during withdrawal are the brainstem nuclei that project via the ventral noradrenergic bundle, i.e., the A1 and A2 cell groups (Aston-Jones et al. 1999; Delfs et al. 2000).

Stress systems

Opioid withdrawal is known to induce a robust peripheral stress response through activation of the HPA axis and the sympathetic nervous system. Release of cortisol and NE from the adrenal glands circulates systemically and impacts circuits throughout the central nervous system. Thus, activation of stress systems has the potential to impact all of the circuits discussed above and more.

The production and secretion of glucocorticoids (e.g., cortisol in humans and corticosterone in rodents) is controlled by the HPA axis. Neurons in the paraventricular nucleus of the hypothalamus (PVN) release CRH into the anterior pituitary, which releases adrenocorticotropic hormone (ACTH) into the bloodstream to trigger glucocorticoid secretion from the adrenal cortex (Fulford and Harbuz 2005; Watson and Mackin 2006; Glynn et al. 2013). During withdrawal from chronic morphine treatment, corticosterone and ACTH levels are increased in rodents (Fuertes et al. 2000; Houshyar et al. 2003; Almela et al. 2012; Navarro-Zaragoza et al. 2021). Additionally, formation of a withdrawal-induced place aversion and induction of delta FosB expression in NAc, BNST, CeA, and PVN is impaired following adrenalectomy (García-Pérez et al. 2012, 2016; García-Pérez and Milanés 2020), although somatic withdrawal signs remain intact. Clinical findings also suggest increased HPA activity in humans during withdrawal. Increased cortisol has been observed during spontaneous and naloxone-precipitated withdrawal in individuals who are dependent on opioids (Tennant et al. 1991; Culpepper-Morgan and Kreek 1997; Nava et al. 2006; Zhang et al. 2008).

HPA axis activation results from activity in the PVN during withdrawal, as evidenced by cFos expression in CRH-containing neurons and increases in CRH mRNA in the PVN (McNally and Akil 2002; Houshyar et al. 2003; Hamlin et al. 2004), although decreases in CRH mRNA have also been observed (Milanés et al. 2002). The important role that CRH plays in withdrawal has also been demonstrated in mice with genetic deletion of CRH-1 or CRH-2 receptors (Contarino and Papaleo 2005; Ingallinesi et al. 2012; García-Carmona et al. 2015a) and through systemic pharmacological blockade of CRH receptors (Stinus et al. 2005; Skelton et al. 2007). It is important to note here that systemic manipulations such as these cannot differentiate between the actions of CRH in the HPA axis and in other extra-hypothalamic brain regions such as the CeA. Noradrenergic projections from the A1 and A2 cell groups are one mechanism through which PVN CRH neurons are recruited during withdrawal. Direct noradrenergic projections to PVN are activated during withdrawal from chronic morphine exposure (Fuertes et al. 2000; Benavides et al. 2003). The BNST is also critical in regulating PVN CRH activity (Forray and Gysling 2004), meaning that NE release here likely also contributes to HPA axis activation during opioid withdrawal, although this has yet to be demonstrated.

Despite robust evidence for glucocorticoid release during opioid withdrawal and the impacts of adrenalectomy on place aversion mentioned above (García-Pérez et al. 2016; García-Pérez and Milanés 2020), blockade of glucocorticoid receptors (GR) during withdrawal has been found ineffective in modifying affective signs of withdrawal (Solecki et al. 2019). The adrenal gland is also an important source of systemic epinephrine/norepinephrine that is activated by sympathetic projections from the spinal cord, which receive inputs from central nervous system nuclei, including the PVN (Hosoya et al. 1991). In human patients, naloxone-precipitated withdrawal increases plasma concentrations of epinephrine and NE (Kienbaum et al. 1998), which may underlie some of the effects of adrenalectomy on withdrawal-induced aversion. Systemic treatment with β adrenergic receptor antagonists attenuates affective signs of opioid withdrawal (Solecki et al. 2019), but much like systemic studies of the CRH system, it is difficult to ascribe these effects to sympathoadrenal vs. central mechanisms. Thus, the exact mechanisms through which the HPA axis contributes to the affective component of withdrawal have yet to be fully clarified.

Neural circuits mediating somatic withdrawal signs

Some of the most prominent aspects of the opioid withdrawal syndrome include the somatic signs and symptoms. In humans, the somatic withdrawal syndrome begins within 24 h of discontinuing opioids and reaches its peak within the next few days, though this time course can vary due to individual differences and whether an individual is discontinuing short or long-acting opioids (Dunn et al. 2019). Typical somatic signs of withdrawal include gastrointestinal issues such as diarrhea and nausea and vomiting, changes in pain sensitivity including muscle and joint aches and hyperalgesia, hot and cold flashes, piloerection, pupil dilation, lacrimation, rhinorrhea, and yawning (Bradley et al. 1987; Dunn et al. 2019). Many of these signs can be measured in rodent models of opioid withdrawal, and species-specific signs can be observed as well. Scales of somatic withdrawal in rodents frequently include signs such as jumping, rearing, wet dog shakes, abdominal constrictions, diarrhea, salivation, hunched posture, and teeth chattering (Gellert and Holtzman 1978; Maldonado et al. 1992a, b; Maldonado and Koob 1993).

In general, somatic signs and symptoms of opioid withdrawal can be attributed to compensatory rebound activation and neuronal hyperexcitability following opioid-induced inhibition. Peripheral MORs are found in the terminals of afferent sensory neurons and in enteric neurons (Stein 2013; Galligan and Akbarali 2014), and these receptors are responsible for the gastrointestinal effects and other peripheral signs such as lacrimation, rhinorrhea, and salivation (Bianchetti et al. 1986; Maldonado, Negus, Koob, 1992a). For example, withdrawal-induced diarrhea was shown to be dependent on peripheral opioid receptors by administration of the peripherally selective opioid antagonist SR 58,002 C, which induces diarrhea, but not jumping, in morphine-dependent mice (Bianchetti et al. 1986). Similarly, the widely available over-the-counter anti-diarrheal medication, loperamide, primarily reduces diarrhea by activating MORs in the gut (Kang et al. 2016).

Within the central nervous system, MOR antagonism in a wide array of brain regions precipitates withdrawal signs in rodents, with especially strong effects observed in two contiguous regions of the brainstem, the periaqueductal gray (PAG) and locus coeruleus (LC) (Maldonado et al. 1992a, b; Koob et al. 1992). Infusion of the opioid receptor antagonist methylnaloxonium into the PAG produced signs such as teeth chatter, jumping, rearing, piloerection, and loss of body weight in rats (Maldonado et al. 1992a, b) as well as severe agitation and development of a conditioned place aversion (Stinus et al. 1990). Manipulation of protein kinases in this region also attenuates withdrawal behaviors (Maldonado et al. 1995; Punch et al. 1997). Further, PAG neurons express cFos following precipitated withdrawal (Chieng et al. 1995) and have been shown to develop electrophysiological tolerance and withdrawal following opioid exposure (Chieng and Christie 1996). This sensitivity of the PAG to withdrawal is not surprising considering the large number of opioid receptors expressed in this region and a well-characterized role in opioid-induced analgesia via descending projections to the dorsal horn of the spinal cord. Because of its important role in modulation of pain, the PAG is also the primary mediator of the enhanced pain sensitivity characteristic of withdrawal. Withdrawal-induced hyperexcitability in GABAergic interneurons of the PAG reduces the ability of PAG output neurons to suppress ascending pain signals (Hack et al. 2003; Bagley et al. 2005, 2011).

Another brain region that has received a great deal of attention for its role in somatic signs of opioid withdrawal is the LC, the primary source of NE in the brain. Rasmussen and colleagues demonstrated that the time course of multiple somatic withdrawal signs paralleled increased activity in the LC (Rasmussen et al. 1990). Increased cFos activity has also been observed in the LC during precipitated withdrawal (Hayward et al. 1990; Alvarez-Bagnarol et al. 2022). Increases in LC activity are mediated by excitatory inputs from the nucleus paragigantoceullularis and orexinergic inputs from the lateral hypothalamus in addition to MOR antagonism (Rasmussen and Aghajanian, 1989; Akaoka and Aston-Jones, 1991; Aghajanian et al. 1994; Hooshmand et al. 2019). Maldonado and Koob showed that lesioning the LC in dependent rats reduces a wide breadth of somatic withdrawal signs, including jumping, hyperactivity, and wet dog shakes (Maldonado and Koob 1993). Additionally, methylnaloxonium injection directly into the LC enhances somatic withdrawal signs in dependent rats (Maldonado et al. 1992a, b) while intra-LC AMPA receptor antagonism (Rasmussen et al. 1996; Taylor et al. 1998), GABA-B receptor agonism (Riahi et al. 2009), and orexin-1 receptor antagonism (Azizi et al. 2010) reduces their expression. A role for noradrenergic involvement in withdrawal symptoms is further demonstrated in clinical data with α2 adrenergic agonists demonstrating promise for reduction of withdrawal signs and improvement of treatment outcomes in addicted individuals (Gowing et al. 2016). Despite this robust evidence suggesting a role for the LC, studies using neurochemical lesion techniques to target LC noradrenergic neurons have failed to find a role for these neurons in the somatic component of withdrawal (Chieng and Christie, 1995; Christie et al. 1997; Caillé et al., 1999). The precise role of the LC in somatic signs of withdrawal therefore remains an unresolved issue (Williams et al. 2001).

Discussion

Taken together, the evidence reviewed suggests that opioid withdrawal results in severely dysregulated reward and stress systems. Because preclinical studies in rodents do not provide insight into an animal’s affective consciousness it is impossible to conclude that brain regions such as the amygdala or NAc mediate the subjective emotional experience of withdrawal. Still, the reward and stress systems reviewed above can be said to work together to produce behaviors that represent negative emotional states, such as aversion, anhedonia, and anxiety. Further, neural changes in these circuits parallel the psychological phenomena of OUD, such as an unusual focus on drug reward at the expense of natural rewards and self-medication or relapse in response to anxiety or stress.

Role in addiction

A number of theoretical frameworks have been proposed to explain the psychology of addictive behavior, and there is ongoing debate over the relative contribution of negative affect, habit, decision-making processes, and incentive motivation to maladaptive drug seeking (Field and Kersbergen, 2020; Hogarth 2020; Epstein 2020). Relief of withdrawal has long been seen as a key driver of continued drug use although thinking about the nature of this contribution has shifted over time (Wikler 1948; Dole et al. 1966; Koob and Volkow 2010). One important observation that strongly supports the supposition that withdrawal is critical to initiating and maintaining the cycle of addiction is the observation that opioids are most likely to be abused when administered intermittently (vs. continuously). For example, rapidly delivered and short-acting heroin has a greater abuse liability than long-acting methadone (Stimmel and Kreek 2000). Preclinical markers of addiction such as psychomotor sensitization and opioid self-administration are similarly observed following intermittent opioid exposure only (Vanderschuren et al. 1997; Rothwell et al. 2010; Yu et al. 2014; Lefevre et al. 2020; Fragale et al. 2021). As such, withdrawal, or perhaps more specifically reexposure to opioids during withdrawal, drives the development of addictive behaviors by creating an altered neural state that contributes to maladaptive drug-induced plasticity. These withdrawal-induced changes in the function of the mesolimbic dopamine system, BLA, CeA, and BNST consequently set the stage for opioids and opioid-paired cues to become more salient, valued, and motivating than other stimuli.

An important line of evidence suggesting a role for opioid withdrawal in promoting addiction comes from the experiences of patients with OUD. For example, a study of patients seeking treatment for prescription opioid use found withdrawal avoidance to be the number one factor motivating current use (Weiss et al. 2014) and multiple other studies have found at least a partial role for withdrawal when examining factors that motivate opioid use (Heiwe et al. 2011; Harocopos et al., 2016; Stumbo et al. 2017; Frank et al. 2016; Cicero and Ellis 2017; AbdelWahab et al. 2018). Fear of withdrawal also contributes to risky decision making and continued use despite negative consequences. Individuals with OUD report making risky medical decisions (Summers et al. 2018) and avoiding treatment for their drug use (Mitchell et al. 2009) over concerns related to opioid withdrawal.

The preclinical literature also supports a role for withdrawal in motivating opioid consumption, although the literature is mixed. Injection of naloxone in morphine-dependent rats increases responding for heroin at low doses (e.g., 0.01 mg/kg), but higher doses of naloxone (e.g., 0.03 mg/kg) do the opposite (Carrera et al. 1999). Stimuli conditioned to acutely precipitate heroin withdrawal can also stimulate heroin self-administration behavior in rats and blunt the sensitivity of the brain reward system, as measured with ICSS (Kenny et al. 2006). Experience with heroin during withdrawal has also been shown to be necessary for withdrawal to motivate future drug seeking in rats (Hutcheson et al. 2001). Work in rhesus monkeys has further demonstrated that heroin (vs. food) choice increases following cessation of 21-h heroin access (Negus, 2006; Negus and Rice 2009). A number of pharmacological manipulations that reduce signs of opioid withdrawal reduce self-administration behaviors, including reinstatement, in rats, and these manipulations can include a wide range of pharmacological targets including α1 adrenergic receptors (Greenwell et al. 2009), CRF (Park et al. 2015), PPAR-γ, (de Guglielmo et al. 2017), dopamine D3 receptors (de Guglielmo et al. 2019), and 5α-Reductase (Bosse et al. 2021). However, drugs targeting withdrawal signs did not reduce heroin choice in monkeys (Negus and Rice 2009) and have not yet successfully translated to the clinic. In contradiction to findings supporting a role for withdrawal in opioid consumption, another line of evidence suggests that opioid withdrawal-induced anhedonia (i.e., increases in ICSS thresholds) is associated with reduced morphine self-administration behaviors (Holtz et al. 2015; Swain et al. 2020). Examinations of the influence of withdrawal on heroin seeking also fail to find an effect of naloxone injection on reinstatement of lever pressing (Shaham and Stewart 1995; Shaham et al. 1996).

While early theories postulated a role for relief of physical signs of withdrawal in addiction (Wikler 1948; Dole et al. 1966), views have shifted to consider the affective signs and symptoms of withdrawal to be of primary motivational significance (Koob and Volkow 2010). One reason for this distinction is that while the somatic component of opioid withdrawal resolves in a few days to weeks, affective signs can persist much longer. Additionally, the low abuse liability of many pharmacologically active substances that nonetheless produce tolerance and a state of withdrawal upon abstinence (e.g., caffeine) argues against the idea that physical dependence alone is sufficient for addiction (Heinz et al., 2020). Indeed, opioids themselves can produce opioid withdrawal syndrome and opioid dependence separately from OUD (Kosten and Baxter 2019; Ballantyne et al. 2019). Preclinical work also suggests a dissociation in the neural substrates mediating the affective and somatic components of withdrawal (see mechanisms reviewed above; Frenois et al. 2002) and that somatic signs are not predictive of affective withdrawal or opioid self-administration behaviors (Mucha 1987; Swain et al. 2020). That said, while affective withdrawal signs may be more reflective of neural plasticity in circuits responsible for addictive behavior, somatic signs should not be completely dismissed when considering the role of withdrawal in addiction.

One reason to consider a role for somatic withdrawal in OUD is that physical and emotional signs are not entirely separable. Although not typically life-threatening, acute somatic withdrawal signs such as aches and pains, nausea and diarrhea, and hot/cold flashes are severely unpleasant (Dunn et al. 2019). As such, the symptoms of opioid withdrawal are thought to resemble a severe flu-like illness (Farrell 1994) that individuals are highly motivated to avoid (Summers et al. 2018). These symptoms therefore construct a highly salient event composed of multiple aversive physical symptoms, and there is evidence to suggest that at least some addicted individuals continue drug use to avoid these symptoms. For instance, opioid-dependent patients cite a fear or concern about increased pain sensitivity as one of the primary reasons why they wish not to experience withdrawal (Stumbo et al. 2017; Frank et al. 2016). Because many of the symptoms of opioid withdrawal are uncomfortable and painful, it is not possible to categorize them as purely “physical.” There may also be overlap between the physical and emotional aspects of withdrawal as anticipation of somatic withdrawal signs produces distress and anxiety (Bruneau et al. 2021). It should further be noted that there is overlap in the neural structures demonstrated to produce affective and somatic signs of withdrawal in animal models. For example, the dopaminergic system contributes to somatic withdrawal signs in rats (Harris and Aston-Jones 1994; Chartoff et al. 2006) and methylnaloxonium infused into the PAG of morphine-dependent rats produces a place aversion (Stinus et al., 1990). Therefore, it is important to recognize that physical dependence also contributes to negative affect and that avoidance of these symptoms is in fact a powerful influence in maintaining opioid use.

Implications for relapse and treatment

In the early stages of abstinence from opioids, negative emotional signs and symptoms of withdrawal contribute to the motivation to continue drug use as individuals take drugs to avoid the associated pain and dysphoria. As discussed above, exposure to drugs during this state promotes the neural plasticity that underlies compulsive use and drug craving (Koob 2020). Although this acute phase of withdrawal is a temporary state, persistent emotional dysregulation and sensitization of pain and stress systems far beyond the detoxification stage leave individuals vulnerable to opioid craving months or even years after abstinence is achieved (Sinha 2009; Koob 2020). Cues conditioned to withdrawal can also trigger craving and promote renewed drug taking (Pantazis et al. 2021). Thus, the state of protracted withdrawal is an important contributor to relapse in individuals recovering from OUD.

In animals, relapse is modeled by examining reinstatement of drug seeking following exposure to drugs, drug-paired cues, stress, or withdrawal-paired cues (Shaham and Stewart 1995; Kenny et al. 2006; Mantsch et al. 2016; Bossert et al. 2019). It is notable that many of the neural circuits and neurotransmitters implicated in affective withdrawal from opioids have also been identified as participants in reinstatement, particularly when it is induced by stress (for a thorough review of the stress-induced reinstatement model see Mantsch et al. 2016). These overlapping neural circuits include the CeA, BNST, VTA, and NAc and within these roles for NE, CRH, DA, and glutamate have been found. Precipitators of relapse such as stress consequently engage the same neural circuits that are sensitized during repeated cycles of drug exposure and withdrawal. Thus, withdrawal is critical in setting the stage for relapse to be initiated by pain, stress, and cues in the long term.

The complexity and interconnectivity of the circuits recruited during opioid withdrawal may explain, at least in part, the difficulty of treating OUD and preventing relapse. Some of the most successful treatments available today help patients achieve abstinence by avoiding the withdrawn state. For instance, pharmacological maintenance therapies such as methadone and buprenorphine are long-acting opioid receptor agonists (Joseph et al. 2000). The chronic, stable levels of opioid receptor agonism provided by these pharmacotherapies prevent the highs and lows associated with opioid use, including the acute and protracted withdrawal states. Of course, despite the success and critical importance of pharmacological maintenance therapies to current OUD treatment strategies, a limitation of these treatments is that they do not correct the long-lasting brain circuit alterations that come along with addiction and influence complex phenomena such as salience, value, or motivation. For example, methadone maintenance significantly reduces heroin reinstatement after a priming injection of heroin; however, that study saw no effect of methadone maintenance on stress-induced reinstatement (Leri et al. 2004). Furthermore, although methadone maintenance therapy is effective at reducing relapse while individuals are currently undergoing treatment (Bell and Strang 2020), relapse to opioids still remains a significant concern after completion of methadone therapy programs (Magura and Rosenblum 2001). Thus, although methadone maintenance therapy is effective in the short-term, it does not adequately address the issue of long-term relapse susceptibility. In this sense, it may allow providers and patients to feel comfortable for as long as the individual is in treatment, but any holistic treatment plan that does not adequately prevent relapse years later ultimately fails to achieve the goal of long-term success. Thus, the challenge for the future is to develop treatment strategies that normalize some of the long-term neural circuit alterations resulting from opioid use and withdrawal.

Future directions

Given the importance of withdrawal in maintaining the cycle of addiction and setting the stage for relapse in abstinent patients, continued research into the neural mechanisms underlying withdrawal-induced negative emotional states is necessary. Contemporary research has begun to assess the influence of additional neurotransmitter systems and brain circuits, and this work has outlined some promising new avenues for OUD treatments. In terms of opioid receptors, most of the studies reviewed above have focused on the role of MORs. A few recent studies suggest that kappa opioid receptor (KOR) stimulation, which is known to inhibit striatal DA release and increase aversion (for review, see Bruijnzeel 2009), may also contribute to aversion during withdrawal. For instance, systemic KOR antagonism reduces morphine withdrawal-induced CPA (Kelsey et al. 2015). This effect may be mediated by KORs in the BLA and CeA, which are thought to be involved in anxiety-like behavior (Knoll et al. 2011). Advanced genetic and neural circuit approaches (e.g., selective gene knockout, optogenetics) will help determine the exact mechanisms by which KORs may be involved in the described circuit. Another popular treatment currently used in OUD patients is α2 adrenergic agonists. As has been discussed, the adrenergic system plays a crucial role in the physical symptoms of withdrawal and α2 agonists are quite effective at reducing these symptoms (Gowing et al. 2016). Furthermore, the use of α2 agonists decreases stress-induced drug seeking in rats as well as craving in humans (Sinha et al. 2011).

Another promising target is the cannabinoid type 1 receptor (CB₁R). This system is increasingly being implicated in anxiety and negative affect (Witkin et al. 2005) and modulation of CB₁R signaling in the BNST, BLA, CeA, and insular cortex all attenuate morphine withdrawal-induced place aversion (Wills et al. 2016, 2017). Other neurotransmitter systems such as adenosine (Jafarova Demirkapu et al. 2020) and orexin (Laorden et al. 2012) have also been implicated in withdrawal and are worthy of further investigation.

In addition to new neurochemical targets, future studies into the role that cortical circuits play in withdrawal are also needed. The insular cortex, known for its role in interoception and self-awareness, has been identified as a potential therapeutic target for addictive disorders (Naqvi et al. 2007; Dinur-Klein et al. 2014; Droutman et al. 2015). This region is highly interconnected with the brain circuits reviewed here, including the BLA, CeA, and BNST, and it receives DAergic inputs from the VTA (Gogolla 2017). Inactivation of the insula in rats has been shown to disrupt the formation of morphine withdrawal-induced CPA (Li et al. 2013) as do manipulations of CB₁R signaling in this area (Wills et al. 2016). The anterior cingulate cortex (ACC) is another cortical region implicated in emotional processes and highly interconnected with subcortical circuits known to participate in withdrawal. Neuroimaging suggests that the ACC is activated during opioid withdrawal in men (Chu et al. 2015) and inhibition of this region has recently been demonstrated to prevent somatic signs of opioid withdrawal in male mice (McDevitt et al. 2021).

Finally, female subjects are currently vastly underrepresented in the literature on the neural mechanisms of opioid withdrawal. The majority of studies reviewed here were conducted exclusively in male subjects. While many areas of addiction neuroscience research have made progress in regard to the study of sex differences (Radke et al. 2021), sex differences in opioid withdrawal remain almost entirely unexplored. Male rodents are known to exhibit greater somatic signs of morphine withdrawal than females (Craft et al. 1999; Kest et al. 2001; Radke et al. 2013; Bobzean et al. 2019), which we speculate may depend on the PAG given demonstrated sex differences in opioid-induced analgesia mediated by this region (Loyd and Murphy 2009). In regard to affective signs of withdrawal, conditioning to morphine withdrawal may be equivalent in male and female mice (García-Carmona et al. 2015b), although at least one report suggests that the acquisition of place aversion is dependent on estradiol in female rats (Martinez-Casiano et al. 2015). Similar increases in the acoustic startle response during withdrawal from acute morphine in male and female rats (Radke et al. 2013, 2015) have also been observed. Many studies additionally find sex differences in self-administration of opioids (Lynch and Carroll 1999; Cicero et al. 2003; Mavrikaki et al. 2017; Fulenwider et al. 2019; Smethells et al. 2020; Monroe and Radke 2021; Radke et al. 2021); however, several studies also demonstrate a lack of sex differences under certain self-administration parameters (Venniro et al. 2019; Venniro et al. 2017; Fredriksson et al. 2020; Bossert et al. 2022). Further research is needed to clarify the influence of biological sex on opioid-seeking behaviors and the extent to which withdrawal contributes to these behaviors in males vs. females.

Summary

In sum, negative affect during opioid withdrawal is a crucial motivational phenomenon in the cycle of addiction. It involves brain systems underlying reward and motivated behavior, learning and memory, and stress. These systems interact with one another under conditions of dysregulated reward systems, hyperactive stress, and enhanced sensitivity to environmental cues, contributing to the complicated pathophysiology underlying opioid addiction. Continued progress in this area over the next several decades has the potential to lead to new and better treatments for OUD and other addictive disorders.