The diabetes drug semaglutide lowers alcohol drinking and changes brain signals that calm nerve activity

For mice drinking sweetened (sweet) alcohol, a main effect of Dose (F_5,65 = 51.81, P < 0.0001) was found; semaglutide at all doses (P < 0.0001), compared with vehicle, reduced intake. There was no main effect of Sex or Dose × Sex interaction. Male and female data were combined for visualization, but the individual data points are depicted by sex-specific symbols (Figure 1A).

For mice drinking unsweetened (unsweet) alcohol, a main effect of Dose (F_5,70 = 9.12, P < 0.0001) was found; semaglutide at 0.003 mg/kg (P = 0.05), 0.01 mg/kg (P = 0.0007), 0.03 mg/kg (P < 0.0001), and 0.1 mg/kg (P < 0.0001), compared with vehicle, reduced intake. A main effect of Sex (F_1,14 = 7.66, P = 0.02; female > male), but no Dose × Sex interaction, was also observed (Figure 1B).

For mice drinking a sweet caloric solution not containing alcohol (glucose + saccharin), a main effect of Dose (F_5,65 = 5.53, P = 0.0003) was found; semaglutide at 0.003 mg/kg (P = 0.021), 0.01 mg/kg (P = 0.001), 0.03 mg/kg (P = 0.002), and 0.1 mg/kg (P = 0.0007), compared with vehicle, reduced intake. There was no main effect of Sex or Dose × Sex interaction (Figure 1C).

For mice drinking water, a main effect of Dose (F_5,35 = 18.64, P < 0.0001) was found; semaglutide at all doses (P < 0.0001), compared with vehicle, reduced intake (Figure 2A). For mice drinking a sweet noncaloric solution (saccharin), a main effect of Dose (F_5,35 = 18.02, P < 0.0001) was found; semaglutide at 0.01 mg/kg (P = 0.005), 0.03 mg/kg (P = 0.002), and 0.1 mg/kg (P = 0.003), compared with vehicle, reduced intake (Figure 2B).

For mice drinking caloric solutions, either an unsweet carbohydrate (maltodextrin) solution or an unsweet fat (corn oil) emulsion, a main effect of Dose (maltodextrin: F_5,35 = 57.14, P < 0.0001; corn oil: F_5,35 = 78.43, P < 0.0001) was found; semaglutide at all doses (P < 0.001), compared with vehicle, reduced intake (Figure 2, C and D).

Chow and water intake were examined in mice that were previously drinking unsweet alcohol. For chow intake, a main effect of Dose (F_5,70 = 36.7, P < 0.0001) was found; semaglutide at all doses (P < 0.0001) except 0.001 mg/kg, compared with vehicle, reduced chow intake. For water intake, a main effect of Dose (F_5,70 = 23.91, P < 0.0001) was found; semaglutide at all doses (P < 0.001), compared with vehicle, reduced water intake (Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.170671DS1↗).

Saline-treated mice were tested on the rotarod to determine whether semaglutide per se affects motor coordination (saline condition; Supplemental Figure 1A). Although a significant Dose effect (F_2,84 = 10.96, P < 0.0001; 0.01 mg/kg < 0 and 0.1 mg/kg) was found, semaglutide did not change motor coordination, compared with baseline (i.e., no Dose × Time interaction). The main effect of Time was not significant.

We also evaluated the effects of semaglutide on alcohol-induced ataxia (Supplemental Figure 1B). A main effect of Time (F_5,195 = 187.0, P < 0.0001) was found, indicating that alcohol induced motor incoordination, and this effect ameliorated over time. However, no Dose or Dose × Time interaction was shown, indicating that semaglutide did not influence alcohol-induced ataxia.

Blood was collected 30 minutes and 90 minutes after alcohol injection, immediately following the rotarod testing, to measure BALs (Supplemental Figure 1C). A main effect of Time (F_1,39 = 231.9, P < 0.0001) was found, indicating that BALs were lower at 90 than 30 minutes. Although the Dose × Time interaction was significant (F_2,39 = 94.5, P = 0.02), post hoc comparison did not show any differences. The main effect of Dose was not significant.

Effects of semaglutide on spontaneous locomotion in mice were evaluated by measuring the distance traveled in the circular corridor test (Supplemental Figure 1D). A main effect of Dose (F_2,14 = 37.37, P < 0.0001) was found; semaglutide at 0.1 mg/kg (P < 0.0001), compared with vehicle, decreased locomotion.

In nondependent rats, a main effect of Dose (F_3,54 = 57.11, P < 0.0001), but no effect of Sex or Dose × Sex interaction, was found for alcohol binge-like drinking. Compared with vehicle, semaglutide at all doses (0.001 mg/kg, P < 0.01; 0.01 mg/kg, P < 0.0001; 0.1 mg/kg, P < 0.0001) reduced self-administration of the sweet alcohol solution (Figure 3A). For water self-administration, a main effect of Dose (F_3,54 = 3.95, P = 0.01; post hoc comparisons did not indicate significant differences) and Sex (F_1,18 = 9.33, P = 0.007; female > male), but no Dose × Sex interaction, was found (Figure 3B).

In alcohol-dependent rats, a main effect of Dose (F_3,60 = 11.24, P < 0.0001), but no effect of Sex or Dose × Sex interaction, was found for dependence-induced drinking. Compared with vehicle, semaglutide at 0.1 mg/kg (P = 0.0007) reduced self-administration of the unsweet alcohol solution (Figure 3C). For water self-administration, a main effect of Sex (F_1,20 = 6.91, P = 0.01; male > female), but no effect of Dose or Dose × Sex interaction, was found (Figure 3D).

The distance traveled in the open field was not significantly different under semaglutide (14.26 ± 7.9 m) and vehicle (15.26 ± 2.0 m).

As shown in Supplemental Figure 2, A–C, and Supplemental Table 2A, and in line with our previous work (31–33), alcohol vapor exposure significantly elevated GABA_A receptor–mediated neurotransmission in the medial subdivision of the CeA, as indicated by significantly increased spontaneous inhibitory postsynaptic current (sIPSC) frequencies (t = 2.94, df = 31, P = 0.006) in alcohol-dependent, compared with alcohol-naive, rats. Other sIPSC characteristics such as amplitudes, rise time, and decay time did not differ between the 2 groups.

As shown in Supplemental Figure 2, D–F, and Supplemental Table 2B, alcohol vapor exposure also significantly elevated GABA_A receptor–mediated neurotransmission in pyramidal neurons located in layer 5 of the ILC. Specifically, increased frequencies (t = 2.08, df = 24, P = 0.04) and amplitudes (t = 3.19, df = 24, P = 0.003) of sIPSCs onto ILC neurons were found in alcohol-dependent, compared with alcohol-naive, rats. The sIPSC kinetics (i.e., current rise and decay times) did not differ between the 2 groups. These data indicate that alcohol vapor exposure induces ILC neuroadaptations at both pre- and postsynaptic sites.

In alcohol-naive rats, acute application of semaglutide significantly increased sIPSC frequency in CeA neurons (130.7% ± 9.2%; t = 3.33, df = 9, P = 0.008), without affecting postsynaptic measures (amplitudes, rise time, or decay time), suggesting enhanced GABA release. In contrast, in alcohol-dependent rats, semaglutide overall did not alter any sIPSC parameter. Of note, semaglutide increased GABA release in a subset of CeA neurons and decreased it in another subset (Figure 4).

In alcohol-naive rats, acute application of semaglutide significantly increased sIPSC frequency in ILC neurons (140.1% ± 11.2%, t = 3.56, df = 8, P = 0.007), without affecting postsynaptic measures (amplitudes, rise time, or decay time), suggesting enhanced GABA release. In contrast, in alcohol-dependent rats, semaglutide overall did not alter any sIPSC parameter. Similar to the CeA, semaglutide increased GABA release in a subset of ILC neurons and decreased it in another subset (Figure 5).

Adult, male (n = 40) and female (n = 37) C57BL/6J mice were obtained from The Jackson Laboratory and weighed between 15 and 25 g upon arrival. Adult, male (n = 21) and female (n = 21) Wistar rats were obtained from Charles River Laboratories and weighed between 180 and 360 g at the start of behavioral experiment. Adult, male (n = 18) Wistar rats used for electrophysiology studies were bred at The Scripps Research Institute and weighed between 380 and 700 g. Mice and rats were single and group housed, respectively, in standard cages and in separate temperature- and humidity-controlled rooms with a reverse 12-hour/12-hour light/dark cycle (22°C ± 2°C, 50%–60%, lights on at 7 p.m.). All behavioral tests were conducted during the dark cycle. Food and water were available ad libitum except during behavioral testing. Animals were habituated to the animal facilities for at least 1 week prior to starting the experiments.

For behavioral testing, semaglutide (Peptide International) was prepared using 1.25% (v/v) dimethyl sulfoxide (DMSO; Thermo Fisher Scientific) and 1.25% (v/v) Tween 80 (Thermo Fisher Scientific), and diluted with 0.9% saline (Hospira). Following a within-subjects, Latin-square design, semaglutide (0.001, 0.003, 0.01, 0.03, 0.1 mg/kg in mice; 0.001, 0.01, 0.1 mg/kg in rats), and vehicle were administered s.c. The volume of injection was 10 mL/kg in mice and 1 mL/kg in rats. The alcohol solution used for systemic injections in the rotarod experiments in mice was prepared with 200-proof ethanol (Pharmco) in 0.9% saline to produce a 20% (v/v) alcohol solution. This solution was administered i.p. at a dose of 2.0 g/kg. For electrophysiology studies, stock solutions of semaglutide (BOC Sciences), CGP55845A (Tocris), 6,7-dinitroquinoxaline-2,3-dione (DNQX; Tocris), and DL-2-amino-5-phosphonovalerate (DL-AP5; Tocris) were prepared in either distilled water or DMSO, aliquoted, frozen, and added to the bath solution.

All drinking solutions were prepared using tap water and 190-proof ethanol (The Warner-Graham Company). Mice with access to alcohol were given either a sweet (20% v/v ethanol, 3% w/v glucose, 0.1% w/v saccharin) or unsweet (20% v/v ethanol) alcohol solution. Mice given access to drinking solutions not containing alcohol received either a sweet caloric (0.3% w/v glucose, 0.01% w/v saccharin) solution, a sweet noncaloric (0.1% w/v saccharin) solution, an unsweet carbohydrate (28% w/v maltodextrin) solution, an unsweet fat (12.5% w/v corn oil, 0.5% v/v Tween 80) emulsion, or tap water. The calorie content of the maltodextrin solution and corn oil emulsion approximates that of 20% v/v ethanol. For operant self-administration, nondependent rats were given access to a sweet alcohol solution (10% w/v ethanol, 3% w/v glucose, 0.1% w/v saccharin), and alcohol-dependent rats were given access to an unsweet alcohol solution (10% w/v ethanol).

A drinking-in-the-dark (DID) test was used to model binge-like drinking in mice (119, 120). Initially, a 4-day protocol was used in which mice had access to drinking solutions for 2 hours for the first 3 days, and for 4 hours on the fourth day. We adhered to this schedule for 3 weeks before switching to a modified 2-day DID procedure. Here, mice received a 2-hour session for 1 day and a 4-hour session the next day. After a day off, a second round of 2-day DID was conducted in the same week. The effects of semaglutide (2 doses per week) were evaluated during the 4-hour test sessions. Semaglutide was administered (s.c.) 30 minutes before mice were given access to the drinking solutions, 3 hours into the dark phase (119). During all DID sessions, food and water were removed from the home cages. Mice with access to only tap water during the DID session were water deprived immediately after the 2-hour DID session and were given access again during the 4-hour DID session the next day. The volume/calories consumed were calculated from weight change of drinking bottles, which were weighed at 0, 2, and/or 4 hours during a DID session.

The effects of semaglutide on chow and water intake were evaluated in mice that were previously drinking unsweet alcohol. Semaglutide or vehicle was administered 3 hours into the dark phase, and food and water were measured 24 hours after treatment.

The effects of semaglutide on motor coordination were evaluated using an accelerating rotarod test in mice (120, 121). Mice that were previously drinking unsweet alcohol were placed on the rotarod apparatus (Rotamex-5, Columbus Instruments) and habituated for 1 minute with the rod rotating at a constant speed of 4 rpm. During training and test trials, mice were placed on the rod set at 4 rpm with a constant acceleration rate of 8 rpm/min up to a maximum of 40 rpm. The latency to fall was automatically recorded by photocell beams, with a maximum cutoff latency of 5 minutes. Immediately following habituation, mice received 5 consecutive training trials, separated by 5-minute rest intervals, and were given a minimum resting period of 24 hours prior to test trials. On testing days, mice were given 2 baseline trials separated by 5-minute rest intervals.

To test the effects of semaglutide per se on motor coordination (saline condition), mice were administered vehicle or semaglutide (0.01, 0.1 mg/kg; s.c.). Thirty minutes later, they were injected with saline (10 mL/kg; i.p.) and were tested on the rotarod 30, 60, and 90 minutes after saline injection. To test the effects of semaglutide on alcohol-induced ataxia (alcohol condition), mice were injected vehicle or semaglutide (0.01, 0.1 mg/kg; s.c.), followed by 2 g/kg alcohol 30 minutes later and were then tested on the rotarod 15, 30, 60, 90, and 120 minutes after alcohol injection. For both saline and alcohol conditions, we used a within-subjects, Latin-square design, with each testing occurring at least 24 hours apart. Blood was collected via the submandibular vein immediately after the 30- and 90-minute test trials to measure BALs, using an Analox Alcohol Analyzer (Analox Technologies North America).

A circular corridor test (120) was used to evaluate the effects of semaglutide on spontaneous locomotion in mice that were given access to tap water during DID. The circular corridor apparatus (Thermal Gradient Ring, Ugo Basile) was at room temperature (22°C) throughout the experiment. Mice were first allowed to explore the apparatus freely for 20 minutes to habituate and were then given a 24-hour minimum rest period. On test days, mice were administered vehicle or semaglutide (0.01, 0.1 mg/kg; s.c.) in a within-subjects, Latin-square design and returned to their home-cages for 3 hours. Mice were then placed in the circular corridor for a 20-minute test session. AnyMaze Video Tracking Software (Stoelting) was used to track the total distance traveled by each mouse.

Sweet and unsweet alcohol solutions were used for operant self-administration in rats (54, 122). To model alcohol binge-like drinking, rats were trained to self-administer a sweet alcohol (10% v/v ethanol, 3% w/v glucose, 0.1% w/v saccharin) solution and water under a free-choice, fixed ratio 1 (FR1) schedule of reinforcement in standard operant conditioning chambers (28 × 26 × 20 cm; Med Associates) (122, 123). Alcohol-dependent rats were trained similarly except that they received unsweet alcohol (10% v/v ethanol) and water (54, 120). Each operant response to the alcohol- or water-associated lever was reinforced with the delivery of 0.1 mL of fluid. Following operant responses to alcohol, a cue light located above the alcohol-associated lever was illuminated for the duration of the alcohol solution delivery (2 seconds). During this time, additional lever presses did not lead to another fluid delivery. No cue light was associated with the delivery of water. After about 16 training sessions, rats underwent 30-minute FR1 self-administration sessions to evaluate the effects of semaglutide. Semaglutide (0.001, 0.01, 0.1 mg/kg; s.c.) or vehicle was administered 3 hours prior to each self-administration session, following a within-subjects, Latin-square design.

Rats that were trained on unsweet alcohol operant self-administration (described above) were made dependent on alcohol by chronic, intermittent alcohol vapor exposure (54, 120, 124). Briefly, rats were exposed to 14 hours of alcohol vapor per day, followed by 10 hours of room air (withdrawal). The target BAL for the rats during alcohol vapor exposure was between 150 and 250 mg/dL. Rats underwent behavioral testing during acute spontaneous withdrawal (i.e., 6–8 hours after vapor turned off). Nondependent rats were exposed to air without alcohol and were tested at the same time as the dependent rats. Semaglutide (0, 0.001, 0.01, and 0.1 mg/kg; i.p.) was administered 3 hours prior to a drinking session.

Rats used for the electrophysiology experiments were also made dependent on alcohol following an alcohol vapor protocol over 5–7 weeks (31, 32, 101). BALs were measured 1–2 times per week (average BAL = 193 ± 27 mg/dL), and air-exposed rats were used as controls (alcohol-naive).

The effects of semaglutide on spontaneous locomotion in rats were assessed using an open-field test. Alcohol-dependent rats were first habituated to the apparatus (40 × 40 cm) for 15 minutes. On testing days, rats were administered with semaglutide (0.1 mg/kg; s.c.) or vehicle, in a randomized order and, 3 hours later, were placed in the center of the open field and allowed free access for 15 minutes. The open field tests were separated by at least 3 days and conducted under red light. AnyMaze Video Tracking software (Stoelting) was used to track the total distance traveled by each rat.

Preparation of brain slices and electrophysiological recordings were performed as previously described (31–33). Briefly, deeply anesthetized rats (3%–5% isoflurane anesthesia) were rapidly decapitated, and their brains were isolated in an ice-cold, oxygenated, high-sucrose cutting solution (composition in mM: 206 sucrose, 2.5 KCl, 0.5 CaCl₂, 7 MgCl₂, 1.2 NaH₂PO₄, 26 NaHCO₃, 5 glucose, and 5 HEPES; Sigma-Aldrich). We then divided the brains with a coronal cut roughly at bregma to enable cutting acute brain slices from 2 different regions at the same time. Specifically, we cut coronal slices containing the medial subdivision of the CeA (300 μM; using a Leica VT 1000S vibratome) and coronal slices containing the ILC (300 μm; using a Leica VT1200 vibratome), which were then incubated for 30 minutes in 37°C warm, oxygenated artificial cerebrospinal fluid (aCSF) (composition in mM: 130 NaCl, 3.5 KCl, 2 CaCl₂, 1.25 NaH₂PO₄, 1.5 MgSO₄, 24 NaHCO₃, and 10 glucose; Sigma-Aldrich), followed by another 30-minute incubation at room temperature. Dependent rats were euthanized within the last hour of their daily alcohol vapor exposure. We did not add ethanol to any of the solutions used for preparation and incubation of brain slices; thus, the slices underwent acute in vitro withdrawal, as previously shown (31, 32, 101).

Using whole-cell patch clamp, we recorded pharmacologically isolated GABA_A receptor–mediated sIPSCs from 33 CeA and 26 ILC neurons held at –60 mV by adding 20 μM DNQX (to block AMPA and kainate receptors), 30 μM AP-5 (to block NMDA receptors), and 1 μM CGP55845A (to block presynaptic GABA_B receptors) to the bath solution (31–33, 109–111). Neurons were visualized with infrared differential interference contrast optics, using either 40× or 60× water-immersion objectives (Olympus BX51WI, Olympus Scientific Solutions), and CCD cameras (EXi Aqua, QImaging Corporation). We did not select a specific neuronal cell type in the CeA (125), while we recorded only from pyramidal neurons in layer 5 of the ILC (capacitance > 70 pF). All recordings were performed in gap-free acquisition mode with a 20 kHz sampling rate and 10 kHz low-pass filtering, using a MultiClamp700B amplifier, Digidata 1440A, and pClamp 10 software (Molecular Devices). Patch pipettes were pulled from borosilicate glass (3–5 mΩ, King Precision Glass) and filled with a KCl-based internal solution (composition in mM: 145 KCl, 5 EGTA, 5 MgCl₂, 10 HEPES, 2 Mg-ATP, and 0.2 Na-GTP; pH 7.2–7.4 adjusted with 1M NaOH, 295–315 mOsms; Sigma-Aldrich). We only recorded from neurons with an access resistance (Ra) < 15 MΩ, which changed less than < 20% during the recording, as monitored by frequent 10 mV pulses. Semaglutide (100 nM) (126) was applied by bath perfusion.

The DID data of mice drinking sweet alcohol, unsweet alcohol, or the sweet caloric solution not containing alcohol as well as the operant self-administration data of rats (binge-like and dependence-induced drinking) were analyzed using 2-way repeated-measures ANOVA with Dose as the within-subjects factor and Sex as the between-subjects factor. Since we did not find interactions between Dose and Sex, all other behavioral data were analyzed combining males and females. Thus, the DID data of mice drinking tap water, the saccharin solution (sweet, noncaloric), the maltodextrin solution (unsweet, caloric), or the corn oil emulsion (unsweet, caloric); chow and water intake data; and spontaneous locomotion data (total distance traveled on the circular corridor and in the open field by mice and rats, respectively) were analyzed using 1-way repeated-measures ANOVA or paired 2-tailed Student’s t test, with Dose as the within-subjects factor. The rotarod and BAL data of mice were analyzed using 2-way repeated-measures ANOVA with Dose and Time as within-subjects factors; saline condition and alcohol condition were analyzed separately. When appropriate, Dunnett’s, Tukey’s, or Duncan’s tests were used for post hoc comparisons.

The electrophysiology data were obtained from 59 individual neurons from 18 different rats. The frequency, amplitude, rise, and decay time of sIPSCs were analyzed semiautomatically using MiniAnalysis software (Synaptosoft Inc.). Each event was visually confirmed, and sIPSCs with amplitudes < 5 pA were excluded. We combined events from 3-minute bins to obtain averaged sIPSC characteristics. To account for cell-to-cell variability, we normalized all relevant parameters (frequency, amplitude, rise, and decay time) to baseline control conditions and pooled data before group analyses. We used 1-sample t tests to examine changes from baseline control conditions. Unpaired Student’s t tests were then used to compare semaglutide effects on sIPSC characteristic between alcohol-naive and alcohol-dependent groups.

All data are represented as mean ± SEM. A P value less than 0.05 (2-tailed) was considered significant. Prism 8 (GraphPad Prism) and Statistica 13 (TIBCO Software) were used for the analyses.

All procedures were performed according to the Guide for the Care and Use of Laboratory Animals (National Academies Press, 2011) and were approved by the IACUC of the NIDA IRP or The Scripps Research Institute.

The data are available from the corresponding authors upon reasonable request.

In-Press Preview

Electronic publication

The data are available from the corresponding authors upon reasonable request.