Reciprocal midbrain-extended amygdala circuit activity in preclinical models of alcohol use and misuse

Abstract

Alcohol dependence is characterized by a shift in motivation to consume alcohol from positive reinforcement (i.e., increased likelihood of future alcohol drinking based on its rewarding effects) to negative reinforcement (i.e., increased likelihood of future alcohol drinking based on alcohol-induced reductions in negative affective symptoms, including but not limited to those experienced during alcohol withdrawal). The neural adaptations that occur during this transition are not entirely understood. Mesolimbic reinforcement circuitry (i.e., ventral tegmental area [VTA] neurons) is activated during early stages of alcohol use, and may be involved in the recruitment of brain stress circuitry (i.e., extended amygdala) during the transition to alcohol dependence, after chronic periods of high-dose alcohol exposure. Here, we review the literature regarding the role of canonical brain reinforcement (VTA) and brain stress (extended amygdala) systems, and the connections between them, in acute, sub-chronic, and chronic alcohol response. Particular emphasis is placed on preclinical models of alcohol use.

Introduction

Alcohol use disorder (AUD) affects over 14 million Americans (SAMHSA, 2018), costing the US ~$249 billion annually (Sacks et al., 2015). The transition to AUD is characterized by cycles of alcohol intoxication, withdrawal, and craving. Each stage of this cycle is typically associated with distinct neural correlates, with mesolimbic circuitry heavily implicated in alcohol intoxication; extended amygdala circuitry involved in withdrawal and negative affect; and cortical structures contributing to preoccupation/anticipation of alcohol use (Koob and Volkow, 2010). A thorough understanding of the neurobiological underpinnings of the acute actions of alcohol, as well as the neural alterations following chronic alcohol use and subsequent alcohol response, is necessary for developing targeted therapeutics aimed at reducing excessive drinking and other symptoms in individuals with AUD.

Positive and negative reinforcement in AUD

Alcohol use disorder is clinically defined by the presence of diagnostic criteria in several categories, including impaired control over consumption (e.g., drinking more alcohol than intended, or facing an inability to reduce drinking), impact on social structures (e.g., drinking despite negative consequences on interpersonal or work relationships), potential for harm (e.g., engaging in risky behavior as a consequence of drinking), and physiological adaptations to alcohol use (e.g., presence of tolerance or withdrawal symptoms; American Psychiatric Association, 2013). The neurobiological alterations associated with this altered behavior state have been described using an allostasis model, wherein the body adapts to sustained change in environmental factors to achieve a new stability in internal parameters (Koob and Le Moal, 1997; Sterling and Eyer, 1988). Associated with this allostatic model of chronic alcohol use is a shift in the motivational forces that drive alcohol consumption, more specifically a transition from positive reinforcement (i.e., increased likelihood of future alcohol drinking based on its rewarding effects) to that of negative reinforcement (i.e., increased likelihood of future alcohol drinking based on alcohol-induced reductions in negative affective symptoms, including but not limited to those experienced during alcohol withdrawal). It has also been argued that the development of AUD is associated with a transition from impulsive (initiating an act in order to obtain a sense of pleasure or gratification) to compulsive (initiating an act in order to obtain a sense of relief of anxiety or stress) behaviors (Koob et al., 2004). The transition to development of AUD is not uniform among all individuals and may be influenced by co-occurring disorders (e.g., PTSD or chronic pain); however, compulsive alcohol intake is a shared characteristic of AUD diagnoses in humans (for review, see Egli et al., 2012; Gilpin and Weiner, 2017; Maleki et al., 2019).

Specific neuronal and circuit changes due to repeated alcohol use, as well as neural regulators of alcohol dependence-associated behaviors, have been the subject of research for decades. Simplistically, early in alcohol use, alcohol is thought to engage brain reinforcement systems (canonically, these are described as projections from the ventral tegmental area [VTA] to the nucleus accumbens [NAc]; Di Chiara, 1997). As an individual transitions to more compulsive use, negative reinforcement mechanisms are thought to become an important contributor to excessive alcohol drinking, and this is thought to occur via gradual recruitment of brain stress circuits, including but not limited to the central amygdala (CeA) and bed nucleus of the stria terminalis (BNST; collectively referred to as the extended amygdala, along with the NAc shell; Koob and Volkow, 2010). Early stages of alcohol use are associated with increases in dopaminergic (DAergic) transmission from VTA cells that are associated with subjective perceptions of reward and updates in reward prediction error encoding (Berridge, 2007; Schultz, 1998). Consumption of high quantities of alcohol over an extended period of time is associated with deficits in DA signaling, along with parallel increases in extended amygdala activity, increases in pro-stress neuropeptide content and signaling, and decreases in anti-stress neuropeptide content and signaling in those regions (Koob and Le Moal, 2008). Generally speaking, canonical brain reinforcement circuitry becomes hypofunctional after chronic high-dose alcohol and during the transition to AUD, whereas brain stress signaling systems become more engaged on a similar timeline. Below, we discuss evidence for connections between brain reinforcement and stress systems, effects of chronic alcohol on circuit connections between brain reinforcement and stress systems, and the potential role of brain reinforcement-stress system circuits in mediating aspects of alcohol use disorder, as modeled in preclinical systems.

Ventral Tegmental Area

Structure and Function

The VTA is canonically thought to mediate the acute positive reinforcing effects of alcohol (Morikawa and Morrisett, 2010). The VTA is a source of DA neurons that send projections to brain regions involved in regulating reinforcement and cognition, such as the NAc and medial prefrontal cortex (mPFC; Beier et al., 2015). While much focus on VTA response to reinforcing stimuli has centered around DA neurons, this represents only 55–65% of the neuronal population, with the remainder of VTA neurons being GABAergic (30–35%) and glutamatergic (2–5%; Breton et al., 2019; Margolis et al., 2006; Nair-Roberts et al., 2008; Yamaguchi et al., 2015). VTA neurons regulate myriad behaviors via projections to the mPFC (e.g., perseverative behaviors; Kabanova et al., 2015), lateral habenula (e.g., reinforcement-related behaviors; Stamatakis et al., 2013), and NAc (e.g., associative learning, reward prediction, and aversion; Brown et al., 2012; Day et al., 2007; Qi et al., 2016).

Alcohol Response

The VTA responds to acute alcohol in the alcohol-naïve organism by displaying an increase in firing activity (Brodie et al., 1990; Gessa et al., 1985; Mrejeru et al., 2015). Preclinical studies have demonstrated that silencing VTA DA cell bodies in alcohol-preferring P and Wistar rats (Gatto et al., 1994; Rodd et al., 2004), as well as blocking DA receptors in male Long Evans and female high-alcohol-drinking rats (Dyr et al., 1993; Hodge et al., 1997) reduces voluntary alcohol self-administration, suggesting that alcohol-induced increased VTA DA activity is associated with alcohol reward.

Sub-chronic alcohol exposure (i.e., alcohol experience associated with reward, but not to levels of dependence) produces plasticity in VTA DA neurons and a greater response to alcohol application. For example, repeated administration of alcohol results in increased expression of GluR1 and NR1, subunits of AMPA and NMDA receptors, respectively, and TH in VTA DA neurons (Ortiz et al., 1995). Alcohol experience also produces corresponding increases in excitatory drive onto VTA DA neurons and expression of long-term potentiation following voluntary alcohol consumption in Wistar rats and systemic administration of alcohol in DBA mice (Stuber et al., 2008; Wanat et al., 2009). This plasticity is thought to facilitate alcohol’s excitatory effects on this region; indeed, following voluntary alcohol intake in TH-GFP adolescent mice, VTA DA neurons demonstrate a sensitized excitatory response to low concentrations of alcohol in vitro (Avegno et al., 2016). Collectively, these data support a role for VTA DA neurons in mediating the rewarding effects of acute and sub-chronic alcohol.

As a consequence of chronic alcohol exposure (i.e., levels sufficient to induce alcohol dependence), VTA DA neurons are often considered to have a deficit in signaling and to exhibit a blunted response to alcohol. Human imaging studies indicate that there is lower DA transmission in the brains of individuals with AUD compared to healthy controls (e.g., Volkow et al., 2013), suggesting a decreased activity of VTA DA neurons. A reduction in firing rate of VTA DA neurons has been reported during withdrawal after repeated alcohol gavage in Sprague-Dawley rats (Diana et al., 1992); similarly, the proportion of spontaneously active VTA DA neurons is reduced during withdrawal from chronic alcohol diet in TO mice (Bailey et al., 1998), vapor exposure in Long Evans rats (Avegno et al., 2020), and intragastric administration in Sprague-Dawley rats (Shen and Chiodo, 1993). These data support the hypothesis that depressed dopaminergic tone during alcohol dependence is a consequence of reduced cellular activity at the level of the VTA.

Beyond Canonical Reinforcement

The information outlined above presents the accepted framework of the VTA as a center mitigating response to reinforcing stimuli, including alcohol; wherein activation of VTA DA neurons is associated with subjective feelings of reward and influences motivation to consume alcohol or to engage in behavior that results in receiving alcohol. As alcohol use transitions from impulsive to compulsive, the relative influence of the VTA wanes, and a desire to reduce the deficit in dopaminergic tone is considered to be one of the driving factors in one’s motivation to continue consuming alcohol. While this conceptual framework is both compelling and supported by a wealth of published observations, it is important to note that this may not capture the full extent of the VTA’s involvement in alcohol response.

At the cellular level, VTA neurons exhibit considerable heterogeneity that has recently become a subject of more detailed investigation (Lammel et al., 2008; Margolis et al., 2010; Margolis et al., 2008). While there is little debate that VTA DA neurons are implicated in the reinforcing effects of alcohol (Morikawa and Morrisett, 2010), evidence also suggests that stress and addiction (i.e., cocaine) produce neuroadaptations in the VTA (Lammel et al., 2014). In response to alcohol specifically, VTA DA neurons demonstrate a heterogeneous response to acute (Mrejeru et al., 2015) and sub-chronic voluntary (Avegno et al., 2016) alcohol exposure in TH-GFP mice, with a subset of medially located VTA DA neurons demonstrating greater excitatory responses to alcohol relative to lateral VTA DA or non-DA neurons. VTA neurons can be subdivided based on projection target and molecular signature (e.g., those that release DA, glutamate, or GABA solely, or co-release more than one neurotransmitter), and these subpopulations exhibit different responses to rewarding and aversive stimuli (Lammel et al., 2011; Morales and Margolis, 2017).

At the systems level, the hypodopaminergic model of alcohol dependence is not fully supported by preclinical and clinical literature. Postmortem studies on the brains of individuals with alcohol use disorder indicate a downregulation of dopamine 1 receptor (D1R) and dopamine transporter (DAT) binding sites in the ventral striatum and nucleus caudatus, with no change in dopamine 2 receptor (D2R) binding sites, indicative of a hyperdopaminergic state during acute or protracted withdrawal from chronic alcohol (Hirth et al., 2016). A follow-up meta-analysis of studies performed in rats indicates that reduced accumbal DA levels observed in the rat occur transiently during days 1–3 of alcohol cessation, with an increase in dopaminergic tone reported at later timepoints (7 and 21 days post alcohol) (Hirth et al., 2016). Additionally, acute alcohol challenge (2 g/kg, i.p.) results in blunted increases in extracellular accumbal dopamine in CIE-exposed male Wistar rats tested during protracted withdrawal (Hirth et al., 2016), and actually results in decreased extracellular dopamine levels in C57BL/6J mice tested during acute withdrawal (Karkhanis et al., 2016). In the context of the hypodopaminergic hypothesis of alcohol dependence, these findings collectively all point towards a dysregulation of the DA system in alcohol dependence; however, the nature and duration of this dysregulation is not consistently reported in the clinical and preclinical literature. Additionally, just as the impact of chronic alcohol on DA levels is not uniformly reported in the literature, the motivation to consume alcohol in heavy drinkers or individuals with AUD is not uniform either. Studies in non-treatment-seeking heavy drinking individuals report that a majority of self-report as “reward” drinkers, who cite a motivation to drink alcohol due to its reinforcing or pleasurable effects; as opposed to “relief” or “habit” drinkers (Burnette et al., 2021; Grodin et al., 2019).

Collectively, these variable results emphasize the complexity of alcohol-mediated neural and behavioral dysregulation and underscore the need for individualized treatment for this disorder. Therefore, the impact of alcohol dependence on VTA neuron activity, as well as the influence of this region on alcohol dependence-associated behaviors, may be more complex than its canonical role.

Extended Amygdala

Structure and Function

The extended amygdala refers to a collection of forebrain structures which share similar morphological and immunocytochemical characteristics, have similar connectivity with other brain regions, and produce similar physiological and behavioral functions. These structures are components of a continuum linking the basal forebrain to the lateral hypothalamus and include the bed nucleus of stria terminalis, sublenticular substantia innominata, centromedial amygdala, and portions of the nucleus accumbens shell (Alheid and Heimer, 1988; Zahm, 1998). Collectively, the extended amygdala receives input from limbic and olfactory cortices and sends efferents to the hypothalamus and midbrain. Neurons within these areas are primarily GABAergic (Duvarci and Pare, 2015; Sun and Cassell, 1993), but there is a large degree of heterogeneity within these neurons, based on factors including connectivity and neuropeptide co-expression (Batten et al., 2002; Jüngling et al., 2015; Kozicz et al., 1997; Pomrenze et al., 2015; Reyes et al., 2008; Walter et al., 1991). For example, in humans, the BNST and CeA both share functional connectivity with regions including the mPFC, hippocampus, thalamus, and periaqueductal gray, while the CeA has stronger connectivity with regions involved in sensory and attention processing, compared to a greater connection between the BNST and regions involved in motivational processing (Gorka et al., 2018). The NAc shares similar connectivity to the CeA and BNST and mediates appetitive and defensive motivation behaviors (Reynolds and Berridge, 2008).

Alcohol Response

Extended amygdala brain regions play a role in voluntary alcohol consumption in rodent models; for example, site-specific administration of a D1 receptor antagonist into the BNST (Eiler et al., 2003); an opioid receptor antagonist into the CeA (Foster et al., 2004); or a GABA_A receptor antagonist into the CeA, BNST, or NAc (Hyytiä and Koob, 1995), all reduce operant self-administration of alcohol in alcohol-preferring and/or Wistar rats. Acute alcohol actions on the CeA are posited to mediate some of the anxiolytic effects of the drug (Koob, 2004). Bath application of alcohol in vitro increases GABAergic transmission in the CeA, and this effect is attenuated by bath application of a CRFR1 antagonist (Nie et al., 2004). Systemic administration of alcohol leads to increases in extracellular CRF concentrations in the rat CeA (Lam and Gianoulakis, 2011), and CeA CRF neurons increase firing activity during binge-like alcohol drinking sessions in Crh^tm1(cre)Zjh mice (Aroni et al., 2021). Pharmacological (Hyytiä and Koob, 1995) or siRNA-mediated (Liu et al., 2011b) reductions of CeA GABA_A receptor activity in Wistar and/or P rats, as well as CeA CRF1R antagonism in C57BL/6J mice (Lowery-Gionta et al., 2012), reduce voluntary consumption of alcohol in rodents that are not dependent on alcohol.

As an individual (animal or human) transitions from impulsive to compulsive alcohol use, the CeA and BNST mediate additional alcohol-related behaviors. Imaging studies in humans with AUD indicate greater connectivity between the amygdala and BNST, as measured by functional magnetic resonance imaging, that is associated with loss of control over alcohol intake (O’Daly et al., 2012). Alcohol withdrawal recruits an ensemble of CeA neurons; inactivation of this ensemble in alcohol vapor-exposed Fos-LacZ rats reduces somatic withdrawal signs and escalated alcohol consumption (de Guglielmo et al., 2016). Furthermore, optogenetic inhibition of CeA CRF-positive neurons in alcohol vapor-exposed Crh-Cre rats prevents recruitment of the neuronal ensemble, decreases the escalation of alcohol drinking, and decreases the intensity of somatic signs of withdrawal (de Guglielmo et al., 2019). Extracellular CRF concentrations are increased in the BNST of Long Evans rats during acute withdrawal from chronic alcohol diet (Olive et al., 2002), and administration of CRF into the BNST leads to sensitization of alcohol withdrawal-induced anxiety-like behavior in alcohol diet-fed Sprague-Dawley rats (Huang et al., 2010), highlighting a potential role for this region in negative affective symptoms during alcohol withdrawal. The CeA and BNST have been implicated in myriad alcohol withdrawal-associated behaviors, including hyperalgesia in vapor-exposed Wistar rats (Avegno et al., 2018), escalated alcohol intake in vapor-exposed Wistar rats and C57BL/6 mice (Ferragud et al., 2021; Finn et al., 2007; Gilpin et al., 2008), escalated binge-like alcohol drinking in C57BL/6J mice (Campbell et al., 2019), cue-induced craving in humans with AUD (Schneider et al., 2001), and stress-induced reinstatement of alcohol seeking in vapor-exposed Long Evans rats (Funk et al., 2019).

Below, we discuss the current understanding of reciprocal connections between the VTA and extended amygdala regions (NAc_shell, BNST, and CeA; Figure 1), as well as what is known about the role of those circuits in mediating alcohol responses and alcohol withdrawal-associated behaviors.

Figure 1:

Summary of VTA-extended amygdala connectivity. Top panel summarizes spatial location of reciprocal projections between the VTA (grey) and NAc (yellow; left), BNST (green; middle), and CeA (blue; right). Bottom panel summarizes the cell types within each projection. Illustration generated by modifying images purchased in the Illustration Toolkit-Neuroscience from Motifolio, Inc.

VTA↔NAc shell

Connectivity and Function

D1R-expressing NAc medium spiny neurons (Bocklisch et al., 2013) send GABAergic projections to the VTA (Kalivas et al., 1993) and synapse onto both GABA_AR-containing GABA neurons and GABA_BR-containing DA neurons (Edwards et al., 2017). Bi-directional connections between the VTA and the lateral and medial subdivisions of the NAc_shell are largely non-overlapping. Projections from the lateral and medial NAc_shell target the lateral and medial VTA, respectively (Yang et al., 2018). Likewise, medial and lateral NAc_shell-projecting VTA neurons are typically located in the medial and lateral VTA, respectively (Lammel et al., 2008), suggesting the presence of parallel reciprocal circuits. Optogenetic and electrophysiological experiments in DAT^IRESCre mice have identified a population of DA-glutamate co-releasing VTA neurons that project to the NAc_shell, but this study did not distinguish between medial and lateral VTA sub-regions (Mingote et al., 2015). The majority of medial NAc_shell-projecting VTA DA neurons are likely capable of glutamate co-release (Mingote et al., 2019), and these neurons appear to preferentially innervate cholinergic interneurons in the NAc. Optogenetic stimulation of VTA DA terminals in the medial NAc_shell of DAT^IRESCre mice elicits a stronger glutamatergic postsynaptic response in cholinergic interneurons than in medium spiny neurons or fast spiking interneurons (Chuhma et al., 2014). Conversely, lateral NAc_shell-projecting VTA neurons are a mix of DA-glutamate co-releasing and DA-only neurons (Mingote et al., 2019). Optogenetic stimulation of lateral NAc_shell terminals in the VTA results in disinhibition of VTA DA projections to lateral NAc_shell, whereas stimulation of medial NAc_shell terminals in the VTA results in inhibition of VTA DA projections to medial NAc_shell neurons, highlighting the complexity of this system and providing a mechanism by which these closed loop systems can function as positive or negative feedback loops as well (Yang et al., 2018).

Alcohol Response

While VTA-NAc_shell circuitry has been studied extensively in the response to alcohol, data on reciprocal (NAc_shell-VTA) circuitry is lacking. Acute alcohol activates VTA-NAc circuitry, although not all studies of this phenomenon have distinguished between the medial and lateral NAc shell and the NAc core. Acute alcohol increases the firing of VTA DA neurons in alcohol-naïve mice and rats (Brodie et al., 1990; Gessa et al., 1985; Mrejeru et al., 2015) and increases local DA concentration in the NAc of Sprague Dawley, Long Evans, and Wistar rats (Di Chiara and Imperato, 1985; Samson et al., 1997; Yim and Gonzales, 2000). The extent of alcohol-induced increased NAc DA concentration is correlated with preference for alcohol over water in inbred rat strains (Katner and Weiss, 2001), and human imaging studies indicate a significant correlation between alcohol induced increases in extracellular striatal DA concentrations and subjective reports of activation (e.g., elation, excitement; Urban et al., 2010; Yoder et al., 2007). These associations indicate that activation of the VTA-NAc DA pathway plays a role in alcohol’s reinforcing effects. Indeed, preclinical studies have demonstrated that silencing VTA DA cell bodies in alcohol-preferring P and Wistar rats (Gatto et al., 1994; Rodd et al., 2004), as well as blocking DA receptors in male Long Evans and female high-alcohol-drinking rats (Dyr et al., 1993; Hodge et al., 1997) or lesioning DA terminals in the NAc of female alcohol-preferring rats (Ikemoto et al., 1997), reduces voluntary alcohol self-administration, providing further support for this.

Following repeated sub-chronic exposure to alcohol, P rats exhibit potentiation of alcohol-induced DA release in the NAc in vivo and increases in voluntary self-administration of alcohol into the VTA (Toalston et al., 2014). In vitro experiments demonstrate a sensitized excitatory response of medial NAc_shell-projecting VTA DA neurons to alcohol following binge-like alcohol intake in TH-GFP mice (Avegno et al., 2016). Site-specific inhibition of D1Rs in the NAc_shell reduces context-induced reinstatement of operant responding for alcohol-associated cues in Long Evans rats, suggesting that this circuitry is involved in strengthening contextual associations related to alcohol use (Chaudhri et al., 2009). Collectively, these data support a role for VTA-NAc_shell DA neurons in mediating the rewarding effects of alcohol, and a strengthening of VTA-NAc_shell innervation following sub-chronic alcohol exposure.

Chronic alcohol results in reduced DA signaling into the NAc_shell. Human imaging studies indicate that there is lower striatal DA transmission in the brains of individuals with AUD compared to healthy controls (e.g., Volkow et al., 2013), suggesting disrupted signaling from VTA DA neurons (e.g., decreased innervation, alterations in dopamine receptor or transporter levels; see also (Hirth et al., 2016). Similar results have been demonstrated using in vivo microdialysis in the rat following repeated systemic alcohol administration (Rossetti et al., 1992). Chronic alcohol reduces NAc DA release and increases uptake rates following vapor exposure in C57BL/6J mice (Karkhanis et al., 2015; Rose et al., 2016). Limited studies have explicitly focused on NAc_shell-projecting VTA neuron activity (or reciprocal circuitry) in alcohol dependent animals; however, the existing literature demonstrating alcohol dependence-associated reduced VTA DA neuron activity and reduced NAc DA levels collectively points to a deficit in VTA-NAc signaling in the context of alcohol dependence.

The existing literature points to a complex reciprocal connectivity between the VTA and NAc_shell. Interestingly, despite the canonical role of the VTA-NAc circuit in mediating reinforcement, relatively few studies have directly assessed this connection in response to acute, sub-chronic, or chronic alcohol, highlighting the need for circuit-specific research. Further, substantial gaps in the literature on the reciprocal NAc-VTA circuit warrant additional research.

VTA↔BNST

Connectivity and Function

Reciprocal connections between the VTA and BNST are relatively well characterized. BNST-projecting neurons are distributed throughout the VTA in the rat, and approximately 25% of these cells are DAergic (Hasue and Shammah-Lagnado, 2002). TH-expressing terminals are most densely expressed in the dorsolateral (including the oval and juxtacapsular nuclei) BNST (Phelix et al., 1992) and synapse onto CRF-positive neurons (Meloni et al., 2006), but the BNST also receives DAergic innervation from the substantia nigra, periaqueductal grey, and dorsal raphe (Hasue and Shammah-Lagnado, 2002). Electrophysiological studies in C57BL/6J and CRF-tomato mice demonstrate that bath-applied DA depolarizes BNST CRF-positive neurons (Silberman et al., 2013) and increases excitatory transmission in BNST via a CRF1R-dependent mechanism (Kash et al., 2008). In the BNST of Long Evans and Sprague-Dawley rats, DA decreases inhibitory synaptic transmission (Krawczyk et al., 2011); these converging excitatory effects highlight a potential feedback mechanism by which VTA DA neurons can regulate BNST efferents.

A heterogeneous population of BNST neurons project to the VTA (Silberman et al., 2013). The VTA receives GABAergic and glutamatergic innervation from the BNST (Dong and Swanson, 2006). In vivo recordings in Sprague-Dawley rats demonstrate an excitatory functional connection between ventral BNST neurons and putative VTA DA neurons, with no significant influence of BNST projections on the activity of putative VTA GABA neurons (Georges and Aston-Jones, 2001, 2002). In over half of neuronal recordings, a long latency was observed between vBNST stimulation and VTA DA neuron activation, indicating a potential indirect mechanism of action in some cells (either via a relayed excitatory signal through an intermediate brain region, or possibly via modulation of VTA neurons that influence VTA DA neuron activity; Georges and Aston-Jones, 2002). Approximately 30% of VTA-projecting BNST neurons in the rat, and approximately 80% of VTA-projecting BNST neurons in the non-human primate, are CRF-positive (Fudge et al., 2017; Rodaros et al., 2007). Anatomical studies in C57BL/6J and GAD67-GFP mice indicate that the majority (>90%) of VTA-projecting BNST neurons are GABAergic and preferentially (80–90%) synapse onto VTA GABAergic neurons (Kudo et al., 2012), implicating a disinhibitory mechanism by which BNST projection neurons produce an excitatory effect on VTA DA neurons. This seemingly conflicting data from mouse (Kudo et al., 2012) and rat (Georges and Aston-Jones, 2001; Georges and Aston-Jones, 2002) studies could reflect species-specific differences in the BNST-VTA circuit or could be reconciled by experimental differences (e.g., Kudo et al (Kudo et al., 2012) combined data from dorsal and ventral BNST subdivisions, whereas Georges and Aston-Jones (Georges and Aston-Jones, 2001; Georges and Aston-Jones, 2002) focused on ventral BNST exclusively). Activation of BNST-VTA GABA neurons produces real-time place preference in Vgat-ires-cre mice, while activation of BNST-VTA glutamate neurons produces real-time place aversion in Vglut2-ires-cre mice (Jennings et al., 2013), indicating that distinct subpopulations of VTA-projecting BNST neurons may mediate different aspects of reward processing and aversion. Further characterization in transgenic mice detailed an anxiolytic BNST-VTA circuit under the control of a local population of BNST CRF neurons, which in turn are modulated by serotonergic projection neurons from the dorsal raphe nucleus. Serotonin release into the BNST activates local BNST CRF neurons via actions on 5HT_2C receptors, resulting in downstream inhibition of VTA-projecting BNST neurons, and ultimately eliciting anxiety-like behavior and enhanced fear learning (Marcinkiewcz et al., 2016).

Alcohol Response

BNST-VTA circuitry is implicated in voluntary alcohol intake in non-dependent individuals. Acute alcohol increases DA concentration in the BNST of Sprague-Dawley rats (Carboni et al., 2000), and local inhibition of D1 receptors in BNST reduces voluntary alcohol consumption in alcohol-preferring rats (Eiler et al., 2003). These observations raise the possibility that increased DAergic signaling from VTA into BNST mediates alcohol reinforcement, although pharmacology alone is insufficient to isolate VTA-BNST inputs specifically (Hasue and Shammah-Lagnado, 2002). Chemogenetic inhibition of VTA-projecting BNST neurons prevents alcohol conditioned place preference expression in DBA/2J mice (Pina and Cunningham, 2017), and selective chemogenetic inhibition of VTA-projecting dlBNST GABA neurons in Vgat-ires-cre mice (Companion and Thiele, 2018) or VTA-projecting BNST CRF neurons in CRF-Cre mice (Rinker et al., 2017) decreases voluntary binge-like alcohol consumption. Therefore, there is behavioral evidence for a role of BNST-VTA projections, and potentially VTA-BNST projections, in mediating response to acute and sub-chronic alcohol (e.g., voluntary alcohol consumption).

In a series of electrophysiological studies in alcohol-dependent, withdrawn C57BL/6J and CRF-Tomato mice, Silberman and colleagues demonstrated that (1) bath application of DA depolarizes BNST CRF neurons, which could facilitate increased local CRF transmission; (2) bath application of CRF increases excitatory transmission onto VTA-projecting BNST neurons; and (3) alcohol dependence increases glutamatergic drive onto VTA-projecting BNST neurons and occludes excitatory CRF effects during acute withdrawal (4 hr post alcohol cessation), indicating a mechanism by which BNST-VTA circuitry is disrupted during dependence (Silberman et al., 2013). Following prolonged withdrawal (72 hr post alcohol cessation) in alcohol dependent CRF-L10A-GFP mice, local BNST CRF neurons display increased excitability; VTA-projecting non-CRF BNST neurons display a corresponding increase in spontaneous inhibitory transmission and decrease in spontaneous excitatory transmission (Pati et al., 2020). Given that inhibition of VTA-projecting BNST neurons can elicit anxiety-like behavior in alcohol-naïve mice (Marcinkiewcz et al., 2016), this net inhibitory influence on BNST-VTA neuron activity could correspond with increased anxiety-like behavior during withdrawal from chronic alcohol. Collectively, these studies indicate engagement of reciprocal circuitry between the BNST and VTA during repeated alcohol intake and raise the possibility that this system may be further disrupted in alcohol dependence. While in vitro and in vivo studies have begun to characterize BNST-VTA circuitry in response to acute, sub-chronic, or chronic alcohol, gaps remain in research on the reciprocal VTA-BNST circuit.

VTA↔CeA

Connectivity and Function

Early insight into the functional relevance of VTA input to the CeA came from studies modulating local DA receptor activity in the CeA, with the hypothesis being that the source of DA input into the CeA was the VTA, and therefore manipulating CeA DA activity would be targeting the VTA-CeA circuit specifically. Using this approach, a role for this circuit in withdrawal and anxiety has been proposed. Pharmacological experiments and CeA neuron recordings suggest that DA reduces evoked inhibitory synaptic transmission in CeA via D1 receptors in Sprague-Dawley rats (Naylor et al., 2010). Transgenic mice with hyperactive DA neurons (as a consequence of selective leptin receptor deletion in this population) exhibit high anxiety-like behavior that is reversed by D1R antagonist infusion into the CeA (Liu et al., 2011a), implicating this circuit in anxiety-like behavior. Additional studies have utilized retrograde or anterograde tracers to evaluate this circuit directly (Avegno et al., 2020; Leshan et al., 2010; Mingote et al., 2015; Zhou et al., 2019), and these have raised two important points that should be considered when interpreting results from DA modulation in the CeA. First, non-DA projection neurons from the VTA also project to the CeA (Avegno et al., 2020; Zhou et al., 2019), indicating that focusing solely on DA modulation of the CeA can miss potentially important VTA subpopulations. Approximately 30% of CeA-projecting VTA neurons in Long Evans rats express Slc17a6 (which transcribes vesicular glutamate transporter 2 [vGluT2], a marker for glutamatergic neurons), and an additional ~30% of neurons express Slc32a1 (which transcribes vesicular inhibitory amino acid transporter [VIAAT], a marker for GABAergic neurons; Avegno et al., 2020), and CeA-projecting VTA GABA neurons have been implicated in defensive response to a looming threat in GAD2::Cre mice (Zhou et al., 2019). Second, the CeA receives DAergic input from the substantia nigra and periaqueductal gray in addition to the VTA (Avegno et al., 2020; Hasue and Shammah-Lagnado, 2002; Leshan et al., 2010), indicating that local CeA DA modulation alone is insufficient to implicate VTA inputs. Optical simulation of VTA terminals in CeA can elicit inhibitory postsynaptic responses in GAD2::Cre mice (Zhou et al., 2019) or excitatory postsynaptic responses in DAT^IRESCre and C57BL/6J mice (Avegno et al., 2020; Mingote et al., 2015). These optically-evoked events are blocked in the presence of GABAergic or glutamatergic antagonists, respectively, indicating functional connectivity between the two regions. Additionally, some CeA-projecting VTA neurons are capable of DA and glutamate co-release in DAT^IRESCre mice (Mingote et al., 2015), with approximately 50% of Th-expressing VTA-CeA neurons co-expressing Slc17a6 (Avegno et al., 2020). Work from the Myers lab has demonstrated a population of leptin receptor (LepRb)-containing VTA neurons in LepRb-EGFPf mice, approximately 75% of which co-express TH and synapse onto cocaine and amphetamine regulated transcript (CART)-expressing CeA neurons (Leshan et al., 2010). Conditional knockout of LepRb in midbrain DA neurons produces an anxiogenic phenotype in mice, which is reversed by D1R antagonism in CeA (Liu et al., 2011a), raising the possibility that this population specifically regulates anxiety-like behavior.

Just as early studies inferred functional importance of VTA input to CeA by modulating local DA, a potential role for CeA input into the VTA has been investigated via modulation of CRF signaling within the VTA (similarly, a limitation of this pharmacological approach is an inability to conclusively isolate CeA-VTA circuitry, as VTA CRF input is not limited to the CeA; Fudge et al., 2017; Rodaros et al., 2007). CeA innervation of dopaminergic midbrain neurons has been characterized in rats (Gonzales and Chesselet, 1990; Wallace et al., 1992; Zahm et al., 2011) and non-human primates (Fudge and Haber, 2000; Price and Amaral, 1981). Tracing studies in non-human primates indicate that CeA neurons project primarily to the retrorubal field (RRF) and parabrachial pigmented nucleus (PBP) subregions within the VTA, and that approximately 80% of these neurons are CRF-positive (Fudge et al., 2017). Approximately 30% of VTA-projecting CeA neurons are estimated to be CRF-positive in the rat (Rodaros et al., 2007). CRF-positive neurons in the CeA of Wistar rats receive input from DA terminals (albeit from an unknown origin; Eliava et al., 2003), raising the possibility of a feedback circuit between these two regions.

Alcohol Response

Direct, circuit-specific effects of acute, sub-chronic, and chronic alcohol on reciprocal VTA-CeA circuitry is not well characterized. Binge-like alcohol consumption increases CRF–mediated potentiation of NMDA receptor currents in putative DA neurons in the VTA of C57BL/6J mice, and intra-VTA administration of a CRF binding protein antagonist (Albrechet-Souza et al., 2015) or CRF1 receptor antagonist reduces voluntary alcohol consumption (under a two-bottle choice or drinking in the dark paradigm) in C57BL/6J mice and Long Evans rats (Hwa et al., 2013; Sparta et al., 2013), indicating a role for VTA CRF1R-mediated signaling in regulating binge-like alcohol intake. The firing of CeA CRF neurons is increased in Crh^tm1(cre)Zjh mice under a similar drinking paradigm (modified drinking in the dark; Aroni et al., 2021), raising the possibility that CeA CRF innervation of the VTA is implicated in alcohol drinking. However, the VTA also receives CRF input from the BNST and paraventricular nucleus of the hypothalamus (Fudge et al., 2017; Rodaros et al., 2007); therefore, the extent to which CeA CRF inputs to VTA specifically regulate alcohol intake remains to be determined.

Characterization of alcohol effects on VTA projections to CeA is extremely limited. Recent work demonstrated activation of CeA-projecting VTA neurons (approximately 30% of which are TH-expressing) during withdrawal from chronic alcohol vapor exposure in Long Evans rats, as well as increased evoked excitatory postsynaptic responses in CeA neurons following stimulation of VTA terminals in alcohol-dependent C57BL/6J mice (Avegno et al., 2020), indicating activation of this circuit in alcohol-dependent animals during withdrawal, and raising the possibility that this circuit plays a role in alcohol dependence-associated behaviors. Therefore, based on the existing literature, we speculate that CeA projections to VTA may have a role in mediating binge-like alcohol intake, whereas VTA projections to CeA may have a role in mediating alcohol dependence- and withdrawal-related behaviors (e.g., increased anxiety-like behavior during withdrawal). Additional research directly assessing acute, sub-chronic, and chronic alcohol effects on each circuit, as well as circuit contribution to alcohol-associated behaviors, is needed.

Conclusions

Above, we reviewed the current literature on reciprocal connections between the VTA and extended amygdala subregions, the effects of alcohol on these individual circuits, and the potential role of these circuits in mediating behavioral responses to acute alcohol, binge-like alcohol exposure, and alcohol dependence and withdrawal. A simplified framework could describe a shift from predominant activation of VTA-NAc_shell neurons in response to acute or sub-chronic alcohol exposure, to a gradual recruitment of VTA-BNST neurons in response to binge-like alcohol intake, then a predominant activation of VTA-CeA neurons as an individual shifts to an alcohol dependent state (Figure 2). However, this does not fully reflect the complexity of VTA-extended amygdala interactions in the context of alcohol response, nor does it account for reciprocal connections between extended amygdala subregions and the VTA. Additionally, the extent to which there is overlap between VTA neurons with shared or distinct afferent/efferent connections is not well characterized. VTA-NAc_shell neurons appear to have minimal shared arborization to BNST or to CeA (Beier et al., 2015), although collateralization has not been extensively studied among all cell groups considered here. The extent to which activation of one circuit within the VTA-extended amygdala system influences activity of a separate circuit is an additional area open for research (e.g., whether and how early engagement of the VTA produces plasticity in extended amygdala projection targets; whether stress or alcohol engagement of extended amygdala subregions produce plasticity in the VTA; or whether engagement of extended amygdala subregions influence each other).

Figure 2:

Summary of VTA-extended amygdala connectivity across stages of alcohol use. Reciprocal connections between VTA (grey), NAc_shell (yellow), BNST (green), and CeA (blue) indicated. Thicker arrows indicate increased strength of connection or demonstrated role of a circuit in alcohol-related behaviors; thinner arrow indicates a deficit in signaling between regions. Dashed arrows are used when additional experiments are needed to confirm circuit specificity. No arrows are used where existing circuit data in a given alcohol context is lacking. BNST-VTA arrow in right panel reflects stronger connectivity during acute (4 hr post alcohol) withdrawal, and decreased connectivity during protracted (72 hr post alcohol) withdrawal. Illustration generated by modifying images purchased in the Illustration Toolkit-Neuroscience from Motifolio, Inc.

Behaviorally, we can speculate that early engagement of VTA-NAc and BNST-VTA circuitry contributes to alcohol reinforcement and escalated consumption of alcohol. In response to sub-chronic alcohol use, continued activation of VTA-NAc and BNST-VTA circuits, as well as recruitment of CeA-VTA inputs, may mediate voluntary alcohol consumption and escalated intake of alcohol. As a result of chronic alcohol use, we speculate that alcohol withdrawal-associated behaviors (e.g., increased anxiety) manifest as a result of both engagement of VTA-CeA circuitry during acute withdrawal and reduced activity of VTA-projecting BNST neurons during protracted withdrawal. The contributions of each circuit to alcohol-associated behaviors outlined above are supported or suggested by the existing literature, although we note that additional research is needed to definitively confirm the contributions of VTA-extended amygdala connections to behaviors present across varying stages of alcohol use.

While much is known about VTA connections between extended amygdala subregions, several gaps in the literature remain. Future research is needed to thoroughly characterize the midbrain-extended amygdala connectivity, as well as alterations in microcircuitry consequent to increasing levels of alcohol use. Increasing emphasis should be placed on understanding brain stress-reward interactions in alcohol use and misuse, as research into the midbrain and extended amygdala regions, and crosstalk between the two, is fundamental to understanding neural response to rewarding levels of alcohol and dysregulation following chronic alcohol intake and may pave the way to improving therapeutics for individuals with AUD.

• We review the role of canonical midbrain reinforcement (VTA) and brain stress (extended amygdala) systems in commonly measured outcomes in preclinical alcohol use disorder (AUD) models

• We summarize what is and is not known about reciprocal circuits between midbrain and extended amygdala brain areas, broken down by cell type and sub-region

• We summarize what is and is not known about the effects of acute, sub-chronic, and chronic alcohol exposure on VTA-extended amygdala circuits and the role of those circuits in alcohol-related behaviors