REINSTATEMENT OF ETHANOL AND SUCROSE SEEKING BY THE NEUROSTEROID ALLOPREGNANOLONE IN C57BL/6 MICE

Abstract

Rationale

Recent work in our laboratory documented that the “sipper” method of operant ethanol self-administration produced high ethanol intake and blood ethanol concentrations as well as the typical extinction “burst” in responding under non-reinforced conditions in male C57BL/6 mice. However, the neurochemical basis for reinstatement of responding following extinction has not been examined in mice with this model.

Objectives

Based on findings that the GABAergic neurosteroid allopregnanolone (ALLO) significantly increased the consummatory phase of ethanol self-administration, the present study determined the effect of ALLO on reinstatement of extinguished ethanol-seeking behavior and compared this effect to reinstatement of responding for sucrose reward.

Methods

Separate groups of male C57BL/6 mice were trained to lever press for access to a 10% ethanol (10E) or a 5% sucrose (5S) solution. A single response requirement of 16 presses (RR16) on an active lever resulted in 30 min of continuous access to the 10E or 5S solution. After the animals responded on the RR16 schedule for 14 weeks, mice were exposed to 30 min extinction sessions where responding had no scheduled consequence. Once responding stabilized below the pre-extinction baseline, mice received an IP injection of ALLO (0, 3.2, 5.6, 10 or 17 mg/kg) 15 min prior to the extinction session in a within-subjects design.

Results

ALLO produced a dose-dependent increase in responding under non-reinforced conditions in both the 10E and 5S groups. Additional work documented the ability of a conditioned cue light or a compound cue (light+lever retraction) to reinstate non-reinforced responding on the previously active lever.

Conclusions

These findings definitively show that conditioned cues and priming with ALLO are potent stimuli for reinstating both ethanol and sucrose seeking behavior in C57BL/6 mice.

INTRODUCTION

Alcohol addiction is a complex phenotype that is difficult to model in its entirety with a single preclinical animal model. As a result, basic research has focused on models related to aspects of the addiction phenotype such as the initiation and maintenance of alcohol intake (e.g., Samson and Hodge, 1996; Cunningham et al., 2000; Samson, 2000; Sanchis-Segura and Spanagel, 2006), alcohol withdrawal (e.g., Littleton, 1998; Becker, 2000; Little et al., 2005; Heilig and Koob, 2007), neuroadaptations in dependent animals and during withdrawal (e.g., Morrow, 1995; Crews et al., 1996; Hashimoto and Wiren, 2007; Kalivas and O’Brien, 2008), and genetic models of high alcohol intake or preference (e.g., McBride and Li, 1998; Grahame et al., 1999; Bell et al., 2006; Bice et al., 2006; Crabbe et al., 2006). Within the last two decades, increased attention also has focused on models of alcohol craving and relapse (e.g., Koob, 2000; Littleton, 2000; Lê and Shaham, 2002; Spanagel, 2003; Rodd et al., 2004), given that relapse is a major obstacle in the treatment of human alcoholics. The reinstatement procedure, whereby non-contingent exposure to drug, non-drug stimuli, or stress after extinction can cause an animal to resume a previous drug-reinforced behavior, is one animal model of drug craving (see reviews by Lê and Shaham, 2002; Shaham et al., 2003; Stewart, 2003; Epstein et al., 2006). Validation of the reinstatement procedure as a model of craving, which can lead to induction of alcohol-seeking behavior in humans, is provided by evidence that factors such as re-exposure to drug or drug-associated cues (Ludwig et al., 1974; Jaffe et al., 1989; O’Brien et al., 1992; Epstein and Preston, 2003) and exposure to certain stressors (Brown et al., 1995; Sinha, 2001) can provoke craving and potentially relapse in humans.

Animal models of reinstatement can utilize operant drug self-administration procedures according to the following general scenario: animals are trained to self-administer drug by performing an operant response, this response is extinguished, and then the ability of acute non-contingent exposure to the drug (i.e., drug priming) or of non-drug stimuli (i.e., conditioned stimuli associated with drug intake or stress) to reinstate drug seeking (i.e., responding on the active lever) is determined under extinction (i.e., non-reinforced) conditions (see Lê and Shaham, 2002; Shaham et al., 2003; Epstein et al., 2006). A curious finding is that even though animals will reliably self-administer ethanol, the ability of a priming dose of ethanol to reinstate non-reinforced responding has been described as very modest and highly dependent on the concurrent presentation of ethanol-associated stimuli (Lê and Shaham, 2002) or difficult to reproduce (Nie and Janak, 2003). However, studies aimed at understanding the neurobiology of relapse to alcohol use can circumvent this potential difficulty by utilizing pharmacological agents that generalize to the discriminative stimulus properties of the drug reinforcer (Stewart and de Wit, 1987). Since drug discrimination studies have determined that GABA_A, NMDA and 5-HT₃ receptors contribute to the stimulus properties of ethanol (reviewed in Grant, 1999), one strategy would be to examine pharmacological agents targeting these receptor systems for their effects on alcohol-seeking behavior with the reinstatement model.

The GABAergic neurosteroid allopregnanolone (ALLO) is a potent positive modulator of GABA_A receptors (e.g., Belelli and Lambert, 2005) that is capable of substituting for an ethanol discriminative stimulus (e.g., Ator et al., 1993; Grant et al., 1997; Bowen et al., 1999; Hodge et al., 2001). As another similar feature to ethanol, ALLO and its stereoisomer pregnanolone possess rewarding properties, as measured by conditioned place preference, two-bottle choice preference drinking, and intravenous self-administration (Finn et al., 1997; Sinnott et al., 2002a; Rowlett et al., 1999). These related properties led to a number of studies examining the modulatory effects of GABAergic neurosteroids on ethanol reinforcement and consumption. Notably, systemic and intracerebroventricular administration of ALLO significantly increased two-bottle ethanol preference drinking and operant ethanol self-administration in male mice and rats by promoting the onset of self-administration and blunting the maintenance of consumption during the latter portion of the limited access sessions (Janak et al., 1998; Sinnott et al., 2002b; Ford et al., 2005b, 2007b). This modulatory effect of ALLO was selective for ethanol in rats (Janak and Gill, 2003), whereas ALLO also increased consumption of a saccharin solution in mice (Sinnott et al., 2002b). In contrast, sub-chronic (7 day) treatment with the 5α-reductase inhibitor finasteride, which blocks the biosynthesis of endogenous ALLO and other GABAergic neurosteroids, suppressed the onset of ethanol self-administration (Ford et al., 2005a, 2008). Notably, pre-treatment with finasteride to mice with established consumption patterns produced a transient suppression of ethanol drinking, whereas it dose-dependently blunted the acquisition of ethanol intake when it was administered prior to the inaugural experience with ethanol consumption. Collectively, these findings suggest that endogenous GABAergic neurosteroid tone may influence the regulatory processes governing ethanol intake.

Recent work in our laboratory documented that the “sipper” method of operant ethanol self-administration produced high ethanol intake and blood ethanol concentrations as well as the typical extinction “burst” in responding under non-reinforced conditions in male C57BL/6 mice (Ford et al., 2007a). Work in other laboratories determined that conditioned stimuli and the alcohol self-administration context could reinstate non-reinforced responding on the alcohol-associated lever in C57BL/6 mice and in rats (Nie and Janak, 2003; Burattini et al., 2006; Tsiang and Janak, 2006; Zironi et al., 2006). Finally, priming injections of the GABAergic neurosteroid ALLO dose-dependently reinstated previously extinguished responding for ethanol, but not for sucrose, in rats (Nie and Janak, 2003), but similar studies have not been conducted in mice. Thus, the primary purpose of the present studies was to determine the effect of ALLO priming doses on reinstatement of extinguished ethanol-seeking behavior in C57BL/6 mice and to compare this effect to reinstatement of responding for sucrose reward. In order to characterize reinstatement following extinction of ethanol (or sucrose) self-administration with the “sipper” model, subsequent studies determined whether a priming dose of ethanol or conditioned stimuli (CS) could reinstate ethanol-seeking or sucrose-seeking behavior in mice.

METHODS

Animals

Twenty-four male C57BL/6 mice were purchased from the Jackson Laboratory West (Davis, CA) at 6 weeks of age. Mice were individually housed and acclimated to a normal light/dark schedule (12hr/12hr; lights on at 600 hrs) for two weeks prior to the onset of operant training. Food was available ad libitum in the home cage. With the exception of the initial sipper training (noted below), mice had ad libitum access to water in the home cage. Animals were weighed and handled daily. All instrumental training and test sessions were conducted between 1300 and 1600 hrs. The local Institutional Animal Care and Use Committee approved all experimental methods and procedures in accordance with the guidelines described in the Guide for the Care and Use of Mammals in Neuroscience and Behavioral Research (National Research Council of the National Academies, 2003).

Apparatus

Daily sessions were run in modular operant conditioning chambers (21.6 × 17.8 × 12.7 cm) with stainless steel grid floors (Med-Associates Inc., St. Albans, VT), as described previously (Ford et al., 2007a, 2007b). In brief, one wall of each chamber was outfitted with a house light, two ultra-sensitive retractable levers, and two stimulus lights (light-emitting diodes; one located above each lever). On the opposing wall a portal was positioned for access to a retractable sipper apparatus that contained a modified 10 ml graduated pipette attached to a metal, double ball bearing sipper. Lickometers were interfaced to an IBM compatible computer operated by MED-PC software (Med-Associates Inc.). Each chamber was placed within a sound-attenuating wooden cabinet (51 × 38 × 33 cm; Fisher Custom Woodworking; Portland, OR) that contained an exhaust fan to facilitate proper chamber ventilation.

Lever Acquisition Phase

The purpose of the acquisition phase was to establish a relationship between an instrumental response (i.e., responding on the active lever) and its consequences (i.e., retraction of both levers, the concomitant activation of a stimulus light and inactivation of the house light for a 5-sec duration, followed by the presentation of a sipper tube containing a 10% w/v sucrose solution). All mice were initially trained to respond on a fixed ratio (FR)-1 schedule of reinforcement for 10% sucrose (10S) during 60-min sessions. Throughout the initial 5 days of training, sipper access periods were reduced from 60- to 30-sec between sessions. Mice were also provided a single 15-hr overnight session in the test chambers. These manipulations were conducted to promote higher levels of total responding and to accelerate training. After the first 5 days, the duration of the daily sessions was reduced to 30-min. Mice were water restricted for 16 hrs prior to the initial 13 sessions to enhance their motivational state, but were then provided ad libitum water in the home cage during all subsequent experimental phases. A minimum acquisition criterion of ten 30-sec sipper presentations per session was achieved by the fifth training day, and the removal of water restriction did not alter the mean number of 10S reinforcers earned. Responding on an inactive lever was recorded during the lever acquisition phase, but had no consequence. The locations of the active and inactive levers were counterbalanced across chambers.

Sucrose Fading and Reinforcement Schedule Alteration

Mice were assigned to one of two groups: One group was trained to lever press for ethanol reinforcement (10% v/v ethanol solution; 10E) and the second group was trained to lever press for sucrose reinforcement (5S solution). For animals assigned to the 10E group (n = 10), consumption of the 10% ethanol solution was accomplished by progressively reducing the sucrose concentration in the following manner via a modification of the sucrose fading procedure: After exposure to a 10E/10S solution for two sessions, fading in the 10E group consisted of step-wise changes to 10E/5S (3 sessions), 10E/2.5S (2 sessions), and finally to 10E. The sucrose-reinforced group (n = 14) was concurrently faded from 10S to 8.5S (3 sessions), 7S (2 sessions), and finally to 5S. The reinforcement schedule remained at FR-1 followed by a 30-sec sipper presentation for both groups throughout the fading procedure. The 5S group remained ethanol-naïve throughout the sucrose fading procedure and during all subsequent self-administration sessions.

Over the ensuing 5 weeks, the reinforcement schedule for both the 10E and 5S groups was incrementally increased from FR-1 to FR-8 in blocks of 2–3 sessions each, and the sipper presentation time was gradually lengthened from 30-sec to 720-sec. The appetitive and consummatory phases of operant self-administration were then procedurally separated such that the completion of a single response requirement of 8 lever presses (RR-8) resulted in 30-min of continuous access to either the 10E or 5S solution. That is, completion of a response requirement of 8 presses (RR-8) on the active lever resulted in the immediate retraction of both levers, the concomitant activation of a stimulus light and inactivation of the house light for a 5-sec duration, followed by the presentation of a sipper tube containing either the 10E or 5S solution for 30-min. Throughout an additional 2-week period, the response requirement was elevated from RR-8 to RR-16. A limit of 20-min was imposed for the fulfillment of the response requirement. Mice consistently completed the RR under this condition and were provided reinforcer access.

Mice were maintained on the RR-16 schedule for 14 weeks. At the conclusion of this maintenance period, the consumed doses of ethanol and sucrose during the 30-min sessions were 0.95 ± 0.10 g/kg and 1.98 ± 0.16 g/kg (mean ± SEM), respectively.

Extinction and Reinstatement Procedures

Mice were exposed to 30-min extinction sessions during which responding on the active and inactive levers had no scheduled consequence. By the tenth extinction session, mean response frequencies on the previously reinforced active lever were consistently below the pre-extinction baseline (i.e., < 16 responses) in both the 10E and 5S groups.

Mice were tested for the ability of pharmacological manipulations or conditioned cues to reinstate extinguished ethanol- and sucrose-reinforced responding in three experimental phases (Table 1). A within-subjects design was employed. Baseline extinction responding was re-established (i.e., non-reinforced responding on the previously active lever was confirmed to be below the pre-extinction baseline; ~ 11 lever presses) between the various drug doses and conditioned cue presentations throughout each experimental phase. For the pharmacological manipulations, extinction sessions with vehicle (VEH) injection pretreatments were conducted between each drug dose test until stable extinction levels of non-reinforced responding were re-established (typically 2–3 sessions). A collapsed baseline (0 mg/kg drug) of responding was calculated from the average of VEH injection sessions that immediately preceded each drug test session. For all experimental phases, average extinction baseline performance was ≤ 11 responses on the previously active lever for the 5S and 10E groups with < 15% variability across 3 days of responding under extinction baseline conditions.

TABLE 1.

Reinstatement Experimental Procedures

Reinstatement Test	Drug Dose/CS Exposure	Pre-treatment
Phase 1 = ALLO dose response	3.2, 5.6, 10 & 17 mg/kg; VEH on intervening days	15-min pretreatment
Phase 2 = Ethanol dose response	0.5, 1.0 & 2.0 g/kg; VEH on intervening days 0.5 g/kg; VEH on intervening days	15-min pretreatment 5-min pretreatment
Phase 3 = CS exposure	stimulus light cue & compound cue (light+lever retraction; no injection on intervening days	CS presented on FR-1 schedule concomitant with previous active lever responding

Separate groups of ethanol- or sucrose-trained mice were stably self-administering 10E (0.95 ± 0.10 g/kg; n = 10) or 5S (1.98 ± 0.16 g/kg; n = 14) for 14 weeks prior to extinction onset. Then, mice underwent extinction during which responding on the active and inactive levers had no scheduled consequence. Once mean response frequencies on the previously reinforced active lever were consistently below the pre-extinction baseline (< 16 responses, typically ~ 11 responses), mice were tested for the ability of pharmacological manipulations or conditioned cues to reinstate extinguished ethanol- and sucrose-reinforced responding in three experimental phases. Baseline extinction responding was re-established between the various drug doses or conditioned cue presentations. Animals were never given the unconditioned stimulus (i.e., 5S or 10E) again, so multiple extinction curves were not conducted between experimental phases.

In the first experimental phase, the ability of ALLO to reinstate extinguished ethanol- and sucrose-reinforced responding was evaluated. Mice were acclimated to injections of 20% w/v β-cyclodextrin (VEH), administered as 15-min pretreatments. Once stable extinction performance was re-established (see above for criterion), all mice received injections of ALLO (3.2, 5.6, 10 and 17 mg/kg; intraperitoneal, IP) in a within-subject design throughout a 2-week period. ALLO doses were tested in ascending order. ALLO doses were selected on the basis of previous reports documenting the influence of this neurosteroid on the reinstatement of ethanol-seeking behavior in male rats (Nie and Janak, 2003) and the modulation of ethanol drinking patterns in male C57BL/6 mice (Ford et al., 2005b).

In a second experimental phase, the ability of a priming dose of ethanol to reinstate responding after extinction of 10E- and 5S-reinforced responding was evaluated. Operant performance under extinction conditions was stabilized following saline injection (VEH) pretreatment that was administered 15-min prior to session onset. Over a 2-week period multiple ethanol doses (0.5, 1.0, and 2.0 g/kg; IP) were assessed in a within-subject design. The 0.5 g/kg dose was comparable to that previously employed to reinstate ethanol-reinforced responding in rats (Lê et al., 1998). The 1.0-g/kg dose of ethanol was selected because it closely approximated the orally self-administered dose exhibited by the 10E-reinforced mice prior to extinction onset. In order to test ethanol-priming effects earlier on the rising limb of the ethanol distribution curve, the 0.5 g/kg dose was re-tested following its administration at 5-min prior to session start. The 5-min pretreatment was the earliest time point that was technically possible to utilize with our operant procedure.

In a third experimental phase, the ability of conditioned cues to reinstate non-reinforced responding was examined in the 10E and 5S groups. Over a 2-week period, two manipulations were tested for their ability to reinstate ethanol- and sucrose-appropriate responding as follows: first, a stimulus light cue (positioned above the active lever), and then a combination of a stimulus light cue plus dual-lever retraction (‘compound cue’). During a single test session, the stimulus light cue (5-sec illumination; see lever acquisition phase above) was presented on a FR-1 schedule concomitant with responding on the previously active lever. Five sessions in the absence of the stimulus light cue were then conducted to re-establish baseline levels of extinction responding. During the ‘compound cue’ test session, the stimulus light cue was again presented on a FR-1 schedule in conjunction with retraction of both levers for 5-sec (as described in the lever acquisition phase above). No injections were administered throughout this experimental phase.

Drug Solutions

Ethanol (200 proof; Pharmco Products, Inc., Brookfield, CT) and sucrose (Sigma-Aldrich Company, St. Louis, MO) drinking solutions were constituted in tap water. A 20% v/v ethanol solution made in physiological saline was used for systemic injections. Allopregnanolone (3α-hydroxy-5α-pregnan-20-one; ALLO) was synthesized by and purchased from Dr. Robert H. Purdy (Veterans Medical Research Foundation, San Diego, CA). ALLO was solubilized in a 20% w/v 2-hydroxypropyl-β-cyclodextrin (Cerestar USA, Inc., Hammond, IN) solution, diluted in saline and injected in a volume of 0.01 ml/gm body weight.

Data Analysis

Consumed doses (g/kg) of 5S and 10E solutions were calculated based on the volume depleted (to the nearest 1/20 ml) following the 30-min access period. Cumulative records of responding on the active and inactive levers also were acquired. Due to the within-subjects design, one-way repeated measures ANOVAs were used to evaluate the influence of pharmacological manipulations and CS presentations on non-reinforced responding on the previously assigned active and inactive levers. Mice trained with 5S and 10E were evaluated separately, since they were treated as independent groups. Where appropriate, pair-wise differences were determined by Tukey’s or Dunnett’s post-hoc tests. Since we wished to compare the extinction time course in the 5S and 10E trained mice, two-way ANOVA with time as a repeated measures factor examined responding on the previously active and inactive levers. Linear regression analysis also was used to calculate the slope of the first six sessions of the initial extinction curve for the 5S and 10E mice, which were compared with a t-test. For all analyses, statistical significance was set at P ≤ 0.05.

RESULTS

Extinction time course

Analysis of non-reinforced responding on the previously active lever with two-way repeated measures ANOVA indicated that there was a significant influence of extinction session [F(11,242) = 87.18; P < 0.001] and a significant group X session interaction [F(11,242) = 9.56; P < 0.001] (Figure 1A). Responding on the previously active lever was significantly different between the 5S and 10E groups on extinction sessions 1 (P < 0.001), 4 (P < 0.01), and 5 (P < 0.05). The slope of the extinction curve across the first 6 sessions tended to differ between the 5S (−20.58 ± 5.92) and 10E (−9.49 ± 2.45) groups [t(10) = 1.73, P = 0.11].

FIGURE 1. Time Course of Responding During Extinction.

Responses on the previously active (panel A) and inactive (panel B) levers are shown for 5S-trained (open squares; n = 14) and 10E-trained (black circles; n = 10) mice. Baseline responding prior to extinction onset was 16 presses on the active lever for all mice and averaged 0.37 ± 0.13 and 1.18 ± 0.22 inactive lever presses for the 5S and 10E groups, respectively. Responding in both groups dropped below the extinction criterion (< 16 responses on the previously active lever) by the 10^th extinction session.

^#P < 0.05, ^##P < 0.02, ^###P < 0.001; between group differences within a given session

***P < 0.001; versus within-group extinction session 1 values for both 5S and 10E groups

Active lever responses were increased by 7.2-fold in the 5S group and by 4-fold in the 10E group during the first extinction session, when compared to baseline responding prior to extinction (i.e., 16 presses for all mice in both groups). Post-hoc tests confirmed that both groups of mice demonstrated a precipitous decline in responding on the previously active lever between the first day of extinction and all subsequent extinction sessions (P < 0.001 for every case). Furthermore, by extinction session 10, both groups were emitting less than the 16 responses on the active lever that previously gained them access to the 5S and 10E solutions (i.e., RR-16 schedule). The mean ± SEM active lever responses during extinction session 12 were 12.1 ± 2.3 and 14.0 ± 1.5 for the 5S and 10E groups, respectively.

Although a two-way repeated measures ANOVA revealed a significant influence of extinction session for non-reinforced responding on the previously inactive lever [F(11,242) = 2.25; P < 0.05], no significant pair-wise comparisons were found (Figure 1B). Effects of group and a group X session interaction were absent, thereby demonstrating that extinction of responding occurred exclusively on the previously active lever.

Effect of ALLO on reinstatement of ethanol and sucrose seeking

Pre-treatment with ALLO increased non-reinforced lever press responding in both the 5S- and 10E-trained groups (Figure 2). A within-subjects repeated measures ANOVA found a significant influence of ALLO dose on previously active lever responding in the 5S [F(4,52) = 3.33; P < 0.05] and 10E [F(4,36) = 4.39; P < 0.01] groups. Subsequent analyses revealed that the 10-mg/kg dose of ALLO significantly enhanced previously active lever responses in the 5S group by 2.2-fold (P < 0.05) and in the 10E group by 1.7-fold (P < 0.05), when compared to the respective VEH (baseline) treatments (Figure 2A). There also was a significant effect of ALLO pretreatment on inactive lever responding in the 5S [F(4,52) = 5.59; P < 0.001] and 10E [F(4,36) = 4.38; P < 0.01] groups. However, the only significant pair-wise comparison was found in the 5S group, whereby the 17 mg/kg ALLO dose significantly increased inactive lever responses versus VEH baseline levels (Figure 2B).

FIGURE 2. ALLO reinstates extinguished ethanol- and sucrose-reinforced responding.

Responses on the previously active (panel A) and inactive (panel B) levers are shown for the 5S (left side of each panel; n = 14) and 10E (right side of each panel; n = 10) groups of animals that were pretreated with ALLO (3.2, 5.6, 10 and 17 mg/kg, IP) at 15 min prior to extinction session onset. Baseline (0 mg/kg, VEH) responding represents the collapsed average of the extinction sessions following VEH injection that immediately preceded each test session with ALLO.

*P < 0.05, **P < 0.01 versus within-subject baseline

Effect of ethanol on reinstatement of ethanol and sucrose seeking

In contrast to the effect of ALLO, a 15 min pretreatment with a priming dose of ethanol had a general suppressive effect on previously active lever-press responding in the 5S group (0.5, 1.0 and 2.0 g/kg), whereas only the highest ethanol dose suppressed responding in the 10E group (2.0 g/kg; see Figure 3A). There was a significant influence of ethanol priming dose on previously active lever responses in the 5S- [F(3,39) = 8.05; P < 0.001] and 10E- [F(3,27) = 7.00; P < 0.001] groups. The 5S group was more sensitive to the suppressive effect of ethanol pretreatment on extinction responding, as each ethanol dose tested significantly suppressed this measure by greater than 50% (P < 0.001 for 0.5 g/kg; P < 0.01 for 1.0 and 2.0 g/kg), when compared to baseline (Figure 3A). In contrast, only the 2.0 g/kg ethanol dose significantly suppressed responses on the previously active lever in the 10E group (P < 0.001). Responding on the inactive lever also was affected by ethanol priming dose in the 10E group [F(3,27) = 3.41; P < 0.05], but not in the 5S group (Figure 3B). In particular, the 2.0 g/kg dose attenuated inactive lever responses in the 10E group by 55% (P < 0.05).

FIGURE 3. Ethanol priming does not reinstate extinguished ethanol- and sucrose-reinforced responding.

Responses on the previously active (panel A) and inactive (panel B) levers are shown for the 5S (left side of each panel; n = 14) and 10E (right side of each panel; n = 10) groups of animals that were pretreated with ethanol (0.5, 1.0 and 2.0 g/kg, IP) at 15 min prior to extinction session onset. Baseline (0 mg/kg, VEH) responding represents the collapsed average of the extinction sessions following VEH injection that immediately preceded each test session with ethanol.

*P < 0.05, **P < 0.01, ***P < 0.001 versus within-subject baseline

We also examined the effect of a 5 min pretreatment with a 0.5 g/kg priming dose of ethanol (or saline) on responding under non-reinforced conditions. Responding on either lever was not altered significantly in the 5S or 10E groups by the 5 min pretreatment with the 0.5 g/kg ethanol dose (data not shown). Thus, the 0.5-g/kg priming dose did not suppress responding in the 10E-trained group with either pretreatment, whereas it only suppressed responding in the 5S-trained group with the 15 min pretreatment.

Cue-induced reinstatement of ethanol and sucrose seeking

During this phase of testing, the ability of conditioned cues to reinstate non-reinforced responding in the 5S and 10E groups was examined. A within-subjects repeated measures ANOVA determined that the light CS significantly increased non-reinforced responding on the previously active lever in the 5S [F(1,13) = 31.28; P < 0.001] and 10E [F(1,9) = 16.19; P < 0.01] groups (Figure 4). Subsequent analyses confirmed that the presentation of the light CS significantly increased responding on the previously active lever by 2.25-fold in the 5S group (P < 0.001) and by 2.4-fold in the 10E group (P < 0.01), when compared to their respective baseline extinction responding (Figure 4A). Presentation of the light CS did not affect previously inactive lever responses in either group (Figure 4B).

FIGURE 4. Cue-induced reinstatement of extinguished ethanol- and sucrose-reinforced responding.

Responses on the previously active (panels A & C) and inactive (panels B & D) levers are shown for the 5S (left side of each panel; n = 14) and 10E (right side of each panel; n = 10) groups of mice following presentation of a stimulus light cue (panels A & B) or a compound cue consisting of the light+lever retraction (panels C & D).

*P < 0.05, **P < 0.01, ***P < 0.001 versus within-subject baseline

Presentation of the compound CS (light+lever retraction) also significantly increased non-reinforced responding on the previously active lever in the 5S [F(1,13) = 76.12; P < 0.001] and 10E [F(1,9) = 11.74; P < 0.01] groups (Figure 4C). The compound cue produced a greater effect on reinstatement of non-reinforced responding than that produced by the light cue alone, since responding on the previously active lever was increased by 5.5-fold in the 5S group (P < 0.001) and by 3.9-fold in the 10E group (P < 0.01), versus their respective baseline. Unlike the light CS, the compound CS also increased non-reinforced responding on the previously inactive lever in the 5S [F(1,13) = 17.20; P < 0.001] and 10E [F(1,9) = 8.37; P < 0.05] groups. Presentation of the compound CS increased previously inactive lever responses by 3.3-fold in the 5S group (P < 0.001) and by 2-fold in the 10E group (P < 0.05) versus their respective baselines (Figure 4D).

DISCUSSION

The present studies are the first characterization of reinstatement following extinction of ethanol and sucrose self-administration with the “sipper” method of operant self-administration in mice. The results indicated that a priming dose of the GABAergic neurosteroid ALLO or exposure to a CS previously paired with reinforcer delivery reinstated ethanol-seeking and sucrose-seeking behavior in mice. The majority of evidence supporting GABAergic involvement in the regulatory processes governing ethanol intake (Rassnick et al., 1993; Hodge et al., 1995, 1996; Hyytiä and Koob, 1995; Petry, 1997; Shelton and Balster, 1997; Janak et al., 1998; June et al., 1998a, 1998b; Nowak et al., 1998; Soderpalm and Hansen, 1998; Samson and Chappell, 2001; Sinnott et al., 2002b; Janak and Gill, 2003; Ford et al., 2005a, 2005b, 2007b; Krystal et al., 2006) indicate a relatively selective effect of GABA_A receptor ligands on ethanol consumption, although benzodiazepines also have been reported to alter saccharin and food consumption in rats (e.g., Wegelius et al., 1994; Shelton and Balster, 1997). Collectively, the present findings in mice support a role of GABAergic neurosteroids such as ALLO in general motivation.

A priming dose of ALLO reinstated responding on the previously active lever in an inverted-U function for both the 10E and 5S groups. Whereas the 10-mg/kg ALLO dose significantly increased non-reinforced responding on the previously active lever in both groups, the 17-mg/kg dose did not. However, the 17-mg/kg dose significantly increased previously inactive lever responses only in the 5S group. Thus, ALLO selectively increased non-reinforced responding on the active lever in the ethanol-trained mice.

The ability of ALLO to reinstate extinguished responding for both ethanol and sucrose is consistent with earlier work in male C57BL/6 mice whereby ALLO increased limited access saccharin and ethanol consumption (Sinnott et al., 2002b). Consistent with this finding, several studies have found that ALLO produces hyperphagia in mice and rats (e.g., Chen et al., 1996; Reddy and Kulkarni, 1998, 1999; Fudge et al., 2006). However, it should be noted that the present findings differ from a recent report, documenting an ALLO dependent reinstatement of ethanol but not sucrose responding in male rats (Nie and Janak, 2003). This discrepancy may reflect species or procedural differences. Although additional experiments are needed to resolve this definitively, it is clear from our data and others that ALLO affects both appetitive processing and consummatory phases of ethanol and natural rewards.

It can be argued that the increase in previously inactive lever responses in the 5S group, which is a common observation during extinction testing regardless of the original reinforcer, reflects a “re-direction” of the animal’s behavior toward alternative responses that may achieve a reward. In this scenario, the increase in inactive lever response following ALLO treatment would be consistent with a selective induction of an alternative behavior directed at achieving the original sucrose reward. It is interesting that we did not observe the same pattern of inactive lever responding in the ethanol group. In contrast to the results in the 5S group, animals extinguished from ethanol reward exhibited the same pattern of inactive lever responding as on the previously active lever (i.e. an inverted-U function). This could be interpreted as a dose-response shift to the left for the effect of ALLO on ethanol compared to sucrose reward. Additional studies will be needed to determine if this is the case.

Ethanol priming did not reinstate extinction responding. In contrast, ethanol priming doses as low as 0.5 g/kg significantly suppressed responding on the previously active lever in the 5S group. In the 10E group, we observed a dose-dependent suppression of responding, with the 2 g/kg ethanol injection significantly reducing previously active lever presses. This finding contrasts with the effects of stimulants and opiates, where priming doses consistently reinstate self-administration behavior (e.g., de Wit and Stewart, 1981; Kantak et al., 2002; Kalivas and McFarland, 2003; Shaham et al., 2003). However, our results were not surprising, given that the ability of systemic ethanol injections to prime reinstatement only has been reported in two laboratories (Lê et al., 1998, 1999; Vosler et al., 2001), with effects that are very modest (Lê and Shaham, 2002) and difficult to reproduce (Nie and Janak, 2003). Consistent with this idea, recent work found that an oral priming dose of ethanol was effective at reinstating ethanol seeking only in the presence of an ethanol-associated CS (Bäckström and Hyytiä, 2004). Taken in conjunction with the conflicting data on ethanol priming, the present observation that neither of the ethanol pretreatment times tested (i.e., 5 or 15 min) reinstated non-reinforced lever pressing is consistent with the idea that the temporal relationship between the lever-press behavior and the pharmacological effects of ethanol following oral self-administration is not as tightly coupled as that with intravenous self-administration of drugs like cocaine or heroin (e.g., Meisch, 2001).

The ethanol-naïve 5S group was more sensitive to ethanol’s suppressive effect on previously active lever responses during extinction than the ethanol-experienced 10E group. It is possible that this difference in sensitivity to ethanol’s suppressive effect on non-reinforced lever responses was due to the development of tolerance to the motor incoordinating effects of the lower ethanol priming doses that were tested in the 10E group (i.e., 0.5 and 1.0 g/kg), as this group stably self-administered approximately 1 g/kg ethanol prior to extinction. However, when we tested an ethanol pretreatment time (5 min) that more closely mimicked the rising limb of the ethanol distribution curve, the 0.5 g/kg ethanol-priming dose did not suppress non-reinforced lever responses in the 5S group. This finding is consistent with a previous report that a 0.5 g/kg priming dose of ethanol did not alter active or inactive lever responses in sucrose-trained rats (Vosler et al., 2001).

The light CS and compound CS (light+lever retraction) that were associated with 5S and 10E reinforcement markedly increased responding under non-reinforced conditions to approximately 50% of the levels that were observed at the onset of extinction (compare Figure 4A & 4C with session 1 in Figure 1A). The compound cue produced a greater increase in reinstatement compared to the light CS alone and also significantly elevated previously inactive lever responses in both the 5S and 10E groups. The ability of conditioned cues that are associated with ethanol self-administration or the environmental context to induce ethanol-seeking behavior has been well documented (e.g., Katner and Weiss, 1999; Lê and Shaham, 2002; Liu and Weiss, 2002, 2004; Nie and Janak, 2003; Bäckström and Hyytiä, 2004; Burattini et al., 2006; Tsiang and Janak, 2006; Zironi et al., 2006). Recent work demonstrated that re-exposure of male rats (Burattini et al., 2005; Zironi et al., 2006) and C57BL/6 mice (Tsiang and Janak, 2005) to the self-administration context following extinction in a separate context reinstated responding on the previously ethanol-reinforced lever. And, the ability of the ethanol self-administration context to support ethanol-seeking behavior was maintained over 3 weeks (Zironi et al., 2006). In the present study, mice were tested for reinstatement following presentation of a light CS and a compound CS after the animals had been maintained under extinction conditions for more than two months. While it is possible that the extended period of withdrawal from ethanol self-administration affected reinstatement behavior, particularly during ethanol priming, the present results demonstrate that conditioned cues can have a persistent impact on ethanol-seeking behavior.

In conclusion, the present study is the first to characterize reinstatement following extinction of ethanol and sucrose self-administration with the “sipper” method of operant self-administration in C57BL/6 mice and to demonstrate that conditioned cues reliably reinstated ethanol- and sucrose-seeking behavior with this procedure. Importantly, the present findings are the first report of the effects of a GABAergic neurosteroid on ethanol and sucrose seeking in the mouse. In conjunction with previous work, the ability of ALLO to reinstate ethanol- and sucrose-seeking behavior in male C57BL/6 mice demonstrates the potential importance of ALLO in facilitating the appetitive and consummatory phases underlying self-administration of ethanol and sweet solutions in mice. Given that genetic mouse models are gaining interest to test mechanistic hypotheses related to ethanol self-administration and reinstatement, the present findings with neurosteroids are important. Considering that relapse to ethanol use is a major obstacle in the treatment of alcoholism and that ethanol is often consumed in sweetened solutions, GABAergic neurosteroids such as ALLO may be an important target for the prevention of ethanol relapse.