Forty female Wistar rats (Harlan, Indianapolis, IN, USA) weighing 250–320 g at the time of surgery were used in experiment 1, and were individually housed and kept on a 12-h light–dark cycle (lights on at 09.00 h). Seventy-three male Wistar rats (Harlan, Dublin, VA, USA) weighing 280–350 g at the time of surgery were used in experiments 2–6, and were individually housed and kept on a reversed 12-h light–dark cycle (lights on at 21.00 h). Food and water were available ad libitum except during testing. Constant temperature and relative humidity were maintained in the colony and experimental rooms. The procedures were approved by the Animal Care and Use Committees of the institutions and were in accordance with National Institutes of Health guidelines.

In experiment 1, a unilateral 22-gauge guide cannula was implanted in rats under halothane anaesthesia, aimed at the DRN, MRN and surrounding regions using the following coordinates (posterior distance from bregma, lateral distance from the midline and ventral distance from skull surface were measured along the trajectory of the angled cannula). These respective coordinates (in mm), purposefully varied to survey in and around these nuclei, were: 6.8–8.0, 2.5–4.0, 5.0–6.9 with 20° for the DRN and its surrounding regions, and 6.8–8.0, 1.5–3.0, 7.8–8.4 with 10° for the MRN and its surrounding regions. While not in use, a 28-gauge stylet maintained the patency of the guide cannula. The stylet extended 0.5 mm beyond the tip of the guide. In experiments 2–6, rats were anaesthetized with sodium pentobarbital (31 mg /kg, i.p.) and chloral hydrate (142 mg /kg, i.p.) and implanted with a unilateral 24-gauge guide cannula that ended 1.0 mm above the target site. The target coordinates were 7.4, 1.6, 8.0 and 10° angle for the MRN, and 7.4, 2.6, 6.2 and 20° angle for the DRN. These stereotaxic surgeries were performed using the flat-skull method, in which bregma and lambda lie in the same coronal plane. Testing began 5–7 days after surgery.

The GABA_A agonist muscimol hydrobromide and the GABA_A antagonist picrotoxin (Sigma Chemical Co., St Louis, MO, USA) were dissolved in an artificial cerebrospinal fluid consisting of (in mM) NaCl, 120.0; KCl, 4.8; KH₂PO₄, 1.2; MgSO₄, 1.2; NaHCO₃, 25.0; CaCl₂, 2.5; and d-glucose, 10.0 (experiment 1); or NaCl, 148; KCl, 2.7; CaCl₂ 1.2; and MgCl₂, 0.85 (experiments 2–6). When necessary, pH levels were adjusted to 7.3 ± 0.2 with NaOH or HCl. The D₁ receptor antagonist R(+)-7-chloro-8-hydroxy-3-methyl-1-phenyl-2,3,4,5-tetrahydro-1H-3-benzazepine hydrochloride (SCH 23390; Sigma) was dissolved in 0.9% saline.

Each rat was placed individually in the operant chamber. To minimize air trapped at the tip of the injection cannula, an infusion current was delivered for 5 s as the injection cannula was inserted into the guide cannula. Depression of the ‘active’ lever resulted in the delivery of 100 nL infusate over a 5-s period followed by a time-out period (55 s), during which depression of the active lever produced no infusion. The time-out period was instituted to avoid continuous delivery of infusions, which may result in adverse drug effects. Depression of the ‘inactive’ lever had no programmed consequence. The assignment of left and right levers for active and inactive functions was counterbalanced among subjects and remained the same for each rat throughout the experiment. No cue was provided to signal the delivery or availability of infusions. No shaping technique was used to facilitate the acquisition of lever responses. The numbers of infusions and responses on the active and inactive levers were recorded, including lever presses during the time-out period. Over four sessions, each rat received four different concentrations of muscimol: 0, 25, 50 or 100 μM (i.e. 0, 2.5, 5 or 10 pmol or 0, 0.5, 1 or 2 ng per 100-nL infusion). The order of testing these concentrations was counterbalanced among rats. The numbers of responses on each lever and subsequent infusions were recorded. Sessions lasted 180 min or until the rats received a total of 40 infusions. The total number of possible infusions was limited to minimize possible adverse drug effects.

Each rat was placed in the operant chamber with two retractable levers. A response on the active lever resulted in a 5-s infusion (75 nL), turned on a tone, extinguished the light above the lever and retracted the active and inactive levers for 20 s. A response on the inactive lever retracted both the active and inactive levers for 20 s and resulted in no other programmed consequences. The assignment of left and right levers for active and inactive functions was counterbalanced among subjects and remained the same for each rat throughout the experiment. Each rat received vehicle in session 1 and 100 μM muscimol (i.e. 7.5 pmol or 1.5 ng per 75-nL infusion) in sessions 2–5. Sessions lasted 90 min or until the rats received a total of 60 infusions. Because we did not observe adverse effects of the drug in experiment 1, we made the total number of infusions greater (60) and timeout period shorter (20 s) in this experiment than experiment 1.

Experimentally naïve rats were placed in the operant chamber with two retractable levers, and received 100 μM muscimol (75 nL) into the MRN over four sessions. The procedure of the first two sessions consisted of key features similar to those of experiment 1: a response on the active lever delivered an infusion of over 5 s followed by a 55-s time-out period, whereas a response on the inactive lever resulted in no programmed consequence. Responding on the levers did not result in lever retraction or provide any programmed cues for infusions. Sessions 3 and 4 used the same two-lever procedure described in experiment 2.

The place-conditioning chamber (MED Associates) consisted of two compartments (21 × 21 × 28 cm³) and an area (21 × 21 × 12.5 cm³) connecting the compartments; a guillotine door separated each compartment from the connecting area. One compartment differed from the other in wall colour (black vs. white), floor type (net vs. grid) and lighting; the amount of light was modulated in each compartment so that rats did not prefer one compartment to the other prior to place conditioning. In session 1, experimentally naïve rats were placed individually in the chamber with guillotine doors open for 15 min without any treatment; each rat had access to both compartments, and the time spent in each compartment was recorded. In sessions 2–5, each rat was placed in a cylinder and received an intracranial injection (500 nL delivered over 60 s via a 31-gauge injection cannula extending 1 mm below the tip of the guide). An additional 30-s period elapsed before the injection cannula was removed. Each rat was then placed in one of the compartments with guillotine doors closed and remained there for 30 min; the placement into the two compartments was alternated over sessions. Muscimol (100 μM) injections were paired with one compartment and vehicle injections with the other for the MRN and DRN muscimol groups. Vehicle injections were paired with both compartments for the MRN vehicle group. The order of drug treatments and the assignment of the compartments with drug treatments were counterbalanced among the subjects for each group. In session 6, the rats were placed individually in the chamber without an injection and given access to both compartments; the time spent in each compartment was recorded for 15 min. Sessions were separated by 24 h. Place preference scores were obtained by subtracting time spent in vehicle-paired compartment from time spent in drug-paired compartment. In order to obtain place preference scores in the vehicle group, one of the compartments was assigned as ‘drug-paired’ before obtaining data, even though the compartment was paired with vehicle for this group.

The subjects for this experiment were the same rats that completed experiment 2. Two of the 17 rats were eliminated because their self-administration rates of 100 μM muscimol into median raphe nucleus were < 0.2 infusions per minute and therefore they were classified as nonresponders. Using the two-lever procedure described for experiment 2, the rats received vehicle, 100 μM muscimol, and 100 μM muscimol plus 100 μM picrotoxin over three sessions. The order of testing these drugs was counterbalanced.

The subjects for this experiment were the same rats that completed experiment 3. Two of the 13 rats were eliminated for the same reason mentioned for experiment 5 (see above). Results from two additional rats were eliminated from the data analysis because of low infusion number (< 3) when muscimol was substituted for vehicle alone; such low numbers prevented analysis of interinfusion intervals (see Results). Thirty minutes before the start of session 1, all rats were treated with 0.9% saline (1 mL /kg, i.p.), and received vehicle infusions. Thirty minutes before muscimol (100 μM) self-administration in session 2, half of the rats received saline and the other half received SCH 23390 (0.025 mg /kg, i.p.). In session 3, pretreatments were reversed between the two groups and they received muscimol infusions.

Upon completion of experiment 1, all rats were killed by CO₂ inhalation and their brains were removed and frozen. The rats from experiments 2–6 were deeply anaesthetized with a mixture of sodium pentobarbital (31 mg /kg) and chloral hydrate (142 mg /kg), and their brains were removed and placed in 10% formalin. Frozen coronal sections (40 μm) encompassing the cannula tracks were cut with a cryostat and stained with Cresyl Violet. Cannulae placements were confirmed by microscopic examination.

The arousal effects following repeated muscimol administration may have interfered with lever discrimination in experiment 1. To overcome arousal interference with lever discrimination, the present experiment implemented a ‘penalty’ for responding on the inactive lever that consisted of a retraction of both levers for a 20-s period in the absence of muscimol infusions; this 20-s timeout without reward would be punishing if rats expected to earn muscimol infusions. On the other hand, when the active lever was depressed both levers were also retracted for 20 s, but delivered 100 μM muscimol accompanied by the extinction of a light cue above the lever and the presentation of a tone. Rats trained in this procedure learned to quickly discriminate between the two levers (Fig. 4A). In session 1, rats received vehicle infusions upon active-lever presses. As illustrated in event records of a representative rat (Fig. 4B), rats typically lever-pressed indiscriminately between the two levers during the first 40 min of the session and decreased responding thereafter. In sessions 2–5, they received 100 μM muscimol into the MRN or DRN upon active-lever presses. Rats responded on the active lever more frequently than the inactive lever throughout the session (Fig. 4B).

Many conditions of experiment 2 were different from those of experiment 1, including the sex of rats, testing chamber, testing length, etc. in addition to response contingencies. To confirm that response contingencies, instead of other factors, determined the successful lever discrimination in experiment 2, we examined whether rats fail to learn to discriminate between the two levers when response contingencies are those of experiment 1 while all other factors are identical to experiment 2. Rats failed to discriminate between the two levers when they obtained muscimol into the median raphe with the two levers available throughout testing and no explicit cues provided in sessions 1 and 2, whereas the rats learned to discriminate between the two levers when they received muscimol with the response contingencies of experiment 2 in sessions 3 and 4 (Fig. 5).

The rewarding effects of muscimol into the MRN and DRN were also examined using a conditioned place preference procedure. The compartment paired with MRN muscimol was preferred over the one paired with vehicle (n = 10) (Fig. 6). The compartment paired with DRN muscimol was not significantly preferred over the vehicle-paired compartment (n = 8). As expected, the compartment paired with vehicle was not preferred over another vehicle-paired compartment after conditioning (n = 12).

To determine whether the reinforcing effects of muscimol are mediated by GABA_A receptors, we examined the effects of coadministration of picrotoxin on muscimol self-administration into the MRN. Co-administration of picrotoxin diminished self-administration of muscimol (Fig. 7).