Cue-conditioned alcohol seeking in rats following abstinence: involvement of metabotropic glutamate 5 receptors

Abstract

Background and purpose:

The current study was designed to: (i) examine whether functional interactions occur between receptors known to regulate alcohol self-administration; and (ii) characterize relapse to alcohol seeking following abstinence.

Experimental approach:

The selective cannabinoid CB₁ receptor antagonist SR141716A (0.03–1.0 mg·kg⁻¹ i.p.) resulted in a dose-dependent reduction in ethanol self-administration in ethanol-preferring Indiana-preferring rats. SR141716A was then co-administered with either the selective glutamate metabotropic glutamate 5 (mGlu₅) receptor antagonist 3-[(2-methyl-1,3-thiazol-4-yl)ethynyl]pyridine (MTEP) or the selective adenosine A_2A receptor antagonist SCH58261.

Key results:

When administered at individually sub-threshold doses, a combination of SR141716A (0.1 mg·kg⁻¹) and SCH58261 (0.5 mg·kg⁻¹ i.p.) produced a reduction (28%) in ethanol self-administration. Combinations of threshold doses of SR141716A (0.3 mg·kg⁻¹) and SCH58261 (2.0 mg·kg⁻¹, i.p.) caused an essentially additive reduction (68%) in alcohol self-administration. A combination of individually sub-threshold doses of CB₁ and mGlu₅ receptor antagonists did not affect alcohol self-administration; however, combined threshold doses of SR141716A (0.3 mg·kg⁻¹) and MTEP (1.0 mg·kg⁻¹ i.p.) did reduce ethanol self-administration markedly (80%). Cue-conditioned alcohol seeking was attenuated by pretreatment with MTEP (1.0 mg·kg⁻¹) co-administered with SR141716A (0.3 mg·kg⁻¹ i.p.). In contrast, SCH58261 (2.0 mg·kg⁻¹) co-administered with SR141716A (0.3 mg·kg⁻¹ i.p.) did not reduce cue-conditioned alcohol seeking.

Conclusions and implications:

Adenosine A_2A and cannabinoid CB₁ receptors regulated alcohol self-administration additively, but combined low-dose antagonism of these receptors did not prevent cue-conditioned alcohol seeking after abstinence. In contrast, combined low-dose antagonism of mGlu₅ and CB₁ receptors did prevent relapse-like alcohol seeking after abstinence, suggesting a prominent role for mGlu₅ receptors in this paradigm.

Introduction

Ethanol is the second most commonly abused psychotropic drug after caffeine (Nevo and Hamon, 1995). The consequence of alcohol abuse is far more serious, however, resulting in significant social and economic costs. The World Health Organization estimates that 2 billion people consume alcohol, and of those 76.3 million have diagnosable alcohol use disorders. In 1998, the estimated cost of alcohol abuse to the United States was $184.6 billion (World Health Organization, 2004). Due to the polymodal nature of alcohol addiction, relapse rates remain high under treatment (Graham et al., 2002; Heidbreder, 2005), and further investigation into the pathophysiology of the disease is required. Early theories on the mechanism of action of ethanol suggested that membrane lipids were a target, but more recent research refutes this simple view, with enzymes, receptors and various ion channels shown to be targets for ethanol (see Vengeliene et al., 2008).

The metabotropic glutamate 5 (mGlu₅) receptor (nomenclature follows Alexander et al., 2008) is a member of the seven-transmembrane, G protein-coupled receptor family. High levels of mRNA encoding the mGlu₅ receptor and associated protein are found within the olfactory tubercle and bulb, nucleus accumbens, caudate–putamen, lateral septum, hippocampus and cortex (Romano et al., 1995; Sahara et al., 2001). The mGlu₅ receptor is positively coupled to adenylate cyclase through G_s/G_olf proteins, and has been associated with phosphoinositide hydrolysis and activation of phospholipase C (Conn and Pin, 1997; Hermans and Challiss, 2001).

The mGlu₅ receptor appears to be involved in the reinforcing properties of a number of drugs of abuse. Concerning ethanol, the mGlu₅ receptor antagonist 2-methyl-6-(phenylethynyl)-pyridine (MPEP) decreased consumption in mice (Olive et al., 2005) and rats (McMillen et al., 2005), and operant self-administration in mice (Hodge et al., 2006) and rats (Schroeder et al., 2005). MPEP also prevented the reinstatement of ethanol-seeking behaviour induced by olfactory cues (Backstrom et al., 2004) or repeated deprivations (Schroeder et al., 2005) in rats. Administration of 3-[(2-methyl-1,3-thiazol-4-yl)ethynyl]pyridine (MTEP), a mGlu₅ receptor antagonist with greater selectivity and bioavailability than MPEP (Anderson et al., 2002; Cosford et al., 2003), decreased ethanol self-administration in two strains of alcohol-preferring rats (Cowen et al., 2005b) and reduced both consummatory and appetitive responding for ethanol in C57/BL6J mice (Cowen et al., 2007). mGlu₅ receptor-deficient mice consume less ethanol than their wild-type littermates, in a two-bottle free-choice paradigm, and are more susceptible to the hypnotic effects of ethanol (Bird et al., 2008).

The adenosine A_2A receptor belongs to a family of four G protein-coupled adenosine receptors that are widely distributed throughout the body, with particularly strong expression in the basal ganglia (Fredholm et al., 2000; Yaar et al., 2005). Adenosine A_2A receptors are localized pre-synaptically on glutamatergic afferents from the prefrontal cortex, as well as post-synaptically on enkephalin/dopamine D₂ receptor expressing GABAergic medium spiny neurones, where they appear to modulate cortico-limbic-striatal glutamatergic neurotransmission (Schiffmann et al., 2007). The A_2A receptor is positively coupled to adenylate cyclase via G_s or G_olf proteins, and, upon activation, elevates cAMP and intracellular calcium levels (Ongini and Fredholm, 1996; Kull et al., 1999; Fredholm et al., 2000; Yaar et al., 2005).

Worldwide, the most commonly used psychoactive drug is caffeine, a non-selective adenosine receptor antagonist. The adenosine A_2A receptor in particular appears to be involved in the reinforcing properties of ethanol and other drugs of abuse (Brown and Short, 2008; Castane et al., 2008; Brown et al., 2009). Adenosine A_2A receptor-deficient mice are observed to be less sensitive to the acute intoxicating effects of ethanol (Naassila et al., 2002), and exhibit blunted ethanol withdrawal effects, a phenotype replicated in wild-type mice following treatment with A_2A receptor antagonists (El Yacoubi et al., 2001). A_2A receptor antagonism also attenuates operant self-administration of ethanol in rats (Arolfo et al., 2004; Thorsell et al., 2007; Adams et al., 2008).

Cannabinoid receptors are also seven-transmembrane-spanning, G protein-coupled receptors. There are two known subtypes, one of which, the CB₁ receptor, is the most abundantly expressed G protein-coupled receptor within the CNS (Solinas et al., 2008). The CB₁ receptor is negatively coupled through G_i/G_o proteins to adenylate cyclase, and positively coupled to mitogen-activated protein kinases (Solinas et al., 2008). The CB₁ receptor antagonist rimonabant (N-(piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide hydrochloride; SR141716A) inhibits phasic dopamine release in the nucleus accumbens evoked by nicotine, ethanol and cocaine administration in freely moving Sprague-Dawley rats (Cheer et al., 2007) without altering basal dopamine release (Cheer et al., 2004). Considering the effect of many drugs of abuse upon phasic dopamine release, it is logical that the CB₁ receptor is more widely implicated in the reinforcing properties of many drugs of abuse (Cheer et al., 2007).

CB₁ receptor knockout mice display reduced consumption of ethanol in a two-bottle free-choice paradigm (Lallemand and de Witte, 2005; Thanos et al., 2005), a reduction in place preference for ethanol in a conditioned place preference paradigm (Thanos et al., 2005), and impaired neuroadaptation of NMDA and GABA_A receptors following chronic ethanol exposure (Warnault et al., 2007). CB₁ receptor knockout mice also show reduced cocaine self-administration, reduced drug-lever pairing discrimination and a lower progressive ratio breakpoint, results which mirror those obtained when the CB₁ antagonist SR141716A is given to wild-type mice (Soria et al., 2005). Interestingly, however, when CB₁ receptor-deficient mice are administered cocaine, they show similar dopamine release in the nucleus accumbens as their wild-type littermates (Soria et al., 2005), a finding perhaps suggestive of developmental adaptations within the nucleus accumbens.

Another CB₁ receptor antagonist, SR147778, attenuates ethanol acquisition, drinking and the deprivation effect in Sardinian alcohol-preferring (sP) rats in a two-bottle free-choice paradigm (Gessa et al., 2005). sP Rats exhibit greater CB₁ receptor mRNA expression than Wistar rats within the mesocorticolimbic system (Cippitelli et al., 2005), suggesting that the differences observed may correlate with alcohol drinking behaviours. SR147778 significantly reduces ethanol preference and intake in a two-bottle free-choice paradigm (Lallemand and De Witte, 2006), and also self-administration under operant conditions (Economidou et al., 2007) in Wistar rats. Furthermore, SR141716A significantly decreases the motivation to self-administer ethanol, and blocks cue-induced reinstatement in both Wistar (Cippitelli et al., 2005; Economidou et al., 2006) and sP rats (Cippitelli et al., 2005), but does not affect stress-induced reinstatement in Wistar rats (Economidou et al., 2007).

We have previously reported that the selective adenosine A_2A receptor antagonist SCH58261 and the selective glutamate mGlu₅ receptor antagonist MTEP interact to produce an apparently synergistic reduction in ethanol self-administration, and in combination completely block cue-induced reinstatement in alcohol-preferring Indiana-preferring rat (iP) rats (Adams et al., 2008). The logic for investigating mGlu₅ and A_2A receptor interactions arose from the co-localization of the A_2A and mGlu₅ receptors within the mesocorticolimbic pathway, a circuit intimately involved in natural and drug-induced reward. The existence of heteromeric receptor complexes (Ferre et al., 2002), signal transduction commonalities (Agnati et al., 2003) and other functional evidence (Nishi et al., 2003; Kachroo et al., 2005; Rodrigues et al., 2005) also highlight the need to explore interactions between these receptors in relation to complex behavioural patterns.

Likewise, interactions between the cannabinoid CB₁ receptor and other receptors have been reported. For example, µ-opioid and CB₁ receptors have been found to interact synergistically via a common signal transduction pathway in cultured primary striatal and accumbal neurones, and this phenomenon was regulated by the adenosine A_2A receptor (Yao et al., 2006). Reductions in tetrahydrocannabinol-induced rewarding or aversive effects have been found in mice lacking the A_2A receptor, which display a normal distribution of the CB₁ receptor (Soria et al., 2004).

Given the ability of the CB₁ receptor to regulate alcohol self-administration, the similarities in the signal transduction pathways of CB₁, mGlu₅ and A_2A receptors, and the co-localization of the receptors within the mesocorticolimbic pathway, here we seek to extend previous work and examine if interactions exist between A_2A, mGlu₅ and CB₁ receptors in an operant, ethanol self-administration paradigm, utilizing alcohol-preferring (iP) rats. The choice of cue-conditioned alcohol seeking following abstinence, rather than typical extinction–reinstatement paradigms, is based on the premise that many human alcohol/drug users do not undergo the equivalent of extinction training via rehabilitation programmes. Indeed, it is common for humans with alcohol use disorders to relapse following a period of abstinence (40–60% within months, 70–80% by 1 year; Dawson et al. (2007). Consequently, we have established a model to address this often overlooked issue.

Methods

Animals

All experiments were performed in accordance with the Prevention of Cruelty to Animals Act 1986, under the guidelines of the National Health and Medical Research Council Code of Practice for the Care and Use of Animals for Experimental Purposes in Australia. Inbred alcohol-preferring (iP) rats were obtained from the breeding colony at the Howard Florey Institute (University of Melbourne). Parental stock had previously been obtained from Professor TK Li (while at Indiana University, Indianapolis, IN, USA). The animals were pair housed with ad libitum access to standard rat chow and water, with a 12 h light/dark cycle; lights on 0700 h.

Operant alcohol self-administration

Alcohol-preferring iP rats (n= 30) were trained to self-administer ethanol (10% v/v; Merck Pty Ltd, Victoria, Australia) under operant conditions using a fixed ratio of 3 (FR3) during 20 min sessions as previously described (Cowen et al., 2005a,b; Lawrence et al., 2006; Liang et al., 2006). Operant chambers supplied by Med Associates (St Albans, VT, USA) were employed. Each chamber was housed individually in sound attenuation cubicles, featuring a fan to provide airflow and mask external noise. The chambers were connected to a computer running Med-PC IV software (Med Associates) to record the activity. Availability of ethanol was conditioned by the presence of an olfactory cue (S+; 2 drops of vanilla essence, placed on the bedding of the operant chamber directly under the active lever), plus a 1 s light stimulus (CS+) occurred when FR3 was obtained. Ultimately, the rats were responding for a 10% ethanol solution under a fixed ratio requirement of 3 (FR3) with the presentation of alcohol and water randomized to minimize side preference. For each session, total ethanol and water responses were recorded with each delivery consisting of 100 µL of either water or ethanol solution, and the difference of fluid in the ethanol reservoir between the beginning and end of the session was also recorded to ensure the correct calibration of the delivery system. Following acquisition of lever-pressing behaviour and stable alcohol self-administration (<10% variation across sessions, 10% ethanol v/v), the rats were given the adenosine A_2A receptor antagonist SCH58261 (5-amino-2-(2-furyl)-7-phenylethyl-pyrazolo[4,3-e]-1,2,4-triazolo[1,5c]pyrimidine, Sigma, St Louis, MO, USA; 1 and 2 mg·kg⁻¹ i.p. and co-administered as detailed) suspended in methyl cellulose 30 min prior to the beginning of operant sessions. The mGlu₅ receptor antagonist MTEP (Ascent Scientific, N. Somerset, UK; 0.25 and 1 mg·kg⁻¹ i.p.) was dissolved in 1% dimethyl sulphoxide (DMSO) as previously described (Cowen et al., 2005b; 2007; Adams et al., 2008) and administered 20 min prior to the operant session. The CB₁ receptor antagonist SR141716A (Sanofi Synthelabo Recherche, Montpelier, France; 0.03, 0.1, 0.3 and 1.0 mg·kg⁻¹ i.p. and co-administered as detailed) was dissolved in Tween 80/saline, and administered 20 min prior to the operant session. Administration of MTEP with either SCH58261 or SR141716A involved two injections both at 0.5 mL·kg⁻¹ at the time-points mentioned; vehicle controls were delivered in the same way. Drug administration weeks were structured so that Mondays and Fridays were no injection days; vehicle was injected either Tuesday or Wednesday (i.p.) followed by drug the following day (i.p.).

Cue-conditioned alcohol seeking

Following standard operant training as detailed earlier, iP rats (n= 30) were returned to their home cages for a period of 4 weeks. Alcohol seeking was then triggered by replacing S+ (i.e. the olfactory cue) under the ‘active’ lever and also reprogramming the software such that the stimulus light (CS+) was activated (for 1 s) after every FR3 response, although there was no delivery of ethanol into the receptacle. Prior to the session, the rats were treated with either vehicle or a combination of SCH58261 (0.5 mg·kg⁻¹ i.p. 30 min prior) and SR141716A (0.1 mg·kg⁻¹ i.p. 20 min prior) or MTEP (0.25 mg·kg⁻¹ i.p. 20 min prior) and SR141716A (n= 10 per treatment).

Statistics

Statistical analysis was performed with SigmaStat (version 3; SPSS Inc., Chicago, IL, USA). The data are presented as mean ± SEM. A significance level of P= 0.05 was used. In general, session totals and time-courses were analysed using a repeated-measures two-way analysis of variance (anova) with Student–Newman–Keuls post hoc tests. For every dose of drug, there is a corresponding vehicle injection, thus the factors for the session totals were treatment versus drug/vehicle; for the time-course analysis, the factors were drug/vehicle versus time-point. The effect of 0.5 mg·kg⁻¹ i.p. SCH58261 was also examined using a paired t-test. Vehicle injections on each day were not significantly different for ethanol or water lever presses (as examined via a paired t-test), and were therefore pooled for graphical representation. The resultant data were analysed using a one-way anova with a Student–Newman–Keuls post hoc test.

Results

Effect of SR141716A on operant responding for ethanol

We have previously published dose–response curves demonstrating the effect of MTEP and SCH58261 upon ethanol self-administration in iP rats (Cowen et al., 2005b; Adams et al., 2008). Therefore, we first conducted a dose–response curve for the selective CB₁ antagonist SR141716A. A cohort of 10 iP rats responding stably and preferentially for 10% v/v ethanol (38.7 ± 11.3 responses representing 0.6 ± 0.1 g·kg⁻¹ ethanol per 20 min session) compared with water (3.8 ± 1.0 responses) were treated with either vehicle or SR141716A (Figure 1). A significant effect of treatment occurred [treatment: F_(6,42)= 1.095, P= 0.381; drug vs. vehicle F_(1,7)= 7.858, P= 0.026] indicating that SR141716A significantly reduced ethanol self-administration at 1.0 mg·kg⁻¹ i.p. (−75%, P= 0.003) and 0.3 mg·kg⁻¹ i.p. (−50%, P= 0.006). SR141716A at 1.0 mg·kg⁻¹ i.p. also significantly attenuated water responding (P= 0.009), but 0.3 mg·kg⁻¹ was without effect on water responding. No significant effects on self-administration of ethanol or water were noted when SR141716A was administered at 0.1 or 0.03 mg·kg⁻¹ i.p.

Figure 1.

The effect of SR141716A and MTEP on operant ethanol self-administration in iP rats (n= 10). Ethanol self-administration was significantly reduced by SR141716A administration at 1.0 mg·kg⁻¹ (P= 0.003) and 0.3 mg·kg⁻¹ (P= 0.006) i.p. When co-administered at a low dose, SR141716A and MTEP had no effect on ethanol self-administration at 0.1 and 0.25 mg·kg⁻¹, respectively, but did significantly reduce ethanol self-administration when co-administered at 0.3 and 1 mg·kg⁻¹ i.p. (P= 0.001). Water responses were significantly reduced by SR141716A at 1 mg·kg⁻¹ (P= 0.009) and SR141716A, and MTEP co-administered at 0.1 and 0.25 mg·kg⁻¹ i.p. (P= 0.011). V, Vehicle; SR0.1 or 0.3, SR141716A at 0.1 or 0.3 mg·kg⁻¹ i.p.; MTEP 0.25 or 1.0, MTEP at 0.25 or 1.0 mg·kg⁻¹ i.p. *Significantly different to vehicle.

Effect of MTEP and SR141716A on operant responding for ethanol

From previously published experiments, the highest sub-threshold dose of MTEP in iP rats was 0.25 mg·kg⁻¹ i.p. (Adams et al., 2008). This was repeated in the current cohort of iP rats to confirm no significant reduction in ethanol self-administration (Figure 1, P > 0.05). When co-administered at individually sub-threshold doses, SR141716A (0.1 mg·kg⁻¹) and MTEP (0.25 mg·kg⁻¹ i.p.) produced no reduction in ethanol self-administration; however, a small, but significant, reduction in water responding was observed (P= 0.011). When co-administered at individually threshold doses, SR141716A (0.3 mg·kg⁻¹) and MTEP (1.0 mg·kg⁻¹ i.p.) in combination produced a significant reduction in ethanol self-administration (−80%, P= 0.001). It should be noted that MTEP administered alone (1.0 mg·kg⁻¹ i.p.) in a previous study with similar protocols and the same strain of rat significantly reduced ethanol (∼−55%), but not water self-administration (Cowen et al., 2005b).

Effect of SCH58261 and SR141716A on operant responding for ethanol

From previously published experiments, the highest sub-threshold dose of SCH58261 in iP rats was 0.5 mg·kg⁻¹ i.p. (Adams et al., 2008). This was repeated in a cohort of iP rats with no significant reduction in ethanol self-administration (Figure 2). SR141716A (0.1 mg·kg⁻¹ i.p.) was also administered to this cohort to reverify that this dose was sub-threshold for alcohol self-administration (Figure 2).

Figure 2.

The effect of SCH58261 and SR141716A on operant ethanol self-administration in iP rats (n= 10). Ethanol self-administration was not significantly altered by SCH58261 at 0.5 mg·kg⁻¹ i.p., although water responding was reduced (P= 0.035). SR141716A at 0.1 mg·kg⁻¹ i.p. did not alter responding. When co-administered, SR141716A and SCH58261 significantly reduced responding for ethanol without affecting responding for water at 0.1 and 0.5 mg·kg⁻¹ (P= 0.030), and 0.3 and 2 mg·kg⁻¹ (P= 0.005) respectively. V, Vehicle; SCH0.5 or 2, SCH58261 at 0.5 or 2.0 mg·kg⁻¹ i.p.; SR0.1 or 0.3, SR141716A at 0.1 or 0.3 mg·kg⁻¹ i.p. *Significantly different to vehicle.

When co-administered at the highest sub-threshold doses, SR141716A (0.1 mg·kg⁻¹) and SCH58261 (0.5 mg·kg⁻¹ i.p.) produced a significant reduction in ethanol self-administration (−28%, P= 0.030, Figure 2). When co-administered at threshold doses, SR141716A (0.3 mg·kg⁻¹) and SCH58261 (2.0 mg·kg⁻¹ i.p.) together produced a significant reduction in ethanol self-administration (−68%, P= 0.005). We have previously shown that SCH58261 (2.0 mg·kg⁻¹ i.p.) significantly reduces ethanol self-administration (−47%) in this rat strain using this paradigm (Adams et al., 2008).

Effect of SR141716A/SCH58261 and SR141716A/MTEP on cue-conditioned alcohol seeking

Following a period of withdrawal (4 weeks), the rats were tested for cue-conditioned alcohol seeking under extinction conditions (cues present, but no delivery of ethanol subsequent to lever pressing). Figure 3 shows that under these conditions, iP rats show robust alcohol seeking in this paradigm. MTEP (1.0 mg·kg⁻¹) co-administered with SR141716A (0.3 mg·kg⁻¹ i.p.) significantly reduced alcohol seeking (P < 0.001) compared to vehicle-treated iP rats. In contrast, SCH58261 (2 mg·kg⁻¹) co-administered with SR141716A (0.3 mg·kg⁻¹ i.p.) did not reduce active lever pressing during relapse-like behaviour (P= 1.0).

Figure 3.

The effect of SCH58261 and MTEP, co-administered with SR141716A, on cue-conditioned alcohol seeking in iP rats (n= 10 per treatment). MTEP (1.0 mg·kg⁻¹) and SR141716A (0.3 mg·kg⁻¹) significantly reduced cue-conditioned alcohol seeking compared with vehicle (P < 0.001). SCH58261 (2.0 mg·kg⁻¹) and SR141716A (0.3 mg·kg⁻¹) co-administered i.p. had no effect on cue-conditioned alcohol seeking. Veh, Vehicle; SCH2.0, SCH58261 at 2.0 mg·kg⁻¹ i.p.; SR0.3, SR141716A at 0.3 mg·kg⁻¹ i.p.; MTEP1.0, MTEP at 1.0 mg·kg⁻¹ i.p. *Significantly different to vehicle (P < 0.001).

Effects of combinations of SCH58261, MTEP and SR141716A on operant ethanol self-administration in iP rats

Figure 4A shows the normalized, synergistic effect of combining individually sub-threshold doses of SCH58261 and MTEP on ethanol self-administration in iP rats (primary data published in Adams et al., 2008). From the present data, it is clear that combining either individually sub-threshold doses of SR141716A (0.1 mg·kg⁻¹ i.p.) and MTEP (0.25 mg·kg⁻¹ i.p.; Figure 4B), or individually sub-threshold doses of SR141716A (0.1 mg·kg⁻¹ i.p.) and SCH58261 (0.5 mg·kg⁻¹ i.p.; Figure 4C), produces effects on alcohol self-administration that are essentially additive.

Figure 4.

Summary figures describing the effects of SCH58261, MTEP and SR141716A on operant ethanol self-administration in iP rats. Open bars represent the percentage change in deliveries of 10% v/v ethanol on an FR3 schedule. The dotted, horizontal line represents the theoretical additive effect of each combination. The solid bar represents the actual percentage reduction in ethanol self-administration achieved with co-administration. (A) The combination of SCH58261 (0.5 mg·kg⁻¹) and MTEP (0.25 mg·kg⁻¹) injected i.p. produces a synergistic reduction in ethanol self-administration in iP rats. Primary data for Figure 4a have been reported previously (Adams et al., 2008). (B) The combination of SR141716A (0.1 mg·kg⁻¹) and MTEP (0.25 mg·kg⁻¹ both i.p.) and (C) the combination of SR141716A (0.1 mg·kg⁻¹) and SCH58261 (0.5 mg·kg⁻¹ i.p.) produced essentially additive effects on ethanol self-administration. SCH: SCH58261, SR: SR141716A.

Discussion

Here, we report that combinations of antagonists that target glutamate mGlu₅, adenosine A_2A and cannabinoid CB₁ receptors can regulate alcohol self-administration and/or seeking following a period of abstinence. We have previously noted that co-administration of low doses of A_2A and mGlu₅ receptor antagonists results in a dramatic reduction of alcohol self-administration and blocks cue-induced reinstatement following extinction (Adams et al., 2008). We now provide more evidence for the specificity of this A_2A/mGlu₅ interaction, because essentially additive effects were observed for combinations of A_2A/CB₁ or mGlu₅/CB₁ receptor antagonists in relation to self-administration of alcohol. Moreover, combined low-dose CB₁/A_2A receptor antagonists had no effect on alcohol seeking following a period of abstinence. In contrast, while an mGlu₅/CB₁ receptor antagonist combination again suggested a simple additive effect in terms of reducing alcohol self-administration, this combination was able to prevent alcohol seeking following abstinence. Consequently, this study demonstrates that in iP rats, following a period of withdrawal the re-exposure to cues previously associated with the availability of alcohol precipitates a relapse-like response that apparently involves mGlu₅-mediated signalling.

At sub-threshold doses, the effect of SR141716A and MTEP in combination produced essentially an additive effect (Figure 4B), suggesting that CB₁ and mGlu₅ receptors may not functionally interact in a co-operative or facilitatory manner within this paradigm to reduce operant ethanol self-administration. Although sub-threshold combinations of SR141716A co-administered with SCH58261 i.p. reduced ethanol self-administration, the effect was no greater than additive (Figure 4C), suggesting that CB₁ and adenosine A_2A receptors probably work independently within this paradigm.

Striatal cannabinoid CB₁ receptors are localized pre-synaptically on GABAergic terminals of interneurones, collaterals from GABAergic efferent neurones and also on glutamatergic but not dopaminergic terminals (Kofalvi et al., 2005). If CB₁ receptors are found post-synaptically, their population is sparse (for review, see Ferre et al., 2009). Adenosine A_2A receptors are localized within the striatum post-synaptically on dopamine D₂ receptor/enkephalin expressing GABAergic medium spiny neurones receiving glutamatergic input, and a smaller population pre-synaptically (Hettinger et al., 2001; Rosin et al., 2003). In the striatum, CB₁ and A_2A receptors are co-localized on glutamatergic nerve terminals and in the dendritic spines of GABAergic/enkephalinergic neurons (Shindou et al., 2008). As a heteromer in cell culture, cannabinoid CB₁ receptor function is co-dependent on adenosine A_2A receptor function, with activation of the heteromer resulting in G_i protein signalling (Carriba et al., 2007).

mGlu₅ receptors are mainly located within the post-synaptic density of glutamatergic synapses (Pin et al., 2003). The anatomical co-localization of the A_2A and mGlu₅ receptors, and their existence as a heteromer in striatal extracts (Ferre et al., 2002) suggest opportunities for interaction. Receptor heteromer interactions are proposed to occur via various mechanisms. These include direct, protein–protein interactions and intra-membrane lipid and scaffolding protein interactions resulting in allosteric modulation of receptors and subsequent alterations in ligand affinity, and phosphorylation and signal transduction interactions (Franco et al., 2003). One interesting finding from Yao et al. (2008) reveals a substantial cross-talk between PKA (mGlu₅ and dopamine D₂ receptors) and PKC (cannabinoid CB₁ and adenosine A_2A receptors) signalling occurs whereby PKC activation leads to potentiation of G_s receptor signalling. Additionally, the activator of G protein signalling 3 (AGS3) levels within the nucleus accumbens core are significantly elevated following abstinence, knockdown of which normalizes heightened alcohol-seeking responses in rats (Bowers et al., 2008). Further work exploring these interactions within signalling transduction pathways would seem essential.

There is evidence to support the existence of adenosine A_2A/mGlu₅/dopamine D₂ receptor mosaics. Indeed, A_2A or mGlu₅ receptor agonists reduce the affinity of dopamine D₂ binding sites, with concurrent stimulation resulting in synergistic interactions for c-fos expression, ERK phosphorylation and DARPP-32 (Ferre et al., 2002; Nishi et al., 2003). Functional interactions also occur between A_2A and mGlu₅ receptor antagonists, which synergistically increase locomotion in reserpinized mice (Coccurello et al., 2004; Kachroo et al., 2005).

The CB₁ receptor, generally located across the synapse from the A_2A/mGlu₅ receptor complex, has no such opportunity for ‘direct’ interaction as a mosaic, although the cannabinoid system has been demonstrated to signal via retrograde messaging (Matyas et al., 2006), and so the possibility exists for A_2A and/or mGlu₅ receptors to modulate the synthesis and/or release of endocannabinoids from medium spiny neurones. Excited cortico-striatal glutamatergic inputs induce retrograde endocannabinoid signalling, which is involved in dopamine D₂ receptor-mediated long-term synaptic plasticity in this region (Giuffrida et al., 1999; Centonze et al., 2004). Ethanol self-administration dose-dependently increases dialysate levels of 2-arachidonoylglycerol within the nucleus accumbens shell of rats (Caille et al., 2007). This plasticity subsequently results in the reduced probability of glutamate release (Choi and Lovinger, 1997), and as previously alluded to, adenosine A_2A or mGlu₅ receptor antagonists effectively function as positive allosteric modulators of dopamine D₂ receptors within the receptor mosaic. Further work to extend the current experiment into a chronic setting is required to examine striatal plasticity within the context of alcohol studies.

The association of instrumental actions (lever pressing) followed by reward is mediated within the dorsal striatum, as lesions to, or dopamine antagonists infused into this area, abolish this association (Faure et al., 2005; Vanderschuren et al., 2005). Drug-primed reinstatement is associated with glutamatergic inputs into the basal ganglia (Kalivas and McFarland, 2003), and reversible inactivation of the anterior cingulate prevents cue-, foot shock stress- and cocaine-primed reinstatement in rats (McFarland and Kalivas, 2001; See, 2002). In fact, glutamatergic innervation of the accumbens core via the anterior cingulate is critical for cue-induced reinstatement (Di Ciano and Everitt, 2001). As mGlu₅ receptors (Homayoun et al., 2004) can modulate glutamatergic transmission onto medium spiny neurones, it is possible that striatal mGlu₅ receptors may play a role in relapse to alcohol seeking following abstinence.

Threshold doses of SCH58261 (2.0 mg·kg⁻¹) and SR141716A (0.3 mg·kg⁻¹) did not block relapse-like alcohol seeking, although SR141716A (0.3 mg·kg⁻¹) and MTEP (1.0 mg·kg⁻¹ i.p.) did (Figure 3), suggesting that attenuation of relapse in the current paradigm is mediated primarily by the mGlu₅ receptor. Importantly, MPEP alone (3.0 mg·kg⁻¹ i.p.) reduces cue-induced reinstatement of ethanol-seeking (Backstrom et al., 2004), and MTEP alone (1.0 mg·kg⁻¹ i.p.) reduces operant ethanol self-administration (Cowen et al., 2005b). MPEP blocks nicotine-induced drug-seeking behaviour and reinstatement in Wistar rats (Tessari et al., 2004; Bespalov et al., 2005), and reduces incentive motivational properties of nicotine, cocaine and food (Paterson and Markou, 2005). While SR141716A can prevent reinstatement of alcohol seeking, this typically occurs at doses of 1 mg·kg⁻¹ or greater, depending on rat strain (Cippitelli et al., 2005). In the present study, we used SR141716A at a dose of 0.3 mg·kg⁻¹ during drug combination trials, which in combination with MTEP (1.0 mg·kg⁻¹) did prevent relapse, although the same dose of SR141716A in combination with SCH58261 (2.0 mg·kg⁻¹) had no impact on alcohol seeking. Higher doses of SR141716A were confounded by altered water responding, and therefore not pursued further. Therefore, while cannabinoid CB₁ receptors and adenosine A_2A receptors can regulate drug seeking, it is likely that under the conditions employed in the present study, the mGlu₅ receptor plays a more relevant role in alcohol seeking following abstinence. It would also appear that mGlu₅ receptors do not synergistically interact with CB₁ receptors in this context, perhaps due to the subcellular localization of the receptor types.

One explanation as to why CB₁–mGlu₅ receptor antagonists attenuate cue-conditioned alcohol seeking while CB₁–A_2A antagonists do not is the mGlu₅–NMDA receptor heterodimer, found in medium spiny neurons on the post-synaptic membrane of striatal glutamate terminals (Fuxe et al., 2007). The attenuation of function of this receptor complex would reduce glutamatergic function from the prefrontal cortex, a structure known to modulate reinstatement (Weitlauf and Woodward, 2008), but less involved in self-administration.

There is debate as to the validity of relapse and reinstatement models in animal research. Issues of volition, for example, are difficult to model. A valid model of relapse is significant considering up to 85% of abstinent alcoholics displaying no withdrawal symptoms relapse (Boothby and Doering, 2005). Extinction–reinstatement and cue-conditioned relapse appear to be mediated via differing neural circuitry (Fuchs et al., 2006). Extinction training, perhaps the rodent correlate of human rehabilitation clinic attendance, has questionable construct validity as the vast majority of drug users do not seek help in achieving abstinence (Cunningham, 1999), but rather become ‘spontaneously abstinent’. Thus, forced abstinence may be a more accurate model of the typical human experience. Abstinence also induces incubation of craving, suggested to be an important component of the persisting susceptibility for relapse in humans (Grimm et al., 2001). Unlike many studies using a brief extinction protocol, the current study used a 1 month abstinence period as human relapse is an enduring phenomenon (Epstein et al., 2006).

We found evidence for an apparently additive effect between antagonists of cannabinoid CB₁ and either adenosine A_2A or glutamate mGlu₅ receptors in relation to alcohol self-administration. Combination treatment approaches may potentially reduce doses of individual drugs, and thus minimize off-target effects. We also demonstrate that relapse to alcohol seeking can be precipitated following a period of abstinence, and this appears to be mediated in part by mGlu₅ receptors.

Glossary

Abbreviations:

AGS3

activator of G protein signalling 3

CS+

conditioned stimulus

DARPP-32

dopamine and cAMP-regulated phosphoprotein of 32 kDa

DMSO

dimethyl sulphoxide

ERK

extracellular signal-regulated kinase

iP

Indiana-preferring rat

MPEP

2-methyl-6-(phenylethynyl)-pyridine

MTEP

3-[(2-methyl-1,3-thiazol-4-yl)ethynyl]pyridine

NAcc

nucleus accumbens

NMDA

N-methyl-d-aspartic acid

PKA

protein kinase A

PKC

protein kinase C

S+

unconditioned stimulus

SCH58261

5-amino-2-(2-furyl)-7-phenylethyl-pyrazolo[4,3-e]-1,2,4-triazolo[1,5c]pyrimidine

sP

Sardinian alcohol-preferring rat

SR141716A

N-(piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide hydrochloride

SR147778

5-(4-bromophenyl)-1-(2,4-dichlorophenyl)-4-ethyl-N-(1-piperidinyl)-1H-pyrazole-3-carboxamide

THC

tetrahydrocannabinol