In contrast to the common view of GLP‐1 in controlling food intake and glucose homeostasis, recent studies pinpoint GLP‐1 as a reward regulator (for review see, Engel & Jerlhag 2014 ). This was initially reported in a study showing that acute and peripheral administration of the GLP‐1 receptor agonist exendin‐4 (Ex4) blocks the alcohol‐induced conditioned place preference (CPP), locomotor stimulation and accumbal dopamine release in mice (Egecioglu et al. 2013c ). Moreover, Ex4 treatment decreased alcohol intake in the intermittent access model as well as alcohol‐seeking behavior, using the progressive ratio schedule of reinforcement, in rats (Egecioglu et al. 2013c ). In accordance are the data showing that GLP‐1 as well as Ex4 decreases, whereas GLP‐1 receptor antagonist increases alcohol intake in rats (Shirazi et al. 2013 ). Ex4 attenuates alcohol‐induced CPP, and GLP‐1‐controlled alcohol intake involves the VTA in rats (Shirazi et al. 2013 ). Opposed to Ex4, which is metabolized rapidly and extensively (Deacon et al. 1995 ), the clinically available GLP‐1 receptor agonist, liraglutide, has protracted and maintained biological activity (for review see, Holst 2004 ). The present series of experiments were undertaken to investigate the effect of liraglutide on several alcohol‐related behaviors in rats that model different aspects of alcohol use disorder (AUD) in humans. Initially, we investigated the effects of acute liraglutide treatment on the rewarding properties of alcohol, as measured by accumbal dopamine release and CPP as well as on blood alcohol concentrations in mice. Thereafter, the ability of acute treatment of liraglutide to influence alcohol intake and relapse‐like drinking in outbred rats were explored. Finally, the effects of repeated liraglutide treatment on alcohol intake in outbred rats as well as on operant and oral self‐administration of alcohol in selectively bred Sardinian alcohol‐preferring (sP) rats were studied. The present experiments, by means of these preclinical models, could therefore elucidate the possibility to use liraglutide as treatment of AUD in humans.

The incretin hormone, glucagon‐like peptide 1 (GLP‐1), is released into the circulation from the distal small intestine in response to food intake (Brubaker & Anini 2003 ) and acts thereby as a signal for meal termination (Pannacciulli et al. 2007 ). The findings that its biological effects include inhibition of gastric emptying (Flint et al. 2001 ), glucose‐dependent stimulation of insulin secretion (Kreymann et al. 1987 ) and suppression of glucagon release (Orskov et al. 1988 ) lead to the approval of GLP‐1 receptor agonists for treatment of type II diabetes(for review, see Holst 2004 ). Experimental evidence from animals (Tang‐Christensen et al. 1996 ), healthy subjects (Flint et al. 1998 ) and patients with type II diabetes (Gutzwiller et al. 1999 ) shows that circulating GLP‐1 reduces food intake and promotes satiety. Along these lines of investigations, GLP‐1 receptor agonists were found to reduce body weight when administered subcutaneously in humans (Zander et al. 2002 ) and intracerebroventricularly in animals (Meeran et al. 1999 ). GLP‐1 regulates food intake via GLP‐1 receptors in the nucleus tractus solitarius (NTS), amygdala and the hypothalamus (Tang‐Christensen et al. 1996 ; McMahon & Wellman 1998 ; Hayes et al. 2009 ). In addition to areas regulating homeostatic feeding, the expression of GLP‐1 receptors is found in reward‐related areas including the ventral tegmental area (VTA) and nucleus accumbens (NAc) (Alvarez et al. 1996 ; Merchenthaler et al. 1999 ). Taken together with the findings that both the VTA and NAc are targeted by GLP‐1‐producing neurons from the NTS (Alvarez et al. 1996 ; Merchenthaler et al. 1999 ), this provides a pathway through which GLP‐1 may regulate reinforcement.

Test sessions started immediately after termination of the 20‐day maintenance phase. In test sessions, response requirement on the alcohol and water lever was kept at FR4 and FR1, respectively. Test sessions lasted for 30 minute and were conducted for 5 consecutive days. Vehicle or liraglutide, at the doses of 0.05, and 0.1 mg/kg, was administered sc to independent groups of n = 9 rats. After completion of the treatment phase, rats were exposed to four additional daily self‐administration sessions (posttreatment phase). Sessions of the treatment and posttreatment phases were conducted with no weekend interruption.

Following acute sc treatment of 0.1 mg/kg liraglutide, these outbred Wistar rats ( n = 15) were subjected to an additional 7 days of liraglutide (0.1 mg/kg, sc) ( n = 7) or vehicle ( n = 8) treatment. The effect of repeated administration of liraglutide on the daily alcohol, water, total fluid and food intake was investigated at test session 1 (Monday), test session 2 (Wednesday) and at test session 3 (Friday), which correspond to alcohol drinking days in the intermittent access model. In addition to the alcohol consumption days, the rats received treatment on water consumption days, i.e. Saturday, Sunday, Tuesday and Thursday. All injections were given 60 minutes before the rats were given access to alcohol and water. In addition, water intake was measured the three following days after discontinuation of the liraglutide treatment.

The alcohol deprivation model is based on the observation that voluntary alcohol intake will increase temporarily when compared with baseline drinking conditions following forced abstinence in alcohol‐experienced rats (Spanagel 2000 ). Rats ( n = 11) were subjected to the intermittent access 20 percent alcohol two‐bottle‐choice drinking paradigm (as described earlier) for 10 weeks, and a stable baseline alcohol intake (g/kg/day) was obtained. Alcohol, water and food intake was measured at 1 and 24 hours during baseline as well as following drug treatment. The rats were deprived of alcohol for 10 days, and alcohol was thereafter reintroduced. Sixty minutes before the reintroduction of alcohol, the rats were treated with either liraglutide (0.1 mg/kg, sc, n = 5) or vehicle ( n = 6) in a balanced design. Thereafter, bottles and food were weighed at 24 hours after the fluids were presented.

In the first drinking experiment, the effects of acute administration of liraglutide (0.1 mg/kg, sc, n = 7) or vehicle ( n = 8) on alcohol, water and food intake as well as body weight in outbred Wistar rats ( n = 15) that had voluntarily consumed 20 percent alcohol for 12 weeks were investigated. The injection was given 60 minutes before the rats were given access to alcohol and water, and the intakes were measured 24 hours following treatment. Thereafter, these rats were repeatedly administered with liraglutide (0.1 mg/kg, sc) or vehicle for another 7 days, allowing us to investigate the effects of repeated liraglutide treatment on the daily alcohol intake (in the succeeding text).

Ten to twelve weeks of intermittent access 20 percent alcohol two‐bottle‐choice drinking paradigm induces voluntary intake of high amounts of alcohol (Wise 1973 ; Simms et al. 2008 ; Loi et al. 2010 ) and pharmacological relevant blood alcohol concentrations (Simms et al. 2008 ; Carnicella et al. 2009 ) in rodents. In addition, the expression of ghrelin receptors, which mediate alcohol reinforcement, is increased in reward‐related areas in high‐alcohol‐consuming compared with low‐alcohol‐consuming rats following 10 weeks of voluntary alcohol intake (Landgren et al. 2011 ). In brief, the rats were given free access to one bottle of 20 percent alcohol and one bottle of water during three 24‐hour sessions per week (Mondays, Wednesdays and Fridays). The rats had unlimited access to two bottles of water between the alcohol‐access periods. Bottles were weighed at 24 hours after the fluids were presented to the rats. The body weight of each rat was measured daily prior to bottle presentation, to allow for calculating the grams of alcohol intake per kilogram of body weight (g/kg). The preference for alcohol over water (the ratio of alcohol to total fluid intake) was calculated at all time points. In addition, water and food intake was measured. Three separate drinking experiments in outbred rats were conducted after a period of 10–12 weeks of intermittent access to alcohol.

The second experiment in separate mice was designed to investigate the effect of liraglutide on memory consolidation of a reward. In this experiment, mice during preconditioning were injected sc with vehicle and were placed in the chamber with free access to both compartments during 20 minutes to determine the initial place preference. Conditioning (20 minutes per session) was performed using a biased procedure in which alcohol (1.75 g/kg, ip) was paired with the least preferred compartment and vehicle with the preferred compartment. All mice received one alcohol and one vehicle injection every day, and the injections were altered between morning and afternoon in a balanced design. At post‐conditioning, mice were injected with liraglutide (0.1 mg/kg, sc) or an equal volume of vehicle solution and 60 minutes later, placed on the midline between the two compartments with free access to both compartments for 20 minutes (creating the following treatment groups: alc–veh and alc–lir).

The first CPP test was designed to investigate the effect of liraglutide on the rewarding properties of alcohol. Mice at preconditioning were placed in the chamber with free access to both compartments during 20 minutes to determine the initial place preference. Conditioning (20 minutes per session) was performed using a biased procedure in which alcohol (1.75 g/kg, ip) was paired with the least preferred compartment and vehicle with the preferred compartment. In these experiments, liraglutide (0.1 mg/kg, sc) or vehicle was administered 60 minutes prior to the alcohol injection on each of the four conditioning days (creating the following treatment groups: veh–alc or lir–alc). At post‐conditioning, the mice were untreated and were placed on the midline between the two compartments with free access to both compartments for 20 minutes. In a control experiment for liraglutide, separate mice were subjected to the same procedure but received vehicle injections instead of alcohol throughout the conditioning (non‐alcohol‐conditioned control group; creating the following treatment groups: vehicle–vehicle and liraglutide–vehicle).

Dopamine was separated and quantified using two different high‐performance liquid chromatography with electrochemical detection as described previously (Clarke et al. 2014 ). In brief, a pump (UltiMate 3000 Pump, Thermo Scientific, Darmstadt, Germany), an ion exchange column (Nucleosil SA, 2.0 × 150 mm, 5 µm diameter, pore size 100 Å; Phenomenex Scandinavia, Västra Frölunda, Sweden) and a detector (Decade, Kovalent AB, Sweden) operated at 400 mV versus the cell were used. The mobile phase was delivered at 0.3 ml/min and consists of 58 mM citric acid, 135 mM NaOH, 0.107 mM Na2–EDTA and 20 percent methanol. The second system consists of a pump (UltiMate 3000 Pump, Thermo Scientific), a reversed‐phase column (2.0 × 50 mm, 3 µm diameter; pore size 100 Å; Phenomenex Scandinavia) and a detector (Dionex, Västra Frölunda, Sweden) operated at 220 mV versus the cell. The mobile phase was delivered at 0.3 ml/min and consists of 150 mM NaH2PO4, 4.76 mM citric acid, 3 mM sodium dodecyl sulfate, 50 μM EDTA, as well as 10 percent MeOH and 15 percent acetonitrile.

The effect of systemic administration of liraglutide [0.1 mg/kg, subcutaneous (sc)] on alcohol‐induced (1.75 g/kg, ip) accumbal dopamine release was investigated using microdialysis in freely moving mice. On the day of the experiment, the probe was connected to a microperfusion pump (U‐864 Syringe Pump, AgnThós AB, Lidingö, Sweden) and perfused with Ringer solution at a rate of 1.5 µl/minute. After 1 hour of habituation to the microdialysis setup, perfusion samples were collected every 20 minutes. The baseline dopamine level was defined as the average of three consecutive samples before the first alcohol or vehicle (saline, ip) challenge (time 0). This initial alcohol‐challenge was given to establish that all mice included in the experiment would respond with an alcohol‐induced release of accumbal dopamine. The challenge‐induced increase in accumbal dopamine was calculated as the percent increase from baseline. Seven consecutive 20‐minute samples were collected after the initial challenge. At 140 minutes, the mice were injected with liraglutide (0.1 mg/kg, sc) or vehicle (second challenge), and 60 minutes later, vehicle or a second injection of alcohol (1.75 g/kg, ip) was administered (third challenge; 200 minutes) and followed by collection of four 20‐minute samples (experiment terminated at 280 min). Collectively, the following treatment groups ( n = 12 in each group) were created: alcohol–vehicle–alcohol (veh–alc), alcohol–liraglutide–alcohol (lir–alc) and vehicle–liraglutide–vehicle (lir–veh). [Correction added on 28 September 2015 after the first online publication: the word ‘in‐veh’ has been changed to ‘lir‐veh’.]

For measurements of extracellular dopamine levels, mice were implanted with a microdialysis probe positioned in the NAc shell as described previously (Jerlhag et al. 2006 ). In brief, mice were anesthetized with isoflurane (Isoflurane Baxter, Univentor 400 Anaesthesia Unit, Univentor Ltd., Zejtun, Malta), placed in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA), and one hole for the probe and one for the anchoring screw were drilled. Xylocaine Adrenalin (5 µg/ml; Pfizer Inic, New York, NY, USA) was used as local anesthetics and carprofen (Rimadyl ® ) (Astra Zeneca, Gothenburg, Sweden) at a dose of 5 mg/kg intraperitoneal (ip) to relieve pain. The probe was randomly alternated to either the left or right side of the brain. The coordinates of 1.4 mm anterior to the bregma, ±0.6 lateral to the midline and 4.7 mm below the surface of the brain surface were used (Franklin & Paxinos 1997 ).

An overall effect of treatment [ F (2, 24) = 9.67, P = 0.0008], of time [ F (4, 96) = 4.95, P = 0.0011] and of treatment × time interaction [ F (8, 96) = 2.26, P = 0.0293] on daily changes in rat body weight during the 5‐day treatment phase was observed. Post hoc analysis indicated that treatment with both doses of liraglutide resulted in an immediate and relatively stable reduction in body weight of 15–20 g in comparison with saline‐treated rats (Fig. 7 b). Daily changes in rat body weight were similar among the three rat groups during the 4‐day posttreatment phase [treatment: F (2, 24) = 0.82, P = 0.4522; time: F (3, 72) = 7.76, P < =0.0002; treatment × time interaction: F (6, 72) = 1.42, P = 0.2198] (Fig. 7 b).

Repeated treatment of low doses of liraglutide reduces food intake and body weight in alcohol‐preferring sP rats. (a) Repeated treatment with 0.05 and 0.1 mg/kg liraglutide reduced daily food intake during the 5‐day treatment phase in sP rats. Daily food intake was identical among the three rat groups during the 4‐day posttreatment phase. (b) Repeated treatment with both doses of liraglutide resulted in an immediate and relatively stable reduction in body weight in comparison with vehicle‐treated rats. Daily changes in rat body weight were similar among the three rat groups during the 4‐day posttreatment phase

An overall effect of treatment [ F (2, 24) = 3.30, P = 0.0587], of time [ F (4, 96) = 7.03, P < 0.0001] and of treatment × time interaction [ F (8, 96) = 3.09, P < 0.0039] on daily food intake in the home cage during the 5‐day treatment phase was obtained. Post hoc analysis indicated that treatment with 0.05 and 0.1 mg/kg liraglutide reduced daily food intake on day 1 and on days 1 and 2, respectively (Fig. 7 a). On continuing treatment, the magnitude of the anorectic effect of both doses of liraglutide tended to decrease (Fig. 7 a). Daily food intake was virtually identical among the three rat groups during the 4‐day posttreatment phase [treatment: F (2, 24) = 0.08, P = 0.9194; time: F (3, 72) = 40.47, P < 0.0001; treatment × time interaction: F (6, 72) = 0.40, P = 0.8774] (Fig. 7 a).

An overall effect of treatment [ F (2, 24) = 6.52, P = 0.0055] but not of time [ F (3, 72) = 1.60, P = 0.1964], and no significant treatment × time interaction [ F (6, 72) = 0.59, P = 0.7347], on number of lever responses for alcohol during the 4‐day posttreatment phase was observed (Fig. 6 a). In addition, an overall effect of treatment [ F (2, 24) = 4.56, P = 0.0209] but not of time [ F (3, 72) = 1.42, P = 0.2441], and no significant treatment × time interaction [ F (6, 72) = 0.53, P = 0.7866], on amount of self‐administered alcohol during the 4‐day posttreatment phase was obtained (Fig. 6 b). After treatment discontinuation, number of lever responses for alcohol (Fig. 6 a) and amount of self‐administered alcohol (Fig. 6 b) in the rat group treated with 0.1 mg/kg liraglutide remained reduced, in comparison with control values, during the first 2–3 days, subsequently tending to return to control values.

Repeated treatment of low doses of liraglutide reduces operant self‐administration of alcohol and alcohol intake in alcohol‐preferring sP rats. Repeated treatment of two doses of liraglutide (0.1 or 0.05 mg/kg) reduced (a) the number of lever responses for alcohol as well as (b) the amount of self‐administered alcohol during the 5‐day treatment phase in sP rats. After treatment discontinuation, the number of lever responses for alcohol and amount of self‐administered alcohol were lower in the rat group treated with 0.1 mg/kg liraglutide in comparison with control values. All values represent mean ± SEM (* P < 0.05)

There was an overall effect of treatment [ F (2, 24) = 5.47, P = 0.0110], of time [ F (4, 96) = 3.15, P = 0.0176] and of treatment × time interaction [ F (8, 96) = 3.24, P = 0.0026] on the number of lever responses for alcohol during the 5‐day treatment phase. There was also an overall effect of treatment [ F (2, 24) = 3.46, P = 0.0479], of time [ F (4, 96) = 3.92, P = 0.0054] and of treatment × time interaction [ F (8, 96) = 4.15, P = 0.0003] on amount of self‐administered alcohol during the 5‐day treatment phase. Post hoc analysis shows that treatment with either 0.1 ( n = 9) or 0.05 ( n = 9) mg/kg of liraglutide was totally ineffective on day 1 on the number of lever responses for alcohol (Fig. 6 a) and the amount of self‐administered alcohol (Fig. 6 b) compared with vehicle ( n = 9). Conversely, on day 2, treatment with both doses of liraglutide produced a similar reduction (40–50 percent in comparison with vehicle‐treated rats) in both variables (Fig. 6 a & 6 b). In the subsequent three daily self‐administration sessions (days 3–5), number of lever responses for alcohol and amount of self‐administered alcohol progressively returned to control values in the rat group treated with 0.05 mg/kg liraglutide, while both variables remained relatively stable and reduced in the rat group treated with 0.1 mg/kg liraglutide (Fig. 6 a & 6 b).

When liraglutide (0.1 mg/kg) was given repeatedly, in total for 8 days from which three are alcohol drinking days, there was an overall main effect of treatment [ F (1, 13) = 5.838, P = 0.031] and of time [ F (2, 26) = 4.642, P = 0.019], and a tendency for the interaction of treatment × time [ F (2, 26) = 2.912, P = 0.072] on alcohol intake. Post hoc test revealed that liraglutide significantly reduced alcohol intake at test session 1 ( P < 0.05) and test session 2 ( P < 0.05), but not at test session 3 (Fig. 5 a). There was an overall effect on water intake following repeated liraglutide treatment [ F (1, 13) = 11.77, P = 0.0045]. There was no overall effect of time [ F (2, 26) = 0.4355, P = 0.6516] nor of treatment × time interaction [ F (2, 26) = 1.555, P = 0.2301]. Post hoc test revealed that liraglutide significantly increased water intake at test session 1 ( P < 0.01) and test session 3 ( P < 0.01), but not at test session 2 (Fig. 5 b). There was an overall effect on total fluid intake of treatment × time interaction [ F (2, 26) = 3.752, P = 0.0370] and tendency of treatment effect [ F (1, 13) = 3.894, P = 0.0701]. There was no overall effect of time [ F (2, 26) = 0.7541, P = 0.4804]. Post hoc test revealed that liraglutide significantly increased total fluid intake at test session 3 (P < 0.05), but not at test session 1 and test session 2 (Fig. 5 c). There was an overall effect on alcohol preference of treatment [ F (1, 13) = 11.40, P = 0.0050]. There was no overall effect of time [ F (2, 26) = 0.1893, P = 0.8286] and of treatment × time interaction [ F (2, 26) = 1.312, P = 0.2865]. Post hoc test revealed that liraglutide significantly decreased alcohol preference at test session 1 ( P < 0.01) and test session 2 ( P < 0.01), but not at test session 3 (Fig. 5 d). There was an overall main effect of treatment [ F (1, 13) = 38.36, P < 0.0001], but not of time [ F (2, 26) = 1.931, P = 0.1652], nor of the treatment × time interaction [ F (2, 26) = 0.9202, P = 0.4110] on food intake. Post hoc test revealed that liraglutide significantly reduced food intake at test session 1 ( P < 0.001), test session 2 ( P < 0.001) and test session 3 ( P < 0.05) (Fig. 5 e). There was an overall main effect of treatment [ F (1, 13) = 5.625, P = 0.0338], but not of time [ F (2, 26) = 0.4649, P = 0.6333], nor of treatment × time interaction [ F (2, 26) = 0.9091, P = 0.4153] on body weight. Post hoc test revealed that liraglutide did not reduce body weight at any treatment day (Fig. 5 f).

After 10 days of forced alcohol abstinence, the rats were treated with liraglutide ( n = 5) or vehicle ( n = 6) 60 minutes before given access to one bottle of water and one bottle of 20 percent alcohol. After 24 hours of alcohol access, there was an overall main effect of treatment [ F (1, 9) = 20.44, P = 0.0014] and an significant interaction of treatment × time [ F (1, 9) = 11.39, P = 0.0082]. However, no effect of time was observed [ F (1, 9) = 1.247, P = 0.2930]. Post hoc analysis revealed a significant alcohol deprivation effect (i.e. significant increase in alcohol intake compared with respective baseline) in vehicle‐treated ( P < 0.05) but not in liraglutide‐treated ( P > 0.05) rats and that the alcohol intake was significantly higher in vehicle‐treated compared with liraglutide‐treated rats ( P < 0.001) (Fig. 4 ). In addition, the liraglutide treatment significantly reduced alcohol intake (g/kg) (vehicle 2.6 ± 0.4; liraglutide 0.6 ± 0.1; P < 0.01) and alcohol preference (vehicle 41 ± 7; liraglutide 12 ± 3; P < 0.01) compared with vehicle. No effect on water (ml) (vehicle 14 ± 5; liraglutide 17 ± 6; P = 0.6180), total fluid (ml) (vehicle 21 ± 4; liraglutide 19 ± 6; P = 0.7530) nor on food (vehicle 13 ± 3; liraglutide 7 ± 2; P = 0.1941) intake was observed.

The effect of acute liraglutide (0.1 mg/kg, sc) treatment on alcohol intake was investigated in a group of rats ( n = 15). After 12 weeks of intermittent alcohol intake, there was no significant difference in 24‐hour baseline consumption of alcohol intake (vehicle 4.0 ± 0.8 g/kg, n = 8; liraglutide 3.4 ± 0.7 g/kg, n = 7; P = 0.5889) between the two groups of rats. Liraglutide treatment significantly reduced alcohol intake (g/kg) compared with vehicle (Fig. 2 a, P < 0.01), alcohol preference (Fig. 2 d, P < 0.05) as well as food intake (Fig. 2 e, P < 0.001). There was a tendency in increased water intake (ml) by liraglutide (Fig. 2 b, P = 0.0936), but there was no effect on total fluid intake (ml) (Fig. 2 c P = 0.6119) or on body weight (g) (Fig. 2 f, P = 0.3901).

During a drug‐free test session (post‐conditioning), alcohol‐induced (1.75 kg/kg) (veh‐alc) CPP was significantly attenuated by concomitant injection of liraglutide (0.1 mg/kg) (lir–alc) on each conditioning day compared with vehicle injection ( P = 0.0065, n = 7 in each group; Fig. 1 b). The control experiment showed that there was no difference between liraglutide–vehicle and vehicle–vehicle treatment on CPP ( P = 0.9788, n = 8 in each group; Fig. 1 c) during the drug‐free test session (post‐conditioning). However, the alcohol‐induced (1.75 kg/kg) (alc–veh, 12 ± 6 percent) CPP was not affected by an acute single injection of liraglutide (0.1 mg/kg) (alc–lir, 14 ± 7 percent) on the post‐conditioning day compared with vehicle injection ( P = 0.7903, n = 8 in each group).

Acute administration of liraglutide attenuates accumbal dopamine release and conditioned place preference (CPP) but does not alter blood alcohol concentrations in mice. (a) Initial injections of alcohol (1.75 g/kg) caused a significant increase in accumbal dopamine release compared with vehicle treatment liraglutide–vehicle (lir–veh) in the alcohol–vehicle–alcohol (veh–alc) and alcohol–liraglutide–alcohol (lir–alc) group. In the subsequent part of the experiment, administration of liraglutide (0.1 mg/kg, at 160 minutes) 60 minutes prior to the second alcohol injection (at 200 minutes) significantly attenuated the alcohol‐induced accumbal dopamine release (lir–alc) compared with vehicle pretreatment (veh–alc). (b) The alcohol‐induced (1.75 g/kg) (veh–alc) CPP was significantly attenuated by concomitant injection of liraglutide (0.1 mg/kg) (lir–alc) on each conditioning day compared with vehicle injection. (c) The control experiment showed that there was no difference between lir–veh (0.1 mg/kg) and vehicle–vehicle treatment on CPP. (d) Compared with vehicle injection, acute treatment of liraglutide (0.1 mg/kg) did not affect the blood alcohol concentrations induced by a peripheral injection of alcohol (1.75 g/kg). Data are presented as mean (g/kg) ± SEM (* P < 0.05, ** P < 0.01, *** P < 0.001, n.s. P > 0.05). [Correction added on 28 September 2015 after the first online publication: the legends of Figure 1a has been updated]

Accumbal microdialysis measurements of dopamine in mice revealed an overall main effect of treatment [ F (16, 187) = 1.98, P = 0.0162], time [ F (32, 374) = 92.35, P < 0.0001] and a significant interaction of treatment × time [ F (32, 374) = 2.236, P = 0.0002]. In the first part of the experiment, the responsiveness to alcohol (1.75 g/kg) per se was investigated (alcohol injection at time point 0 minute). Initial injections of alcohol caused a significant increase in accumbal dopamine release compared with vehicle treatment in both groups treated with alcohol (veh–alc and lir–alc). Specifically, in the veh–alc group, alcohol increased accumbal dopamine at time points 40 ( P < 0.01), 60–100 ( P < 0.001) and 120–140 minutes ( P < 0.01). In addition, alcohol increased NAc dopamine in the lir–alc group at time points 60 and 1000 minutes ( P < 0.01) and 120 minutes ( P < 0.05) (Fig. 1 a). In the subsequent part of the experiment, administration of liraglutide (0.1 mg/kg, at 160 minutes) 60 minutes prior to the second alcohol injection (1.75 g/kg, at 200 minutes) significantly attenuated the alcohol‐induced accumbal dopamine release (alc–lir–alc) compared with vehicle pretreatment (alc–veh–alc) at time points 200 ( P < 0.05) and 240–280 minutes ( P < 0.01) (Fig. 1 a).

Discussion

Alcohol use disorder is a major healthcare challenge, causing an enormous cost to society, and novel treatment strategies are warranted. The present study provides evidence that the GLP‐1 receptor agonist liraglutide have an important role in regulating alcohol‐mediated behaviors in rodents, suggesting that this clinically available medication could be used for treatment of AUD in humans. Indeed, we found that liraglutide suppresses the well‐documented effects of alcohol on the mesolimbic dopamine system, namely alcohol‐induced accumbal dopamine release and CPP in mice (Engel et al. 1988; Spanagel 2000). Acute liraglutide treatment reduced alcohol intake and prevented the alcohol deprivation effect in outbred rats. In addition, repeated treatments of liraglutide decreased alcohol intake in outbred rats as well as reduced the operant self‐administration of alcohol in alcohol‐preferring sP rats.

Given that liraglutide was administered peripherally, the mechanisms through which the GLP‐1 receptor ligand attenuates alcohol‐mediated behavior cannot be determined. However, GLP‐1 receptors are distributed throughout the mesolimbic dopamine system (Alvarez et al. 1996; Merchenthaler et al. 1999), a reward pathway intimately associated with development of AUD (for review, see Engel & Jerlhag 2014 and Soderpalm et al. 2009). The possibility should therefore be considered that liraglutide attenuates the alcohol‐mediated behaviors through NAc and VTA GLP‐1 receptor. This is further substantiated by the findings that liraglutide passes the blood brain barrier (Hunter & Holscher 2012) and that local administration of a GLP‐1 receptor agonist into the VTA reduces alcohol intake as well as abolishes the ability of alcohol to cause a CPP in rats (Shirazi et al. 2013). In addition, GLP‐1 receptors are distributed throughout the brain, suggesting that GLP‐1 receptors outside of limbic areas may also control alcohol reinforcement. For instance, GLP‐1 receptors in the lateral septum, which is highly interconnected with the mesolimbic dopamine system, regulate the activity of VTA‐dopamine neurons as well as stress‐induced drug relapse (Highfield et al. 2000; Luo et al. 2011; Harasta et al. 2015). The recent data showing that GLP‐1 receptors within the lateral septum regulate cocaine‐induced behaviors in mice raise the possibility that these receptors may be involved in mediating the ability of liraglutide to attenuate alcohol‐induced behaviors in rodents (Harasta et al. 2015). Moreover, GLP‐1 receptors within the NTS are known to regulate food intake (Tang‐Christensen et al. 1996; McMahon & Wellman 1998; Hayes et al. 2009) as well as food reward behavior in rats (Richard et al. 2015) and may thus be involved in mediating the rewarding properties of alcohol. The findings that liraglutide activates GLP‐1 receptors in the arcuate nucleus of the hypothalamus, regulating liraglutide‐dependent weight loss (Secher et al. 2014), and that there are hypothalamic projections to the VTA provide another possible indirect pathway through which GLP‐1 may regulate alcohol‐mediated behaviors. In support for an indirect regulation of the mesolimbic dopamine system by liraglutide are the present data showing that liraglutide has no effect on accumbal dopamine release per se. Liraglutide may also regulate alcohol‐mediated behaviors via peripheral mechanisms because circulating GLP‐1 acts through activation of vagal afferents to GLP‐1 containing neurons in the NTS (Larsen & Holst 2005), which project directly to the VTA and NAc (Alvarez et al. 1996; Merchenthaler et al. 1999). To explore the role of GLP‐1 receptors within various brain areas, studies examining the effect of intra‐nuclei infusion of GLP‐1 receptor ligands on alcohol‐mediated behaviors are warranted.

In the present study, we firstly showed that acute administration of liraglutide, with no effect per se on accumbal dopamine releases or CPP, attenuated the rewarding properties of alcohol as measured by accumbal dopamine release as well as the CPP. The CPP experiments showed that liraglutide was able to block the rewarding properties of alcohol, but not the memory consolidation of alcohol reward. Moreover, acute administration of liraglutide did not influence the blood alcohol concentrations in mice. Secondly, acute administration of two different doses of liraglutide reduced alcohol intake in high‐alcohol‐consuming rats. We also showed that acute liraglutide treatment of a low dose in rats prevented the alcohol deprivation effect, an important characteristics of AUD. The alcohol deprivation effect in rodents has been suggested to reflect relapse caused by craving in the clinical setting (Spanagel 2000). Indeed, two currently available agents for treatment of alcohol dependence, naltrexone and acamprosate, prevent the alcohol deprivation effect in rats (Spanagel & Zieglgansberger 1997; Heyser et al. 2003) as well as craving‐induced relapse in humans (Soyka & Rosner 2008). Thirdly, repeated administration of a liraglutide reduced alcohol intake as well as alcohol preference in high‐consuming Wistar rats. Most importantly, the effect of liraglutide is long‐lasting and is pronounced during 24 hours. Repeated administration of liraglutide increased the water intake and did not affect total fluid intake, suggesting that the effects of liraglutide are selective for alcohol. Fourthly, repeated liraglutide treatment reduced the reinforcing effects of alcohol in alcohol‐preferring sP rats exposed to a standard procedure of operant self‐administration of alcohol. This set of data are of relevance as they extend the reducing effect of liraglutide on alcohol intake in unselected Wistar rats to (i) an operant procedure of oral alcohol self‐administration, providing therefore a first line of evidence on liraglutide potential on the reinforcing properties of alcohol (beside on its mere consumption), and (ii) a rat line selectively bred for high alcohol preference and intake, whose alcohol‐seeking and alcohol‐taking behaviors have been proposed to model several aspects of excessive alcohol consumption in humans (Colombo et al. 2006). Finally, we showed that the GLP‐1 receptor agonist did not induce a rebound increase in alcohol self‐administration in sP rats after the treatment was terminated. Indeed, we actually showed that the number of lever responses for alcohol and amount of self‐administered alcohol are lower, over the 4‐day posttreatment period, in sP rats previously treated with liraglutide compared with vehicle treatment. Of clinical interest is that we have used lower doses of liraglutide than in studies showing that liraglutide affects body weight and blood glucose levels in rodents (Raun et al. 2007; Secher et al. 2014). Collectively, these data show that the physiological role of GLP‐1 extends outside regulation of food intake and glucose homeostasis (Flint et al. 1998) to include reinforcement mediation.

In the present series of experiment, we show that liraglutide attenuates the ability of a second injection of alcohol to increase accumbal dopamine. Therefore, the possibility should be considered that we investigate a priming effect, rather than an acute reward effect, of alcohol. It should be considered a limitation that only one single dose of liraglutide was used to investigate the effects on alcohol‐induced activation of the mesolimbic dopamine system. In addition, we investigated the effect of alcohol in NAc shell (for schematic placements, see Figure S1), because we obtain a robust increase in accumbal dopamine in NAc shell as compared with NAc core (unpublished data). In addition, ghrelin increases NAc dopamine in shell but not in the core (Quarta et al. 2009). Even though we showed that acute administration or liraglutide did not influence the blood alcohol concentrations in mice, we cannot rule out that repeated liraglutide treatment affects the metabolism of alcohol. In addition, the blood alcohol levels were investigated following peripheral injections of alcohol, rather than following alcohol consumption. We also showed that co‐adjuvant, in contrast to acute, treatment of liraglutide attenuates alcohol‐induced CPP, suggesting that GLP‐1 receptor agonist interfere with the primary motivational properties of alcohol rather than the expression of CPP (Sanchis‐Segura & Spanagel 2006). The findings might be somewhat surprising because GLP‐1 receptor ligands, including liraglutide, previously has been attributed profound effects on memory formation, synaptic plasticity as well as hippocampal neuroprotection in a mouse model of Alzheimer disease (Holscher 2014). However, the possibility that memory consolidation of alcohol‐induced CPP and synaptic plasticity involves various brain circuits should therefore be considered. In the present experiment, we showed that liraglutide reduced both food and alcohol intake, raising the possibility that the GLP‐1 receptor agonist reduced ingestive behaviors of reinforcers with caloric value. However, repeated administration of liraglutide increases, and acute administration has a tendency of increasing water intake in rats. Moreover, this appears less likely because GLP‐1 receptor ligands attenuate drug‐induced reward (Erreger et al. 2012; Graham et al. 2013; Egecioglu et al. 2013a,2013b). Investigation of the possibility that liraglutide alters ingestive behavior and tasting of the effects of liraglutide on intravenous alcohol self‐administration in rodents or humans should be explored. In corroboration, intravenous ghrelin administration increases the craving for alcohol in alcohol‐dependent individuals (Leggio et al. 2014).

The direct involvement of GLP‐1 receptors for reinforcement is further supported by the data showing that another GLP‐1 receptor agonist, Ex4, reduces alcohol intake, attenuates alcohol‐seeking behavior as well as blocks the alcohol‐induced reward in mice (Egecioglu et al. 2013c). These data were extended and corroborated by others showing that a GLP‐1 receptor antagonist increases alcohol intake in rats (Shirazi et al. 2013). In addition, Ex4 reduced food reward and motivation to consume palatable food, and a GLP‐1 antagonist blocked these effects in rats (Dickson et al. 2012). Taken together with the data showing that liraglutide reduced weight and fat gains, decreased calorie intake as well as shifted food preference from candy to regular chow in candy‐fed rats (Raun et al. 2007), it may be suggest that GLP‐1‐sensitive mechanisms regulate reward processes in general. In corroboration, Ex4 attenuate nicotine‐induced, amphetamine‐induced and cocaine‐induced reward as measured by accumbal dopamine release, CPP and locomotor stimulation in rodents (Erreger et al. 2012; Graham et al. 2013; Egecioglu et al. 2013a,2013b). It was recently found that withdrawal‐induced anxiety as well as tolerance to the anxiolytic effects of alcohol is delayed by liraglutide or by inhibition of the enzyme, dipeptidyl‐peptidase IV, responsible for degradation of endogenous GLP‐1 (Sharma et al. 2014a,2014b). The findings that this gut‐brain peptide reduces alcohol as well as nicotine reinforcement in rodents and that there is a co‐morbidity between smoking and alcohol dependence (for review, see Soderpalm et al. 2009) hold potentially important clinical implications because both addictions might be affected by GLP‐1 receptor ligands.

Acute treatment with high doses of liraglutide has previously been shown to induce a condition taste aversion (Kanoski et al. 2012), raising the possibility that a reduction in alcohol intake is due to nausea rather than reduced alcohol reward. However, this appears less likely because we evaluated the effects of a low dose of liraglutide on conditioned place aversion, which has been suggested to correlate to conditioned taste aversion (Cagniard & Murphy 2012). We showed that repeated liraglutide treatment did not induce a conditioned place aversion in vehicle‐paired mice during a drug‐free session, indicating that the selected doses of liraglutide do not induce aversion. Moreover, we used lower doses than those causing a taste aversion (Kanoski et al. 2012), and we followed up the acute studies with repeated administration of liraglutide, a design with documented tolerance for nausea. This is further substantiated by the findings that liraglutide, in contrast to Ex4, does not condition a taste avoidance to saccharine in mice (McKay & Daniels 2013).

The present study show that repeated administration of the GLP‐1 analog decreases food intake as well as body weight. Rodent and human studies collectively show that liraglutide induces weight loss by reducing appetite and energy intake (Tang‐Christensen et al. 1996; Flint et al. 1998; Gutzwiller et al. 1999; ). The findings showing that liraglutide does not alter the intake of chow nor high‐fat diet in neuronal GLP‐1 knockout mice, but still affects glucose levels, suggest that central GLP‐1 receptors are required for regulation of food intake in mice (Sisley et al. 2014). Previous studies show that the anorexigenic properties of GLP‐1 are mediated via GLP‐1 receptors in the NTS, amygdala and hypothalamus (Tang‐Christensen et al. 1996; McMahon & Wellman 1998; Hayes et al. 2009) as well as NAc (Dossat et al. 2011). In addition, intra‐VTA or intra‐NAc administration of Ex4 decreased the motivation for sucrose, reduced the intake of palatable food as well as attenuated CPP for palatable food (Alhadeff et al. 2012; Dickson et al. 2012). The liraglutide‐induced weight loss, at least in part, is mediated via pro‐opiomelanocortin (POMC)/cocaine‐ and amphetamine‐regulated transcript (CART) neurons in the hypothalamic arcuate nucleus in mice (Secher et al. 2014). We therefore suggest that the obtained reduction in food intake and body weight following peripheral administration of liraglutide is due to GLP‐1 signaling within as well as outside of brain reward areas. Given that liraglutide reduced food consumption in rats and humans (Tang‐Christensen et al. 1996; Flint et al. 1998; Gutzwiller et al. 1999; ), the possibility that the reduction in alcohol intake is due to calories rather than reward should be considered. This appears less likely because we showed that liraglutide did not induce a CPP in vehicle‐paired mice. Moreover, other GLP‐1 receptor agonists attenuate calorie‐independent reward, namely nicotine‐induced, amphetamine‐induced and cocaine‐induced CPP, locomotor stimulation and accumbal dopamine releases (Erreger et al. 2012; Graham et al. 2013; Egecioglu et al. 2013a,2013b).

In addition to GLP‐1, several other neuropeptides modulate the responses to alcohol. Ghrelin increases, whereas ghrelin receptor antagonists reduce alcohol intake, the rewarding properties of alcohol as well as the motivational properties of alcohol in rodents (for review, see Engel & Jerlhag 2014). It was initially shown that the alpha‐melanocyte‐stimulating hormone–melanocortin (MC)‐4 receptor system reduces alcohol intake and ease alcohol withdrawal symptoms (for review, see Olney et al. 2014). In addition, a recent study showed that intra‐VTA injection of alpha‐melanocyte‐stimulating hormone or an MC4 agonists increased lever presses for alcohol in rats (Shelkar et al. 2015). Cocaine‐regulated and amphetamine‐regulated transcript, which interacts prominently with the mesolimbic dopamine system, inhibits context‐induced reinstatement of alcohol‐seeking behaviors, and cocaine‐regulated and amphetamine‐regulated transcript knockout mice consume and prefer alcohol less than wild‐type mice (King et al. 2010; Salinas et al. 2014). In comparison with wild‐type mice, neuropeptide Y (NPY)‐deficient mice display an increased alcohol intake whereas NPY‐overexpressing mice have suppressed alcohol consumption (for review, see Thorsell 2007). Moreover, an NPY2 antagonist reduces alcohol intake in alcohol‐naïve as well postdependent animals, and NPY blocks yohimbine‐induced reinstatement of alcohol seeking in rats (Thorsell 2007; Cippitelli et al. 2010). The anorexigenic peptides leptin, cholecystokinin and galanin decrease preference for as well as intake of alcohol in rodents (for review, see Engel & Jerlhag 2014).

The present study shows that acute administration of a low dose of liraglutide attenuates the ability of alcohol to release accumbal dopamine, to cause a CPP in mice and to reduce alcohol consumption as well as prevents the alcohol deprivation effect in rats. Moreover, repeated administration of a low dose of liraglutide reduces alcohol intake as well as decreases the reinforcing properties of alcohol in rats. Liraglutide is currently used as treatments for type II diabetes because it enhances glucose‐dependent insulin secretion (Holst & Seino 2009), reduces gastric emptying as well as decreases glucagon secretion (Kreymann et al. 1987; Orskov et al. 1988; Flint et al. 2001). We therefore argue that GLP‐1 and its receptor play a role in the pathophysiology of alcohol‐mediated behavior and that GLP‐1 receptor agonists, such as liraglutide, deserve to be evaluated as potential therapeutics for AUD.