The amygdala noradrenergic system is compromised with alcohol use disorder

Abstract

Background:

Alcohol use disorder (AUD) is a leading preventable cause of death. The central amygdala (CeA) is a hub for stress and AUD, while dysfunction of the noradrenaline (NA) stress system is implicated in AUD relapse.

Methods:

Here we investigated whether alcohol (ethanol) dependence and protracted withdrawal alter noradrenergic regulation of the amygdala in rodents and humans. Male adult rats were housed under control conditions, subjected to chronic intermittent ethanol vapor exposure (CIE) to induce dependence, or withdrawn from CIE for two weeks, and ex vivo electrophysiology, biochemistry (catecholamine quantification by HPLC), in situ hybridization and behavioral brain-site specific pharmacology studies were performed. We also used qRT-PCR to assess gene expression of the α1_B, β1, and β2 adrenergic receptor in human post-mortem brain tissue from men diagnosed with AUD and matched controls.

Results:

We found that α1 receptors potentiate CeA GABAergic transmission and drive moderate alcohol intake in control rats. In dependent rats, β receptors disinhibit a subpopulation of CeA neurons, contributing to their excessive drinking. Withdrawal produces CeA functional recovery with no change in local noradrenaline tissue concentrations, though there are some long-lasting differences in the cellular patterns of adrenergic receptor mRNA expression. Additionally, post-mortem brain analyses reveal increased α1B receptor mRNA in the amygdala of humans with AUD.

Conclusions:

Thus, CeA adrenergic receptors are key neural substrates of AUD. Identification of these novel mechanisms that drive alcohol drinking, particularly during the alcohol-dependent state, support ongoing new medication development for AUD.

Introduction

Stress is a major contributor to alcohol use disorder (AUD), promoting escalation of drinking in dependence and triggering craving and relapse during abstinence (1). Individuals with AUD display heightened stress reactivity (1–3), and exhibit a high rate of comorbidity with stress-related psychiatric diseases (4). However, there are no FDA-approved medications targeting brain stress systems to control excessive drinking.

One promising target is the noradrenaline (or norepinephrine; NA) system, which regulates the brain’s response to stress and alcohol, and is implicated in craving and relapse in abstinent AUD patients (2,5). Alcohol intoxication increases central NA in humans, and NA dysregulation persists in abstinence (6). Preclinically, alcohol activates the locus coeruleus (LC) and nucleus tractus solitarius (NTS), two major noradrenergic brain nuclei (7). Additionally, α1 (prazosin) or β1/2 (propranolol) adrenergic receptor inverse agonists decrease dependence and relapse-related drinking (8,9). Both drugs are FDA-approved for cardiovascular diseases, but not indicated for AUD. These studies (10–13) have renewed translational interest, though more information is needed about the AUD symptoms and subtypes best suited for adrenergic receptor treatment (5,14). Thus, there is an urgent need to understand how dysregulation of brain region-specific adrenergic mechanisms contribute to escalated drinking in AUD.

The central amygdala (CeA) is crucial for alcohol’s reinforcing actions, and its overactivation marks the transition to dependence (15,16). CeA GABAergic neurons are interconnected and send downstream projections including to the LC and NTS (17–20). The LC and NTS project back to the CeA, and activation of these feed-forward stress circuits produces persistent brain-wide NA release (20). Despite the role of these circuits in negative emotional states, little is known about how NA regulates CeA activity. Given current efforts to develop adrenergic receptor-based therapeutics for AUD, we used a systems biology translational approach to investigate mechanistically how alcohol-induced CeA noradrenergic dysfunction promotes the escalation of drinking in dependence in rodents and is associated with alcohol drinking-related behaviors in humans with AUD.

Materials and Methods

Experimental design

Rodent procedures were approved by TSRI Institutional Animal Care and Use Committee, consistent with the NIH Guide for the Care and Use of Laboratory Animals. Adult male rats (Charles River Laboratories, Raleigh, NC) were housed 2–3 per cage under a reverse 12h/12h light/dark cycle with ad libitum food and water. All preclinical studies were performed in control, ethanol dependent and withdrawn rats (Fig. S1–2).

The human post-mortem brain project was approved by the NIAAA Scientific Advisory Board and the NIH Office of Human Subjects Research Protections and was exempt from review by the NIH Institutional Review Board (21). Brain tissue (N=27) from men diagnosed with severe AUD and controls was obtained from the New South Wales Tissue Resource Centre (NSWBTRC) at the University of Sydney, Australia (22).

Chronic intermittent ethanol (CIE) exposure to induce ethanol dependence

Sixty-nine adult, male Sprague Dawley rats (418.5±13.5 g terminal weight) received 5–7 weeks of chronic intermittent ethanol vapor (CIE; 14 h vapor/10 h air daily) to produce ethanol dependence (23–26) (Fig. S1). The mean blood alcohol level (BAL) was 181±5 mg/dL. Forty-one CIE rats were sacrificed 15–30 min prior to the end of daily vapor exposure (alcohol-dependent group), while the remaining CIE rats underwent 2 weeks of withdrawal. Naive rats (N=46) received continuous air.

Electrophysiology

Rats were anesthetized and brain slices prepared (25,27). Neuronal intrinsic membrane properties and excitability, spontaneous/miniature inhibitory postsynaptic currents (s/mIPSCs) and spontaneous cell firing were recorded in 236 cells in the medial central amygdala (CeA) (25–28). Each experiment includes data from a minimum of 4 rats, with 1–2 cells per animal. Recordings were analyzed with pClamp (Molecular Devices, Sunnyvale, CA), MiniAnalysis (Synaptosoft Inc., Fort Lee, NJ) or NeuroExpress software by Dr. A. Szucs (26,29). Cells were classified as increased, decreased, or no change based on ±20% change in the maximum drug effect from baseline.

DL-AP5, CGP 55845A, DNQX, noradrenaline bitartrate, prazosin hydrochloride and propranolol hydrochloride (Tocris, Bristol, UK); bicuculline and tetrodotoxin (TTX; Sigma. St. Louis, MO); and alcohol (Remet, La Mirada, CA) were dissolved in ACSF.

Catecholamine quantification

Rats (5–7 per group) were anesthetized, and the CeA dissected. Briefly, the sample was homogenized in 200 μL of 100 mM perchloric acid solution (30). After centrifugation and extraction, 20 μL of sample was combined 1:1 with 100 mM potassium hydroxide solution. The resulting sample was analyzed for NA and dopamine (DA) content with a reversed-phase high performance liquid chromatography (HPLC) system coupled to an electrochemical detector. Final values were adjusted for wet tissue weight.

In situ hybridization and confocal microscopy

Rats (4–6 per group) were anesthetized and perfused transcardially. Brains were fixed and flash-frozen. ISH was performed using a RNAscope kit (Biotechne ACD 320850; manuals: #320535 and #320293) and negative control (320751), Adra1a (592781-C2), Adrb1 (468121-C1), Adrb2 (468131-C3) probes (31).

Medial CeA images were obtained using a Zeiss LSM 780 laser scanning confocal microscope (40X oil immersion, 1024×1024 pixel, 5-μm z-stacks), and quantified with CellProfiler (31). For each image, the percent of nuclei expressing each gene transcript was determined. Brightness/contrast/pixel dilation are the same for all representative images.

Alcohol self-administration and CIE

Thirty-two male Wistar rats received alcohol (10% w/v) self-administration training. Half were made alcohol-dependent using a modified CIE procedure, while the remaining non-dependent rats received air (32). During CIE week 4–5, all rats underwent alcohol self-administration during acute withdrawal (8 h post-vapor), three times per week (Fig. S2).

At the beginning of CIE week 2, rats were implanted with a double guide cannula (Plastics One, Roanoke, VA) aimed at the CeA (anterior/posterior, −2.6 mm; medial/lateral, ±4.2 mm; dorsal/ventral, −6.1 mm from bregma (33)). Two non-dependent (cannula failure) and two dependent rats (health complications and cannula failure) were excluded (Fig. 7E).

Figure 7.

CeA α1 and β receptor signaling drive alcohol intake in non-dependent and dependent rats. A: Intra-CeA administration of prazosin decreased alcohol intake only in the non-dependent rats. B: Intra-CeA propranolol decreased alcohol intake only in dependent rats. with no effect on the inactive lever across both groups. C and D: Responses at the inactive lever were unaffected. E: Histological reconstruction showing CeA microinfusion sites (drawing from the atlas of Paxinos and Watson). All data are presented as mean±SEM. **p<.01 vs. basal intake by Sidak post hoc test; ^p<.05, ^^^p<.001 vs. vehicle by Sidak post hoc test; ^#p<.05, ^##p<.01 vs. dependent rats by Sidak post hoc test. N=14 rats/group. Bsl, baseline during training before CIE vapor exposure; Veh, vehicle.

Starting at CIE week 6, rats were tested for alcohol self-administration after bilateral CeA microinfusion of prazosin or propranolol (0, 6 and 60 nmol) in random order using a Latin-square design (Fig. S2). After microinfusion (0.5 μl/site; 0.5 μl/min over 1 min), rats were returned to their home cages for 2 min and then tested for self-administration.

Human α1_B, β1, and β2 adrenergic receptor mRNA expression

Fold change in mRNA levels of the α1_B, β1, and β2 adrenergic receptor was measured in the amygdala, hippocampus, PFC, ventral tegmental area (VTA) and nucleus accumbens (NAc) using real-time quantitative polymerase chain reaction (qRT-PCR).

Statistical analyses

Statistical analyses were performed using Prism (v.9, GraphPad, San Diego, CA), R software or Statistical Package for the Social Sciences (SPSSv.26) (IBM, Armonk, NY) with differences significant at p<.05. Data are represented as mean±SEM.

For the electrophysiology experiments, drug effects were normalized to their own neuron’s baseline prior to group analyses. If the data had equal variance and normal distributions according to the Kolmogorov-Smirnov test, final values were analyzed using one-sample or paired t-tests or one-way ANOVAs. Otherwise, non-parametric measures were used. For the catecholamine concentration study, we conducted one-way ANOVAs. For the in situ hybridization study, each probe’s data was normalized to the control group. Final values were analyzed using one-way ANOVA and Tukey’s post hoc test. For the behavioral study, two-way ANOVA for repeated measures (RM) and the Sidak post hoc test were used to analyze non-dependent and dependent rats’ responses during self-administration training and following drug microinfusions.

For the human post-mortem study, data were analyzed using GLM that tested the main effects of group, brain regions, and group by regions interaction. Post hoc independent samples t-tests were conducted to determine the direction of difference in receptor mRNA expression. Effect sizes were calculated as η² for GLM and Cohen d for t-test. Partial correlation (smoking status as covariate) was used to analyze the effects of alcohol/smoking-related outcomes on mRNA expression level.

Results

Noradrenaline regulates CeA activity

We first assessed CeA neuronal excitability in naïve rats. Only a subset of neurons displayed spontaneous action potential firing with a regular discharge pattern (Table S1) (25,26). NA (1 μM, 15 min) (34) decreased firing to 37.6±8.3% of baseline in all cells [t(7)=7.51, p<.001 by one-sample t-test], with recovery after 15 min drug washout [t(7)=8.64, p<.001 by paired t-test] (Fig. 1A–B; Fig. S3; Table S3). Since NA has only minor effects on membrane properties, its firing effects are likely due to extrinsic factors (Fig. S4).

Figure 1.

NA increases CeA GABA release via α1 receptors in naïve rats. A: Noradrenaline (1 μM NA for 15 min) decreased action potential firing in all neurons (n=8 cells from 5 rats). B: Representative firing traces and graph demonstrating that NA decreased the neuronal firing rate, with recovery after a 15 min ACSF wash period. C: Pharmacological blockade of GABA transmission prevented the NA-induced decrease in neuronal firing (n=7 cells from 5 rats, though 1 cell did not complete washout). D: NA increased the sIPSC frequency in 9/15 CeA neurons and had no effect in the remaining neurons (N=14 rats). E: Representative sIPSC traces and graph demonstrating that NA increased the sIPSC frequency in the majority of CeA neurons, with no effect on sIPSC amplitude or kinetics. F: NA did not alter the sIPSC properties of the remaining neurons. G: Representative mIPSC traces and graph demonstrating that NA had no effect on mIPSC characteristics (n=7 cells from 4 rats). H: Pretreatment with the α1 receptor inverse agonist (10 μM prazosin) prevented NA’s actions on the sIPSC frequency in 6/8 neurons (N=6 rats), indicating that NA-induced GABA release is mediated by the α1 receptor. I: After pretreatment with the β receptor inverse agonist (20 μM propranolol), NA was still able to increase the sIPSC frequency (n=6 cells from 4 rats). All data are normalized to a pre-drug baseline and presented as mean±SEM. **p<.01, ***p<.001 by one-sample t-test; ^###p<.001 by paired t-test.

Blocking GABA transmission prevented NA’s firing effects, suggesting it is mediated by inhibitory synaptic activity (Fig. 1C). Thus, we recorded sIPSCs (Table S2). NA increased sIPSC frequency to 212.6±24.0% of baseline in 9/15 cells [t(8)=4.69, p<.01 by one-sample t-test] (Fig. 1D–E), with no effect in the remaining cells (Fig. 1D,F). NA had no effects on sIPSC amplitude or kinetics, regardless of frequency effects. In contrast to sIPSCs, NA had no effect on the action potential-independent mIPSC frequency (Fig. 1G). s/mIPSC frequencies index probability of vesicle release, while amplitude and kinetics reflect postsynaptic receptor function (25,26). Thus, NA increases CeA GABA release in an action potential-dependent manner in naïve rats.

α1 receptors drive CeA noradrenergic function

We next investigated the synaptic role of specific adrenergic receptor subtypes. Fifteen min pretreatment with α1 (10 μM prazosin) or β receptor (20 μM propranolol) inverse agonists had no effect on sIPSC frequencies (praz or prop alone in Fig. 1H–I), indicating a lack of tonic activity. Prazosin blocked NA’s potentiation of the sIPSC frequency in 6 cells, though NA still had an effect in 2 cells (Fig. 1H). Note, all data were normalized to a single pre-prazosin baseline to track time-course effects. Propranolol did not hinder NA’s potentiating effect [t(5)=5.22, p<.01 by one-sample t-test; t(5)=7.00, p<.001 by paired t-test] (Fig. 1I). Thus, NA/α1 signaling potentiates CeA GABA release in naïve rats.

Acute ethanol does not interact with noradrenaline

Since acute ethanol (EtOH) increases CeA GABA release (25), we assessed possible NA-EtOH shared mechanisms. EtOH pretreatment (44 mM for 15 min, which produces a maximal CeA effect (35)) did not hinder NA-induced GABA release [t(6)=3.21, p<.05 by paired t-test] (Fig. 2A). Specifically, EtOH increased sIPSC frequency [t(6)=4.81, p<.01 by one-sample t-test], and subsequent NA co-application further increased it [t(6)=3.76, p<.01]. Note, all data were normalized to a single pre-EtOH baseline to track time-course effects.

Figure 2.

Acute alcohol does not interact with the CeA noradrenergic system. A: Pretreatment with acute ethanol (44 mM EtOH) increased the sIPSC frequency, and noradrenaline (1 μM NA) further potentiated the sIPSC frequency in 4/7 neurons compared to EtOH alone (N=5 rats). B: NA pretreatment either increased or had no effect on the sIPSC frequency, and EtOH co-application furthered increased it compared to NA alone (n=10 cells from 8 rats). C: The magnitude of EtOH’s effect on the sIPSC frequency in neurons pretreated with NA was similar to EtOH alone (cells from panel A and B). D: The magnitude of NA’s effect on the sIPSC frequency in neurons pretreated with EtOH was similar to NA alone (cells from panel A and B). All data are normalized to a pre-drug baseline and presented as mean±SEM. *p<.05, **p<.01 by one-sample t-test; ^#p<.05 by paired t-test.

Similarly, NA pretreatment did not alter EtOH’s ability to increase GABA release [p<.01 by Wilcoxon matched-pairs signed rank test] (Fig. 2B). Overall, NA increased sIPSC frequency [t(9)=2.32, p<.05 by one-sample t-test], though it had no effect in half the cells. Of the 5 NA-responsive cells, 4 responded to subsequent EtOH with a further increase in sIPSC frequency [across all cells: t(9)=2.78, p<.05 by one-sample t-test]. Notably, the magnitude of EtOH and NA’s potentiating effects were similar in the absence or presence of the other drug (Fig. 2C–D). Thus, acute alcohol and NA do not interact at CeA GABAergic synapses in naïve rats, suggesting divergent downstream mechanisms to regulate GABA release.

Alcohol dependence dysregulates noradrenaline’s cellular effects

Alcohol dependence (CIE; Fig. S1) dysregulated noradrenergic signaling, such that NA had mixed effects on cellular and synaptic activity. NA decreased firing to 40.7±7.6% of baseline in 10 cells [t(9)=7.82, p<.001 by one-sample t-test], increased it to 189.2±24.5% in 4 cells [t(3)=3.65, p<.05], and had no effect in 3 cells (Fig. 3A–B; Fig. S3; Table S3). NA also increased the sIPSC frequency to 175.8±18.7% of baseline in 8 cells [t(7)=4.05, p<.01], decreased it to 49.9±4.5% in 9 cells [t(8)=11.07, p<.001], and had no effect in 1 cell (Fig. 3C–D), without altering the other sIPSC properties. Thus, in alcohol-dependent rats, NA can enhance or reduce inhibitory input onto subsets of CeA neurons, which may drive the observed effects of NA on firing as in naïve rats (see Fig. 1C).

Figure 3.

Alcohol dependence recruits tonic α1 and NA-induced β receptor signaling. A: Noradrenaline (1 μM NA) decreased action potential firing in 10/17 neurons, increased it in 4/17 neurons, and had no effect in the remaining neurons (N=14 rats). B: Representative firing traces and graphs demonstrating that NA either decreased or increased the neuronal firing rate. C: NA increased the sIPSC frequency in 8/18 neurons, decreased it in 9/18 neurons, and had no effect in the remaining neuron (N=14 rats). D: Representative sIPSC traces and graphs indicating that NA either increased or decreased the sIPSC frequency. E: 10 μM prazosin decreased sIPSC frequency, indicating tonic α1 receptor activity. After prazosin pretreatment, NA either decreased or had no effect on the sIPSC frequency (n=9 cells from 7 rats). F: After 20 μM propranolol pretreatment, NA increased or had no effect on the sIPSC frequency, revealing β receptor recruitment in dependence (n=8 cells from 5 rats). All data are normalized to a pre-drug baseline and presented as mean±SEM. *p<.05, **p<.01, ***p<.001 by one-sample t-test; ^#p<.05, ^##p<.01 by paired t-test.

Dependence enhances α1 activity and recruits β signaling

In dependent rats, prazosin alone decreased sIPSC frequency, indicating that CIE leads to basal α1 activation enhancing GABA tone [t(8)=3.54, p<.01 by one-sample t-test] (Fig. 3E). Additionally, NA’s ability to enhance GABA release was prevented by prazosin (Fig. 3E), similar to naïve rats (see Fig. 1H). Instead, NA co-application with prazosin decreased sIPSC frequency [t(8)=8.59, p<.001 by one-sample t-test; t(8)=3.38, p<.01 by paired t-test], revealing a second adrenergic mechanism that reduces GABA input in a subset of CeA neurons (Fig. 3E). Interestingly, propranolol prevented NA’s ability to reduce GABA release, suggesting that NA’s disinhibitory effect is mediated by β receptors (Fig. 3F). Finally, NA co-application with propranolol significantly increased the sIPSC frequency in 5/8 cells [t(7)=3.93, p<.01 by one-sample t-test; t(7)=2.9, p<.05 by paired t-test], highlighting that NA’s potentiating effects are not β-mediated (Fig. 3F). Thus, NA’s dual effects on CeA GABA release after alcohol dependence require distinct adrenergic receptor subtypes, such that NA activates both α1 and β receptors to promote and limit local inhibition, respectively.

Protracted withdrawal produces CeA functional recovery

After 2 weeks of withdrawal, NA decreased firing to 33.2±7.1% of baseline in 10/11 cells [t(9)=9.45, p<.001 by one-sample t-test] (Fig. 4A–B; Fig. S3; Table S3). NA also increased sIPSC frequency to 264.1±54.7% of baseline in 9/11 cells [t(8)=3.00, p<.05 by one-sample t-test] (Fig. 4C–D), without altering other sIPSC properties. Moreover, prazosin blocked NA’s potentiation of sIPSC frequency in 7/8 cells (Fig. 4E), while propranolol did not [t(7)=2.58, p<.05 by one-sample t-test; t(7)=9.28, p<.001 by paired t-test] (Fig. 4F). Thus, withdrawal produces a functional recovery that resembles the naïve state.

Figure 4.

Noradrenergic regulation of CeA activity recovers with protracted withdrawal. A: Noradrenaline (1 μM NA) decreased action potential firing in 10/11 neurons and had no effect in the remaining neuron (N=9 rats). B: Representative firing traces and graph demonstrating that NA decreased the firing rate in most neurons. C: NA increased the sIPSC frequency in 9/11 CeA neurons and had no effect in the remaining neurons (N=10 rats). D: Representative sIPSC traces and graph demonstrating that NA increased the sIPSC frequency in most neurons. E: After 10 μM prazosin pretreatment, NA either decreased or had no effect on the sIPSC frequency in 7/8 neurons (N=6 rats). F: After 20 μM propranolol pretreatment, NA increased the sIPSC frequency (n=8 cells from 6 rats). All data are normalized to a pre-drug baseline and presented as mean±SEM. *p<.05, ***p<.001 by one-sample t-test; ^###p<.001 by paired t-test.

Dependence and withdrawal do not alter CeA NA levels

Given this alcohol-induced noradrenergic dysfunction, we assessed CeA NA tissue content and found no differences in NA concentration across naïve (M=1.56 pmol/mg), dependent (M=1.70 pmol/mg) and withdrawn (M=1.08 pmol/mg) rats [F(2, 15)=2.70, p=0.09] (Fig. 5A). CeA dopamine (DA) concentrations were also unchanged [F(2, 15)=1.18, p=0.33] (Fig. 5B). Therefore, CeA NA release likely does not contribute to the enhanced adrenergic receptor signaling observed during dependence.

Figure 5.

Dependence and withdrawal do not alter CeA NA concentration. Tissue concentrations (pmol/mg) of A: noradrenaline (NA) or B: dopamine (DA) in the CeA of naïve, alcohol-dependent (Dep), or withdrawn (WD) rats. All data are presented as mean±SEM. N=5–7 rats/group.

Dependence and withdrawal selectively alter CeA adrenergic receptor-expressing cell populations

We next examined how dependence and withdrawal engages CeA adrenergic receptor signaling using in situ hybridization (ISH)/RNAscope to quantify the percent of cells expressing α1 (Adra1a+), β1 (Adrb1+) and β2 mRNA (Adrb2+) (Fig. 6A). The number of Adra1a+ [F(2,40)=3.613, p<0.05 by one-way ANOVA; p<0.05 by Tukey’s test] and Adrb1+ [F(2,40)=3.37, p<0.05] cells were significantly decreased after chronic ethanol (Fig. 6B,C). In contrast, Adrb2+ cells (Fig. 6D) and cells co-expressing Adrb1 and Adrb2 were unchanged (Fig. 6E). Therefore, CeA cell populations expressing α1 and β1 mRNA were selectively altered by dependence and withdrawal, perhaps in direct response to treatment and/or as part of a compensatory mechanism involved in functional recovery.

Figure 6.

Withdrawal alters α1 and β mRNA expression patterns in the CeA. A: Representative images of Adra1a (red), Adrb1 (green), Adrb2 (yellow) and DAPI (blue) for naive, dependent (Dep), and withdrawal (WD) rats. Scale bar = 10μm. Summary bar graphs indicate the change in the percent of nuclei expressing B: Adra1a, C: Adrb1, D: Adrb2 and E: co-expressing Adrb1+Adrb2 in the medial CeA of Dep and WD rats relative to the Naïve group. All data are presented as mean±SEM. ^p<.05 by Tukey’s post hoc test; ^$p<.05 by one-way ANOVA; n=13–15 images from 4–6 rats/group.

Dissociable CeA adrenergic mechanisms drive alcohol drinking in rodents

The adrenergic receptor subtype-specificity of chronic alcohol’s molecular and synaptic actions prompted investigation of their role in alcohol consumption. Alcohol self-administration was higher in dependent rats compared to baseline pre-CIE levels and to non-dependent rats (see Supplemental Materials; Fig. S2,5; Fig. 7A–B). Intra-CeA infusion of the α1 inverse agonist prazosin (60 nmol) reduced alcohol intake in non-dependent, but not dependent rats [p<.05 vs. vehicle by Sidak post hoc test following two-way RM ANOVA with non-dep vs. dep rats: F(1,26)=4.98, p<.05; prazosin: F(3,78)=5.79, p<.01; interaction: F(3,78)=2.46, p>.05] (Fig. 7A). Inactive lever responses were unchanged [non-dep vs. dep: F(1,26)=0.75, p>.05; prazosin: F(3,78)=0.21, p>.05; interaction: F(3,78)=0.57, p>.05] (Fig. 7C). In contrast, the dependence-induced escalation in alcohol intake was reduced by intra-CeA infusion of the β inverse agonist propranolol (60 nmol) [p<.001 vs. vehicle by Sidak post hoc test following two-way RM ANOVA with non-dep vs. dep: F(1,26)=12.80, p<.01; propranolol: F(3,78)=6.86, p<.001; interaction: F(3,78)=2.65, p>.05] (Fig. 7B). Propranolol did not alter alcohol consumption in the non-dependent rats (Fig. 7B). Inactive lever responses were unchanged [non-dep vs. dep: F(1,26)=2.34, p>.05; propranolol: F(3,78)=1.72, p>.05; interaction: F(3,78)=0.66, p>.05]. Of note, baseline inactive lever responses were similar (Fig. 7C–D), excluding non-specific locomotor effects induced by the CeA microinfusions. Therefore, our preclinical studies reveal dissociable CeA noradrenergic mechanisms that regulate state-dependent alcohol consumption, with α1 primarily mediating responses in naïve/non-dependent rats and β recruitment occurring during dependence.

α1_B, but not _β1 or β2 receptor expression, is higher in the amygdala of humans with AUD

Finally, we investigated whether the noradrenergic system is also compromised in humans with AUD (individual characteristics described in (21,22)). The amygdala, hippocampus, prefrontal cortex (PFC), VTA and NAc were included for β1 and β2 mRNA analysis, but the VTA and NAc were excluded for α1_B due to a reduced level of mRNA (21). General Linear Model (GLM) analysis of α1_B adrenergic receptor mRNA levels in the amygdala, hippocampus and PFC revealed no main effect on group (AUD and controls) or on brain regions (p>.05); however there was a group by brain region interaction [F(2,62)=5.273, p<.01, η²=0.145]. Post hoc analysis showed a significant increase in amygdala α1_B mRNA levels [t(22)=−2.77, p=.011, d=1.1], with no differences in the hippocampus and PFC (Fig. 8A). GLM analysis of β1 and β2 receptor expression level in all regions revealed no main effect on group, brain regions, or group by brain region interaction (Fig. 8B–C). However, there was a trend for β1 mRNA overexpression selectively in the amygdala [t(20)=−2.7729, p=.165, d=.25] (Fig. 8B). We next examined whether specific alcohol/smoking-related behaviors correlated with α1_B, β1, and β2 amygdalar mRNA expression level. Partial correlation analysis (smoking status as covariate) did not detect differences between adrenergic receptor mRNA expression and alcohol/smoking-related behaviors (Table S4). Overall, these findings support a relationship between human α1_B adrenergic receptor expression in the amygdala and diagnosis of AUD.

Figure 8.

Noradrenergic receptor expression in post-mortem human brain. A: α1_B receptor mRNA expression level (2^−ΔΔCt) is higher in the amygdala in humans with AUD compared to controls, but not in the hippocampus or prefrontal cortex (PFC). There was no α1_B receptor mRNA expression in the VTA and NA and these brain regions were excluded in the analysis. B and C: There was no significant difference in the β1 and β2 receptor mRNA expression level, however, there was a trend for β1 overexpression toward the same direction of the α1_B mRNA only in the amygdala in humans with AUD compared to controls. All data presented as mean±SEM. ^$p<.05 by GLM; *p<.05 by post hoc t-test.

Discussion

Noradrenaline is a potent neuromodulator of CeA activity. We investigated CeA noradrenergic dysregulation after alcohol dependence and protracted withdrawal, focusing specifically on α1 and β1/2 receptors to provide pharmacological and mechanistic insights into ongoing AUD clinical trials. We identified alterations in adrenergic receptor function that modulate GABAergic transmission, thus impacting overall CeA function and likely contributing to AUD. Our preclinical work revealed that NA/α1 potentiates CeA GABA release and α1 activity contributes to the moderate levels of alcohol intake associated with social drinking in humans. Chronic alcohol recruits tonic α1 and NA/β mechanisms that may account for overall dysregulation of CeA GABA signaling, and the escalation of alcohol drinking that defines the dependent state. Protracted withdrawal shifts adrenergic in situ hybridization patterns despite a functional recovery of NA’s synaptic influence, suggesting engagement of compensatory mechanisms. Finally, we identified a relationship between AUD diagnosis and α1_B mRNA expression level in the human amygdala. Of note, this post-mortem analysis has limitations including small sample size and individuals with AUD also smoked nicotine (entered as a covariate). Another limitation is that all investigations were done on male rodents and humans. Collectively, our study suggests that dependence/withdrawal enhances local noradrenergic signaling, leaving the CeA susceptible to further excessive alcohol consumption. These results provide significant mechanistic insights into the noradrenergic system that further support clinical study.

Dissociable adrenergic mechanisms regulate CeA activity

The CeA is a complex brain region with significant cellular heterogeneity (25,26,31,36,37). To date, NA’s regulation of CeA cellular physiology has not been directly investigated. We found that NA engages α1 and β receptors to produce opposing effects on CeA GABA activity, with α1 inhibiting CeA neurons and dependence-induced β activity causing disinhibition. Of note, NA’s dual effects are not restricted to low threshold bursting, regular spiking, or late spiking neurons (data not shown) (38,39).

Interestingly, β activity enhances glutamatergic input to the lateral CeA (40), which likely increases GABA release in the medial CeA in line with our findings in control rats. Noradrenergic signaling in the basolateral amygdala (BLA) regulates inhibitory transmission in a circuit-specific manner with NA/α1 signaling enhancing GABA release from BLA interneurons (41,42), while β activity facilitates lateral paracapsular GABAergic inputs (43) Since we observed no change in CeA NA levels, we speculate that a similar shift in adrenergic function could produce NA’s dissociable effects within the CeA microcircuitry after dependence. Our recent ex vivo fast-scan cyclic voltammetry study showed that evoked rat CeA NA release is decreased by CRF, but not acute ethanol (44). These interactions are further complicated by the intricate nature of inhibitory CeA microcircuits (15), and neuronal subpopulations (e.g., CRF1+) that are ethanol sensitive (31,39). Future studies using viral strategies and/or transgenic animals are needed to resolve adrenergic receptors expression patterns and the cell specificity of NA responses.

Alcohol dependence and withdrawal dysregulate the CeA noradrenergic system

We observed that dependence induces tonic α1 activity and disinhibits a subset of CeA neurons via NA/β signaling. It is unclear whether these shifts in adrenergic function are causes or consequences of dependence. Regardless, these neuroadaptations reflect the prominent role of the CeA in the escalation of drinking, and may contribute to CeA overactivation that marks dependence (15,16). Protracted withdrawal functionally recovers NA’s effects, but there is an associated decrease of α1- and β-expressing cells. In humans, AUD produces widespread noradrenergic dysfunction that persists into withdrawal (45,46), with elevations in NA and its metabolites in the brain and cerebral spinal fluid (6,47,48). In contrast, we found no change in CeA NA content after dependence or protracted abstinence, though these tissue levels may not reflect extracellular concentrations. Nonetheless, collectively our data suggest that compensatory mechanisms are engaged to achieve functional recovery, highlighting the importance of the noradrenergic system in the CeA remediation during withdrawal. Likewise, human post-mortem brain analysis revealed higher amygdala α1_B mRNA levels in AUD. Notably, these results are specific to the amygdala, among the regions investigated. No differences in β1 and β2 mRNA receptor expression were observed; however for β1 mRNA, we found a trend for overexpression only in the amygdala. Notably, in clinical settings, β-blockers mitigate withdrawal syndrome (5,49) and individuals with AUD in this study were actively drinking (confirmed by high BAC at the time of death). Thus, our molecular results across rodents and humans support a role of amygdalar adrenergic receptors in AUD.

Dynamic effects of alcohol’s noradrenergic effects may underlie the observed differences across studies. Most prominently, alcohol decreased LC neuronal firing (50), but increased LC c-fos expression and downstream NA release (7,51,52). Some of these effects are sexually dimorphic, with females lacking LC habituation to repeated ethanol exposures (53,54). One limitation of this study is the use of only male subjects; future studies should explore sex differences in how ethanol impacts CeA adrenergic mechanisms. Additionally, in the post-mortem human tissue study, all individuals with AUD also smoked, though there was no correlation between cigarette packs/year and amygdala α1_B mRNA expression. This finding may hold clinical neuroscience relevance as AUD and nicotine use disorder are highly comorbid. Finally, the post-mortem tissue study only examined mRNA expression of the α1_B subtype, but prazosin has non-selective actions on all three α1 subtypes (α1_A, α1_B and α1_D) (5,56)_.

Targeting α1 and β receptors to control drinking

The precise noradrenergic mechanisms that promote AUD are unknown. Previous preclinical studies have implicated α1 and β receptors using various alcohol exposure paradigms. Prazosin reduced alcohol responding in dependent (1.5 and 2 mg/kg) and non-dependent rats (2 mg/kg only) (8), alcohol intake and anxiety-like behavior after chronic home cage drinking (57), and alcohol seeking, intake and relapse in alcohol-preferring P rats (58–61). In contrast, propranolol (2.5–10 mg/kg) selectively reduced drinking in dependent rats, with only the highest dose impacting non-dependent responding (9), though its intake effects (5–15 mg/kg) in early withdrawal were inconsistent in free-choice drinking P rats (62). Interestingly, propranolol (2–5 mg/kg) also suppressed tremors and audiogenic convulsions in chronic ethanol-exposed rats, though it did not alleviate other withdrawal symptoms (63). Since these studies all used systemic drug administration, we examined CeA-specific α1 vs. β mechanisms.

To the best of our knowledge, this is the first study to investigate adrenergic effects on dependence-induced alcohol drinking in a brain region-specific manner. We found that CeA microinfusion of prazosin only affected alcohol consumption in non-dependent rats, suggesting that its systemic effects in previous studies may be due to actions in other brain regions. Conversely, intra-CeA propranolol selectively reduced drinking in dependent rats. Since propranolol alleviates some anxiety/stress-related aspects of alcohol withdrawal syndrome (63,64) and reduces the motivation to consume alcohol (9), our data suggests that CeA β signaling may significantly contribute in multiple ways to the negative reinforcing effects of alcohol that define the dependence/withdrawal cycle. Thus, the noradrenergic system plays a crucial role in the CeA’s neurobiological response to alcohol, and its recruitment can drive excessive drinking in AUD.

Adrenergic receptors represent promising targets for treating AUD, with a few clinical trials already underway. Prazosin generally improved drinking outcomes, and reduced alcohol craving and anxiety in AUD patients ((10–13), though see (65)). Moreover, doxazosin (α1 receptor blocker) specifically benefits individuals with family history of alcoholism and higher blood pressure, highlighting the importance of personalized treatment in AUD (66,67). Likewise, β-blockers (propranolol or atenolol) reduce alcohol withdrawal symptoms, alcohol craving, and anxiety symptoms in AUD patients (68,69). These studies provide strong evidence that targeting adrenergic receptors may be effective in treating certain subtypes of AUD, particularly based on family history and the presence of stress symptoms.

Developing individualized, highly specific treatment requires a precise neurobiological understanding of the complex interactions between alcohol and the noradrenergic stress system. In this study, we have established CeA adrenergic receptors as a key neural substrate of AUD. Specifically, α1 potentiates CeA inhibition and drives the moderate levels of alcohol intake associated with social drinking in humans. β activity disinhibits a subset of CeA neurons, leading to excessive drinking in dependence. While protracted withdrawal produces functional recovery of the CeA noradrenergic system, there are long-lasting changes in adrenergic receptor-expressing cells. Additionally, α1_B receptors were over-expressed in the amygdala in individuals with AUD. Identification of these novel noradrenergic mechanisms as potential significant drivers of alcohol intake can guide ongoing medication development, particularly with regards to the identification of AUD symptoms and patient subtypes that may especially benefit from specific adrenergic compounds.

KEY RESOURCES TABLE.

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
Biological Sample	Human post-mortem brain tissue, male	New South Wales Tissue Resource Center (NSWBTRC) at the University of Sydney, Australia (PMID: 27139235)
Chemical Compound or Drug	noradrenaline bitartrate	Tocris	Cat. No. 5169/50
Chemical Compound or Drug	prazosin hydrochloride	Tocris	Cat. No. 0623/100
Chemical Compound or Drug	propranolol hydrochloride	Tocris	Cat. No. 0624/100
Chemical Compound or Drug	DL-AP-5	Tocris	Cat. No. 0105/10
Chemical Compound or Drug	CGP 55845A	Tocris	Cat. No. 1248/10
Chemical Compound or Drug	DNQX	Tocris	Cat. No. 2312/10
Chemical Compound or Drug	bicuculline	Sigma	14340
Chemical Compound or Drug	tetrodotoxin	Sigma	554412
Chemical Compound or Drug	alcohol	Remet
Commercial Assay Or Kit	RNAscope fluorescent multiplex kit	ACD Biotechne	ACD 320850; Probes: negative control (320751), Adra1a (592781-C2), Adrb1 (468121-C1), Adrb2 (468131-C3).
Commercial Assay Or Kit	RNeasy Lipid Tissue mini Kit	Qiagen	74804
Commercial Assay Or Kit	RNA 6000 Nano kit	Agilent	5067-1511
Commercial Assay Or Kit	SuperScript® III First-Strand Synthesis SuperMix	Invitrogen	18080400
Commercial Assay Or Kit	TaqMan PCR assay	Thermo Fisher	adrenoceptor alpha 1B, ADRA1B, Hs00171263_m1; adrenoceptor beta 1, ADRB1, Hs02330048_s1; and adrenoceptor beta 2, ADRB2, Hs00240532_s1
Organism/Strain	Rat: Sprague Dawley, male	Charles River Laboratories (Raleigh, NC)
Organism/Strain	Rat: Wistar, Male	Charles River Laboratories (Raleigh, NC)
Software; Algorithm	Mini Analysis software	Synaptosoft Inc.
Software; Algorithm	pClamp 10 software	Molecular Devices
Software; Algorithm	NeuroExpress software	Dr. A. Szucs (PMID: 32769108; PMID: 29471053)
Software; Algorithm	Prism software (v.9)	GraphPad
Software; Algorithm	R software	https://www.r-project.org/
Software; Algorithm	Statistical Package for the Social Sciences software (SPSSv.26)	IBM

Data and materials availability:

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials.