Ethanol blocks a novel form of iLTD, but not iLTP of inhibitory inputs to VTA GABA neurons

Abstract

The ventral tegmental area (VTA) is an essential component of the mesocorticolimbic dopamine (DA) circuit that processes reward and motivated behaviors. The VTA contains DA neurons essential in this process, as well as GABAergic inhibitory cells that regulate DA cell activity. In response to drug exposure, synaptic connections of the VTA circuit can be rewired via synaptic plasticity—a phenomenon thought to be responsible for the pathology of drug dependence. While synaptic plasticity to VTA DA neurons as well as prefrontal cortex to nucleus accumbens GABA neurons are well studied, VTA GABA cell plasticity, specifically inhibitory inputs to VTA GABA neurons, is less understood. Therefore, we investigated the plasticity of these inhibitory inputs. Using whole cell electrophysiology in GAD67-GFP mice to identify GABA cells, we observed that these VTA GABA cells experience either inhibitory GABAergic long-term potentiation (iLTP) or inhibitory long-term depression (iLTD) in response to a 5 Hz stimulus. Paired pulse ratios, coefficient of variance, and failure rates suggest a presynaptic mechanism for both plasticity types, where iLTP is NMDA receptor-dependent and iLTD is GABA_B receptor-dependent—this being the first report of iLTD onto VTA GABA cells. As illicit drug exposure can alter VTA plasticity, we employed chronic intermittent exposure (CIE) to ethanol (EtOH) vapor in male and female mice to examine its potential impact on VTA GABA input plasticity. Chronic EtOH vapor exposure produced measurable behavioral changes illustrating dependence and concomitantly prevented previously observed iLTD, which continued in air-exposed controls, illustrating the impact of EtOH on VTA neurocircuitry and suggesting physiologic mechanisms at play in alcohol use disorder and withdrawal states. Taken together, these novel findings of unique GABAergic synapses exhibiting either iLTP or iLTD within the mesolimbic circuit, and EtOH blockade specifically of iLTD, characterize inhibitory VTA plasticity as a malleable, experience-dependent system modified by EtOH.

Introduction

The mesocorticolimbic dopamine (DA) circuit contains important, evolutionarily conserved pathways that identify and promote species-perpetuating and survival behaviors [1]. Dopamine neurons in the ventral tegmental area (VTA) project to the nucleus accumbens (NAc) where they release DA onto medium spiny neurons (MSNs) [2]. A common feature of drugs such as opiates, amphetamines, tetrahydrocannabinol (THC), cocaine, nicotine, ethanol (EtOH), and other addictive substances is an increase in DA release or synaptic DA accumulation in the mesocorticolimbic pathway with acute exposure [3, 4]. Substance use disorder is a debilitating brain disease involving adaptations in neural circuitry and synaptic transmission in response to chronic drug exposure [5, 6]. Persons experiencing substance use disorder can no longer process salience normally and experience withdrawals, cravings, and deterioration of executive function in both the acute and chronic phases of addiction.

Synaptic plasticity induced or altered by drug exposure was initially observed in excitatory inputs to VTA DA cells [7], and drug-dependent plasticity has since been thoroughly described at both excitatory and inhibitory inputs onto DA cells [8–16]. While the DA cells involved in dependence have been well studied, research shows that the inhibitory GABA neurons or GABA receptors modulating DA cells are also altered in response to drug exposure [16–18]. However, less is understood about GABA neuron inputs, and how they change following drug exposure.

Ethanol is one of many substances that alters mesolimbic circuitry and can lead to addiction, which is an expensive and devastating problem for societies around the world, with over 100 million individuals worldwide suffering from alcohol use disorder (AUD; [19]. In the United States, 3 in 10 individuals meet criteria for unhealthy or risky alcohol use (NIAAA, 2004), and this unhealthy alcohol use increases mortality and suicidality in middle aged adults in the United States [20, 21]. Clinical treatments for AUD are limited and have only modest efficacy for most patients [22]. In the rodent NAc Acute EtOH exposure increases DA release [23, 24] and causes paired pulse depression at the GABA to DA synapse [25], while chronic EtOH exposure leads to decreased DA release in the NAc [26, 27]. This may explain EtOH-seeking behavior and cravings during withdrawal. Acute in vivo EtOH exposure in rodent models decreases GABA neuron activity in the VTA, specifically in GABA neuron firing rate and GABA release [28]. While there is a wide array of studies illustrating EtOH impact on VTA circuitry and receptor activity, this study focused on EtOH-induced effects on synaptic plasticity. For example, while EtOH is known to modulate synaptic plasticity of inputs to VTA DA cells including at glutamatergic synapses [10, 12] and GABAergic synapses [25], its effects on synaptic plasticity of inputs to VTA GABA cells are unknown, but may be physiologically and behaviorally relevant as illustrated above.

Therefore, we used whole cell electrophysiology ex vivo to study the modulation of inhibitory inputs to VTA GABA neurons. We not only observed the previously reported LTP at this synapse [16], but also describe a novel form of iLTD. Next, we investigated how this plasticity at the GABA inputs to VTA GABA cells was influenced by chronic EtOH exposure using behavioral models and whole cell electrophysiology.

Methods

Materials and methods

All experiments were performed in accordance with Institutional Animal Care and Use Committee protocols, following National Institutes of Health Guide for the care and use of laboratory animals. Experiments were approved by the Brigham Young University Institutional Animal Care and Use Committee, Animal Welfare Assurance Number A3783-01.

Animals

Male and female juvenile (P15–P40) or young adult (P60-P90) CD1-background GAD67/GFP knock-in mice were used to identify GABA neurons using a GFP reporter [29]. Chronic EtOH and chronic air experiments were performed on P60-P90 mice due to the few weeks in the vapor chambers, while the remainder of the experiments were performed on P15-P40 mice. No differences were noted between plasticity based on age or gender, and thus all mice were combined unless noted otherwise. Most mice were used for one experiment each. Glutamate decarboxylase (GAD65/GAD67) is required to convert glutamate to GABA and is a key cell marker of GABA neurons. The GAD65 gene can co-express in some tyrosine hydroxylase+ cells [30–32]. However, cells expressing GAD67 comprise the largest subset of GABA neurons within the VTA and have not been found in co-expression with DA neuron proteins [32].

Slice preparation

Mice were anesthetized with isoflurane (1–2%) and decapitated with a rodent guillotine. Brains were rapidly removed and sectioned transversely on a vibratome at 250 μm (young adult) or 300 μm (juvenile). Juvenile brains were sliced using an ice-cold, sucrose-based cutting solution composed of 220 mM sucrose, 0.2 mM CaCl₂, 3 mM KCl, 1.25 mM NaH₂PO₄, 25 mM NaHCO₃, 12 mM MgSO₄, 10 mM glucose, and 400 μM ascorbic acid. After sectioning, the slices were placed in oxygenated ACSF composed of 119 mM NaCl, 26 mM NaHCO₃, 2.5 mM KCl, 1 mM NaH₂PO₄, 2.5 mM CaCl₂, 1.3 mM MgSO₄, and 11 mM glucose in an incubator at 35° Celsius for an hour and then transferred to room temperature until recording. To better preserve adult VTA slices, adult brains were sliced using an ice-cold NMDG-based cutting solution composed of 92 mM NMDG, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 30 mM NaHCO₃, 20 mM HEPES, 25 mM glucose, 2 mM thiourea, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 0.5 mM CaCl₂·2H₂O, and 10 mM MgSO₄·7H₂O, titrated to 7.5 pH with HCl. After sectioning, adult slices were kept in a warm bath of the NMDG-based cutting solution and were spiked with increasing amounts of NaCl over 20 min before being placed in an oxygenated HEPES holding solution composed of 92 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 30 mM NaHCO₃, 20 mM HEPES, 25 mM glucose, 2 mM thiourea, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 2 mM CaCl₂·2H₂O, and 2 mM MgSO₄·7H₂O titrated pH to 7.5 with several drops of concentrated 10 N NaOH [33]. Several 5 Hz experiments were performed in adults cut using the sucrose ACSF protocol, as well as several 5 Hz experiments from adolescent animals sliced with the NMDG perfusion protocol, to confirm the NMDG slicing protocol did not influence plasticity and no significant difference was found between methods (data not shown).

Recording Protocol

Recordings began at least 1 h after cutting. Slices were placed in the recording chamber and bathed with oxygenated (95% O2, and 5% CO2) high divalent ACSF (4 mM CaCl₂ and 4 mM MgSO₄) at 30–32 degrees Celsius. High divalent ACSF reduced cell-spiking currents resulting from evoking IPSCs in GABA cells, which ‘contaminate’ synaptic currents [34, 35]. Regular ACSF experiments confirmed that control iLTD and iLTP still occurred in typical divalent ion concentrations. Excitatory glutamate currents were blocked using 10 μM CNQX (Alomone Labs). NMDA antagonist APV was applied in some cases to determine NMDA-dependence. The VTA was visualized using an Olympus BX51W1 microscope. GAD67-GFP^+/− cells were identified by fluorescence and located in approximately the following coordinates from adult mouse bregma: anteroposterior −2.9 to −3.1, mediolateral 0.1 to 0.4, dorsoventral −3.9 to −4.1. GFP⁺ cells were patched with a borosilicate glass pipette (3–6 MΩ) filled with internal solution composed of 117 mM KCl, 2.8 mM NaCl, 20 mM HEPES, 5 mM MgCl₂, 2 mM ATP-Na, 0.3 mM GTP-Na, and 1 mM QX-314 at pH 7.28 with an osmolarity at 275–285 mOsm. Recordings were made in voltage clamp with cells held at −65 mV. Plasticity was induced using a 5 Hz or 100 Hz stimulation in current clamp mode based on prior published ranges to induce plasticity [36] and delivered using a concentric bipolar electrode (Microprobes for Life Science) 200–400 µm from the patched cell. Currents were amplified using Multiclamp 700 B and digitized with an Axon 1440 A digitizer (Molecular Devices). Signals were filtered at 4 kHz and recorded using Clampex 10.7 (Molecular Devices). Electrophysiological data was analyzed using Clampfit software (Molecular Devices), Microsoft Excel, and Origin 10.8 (OriginLab Corporations, Northampton, MA, USA). Cell input and series resistance were monitored continuously throughout each experiment by 5 mV 100 ms hyperpolarization pulses; data were discarded if resistance changed by more than 15%. Input-output IPSC curves were created by post-hoc analysis of raw data using two-minute averages of IPSCs from CIE and chronic air controls.

Single-cell PCR

Cells from the wild type littermates of GFP knock-in mice were used for PCR analysis and extracted using gentle suction and placed into chilled reverse transcriptase reagents (BioRad) within 2 h. One control sample of artificial cerebral spinal fluid was obtained for each slice and used to identify contamination from extracellular mRNA. Using iScript cDNA Synthesis kit (BioRad), extracted cells were reverse transcribed to cDNA under the manufacturer’s protocol and cycled in a C1000 Thermocycler (BioRad) at 25 °C for 8 min, 42 °C for 60 min, and 70 °C for 15 min. Following reverse transcription, cells were multiplexed using primer sequences as described previously [32]. Each pre-amplified cell was run for every target individually in triplicate with the appropriate FAM-TAMRA probe. Each cell was run in a CFX96 qPCR machine (BioRad) with a 95 °C hot start for 3 min, followed by 60 cycles of 95 °C for 15 s, 57 °C for 25 seconds, and 72 °C for 25 s. Data was analyzed as described previously [32].

Chronic intermittent EtOH exposure

Chronic Intermittent Ethanol (CIE) exposure models EtOH withdrawal. The methods followed our previously cited research [37, 38]. Briefly, once weaned at postnatal Day 21, GAD67/GFP^+/− mice were placed on a reverse light/dark cycle. CIE started at post-natal day 28. We modified the vapor chamber system developed in the lab of Graeme Mason at Yale [39]. The alcohol vapor concentration was set in the feedback system with the breathalyzer at 3 L/min level to deliver 200 mg% blood alcohol levels. To confirm blood alcohol levels at this target level, cheek blood samples from some mice were taken and blood alcohol levels were measured using an enzymatic kit (Sigma-Aldrich, St. Louis, MO). The target blood alcohol levels level did not vary by more than 10% in the animals tested. The control animals were housed in the sealed chambers alongside of the alcohol chambers in the same ventilation hood, but only received normal air during the three to six-week period.

Marble burying test

This test measured ethanol withdrawal-induced, compulsive-like behaviors in CIE versus control air mice [40]. Twenty-four hours after vapor treatment, mice were placed for 30 min in a polycarbonate cage (26.5 × 15.5 × 12.5 cm) that contained 5 cm of corncob bedding with 28 clean marbles (13 mm in diameter) evenly distributed throughout the cage. To be counted as buried a minimum of 2/3 of the marble needed to be shielded by the bedding. The number of marbles buried was scored separately by two individuals and the final score was averaged. A higher score correlates to increased compulsive-like behavior [41].

Elevated plus maze

Elevated plus maze (EPM) was used to assess withdrawal-induced changes in anxiety-like behaviors. The apparatus used for the test included two white open arms (25 x 5 x 0.5 cm) across from each other and perpendicular to two black closed arms (25 x 5 x 16 cm) with a square center platform (5 x 5 x 0.5 cm); the entire apparatus is 50 cm above the floor [42]. Each mouse tested was placed in the maze for a 5 min run. The iSpy software was used to record the movement of the subject throughout the maze.

Drugs

APV, L-NAME, baclofen, eticlopride,QX-314, strychnine were obtained from Tocris and dissolved in ddH₂O to make stock solutions. AM251, CPG 54626, SNAP, bicuculline, quinpirole, and LY225910 were obtained from Tocris and dissolved in DMSO to make stock solutions. CNQX was obtained from Alomone Labs and dissolved in ddH2O. All stock solutions were frozen at −20 °C until dilution into ACSF to final concentrations.

Statistical Analysis

Electrophysiology data was analyzed using Clampfit software and statistical tests were computed using Microsoft Excel or SPSS. One-way ANOVA analysis was employed to determine if plasticity was statistically significant by comparing baseline to post-conditioning within every individual experiment. If an experiment was confirmed by ANOVA to display a significant (p < 0.05) increase in averaged post-conditioning responses compared to the baseline, it was grouped with iLTP; if the averaged post-conditioning responses showed a significant (p < 0.05) decrease compared to the baseline, the experiment was grouped with iLTD. Next, the average of the last 5–10 min of the baseline was compared in iLTD to 10–20 min post-conditioning, and for iLTP, to 20–30 min post-conditioning, as iLTP had a delayed onset to full plasticity induction. Grouped data of either iLTP or iLTD was analyzed by ANOVA during these time points to confirm significance in post-conditioning responses to baseline. Normality was assessed and if normally distributed, experiments were analyzed using student T-test (two-tail, unequal variance) to compare between two different experimental groups post-conditioning (i.e., control iLTP/iLTD to experimental iLTP/iLTD). For this analysis, the same 10-minute window at post-conditioning described above was picked from each group. To control for multiple comparisons, a simple Bonferroni correction was implemented for electrophysiology data. A total of 61 comparisons for whole-cell experiments were performed, and a critical level for significance of 0.05/62 (0.00081) was used. PPR and 1/CV² were calculated as described previously [43]. Statistical significance for PPRs, 1/CV², and LTD failure rate were calculated using the Wilcoxon ranked sign test in SPSS. The Wilcoxon ranked test was used due to the high variability of these data, which were not normally distributed. Normally distributed behavioral data was compared using Student T test between experimental and control groups. Lastly, an ANCOVA determined significance of fitted input-output data.

Results

Inhibitory afferents to VTA GABA neurons exhibit iLTP or iLTD

GAD67-GFP knock-in mice [29] positively identify VTA GABA neurons. After 5 Hz conditioning protocol we observed inhibitory synapses onto GABA neurons experienced either inhibitory long-term potentiation (iLTP) or inhibitory long-term depression (iLTD). Both iLTP and iLTD were significantly induced in male and female animals (Fig. 1A, B). Experiments were individually classified as exhibiting either iLTP or iLTD based on single factor ANOVA (see methods section). Either iLTP or iLTD were induced in response to a 5 Hz stimulus (3 min, 900 stimuli) in every case. We never noted experiments that did not exhibit either iLTP or iLTD under control conditions. To isolate GABA_A currents, we eliminated AMPA receptor currents using antagonist CNQX (10 µM) in all electrophysiology experiments unless otherwise noted. Application of CNQX reduced the average currents in GABA cells to 60.4 ± 1.6% of the baseline (Fig. 1C). In addition, we confirmed that currents we recorded were chloride-mediated, as 92.9 ± 1.8% of currents were abolished by GABA_A receptor/chloride channel antagonist picrotoxin (100 µM; Fig. 1D). As picrotoxin is nonselective, we recorded IPSCs in the presence of the selective GABA_A antagonist, bicuculline, which eliminated 90.6% ± 3% of the IPSC (Fig. 1E). We further acknowledge that IPSCs can also be mediated by the neurotransmitter glycine. We determined that the prototypical glycine receptor antagonist, strychnine (1 µM), did reduce IPSCs by 28.4 ± 4.2% (Fig. 1F). To ensure the aforementioned iLTP and iLTD were GABA_A and not glycine receptor-dependent; we performed plasticity experiments in strychnine and continued to observe iLTP and iLTD (Fig. 1G). Additionally, we analyzed paired pulse ratio (PPR) and coefficient of variation (1/CV²) for iLTP and iLTD experiments. In iLTP, PPR significantly decreased and 1/CV² significantly increased, suggesting a presynaptic mechanism for iLTP (Supplementary Fig. 1A). In iLTD, PPR increased and 1/CV² decreased, though neither significantly (Supplementary Fig. 1B). The iLTD-exhibiting GABA neurons displayed notable failure rates, which increased significantly following the stimulus protocol; while iLTP failure rates did not change significantly (Supplementary Fig. 1C, D).

Fig. 1. Inhibitory afferents to VTA GABA neurons exhibit either iLTP or iLTD in response to a 5 Hz stimulus.

A In VTA brain slices from male GAD67/GFP^+/− mice, a 5 Hz stimulus induced iLTP or iLTD of inhibitory postsynaptic currents (IPSCs) to VTA GABA neurons (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 7; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 8). In this and all other electrophysiological experiments, 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX) was included in the bath solution except as noted. B In VTA brain slices from female GAD67/GFP^+/− mice, a 5 Hz stimulus induced iLTP or iLTD of GABA inputs to VTA GABA neurons (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 5; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 7). C AMPA receptor antagonist CNQX (10 µM) reduced currents from all afferents to VTA GABA neurons to 60.4 ± 1.6% (P < 0.0001, ANOVA, n = 20 cells). D GABA_A receptor/chloride channel antagonist picrotoxin (100 µM) eliminated 92.9 ± 1.8% of IPSCs (p < 0.0001, ANOVA, n = 11). E Bicuculline (10 µM), a more selective GABA_A receptor antagonist, eliminated 90.6 ± 3% of IPSCS (p < 0.0001, ANOVA, n = 5). F Glycine receptor antagonist strychnine (1 µM) eliminated 28.4 ± 4.2% of IPSCs (P < 0.0001 compared to baseline, ANOVA, n = 9). G With strychnine (1 µM) in the bath solution, 5 Hz stimulus continued to induce iLTP or iLTD of IPSCs in the VTA GABA neurons (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 7; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 5). In all figures, data are represented as mean ± SEM and scale bars in all trace insets represent 100 pA and 10 ms. Averaged traces (15 traces) in all figures were taken from baseline (black) and post-conditioning or post-drug application (light gray). In 1 A, 1B, 1 G, and throughout the remaining figures, graph insets next to the plasticity data illustrate every cell recorded from showing neuron-by-neuron analysis grouped by those exhibiting iLTP, iLTD or no change (NC). Open circles represent each cell, while filled circles illustrate the group average ± SEM. N-values for each cell are include for ease of assessing the data.

Both iLTP and iLTD are unique synaptic events

As two unique forms of plasticity from similar GABA cells in response to identical stimulus parameters is rare, we did several experiments to test the veracity of our observations. To rule out the possibility that tagging of a GFP reporter to GAD67 might affect neural transmission and alter plasticity, we performed the same experiments on wild-type littermate GABA cells. Again, we note iLTP or iLTD in these GFP-negative animals (Fig. 2A). In the absence of GAD67-GFP fluorescence, we confirmed these were GABA cells by employing spiking during whole-cell patch clamp recordings and post-hoc single cell PCR analysis, which demonstrated respectively a lack of I_h current and expression of GABA cell marker, GAD67, while TH was absent (Fig. 2B). Note that in the absence of conditioning stimulus, no IPSC change was observed in any of our controls, thus confirming IPSCs would stay constant without the introduction of a stimulus (Fig. 2C). While high divalent cation ACSF avoided spiking that ‘contaminates’ synaptic responses, we performed control experiments in regular ACSF to ensure plasticity was not impacted by ACSF type. Both iLTP and iLTD continue in normal ACSF to the same degree and ratio as in high divalent ACSF (Fig. 2D). Next, to ascertain whether the observed iLTP and iLTD were specific to the induction protocol, we employed a different conditioning method, high frequency stimulus (HFS: 2 pulses of 100 Hz). Like the 5 Hz protocol, HFS conditioning induced either iLTP or iLTD of IPSCs (Fig. 2E). We next tested the quantitative nature of this plasticity, (i.e., whether further stimulation could convert iLTD to iLTP), by introducing a second 5 Hz stimulus to an already established iLTP/iLTD. A second application of 5 Hz reinforced the initial plasticity type (Fig. 2F). Our findings suggest that dual plasticity displayed at the VTA GABA-to-GABA synapses is a genuine phenomenon and did not vary based on stimulation type or duration. Subsequently, we continued to use the 5 Hz stimulus throughout our study, as the most physiologically relevant stimulus.

Fig. 2. The iLTP and iLTD observed at the inhibitory VTA GABA synapses are unique synaptic events.

A Inhibitory inputs to VTA GABA neurons from the wild-type littermates (GAD67/GFP^−/−) also exhibited iLTP or iLTD in response to a 5 Hz stimulus (iLTP: p < 0.0001, compared to baseline, ANOVA, n = 5; iLTD: p < 0.0001, compared to baseline, ANOVA, n = 4). Additionally, 15 other cells examined were determined to be non-GABAergic (likely DA) neurons and not included in the data set. B (a) VTA GABA neurons of wild-type animals lacked a hyperpolarization-activated I_h current and (b), (c) were confirmed as GABA cells by post-hoc single-cell PCR. The gel illustrates one example cell, which includes a 50 base pair (bp) ladder, where GAD67 is present at the appropriate amplicon size and TH is absent. Note in the gel, the absence of amplicon for GAD67 in the ACSF background control (GAD67+ACSF). The 18S was our control housekeeping gene. RFU is relative florescent units.  C In the absence of a conditioning stimulus, no change was noted to IPSCs of VTA GABA neurons (P = 0.7523 compared to baseline, ANOVA, n = 9; and P < 0.0001 compared to control iLTP and iLTD in Fig. 1A, Student’s T-test). D In the presence of regular ACSF, iLTP, and iLTD both are induced at levels that are not significantly different from controls in high-divalent ACSF (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 5, and P = 0.6089 compared to iLTP in Fig. 1A, Student’s T-Test; iLTD: p < 0.0001, compared to baseline, ANOVA, n = 6, and P = 0.0864 compared to iLTD in Fig. 1A, Student’s T-Test). E The application of a high-frequency stimulus (100 Hz) at the VTA GABA-to-GABA synapses induced either iLTP or iLTD just like the low-frequency 5 Hz stimulus (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 8; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 7). We also compared the 100 Hz-induced iLTP and iLTD with control iLTP and iLTD (Fig. 1A) at different time points (Minute 10 to 15: iLTP, p = 0.1045, Student’s T-Test, compared to control iLTP; iLTD, p = 0.0897, Student’s T-Test, compared to control iLTD. Minute 20–25, iLTP, p = 0.6249, Student’s T-Test, compared to control iLTP; iLTD, P = 0.168, Student’s T-Test, compared to control iLTD. Minute 30–35, Student’s T-Test, iLTP, P = 0.0968, compared to control iLTP, iLTD, P = 0.2826, Student’s T-Test, compared to control iLTD). F The application of a second 5 Hz stimulus to an already-established iLTP or iLTD at the VTA GABA-to-GABA synapse preserved the plasticity in the same direction in every case (iLTP after first 5 Hz: p < 0.0001 compared to baseline, ANOVA, n = 7; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 8). After the second 5 Hz stimulation, further potentiation of iLTP or depression of iLTD were not significantly different from the first stimulation (2^nd iLTP, p = 0.3190, compared to the 1st iLTP at the last 5 min, ANOVA; 2nd iLTD, p = 0.1657, compared to the 1st LTD at the last 5 min, ANOVA). Scale bar in Fig. 2B (a) represents 200 pA and 2 s. Trace inset in Fig. 2F show baseline (black), first 5 Hz stimulation (light gray), and second 5 Hz stimulation (dark gray).

iLTP is somewhat NMDA receptor dependent

We next examined known mechanisms of presynaptic iLTP to determine cellular mechanisms of observed plasticity at the VTA GABA-to-GABA synapse. Nitric oxide (NO) participates in several types of presynaptic LTP, including LTP in the hippocampus [44], and at the VTA GABA to DA synapses [9, 11]. The NO synthase (NOS) antagonist L-NAME (200 µM) failed to block iLTP (Fig. 3A). However, the NO donor SNAP induced synaptic potentiation in ~50% of cells (Fig. 3B). Thus, some synapses express the machinery for nitric oxide potentiation, but this signaling does not likely mediate iLTP as L-NAME did not eliminate it. As NMDA receptors (NMDARs) are often involved in many LTP forms, we applied NMDAR antagonist APV (50 µM) and interestingly blocked iLTP in some, but not all cases, while leaving iLTD intact, suggesting iLTP is somewhat NMDAR dependent (Fig. 3C).

Fig. 3. The iLTP does not involve the nitric oxide pathway but is somewhat NMDAR-dependent.

A Both iLTP and iLTD were still observed in the presence of NOS antagonist L-NAME (200 µM) (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 6; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 6). B The NO donor SNAP (200 µM) induced significant potentiation in 8 out of 13 cells examined (Potentiated group, P < 0.0001, ANOVA, n = 8; No change group, p = 0.1269 compared to baseline, ANOVA, n = 5, and P < 0.0001 compared to control iLTP and iLTD in Fig. 1A, Student’s T-Test). C The NMDA receptor antagonist APV (50 µM) blocked the previously observed iLTP among some, but not all cells (no iLTP group: p = 0.983 compared to baseline, ANOVA, n = 4, and significantly different compared to control iLTP in Fig. 1A, p < 0.0001 Student’s T-Test; the iLTP group: p < 0.0001 compared to baseline, ANOVA, n = 4). The iLTD group was not effected (p < 0.0001 compared to baseline, ANOVA, n = 7).

iLTD is independent of the cannabinoid 1 (CB1) and dopamine 2 (D2) receptors

Presynaptic LTD forms commonly require CB1 receptors [45]. However, iLTD and iLTP both persisted in the presence of CB1 antagonist AM251 (2 µM; Fig. 4A) and in GAD67/GFP^+/−/CB1^-/- animals (Fig. 4B). Moreover, bath application of the selective CB1 agonist WIN55 (10 µM) did not alter synaptic responses (Fig. 4C), suggesting CB1 is not expressed at this synapse or involved in plasticity. In addition, a subset of VTA GABA neurons respond to DA receptor D2 activation [46], therefore we examined the possible role of D2. Application of D2 antagonist eticlopride (10 µM) did not block either iLTD or iLTP (Fig. 4D), suggesting iLTP/iLTD are D2 receptor independent. Furthermore, the D2 agonist quinpirole did not significantly alter synaptic transmission (Fig. 4E), or holding potential of the GABA cells in voltage-clamp mode (data not shown).

Fig. 4. The iLTD is GABAB receptor-dependent, but independent of the endocannabinoid system and dopamine receptor D2.

A Bath application of CB1 antagonist AM251 (2 µM) failed to block either iLTP or iLTD (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 9; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 7). B Both types of plasticity were present in GAD67/GFP^+/–/CB1^–/– animals (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 10; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 5). C The CB1 receptor agonist WIN55, 212-2 (10 µM) failed to induce depression at this synapse (p = 0.448 compared to baseline, ANOVA, n = 8, and p < 0.0001 compared to control iLTP and iLTD in Fig. 1A, Student’s T-Test). D Blockade of D2 receptors using eticlopride (10 µM) also failed to alter plasticity (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 5; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 8). E Quinpirole (0.1 µM) did not alter synatpic IPCS responses (No change group: p = 0.1891 compared to baseline, ANOVA, n = 6, and p < 0.0001 compared to control iLTP and iLTD in Fig. 1A, Student’s T-Test,), nor holding potential in voltage-clamp mode (data not shown). F Baclofen (50 µM) strongly depressed IPSCs onto VTA GABA neurons (n = 10). Trace inset shows baseline (black), baclofen application (light gray), and washout (dark gray). G GABA_B antagonist CGP 54626 (2 µM) prevented the previously observed iLTD (no iLTD group: p = 0.2570 compared to baseline, ANOVA, n = 5, and p < 0.0001 compared to control iLTD in Fig. 1A, Student’s T-Test), while iLTP remained (p < 0.0001 compared to baseline, ANOVA, n = 7). H A lower concentration of baclofen (1 µM, closer to EC₅₀) eliminated iLTD, while keeping the iLTP intact (no iLTD group: p = 0.484 compared to baseline, ANOVA, n = 6, and p < 0.0001 compared to control iLTD in Fig. 1A, Student’s T-Test; iLTP: p < 0.0001 compared to baseline, ANOVA, n = 5). Interestingly, there was a significant difference in the degree of IPCS depression from baseline, induced by baclofen in GABA cells exhibiting iLTD versus iLTP (p < 0.0001, Student’s T-Test).

iLTD is GABAB receptor-dependent

Next, we examined the potential role of GABA_B receptors in plasticity. We first confirmed the presence of synaptic GABA_B receptors at inputs to VTA GABA cells using the classic GABA_B receptor agonist baclofen. Baclofen (50 µM) strongly depressed recorded IPSCs in all cells (Fig. 4F). We then applied the GABA_B antagonist CGP 54626 (2 µM), which completely blocked iLTD, while leaving iLTP intact (Fig. 4G). Finally, we attempted to occlude this GABA_B-mediated LTD using a lower concentration of baclofen (1 µM, closer to EC₅₀; [47]), allowing for GABA_B activation without eliminating the IPSC. After a stable baseline, baclofen (1 µM) was added, and once a stable baclofen baseline was established the 5 Hz conditioning was introduced. Following the conditioning, we observed that iLTD was indeed occluded while the iLTP was not affected (Fig. 4H). Low-concentration baclofen (1 µM) significantly reduced IPSCs from baseline, with those exhibiting iLTP reduced to 46.7% ± 0.34, and exhibiting iLTD to 61.3% ± 0.54. Interestingly, these were significantly different from each other (Fig. 4H). PPR and 1/CV² analysis of baseline compared to post-baclofen application illustrated no significant change in PPR, but a significant depression in 1/CV² (Supplementary Fig. 2). Finally, baclofen (1 µM) had no effect on holding current levels in voltage-clamp mode (data not shown).

Chronic ethanol eliminated iLTD and increased anxiety and obsessive-compulsive behavior during withdrawal

As ethanol can alter plasticity, we examined the impact of long-term ethanol exposure by examining intermittent ethanol exposure (CIE) treatment (Fig. 5A). We chose the marble burying test (MBT) to examine compulsive-like behavior, and the elevated plus maze (EPM) to examine anxiety-like behavior. All behavioral testing was conducted 24 h after the last ethanol exposure. MBT demonstrated that ethanol withdrawal produced a significant increase in the marbles buried compared to the air-treated controls (Fig. 5B), suggesting an increase in compulsive-like behavior. Likewise, ethanol withdrawal mice spent significantly less time in the EPM open arm and more time in the closed arm compared to air-treated controls (Fig. 5C, D), indicating an increase in anxiety-like behavior. This data demonstrates our methodology created a behaviorally relevant EtOH concentration in the brain. As chronic EtOH affects GABA_A currents, we examined GABA_A kinetics and input-output relationship. Significant differences were not noted in GABA_A current kinetics between CIE and air exposed mice (data not shown). However, there was a significant difference in input/output curves, which examine the correlation between stimulating electrode intensity and GABA_A current amplitude. As expected, chronic EtOH exposed mice exhibited significantly smaller currents than air controls (Fig. 5E). Regarding plasticity, CIE exposure eliminated iLTD, while iLTP remained unchanged (Fig. 5F). Among the air-control treated animals, both iLTP and iLTD were still observed (Fig. 5G). These findings confirm the influence of ethanol on the mesolimbic circuit to alter GABA cellular plasticity and currents.

Fig. 5. Chronic ethanol treated mice displayed anxiety and compulsive-like behavior during the ethanol withdraw period and eliminated iLTD.

A Schematic illustration of the time line of chronic ethanol exposure. B During the marble burying tests, chronic ethanol treated GAD67/GFP^+/– mice consistently buried more marbles compared to the air-treated GAD67/GFP^+/– controls (Chronic ethanol: n = 11 mice, buried on average 24.23 ± 0.35 marbles; Chronic air: n = 12 mice buried on average 20.25 ± 0.51 marbles; P < 0.0001, Student’s T-Test). Insets show the display of marbles at the start (left), buried by a control mouse (middle) and buried by a chronic ethanol-exposed mouse. C, D During the elevated plus maze (EPM) tests, chronic ethanol treated GAD67/GFP^+/– mice spent significantly more time in the closed arm and significantly less in the open arm than air-treated controls. Time in closed arm (Chronic ethanol, n = 9 mice, average time spent = 205 ± 6.53 s; Chronic air: n = 5 mice, average time spent=104.4 ± 13.25 s; p < 0.0001, Student’s T-Test). Time in open arm (Chronic ethanol, n = 9 mice, average time spent = 89.2 ± 8.09 s; Chronic air: n = 5 mice, average time spent = 208.8 ± 12.4 s; P < 0.0001, Student’s T-Test). E The input-output graph is a measure of recorded IPSCs amplitude based on stimulation electrode intensity to examine differences in GABA_A-evoked currents in chronic EtOH versus chronic air mice. Individual data points from 14 chronic EtOH cells (black squares) and 13 chronic air cells (light gray circles) were graphed as a line using a polynomial fit. These lines were significantly different from each other (P < 0.05; ANCOVA). F The previously observed iLTD was prevented in GAD67/GFP^+/– mice treated with ethanol in the vapor chamber (No plasticity group: p = 0.1792 compared to baseline, ANOVA, n = 7, and p < 0.0001 compared to control iLTD in Fig. 1A, Student’s T-Test; iLTP: p < 0.0001, compared to baseline, ANOVA, n = 6). G Both iLTP and iLTD were observed in GAD67/GFP^+/– mice treated with air in the vapor chamber (iLTP: p < 0.0001 compared to baseline, ANOVA, n = 6; iLTD: p < 0.0001 compared to baseline, ANOVA, n = 6). We also compared iLTP and iLTD from the chronic air exposed animals to that of the control animals in Fig. 1A and found no significant difference (chronic-air iLTP: P = 0.2502 compared to control iLTP in Fig. 1A, Student’s T-Test; chronic-air iLTD: p = 0.8913 compared to control iLTD in Fig. 1A, Student’s T-Test). This suggests iLTP and iLTD occur to a similar degree in both juvenile and young adult mice. ***P < 0.0001.

Discussion

We identified a rare plasticity response at inhibitory synapses onto VTA GABA neurons, where we elicit either iLTP or iLTD in response to the same conditioning stimulus. This report is the first description of iLTD of inhibitory inputs to VTA GABA cells, which could potentially depress DA cell activity, and thus DA release. This provides a more complex view of VTA GABA cell plasticity and the important potential functional role that these cells play in modulating the mesolimbic DA circuit. As chronic EtOH exposure eliminated iLTD, this illustrates another mechanism of EtOH regulation of the reward circuit and could have implications for understanding the physiology of AUD and alcohol withdrawal, perhaps providing new perspectives to guide clinical research.

Long-Term Potentiation

Previously, Bocklisch et al. (2013) studied inhibitory inputs to VTA GABA neurons utilizing an optogenetic approach to target projections from the NAc to the VTA, and they identified LTP at the inhibitory synapse of D1-MSNs to VTA GABA neurons in response to high frequency optical stimulation (trains of 50 Hz). LTP appeared to be presynaptic, was eliminated by chronic cocaine, and was dependent on L-type calcium channels, and the D1 receptor. Here we additionally determined iLTP is also dependent in some cases on NMDAR activation, using an electrical stimulus, which is interesting as Bocklisch et al. used an optical stimulus that only activated GABAergic afferents from the NAc to the VTA. However, NMDAR-dependent modulation of GABAergic plasticity to DA neurons has been previously noted [9] and supports our results. Nugent & Kauer (2007) suggest that this trigger from excitatory synapse activity may induce plasticity at a proximal inhibitory synapse and may serve to balance excitation and inhibition within the mesolimbic DA circuit. Further work is necessary to understand if the iLTP that we observed in response to a 5 Hz electrical stimulus is the same LTP observed by Bocklisch et al. in response to an optical stimulus. Collectively, our data somewhat suggest a presynaptic mechanism of iLTP as PPR decreases and 1/CV² increases significantly, and thus production of a retrograde factor may be involved. However, failure rates do not change after iLTP induction, thus we do not rule out alternatives to this plasticity mechanism. The iLTP also appeared to be partially NMDAR-dependent. However, two forms of iLTP could exist, one of which is NMDAR-dependent and the other NMDA-independent. In addition, NMDA receptors involved could potentially be autoreceptors on GABAergic terminals or on nearby glutamatergic synapses, producing signaling factors. Further work would be required to determine the complete iLTP induction mechanism, in addition to data from this study and others [16]. Lastly, SNAP potentiating synapses in some, but not all cells, supports the notion that distinct inputs innervate unique groups of GABA cells, as NO signaling is likely presynaptic, similar to GABA inputs to DA neurons [9].

Long-term depression and GABA_B receptors

This is the first report of iLTD of GABAergic inputs to VTA GABA neurons, which was determined to be GABA_B-dependent. GABA_B receptor involvement in LTD mediation or modulation in other brain regions was previously noted [48–50]. GABA_B receptors also usually function pre-synaptically as autoreceptors on GABA cells [51] through modulation of voltage-gated calcium channels, and post-synaptically through modulation of G-protein-coupled inwardly rectifying potassium channels (GIRKs; [52, 53]. In the VTA, GABA_B receptors are expressed postsynaptically by DA and GABA neurons [54–56], and presynaptically on GABA inputs [57] and glutamate inputs to GABA cells [8]. Therefore, GABA_B required for iLTD could be located presynaptically and/or postsynaptically. In our data whereas, baclofen reduces IPSC amplitude without altering holding potential and iLTD occurs with increased failures rates, suggesting iLTD could be presynaptic, our 1/CV² and PPR data is inconclusive and thus our data is ambiguous. Further investigation is required to confirm location of synaptic plasticity. As the proportion of cells exhibiting iLTP is not enhanced in the presence of GABA_B antagonist CGP 54626, this further suggests each cell is only capable of only one form of plasticity using our experimental conditions. Finally, the degree of GABA_B-induced depression (see Fig. 4G) could be a method to differentiate preemptively cells that will exhibit iLTP or iLTD, also suggesting the source of GABAergic input determines plasticity type.

Relevant to this study is that GABA_B receptors may be an important target for understanding and treating drug use disorders as well as psychiatric disorders generally [58]. There is a significant body of research demonstrating GABA_B receptor agonists, such as baclofen, may attenuate certain drug-seeking behaviors in animals [47, 59–62]. The present study is the first report of GABA_B receptor involvement in plasticity of inhibitory inputs to GABA neurons, which may help to explain in part the results of previous behavioral studies. Interestingly, GABA_B receptor knockout animals display marked deficits in DA cell LTP and exhibit behavioral deficits [63, 64]. The previous work studying the effects of baclofen on drug-seeking behavior taken together with our current finding of GABA_B receptor-mediated iLTD in the VTA, corroborate the existence of a GABA_B receptor-dependent mechanism, though we are not attempting to show causality between the two in this study.

Lastly, it is worth noting that functional synaptic CB1 receptors did not participate in plasticity or synaptic depression of GABAergic inputs to VTA GABA neurons (see Fig. 4C). This is unique in that CB1 receptors are selectively expressed at GABAergic inputs to VTA DA neurons [65–68], but apparently not to VTA GABA neurons.

Functional rationale for plasticity

The occurrence of iLTP or iLTD at the same synapse is a unique finding here, as seemingly similar cells have such distinct plasticity types even when stimulating with different conditioning parameters (100 Hz or 5 Hz) to induce plasticity or when using multiple stimulations to examine potential qualitative differences in plasticity type. Hypotheses to explain this phenomenon of distinct plasticity outcomes include (1) inhibitory inputs arriving from different brain areas exhibit iLTP or iLTD, (2) GABAergic afferents to VTA GABA cells contain the capacity for iLTP or iLTD depending on context, (3) differential outputs of various GABA cells determine plasticity based on their efferent circuit connections, or (4) distinct GABA cell subtypes within the VTA experience inhibitory plasticity differently. As discussed previously, Bocklisch et al. (2013) described iLTP in their optogenetic model targeting GABAergic inputs from NAc to VTA GABAs, which could represent the iLTP we describe here. As they did not observe LTD, afferents that exhibit iLTD could come from another brain area such as the lateral hypothalamus or ventral pallidum [2]. In addition, GABAergic inputs to VTA DA neurons arriving from distinct brain regions demonstrated different forms of plasticity [69, 70]; supporting the notion that plasticity type can vary depending on the input location. As GABA cells displayed statistically different baselines compared to post baclofen (1 μM) application (Fig. 4H), this also possibly suggests the presence of two distinct cell inputs with different GABA_B receptor levels prior to stimulation. This is a matter of current investigation, including utilization of optogenetic approaches. In addition, baclofen could be used to predetermine the plasticity type a cell would exhibit, along with potentially SNAP.

In support of our second hypothesis, a unique intrasynaptic interplay may exist between NMDA and GABA_B receptors. If GABA and glutamatergic inputs are synapsing in proximity onto VTA GABA neurons, then postsynaptic potentials may be influenced by these different inputs simultaneously. Perhaps if the predominant input comes from stimulated glutamatergic afferents, postsynaptic calcium rises sufficiently via NMDA receptors to elicit a change that increases presynaptic GABA release [9, 71]. Conversely, if most of the stimulation comes from GABAergic afferents, increased GABA release may activate presynaptic GABA_B autoreceptors on the presynaptic GABA terminal as well as heteroreceptors on the presynaptic glutamate terminal to depress glutamate release and consequently decrease NMDA activation [52]. Increased GABAergic stimulation and GABA release may also activate postsynaptic GABA_B receptors coupled to GIRKs, which could hyperpolarize the postsynaptic cell, strengthening the NMDA magnesium blockade and further decreasing the likelihood of NMDA receptors allowing calcium to depolarize the postsynaptic cell [72–74]. Hence, an intrasynaptic “switch” may exist at this and possibly other synapses in the brain that integrates differential signals into a distinct, experience-dependent outcome, though further work is required to solidify this theory.

The output hypothesis is interesting as VTA GABA cells are a heterogeneous population containing both projecting GABA cells and interneurons. Different GABA populations are also thought to potentially be part of unique reward versus aversion circuits [75]. Therefore, different groups of GABA cells could play distinct roles within the GABAergic circuit involved in reward behavior based on their output, and therefore demonstrate different plasticity forms, which we are also currently investigating.

Lastly, the variety of GAD and VGAT expression among “GABA” cells within the VTA should certainly not be overlooked [76]. We previously examined GABA subtype classification using prototypical calcium-binding proteins that differentiate cortical and hippocampal GABA subtypes [32]. However, these did not differentiate into known subtypes, which we hoped could be predictive of plasticity type exhibited. More recently, novel cellular markers classified VTA GABA cells into unique subtypes, though their function is still unknown [77]. A subsequent single nucleus RNA-sequencing study identified three main populations of VTA GABA cells, with five subclasses [78]. Importantly, all three of these major populations contain GAD-67, the marker used for GABA cell identification in this study. In the future, these new criteria could be used to examine whether plasticity types are specific to any cellular subtypes.

There are also a few caveats to consider. First, age is an important variable as developmental changes occur. However, control air exposed iLTP and iLTD in young adult mice were not significantly different from juvenile-aged control iLTP and iLTD (see Fig. 5G), thus iLTP/iLTD are consistent through the ages of mice employed here. Second, antagonists were applied before baseline, and therefore any constitutive receptor activity cannot be determined, though we feel this will not affect the dependence/independence of receptors involved in iLTP/iLTD. Third, iLTP in all cases, including from 5 Hz or 100 Hz stimulation, had a consistent 5–7 min delay before growing greatly. This was extremely consistent in our experiments and appears to follow the trend in Bocklisch et al. (2013), and likely is the nature of this unique iLTP. Fourth, the ratio of iLTP and iLTD is mostly ~50/50. While some variation did occur (i.e. CB1 experiments), we think the ratio difference are due to the random cell selection, as we do not have a marker yet to select by predetermine plasticity type.

Ethanol and GABA_B receptor-dependent iLTD

Ethanol exposure causes short- and long-term changes in the brain, thought to drive the phenotypic and behavioral pathology of reward and dependence [79, 80], and differentially effects VTA DA cell activation based on the cell circuit [81]. Ethanol also has broad effects on glutamatergic synaptic transmission [82, 83], including NMDARs [84], as well as on GABAergic transmission throughout the brain [85] and enhanced GABA transmission to VTA DA neurons [82]. A single exposure of EtOH also reduced, but did not eliminate glutamate LTP onto DA neurons [82]. Within the VTA, EtOH exposure increases DA neuron firing [24, 86] and decreases VTA GABA neuron firing rate in rats in vivo with an IC₅₀ of 1.0 g/kg [28, 87–89]. This is an order of magnitude more sensitive than EtOH effects on DA neurons [90]. In the striatum, ethanol also can directionally alter plasticity [91], and impairs metabotropic mGluR2-dependent LTD plasticity [92]. Lastly, we note that GABA_A receptors have depressed evoked responses in CIE versus air control exposed (see Fig. 5E), however, we did not expect that at higher stimulation intensities this trait would reverse.

In addition, the metabotropic GABA_B receptor is known to be an important autoreceptor in the mesolimbic DA circuit [93]. In support of a role of GABA_B receptors on VTA GABA neurons and glutamate receptor involvement, Steffensen et al. demonstrated previously that EtOH and baclofen inhibit the activity of VTA GABA neurons, and that EtOH inhibition of NMDAR-mediated glutamate receptor activation of VTA GABA neurons involves pre-synaptic GABA_B receptors [94]. In addition, when mice were treated with a single IP exposure to EtOH the GABAergic synapses of VTA DA neurons exhibited long-lasting potentiation that was likely GABA_B-dependent and possibly attributable to GABA spillover [25].

In the present study, the GABA_B-dependent iLTD we identified is eliminated or occluded by chronic EtOH vapor exposure, perhaps suggesting a mechanism like that described by Melis et al. (2002) in which GABA spillover during chronic EtOH exposure activates presynaptic GABA_B receptors to induce plasticity in vivo at the VTA GABA-GABA synapse. Whether iLTD is occluded or blocked, we cannot currently say. Behaviorally, we demonstrated that ethanol-exposed mice exhibit anxiety and dysphoria consistent with a withdrawal syndrome, illustrating EtOH exposure was significant enough to induce neuroplasticity sufficient to cause behavioral changes. Furthermore, the GABA_B receptor agonist baclofen alleviates withdrawal symptoms and reduces voluntary EtOH consumption in ethanol-dependent rats [95–97]. GABA_B receptor activation also has a bigger contributing effect on VTA GABA versus DA neurons [8, 47], thus GABA_B receptor activation effect likely reflect changes in VTA GABA cell activity. Therefore, the ethanol-induced prevention of iLTD may be a mechanism contributing to the impact of chronic alcohol exposure on reward circuitry, indicating a potential correlation between alcohol use disorder and GABA_B-dependent plasticity in the brain. That being said, making a causality connection of iLTD elimination directly to withdrawal would be overstating our results, which we did not investigate. However, it does illustrate the impact of EtOH on VTA GABA cell plasticity, marking this as a potential mechanism for EtOH impact on reward behavior, and an impactful line of future investigation.

Conclusion

We herein provide evidence of two distinct plasticity types (iLTP and iLTD) that can both occur at the VTA GABA-GABA synapse—a unique finding that we believe signals circuit specificity within the mesolimbic DA system with different brain region inputs synapsing onto unique VTA GABA neurons. This is the first report of iLTD at this synapse, which is prevented by chronic EtOH vapor exposure. A greater understanding of these physiological circuits will ultimately aid in the discovery of safe and effective clinical treatments for those struggling with drug use disorders and addiction [98], such as AUD. Ethanol abuse and addiction are serious problems in the United States and around the world [19]; however, current treatments are often only modestly effective and do little to decrease the widespread problem of AUD. A greater understanding of GABAergic plasticity in the inhibitory VTA circuit we hope is a step toward the development of technology and therapeutic interventions for AUD.

Author contributions

TMN, BJW, and JGE designed the research and wrote the paper. TMN and BJW preformed electrophysiology experiments. ZB performed quantitative PCR experiments. SCS provided funding, resources and expertise for the project. All authors performed the studies and analyzed the data, or edited the paper, and approve the final version.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41386-023-01554-y.