Ethics statement

For all behavioral and electrophysiological experiments involving anesthesia, isoflurane inhalation was used. For electrophysiological experiments, anesthesia was followed by rapid decapitation to allow brain slice preparation. All behavioral studies were approved by the Institutional Animal Care and Use Committee (ACUC) of the National Institute on Drug Abuse Intramural Research Program. All electrophysiological procedures were approved by the IACUC of The Scripps Research Institute. All procedures were conducted according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (8th edition).

Animals

Rats used in behavioral (male Wistar, n = 86; Charles River, Kingston, NY) and electrophysiology (male Sprague-Dawley, n = 87; Charles River, Raleigh, NC) experiments weighed 225 to 275 g upon arrival, were group-housed (2 to 3 per cage) in standard plastic cages in a temperature- and humidity-controlled room, and were maintained under a reverse 12 h/12 h light/dark cycle with food and water available ad libitum except during behavioral testing.

Drugs

For behavioral experiments, alcohol (ethanol; Warner Graham, Cockeysville, MD) was dissolved in tap water. Oxytocin (ChemPep Inc., Wellington, FL) was dissolved in 0.9% saline. For peripheral administration, the non–blood-brain barrier-penetrant oxytocin receptor agonist PF-06655075 was initially dissolved in 10% (v/v) Self Emulsifying Drug Delivery System that consisted of 3:4:3 Miglyol 812:Cremophor RH40:Capmul MCM (v/v/v). The emulsion was completed with 90% (v/v) 50 mM aqueous phosphate buffer (pH 7.4). For intracerebroventricular administration, PF-06655075 was initially dissolved in 5% (v/v) dimethyl sulfoxide, then combined with 5% (v/v) polyethylene glycol 300, and completed with 90% (v/v) saline. PF-06655075 and the vehicle used for its peripheral administration were kindly provided by Pfizer. The non–blood-brain barrier penetrant antagonist L-371,257 (Tocris Bioscience, Ellisville, MO) was prepared in a vehicle of 5% (v/v) dimethyl sulfoxide, 5% (v/v) Cremophor, and 90% (v/v) sterile water.

For electrophysiology experiments, oxytocin and the selective vasopressin receptor 1A antagonist (d(CH2)5,Tyr(Me)2,Arg8)-Vasopressin (TMA) [34] were obtained from Tocris (Ellisville, MO). The selective oxytocin receptor antagonist desGly-NH2-d(CH2)5[D-Tyr2,Thr4]OVT (OTA) was provided by Dr. Maurice Manning (University of Toledo, OH) [34]. CGP 55845A, DL-2-amino-5-phosphonovalerate (DL-AP5), and 6,7-dinitroquinoxaline-2,3-dione (DNQX) were obtained from Tocris (Ellisville, MO). Tetrodotoxin (TTX) was purchased from Biotium (Hayward, CA). Alcohol was purchased from Remet (La Mirada, CA, USA). All drugs were dissolved in artificial cerebrospinal fluid (aCSF).

Operant alcohol self-administration and alcohol vapor exposure

Oral alcohol self-administration experiments were conducted in standard operant chambers (Med Associates, St. Albans, VT) fitted with 2 retractable levers and a dual-cup liquid receptacle. After initial training, as previously described [29,35], the animals were allowed to lever press for alcohol (10%, w/v; 0.1 ml) and water (0.1 ml) on separate levers according to a concurrent fixed-ratio 1 (FR1) schedule of reinforcement (each lever press resulted in fluid delivery) in 30-min operant sessions. After each training session, the liquid receptacle and surrounding area were inspected to confirm the consumption of earned reinforcers. After response acquisition, rats were split into 2 groups matched by their alcohol consumption. For the remainder of these experiments, rats in the nondependent group were exposed to air without alcohol, whereas rats in the dependent group were exposed to alcohol vapor in daily cycles designed to cause intoxication (14 h vapor “on”; 200 mg/dl target blood alcohol levels) and withdrawal (10 h vapor “off”) to induce alcohol dependence, as previously described [29,35]. Dependence is characterized by somatic and motivational signs of withdrawal that include increased anxiety-like behavior, reward deficits, and enhanced motivation to self-administration of alcohol [36–39]. In all operant alcohol self-administration experiments, steady baseline consumption of alcohol was established in dependent and nondependent rats before pharmacological testing began. Operant alcohol self-administration sessions including pharmacological tests were conducted 2 to 3 sessions per week (never on consecutive days to minimize potential carry-over effects) during the 10 h “off” period, 6 to 8 h into withdrawal.

Effect of intraperitoneal oxytocin on alcohol intake (FR1) and motivation (progressive ratio)

Intraperitoneal oxytocin (0, 0.125, 0.25, 0.5, and 1 mg/kg; 0.5 or 1 ml/kg) was administered 30 min prior to FR1 alcohol self-administration sessions in dependent (n = 10) and nondependent (n = 10) rats. The pretreatment time was selected based on previous work that determined peak brain oxytocin concentrations following intraperitoneal oxytocin administration [22]. The test order of intraperitoneal doses was counterbalanced using a within-subjects Latin-square design. Self-administration of alcohol and water was recorded during tests. Based on results of the FR1 test, the 0, 0.125, and 0.25 mg/kg intraperitoneal doses were then tested on a progressive ratio (PR) schedule of reinforcement, in which the number of lever presses required to obtain the next alcohol reinforcer increased progressively, as follows: 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 7, 7, 9, 9, 11, 11, 13, 13, etc. The last completed ratio (breakpoint) was used as an indication of motivation for alcohol. During the PR test, only the alcohol lever was made available. Sessions lasted 90 min or until 15 min had elapsed without a response. The pretreatment time during the PR test was the same as for the FR1 test. Again, these doses were administered in a Latin-square design, with intervening nontesting days.

Effect of intranasal oxytocin on alcohol intake (FR1) and motivation (PR)

To reestablish a stable baseline of intake, the same rats described above were moved to different operant chambers with matching operant manipulanda and reinforcer receptacles. They were allowed eight 30-min FR1 sessions to obtain alcohol and water (S1 Fig). Next, intranasal oxytocin was administered 1 h prior to testing responding for alcohol on FR1 and PR schedules of reinforcement (described above). The pretreatment time was selected based on previous work that determined peak brain oxytocin concentrations following intranasal oxytocin administration [22]. Again, tests were conducted in Latin-square designs with intervening nontesting days. For intranasal drug delivery, general anesthesia was rapidly induced by isoflurane (5% for 2 to 5 min), and then rats were immediately administered intranasal oxytocin (0, 0.25, 0.5, and 1 mg/kg/20 μl; based on the FR1 data, the intranasal doses of 0.5 and 1 mg/kg/20 μl were selected for testing on the PR schedule) using the rat Precision Olfactory Device (rPOD; Impel NeuroPharma, Seattle, WA) and allowed to recover before being returned to their home cage.

Effect of systemic and intranasal oxytocin on locomotion, grooming, motor coordination, and consumption of nonalcoholic palatable solutions

Based on the effects of oxytocin on alcohol drinking and the motivation for alcohol, we selected the doses of intraperitoneal (0.25 mg/kg) and intranasal (1 mg/kg) oxytocin for further behavioral testing. These doses by each route were the lowest dose by each route that significantly reduced alcohol drinking and the motivation for alcohol specifically in dependent rats. The pretreatment times were the same as used for testing on alcohol self-administration.

Behavioral tests were conducted in separate cohorts of rats to assess potential non–alcohol-specific effects of oxytocin on locomotion and grooming in an open field (dependent, n = 6; nondependent, n = 6) and motor coordination on a rotarod (dependent, n = 5; nondependent, n = 7). Two blind observers scored any instances of grooming behavior according to previously published work. A single point was given for each instance of vibration of forepaws, face washing, body grooming, body scratching, paw licking, head shaking, body shaking, and genital grooming, according to previously reported studies [40,41]. Points were totaled for each animal to yield a grooming score.

In addition to alcohol’s pharmacological effect, alcohol contains calories and has a sweet-taste component that contribute to the reinforcer efficacy of alcohol. To assess the potential role of calories and sweet taste in the ability of intraperitoneal and intranasal oxytocin to reduce alcohol consumption, we tested oxytocin on the consumption of a nonalcoholic caloric reinforcer without a sweet taste (5% maltodextrin; n = 7) and on a nonalcoholic sweet solution without caloric content (0.1% saccharin; n = 8; additional details provided in S1 Text).

Central versus peripheral mediation of oxytocin’s effect on alcohol intake

Three separate cohorts of alcohol-dependent rats were used (central oxytocin administration: n = 6; peripheral oxytocin receptor agonist: n = 9; peripheral oxytocin receptor antagonist combined with intranasal oxytocin administration: n = 12). A subset of the group used for testing the peripheral oxytocin receptor antagonist combined with intranasal oxytocin administration was given 2 sessions in the absence of any treatment to reestablish a baseline and then used to test the effect of central administration of the oxytocin receptor agonist PF-06655075 (central oxytocin receptor agonist: n = 7). The central oxytocin and PF-06655075 administration cohorts were surgically implanted with a guide cannula to allow intracerebroventricular administration of these compounds.

The oxytocin receptor antagonist L-371,257, which does not cross the blood-brain barrier, was tested in combination with intranasal oxytocin administration to test the ability of peripheral antagonism to reverse the ability of intranasal oxytocin to reduce alcohol drinking in dependent rats. L-371,257 is a potent and competitive antagonist of the oxytocin receptor (pA2 = 8.4) with high affinity at both the oxytocin receptor (Ki = 19 nM) and vasopressin V1a receptor (Ki = 3.7 nM) [42]. The pretreatment time and dose (45 min; 5 mg/kg) were selected to be in excess of doses that provided blockade of oxytocin-induced uterine contractions for at least several hours following peripheral (intravenous or intraduodenal) administration in rats [42]. Doses much lower than this in rats (0.5 mg/kg intraperitoneal) [43] and mice (300 μg/kg intranasal) [44,45] have been demonstrated to provide behaviorally effective antagonism. Additionally, in previous studies, comparable doses of L-371,257 versus oxytocin (0.5 mg/kg versus 0.5 mg/kg intraperitoneal in rats [43]; 300 μg/kg versus 200 μg/kg intranasal in mice [45]) were administered. Here, we administered 5 times the dose of the antagonist compared to oxytocin (5 mg/kg versus 1 mg/kg) via a preferred route of peripheral absorption (intraperitoneal injection versus intranasal application [23]) to conservatively test for the ability of the peripherally acting antagonist to alter the actions of intranasal oxytocin. Rats were tested in a within-subjects Latin-square design with the following combinations of intraperitoneal and intranasal treatments: intraperitoneal vehicle with intranasal vehicle, intraperitoneal vehicle with intranasal oxytocin (1 mg/kg/20 μl), intraperitoneal L-371,257 (5 mg/kg/ml) with intranasal vehicle, and intraperitoneal L-371,257 (5 mg/kg/ml) with intranasal oxytocin (1 mg/kg/20 μl). Operant training and induction of dependence were conducted, as described above.

To test the effect of central versus peripheral oxytocin receptor agonism on dependence-induced alcohol drinking, we administered centrally or peripherally a long-acting, large molecule that does not penetrate the blood-brain barrier (PF-06655075). For central administration, 7 rats were surgically implanted with a guide cannula to allow intracerebroventricular administration of PF-06655075. These rats were previously used to test the peripheral oxytocin receptor antagonist combined with intranasal oxytocin administration (data shown in Fig 3B). The PF-06655075 dose of 30 μg was selected for intracerebroventricular administration to match the highest dose of oxytocin administered by the same route, a dose that blocked drinking in alcohol-dependent rats. For peripheral administration, the rats were administered 1 mg/kg of PF-06655075 subcutaneously. This dose was selected taking into account the high plasma protein binding of PF-06655075 (i.e., resulting in lower unbound concentrations than oxytocin) as well as its oxytocin receptor binding Ki. Subcutaneous administration of 1 mg/kg of PF-06655075 versus 1 mg/kg of oxytocin would be expected to produce comparable receptor occupancy (94.2% versus 99.7% at Cmax; see S1 Text for calculations based on data of Modi and colleagues [26]). Therefore, 1 mg/kg of subcutaneous PF-06655075 was expected to recapitulate the putative peripheral binding of oxytocin. This hypothesis was based on the observation that lower doses of intraperitoneal oxytocin were sufficient to block drinking in alcohol-dependent rats in the present study. Specifically, 0.25 mg/kg of intraperitoneal oxytocin reduced alcohol drinking in dependent rats to a similar extent as 30 μg of intracerebroventricular oxytocin. Finally, this systemic dose of 1 mg/kg PF-06655075 has been reported to reduce fear behavior potentially via anti-sympathetic action in the periphery [26].

Electrophysiology

Dependent and nondependent rats were deeply anesthetized with isoflurane followed by rapid decapitation and immediate removal of the brain into an ice-cold high-sucrose brain slice cutting solution (sucrose 206 mM; KCl 2.5 mM; CaCl2 0.5 mM; MgCl2 7 mM; NaH2PO4 1.2 mM; NaHCO3 26 mM; glucose 5 mM; HEPES 5 mM [pH 7.4]). Coronal slices (300 to 400 μm) containing the CeA were continuously superfused (flow rate of 2 to 4 ml/min) with 95% O₂/5% CO₂ equilibrated aCSF of the following composition: NaCl 130 mM, KCl 3.5 mM, NaH2PO4 1.25 mM, MgSO4·7H2O 1.5 mM, CaCl2 2.0 mM, NaHCO3 24 mM, and glucose 10 mM. Recordings were performed in neurons from the medial subdivision of the CeA. Each experimental group contained neurons from a minimum of 3 rats. GABAergic activity was pharmacologically isolated with DNQX, DL-AP5, and CGP. All drugs were applied by bath superfusion.

We recorded with sharp micropipettes filled with 3M KCl and evoked GABAergic inhibitory postsynaptic potentials (eIPSPs) by stimulating locally within the medial subdivision of CeA through a bipolar electrode. Neurons were held near their resting membrane potential (−82.4 ± 0.8 mV). We performed an input–output (I/O) protocol consisting of a range of 5 current stimulations, starting at the threshold current required to elicit an eIPSP, up to the strength required to elicit the maximum subthreshold amplitude. The middle stimulus intensity was used to monitor drug-induced changes throughout the duration of the experiment. Paired-pulse ratio (PPR) was performed at the stimulus intensity giving approximately 50% of the maximal amplitude determined in the I/O protocol.

Whole-cell voltage-clamp recordings of GABAergic spontaneous inhibitory postsynaptic currents (sIPSCs) and miniature inhibitory postsynaptic currents (mIPSCs) were from visualized CeA neurons clamped at −60 mV for the duration of the recordings. Patch pipettes (3 to 6 MΩ) were filled with an internal solution composed of the following (in mM): 145 KCl, 0.5 EGTA, 2 MgCl2, 10 HEPES, 2 Na-ATP, and 0.2 Na-GTP. In all experiments, cells with a series resistance greater than 25 MΩ were excluded from analysis, and series resistance was continuously monitored during gap-free recording with a 10-mV pulse. Cells in which series resistance changed more than 25% during the course of the experiment were excluded from analysis. All measures were performed prior to (baseline) and during drug application (details in S1 Text).

Statistical analysis

Results are presented as the mean ± standard error of the mean. The level of significance was established as p < 0.05. Statistical analyses were performed in Prism 6 (Graphpad Software, Inc., La Jolla, CA). Behavioral data were analyzed with one-way (Dose, Session, or Treatment) repeated-measures analysis of variance (R.M. ANOVA), two-way (Group × Dose) R.M. ANOVA, or by paired-samples t test. The Holms-Sidak test was used for post hoc comparisons. Electrophysiology data were analyzed with two-way ANOVA (Group × Concentration or Group × Treatment) followed by Bonferroni post hoc comparisons, one-sample t test, or independent-samples t test, as appropriate.

Effect of intraperitoneal oxytocin on alcohol intake (FR1) and motivation (PR)

Passive exposure to alcohol vapor has been demonstrated to cause somatic signs of dependence in alcohol withdrawal as well as a dysregulation of reward and stress systems. The cardinal feature of the model is the increased consumption and motivation for alcohol exhibited by alcohol-dependent rats when allowed to perform an operant response for access to alcohol [3]. Prior to pharmacological testing, alcohol-dependent rats exposed to chronic-intermittent alcohol vapor exhibited significantly enhanced consumption of alcohol in comparison to nondependent rats that were exposed to air in their home cage throughout the study (S1 Fig).

Oxytocin abolished the difference in alcohol drinking between dependent and nondependent rats at doses of ≥ 0.25 mg/kg (Fig 1). A 2 × 5 (Group × Dose) R.M. ANOVA yielded a significant Group × Dose interaction (F_{4, 72} = 7.98, p < 0.0001). Post hoc analyses indicated that dependent rats self-administered significantly more alcohol than the nondependent rats following 0 (p < 0.0001) and 0.125 mg/kg (p < 0.001) oxytocin doses, but this difference disappeared at higher doses. Responding was significantly lowered at the 0.25, 0.5, and 1 mg/kg doses (all p < 0.0001) in dependent rats compared with the vehicle condition, whereas only the highest dose of 1 mg/kg significantly lowered lever pressing for alcohol in nondependent rats (p < 0.05). Water intake data during all tests are presented in S2 Fig.

Oxytocin (0.125 or 0.25 mg/kg) selectively eliminated the increased motivation for alcohol in dependent rats (Fig 1B). A 2 × 3 (Group × Dose) R.M. ANOVA yielded a significant Group × Dose interaction (F_{2, 36} = 3.59, p < 0.05). The difference in PR breakpoint between groups following vehicle treatment was significant (p < 0.05). When treated with either 0.125 or 0.25 mg/kg doses of oxytocin, dependent rats no longer differed from nondependent rats. Oxytocin (0.125 and 0.25 mg/kg) significantly reduced PR breakpoint for alcohol in dependent rats (all p < 0.01), whereas responding in the nondependent group was not altered by oxytocin.

Effect of intranasal oxytocin on alcohol intake (FR1) and motivation (PR)

Intranasal oxytocin dose-dependently decreased alcohol drinking in dependent rats without affecting drinking in nondependent rats (Fig 1C). The 2 × 4 (Group × Dose) R.M. ANOVA yielded a significant Group × Dose interaction (F_{3, 54} = 14.18, p < 0.0001). Dependent rats drank more alcohol than nondependent rats treated with saline, 0.25, and 0.5 mg/kg oxytocin (all p < 0.001). This difference was no longer observed following treatment with 1 mg/kg oxytocin. Alcohol drinking in dependent rats significantly decreased across the doses of oxytocin compared to the saline condition (all p < 0.0001), whereas drinking in the nondependent group was not significantly altered by oxytocin.

On the PR test (Fig 1D), two-way R.M. ANOVA yielded a significant Group × Dose interaction (F_{2, 36} = 4.27, p < 0.05). Following saline treatment, dependent rats had a significantly higher breakpoint for alcohol than nondependent rats (t₅₄ = 4.50, p < 0.001). This difference was no longer observed following 0.5 or 1 mg/kg oxytocin administration. Oxytocin (1 mg/kg) reduced PR breakpoint in the dependent group (p < 0.001) without altering the behavior of nondependent rats.

Effect of intraperitoneal and intranasal oxytocin on locomotion, grooming, rotarod motor coordination, and consumption of nonalcoholic palatable solutions

Dependent and nondependent rats were not significantly different in their spontaneous locomotion or grooming behavior. Intraperitoneal oxytocin (0.25 mg/kg) significantly decreased locomotion in both dependent and nondependent rats (Fig 2A; Dose: F_{1, 10} = 14.63, p < 0.01; post hoc tests all p < 0.05), whereas intranasal oxytocin (1 mg/kg) had no effect in either group. Intraperitoneal and intranasal oxytocin treatments did not affect grooming behavior (which remained minimal throughout testing; Fig 2A inset). Dependent and nondependent rats did not significantly differ in their rotarod performance, and this performance was not significantly altered by intraperitoneal or intranasal oxytocin treatments (Fig 2B). Intraperitoneal oxytocin reduced saccharin (t₆ = 2.80, p < 0.05) and maltodextrin (t₆ = 3.20, p < 0.05) consumption, whereas intranasal oxytocin had no effect (Fig 2C).

Central versus peripheral effect of oxytocin on alcohol intake

Baseline sessions conducted between intracerebroventricular oxytocin test sessions did not significantly differ, so they were combined for analysis (Fig 3A). All doses of intracerebroventricular oxytocin reduced alcohol consumption in dependent rats (Dose: F_{3, 15} = 10.25, p < 0.001; post hoc tests: all p < 0.05).

Intranasal oxytocin treatment significantly reduced drinking in dependent rats whether the rats were pretreated with either vehicle (VEH + VEH versus VEH + OXT) or with the peripherally restricted oxytocin receptor antagonist L-371,257 (L-371,257 + VEH versus L-371,257 + OXT; Treatment: F_{3, 33} = 10.73, p < 0.0001; post hoc tests: all p < 0.05). L-371,257 did not significantly alter drinking by itself (VEH + VEH versus L-371,257 + VEH) nor did it reverse the ability of intranasal oxytocin to reduce alcohol drinking in dependent rats (VEH + OXT versus L-371,257 + OXT; Fig 3B).

Baseline sessions conducted between test sessions with intracerebroventricular administration of PF-06655075 did not significantly differ and thus were combined for analysis (Fig 3C). Intracerebroventricular administration of PF-06655075, which does not cross the blood-brain barrier and thus was expected to not diffuse to the periphery, significantly reduced responding for alcohol relative to baseline responding and relative to vehicle (Treatment: F_{2, 12} = 42.25, p < 0.0001; post hoc tests: all p < 0.0001). Vehicle administration did not significantly alter responding relative to baseline (Fig 3C). Systemic administration of PF-06655075 (expected to not reach the brain) did not significantly alter responding for alcohol relative to baseline responding or the vehicle condition nor did the vehicle itself significantly alter responding relative to baseline (Fig 3D).

Oxytocin decreases evoked CeA GABAergic signaling in nondependent but not dependent rats

We recorded from neurons in the medial subdivision of the CeA, using sharp intracellular whole cell configuration for locally evoked inhibitory GABAergic eIPSP and whole-cell patch-clamp configuration for GABAergic postsynaptic currents. We found no difference in input resistance of neurons from nondependent (154.4 ± 11.4 MΩ) and dependent (159.7 ± 7.6 MΩ) rats and no difference in spike frequency (S3 Fig) between groups. Baseline GABAergic eIPSP amplitudes stimulated locally within the medial subdivision of CeA were not different between nondependent (9.7 ± 0.7 mV) and dependent (9.9 ± 0.7 mV) animals (S4 Fig). No differences were observed in PPR of eIPSPs at 100 ms interstimulus interval in nondependent (1.06 ± 0.08) and dependent (1.01 ± 0.09) neurons.

In CeA neurons of nondependent rats, 100 nM oxytocin (10 to 15 min) did not alter eIPSP amplitudes, whereas 500 nM and 1,000 nM significantly decreased amplitudes to 83.4% ± 4.9% (t₁₀ = 3.40, p < 0.01) and 74.1% ± 7.7% (t₄ = 3.37, p < 0.05) of baseline, respectively (Fig 4B). Notably, in dependent rats, oxytocin at all 3 concentrations did not affect eIPSPs. A two-way ANOVA (Group × Concentration) yielded a significant Group effect (F_1,44 = 4.55, p < 0.05). Oxytocin did not alter input resistance of nondependent (baseline: 144.3 ± 13.8 MΩ, oxytocin: 148.3 ± 13.7 MΩ) or dependent (baseline: 148.2 ± 10.6 MΩ, oxytocin: 141.8 ± 12.9 MΩ) CeA neurons, suggesting that oxytocin effects on eIPSP amplitudes were not due to excitability changes but due to decreased GABAergic transmission. Oxytocin (500 nM) significantly increased the PPR of eIPSPs (baseline PPR: 0.98 ± 0.11; oxytocin PPR: 1.35 ± 0.14; t₁₀ = 2.56, p < 0.05) in nondependent rats, suggesting a possible decrease in the presynaptic release of GABA (as changes in PPR are inversely related to changes in release) [46]. Oxytocin did not alter PPR in dependent rats (baseline PPR: 1.17 ± 0.22, oxytocin PPR: 1.03 ± 0.07).

Alcohol increases evoked CeA GABAergic transmission

Acute alcohol (44 mM; maximal dose [47]) application significantly increased eIPSP amplitudes in CeA of both nondependent (to 140.8% ± 8.5% of baseline; t₇ = 4.79, p < 0.01) and dependent rats (to 133.9% ± 13.4% of baseline; t₆ = 2.53, p < 0.05; Fig 4C), indicating a lack of tolerance for the acute effects of alcohol on stimulated GABAergic signaling, as previously reported [47]. Additionally, alcohol did not alter input resistance of nondependent (baseline: 169.0 ± 19.2 MΩ, alcohol: 167.6 ± 15.8 MΩ) or dependent (baseline: 167.6 ± 10.4 MΩ, alcohol: 161.4 ± 11.4 MΩ) neurons. Because oxytocin also binds to vasopressin receptors and there is a CeA subpopulation of neurons sensitive to vasopressin binding (specifically, via the V1a but not V1b vasopressin receptor), we always preapplied the selective vasopressin receptor 1A antagonist TMA (500 nM) to isolate specific oxytocin receptor-mediated actions [34,48]. TMA had no effect on eIPSP amplitudes (nondependent: 8.51 ± 1.08 mV and dependent: 9.68 ± 1.31 mV without TMA) and did not affect the magnitude of the alcohol-induced increase of eIPSPs in both nondependent (to 148.6% ± 10.5% of baseline; t₅ = 4.64, p < 0.01) and dependent rats (129.8% ± 10.6% of baseline; t₆ = 2.80, p < 0.05; S5 Fig).

To determine potential interactions between oxytocin and acute alcohol, in most of the neurons that received 500 nM oxytocin, we applied alcohol in the presence of oxytocin. In this subset of neurons (7 of 9) from nondependent slices, coapplication of alcohol only induced a moderate, nonsignificant increase of eIPSP amplitudes (oxytocin: 83.4% ± 4.9%; oxytocin + alcohol: 101.7% ± 8.3% of baseline; Fig 4C). In the subset of neurons (12 of 15) from dependent slices, alcohol did not further increase eIPSPs (oxytocin: 107.6% ± 9.3%; oxytocin + alcohol: 123.0% ± 16.8% of baseline; Fig 4C), suggesting a blockade of oxytocin-induced decreased GABA transmission in dependent rats. A two-way ANOVA (Group × Treatment) yielded a significant main effect of Treatment (F_2,53 = 6.33, p < 0.01), and a Bonferroni post hoc test indicated a significant difference between the effects of alcohol and oxytocin (t₅₃ = 3.35, p < 0.01). Similarly, oxytocin at 100 and 1,000 nM blunted the alcohol-induced facilitation of eIPSPs in both nondependent (84.2% ± 11.0% and 82.1% ± 13.2% of baseline, respectively) and dependent slices (124.0% ± 18.6% and 125.5% ± 15.3% of baseline, respectively). Overall, oxytocin blunted the alcohol-induced increase of GABA transmission (see Fig 4C).

Oxytocin decreases CeA GABA_A receptor function

To further investigate the pre- versus postsynaptic action of oxytocin on GABA signaling, we performed whole-cell patch-clamp recordings of sIPSCs and mIPSCs in CeA neurons. Generally, changes in IPSC frequency reflect altered transmitter release, and changes in amplitude or kinetics reflect alterations in postsynaptic GABA_A receptor sensitivity. However, altered amplitude may also reflect a mix of pre- and postsynaptic effects [49,50].

We first investigated sIPSCs, finding no significant differences between nondependent and dependent rats in sIPSC frequency (1.1 ± 0.2 and 0.8 ± 0.1 Hz, respectively), amplitude (82.0 ± 8.4 and 66.6 ± 5.8 pA, respectively), rise time (2.6 ± 0.1 and 2.8 ± 0.1 ms, respectively), or decay time (8.0 ± 0.8 and 9.4 ± 1.0 ms, respectively). We next used the vasopressin receptor 1A antagonist TMA and found no effects of TMA alone in either nondependent or dependent rats (S6 Fig), as observed for evoked GABAergic responses. In CeA neurons of nondependent rats, application of 500 nM oxytocin significantly decreased the amplitude of sIPSCs to 85.2% ± 6.7% of baseline (t₁₁ = 2.22, p < 0.05) and increased rise times to 110.5% ± 4.7% of baseline (t₁₁ = 2.24, p < 0.05) with no changes in frequency or decay times (Fig 5A), indicating mainly postsynaptic actions of oxytocin to decrease GABA_A receptor function. In dependent rats (Fig 5A), oxytocin also significantly decreased sIPSC amplitude to 83.1% ± 5.0% of baseline (t₈ = 3.37, p < 0.01), with no changes in sIPSC frequency or kinetics.

We next investigated mIPSCs in the presence of the sodium channel blocker TTX to isolate action potential independent currents. Generally, mIPSC analysis reveals specific effects of drugs on vesicular release of GABA, which results from exocytosis of neurotransmitter containing vesicles in a manner independent of action-potential–induced release mechanisms. In addition, sIPSCs and mIPSCs likely result from distinct presynaptic neurotransmitter release mechanisms, e.g., vesicle fusion machinery, spatial segregation of the vesicles and/or vesicle populations, and synaptic vesicle pool dynamics [51]. Similar to sIPSCs, we did not find any baseline differences between nondependent and dependent rats in mIPSC frequency (0.8 ± 0.1 and 0.9 ± 0.3 Hz, respectively), amplitude (53.2 ± 4.0 and 60.7 ± 12.3 pA, respectively), rise time (2.5 ± 0.1 and 2.5 ± 0.1 ms, respectively), or decay time (5.5 ± 0.6 and 6.2 ± 0.8 ms, respectively). Oxytocin did not alter mIPSC frequency, amplitude, or kinetics for neurons from nondependent or dependent rats (Fig 5B), suggesting no effect of oxytocin on this form of GABA release in the medial subdivision of CeA.

Oxytocin blocks alcohol’s effects on GABA release in dependent rats

We next examined the interaction of oxytocin and acute alcohol on CeA sIPSCs in nondependent and dependent rats. In 8 of the 12 CeA neurons from nondependent rats that received 500 nM oxytocin, and 7 of the 9 neurons from dependent rats, we coapplied alcohol (44 mM) and compared effects between groups on sIPSC measures (Fig 5D–5G). A two-way ANOVA (Group × Treatment) yielded a significant effect of Treatment (F_1,13 = 4.87, p < 0.05) and a Group × Treatment interaction (F_1,13 = 9.84, p < 0.01) for sIPSC frequency (Fig 5D). The Bonferroni post hoc comparisons indicated that alcohol with oxytocin significantly increased sIPSC frequency compared with oxytocin alone in nondependent rats (t₁₃ = 3.91, p < 0.01). Therefore, oxytocin blocked the alcohol-induced increase in GABA release only in CeA neurons of alcohol-dependent rats. Additionally, a two-way ANOVA (Group × Treatment) yielded a significant Group × Treatment interaction (F_1,13 = 20.88, p < 0.001) for sIPSC decay time (Fig 5G). The Bonferroni post hoc comparisons indicated that alcohol with oxytocin significantly increased decay times compared with oxytocin alone in nondependent rats (t₁₃ = 4.86, p < 0.001).

As previously shown [52], we found that in nondependent and dependent rats, acute alcohol alone significantly increased sIPSC frequency to 145.5% ± 13.4% of baseline (t₅ = 3.40, p < 0.05) and 131.4% ± 8.6% of baseline (t₉ = 3.65, p < 0.01), respectively (Fig 5H), with no changes in sIPSC amplitude or kinetics, suggesting increased action-potential–dependent GABA release. We compared the effects of alcohol alone with the effects of alcohol and oxytocin using two-way ANOVAs (Group × Treatment) for each sIPSC measure (Fig 5H–5K). We found a significant main effect of Treatment (F_1,27 = 9.00, p < 0.01) and a significant Group × Treatment interaction (F_1,27 = 5.61, p < 0.05) for sIPSC rise times (Fig 5J). The Bonferroni post hoc comparisons indicated a significant difference in the effects of alcohol with oxytocin versus alcohol alone for nondependent neurons.

Finally, we found that the selective oxytocin receptor antagonist desGly-NH2-d(CH2)5[D-Tyr2,Thr4]OVT alone did not alter sIPSCs (S7 Fig), suggesting no basal activity of these receptors on spontaneous GABAergic transmission in dependent rats. Notably, in neurons from dependent rats, oxytocin in the presence of OTA had no effect on sIPSCs when compared to baseline (Fig 5L and 5M). In a subset of these neurons (8 out of 11), we applied alcohol in the presence of both OTA and oxytocin, resulting in a significant increase in sIPSC frequency (t₇ = 3.01, p < 0.05; Fig 5M). We found a similar effect of OTA to rescue the alcohol effect on eIPSPs from nondependent neurons (t₇ = 2.70, p < 0.05; S8 Fig). These results confirm that oxytocin receptor antagonism blocked postsynaptic effects of oxytocin and restored the acute alcohol-induced facilitation of GABA release in the CeA of alcohol-dependent rats.