Chronic Ethanol Exposure Modulates Periaqueductal Gray to Extended Amygdala Dopamine Circuit

Abstract

The bed nucleus of the stria terminalis (BNST) is a component of the extended amygdala that regulates motivated behavior and affective states and plays an integral role in the development of alcohol-use disorder (AUD). The dorsal subdivision of the BNST (dBNST) receives dense dopaminergic input from the ventrolateral periaqueductal gray (vlPAG)/dorsal raphe (DR). To date, no studies have examined the effects of chronic alcohol on this circuit. Here, we used chronic intermittent ethanol exposure (CIE), a well-established rodent model of AUD, to functionally interrogate the vlPAG/DR-BNST dopamine (DA) circuit during acute withdrawal. We selectively targeted vlPAG/DR^DA neurons in tyrosine hydroxylase-expressing transgenic adult male mice. Using ex vivo electrophysiology, we found hyperexcitability of vlPAG/DR^DA neurons in CIE-treated mice. Further, using optogenetic approaches to target vlPAG/DR^DA terminals in the dBNST, we revealed a CIE-mediated shift in the vlPAG/DR-driven excitatory-inhibitory (E/I) ratio to a hyperexcitable state in dBNST. Additionally, to quantify the effect of CIE on endogenous DA signaling, we coupled optogenetics with fast-scan cyclic voltammetry to measure pathway-specific DA release in dBNST. CIE-treated mice had significantly reduced signal half-life, suggestive of faster clearance of DA signaling. CIE treatment also altered the ratio of vlPAG/DR^DA-driven cellular inhibition and excitation of a subset of dBNST neurons. Overall, our findings suggest a dysregulation of vlPAG/DR to BNST dopamine circuit, which may contribute to pathophysiological phenotypes associated with AUD.

SIGNIFICANCE STATEMENT The dorsal bed nucleus of the stria terminalis (dBNST) is highly implicated in the pathophysiology of alcohol-use disorder and receives dopaminergic inputs from ventrolateral periaqueductal gray/dorsal raphe regions (vlPAG/DR). The present study highlights the plasticity within the vlPAG/DR to dBNST dopamine (DA) circuit during acute withdrawal from chronic ethanol exposure. More specifically, our data reveal that chronic ethanol strengthens vlPAG/DR-dBNST glutamatergic transmission while altering both DA transmission and dopamine-mediated cellular inhibition of dBNST neurons. The net result is a shift toward a hyperexcitable state in dBNST activity. Together, our findings suggest chronic ethanol may promote withdrawal-related plasticity by dysregulating the vlPAG/DR-dBNST DA circuit.

Introduction

Alcohol dependence is a major public health crisis in the United States, contributing to socioeconomic burdens, and is the leading cause of preventable deaths (Murray et al., 2015). Alcohol-use disorder (AUD) is a complex disease characterized by repeated cycles of drug intake, withdrawal, and drug-seeking behaviors regulated by multiple neural circuits involved in reward and negative affect (Koob and Volkow, 2010). An increasing body of literature from human brain imaging (Volkow et al., 2009) and preclinical studies has implicated the mesocorticolimbic dopaminergic system in the etiology of AUD (Nestler, 2001; Volkow et al., 2009). For example, acute ethanol administration increases extracellular dopamine (DA) levels in the nucleus accumbens (NAc) of freely moving rats (Di Chiara and Imperato, 1988). Conversely, in vivo activity of ventral tegmental area (VTA) DA neurons is significantly decreased during withdrawal, and this hypodopaminergic state is hypothesized to drive ethanol seeking (Diana et al., 1993; Weiss et al., 1996).

A growing body of literature suggests that the extra-VTA DA population in the ventrolateral columns of periaqueductal gray (vlPAG) and dorsal raphe (DR) show similar phasic activation patterns in response to salient stimuli as VTA DA neurons to encode a diverse set of rewarding and aversive behaviors (Cho et al., 2017; Lin et al., 2020). Recent work from our lab and others has implicated these neurons in the regulation of wakefulness/arousal (Cho et al., 2017), pain processing (Li et al., 2016; Yu et al., 2021), social interaction (Matthews et al., 2016), and opiate reward (Flores et al., 2004; Lin et al., 2020). Given a high degree of overlap between alcohol withdrawal, dysregulation of sleep (Brower and Perron, 2010), increased pain sensitivity (Gatch and Lal, 1999; Jochum et al., 2010; Avegno et al., 2018), and anxiety-like behaviors (Kliethermes, 2005), examining the impact of alcohol on vlPAG/DR DA system could provide a common neuronal basis for these comorbidities. Notably, these neurons project to brain regions integral to the regulation of alcohol-related behaviors, including the central amygdala and the dorsal subdivision of the bed nucleus of the stria terminalis (dBNST; Freedman and Cassell, 1994) to co-release DA and glutamate (Li et al., 2016; Yu et al., 2021).

The vast majority of neurons in the dBNST are GABAergic in phenotype (Sun and Cassell, 1993) and form both intrinsic connections within the BNST (Sun and Cassell, 1993) as well as send projections to areas involved in the reward circuit, such as the VTA and the lateral hypothalamus (Dong et al., 2001; Dong and Swanson, 2006). Similar to the NAc, ethanol dose dependently increases extracellular BNST DA concentrations (Carboni et al., 2000). Pharmacological blockade of D1-like receptors in the BNST reduces ethanol and sucrose self-administration in male and female alcohol-preferring P rats (Eiler et al., 2003). At a synaptic level, bath application of dopamine increases excitatory activity (Kash et al., 2008) and induces long-term plasticity of inhibitory inputs in a dopamine-receptor-subtype-dependent manner (Krawczyk et al., 2011, 2013). Thus, given the well-established role of the BNST in neuroadaptations related to ethanol dependence and withdrawal (Lovinger and Kash, 2015; Vranjkovic et al., 2017), it is reasonable to postulate that chronic ethanol could modulate the vlPAG/DR to dBNST dopamine circuit. In the present study, we used a combination of technical approaches, including slice physiology, channelrhodopsin-assisted circuit mapping, and pathway-specific fast-scan voltammetry to explore the role of vlPAG/DR^DA neurons in driving plasticity of the dBNST during withdrawal from chronic intermittent ethanol exposure (CIE). We discovered distinct neuroadaptations in the vlPAG/DR to dBNST dopamine circuit, resulting in an overall shift toward a hyperexcitable state in dBNST of CIE-treated male mice. These data collectively establish the role of vlPAG/DR to dBNST dopamine circuit in the pathophysiology of AUD.

Materials and Methods

Mice

All experiments were performed on adult male mice, aged nine to 16 weeks at the start of the procedures, in accordance with the NIH guidelines for animal research and with the approval of the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill. Mice were group housed and maintained on a standard 12-h cycle (lights on at 7 A.M.) with ad libitum access to food (5V5R-PicoLab Select Rodent 50 IF/6F) and water. Tyrosine hydroxylase expressing-cre (TH-cre) mice were generated as previously described (Gerfen et al., 2013; Stock from Mutant Mouse Regional Resource Centers; catalog #029177-UCD; RRID: MMRRC_029177-UCD), were bred in-house and maintained on a C57BL6/J background.

Surgical procedure

Adult male mice were anesthetized with isoflurane (2–3%) in oxygen (0.8–1 l/min) and then secured on a stereotaxic frame (Model 1900, Kopf Instruments) for intracranial viral infusions. To minimize postoperative discomfort, meloxicam (5 mg/kg) was administered at the time of the surgery and for two additional days. The vlPAG/DR (from bregma: AP −4.60 mm, ML 0.00 mm, DV −3.2 mm) was targeted using standard coordinates from Paxinos and Franklin. Microinjections were performed with a 1-μl Neuros Hamilton syringe (Hamilton) controlled by a microinfusion pump (WPI) at a rate of 100 nl/min. For channelrhodopsin-2-assisted circuit mapping, 500 nl of AAV5-EF1α-DIO-hChR2(H134R)-eYFP (Addgene; catalog # 20298; ≥1 × 10¹³ vg/ml) was administered unilaterally at an angle of 20°. After infusion, injectors were left in place for 8–10 min to allow for viral diffusion. All surgeries were conducted using aseptic techniques in a sterile environment. Mice were closely monitored and allowed to recover before starting experiments.

Chronic intermittent ethanol (CIE) exposure

Four to five weeks after surgery, age-matched mice were randomly assigned to experimental groups. Mice were housed in pairs to reduce the stress of CIE. Chronic ethanol exposure was achieved via vapor inhalation as previously described (Becker and Lopez, 2004). Briefly, mice were placed in vapor chambers and exposed to volatized ethanol (95%) mixed with fresh air to deliver at a rate of ∼10 l/min. Mice in the ethanol group received intraperitoneal injections of an alcohol dehydrogenase inhibitor, pyrazole (1 mmol/kg) combined with 1.5 g/kg ethanol to induce intoxication and stabilize blood ethanol concentrations before placement in the chambers. These conditions yield stable blood ethanol levels in the range of 150–200 mg/dl. Air controls received only pyrazole and were placed in dedicated chambers (located adjacent to the ethanol chambers) in which air was exchanged at a rate of ∼10 l/min. Each cycle of CIE exposure lasted 16 h per day (in at 5 P.M., out at 9 A.M.), followed by an 8-h withdrawal for four consecutive days of exposure (Monday–Friday). This was followed by a more extended 80-h withdrawal (Friday–Monday). This procedure was repeated for a total of four cycles.

Slice electrophysiology

For whole-cell recordings, mice were anesthetized with isoflurane, 24 h after the last CIE cycle, and rapidly decapitated. Coronal sections through the PAG (250 μm) and the BNST (300 μm) were prepared as previously described (Pati et al., 2019). Briefly, brains were quickly extracted, and slices were made using a Leica VT 1200s vibratome (Leica Biosystems) in ice-cold, oxygenated sucrose solution containing in mm: 183 sucrose, 20 NaCl, 0.5 KCl, 2.5 MgCl₂, 1.2 NaH₂PO₄, 10 glucose, and 26 NaHCO₃ saturated with 95% O₂/5% CO₂. Slices were incubated for at least 30 min in artificial CSF (ACSF) maintained at 35°C that contained in mm: 124 NaCl, 4.0 KCl, 1 NaH₂PO₄, 1.2 MgSO₄, 10 D-glucose, 2 CaCl₂, and 26 NaHCO₃, saturated with 95% O₂/5% CO₂ before transferring to a submerged recording chamber (Warner Instruments) for experimental use. For whole-cell recordings, slices were continuously perfused at a rate of 2.0–3.0 ml/min with oxygenated ACSF maintained at 30 ± 2°C.

Neurons were identified using infrared differential interference contrast on a Scientifica Slicescope II. Fluorescent cells were visualized using a 470-nm LED. Whole-cell patch clamp recordings were performed using micropipettes pulled from a borosilicate glass capillary tube using a Flaming/Brown electrode puller (Sutter P-97; Sutter Instruments). Electrode tip resistance was between 3 and 6 MΩ. All signals were acquired using an Axon Multiclamp 700B (Molecular Devices). Data were sampled at 10 kHz and low pass filtered at 3 kHz. Access resistance was continuously monitored, and changes >20% from the initial value were excluded from data analyses.

Excitability experiments were performed in current-clamp mode using a potassium gluconate-based intracellular solution (in mm: 135 K-gluconate, 5 NaCl, 2 MgCl₂, 10 HEPES, 0.6 EGTA, 4 Na₂ATP, and 0.4 Na₂GTP, pH 7.3, 285–295 mOsm). Membrane resistance was measured immediately after breaking into the cell. Following stabilization, current was injected to hold cells at a common membrane potential of −70 mV to account for intercell variability. Changes in excitability were evaluated by measuring rheobase [minimum current required to elicit an action potential (AP)], AP threshold, and the frequency of action potentials fired at increasing 20-pA current steps (0–160 pA).

For the assessment of spontaneous synaptic activity, two different intracellular solutions were used. Spontaneous excitatory postsynaptic currents (sEPSCs) were assessed in voltage clamp at −80 mV using the same potassium gluconate-based internal as described above. Spontaneous inhibitory postsynaptic currents (sIPSCs) were pharmacologically isolated by adding kynurenic acid (3 mm) to the ACSF to block AMPA and NMDA receptor-dependent postsynaptic currents. Cells were clamped at −70 mV and recorded using a potassium chloride-based intracellular solution (in mm: 80 KCl, 70 K-gluconate, 10 HEPES, 1 EGTA, 4 Na₂ATP, and 0.4 Na₂GTP, pH 7.2, 285–290 mOsm with 1 mg/ml QX-314-bromide).

For optogenetic experiments, channelrhodopsin-2 (ChR2)-expressing neurons in the vlPAG/PAG were evaluated for optically evoked action potentials using stimulation parameters of 10, 5-ms blue light (470 nm) pulses at 1, 5, and 20 Hz delivered through a 40× 0.8 NA immersion objective (Olympus). Identical optical stimulation parameters (1–5 ms in duration with a power of 2.5 mW) were used to compare optically evoked currents in the dBNST between the two groups. The response threshold was set at 10 pA. Optically evoked currents in the dBNST were measured in voltage clamp using a cesium methanesulfonate-based intracellular solution (Cs-Meth; in mm: 135 cesium methanesulfonate, 10 KCl, 1 MgCl2, 0.2 EGTA, 4 MgATP, 0.3 Na2GTP, and 20 phosphocreatine, pH 7.3, 285–290 mOsm with 1 mg/ml QX-314) to detect light-evoked EPSCs (−55 mV) and IPSCs (+10 mV) within the same neuron. Light-evoked AMPA and NMDA currents were recorded in the presence of picrotoxin (25 μm) using a cesium-gluconate based intracellular solution (in mm: 117 D-gluconic acid, 20 HEPES, 0.4 EGTA, 5 TEA, 2 MgCl₂, 4 Na₂ ATP, and 0.4 Na₂ GTP, pH 7.3, 285–290 mOsm with 1 mg/ml QX-314). Cells were held at −70 mV to record AMPA receptor-mediated currents and at +40 mV to record a mixture of AMPA and NMDA receptor-mediated currents. The NMDA current at +40 mV was measured at least 50 ms after the onset of an evoked response to avoid contamination with AMPA current. To assess the putative changes in presynaptic glutamate release probability, fibers expressing ChR2 in the dBNST were stimulated using two 5-ms blue light pulses with an interstimulus interval of 50 ms at −70 mV. Changes in membrane potential following optical stimulation trains (20 pulses at 20 Hz) were recorded in current clamp mode in the presence of kynurenic acid (3 mm) and picrotoxin (25 μm) to block ionotropic glutamate and GABA receptors, respectively. Cells were held at their resting membrane potential (0-pA current) except in spontaneously active cells (n = 6 in air and n = 1 in CIE group), where current was injected to hold the cells close to −70 mV.

FSCV

Optically stimulated dopamine release was detected using ex vivo fast-scan cyclic voltammetry as previously described (Li et al., 2016). Briefly, a carbon-fiber microelectrode (CFME; made in-house) was positioned in the dorsal BNST, and a potential of −0.4 V (vs Ag/AgCl) was applied, then rapidly ramped up to 1.3 V (at 400 V/s) at a rate of 10 Hz (Tarheel CV, Labview; National Instruments). Fluorescently labeled terminals expressing ChR2 in the BNST were then stimulated with light pulses from a 437-nm LED. Stimulation protocols varied by frequency (2, 5, 10, and 20 Hz with 20 pulses with 5-ms pulse width). Background subtracted cyclic voltammograms (CVs) were then analyzed using HDCV (UNC-Chapel Hill). In a subset of these experiments, 2 μm sulpiride was bath applied for 20 min to assess the effects of Gi/o-coupled dopamine D2 receptor (D2R) antagonism on DA release.

Drugs

All chemicals used for slice electrophysiology and FSCV were obtained from either Tocris Bioscience, Abcam, or Hello Bio. Pyrazole was purchased from Sigma-Aldrich. Cre-dependent channelrhodopsin virus was purchased from Addgene.

Data and statistical analysis

Differences in various electrophysiological measures were analyzed using either Clampfit 10.7 (Molecular Devices) or Easy Electrophysiology (RRID:SCR_021190, v2.3.2-β; Garcia et al., 2014) and compared between the air and CIE groups. Group comparisons were made using either a standard unpaired Student's t test or two/three-way ANOVA depending on the number of independent and within-subjects variables in a dataset. Following significant interactions or main effects, post hoc pairwise t tests were performed and corrected using Šídák's post hoc tests to control for multiple comparisons. Fisher's exact test was used to compare proportions. No statistical method was used to predetermine sample size, and no blinding procedures were used. Two to four cells were recorded from each animal per set of experiments. Grubbs' test was used for identifying outliers. All data are expressed as mean ± SEM. p-values ≤ 0.05 were considered significant. Statistical analysis was performed using GraphPad Prism v.9.

Results

CIE withdrawal increases the neuronal excitability of vlPAG/DR^DA neurons

To explore how alcohol withdrawal modulates the neuronal activity of vlPAG/DR^DA neurons, we selectively expressed a CHR2-encoding virus in TH-cre male mice before four cycles of CIE. Following 24-h withdrawal from the last exposure, we assessed CIE-induced changes in neuronal function (Fig. 1A). We recorded from ChR2-expressing neurons in the vlPAG/DR and measured optically evoked action potentials across different frequencies (Fig. 1B). Intrinsic properties of vlPAG/DR^DA neurons were calculated from data generated in voltage clamp using steps from −70 to −80 mV. Data were collected from 20 to 24 cells from nine mice per group. While there was no difference in membrane capacitance between the CIE and air groups (Fig. 1C; t₍₄₂₎ = 0.66; p = 0.515; unpaired Student's t test), we found a trend toward an increase in membrane resistance in the CIE group (Fig. 1D; t₍₄₂₎ = 1.91; p = 0.063; unpaired t test). When we examined the intrinsic excitability of vlPAG/DR^DA neurons, we found that withdrawal from CIE increased the excitability of these neurons (Fig. 1E–I). The action potential threshold and the amount of current required to fire an action potential (rheobase) were assessed through a ramp protocol of 120 pA/1 s. There was a significant effect of CIE on rheobase (Fig. 1E,F; t₍₄₃₎ = 2.08; p = 0.044; unpaired Student's t test) but not on action potential threshold (Fig. 1G; t₍₄₃₎ = 0.59; p = 0.556; unpaired Student's t test). This was also supported by a significant increase in firing frequency to increasing current steps (0–160 pA for 500 ms, at an increment of 20 pA; Fig. 1H,I) in CIE-treated mice (treatment × current interaction: F_(8,344) = 2.35; p = 0.018; current main effect: F_(8,344) = 55.36; p < 0.0001; and a nonsignificant trend for treatment main effect: F_(1,43) = 3.29; p = 0.077, as revealed by repeated-measures two-way ANOVA).

Figure 1.

CIE withdrawal increases intrinsic excitability of vlPAG/DR^DA neurons. A, Schematics for whole-cell recordings from ChR2-expressing neurons in vlPAG/DR. After recovery from AAV injections, male Th-cre mice underwent four CIE cycles. Twenty-four hours after the final CIE exposure, whole-cell recordings were made from vlPAG/DR^DA neurons. B, Representative traces of ChR2+ neurons firing action potentials in response to optogenetic cell body stimulation (blue bars) at 1 and 20 Hz. C, D, There was no change in cell membrane capacitance between the two groups but a trend toward an increase in input resistance in the CIE-exposed mice. E, Representative data from vlPAG/DR^DA neurons in response to a 120-pA/s current ramp while injecting a constant current to hold the cells at −70 mV. The minimum current required to fire an action potential (rheobase) was lowered following CIE withdrawal (F) without any changes (G) in AP threshold. H, Representative traces of action potentials fired in response to a step protocol of increased current steps of 20 pA/500 ms while holding the cells at −70 mV. I, Withdrawal from CIE significantly increased the frequency of APs in response to a graded current injection. J, Representative traces of steady state current in response to graded changes in holding voltage from −70 to −120 mV. CIE significantly reduced current density at more hyperpolarized holding potentials (L) without any changes in the holding current at −70 mV (K). Data expressed as mean ± SEM; # represents significant interaction between groups. *p < 0.05.

An extensive body of literature suggests that ethanol can differentially regulate excitability by acting on a multitude of ion channels. For example, repeated ethanol administration significantly reduced small conductance calcium-dependent potassium channel function in VTA and contributed to increased in vivo DA burst firing (Hopf et al., 2007). We evaluated the relationship between current density and hyperpolarizing command potentials to identify possible ion channel alterations driving increased excitability of vlPAG/DR^DA neurons post-CIE. We altered the command potential between −70 mV to −120 mV with 10 mV incremental voltage steps to evaluate the relationship between steady-state current and voltage potential (Fig. 1J–L; n = 21–26 cells from 9 mice/group). The holding current at −70 mV did not differ between the two groups (Fig. 1K; t₍₄₅₎ = 0.37; p = 0.716; unpaired Student's t test). Withdrawal from CIE did result in a reduced current density as revealed by a two-way repeated-measures ANOVA with Geisser–Greenhouse correction with significant treatment × voltage interaction (F_(5,225) = 4.71; p = 0.0004), and a main effect of both voltage (F_{(1.08, 48.59)} = 274.8; p < 0.0001) and treatment (F_(1,45) = 4.66; p = 0.036). Together, these results suggest that withdrawal from CIE results in hyperexcitability of vlPAG/DR^DA neurons.

Spontaneous synaptic drive on vlPAG/DR^DA neurons is unaffected by withdrawal from CIE

Next, we evaluated the impact of CIE on spontaneous synaptic transmission on vlPAG/DR^DA neurons (Fig. 2). Spontaneous excitatory transmission was measured at −80 mV from ChR2-expressing neurons in TH-cre mice. We found no effect of CIE on sEPSC parameters (Fig. 2A–C; n = 20–25 cells from 9 mice/group; sEPSC frequency: t₍₄₃₎ = 0.14; p = 0.889; sEPSC amplitude: t₍₄₃₎ = 1.13; p = 0.267; unpaired Student's t test).

Figure 2.

CIE withdrawal does not affect spontaneous synaptic transmission in vlPAG/DR^DA neurons. A, Representative traces of sEPSCs from ChR2-expressing neurons vlPAG/DR in TH-cre male mice from air-exposed and CIE-exposed groups. No changes were observed in either mean sEPSC frequency (B) or amplitude (C). D, Representative traces of sIPSCs from ChR2-expressing neurons vlPAG/DR in TH-cre male mice. Similar to sEPSCs, CIE did not alter average sIPSC frequency (E) or amplitude (F). Data expressed as mean ± SEM.

To assess GABAergic transmission on vlPAG/DR^DA neurons, we recorded from ChR2-expressing neurons in TH-cre male mice in the presence of kynurenic acid (3 mm). Similar to sEPSCs, we found no significant effect of CIE on sIPSC parameters (Fig. 2D–F; n = 13–16 cells from 4–5 mice/group). There was no effect on either sIPSC frequency (Fig. 2E; t₍₂₇₎ = 0.23; p = 0.818; unpaired Student's t test) or amplitude in CIE-treated mice (Fig. 2F; t_(23.88) = 1.31; p = 0.204; unpaired Student's t test with Welch's correction). These findings suggest that withdrawal from CIE does not impact spontaneous neurotransmission on vlPAG/DR^DA neurons.

CIE withdrawal disrupts the excitatory to inhibitory balance in the vlPAG/DR^DA-dBNST pathway

Previously, we have functionally characterized dopaminergic inputs from vlPAG/DR to dBNST (Li et al., 2016; Yu et al., 2021) and showed stimulation of these terminals in the BNST co-released glutamate resulting in both optically evoked EPSCs (oEPSCs) and optically evoked IPSCs (oIPSCs). Here, we examined the role of vlPAG/DR^DA-dBNST pathway in driving plasticity during withdrawal from CIE. Overall, there was no difference in the percentage of total light-responsive BNST neurons between the two groups (Fig. 3B; n = 28–36 cells from eight to 10 mice per condition; p = 0.721 for total responsivity; p = 0.508 for oEPSC vs oIPSC responsivity; Fisher's exact test). Similarly, the onset latency of optically evoked responses did not differ across groups. The average onset duration of oEPSCs was <3 ms for both groups (Air: 2.67 ± 0.11 ms; CIE: 2.64 ± 0.07 ms; t₍₄₈₎ = 0.19; p = 0.853; unpaired Student's t test). As expected, compared with oEPSCs, oIPSCs had a slower onset duration that was not altered by CIE (Air: 8.75 ± 0.77 ms; CIE: 8.22 ± 0.37 ms; t_(17.51) = 0.62; p = 0.543; unpaired Student's t test with Welch's correction). While withdrawal from CIE did not affect the amplitudes of optically evoked excitatory inputs (Fig. 3D; n = 23–27 cells from eight to 10 mice per condition; t₍₄₈₎ = 0.18; p = 0.858; unpaired Student's t test), there was a trend toward a reduction in the amplitudes of optically evoked inhibitory neurotransmission (Fig. 3E; n = 13–21 cells from eight to 10 mice per condition; t_(17.61) = 1.80; p = 0.088; unpaired Student's t test with Welch's correction). Next, we compared the amplitudes of excitatory and inhibitory neurotransmission (E/I ratio) obtained from the same neuron. CIE mice had a significantly greater E/I ratio compared with air mice (Fig. 3F; n = 13–18 cells from eight to 10 mice per condition; t_(21.18) = 2.26; p = 0.035; unpaired Student's t test with Welch's correction). To further examine CIE effects on nonspecific synaptic activity, we recorded spontaneous EPSCs and IPSCs from light-responsive BNST neurons that also received both excitatory and inhibitory inputs from vlPAG/DR^DA (Fig. 3G–L; n = 9–13 cells from 6 mice/group). CIE had no effect on sEPSC parameters (Fig. 3H,I; sEPSC frequency: t₍₂₀₎ = 0.66; p = 0.517; sEPSC amplitude: t₍₂₀₎ = 0.52; p = 0.610; unpaired Student's t test) or sIPSC parameters (Fig. 3J,K; sIPSC frequency: t_(9.46) = 1.5; p = 0.166; sIPSC amplitude: t₍₂₀₎ = 0.16; p = 0.874; unpaired Student's t test). Similar to the optically evoked increase in E/I ratio, withdrawal from CIE also resulted in a significantly greater E/I ratio of spontaneous synaptic frequency compared with air mice (Fig. 3L; t₍₂₀₎ = 2.29; p = 0.033; unpaired Student's t test).

Figure 3.

CIE shifts the excitatory-inhibitory balance in the vlPAG/DR^DA-dBNST pathway. A, Schematics for whole-cell recordings from dBNST neurons 24 h post-CIE. ChR2-assisted circuit mapping was used to record cell populations in the dBNST innervated by vlPAG/DR inputs. B, There was no change in the percentage of total light-responsive dBNST neurons between the two groups. C, Representative traces of monosynaptic oEPSC amplitudes (shorter latency; gray traces) and polysynaptic oIPSC amplitudes (longer latency; red traces) from the same neuron held at −55 and +10 mV, respectively, in response to 1- to 5-ms blue light stimulation. While there was no difference in oEPSC amplitudes between air and CIE mice (D), there was a nonsignificant trend toward a reduction in oIPSC amplitudes in CIE-exposed mice (E). F, Comparison of E/I ratio between the two groups, revealed a significant increase in E/I ratio in CIE mice, suggesting a shift toward a hyperexcitable state in the vlPAG/DR^DA-dBNST pathway. G, Representative traces of sEPSCs (black/gray traces; at −55 mV) and sIPSCs (red traces; at +10 mV) recorded from the neurons that were also innervated by vlPAG/DR inputs. There was no difference between groups in either spontaneous excitatory transmission [sEPSC frequency (H) and amplitude (I)] or spontaneous inhibitory transmission [sIPSC frequency (J) and amplitude (K)]. However, CIE-exposed mice had a significantly greater SE/I frequency ratio (L). Data expressed as mean ± SEM; *p < 0.05.

Taken together, this suggests that CIE disrupts the ratio between vlPAG/DR-driven excitatory and inhibitory inputs to drive a hyperexcitable state in dBNST neurons while also shifting the spontaneous activity at a network level toward increased excitability.

Withdrawal from CIE drives glutamatergic plasticity in the vlPAG/DR^DA-dBNST pathway

To further characterize CIE-mediated plasticity of vlPAG/DR^DA-dBNST glutamatergic neurotransmission, we isolated glutamatergic transmission by blocking GABA-A receptors. Postsynaptic glutamate plasticity plays a key role in the pathophysiology of addiction (Lüscher and Malenka, 2011) and is commonly associated with a change in the ratio between AMPAR-mediated and NMDAR-mediated currents. Therefore, we measured the AMPA-NMDA ratio from the same neuron by holding the cell at −70 mV (oAMPA) and at +40 mV (oNMDA). We found that withdrawal from CIE significantly increased both AMPAR-mediated oEPSC amplitudes (Fig. 4B; n = 15 cells from 7–9 mice/group; t_(15.80) = 2.27; p = 0.038; unpaired Student's t test with Welch's correction) and NMDAR-mediated oEPSCs (Fig. 4C; n = 13–14 cells from seven to nine mice per group; t_(12.98) = 2.23; p = 0.044; unpaired Student's t test with Welch's correction). However, we found no difference in AMPA-NMDA ratio between air and CIE-treated mice (Fig. 4D; n = 13–14 cells from seven to nine mice per group; t_(14.19) = 1.34; p = 0.200; unpaired Student's t test with Welch's correction).

Figure 4.

CIE induces glutamatergic plasticity in the vlPAG/DR^DA-dBNST pathway. A, Representative traces of oEPSCs recorded in the presence of picrotoxin at −70 mV (gray traces) to isolate AMPA-mediated EPSCs and at +40 mV to obtain NMDA-mediated amplitude (red traces) within the same dBNST neurons in response to 1- to 5-ms blue light stimulation. B, AMPA-mediated oEPSCs are increased in dBNST neurons of CIE mice. C, Compared with air mice, CIE-treated mice also have higher NMDA-mediated amplitudes. D, No change was observed in the AMPA/NMDA ratio. E, Representative traces of oEPSCs evoked at 50-ms interstimulus intervals in response to 1–5 ms of 473-nm LED (blue bar). Responses normalized to the first peak: air (gray traces) and CIE (red traces). F, Withdrawal from CIE reduced the paired-pulse depression of oEPSCs observed in air mice without a change in the decay kinetics of the first peak (G). H, There was a trend but no significant correlation between the decay kinetics of the first peak and the magnitude of PPR. Data expressed as mean ± SEM; *p < 0.05.

Next, we measured paired-pulse ratio (PPR) from two peaks evoked 50 ms apart as a surrogate for measuring short-term plasticity at vlPAG/DR^DA-dBNST synapses. In air-treated mice, the synaptic responses following the second stimuli were attenuated resulting in paired-pulse depression. Interestingly, we observed a significant reduction in paired-pulse depression in the CIE-treated mice (Fig. 4E,F; n = 12 cells from 7–8 mice/group; t₍₂₂₎ = 2.12; p = 0.046; unpaired Student's t test). The decay kinetics of the first peak in the CIE group generally tended to be faster, though there was no significant difference between the two groups (Fig. 4G; t₍₂₂₎ = 0.80; p = 0.434; unpaired Student's t test). We also found a trend toward a correlation between decay kinetics and PPR (Fig. 4G; r = −0.31; p = 0.146; Spearman correlation). These findings suggest that CIE increases the synaptic strength of vlPAG/DR^DA-dBNST synapses through modulation of AMPAR-mediated currents, although the contribution of presynaptic and postsynaptic loci in driving this plasticity is unclear.

CIE withdrawal modulates vlPAG/DR^DA-dBNST dopaminergic transmission

Prior studies have found that ethanol can dose dependently increase extracellular BNST dopamine concentrations (Carboni et al., 2000) and bath-applied dopamine can modulate both short-term and long-term synaptic activity in the dBNST (Kash et al., 2008; Krawczyk et al., 2011, 2013). Previously, we directly measured endogenous DA from vlPAG/DR^DA terminals using ex vivo fast-scan voltammetry combined with optogenetics in naive male mice (Li et al., 2016). Here, we investigated how CIE alters different aspects of dopamine signaling within the vlPAG/DR^DA-dBNST circuit. Cyclic voltammetry recordings were performed in the dBNST, and optical stimulation trains were delivered as 20 pulses of different frequencies (Fig. 5; five mice per group). Stimulation trains at frequencies ranging from 2 to 20 Hz (Fig. 5C,D) resulted in increases in DA release that was further augmented in the presence of D2 autoreceptor antagonist sulpiride (2 μm; drug × frequency interaction: F_(3,48) = 2.81; p = 0.049; frequency main effect: F_(1.49,23.88) = 70.76; p < 0.0001; drug main effect: F_(1,16) = 11.46; p = 0.004, as revealed by repeated-measures three-way ANOVA with Geisser–Greenhouse correction). Comparisons by treatment, however, revealed no effect of CIE on DA release or D2R regulation of release (treatment main effect: F_(1,16) = 0.04; p = 0.853; treatment × frequency interaction: F_(3,48) = 0.006; p = 0.999; drug × treatment interaction: F_(1,16) = 0.15; p = 0.707; frequency × drug × treatment interaction: F_(3,48) = 0.09; p = 0.966; as revealed by repeated-measures three-way ANOVA with Geisser–Greenhouse correction). The average signal half-life for dopamine release (20 pulses at 20 Hz) in the CIE-treated mice was significantly lower than in air-treated mice (Fig. 5E; t₍₈₎ = 2.52; p = 0.036; unpaired Student's t test). This suggests that CIE results in faster uptake of DA. Interestingly, the percentage change in dopamine half-life in the presence of sulpiride (2 μm) was significantly greater in mice exposed to CIE than their counterparts (Fig. 5F; t₍₈₎ = 2.79; p = 0.024; unpaired Student's t test). Our findings show that chronic ethanol exposure alters the duration of DA signaling by increasing the uptake rate putatively through D2 autoreceptors.

Figure 5.

CIE alters optically stimulated dopamine signals in the vlPAG/DR^DA-dBNST pathway. A, Schematics for recordings from dBNST neurons 24 h post-CIE. ChR2-assisted FSCV was used to record dopamine signals in the vlPAG/DR^DA-dBNST pathway. B, Representative cyclic voltammetry traces of dopamine signals elicited with a 20 pulse, 20 Hz optical stimulation train (air traces in gray and CIE traces in orange). Color plot represents current (z-axis) as a measure of voltage (y-axis) and time (x-axis). C, D, In both air-treated and CIE-treated mice the magnitude of optically stimulated dopamine release increased with increasing frequency of stimulation of ChR2-expressing terminals ranging from 2 to 20 Hz (air: gray dots and CIE: orange dots). Application of D2 autoreceptor antagonist sulpiride (2 μm; white dots) increased release in a frequency-dependent manner. Compared with the air mice, CIE did not alter optically stimulated release of dopamine both at baseline and in the presence of sulpiride. E, Grouped data showing dopamine signal half-life (s) measured from 20 pulses 20-Hz optical stimulation train. Following CIE exposure, the signal half-life was significantly reduced in dBNST compared with air-exposed animals, suggesting faster uptake in CIE mice. F, In the presence of sulpiride, withdrawal from CIE resulted in a significantly greater percentage change in dopamine signal half-life CIE mice compared with air animals. Data expressed as mean ± SEM; *p < 0.05.

Next, we examined how CIE alters cellular responses to optically stimulated DA release in the dBNST. Previous studies characterizing DA-mediated synaptic changes in dBNST focused on fast synaptic transmission following bath application of DA (Kash et al., 2008; Francesconi et al., 2009; Krawczyk et al., 2011). These studies fail to consider spatiotemporal aspects of phasic DA signaling. To address these limitations, we measured changes in the membrane potential of dBNST in response to optical stimulation trains (20 pulses at 20 Hz) in the presence of kynurenic acid (3 mm) and picrotoxin (25 μm) to block ionotropic glutamate and GABA receptors, respectively. We found that in dBNST neurons, optical stimulation of vlPAG/DR^DA terminals bidirectionally altered the resting membrane potential to either drive depolarization or hyperpolarization (Fig. 6A–D). Specifically, only 17.1% of cells recorded from CIE mice showed a clear hyperpolarization compared with 52.8% of cells in air-exposed mice (Fig. 6B; n = 35–36 cells from 10–12 mice/group; p = 0.0026; Fisher's exact test). The average peak depolarization amplitude did not vary between the two groups (Fig. 6C; n = 17–29 cells; t_(23.43) = 1.00; p = 0.326; unpaired Student's t test with Welch's correction). Interestingly, while CIE reduced the percentage of hyperpolarized cells, there was a nonsignificant trend toward increased magnitude of hyperpolarization compared with air mice (Fig. 6D; n = 6–19 cells; t_(5.59) = 2.29; p = 0.065; unpaired Student's t test with Welch's correction). In a small subset of neurons, we recorded optically stimulated responses in the presence of tetrodotoxin (500 nm) and 4-AP (200 μm) to isolate monosynaptic responses (data not shown). Both depolarization (n = 2; 1.1 ± 0.29 mV) and hyperpolarization (n = 3; −2.65 ± 0.79 mV) persisted in the presence of the blockers suggesting the monosynaptic nature of the bidirectional changes in membrane potential.

Figure 6.

CIE disrupts the optically evoked depolarization to hyperpolarization balance in the vlPAG/DR^DA-dBNST pathway. A, Schematic illustrating whole-cell patch clamp recordings in the presence of 3 mm kynurenic acid + 25 μm picrotoxin. Representative traces of depolarization (top; air traces in gray and CIE traces in orange) and hyperpolarization in response to optical stimulation train (20 pulses, 20 Hz; blue bar). B, CIE significantly altered the cellular responses of dBNST neurons in response to phasic stimulation of ChR2-expressing terminals. Compared with air mice, optical stimulation at 20 Hz resulted in a lower percentage of hyperpolarization in CIE-treated animals. C, D, Grouped data showing the peak amplitude of voltage responses from dBNST neurons. There were no statistical differences between groups in average amplitude of depolarization or hyperpolarization, though a trend toward increased amplitude in CIE mice. Intrinsic properties of cells that depolarized varied from cells that hyperpolarized. Specifically, neurons that depolarized had a higher membrane capacitance (E) and were more hyperpolarized at rest (G) compared with neurons that hyperpolarized. CIE did not affect membrane capacitance but altered the resting membrane potential. The hyperpolarized neurons in CIE mice rested at a more positive membrane potential than the air. F, Membrane resistance was not altered by either cellular response or CIE exposure. H, Correlation analysis showed a negative relationship between resting membrane potential and cellular response to optical stimulation. I, Representative traces of hyperpolarization in response to optical stimulation train (20 pulses, 20 Hz; blue bar) at baseline (black trace) and following bath application of sulpiride (2 μm; dark red trace). J, D2 receptor antagonism blocked hyperpolarization. K–M, Representative traces of depolarization in response to optical stimulation train (20 pulses, 20 Hz; blue bar) at baseline (black trace) and in the presence of either SCH-23390 (2 μm; top blue trace) or amiloride (500 μm; bottom green trace). D1 receptor antagonists at two different concentrations (2 μm in blue circles, 10 μm in gray circles) failed to block the response (L). Amiloride (500 μm), an ASIC inhibitor, partially reduced the depolarization (M). Data expressed as mean ± SEM; $ represents the main effect of group. *p < 0.05.

We also compared the intrinsic properties of responsive neurons to test whether there was a difference based on the cellular response type and whether chronic ethanol exposure altered these properties (Fig. 6E–G). A two-way ANOVA revealed a main effect of response type on membrane capacitance (F_(1,67) = 12.65; p = 0.0007), and a trend toward a main effect of treatment (F_(1,67) = 3.244; p = 0.076) and treatment × response type interaction (F_(1,67) = 3.18; p = 0.079). Depolarized neurons had larger capacitance than the hyperpolarized responders, and post hoc analysis revealed a significant difference in membrane capacitance in the CIE-exposed mice (t₍₆₇₎ = 3.33; p = 0.003; Šídák's multiple comparisons test). There was no effect on input resistance (treatment × response type interaction: F_(1,67) = 0.19; p = 0.665; response type main effect: F_(1,67) = 2.66; p = 0.108; treatment main effect: F_(1,67) = 0.73; p = 0.397, as revealed by two-way ANOVA). Interestingly, both CIE and cellular response type had a significant effect on the resting membrane potential (Fig. 6G). Neurons that depolarized in response to optical stimulation rested at a more negative membrane potential compared with neurons that hyperpolarized (treatment × response type interaction: F_(1,60) = 2.98; p = 0.089; response type main effect: F_(1,60) = 30.14; p < 0.0001; and treatment main effect: F_(1,60) = 9.59; p = 0.003; post hoc Šídák's: depol vs hyperpol, t₍₆₀₎ = 3.12; p = 0.006 for air mice and depol vs hyperpol, t₍₆₀₎ = 4.52; p < 0.0001 for CIE mice as revealed by two-way ANOVA). Additionally, the cells in CIE group that hyperpolarized in response to optical stimulation were more depolarized in comparison to hyperpolarizing cells in the air group (t₍₆₀₎ = 2.81; p = 0.013; Šídák's multiple comparisons test). There was also a strong correlation between resting membrane potential and peak amplitude of cellular response (Fig. 6H; R² = 0.19; p = 0.015 for air mice and R² = 0.25; p = 0.003 for CIE-exposed mice).

Prior work from our lab suggested that optically stimulated depolarization and hyperpolarization responses in naive male mice were largely dependent on D1 and D2 receptor activation, respectively (Yu et al., 2021). Therefore, we measured the amplitude of voltage responses in the presence of dopamine receptor antagonists (data from air and CIE mice were pooled together). As expected, bath application of the D2 dopamine receptor antagonist sulpiride (2 μm) significantly reduced the hyperpolarizing potential (Fig. 6I,J; n = 4 cells; t₍₃₎ = 5.98; p = 0.009; paired Student's t test), suggesting that optically stimulated dopamine release hyperpolarized BNST neurons via D2 receptors. Surprisingly, in contrast to our previous findings, optically stimulated depolarization was not occluded in the presence of D1 receptor antagonist, SCH-23390 at two different concentrations, 2 μm (blue circles) and 10 μm (gray circles; Fig. 6K,L; n = 5 cells; t₍₄₎ = 1.33; p = 0.254; paired Student's t test). To further identify the mechanism driving this D1-resistant depolarization, we quantified the amplitude in the presence of various pharmacological agents. A recent study by Chuhma and colleagues reported a similar excitation in the dorsal striatum in response to optogenetic activation of DA terminals that was mediated by a combination of D1 and metabotropic glutamate type 1 (mglur1) receptors (Chuhma et al., 2018). Therefore, we repeated the experiment in the presence of both SCH-23390 (10 μm) and a mglur1 receptor antagonist, CPCOOEt (100 μm). Contrary to Chuhma and colleagues, the optically stimulated depolarization persisted in the presence of these antagonists (n = 2 cells; t₍₁₎ = 0.95; p = 0.518; paired Student's t test). The optically stimulated depolarization was also insensitive to other pharmacological blockers of receptors, such as metabotropic glutamate type 5, D2, α-adrenergic, β-adrenergic, and general dopamine receptors. This was perplexing and made us question whether this slow depolarization was driven by extracellular protons. Several studies have shown that prolonged presynaptic stimulation can increase extracellular proton concentrations, activating acid-sensing ion channels (ASICs) to modulate synaptic transmission (Zha et al., 2006; Chiang et al., 2015; Storozhuk et al., 2016; González-Inchauspe et al., 2017). ASICs are voltage-insensitive, sodium-permeable ion channels belonging to the degenerin/epithelial family (Price et al., 1996; Waldmann et al., 1997). In a study by Ferenczi et al., hippocampal optogenetic pulse trains at 20 Hz resulted in a slow nonsynaptic depolarizing current that was reduced by ∼50% in the presence of amiloride, a nonspecific inhibitor of ASICs (Ferenczi et al., 2016). Consistent with this idea, amiloride (500 μm) partially reduced the amplitude of the optically stimulated depolarizing potential (Fig. 6M; n = 5 cells; t₍₄₎ = 4.73; p = 0.009; paired Student's t test). This implies that while ASICs play a role, there are other mechanisms through which the vlPAG/DR^DA projections depolarize dBNST neurons. Collectively, this suggests that withdrawal from CIE could further facilitate the hyperexcitability of dBNST through the loss of D2-mediated hyperpolarization responses.

Discussion

Understanding how alcohol modulates neurochemically defined neurons in distinct brain circuits is crucial to identification of novel mechanisms involved in the development and maintenance of AUD. In the present study, we used converging technical approaches to gain insight into the plasticity of vlPAG/DR-dBNST dopaminergic synapses in male mice during 24-h withdrawal from chronic ethanol exposure. We identified numerous withdrawal-induced alterations of vlPAG/DR^DA intrinsic neuronal function along with synaptic plasticity of vlPAG/DR^DA-specific inputs to dBNST. Together, our data highlight the modulation of an understudied population of DA neurons and their projections to dBNST by chronic ethanol.

Increased excitability of vlPAG/DR^DA neurons following CIE withdrawal

Emerging evidence suggests that ventrolateral periaqueductal gray/dorsal raphe (vlPAG/DR) harbor a significant extra-VTA DA population that is alcohol-sensitive (Li et al., 2013) and is activated by both rewarding and aversive stimuli (Cho et al., 2017; Lin et al., 2020). Previously, we found that acute alcohol increased firing in male Swiss Webster mice, while two cycles of CIE had no effect on synaptic transmission on vlPAG/DR DA neurons (Li et al., 2013). Expanding on this work, here we saw an increase in the intrinsic excitability of vlPAG/DR^DA neurons in CIE-treated male mice with no alterations in spontaneous synaptic transmission. Our data also suggest a reduction in activation of an inwardly rectifying current is potentially associated with increased excitability. However, additional work is needed to identify the specific ion channel involved in driving CIE-induced hyperexcitability.

CIE-induced plasticity of glutamate co-transmission from vlPAG/DR^DA projections

DA neurons in the vlPAG/DR region co-express markers for glutamate (vglut2; Li et al., 2016) and send projections to both BNST and central amygdala (Meloni et al., 2006; Li et al., 2016). Consistent with our previous observations, we found that dopaminergic terminals originating from the vlPAG/DR region were most abundant in the lateral half of the dBNST (including the anterolateral nucleus, the oval nucleus, and the juxtacapsular nucleus) and co-released dopamine and glutamate. Photoactivation of vlPAG/DR^DA terminals resulted in excitatory monosynaptic and inhibitory polysynaptic connections (Li et al., 2016; Yu et al., 2021). In vlPAG/DR-dBNST specific synapses, we found a shift in the overall balance toward hyperexcitability in the CIE-treated mice, as evidenced by an increase in the E/I ratio and strengthening of AMPAR-mediated currents on dBNST neurons. Additionally, when we examined spontaneous synaptic activity onto light-responsive dBNST neurons, we saw a similar increase in the E/I ratio in CIE-treated mice. As changes in spontaneous synaptic activity predominantly depend on local BNST circuitry, this suggests an overall increase in dBNST network excitability. These results are consistent with prior literature supporting a general hyperactive state of the BNST in acute withdrawal following CIE. For example, work from our group and others indicates withdrawal from CIE increases neuronal excitability (Kash et al., 2009; Pati et al., 2020), excitatory drive (Silberman et al., 2013; Pleil et al., 2015), and long-term potentiation of glutamatergic synapses in the BNST (Wills et al., 2012).

In the present study, we saw an increase in the amplitude of EPSCs at −70 mV (AMPAR-mediated) and +40 mV (NMDAR-mediated) but no change in AMPA/NMDA ratio. Within the BNST, chronic ethanol exposure has been shown to increase temporal summation of NMDAR-mediated EPSCs (Kash et al., 2009) and enhance long-term potentiation (Wills et al., 2012) through a functional upregulation of NR2B subunit of NMDA receptor. Given that we did not pharmacologically isolate the NMDAR-mediated EPSCs, we cannot rule out the contribution of NMDAR plasticity in the vlPAG/DR-dBNST circuit.

Next, we also evaluated the impact of CIE on short-term plasticity by measuring paired pulse ratio at the vlPAG/DR-dBNST synapse. In air-exposed mice, photostimulation of vlPAG/DR^DA terminals resulted in paired-pulse depression which was reduced following withdrawal from CIE. While paired-pulse ratio is generally considered a measure of presynaptic release probability that relies on residual calcium at the release site (Zucker and Regehr, 2002), other cellular mechanisms may influence short-term synaptic plasticity. Therefore, chronic ethanol exposure could reduce paired-pulse depression by either blocking desensitization of AMPARs (Yamada and Tang, 1993), increasing faster recovery from desensitized states (Constals et al., 2015) or through the use-dependent removal of polyamine block of calcium-permeable AMPARs (Rozov and Burnashev, 1999). Short-term depression at excitatory synapses shapes neuronal adaptation by acting as a low-pass filter and dampening the transmission of information (Chung et al., 2002; Fuhrmann et al., 2002). Therefore, by reducing paired-pulse depression, CIE could strengthen the synapse and act as a high-pass filter, increasing information transfer at high firing frequencies during prolonged neural activity.

Complex effects of CIE on phasic vlPAG/DR-dBNST dopamine signals

Given the significance of phasic DA release in regulating alcohol-related behaviors, we next wanted to evaluate the impact of CIE on endogenous dopamine signaling in the vlPAG/DR-dBNST circuit. Complementary to our previous work and others (Li et al., 2016; Melchior et al., 2021; Yu et al., 2021), photostimulation of these terminals at different frequencies led to phasic DA release that was further augmented by inhibition of G_i-coupled D2 autoreceptors. Interestingly, acute withdrawal from CIE had no effects on optically targeted DA release and the sensitivity of terminal release regulating D2 autoreceptors. However, it significantly reduced the signal half-life putatively through alterations in D2 receptor-mediated uptake dynamics. D2 autoreceptors have been shown to regulate the clearance of DA by altering dopamine transporter function specifically during prolonged trains of stimulation (Wu et al., 2002; Benoit-Marand et al., 2011; Ford, 2014). While there are known regional and species-dependent variabilities in alcohol-mediated alterations of DA dynamics, overall, our finding is consistent with chronic ethanol driving a hypodopaminergic state through the modulation of DA release parameters (Budygin et al., 2007; Siciliano et al., 2015; Rose et al., 2016; Melchior and Jones, 2017). Most of these studies focused on DA dynamics in the nucleus accumbens; our work is the first to characterize the effects of chronic alcohol exposure on real-time DA dynamics from vlPAG/DR terminals in the dBNST.

Acute withdrawal from CIE also reduced the proportion of D2-mediated slow hyperpolarization of dBNST neurons tilting the ratio toward increased depolarization during phasic stimulation of vlPAG/DR^DA terminals at 20 Hz. Although not significant, the hyperpolarizing potential tended to be higher in CIE-exposed mice. This could reflect a potential compensatory mechanism to counter the loss of hyperpolarization. While this is the most parsimonious explanation of our findings, there are some caveats. Since both subsets of neurons had different intrinsic properties, and the direction of the optically stimulated membrane responses depended on the initial membrane potential, it is plausible that the CIE group had a more significant subset of neurons that depolarized because of sampling errors.

Surprisingly, while the hyperpolarization required D2 receptors as expected, the depolarization responses did not require D1-like receptors. This contradicted our previous findings, where we showed partial occlusion of optically stimulated depolarization in the presence of D1 antagonists (Yu et al., 2021). Given that neurons in both groups expressed D1-resistant depolarization, it implies that this altered response is not because of ethanol exposure. Plausible explanations include chronic stress associated with vapor chambers and the nonspecific effects of pyrazole (Pereira et al., 1992) since our previous findings were in naive male mice. While the exact mechanism through which phasic stimulation of dopaminergic terminals depolarizes dBNST neurons remains to be elucidated, our data suggest the contribution of ASICs. Supporting the role of ASICs in modulating dBNST neuronal activity is the observation that BNST is densely enriched with ASIC1a subunit and is critical for acid-evoked depolarization (Taugher et al., 2014). Further, Taugher et al., showed that deletion of ASIC1a specifically in the BNST reduced carbon dioxide-evoked freezing. Notably, ASICs have also been implicated in pain, addiction, fear learning, and anxiety-like behaviors (Wemmie et al., 2013; Kreple et al., 2014). Specifically, a recent study from the Wemmie lab identified a novel role of ASIC1a in acute effects of alcohol (Harmata et al., 2022). They found bath application of alcohol potentiated ASIC1a-mediated currents in the basolateral amygdala, and loss of ASIC1A increased the stimulant effects of alcohol while reducing the sedative effects.

Together, our work identifies the role of D2 receptors and ASICs in regulating ethanol withdrawal-induced plasticity in the vlPAG/DR^DA-dBNST circuit. Additional studies will be necessary to discern the complex mechanisms driving cellular and circuit activity in this pathway.

Functional implications

In conclusion, we provide evidence of neuroadaptations in the vlPAG/DR-dBNST dopamine circuit in male mice following withdrawal from chronic ethanol. Notably, adaptations include strengthening of glutamatergic transmission at vlPAG/DR-dBNST synapses and an overall hyperactive state. A major limitation of our study is the exclusion of female subjects given the differential impact of alcohol use disorder on men and women (Grant et al., 2015). Our group and others have identified sex-specific differences in rodents in ethanol-related behaviors following CIE (Jury et al., 2017; Xie et al., 2019). Chronic ethanol is also associated with sex-dependent differential neuroadaptations in multiple brain regions (Bach et al., 2021; Munier et al., 2022; Price and McCool, 2022), including BNST (Kasten et al., 2020; Levine et al., 2021; Marino et al., 2021). Additionally, we have identified sexually dimorphic behaviors following activation of vlPAG/DR^DA terminals in the BNST (Yu et al., 2021). We found activation of vlPAG/DR-BNST circuit reduces pain sensitivity in male mice, whereas in female mice it increases locomotion in the context of salient stimuli. Future studies will be required to examine whether withdrawal from ethanol also differentially engages this circuit in females.

The dBNST is highly heterogenous and forms reciprocal connections with brain regions implicated in stress responses and aversive and appetitive behaviors (Lebow and Chen, 2016; Vranjkovic et al., 2017). Since we did not target any specific neuronal population, it is plausible that we might have failed to capture a more selective and nuanced understanding of this pathway. Functional manipulation of dBNST subpopulations distinguished either by classical neurotransmitters (Jennings et al., 2013), pathway-specificity (Kim et al., 2013), interneurons versus projection neurons (Marcinkiewcz et al., 2016), or neuropeptide expression (Giardino et al., 2018) could result in divergent behavioral responses. However, it is well-established that nonspecific activation of dBNST is anxiogenic (Resstel et al., 2008; Sajdyk et al., 2008; Kim et al., 2013). Therefore, one potential outcome of an increased excitation-inhibition ratio and the loss of D2 receptor-mediated inhibition would be a net excitation of the dBNST network. Additionally, given that dopaminergic terminals target CRF neurons in the dBNST (Meloni et al., 2006) and DA application can increase glutamate release in the BNST in a CRF-dependent manner (Kash et al., 2008), one of the ways alcohol may influence CRF signaling at a circuit level is through manipulation of vlPAG/DR^DA terminals. This could ostensibly drive withdrawal-induced negative affective behaviors. The contribution of this understudied pathway in pathophysiological phenotypes associated with AUD is a potentially exciting future direction.