Ethanol Modulation of Cortico-Basolateral Amygdala Circuits: Neurophysiology and Behavior

Abstract

This review highlights literature relating the anatomy, physiology, and behavioral contributions by projections between rodent prefrontal cortical areas and the basolateral amygdala. These projections are robustly modulated by both environmental experience and exposure to drugs of abuse including ethanol. Recent literature relating optogenetic and chemogenetic dissection of these circuits within behavior both compliments and occasionally challenges roles defined by more traditional pharmacological or lesion-based approaches. In particular, cortico-amygdala circuits help control both aversive and reward-seeking. Exposure to pathology-producing environments or abused drugs dysregulates the relative ‘balance’ of these outcomes. Modern circuit-based approaches have also shown that overlapping populations of neurons within a given brain region frequently govern both aversion and reward-seeking. In addition, these circuits often dramatically influence ‘local’ cortical or basolateral amygdala excitatory or inhibitory circuits. Our understanding of these neurobiological processes, particularly in relation to ethanol research, has just begun and represents a significant opportunity.

1. Introduction

This review will focus on behavior contributions of specific circuits within cortical-basolateral amygdala projections. As highlighted below, these projections are largely reciprocal. The resulting cortico-limbic ‘loops’ are well positioned to integrate emotional information within the decision making/learning process. For example, basolateral amygdala inputs within these loops may help assign emotional ‘valence’ (either positive or negative) to learned associations between cues and outcomes and help drive either aversive or seeking-like responses to these cues. Pathology-producing environmental exposures, such as those that produce stress or those that involve abused drugs, can ultimately alter the relative balance within the cortical/limbic ‘arms’ within these loops to help produce behavioral dysregulation like over-valuation of cues or hyper/inappropriate aversive responses. The recent development of experimenter-directed, real-time circuit manipulations has substantially facilitated our understanding of cortico-limbic loops within alcohol or drug abuse. I will primarily focus on recent literature employing heterologous expression of light-gated channels (opsins or optogenetics) or ‘designer receptors’ activated by highly selective, experimentally-designed ligands (for example, DREADD receptors or chemogenetics). Several exceptional reviews highlight the development and deployment of these technologies within neuroscience (Kim et al., 2017; Nectow and Nestler, 2020; Rajasethupathy et al., 2016; Roth, 2016; Saunders et al., 2015). Here I specifically review studies employing optogenetic or chemogenetic approaches designed in a circuit-specific fashion. This is often achieved by selectively activating opsin- or receptor-expressing terminals via localized light delivery through in-dwelling optical fibers or ligand delivery via microinjection, respectively (Fig 1A). Where possible, I describe direct illustrations of the physiology and behavioral contributions related to cortico-limbic circuits within these approaches. On occasion, I infer this information from ‘intersectional’ approaches (Weinholtz and Castle, 2021) using recombination-dependent opsins or DREADD receptors that are selectively expressed within circuits when recombinase is delivered to terminals using retrogradely transported viral constructs. However, the reader should note that such studies may not always consider or control for the potential contributions of collateral projections arising from opsin- or DREADD-expressing neurons. I have attempted to emphasize those studies that do so by virtue of experimental design or approach.

Figure 1. Optogenetics- or chemogenetics-based dissection of cortico-basolateral amygdala circuits.

(A) Representative experimental design for circuit-based analysis of behavior. Microinjection of virally-encoded, light-gated opsins or designer receptors (not shown) into region “A” expresses these proteins within both local principal neurons and their downstream terminal fields (region “B”). Selective delivery of light or ligands into region “B” permits some understanding of the behavioral contributions by this A–B circuit. (B) The rat basolateral amygdala (BLA), including the lateral and basolateral subdivisions, is here represented in a coronal brain slice from the rat (left). The region contains of both glutamatergic principal neurons and local GABAergic interneurons that provide feed-back and feed-forward inhibition. Both glutamatergic and GABAergic BLA neurons receive excitatory input from frontal cortical areas.

All of the discussion revolves around the role of the basolateral amygdala within cortico-limbic ‘loops’. Located within the temporal lobe, the basolateral complex consists of the lateral and basolateral nuclei (Fig 1B). The lateral amygdala is found dorsally and extends along most of the complex’s anterior/posterior axis. The basolateral amygdala is ventral to the lateral amygdala and consists of two subdivisions, the anterior basolateral amygdala that is most prominent along the rostral half of the basolateral nucleus and the posterior subdivision that becomes most prominent along the caudal half of the nucleus. Cellular anatomy (McDonald, 1982) and afferent/efferent connectivity distinguish these basolateral subdivisions from one another. The anterior subdivision receives unique afferents from the prefrontal, auditory, visual, and perirhinal cortices while extensive afferents to the posterior division arise from the hippocampus and entorhinal cortex (McGinnis et al., 2020a; Pitkänen et al., 2000; Sah et al., 2003). Both subdivisions send projections to the nucleus accumbens (Krettek and Price, 1978a; Phillipson and Griffiths, 1985) and the prefrontal cortex (Fig. 2) including the prelimbic, limbic, and anterior cingulate areas (Krettek and Price, 1977; Sripanidkulchai et al., 1984). Within the lateral/basolateral amygdala, approximately eighty percent of the neurons are glutamatergic projection neurons; and the remaining population of feed-forward and feed-back inhibitory GABAergic neurons tightly controls principal neuron activity. Recent work suggests that unique populations of basolateral amygdala neurons project to distinct brain regions (Beyeler et al., 2018, 2016a; Kim et al., 2016; Namburi et al., 2015; Zhang et al., 2020). As discussed extensively below, these populations help provide limbic regulation of both reward-seeking and aversive behaviors.

Figure 2. Summary of regions/projections discussed in the current review and effects of ethanol exposure on these synapses.

Projections are indicated by lines with known connections denoted by the arrowheads. Ethanol facilitation of projections onto basolateral amygdala principal neurons are indicated by green lines while inhibitory effects are noted by red lines. (→) indicates known effects on projections within reciprocal circuits. Black arrows indicate known projections with clearly established effects of ethanol within that particular prefrontal region, but unknown effect on the projection itself. Abbreviations: BLA – basolateral amygdala; PLC – prelimbic prefrontal cortex (section 2.1); ILC – infralimbic prefrontal cortex (section 2.2); rACC and cACC – rostral and caudal anterior cingulate cortex, respectively (section 2.3); lOFC & vOFC – lateral and ventral orbitofrontal cortex, respectively (section 3); aAIC – anterior agranular insular cortex (section 4).

2. Basolateral Amygdala and the Medial Prefrontal Cortex

The rodent medial prefrontal cortex (mPFC) plays a prominent role in a wide range of behaviors including working memory, attention, decision-making, and executive control over aversive and reward-seeking behaviors. But, as outlined below, the mPFC also receives considerable inputs from limbic areas including the basolateral amygdala. This suggests that ‘bottom-up’ emotional systems strongly influence medial prefrontal contributions to behavior. As discussed here, the ‘mPFC’ includes the prelimbic, infralimbic, and anterior cingulate cortex. Readers should note this nomenclature dominates anatomical descriptions in much of the rodent/preclinical literature. However, some publications or atlases may use a numerical designation for these medial prefrontal areas (prelimbic = Area 32, infralimbic = Area 25, rostral anterior cingulate = Area 24a/b, caudal anterior cingulate = Area 24a’/b’). These numerical designations provide some continuity with mid-cingulate cortical areas in the primate literature (Vogt and Paxinos, 2014). These mPFC regions are interconnected and form a largely continuous structure along the dorsal-ventral axis within the rostral midline of the cortex (Jones et al., 2005; Riga et al., 2014). They are distinguished from each other based on functional behavioral contributions, neurochemistry, and anatomy (Heidbreder and Groenewegen, 2003). While we examine these subregions independently, they interact both directly through regional circuits and indirectly through reciprocal circuits with downstream brain regions like the basolateral amygdala. The prelimbic and infralimbic cortices have received considerable attention in the literature due to their roles in regulated conditioned/learned behaviors. There are many excellent reviews covering the details of these contributions (Izquierdo et al., 2016; Marek et al., 2013; Maren et al., 2013). Our focus here is on direct projections between mPFC regions and the basolateral amygdala, the contributions of these projections to behavior, and the impact of ethanol on their physiology and function.

2.1. Basolateral Amygdala-Prelimbic Cortex Circuits

The prelimbic area, located just dorsal to the infralimbic cortex, is extensively involved with the acquisition of conditioned behaviors (Maren et al., 2013). During auditory fear conditioning for example, repeated exposures to a unique tone followed by inescapable foot shock rapidly produces freezing behavior to just the tone alone – a representation of a learned association between a normally innocuous cue and an aversive experience. These associations can persist for many weeks unless subjected to repeated cue presentations without the foot shock (‘extinction’ training). Extinction does not ‘erase’ the conditioned response but instead produces a unique memory (extinction memory) that competes with the originally conditioned associations. Inactivation of the prelimbic cortex with the GABA_A receptor agonist muscimol reduces the expression of conditioned freezing behavior but does not influence the formation of extinction memories (Sierra-Mercado et al., 2011). In addition to aversive conditioned behaviors, the prelimbic cortex also regulates conditioned responses related to drugs of abuse. Microinjection of muscimol/baclofen to inactivate the prelimbic cortex reduces cue-induced reinstatement of heroin responding (seeking behavior) following extinction training (Schmidt et al., 2005). Similarly, high frequency, repeated electrical stimulation of the prelimbic cortex reduces cocaine seeking-like behaviors (Levy et al., 2007). However, neither manipulation reduced seeking-like behaviors directed at the natural reward, sucrose. This suggests that the prelimbic cortex regulates the formation/expression of learned drug-outcome associations established during the acquisition of self-administration. Finally, during prolonged instrumental conditioning for reward, responding shifts from being goal-directed (reward-dependent) early during the training to more habit-like (reward-independent) as animals become more extensively trained. Animals with prelimbic lesions exhibit only habit-like responding regardless of the training time-course (Killcross and Coutureau, 2003). These findings, along with the fear conditioning and drug-related studies, highlight how the prelimbic cortex integrates executive function within a diversity of environmental or interoceptive outcomes to flexibly control behavior.

2.1.1. Prelimbic-to-Basolateral Amygdala Projections

Anatomy & Physiology.

Prelimbic projections to the basolateral amygdala largely target the anterior subdivision (aBL) of the basolateral nucleus (Arruda-Carvalho and Clem, 2014; Hübner et al., 2014; McGarry and Carter, 2017; McGinnis et al., 2020a) and appear to arise from both layer 2/3 and layer 5 principal neurons (Bloodgood et al., 2018; Gabbott et al., 2005; Joffe et al., 2021; Lu et al., 2017). The prelimbic neurons projecting to the aBL are composed of two populations – those that receive reciprocal projections from the basolateral amygdala (discussed below) and those that do not (Liu et al., 2020). These prelimbic-to-aBL neurons are also distinct from other prelimbic neuron populations projecting to various downstream brain regions like the ventral tegmentum and the nucleus accumbens (Murugan et al., 2017). Compared to these other ‘circuits’, prelimbic-aBL neurons are preferentially regulated by a unique population of cortical feed-forward inhibitory neurons, the chandelier cells (Lu et al., 2017). Prelimbic projections within the aBL produce monosynaptic glutamatergic responses onto principal neurons (Hübner et al., 2014; Kiritoshi and Neugebauer, 2018; McGarry and Carter, 2017; McGinnis et al., 2020a) that project to nucleus accumbens, ventral hippocampus, and reciprocally with the prelimbic cortex (McGarry and Carter, 2017). These prelimbic projections also synapse onto local aBL interneurons and thus likely produce both feed-forward inhibition onto aBL principal neurons as well as feed-back inhibitory responses following principal neuron activation (Hübner et al., 2014; Kiritoshi and Neugebauer, 2018; McGinnis et al., 2020a; Tan et al., 2019). Glutamate release from prelimbic inputs onto aBL principal neurons appears to occur with a relatively low release probability compared to synapses arising from other cortical areas (McGinnis et al., 2020a). This suggests that prelimbic inputs may have greater dynamic range and thus greater potential for modulation by either intrinsic signaling pathways or environmental experience. Indeed, fear conditioning shifts the balance between excitation (glutamate) and inhibition (GABA) towards excitation through increased postsynaptic function of these synapses (Arruda-Carvalho and Clem, 2014; Park and Chung, 2020). Conversely, chronic restraint stress shifts this synaptic balance towards excitation by increasing the number of synapses or increasing glutamate release probability (Liu et al., 2020). This synaptic regulation can be normalized by experimentally induced long-term depression at these inputs which also normalizes the behavioral dysregulation produced by restraint stress.

Behavior.

The advent of circuit-specific manipulations within these prelimbic-to-aBL projections has dramatically increased our appreciation of the behavioral role played by these inputs. Optical activation of prelimbic terminals within the aBL enhances a number of innate avoidance-like behaviors including decreased social interaction, reduced time spent in preferred locations (Huang et al., 2020), and decreased open-arm time in the elevated plus-maze (Hübner et al., 2014). It should be noted that optical stimulation of prelimbic-BLA terminals arising from unique populations of cortical neurons that express D1 dopamine receptors surprisingly produces anxiolytic/anti-depressant effects (Hare et al., 2019) and have positive effects on reward-seeking behavior (Land et al., 2014). These data together suggest that prelimbic neurons projecting to the basolateral amygdala may be composed of diverse populations that produce equally diverse control over innate behaviors. Regardless, circuit-specific manipulations of the prelimbic-aBL projections generally suggest that these inputs promote learning related to aversive behaviors. For example, optical activation of prelimbic terminals in the aBL facilitates the expression of conditioned freezing and avoidance (Diehl et al., 2020; Kirry et al., 2020). Unit recordings show that prelimbic neuron activity primarily occurs during cue presentations while BLA neuron activity is more prominent during the expression of conditioned behavior (Karalis et al., 2016). This suggests that prelimbic neurons drive aBL neurons during expression of learned associations. Importantly, prelimbic neurons that project to aBL also facilitate learned associations involving drugs of abuse. Naloxone-precipitated withdrawal in morphine-dependent animals produces conditioned place aversion to the environment where naloxone was administered. This withdrawal increases the expression of c-Fos, a molecular marker for neuronal activity, in prelimbic neurons that project to the aBL; and optical silencing of prelimbic terminals in the aBL can suppress naloxone-precipitated conditioned place aversion (Song et al., 2019). Recent work with circuit-based approaches therefore generally support the hypothesis that manipulations of prelimbic-to-basolateral amygdala projections modulate many of the ‘classic’ behavioral contributions of the prelimbic cortex as defined by pharmacological inactivation or lesion studies.

2.1.2. Basolateral Amygdala-to-Prelimbic Projections

Anatomy & Physiology.

In addition to the anterior basolateral subdivision receiving most of the inputs from the prelimbic cortex, it is also the source of most basolateral amygdala projections back to the prelimbic cortex (Keefer and Petrovich, 2017; Kim et al., 2016). aBL inputs robustly synapse on layer 2/3 pyramidal neurons (Cheriyan et al., 2016; Klavir et al., 2017; Little and Carter, 2012) as well as some principal neuron populations within layer 5 (Cheriyan et al., 2016; Huang et al., 2019). Relative to synaptic inputs from the thalamus or ventral hippocampus, aBL inputs to layer 2/3 are anatomically ‘privileged’ in that they are distributed very close to the neuronal soma (Little and Carter, 2012) suggesting they exert a robust influence over prelimbic neuron activity. High-frequency stimulation of these aBL inputs produces postsynaptic long-term depression (Klavir et al., 2017) that is mimicked by M1 muscarinic receptor modulation of these inputs onto layer 5 prelimbic principal neurons (Maksymetz et al., 2019). aBL terminals also produce excitatory synaptic responses on fast-spiking, parvalbumin-expressing (PV⁺) prelimbic interneurons (Huang et al., 2019; Klavir et al., 2017). For example, chronic pain (nerve injury model) increases excitatory neurotransmission at aBL synapses onto both PV⁺ interneurons and layer 5 prelimbic neurons projecting to the periaqueductal gray (Cheriyan and Sheets, 2020; Huang et al., 2019). These projections also appear sensitive to experience-dependent plasticity in response to chronic stress (Lowery-Gionta et al., 2018). And finally, chronic morphine exposure enhances the frequency of miniature excitatory postsynaptic currents and drives D1 dopamine receptor-dependent excitability of aBL neurons projecting to the prelimbic cortex (Song et al., 2019). Thus, like the prelimbic-to-aBL projections, reciprocal aBL-to-prelimbic synapses appear to be sensitive both to ‘natural’ aversive experiences and to drugs of abuse.

Behavior.

Like the prelimbic-to-aBL projections, aBL inputs into the prelimbic cortex modulate innate and learned behaviors. Optical stimulation of aBL terminals decreases open arm time in the elevated plus maze, decreases center time in the open field, and decreases social interactions while optical inhibition produces the opposite phenotypes (Felix-Ortiz et al., 2016). aBL neurons projecting to the prelimbic also increase c-Fos expression during fear conditioning indicating increased neuronal activation (Senn et al., 2014). Optical inhibition of aBL synapses in the prelimbic reduces prelimbic neuron firing in response to conditioned cues and facilitates the extinction of contextual fear conditioning (Klavir et al., 2017). This extends to conditioned behaviors related to drugs of abuse. For example, conditioned place aversion produced by naloxone-precipitated withdrawal from chronic morphine is blocked by optical silencing of aBL terminals in the prelimbic cortex (Song et al., 2019). Optical inhibition of these inputs also suppresses cue-primed reinstatement behavior for cocaine self-administration (Stefanik and Kalivas, 2013). Like the reciprocal prelimbic-to-aBL inputs, the behavioral dissection of this aBL-to-prelimic circuit has focused on known behavioral contributions by individual brain regions. The somewhat paradoxical findings with aversion/reward seeking and the unique role of D1-expressing prelimbic neurons (Hare et al., 2019; Land et al., 2014) highlight the diversity of behaviors controlled by unique neuronal populations within these circuits.

2.1.3. Ethanol Modulation of Prelimbic-Basolateral Amygdala Circuits

Studies examining ethanol modulation of these important circuits are largely in their infancy and represent significant opportunities to understand the relationships between cortical-basolateral amygdala circuits, ethanol, and the broader literature implicating these circuits in both aversion- and reward-related behaviors. Early studies showed that repeated intraperitoneal ethanol delays fear extinction and enhances fear retrieval while increasing expression of c-fos in both the prelimbic cortex and basolateral amygdala (Quiñones-Laracuente et al., 2015). These observations suggest a central role of prelimbic-aBL circuitry in ethanol-specific modulation of conditioned behaviors. This is further supported by studies showing that chronic ethanol vapor inhalation, which produces robust dependence, dynamically alters prelimbic neuronal anatomy including increased dendritic length, increased dendritic branches, and increased number of dendritic spines including mature, mushroom-shaped spines in both layer 2/3 (Frost et al., 2019; Kim et al., 2015) and layer 5 prelimbic principal neurons (Kroener et al., 2012). However, these effects may be species- or strain-dependent (Chen et al., 2009; Jury et al., 2017).

With respect to synapses, prolonged drinking enhances glutamatergic neurotransmission onto prelimbic neurons that project to the aBL (Joffe et al., 2021). In a recent study, McGinnis and colleagues (McGinnis et al., 2020a) showed that chronic ethanol vapor exposure/withdrawal enhances glutamate neurotransmission from prelimbic synapses onto aBL principal neurons by increasing apparent presynaptic release probability. Activation of exogenously expressed Gi-coupled DREADD receptors within these terminals attenuates this increased glutamate release; and DREADD-mediated inhibition of these terminals can attenuate the heightened anxiety-like behavior that follows chronic ethanol exposure. It is also noteworthy that chronic ethanol appeared to diminish prelimbic input-driven activation of local BLA inhibitory neural circuitry. These findings illustrate that chronic ethanol facilitation of prelimbic-to-aBL inputs can enhance innate aversive behaviors via modulation of both prelimbic inputs and local aBLA circuits. These neurophysiological effects enhance aversive behaviors. In support of this, knockdown of the presynaptic protein synaptotagmin 1 within prelimbic neurons projecting to the aBL enhances self-administration of quinine-adulterated ethanol suggesting that attenuating neurotransmission within the prelimbic-aBL circuit reduces sensitivity to the aversive tastes (Barbier et al., 2021). Although little is known about the reciprocal projections from aBL neurons to prelimbic cortex in the context of ethanol, genetic ablation of basolateral amygdala neurons projecting to the medial prefrontal cortex enhances ethanol-seeking behaviors (Dong et al., 2017) suggesting these aBL-to-prelimbic inputs may help to limit drinking. Together, studies focused on ethanol effects within the basolateral amygdala-prelimbic circuits largely parallel the extensive literature showing these projections modulate aversion-like and perhaps reward-seeking behaviors. Whether these distinct outcomes reflect the divergent actions of independent circuits or distinct neural ensembles within these pathways would be of considerable interest.

2.2. Basolateral Amygdala-Infralimbic Cortex Circuits

Like the prelimbic cortex, the behavioral role of the infralimbic cortex has been best characterized within the context of conditioned cue-outcome learning. Unlike the prelimbic cortex however, the infralimbic cortex appears to control the formation of ‘extinction memories’ formed by repeated exposures to the conditioned cue in the absence of the aversive stimulus used to establish these associations. Extinction training does not ‘erase’ a conditioned association but forms a new memory that competes with the fear memory. Infralimbic neurons fire action potentials during the recall of extinction memories (Milad and Quirk, 2002). Experimental facilitation of infralimbic neuron activity during extinction training using either electrical (Vidal-Gonzalez et al., 2006) or optogenetic stimulation (Do-Monte et al., 2015) facilitates the acquisition and recall of this extinction memory as well. Conversely, pharmacological (Sierra-Mercado et al., 2011) and optical (Do-Monte et al., 2015) inactivation of these neurons blocks both extinction memory acquisition and recall. Thus, within the context of fear conditioning, the infralimbic cortex functionally ‘opposes’ behavioral outcomes facilitated by the prelimbic cortex, presumably due to unique afferents/efferents or local circuits. Infralimbic cortical neurons similalry appear to regulate conditioned behaviors related to drugs of abuse. Repeated presentation of drug-associated cues without available drug extinguishes drug-appropriate responding, equivalent to extinction training with fear conditioning. Following this extinction, exposure to either sensory or interoceptive cues associated with the drug itself can overcome extinction memories and promote drug-appropriate responding even in the continued absence of drug. Notably, pharmacological inactivation of the infralimbic cortex facilitates drug-appropriate responding (‘reinstatement’) even in the absence of cue presentation (Peters et al., 2008). Infralimbic activation, using either pharmacological (Peters et al., 2008) or electrical (Guercio et al., 2020) stimulation, potently suppresses cue- and cocaine-induced reinstatement. These relationships are paralleled by the influence of the infralimbic cortex on drug-place associations as well (Van den Oever et al., 2013). Thus, the infralimbic cortex is intimately involved with the formation of extinction memories in a number of natural and drug-associated contexts.

2.2.1. Infralimbic-to-Basolateral Amygdala Projections

Anatomy & Physiology.

Like the prelimbic cortex, the infralimbic cortex makes prominent projections to the anterior subdivision of the basolateral amygdala (Arruda-Carvalho and Clem, 2014; McGinnis et al., 2020a) and may project to the posterior subdivision as well (Bukalo et al., 2021; Kiritoshi and Neugebauer, 2018; Park and Chung, 2020; Wood et al., 2019). Infralimbic terminals in the basolateral amygdala arise from principal neurons in layers 2/3 and 5 (Bloodgood et al., 2018; Gabbott et al., 2005; McGarry and Carter, 2017; Wood et al., 2019) that are tightly regulated by local parvalbumin- and somatostatin-expressing GABAergic interneurons (McGarry and Carter, 2017). Distinct populations of infralimbic principal neurons project to the basolateral amygdala and the nucleus accumbens shell (Halladay et al., 2020). Their inputs to the basolateral amygdala drive principal neurons to fire action potentials (Strobel et al., 2015) via monosynaptic glutamatergic synapses (Kiritoshi and Neugebauer, 2018; McGinnis et al., 2020a). Infralimbic activation of basolateral principal neurons produces large, feed-back inhibitory synaptic responses (McGinnis et al., 2020a); and these inputs may also activate local feed-forward inhibitory circuitry (Kiritoshi and Neugebauer, 2018). It is unclear whether distinct populations of infralimbic principal neurons project to basolateral amygdala principal neurons and interneurons. Although infralimbic inputs share many characteristics with prelimbic inputs to the basolateral amygdala, they are not functionally altered by the acquisition of conditioned aversive behaviors (Arruda-Carvalho and Clem, 2014). Rather, extinction learning increases the excitability of infralimbic neurons projecting to the basolateral amygdala (Bloodgood et al., 2018) and increases infralimbic-dependent activation of local basolateral amygdala inhibitory circuitry (Park and Chung, 2020). Thus, prelimbic and infralimbic inputs may differentially engage local basolateral amygdala circuitry during the acquisition and extinction of conditioned behaviors, respectively.

Behavior.

Modulation of infralimbic inputs to the basolateral amygdala appears to recapitulate many of the behavioral outcomes associated with manipulating activity within infralimbic cortex itself. For example, chemogenetic inhibition of these inputs decreases affiliative social interactions (Huang et al., 2020). With conditioned responses, optical stimulation of infralimbic terminals in the basolateral amygdala accentuates the formation of extinction memories and decreases retrieval of conditioned fear-related responses (Bukalo et al., 2021). Chemogenetic inhibition of the infralimbic neurons projecting to the basolateral amygdala similarly blocks extinction memory recall (Bloodgood et al., 2018). In an interesting intersection between social behavior and conditioned aversive responses, Dulka and colleagues (Dulka et al., 2020) recently showed that repeated social defeat in Syrian hamsters produces conditioned ‘defeat’ responses that include increased submissive behavior and lack of aggression. In this same study, chemogenetic activation of infralimbic neurons projecting to the basolateral amygdala decreased the expression of conditioned submissive behaviors in subordinate animals, similar to the ‘oppositional’ regulation of classic conditioned fear responses by these projections.

2.2.2. Basolateral-to-Infralimbic Projections

Anatomy, Physiology, & Behavior.

Reciprocal basolateral amygdala projections to the infralimbic cortex may arise more from the posterior subdivision (Kim et al., 2016) and synapse onto infralimbic neurons projecting back to the basolateral amygdala and to the nucleus accumbens (McGarry and Carter, 2017). Basolateral amygdala synapses in the infralimbic cortex may be more abundant on layer 5 principal neurons relative to layers 2/3 (Cheriyan et al., 2016; Kiritoshi and Neugebauer, 2018). The synapses provide direct, monosynaptic glutamatergic inputs to both layer 5 neurons (Cheriyan et al., 2016) as well as to parvalbumin- and somatostatin-expressing interneurons (McGarry and Carter, 2017). This suggests that basolateral amygdala inputs to the infralimbic cortex, like their reciprocal infralimbic-to-aBL projections, may activate both principal neurons and local, feed-forward inhibitory GABAergic circuitry.

Considerably less is known about how environmental experiences might modify these synapses. Extinction training increases c-Fos expression in basolateral amygdala neurons that project to the infralimbic cortex; and tone presentation following extinction training increases the excitability of these neurons (Senn et al., 2014). In addition, an arthritic pain model enhances inhibitory responses recorded from infralimbic principal neurons when basolateral amygdala inputs are optically stimulated (Kiritoshi and Neugebauer, 2018). These findings together suggest that basolateral amygdala-to-infralimbic cortex projections may act as feed-back inhibitory regulators of the neurons projecting back to the basolateral amygdala. While there is currently little evidence in the literature, these projections are well situation to play prominent roles within conditioned behaviors including those associated with drugs of abuse.

2.2.3. Ethanol Modulation of Infralimbic-Basolateral Amygdala Circuits

Most of the literature describing the role of infralimbic-basolateral amygdala circuits has come from models employing chronic ethanol exposure. Chronic ethanol vapor decreases the spine density of infralimbic neurons (Jury et al., 2017). McGinnis and colleagues (McGinnis et al., 2020a) recently showed that a similar exposure decreases glutamate release from infralimbic terminals onto aBL principal neurons while simultaneously enhancing the engagement of local aBL GABAergic circuits by these inputs. Both studies used adolescent animals, a developmental time point during which mPFC projections are continuing to develop. Understanding age-dependent modulation of basal circuit function and the effects of ethanol represents a significant opportunity for future studies. Regardless, these findings together potentially suggest a reduced efficacy of infralimbic/basolateral amygdala circuitry following chronic ethanol. In support of this, chronic ethanol impairs the retrieval of an extinction memory and generalizes conditioned fear responses to a novel tone (Scarlata et al., 2019). This ‘generalization’ is in fact suppressed by chemogenetic activation of infralimbic neurons. Similarly, chronic ethanol increases rat impulsive-like behavior during the five-choice serial reaction time task; and this is associated with a neurochemical shift towards more excitatory amino acids and significantly decreased extracellular concentrations of glycine and serine in the infralimbic cortex (Irimia et al., 2017). Thus, dysregulation of infralimbic/basolateral amygdala circuits by chronic ethanol may impact a substantial number of behaviors relevant to drug and alcohol abuse.

2.3. Basolateral Amygdala-Anterior Cingulate Circuits

Compared to the infralimbic and prelimbic cortex, the impact of ethanol or other drugs of abuse on the anterior cingulate cortex (ACC)-basolateral amygdala circuits is not well understood. This likely reflects the wide variety of models that have been used to characterize behavioral contributions by this brain region to date. These studies suggest that the ACC encodes prediction-error signaling, controls attention, regulates economic evaluation during conflict/reward decision making, and modulates impulsive choice (Bissonette and Roesch, 2016; Floresco and Ghods-Sharifi, 2007; Hart et al., 2017; Hosking et al., 2014; Kim and Lee, 2011). The relatively extensive anatomical distribution of the anterior cingulate cortex along the anterior/posterior axis, with the rostral pole located just dorsal to the prelimbic cortex and posterior portions extending well caudal, further complicates understanding the roles of the ACC or its efferent/afferent connections. For example, the rostral, but not the caudal, ACC controls the acquisition/retention of conditioned memories associated with aversive experiences (Bissière et al., 2008; Johansen et al., 2001; Johansen and Fields, 2004; Malin et al., 2007; Malin and McGaugh, 2006); and, these regions differentially modulate innate aversive behaviors in response to noxious stimuli (Mussio et al., 2020). Literature relating the effects of abused drugs also suggests complex and differential contributions by the rostral and caudal ACC. Like aversive conditioning, conditioned place preference for both cocaine (Neisewander et al., 2000) and morphine (Harris and Aston-Jones, 2003) enhance c-Fos expression in rostral ACC. Pharmacological inactivation of rostral ACC also disrupts reinstatement of cocaine-seeking behavior (McLaughlin and See, 2003). However, cue-induced reinstatement for both cocaine and morphine enhances c-Fos expression in the caudal ACC (Madsen et al., 2012; Thiel et al., 2010). Notably, caudal ACC lesions do not affect the acquisition or maintenance of cocaine self-administration using a fixed ratio schedule but disrupt the attention for/salience of cues on a second-order reinforcement schedule (Weissenborn et al., 1997). These data together suggest that anterior cingulate-basolateral amygdala circuits are likely to make complex behavioral contributions depending upon the specific anatomical sub-region and behavioral model.

Anatomy, Physiology, & Behavior.

Despite these behavioral complexities, both the rostral and caudal ACC send projections to the basolateral amygdala (Buchanan et al., 1994; Cassell and Wright, 1986; Fillinger et al., 2018). Rostral ACC projections arise from layers 2, 3, and 5 principal neurons (Gabbott et al., 2005) while caudal ACC projections may only come from layer 5 neurons (Jhang et al., 2018). These later projections produce monosynaptic glutamatergic responses measured from basolateral amygdala principal neurons and may also activate local feed-forward inhibitory circuits (Allsop et al., 2018; Jhang et al., 2018). There is also some indication that the lateral amygdala also receives inputs from the rostral ACC (Bissière et al., 2008). Anterior basolateral amygdala principal neurons send reciprocal projections back to the rostral ACC (Bissière et al., 2008); but, it is not clear whether lateral or basolateral amygdala neurons send efferents to the caudal ACC.

While nothing appears to be known about the function of basolateral amygdala-to-ACC projections, there is clear indication that the reciprocal projections from caudal ACC can modulate conditioned responses to aversive stimuli. In an “observational fear conditioning” model where a ‘naïve’ animal exhibits conditioned freezing after witnessing pain/noxious stimuli delivered to a cage-mate, optical inhibition of caudal ACC terminals within the basolateral amygdala during conditioning had no effect on fear acquisition but disrupted fear memory recall (Allsop et al., 2018; Smith et al., 2021). Interestingly, photo-inhibition of these terminals did not impair classical auditory fear conditioning (Allsop et al., 2018) suggesting a more specialized role in the transfer of socially relevant cues. Similar to local pharmacological inactivation studies (Cullen et al., 2015), Ortiz and colleagues (Ortiz et al., 2019) also showed that chemogenetic inhibition of caudal ACC terminals in the basolateral amygdala prevents ‘fear generalization’ whereby animals express inappropriate conditioned fear responses to novel environments after prolonged post-training periods.

2.3.1. Ethanol Modulation of ACC-to-Basolateral Amygdala Circuits

Studies relating ethanol modulation of these circuits are in their infancy. Ethanol drinking experience increases the expression of cFos in the ACC (Hamlin et al., 2007; Li et al., 2019) and dynamically regulates the excitability of ACC deep layer (presumably layer 5/6) principal neurons (Cannady et al., 2020). In one of the only circuit-specific ethanol studies in the literature, Sakaguchi and colleagues (Sakaguchi et al., 2018) employed the observational fear conditioning model and showed that non-contingent ethanol given prior to the observation conditioning enhances the learned fear response. This treatment also enhanced the expression of the immediate early gene Arc within caudal ACC neurons that project to the basolateral amygdala. These data suggest, at least indirectly, that these ACC-to-basolateral amygdala projections may influence the expression of socially-mediated conditioned fear responses. Given the paucity of circuit-specific ethanol literature, the range of behavioral models modulated by the ACC, and the unique role of the ACC in socially-relevant learning/memory, studies focused on these projections represent a remarkable opportunity.

3. Basolateral Amygdala and the Orbitofrontal Cortex

There are several exceptional reviews highlighting the anatomy of orbitofrontal cortex (OFC) and its role in various and diverse behaviors (Balleine et al., 2011; Izquierdo, 2017; Moorman, 2018; Rolls, 2004; Schoenbaum and Shaham, 2008; Stalnaker et al., 2015). In general, the OFC receives visual, somatosensory, and gustatory information from cortical processing areas and, through equally diverse projections, regulates several essential executive processes. Within this review, we will focus on the lateral/ventrolateral OFC given their robust reciprocal connections with the basolateral amygdala (Lasseter et al., 2014; Zimmermann et al., 2017) and the robust literature describing the effects of ethanol in this subregion. Pharmacological or DREADD-dependent inactivation of the lateral OFC does not appear to alter drug self-administration but rather disrupts relapse-like responding for both natural rewards (Hernandez et al., 2020) and abused drugs like ethanol (Arinze and Moorman, 2020; Hernandez et al., 2020), cocaine (Lasseter et al., 2014), and opiates (Reiner et al., 2020). Chemogenetic inhibition similarly reduces extinction memory expression following auditory fear conditioning/extinction training (Zimmermann et al., 2017). Oddly, pharmacological inactivation of the lateral OFC increases the expression of innate anxiety-like behavior (Kuniishi et al., 2017). The behavioral regulation by the lateral OFC thus appears quite complex. Yet, across the frontal cortical regions covered within this review, the orbitofrontal cortex is arguably the most represented within the ethanol preclinical literature. There is also an extensive body of work highlighting the role of the OFC in human alcoholics. However, it is probably the least understood prefrontal area with respect to OFC-basolateral amygdala circuitry and the impact of ethanol on these projections.

3.1. Lateral OFC-Basolateral Amygdala Circuits

Lateral and ventrolateral OFC neurons provide strong, monosynaptic glutamatergic projections to the basolateral amygdala (Kuniishi et al., 2017; Lasseter et al., 2014; Zimmermann et al., 2017). These projections appear to drive basolateral neuronal activity during cue-dependent learning (Saddoris et al., 2005). Repeated exposures to stress increases both the AMPA/NMDA ratio and synaptic contributions by inwardly rectifying AMPA receptors produced by optical stimulation of lateral OFC synapses in the basolateral amygdala (Kuniishi et al., 2017). These findings suggest that OFC projections to the basolateral amygdala are critically important for evaluative learning and are dynamically regulated by environmental exposures. Indeed, optogenetic and chemogenetic manipulation of OFC terminals in the basolateral amygdala modulate the expression of stress-related depressive-like behaviors (Kuniishi et al., 2017) and disrupt the encoding of reward value in food-deprived animals during operant procedures (Malvaez et al., 2019). Finally, using the ‘intersectional’ approach to specifically express inhibitory opsins in OFC neurons projecting to the basolateral amygdala, Arguelo and colleagues (Arguello et al., 2017) showed that optical inhibition of these neurons disrupts relapse-like responding for cocaine following extinction training. Similar inhibition of basolateral amygdala neurons projecting to the OFC was notably without effect, suggesting these reciprocal projections may have unique but undefined effects on behavior. The reciprocal basolateral amygdala projections to the lateral OFC appear to arise primarily from the anterior subdivision and target both the posterior aspect of the lateral OFC and the ventrolateral OFC (Barreiros et al., 2021; Lasseter et al., 2014; Murphy and Deutch, 2018). These projections are likely glutamatergic (Lichtenberg et al., 2017), although this has not been explicitly demonstrated using receptor antagonists. And, like their reciprocal connections, basolateral amygdala inputs drive OFC neuron action potentials during the learning of cue/outcome associations (Schoenbaum et al., 2003). Chemogenetic inhibition of basolateral amygdala terminals in the OFC reduces cue-specific responding for food rewards during the expression of Pavlovian instrumental transfer (Lichtenberg et al., 2017); but there is no effect of OFC terminal inhibition in the basolateral amygdala. Thus, both the Lichtenberg et al. and the Arguello et al. studies indicate that, while OFC and basolateral amygdala have strong reciprocal connections, these projections make unique behavioral contributions.

3.2. Ethanol Modulation of the OFC

There are no studies in the literature focused on ethanol modulation of OFC-basolateral amygdala circuitry. Despite this, ethanol has well-documented effects within the OFC itself. For example, acute ethanol inhibits lateral OFC neuron excitability (Badanich et al., 2013; Nimitvilai et al., 2020) through local activation of inhibitory strychnine-sensitive glycine receptors (Badanich et al., 2013). Both non-contingent chronic ethanol exposure (Nimitvilai et al., 2020, 2018, 2017, 2016) and long-term ethanol drinking (Cannady et al., 2020) produce adaptive increases in OFC neuron excitability by down-regulating the contributions of hyperpolarizing SK-type potassium channels (Nimitvilai et al., 2020, 2016) and potentially by enhancing postsynaptic glutamatergic function (Nimitvilai et al., 2016). Chemogenetic inhibition of lateral OFC neurons and OFC lesions enhance ethanol drinking, but only after chronic ethanol exposure (Den Hartog et al., 2016). It is notable too that OFC neurons exhibit robust, biphasic activity during ethanol self-administration, with action potentials and intracellular calcium increasing during operant responding/approach behavior and then decreasing during consumption (Cazares et al., 2021; Gioia and Woodward, 2021; Hernandez and Moorman, 2020). Chronic ethanol enhances the dynamic range of this processing. These data together suggest that lateral OFC neurons and their outputs make increasingly important contributions to the regulation of ethanol drinking during a dependence like exposure. Given the OFC’s role in relapse-like behavior, conditioned responding, goal-directed and habitual behavior, and decision making, studies specifically focused on OFC-basolateral amygdala circuity represent a significant opportunity to understand the loss of control over drinking following chronic ethanol exposure.

4. Basolateral Amygdala and the Insula Cortex

In the rat, the insular cortex receives multi-modal sensory input from olfactory, gustatory, visceral, and nociceptive from thalamic and cortical areas. It stretches from the dorsolateral orbitotfrontal cortex rostrally and extends caudally to the piriform along the dorsal rhinal fissure. The insula’s three subregions are named based on the organization of principal neuron ‘layers’ used to characterize most cortical regions. The granular insular cortex is located along the dorsal boundary and has six well-defined neuronal layers (1–6). In contrast the agranular insular cortex (AIC) lacks a well-defined layer 6 and is located along the ventral border. Finally, the dysgranular insular cortex is located between the agranular and granular insula along the dorsal/ventral axis and has intermediate layer 6 anatomy (Sewards and Sewards, 2001). These regions are all highly interconnected with one another and with limbic structures (Kobayashi, 2011). The granular and dysgranular insula play important roles in gustatory and nociceptive processing, respectively. The agranular insula, in particular the anterior agranular insula (aAIC), has strong reciprocal connections with the basolateral amygdala (Allen et al., 1991; Mátyás et al., 2014; McDonald and Mascagni, 1996). Blockade of glutamatergic neurotransmission in the aAIC reduces the expression of innate anxiety-like behaviors (Méndez-Ruette et al., 2019). Similarly, optical activation of local aAIC GABAergic interneurons, which inhibits aAIC principal neurons, both reduces innate anxiety-like behaviors and blocks the expression of contextual conditioned fear memories (Shi et al., 2020). These findings clearly support the role of aAIC in aversive behaviors. In contrast, optical silencing of aAIC also reduces affiliative social behavior (Rogers-Carter et al., 2018). This suggests that the aAIC has roles in both aversive and reward-like behaviors. Literature relating aAIC to drugs of abuse supports this suggestion (Naqvi et al., 2014) although the aAIC role is complex and involves evaluations when there is a conflict between drug seek/taking and competing goals or behaviors (Pelloux et al., 2018). This is potentially related to studies showing that stroke-damage to the insula disrupts cigarette smoking in nicotine addicts (Naqvi et al., 2014). However, the extent to which anterior insula-basolateral amygdala circuits contribute to these behaviors is only beginning to be understood.

4.1. Basolateral Amygdala-Insula Cortex Circuits: Anatomy, Physiology, & Behavior

The anatomy and physiology of reciprocal anterior insula-basolateral amygdala projections is less characterized than their behavioral contributions. For example, layer 5 AIC neurons project to the basolateral amygdala (Gabbott et al., 2005). These projections appear to target much of the basolateral nucleus and are both monosynaptic and glutamatergic (McGinnis et al., 2020b). Conversely, basolateral amygdala axons within the AIC target both the deep and superficial layers (Kayyal et al., 2019; Livneh et al., 2017). Importantly, these axons represent collaterals (Livneh et al., 2017) from basolateral amygdala principal neurons that project to the nucleus accumbens, mPFC, and central amygdala. Although it has not been explicitly examined, these basolateral amygdala inputs are likely glutamatergic.

Much of the literature describing the behavioral contributions of insula-basolateral amygdala circuitry has focused on well-established roles for these regions, typically avoidance behaviors related to taste. During conditioned taste aversion for example, animals are given access to a naturally rewarding tastant (e.g. saccharine, the conditioned stimulus) that is subsequently paired with an intraperitoneal injection of the malaise-inducing salt lithium chloride (unconditioned stimulus). The result is an aversive memory characterized by ‘avoidance’ behavior directed at the normally rewarding sweet solution. The memory is long-lasting and relies upon both the basolateral amygdala and insular cortex (Molero-Chamizo and Rivera-Urbina, 2020). This conditioning increases c-Fos expression in (Abe et al., 2020) and firing of (Lavi et al., 2018) AIC neurons projecting to the basolateral amygdala; and, chemogenetic manipulation of these neurons controls the acquisition and expression of conditioned taste responses (Kayyal et al., 2019). Notably, these same chemogenetic manipulations have no impact on conditioned freezing behaviors following auditory fear conditioning. The reciprocal projections from the basolateral amygdala to the anterior insula appear to play related roles (Abe et al., 2020). These anterior insula-basolateral amygdala circuits also appear to regulate a diversity of additional behaviors. For example, functional imaging in patients with post-traumatic stress finds increased connectivity between basolateral amygdala and insula along with positive correlations between symptom severity and these connections (Nicholson et al., 2016). In rodents, these circuits also appear to maintain the relationships between cues and reward during conditioned approach behavior (Nasser et al., 2018), accentuate the extinction during real-time conditioned place preference (Gil-Lievana et al., 2020), and modulate responses to conditioned food cues in hungry animals (Livneh et al., 2017). While there has been little work relating these anterior insula-basolateral amygdala circuits to drugs of abuse, it is clear from this diversity of behavioral outcomes that these projections are poised to make substantial contributions there as well.

4.2. Ethanol Modulation of Basolateral Amygdala-Insula Cortex Circuits

Our current understanding related to ethanol modulation of these circuits represents a potentially significant opportunity for future study. For example, acute ethanol can inhibit NMDA receptor-mediated synaptic responses recorded from layer 2/3 anterior agranular insula neurons and can block NMDA receptor-dependent long-term depression at these synapses (Shillinglaw et al., 2018). Notably, acute ethanol had little impact on GABA-mediated spontaneous transmission in these same neurons. Prolonged abstinence from ethanol drinking by inbred ethanol-preferring P rats increases c-Fos expression in the anterior insula; and this is blocked by local pharmacological inactivation (Campbell et al., 2019). This suggests some involvement of the anterior insula in withdrawal-related processes despite minimal impact of chronic ethanol exposure on insula principal neuron morphology (Frost et al., 2019). Regardless, long-term ethanol self-administration by non-human primates significantly depolarizes the resting membrane potential of deep layer agranular insular neurons and significantly increases glutamatergic neurotransmission onto these neurons (Alexander et al., 2012). This highlights the potential for chronic ethanol to influence activity in brain regions down-stream from the insula, including the basolateral amygdala. Indeed, recent work from McGinnis and colleagues (McGinnis et al., 2020b) used optical stimulation of anterior insula-to-basolateral amygdala projections and showed that chronic ethanol exposure enhances postsynaptic function at these synapses. This facilitation was blocked by in vivo ketamine administration, suggesting a role for NMDA receptor-dependent plasticity. Finally, chemogenetic inhibition of insula terminals in the basolateral amygdala attenuated the enhanced expression of anxiety-like behavior during withdrawal. Although the impact of these insula-basolateral amygdala circuits on ethanol drinking behaviors has not yet been examined, anterior insula projections to the nucleus accumbens appear to positively regulate ‘basal’ ethanol drinking (Jaramillo et al., 2018) as well as ‘compulsive-like’, aversion-resistant ethanol consumption (Seif et al., 2013). Given that the basolateral amygdala also projects to the nucleus accumbens and these inputs can modulate reward-seeking (Millan et al., 2017; Namburi et al., 2015; Stefanik and Kalivas, 2013; Stuber et al., 2011), chronic ethanol facilitation of both anterior insula neuron excitability and their projections to the basolateral amygdala may make significant contributions to withdrawal-dependent facilitation of ethanol drinking.

5. Conclusions

Among the most interesting observations that can be derived from the literature within this review is that many cortico-basolateral amygdala circuits appear to regulate both avoidance (‘negative’) and reward-seeking (‘positive’) behaviors. These discordant outcomes are specifically represented within both prelimbic cortex- and orbitofrontal cortex-basolateral amygdala circuits and are suggested for anterior cingulate- and insula-circuits as well. Within the prelimbic circuits, prelimbic neurons projecting to the anterior basolateral amygdala are composed of distinct populations of neurons that can have opposing influences over innate behaviors like anxiety (Huang et al., 2020; Hübner et al., 2014) and reward-seeking (Hare et al., 2019; Land et al., 2014). The reciprocal aBL-to-prelimbic projections likewise facilitate these same innate and learned behaviors. This suggests a positive, feed-forward/back loop between prelimbic cortex and the anterior basolateral amygdala related to both avoidance and seeking-like behavior. There is also strong evidence for unique populations of ‘valence-specific’ neurons in the basolateral amygdala that project to nucleus accumbens and extended amygdala (Beyeler et al., 2016a, 2016b; Lee et al., 2017; Namburi et al., 2015). Target-specific populations in the basolateral amygdala may also be complimented by neuronal populations across diverse projection sites. For example, basolateral projections to the anterior insula represent collaterals from neurons that also project to the nucleus accumbens which may modulate reward-seeking behavior (Millan et al., 2017; Namburi et al., 2015; Stefanik and Kalivas, 2013; Stuber et al., 2011), to the mPFC which regulates both avoidance and reward seeking (see above), and to central amygdala. Given the anatomical and developmental parallels between the basolateral amygdala and cortical regions, understanding the interactions between, and ethanol modulation of, target-specific and target-diverse projection neurons within the cortex represent impactful future studies.

Finally, the regulation of avoidance and reward behaviors by cortico-amygdala circuits may be directly related to a broader contribution by interactions between intra-region components within these circuits. For example, the anterior and posterior basolateral amygdala, which predominantly send projections to prelimbic and infralimbic cortex respectively, have reciprocal connections with one another (Krettek and Price, 1978b; McDonald and Culberson, 1986); and ethanol modulation of synaptic interactions between anterior and posterior basolateral amygdala might ultimately influence the relative activity of their downstream targets. Basolateral amygdala projections, reciprocal projections from cortex, and even local intra-nuclear projections (Samson and Paré, 2006; Smith et al., 2000) synapse onto both principal neurons as well as local inhibitory GABAergic neurons. Likewise, a recent study by Hagihara et al. (Hagihara et al., 2021) showed that feed-forward GABAergic intercalated cell clusters surrounding the lateral and basolateral nucleus inhibit each other via cluster-to-cluster projections that differentially regulate the acquisition and extinction of fear memories. Chronic ethanol exposure (Diaz et al., 2011) greatly diminishes GABA release onto BLA principal neurons from one of these clusters (“LPC” neurons, Fig. 1B). These data suggest that complex interactions between GABAergic circuits are as important to understand as intra-nuclear glutamatergic projections. These data importantly suggest that ethanol shifts the relative contributions of these local excitatory and inhibitory circuits; and the direction of that shift is circuit-dependent. For example, chronic ethanol shifts the relative influence of prelimbic- and infralimbic- basolateral amygdala circuits to favor the prelimbic projections. Notably, these effects together may reflect the synaptic correlate of an animal’s reduced ability to form ‘extinction memories’ following chronic ethanol (Scarlata et al., 2019) and might ultimately help undermine abstinence. Circuit-based investigations of anterior cingulate- and orbitofrontal cortical-basolateral amygdala projections also hold great promise for understanding both the complex roles these circuits play in ‘normal’ behaviors as well as those associated with alcohol abuse and alcoholism.

Highlights.

• Cortical-basolateral amygdala circuits control a variety of innate and learned behaviors

• Optogenetic and chemogenetic approaches have allowed detailed understanding of these circuits

• Cortical-basolateral amygdala circuits modulate behaviors associated with drugs of abuse and are modified by exposure

• This review relates ethanol modulation of these circuits