Nr3c1 (GR) gene inactivation was selectively targeted in dopaminoceptive (Nr3c1^loxP/loxP;(Tg)D1aCre (Lemberger et al., 2007), hereafter designed GR^D1Cre) or dopamine (Nr3c1^loxP/loxP;(Tg)BAC-DATiCrefto (Turiault et al., 2007), hereafter designed GR^DATCre) neurons as described in Ambroggi et al. (2009). Experimental animals were obtained by mating Nr3c1^loxP/loxP females with either Nr3c1^loxP/loxP;Tg:D1aCre or Nr3c1^loxP/loxP;(Tg)BAC-DATiCrefto mice. Half of the progeny were mutant animals, the other half were control littermates. Animals were bred and raised under standard animal housing conditions, at 22°C, 55–65% humidity, with a 12-h light/dark cycle (7 am/7 pm) and free access to water and a rodent diet. All experiments were performed in accordance with French (Ministère de l'Agriculture et de la Forêt, 87-848) and the European Directive 2010/63/UE and the recommendation 2007/526/EC for care of laboratory animals. Mice were 2–4 month old males and backcrossed for more than 8 generations on C57BL/6 genetic background. All the experiments have been performed within the hours preceding or encompassing the beginning of dark phase (7 pm), when corticosterone levels are elevated (Le Minh et al., 2001). The behavioral sensitization experiments have been carried out from 6 pm to 9 pm; The CPP experiments from 7 pm to 11 pm and the food progressive ratio (PR) experiment were performed from 4 pm to 6 pm. All drugs were dissolved in saline 0.9%. D-amphetamine (freebase; Sigma-Aldrich, Saint-Quentin Fallavier, France), SKF81297 (salt; Tocris Cookson, Bristol, UK), and MK801 (salt; Tocris Cookson, Bristol, UK) were administered intraperitoneally (ip).

Locomotor activity and sensitization were performed as described in Barik et al. (2010). Briefly, locomotor activity was assessed in circular chamber (4.5-cm width, 17-cm external diameter) crossed by four infrared captors (1.5 cm above the base) placed at every 90° (Imetronic, Bordeaux, France). The locomotor activity was counted when animals interrupted two successive beams and thus, had travelled a quarter of the circular corridor. Mice were habituated to the apparatus for 3 h, for 3 consecutive days, and received a saline injection on days 2 and 3. To assess the response to SKF81297 or MK801, on day 4 mice were placed in the apparatus for 90 min before receiving an acute injection of SKF81297 (1.5 or 3 mg/kg) or MK801 (0.2 mg/kg). In the case of amphetamine-induced locomotor sensitization, from day 4 to 8, mice were daily treated with amphetamine or saline after a 90 min habituation to the apparatus. Following 8 days of withdrawal, mice received an acute challenge of amphetamine. The locomotor activity post-injection was acquired for 1 h. For acute drug responses data are recorded as ¼turn per 5 min. For clarity reason, data are presented every 10 min. For locomotor sensitization, data are presented as the sum of activity over 1 h.

The conditioned place preference (CPP) apparatus consisted of two chambers (20 × 20 × 25 cm) with distinct visual and tactile cues connected by a neutral area. On day 1 (pre-conditioning), mice were placed in the neutral area allowed to freely explore the apparatus for 18 min. The time spent in each chamber was measured. On days 2, 4, 6, and 8, amphetamine-paired mice received an amphetamine injection (1 or 2 mg/kg) and were confined to one chamber for 25 min. On days 3, 5, 7, and 9, amphetamine-paired mice received saline in the opposite chamber and were also confined for 25 min. Saline-paired animals received saline in both chambers. To examine food-induced CPP, mice had limited access to chow pellet in their home-cage for a week, to stabilize their bodyweight to 85% of their original weight. Conditioning for food was similar to that of amphetamine. Food-paired mice received a chow pellet (1 g; standard food CPP) or chocolate with cereals (“chocapic,” Nestlé, 0.5 g; palatable food CPP) in the paired chamber on days 2, 4, 6, 8 and confined for 30 min. On days 3, 5, 7, 9, mice were confined to the other, unpaired, chamber but had no access to food. No food-paired mice were alternatively placed in each chamber with no access to food at any time. During the post-conditioning (day 10), mice, in absence of any reward, were allowed to freely explore both chambers for 18 min. The CPP scores were expressed as the increase of time spent in the paired chamber between the post- and the pre-conditioning sessions.

Immunohistochemistry was performed as described in Barik et al. (2010). Briefly, mice were deeply anaesthetized with pentobarbital (Centravet, France) and transcardially perfused with cold phosphate buffer (PB: 0.1 M Na₂HPO₄/NaH₂PO₄, pH 7.4), followed by 4% PFA in PB. Brains were post-fixed overnight in 4% PFA-PB. Free-floating vibratome sections (30 μm) were rinsed twice with PBS (20 min) and incubated (30 min) in PBS-BT (PBS 0.5% BSA, 0.1% Triton X-100) with 10% normal goat serum (NGS). Sections were incubated (4°C) in PBS-BT, 1% NGS, with rabbit anti-c-Fos (1:500, Abcam, Cambridge, MA) for 36 h. Sections were rinsed in PBS and incubated (2 h) in goat anti-rabbit biotinylated secondary antibody (1:1000, Vector Laboratories, Burlingame, CA) in PBS-BT, 1% NGS. PBS-rinsed sections were incubated in avidin-biotin-peroxydase complex (ABC reagent; Vector Laboratories, 1:1000) for 1 h. Signal was revealed using the peroxidase-substrate-kit-DAB, as recommended by Vector Laboratories. Quantification of c-Fos immunopositive cells was done semi-automatically using Mercator Explora-Nova software (La-Rochelle, France). CPU and NAc regions were delineated according to Paxino's mouse brain atlas. For, drug-induced c-Fos expression, mice received an acute challenge of saline or amphetamine, and were perfused 1 h later.

Electrophysiological recordings of NAc medium spiny neurons were performed during the diurnal phase. The experimenter was blind to the genotype during recordings. Mice were anesthetized with chloral hydrate (5.0 mg/kg, i.p.) and mounted in a stereotaxic apparatus. The lateral tail vein was catheterized to administer additional anesthetic or drugs. Body temperature was monitored and maintained with a heating pad at 36.5–37.0°C. Standard electrophysiological procedures were employed. The electrode signal was amplified 2000 times with an AC high impedance amplifier, band pass filtered at 0.4–1 kHz (Dagan 2400A, Minneapolis, MN), and digitized with an interface board at 10 kHz (Digidata 1440A, Axon Instruments Inc., Foster City, CA) and fed to a computer for offline analysis.

For single unit recordings of NAc medium spiny neurons, five barrels manufactured electrodes (ASI instruments, Warren, MI) were pulled and broken to a tip diameter of 8–15 μm. The center barrel was filled with 2 M NaCl containing 1% Fast Green dye (impedance 2–6 MΩ) and was used to record neuronal activity. One side barrel (impedance 20–60 MΩ) was filled with 150 mM NaCl for automatic current balancing. The other barrels were filled with L-glutamate (100 mM, pH 8), which was ejected as an anion. A retaining current (5–10 nA) was applied during non-ejection periods to minimize passive diffusion.

Electrodes were lowered in the NAc as followed: AP+1.1/+1.7, L+0.6/1.2 and DV−3.9/−5.0 mm from the cortical surface. Because most NAc neurons are quiescent in the basal state, glutamate was ejected by micro-iontophoresis while searching for neurons. Once a neuron was detected, the stability of the signal to noise ratio and waveform characteristics were assessed. Recorded neurons were identified as medium spiny neurons NAc neurons by their anatomical location and waveform durations comprised between 1.1 and 1.8 ms (White, 1996; Kish et al., 1999; Mallet et al., 2005). To generate current-response curves, glutamate was ejected by micro-iontophoresis using escalating currents applied in 15 s pulses interspersed with 15 s of non-ejection periods.

Data are presented as means ± s.e.m. Statistical analysis was carried out using Two-Way analysis of variance (ANOVA) for CPP, drug-elicited c-Fos induction. Acute locomotor responses and locomotor sensitization experiments were analyzed with Three-Way ANOVA with repeated measures. Post-hoc Bonferroni's test or Dunnett's for multiple comparison tests were used when appropriate.

We studied neuronal activation upon amphetamine response by quantifying c-Fos expression in mutant and control littermates. Consistent with previous findings (Moratalla et al., 1996), amphetamine (1 and 2 mg/kg) elicited a significant increase in the number of c-Fos-positive cells within the CPu and the NAc core and shell of control animals (Figures 1A,B). This effect was significantly diminished within the NAc subdivisions and displayed a trend toward a decrease in the CPu when GR^D1Cre mice were administered 1 mg/kg of the drug (Figures 1A,B). No significant genotype difference was observed when animals were administrated a higher dose (2 mg/kg) of amphetamine (Figures 1A,B). These results indicate a hyporesponsiveness of the NAc of GR^D1Cre mice to low doses of amphetamine.

In many species including rodents, psychostimulant injection triggers a typical increase in locomotor responses. Thus, locomotor activity of GR^D1Cre mice and their control littermates was measured following acute amphetamine administration. To ascertain the lack of involvement of GR in dopamine-releasing neurons we also examined responses in GR^DATCre mice and their respective controls. While saline injection failed to produce any locomotor hyperactivity, amphetamine increased locomotor activity in control mice with a stronger response at 2 mg/kg compared to 1 mg/kg (Figures 1C–F). The locomotor response to single amphetamine injection was the same in both GR^DATCre (Figures 1C,D) and GR^D1Cre mice (Figures 1E,F) compared to their respective control littermates, for both doses tested. Thus, the absence of GR in dopaminoceptive neurons does not alter the acute behavioral response to amphetamine.

One of the key features of abused drugs is their ability to trigger locomotor sensitization (Vanderschuren and Pierce, 2010), i.e., a progressive and enduring augmentation in locomotor activity following repeated drug injection. We assessed the sensitizing properties of amphetamine in GR^D1Cre and GR^DATCre mice, and respective control littermates. Five consecutive daily injections of amphetamine (1 mg/kg), but not saline, induced significant locomotor sensitization in control mice, which was still persistent following an 8-day withdrawal period (Figures 2A,B). In contrast GR^D1Cre mice failed to develop locomotor sensitization (Figure 2A) whereas the absence of GR in dopamine neurons (GR^DATCre mice) had no effect (Figure 2B). When tested at a higher dose (2 mg/kg), amphetamine induced a more robust locomotor sensitization that was similar in both mutant lines and their respective control littermates (Figures 2C,D). Hence, GR within dopaminoceptive neurons is selectively required for enabling locomotor sensitization to low doses of amphetamine. This suggests that elevated doses of amphetamine are likely to provoke stronger molecular activations that hence may overcome GR's modulatory effects.

Repeated pairings of abused drugs in a specific environment triggers reward-associated memories (Kelley, 2004) thought to reflect changes in the motivational state of the subject (Bardo and Bevins, 2000). We next studied amphetamine CPP, a commonly employed context-dependent paradigm, to measure the effects of rewarding stimuli in GR mutant animals. On the pre-conditioning day, all four groups of animals spent similar amount of time in the two distinct chambers (Figure 3A). Pairing injections of 1 or 2 mg/kg of amphetamine produced a significant CPP in control mice (Figures 3B,C). Mirroring the results obtained for locomotor sensitization, these rewarding effects were abolished in GR^D1Cre mice at the lowest dose of amphetamine tested, but were not significantly different from controls when the dose was increased up to 2 mg/kg (Figure 3B). The absence of GR in pre-synaptic dopamine neurons had no effect as GR^DATCre mice displayed normal CPP to 1 mg/kg amphetamine (Figure 3C).

In response to abused drugs, the increase of dopamine release within the CPu and NAc is thought to filter and selectively reinforce connections arising from excitatory corticostriatal projections (Bamford et al., 2004). Hence this dopamine/glutamate interaction is key to shape medium spiny neurons responsiveness at both electrophysiological and molecular levels, with a central implication of D1 dopamine receptors and NMDA glutamate receptors in these processes (Nicola et al., 2000; Pascoli et al., 2011). We therefore, examined whether GR gene inactivation within dopaminoceptive neurons could impact on dopamine and glutamate receptor functions that may explain the observed phenotype. To challenge D1 dopamine receptor, we injected SKF81297, a selective D1-like receptor agonist and measured subsequent locomotor responses. As previously reported (Corvol et al., 2007), acute systemic SKF81297 injection elicited hyperlocomotion in control animals (Figure 4A). At the 2 doses examined (1.5 and 3 mg/kg), GR^D1Cre mice did not differ from their respective control littermates (Figure 4A) ruling out an impaired functionality of D1 dopamine receptors. To determine the state of glutamate transmission in GR^D1Cre mice, we then assessed the ability of MK801, a non-competitive NMDA antagonist, to elicit hyperlocomotion (Qi et al., 2008). Systemic injection of MK801 (0.2 mg/kg) triggered a robust hyperlocomotion in controls that was significantly decreased in mutant mice (Figure 4B). Therefore, this set of experiments suggests that the impaired glutamate response may play a role in the diminished response to amphetamine.

Altered glutamatergic neurotransmission within the NAc might contribute to the impaired behavioral responses to amphetamine and MK801 as well as to the decrease of accumbal c-Fos induction observed in GR^D1Cre mice. To investigate the functional effects of GR inactivation on glutamatergic neurotransmission within the NAc, we analyzed the reactivity to glutamate of NAc medium spiny neurons. We performed in-vivo recordings of NAc medium spiny neurons coupled to glutamate micro-iontophoresis in control and GR^D1Cre mice. For each neuron, incremental glutamate ejection currents were applied until the neuron reached its maximal firing frequency. In control mice, the maximal frequencies observed were normally-distributed, ranging from 7 to 17 Hz (Figures 5A–C). In contrast in GR^D1Cre mice, the distribution was bimodal: one population had maximal frequencies in a range similar to that observed in control mice, whereas another population was shifted toward lower frequencies; these neurons were unable to fire above 6 Hz. The analysis of the dose-response functions revealed that the EC₅₀ was similar between controls, fast and slow neurons (Figure 5D). We found no evidence of anatomical segregation of fast and slow neurons; in particular, they were found equally in the core and the shell (P > 0.05). This experiment shows that GR positively controls the reactivity to glutamate of a subset of NAc medium spiny neurons.

As the processing of natural rewards and addictive drugs activate overlapping pathways, we sought to determine whether GR gene inactivation in dopaminoceptive neurons resulted in a general impairment of natural reward-seeking. As we did for amphetamine, we first tested control and mutant mice in two CPP experiments in response to normal (chow pellets) or palatable (chocolate) food. Mice were exposed for 30 min to either food in the paired chamber and spent the same amount of time, without food, in the opposite (unpaired) chamber on alternate days. Following 8 days of conditioning, the time increase in the paired chamber was used as an index of place preference. Both normal and palatable food elicited significant CPP in control animals. However, unlike our results with amphetamine, CPP remained unaltered in GR^D1Cre mice (Figures 6A,B respectively).

Next, as GR inactivation within dopaminoceptive neurons has been reported to decrease motivation for cocaine in a PR schedule (Ambroggi et al., 2009), we thus, tested motivation of GR^D1Cre mice to respond instrumentally for food rewards. During initial instrumental training, under a fixed-ratio1 schedule, control and GR^D1Cre mice did not show significant differences in the number of responses (Figure 6C) and ate similar amount of pellets (control: 6.04 ± 0.29 g, GR^D1Cre: 6.3 ± 0.23 g). During the learning phase of the PR schedule (where the response-requirement increased after each reward obtained), control and mutant mice exhibited a comparable increase in their responding for food (Figure 6D left panel). Analysis of the breaking points (defined as the last ratio completed by the animal followed by 15 min during which no additional reward was earned) revealed no difference between controls and GR^D1Cre mice (Figure 6D right panel). While during initial instrumental training GR^D1Cre mice showed a trend toward a decrease in instrumental responses compared to controls, the opposite was rather observed during the last two sessions of PR. This set of data suggests that GR in dopaminoceptive neurons does not modulate food reward responses.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.