Glutamate input to noradrenergic neurons plays an essential role in the development of morphine dependence and psychomotor sensitization

Abstract

The brain's noradrenergic system is involved in the development of behaviours induced by drugs of abuse, e.g. dependence and withdrawal, and also reward or psychomotor effects. To investigate how noradrenergic system activity is controlled in the context associated with drug-induced behaviours, we generated a Cre/loxP mouse model in which the essential glutamate NMDA receptor subunit NR1 is ablated in cells expressing dopamine β-hydroxylase (Dbh). As a result, the noradrenergic cells in NR1^DbhCre mice lack the NMDA receptor-dependent component of excitatory post-synaptic currents. The mutant mice displayed no obvious behavioural alterations, had unchanged noradrenaline content and mild increase in dopamine levels in the nucleus accumbens. Interestingly, NR1^DbhCre animals did not develop morphine-induced psychomotor sensitization. However, when the morphine injections were preceded by treatment with RX821002, an antagonist of α₂-adrenergic receptors, the development of sensitization was restored. Conversely, pretreatment with clonidine, an agonist of α₂-adrenergic receptors, blocked development of sensitization in wild-type mice. We also found that while the development of tolerance to morphine was normal in mutant mice, withdrawal symptoms were attenuated. These data reveal that NMDA receptors on noradrenergic neurons regulate development of opiate dependence and psychomotor sensitization, by controlling drug-induced noradrenaline signalling.

Morphine, NMDA receptor, noradrenaline, reinforcement, sensitization, withdrawal

Introduction

The noradrenergic system of the brain controls arousal, selective attention, response to stress, and plays an important role in learning and memory (Aston-Jones & Cohen, 2005; Sara, 2009). It also plays a central role in drug dependence, becoming strongly activated during opiate withdrawal (Aghajanian, 1978; Gold et al.1978; Rasmussen et al.1990). However, beyond physical dependence, the role of the noradrenaline (NA) system in the development of behaviours associated with drugs of abuse has been a matter of controversy (Sofuoglu & Sewell, 2009; Tassin, 2008; Weinshenker & Schroeder, 2007; Wise, 1978). Although drugs of abuse interfere with both NA and dopamine (DA) signalling, the accumulated evidence points to the latter as the main substrate for the actions of drugs of abuse (Hyman et al.2006; Kauer & Malenka, 2007; Tzschentke, 1998).

Studies with genetically modified mice have reopened the discussion on the respective roles of NA and DA in drug-conditioned behaviours. It was found that the ablation of the DA transporter (Sora et al.1998, but see Chen et al.2006), D₁ (Karasinska et al.2005) or D₂L dopamine receptors (Smith et al.2002) did not abolish cocaine-conditioned place preference (CPP), suggesting that DA-dependent signalling is not the only system involved. In contrast, the ablation of adrenergic α₁b receptors (Adra1b) completely prevented the development of drug-induced psychomotor sensitization or CPP (Drouin et al.2002). Furthermore, in mice lacking dopamine β-hydroxylase (Dbh), an enzyme necessary for NA synthesis, morphine CPP and psychomotor activation were abolished (Olson et al.2006). When mutant mice were subjected to injections of virus vector expressing Dbh into the solitary tract nucleus (NTS), morphine-induced behaviours were rescued. In addition, morphine-induced CPP was still possible in DA-deficient mice (Hnasko et al.2005). Thus the NA system, acting independently of DA, could support development of opiate-induced behaviours.

While drug-induced long-term adaptations in DA pathways were found to be associated with the plasticity of glutamatergic transmission in the ventral tegmental area (VTA) (Chen et al.2010; Ungless et al.2001), the mechanisms controlling NA pathway-related behaviours are poorly understood. The cells involved receive glutamatergic afferents originating mainly from the nucleus paragigantocellularis of the medulla as well as descending projections from the forebrain including the paraventricular nucleus of the hypothalamus and the insular and limbic cortices (Aston-Jones et al.1986; Baude et al.2009; Reyes et al.2005). A large proportion of these glutamatergic fibres in the locus coeruleus (area A6, LC) were found to contain opioid peptides and/or corticotrophin-releasing factor, potentially allowing interaction between the stress axis and the endogenous enkephalin system (Van Bockstaele et al.2010).

In our study, we sought to define the role of the brain's NA system in relation to the development of opiate-induced behaviours. We reasoned that in the absence of functional NMDA receptors, the NA neurons would display disrupted long-term excitatory modulation, thereby altering development of drug-induced behaviours dependent on changes in NA signalling. Therefore, we generated a transgenic mouse, NR1^DbhCre, through the ablation of the essential NMDA receptor subunit NR1 (Grin1) in NA cells.

Methods

Animals

The DbhCre and NR1 flox strains have been described previously (Niewoehner et al.2007; Parlato et al.2007). Transgenic animals from both parental strains were crossed into the C57BL/6N background for at least six generations. Animals aged 8–20 wk at the beginning of experimental procedures were housed in Plexiglas home cages (30×40×20 cm) (n=2–6 animals per cage) on a 12-h light/dark cycle (lights on 08:00 hours) in a temperature-controlled room (24±1°C) at 55–65% humidity. Standard laboratory chow (Labofeed H, WPiK, Poland) and water were available ad libitum. Behavioural tests started 2 wk after the arrival of the animals at the laboratory and were conducted during the light phase (08:00–20:00 hours) by experimenters who were blinded to the genotype and drug treatment. NR1^loxP/loxP or NR1^{loxP/wild-type} animals were used as controls. Behavioural phenotyping was performed on four cohorts of male mice, totalling 93 animals and including 47 NR1^DbhCre mice. Three additional separate groups of mice were used in electrophysiology, microdialysis and tissue neurotransmitter measurements. The size of each experimental group is given in the figure legends. Experiments were conducted in accordance with European Union guidelines for the care and use of laboratory animals (ECC Directive of 24 November 1986) and were approved by the II Local Bioethics Committee (Poland).

Immunostaining

Two control and two NR1^DbhCre mice were deeply anaesthetized and transcardially perfused with saline followed by 4% paraformaldehyde in phosphate-buffered saline (PBS). Dissected brains were fixed for an additional 12 h in 4% paraformaldehyde and then cut with a vibratome (Leica, Germany) at 40 or 50 µm. For immunohistochemistry, coronal sections were incubated with antibodies against tyrosine hydroxylase (TH; Millipore, USA, MAB318, 1:2000) or NR1 (Sigma, USA, G8913, 1:100) and stained with diaminobenzidine (DAB) using the ABC kit (Vector Laboratories, USA) according to the manufacturer's instructions. For each animal 4–6 sections were stained at the levels of the aqueduct, LC and NTS. Images of stained sections were acquired using a Leica DMNB microscope equipped with a Basler ‘Scout’ digital camera and Visiopharm software (Denmark).

Whole-cell recording

Recordings were performed as previously described (Tokarski et al.2003). Briefly, five control and five NR1^DbhCre mice aged 8–12 wk were decapitated, the brains were quickly removed and immersed in ice-cold artificial cerebrospinal fluid (aCSF) containing (mm) 130 NaCl, 5 KCl, 2.5 CaCl₂, 1.3 MgSO₄, 1.25 KH₂PO₄, 26 NaHCO₃, 10 d-glucose and bubbled with a mixture of 95% O₂/5% CO₂. Horizontal slices (350 µm thick) were cut on a vibrating microtome (Leica).

Slices were placed in the recording chamber mounted on the Zeiss Axioskop microscope (Zeiss, Germany) and superfused (3 ml/min) with modified, Mg-free aCSF (32±0.5°C). Neurons were visualized using Nomarski optics and a 40× lens. NA cells of the LC were identified by their position in relation to the IV ventricle, the shape of the soma and the response to a depolarizing current pulse (Andrade & Aghajanian, 1984). Only cells with a resting membrane potential of at least −50 mV and overshooting action potentials were accepted for analysis. Series resistance did not change appreciably during the experiments, indicating stable recording conditions. Stimulus–response characteristics of recorded neurons were evaluated using rectangular current pulses (500 ms) of increasing intensity in 20 pA steps (60–300 pA). Then, cells were depolarized with current injection to −50 mV and spontaneous spiking activity was recorded for 4 min in the current-clamp mode. Next, cells were voltage-clamped at −76 mV, and spontaneous excitatory post-synaptic currents (sEPSCs) were recorded for 12 min (3×4 min). To block NMDA receptor-mediated currents, a specific antagonist, CGP37849 (5 µm), was added to the aCSF. After 20 min perfusion, sEPSCs were again recorded for 12 min. Spontaneous EPSCs were detected offline and analysed using Mini Analysis software (Synaptosoft, USA). The amplitude and area thresholds for the detection of an event were set to 7 pA and 25 fC, respectively. Recorded traces were visually verified following automated analysis.

Tissue levels of neurotransmitters

After each experiment was completed, mice were decapitated, brains were dissected and brain regions were separated on ice. Tissue samples were weighed and homogenized in ice-cold 0.1 m perchloric acid. Then, homogenates were centrifuged at 10 000 g. Supernatants were filtered through membrane filters (0.1 µm pore size) and injected into HPLC for the determination of tissue levels of NA, DA, 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA).

Microdialysis in freely moving animals

Microdialysis probes were constructed by inserting two fused silica tubes (30 and 35 mm long, 150 µm outside diameter (o.d.); Polymicro Technologies Inc., USA) into a microdialysis fibre (220 µm o.d.; AN69, Hospal, Italy). The tube assembly was placed in a Peek cannula (0.3 mm o.d., 6 mm long) to form the shaft of the probe. Portions of the inlet and outlet tubes were individually placed inside polyethylene PE-10 tubing and glued. The free end of the dialysis fiber was sealed, and 2 mm of the exposed length was used for dialysis. The molecular cut-off value of the membrane was 6000 Da. Recovery rate was determined by an in-vitro experiment; the probe was placed in a solution containing 50 pg/µl DA and the DA concentration was measured in dialysate fractions. The recovery efficiency was between 10% and 15%.

Mice were anaesthetized with ketamine (7.5 mg/kg i.p.) and xylazine (1 mg/kg i.p.) and placed into a stereotaxic apparatus (David Kopf Instruments, USA). The skulls were exposed, and small holes were drilled for the insertion of the microdialysis probes using the following coordinates: 1.5 mm anterior to bregma, 1.0 mm lateral to the sagittal suture, −4.0 mm ventral to the dural surface [nucleus accumbens (NAc)]. Animals were allowed to recuperate until the next day (24 h). Then probes were connected to a syringe pump (BAS, USA), which delivered aCSF composed of (mm) 145 NaCl, 2.7 KCl, 1.0 MgCl₂, and 1.2 CaCl₂ (pH 7.4) at a flow rate of 1.5 µl/min. After a 2-h washout period, baseline samples were collected every 30 min over 2 h. Next, mice were injected subcutaneously with morphine (10 mg/kg), and dialysate fractions were collected every 30 min for 4 h. Elution times were not corrected for the volume of the tubing and swivel. At the end of the experiment, the mice were sacrificed, and their brains examined histologically to validate probe placement.

Neurotransmitter analysis

NA, DA and metabolites (DOPAC and HVA) were analysed by HPLC with coulochemical detection. Chromatography was performed using an Ultimate 3000 System (Dionex, USA), coulochemical detector Coulochem III (model 5300, ESA, USA) with 5020 guard cell, 5014B microdialysis cell and Hypersil Gold-C₁₈ analytical columns (3×100 mm). The mobile phase was composed of 0.05 m potassium phosphate buffer adjusted to pH 3.9, 0.5 mm EDTA, 13 mg/l 1-octanesulfonic acid sodium salt, 3.1% methanol and 0.93% acetonitrile. The flow rate during analysis was 0.7 ml/min. The applied potential of a guard cell was +600 mV, whereas that of the microdialysis cell was E₁=−50 mV, E₂=+300 mV. Sensitivity was set at 50 nA/V. The chromatographic data were processed by Chromeleon v. 6.80 software (Dionex). The values were not corrected for in-vitro probe recovery, which was approximately 15%.

CPP/conditioned place aversion (CPA) and locomotor activity measurements

The procedure for studying changes in locomotor activity and CPP/CPA was adapted from (Itzhak & Martin, 2002). Doses of morphine hydrochloride (10 or 7.5 mg/kg i.p., Polfa, Poland; dissolved at 0.75 or 1 mg/ml in saline) and naloxone (1.5 or 2 mg/kg i.p., Sigma Aldrich, USA, dissolved at 0.15 or 0.2 mg/ml in saline) were selected based on their ability to reveal genotype-dependent differences in behaviour as observed in a previous study from our laboratory (Solecki et al.2009). Clonidine hydrochloride (α₂-adrenergic receptor agonist; Tocris, UK) and RX821002 hydrochloride (α₂-adrenergic receptor antagonist; Tocris) were dissolved in sterile water at 0.003 and 0.05 mg/ml, and administered i.p. at 0.03 or 0.5 mg/kg, respectively, 30 min prior morphine or vehicle. Final injected volume was always 0.1 ml per 10 g of animal weight.

CPP/CPA and locomotor activity measurements were conducted in eight automatic boxes (Med Associates, USA) consisting of two conditioning chambers separated by guillotine doors and a central platform. The design of the experiment is shown in Fig. 5a. During the pre-conditioning and post-conditioning tests, mice were placed individually on the central platform of the apparatus and had free access to both arms for 20 min. The time spent in each arm and movements were recorded with the use of an infrared beam system. After the pre-conditioning phase, one arm was paired with drug administration, and the other with saline. The assignment of treatments to the arms was counterbalanced. During conditioning, the mice were treated with vehicle, morphine, cocaine or naloxone (i.p.) immediately before being placed in the appropriate arm for 40 min. Locomotor activity measured on conditioning sessions was used to assess the development of psychomotor sensitization. The extinction of CPP/CPA consisted of 10 conditioning sessions during which mice were treated with saline (i.p.) in both previously drug- and vehicle-paired arms. To induce reinstatement of CPP/CPA mice were treated with 75% of the drug dose used for conditioning, either i.p. morphine (7.5 mg/kg) or naloxone (1.5 mg/kg) and immediately placed in the apparatus for 20 min.

Physical dependence after chronic morphine treatment

Physical dependence was evaluated by measuring behavioural manifestations of naloxone-precipitated withdrawal in mice treated with chronic morphine (Fig. 7a). Three hours after the last morphine treatment, each mouse was injected s.c. with naloxone (4 mg/kg dissolved at 0.4 mg/ml in saline) and then placed in a Plexiglas tube. The number of jumps (defined as no contact with the surface for all four paws), paw tremors, teeth chattering and defecations were scored by observers blind to the genotype and treatment.

Data analysis

CPP score was defined as the difference in time spent in the drug- and vehicle-paired arms during the test day. In general data were analysed by factorial analysis of variance (ANOVA) followed by a post-hoc test or a t test. Results from microdialysis, electrophysiology and conditioned reinforcement were analysed by two-way ANOVA with repeated measures with variables (between-subjects: genotype; within-subjects: time, treatment or CPP stage). Locomotor activity data was analysed in two separate tests. Effects of saline treatment and first drug treatment on locomotor activity were analysed by two-way ANOVA with repeated-measures between-subjects genotype or treatment (i.e. clonidine or RX821002) and within-subjects treatment (i.e. saline or morphine) with activity counted in 5-min bins over a 30-min session. Effects of intermittent drug treatment on locomotor activity were analysed by two-way ANOVA with repeated-measures between-subjects genotype or treatment (i.e. clonidine or RX821002) and within-subjects day, with activity counted as mean values from 30-min test sessions on drug-injection days. Frequencies of morphine withdrawal symptoms were analysed using two-way ANOVA (genotype, treatment).

Additional description of behavioural phenotyping methods is included in the Supplementary material (available online).

Results

Generation of NR1^DbhCre mice

The ablation of the NR1 (Grin1) gene was restricted to cells expressing DA β-hydroxylase (Dbh) using the Cre/loxP system. Transgenic mice harbouring the Cre recombinase under the control of the PAC-derived Dbh gene promoter were crossed with NR1 flox mice, in which loxP sequences flank exons 11–18 of the gene (Fig. 1a). Resulting NR1 flox/flox; DbhCre Tg/0 (NR1^DbhCre) animals were born at the expected ratio and did not differ from their NR1^loxP/loxP littermates or wild-type C57BL6/N mice of similar age. The loss of NR1 protein was restricted to NA cells, as observed in the LC by immunostaining with NR1-specific antibodies (Fig. 1b–e). The mutation did not cause any gross morphological changes in the LC or NTS (Fig. 1f–i). These results are in agreement with the previously reported specificity and efficiency of Cre/loxP-driven deletions using the DbhCre line (Parlato et al.2007, 2010).

Fig. 1

Generation of mice with NR1 gene ablation restricted to noradrenergic cells. (a) Schematic representation of the bacterial artificial chromosome-derived transgene (DbhCre) expressing the Cre under the control of the Dbh gene promoter used for the recombination of a ‘floxed’ variant of the NR1 gene. (b–e) Micrographs show immunohistological staining of the NR1 protein (brown) on representative coronal sections from control (b, d) and NR1^DbhCre mice (c, e). Panels (d, e) are higher magnifications of the corresponding areas from (b, c) marked by dashed line rectangles. (f–i) The loss of NR1 was not associated with any apparent anatomical alterations in the locus coeruleus (f, g) or solitary tract nucleus (h, i) as shown on representative tyrosine hydroxylase immunostainings (brown). Scale bars: (b, c) 25 µm; (d, e) 8.3 µm; (f, g) 200 µm; (h, i) 400 µm.

Phenotypic characterization of NR1^DbhCre mice

The NR1^DbhCre animals displayed no impairments or major phenotypic alterations. They had normal muscular strength and coordination as assessed by the wire-hang test (Fig. 2a, b). The activity pattern of NR1^DbhCre animals in the open field was similar to that of control littermates; i.e. animals habituated to the environment at comparable rates and avoided the illuminated centre of the field (Fig. 2c, d). Accordingly, exploration of the light-dark box apparatus by NR1^DbhCre mice did not differ from that of control animals, indicating typical anxiety-like behaviours (Fig. 2e–g). Finally, the NR1^DbhCre animals demonstrated the same pattern of Y-maze exploration as controls, alternating the explored arms of the apparatus with normal frequency and thus indicating normal spatial memory performance (Fig. 2h–j). In conclusion, the loss of NMDA receptors on NA neurons did not cause observable impairments, abnormal anxiety levels or learning deficits.

Fig. 2

Basic behavioural phenotyping of NR1^DbhCre mice. (a, b) The motor performance of the NR1^DbhCre animals (n=14) and controls (n=11) was similar in the wire hang test, including the ability of mice to hang on to the wire (a, |t₂₃|<1, n.s.) and time spent taking steps (b, t₂₃=1.4, n.s.). (c, d) Locomotor activity of NR1^DbhCre animals (n=8) and controls (n=8) in the open field. There was no genotype effect on locomotor activity; all mice showed decreased activity in subsequent sessions over 4 d (c, genotype: F_1,42=1.9, n.s.; time: F_3,42=68.7, p<0.001; genotype×time: F_3,42<1, n.s.). The loss of NR1 had no effect on the exploration of the illuminated centre of the field (d, genotype: F_1,42<1, n.s.; time: F_3,42=5.6, p<0.01; genotype×time: F_3,42<1, n.s.). (e–g) NR1^DbhCre mice (n=10) and littermate controls (n=8) did not differ with regard to anxiety-related behaviours in the light-dark box test. Irrespective of genotype, mice spent the same amount of time in the light side of the box (e, |t₁₆|<1, n.s.), with a similar number of transitions between compartments (f, t₁₆=1.05, n.s.) and rearings (g, compartment: F_1,30=10.36, p<0.01; genotype: F_1,30<1, n.s.; genotype×compartment: F_1,30<1, n.s.). (h–j) Spatial memory-dependent performance in the Y-maze was normal in NR1^DbhCre mice (h, |t₁₄|<1, n.s.). Both groups (n=8 mice per group) explored the maze and tended not to re-enter previously visited arms (i, j, t₁₄=1.7 and |t₁₄|<1 respectively, both n.s.).

Loss of the NR1 subunit in noradrenergic cells blocks the function of NMDA receptor channels

In order to characterize the functional consequences of NR1 ablation at the cellular level, we performed whole-cell patch-clamp recordings from LC neurons under Mg²⁺-free conditions, which promote NMDA receptor-dependent glutamatergic transmission. The stimulus–response characteristic of NA neurons in NR1^DbhCre mice was normal and similar to that of controls (Fig. 3a). Noradrenergic cells from mutant mice exhibited a non-significant trend towards higher spontaneous activity when depolarized to −50 mV (Fig. 3b; genotype: F_1,13=2.43, n.s.). Similar to controls, neurons lacking NR1 showed a reduction in spiking frequency after administration of CGP37849, a NMDA receptor antagonist (Fig. 3c; treatment: F_1,13=13.60, p<0.01). When the neurons were voltage-clamped at −76 mV, we observed increased frequency of sEPSCs in NR1^DbhCre animals compared to controls (Fig. 3d, e; genotype: F_1,16=4.52, p<0.05), without a significant difference in the mean sEPSC amplitude (Fig. 3f).

Fig. 3

Whole-cell patch-clamp recordings of neurons in the locus coeruleus (LC). (a) Stimulus–response curves of LC putative NA neurons in slices obtained from NR1^DbhCre mice (●) and control animals (◯). Data points represent the mean number of spikes evoked by 500 ms depolarizing current pulses±s.e.m. Analysis was performed on six NA neurons from five control mice and 10 NA neurons from five NR1^DbhCre animals. (b, c) The spontaneous activity of LC neurons maintained at −50 mV in the absence of Mg²⁺ ions, before and after administration of the NMDA receptor antagonist CGP37849. Examples shown in (b) were recorded before CGP37849 administration. Several recorded neurons displayed no spontaneous activity, and overlap at 0 frequency (c). Data were collected from eight NA neurons from five control mice and 13 neurons from five NR1^DbhCre animals. (d–f) Spontaneous excitatory post-synaptic current (sEPSC) recordings in LC neurons under Mg²⁺-free conditions indicate higher frequencies in cells from NR1^DbhCre mice compared to controls (d, e) without a significant difference in amplitude (f). CGP37849 treatment had no effects on either frequency or amplitude. The examples shown in (d) were recorded before CGP37849 administration. Recordings were performed on eight NA neurons from five control mice and 13 NA neurons from five NR1^DbhCre animals. (g–i) Spontaneous EPSCs from the LC of control (left) and NR1^DbhCre (right) mice are differentially affected by administration of the NMDA receptor antagonist CGP37849 (g). While the ‘rise’ intervals (time elapsed from 10% to 90% of amplitude) are similar (h), the ‘tau’ (time elapsed from 100% to 37% of amplitude) is reduced by administration of CGP37849 in control mice (i) but not NR1^DbhCre animals. Analyses were performed on eight NA neurons derived from five control mice and 13 NA neurons from five NR1^DbhCre animals. Significant difference (paired t test with Bonferroni's correction p<0.01) in mean tau time in control neurons before and after CGP37849 administration is indicated by ‘**’.

The administration of CGP37849, an antagonist of NMDA receptors, had no effect on the sEPSC decay time constant in mutant NA neurons (Fig. 3g–i), whereas in controls it resulted in a faster decay of the synaptic current (genotype×treatment: F_1,16=28.81, p<0.001). All noradrenergic cells recorded from NR1^DbhCre mice were insensitive to CGP37849 treatment and exhibited sEPSC decay times in the same range as those of LC neurons in wild-type mice after administration of the NMDA antagonist (Fig. 3i). At the same time, the rise time and amplitude were not affected, indicating that AMPA/kainate receptor-dependent components of sEPSCs were not altered by the mutation.

Notably, the mean frequency of sEPSCs recorded from NR1^DbhCre cells was higher than that observed in controls. Because the frequency of sEPSCs is determined mainly by presynaptic factors, this observation suggests a strengthening of the glutamatergic input, which might partly compensate for the reduced reactivity of the post-synaptic side due to NMDA receptor ablation. This result could also explain the trend towards higher spontaneous activity observed in LC cells from NR1^DbhCre mice when held at −50 mV, and resembles the up-scaling of activity that occurred after the ablation of NR1 in DA cells (Engblom et al.2008; Zweifel et al.2008). The observed NMDA receptor antagonist-induced decrease in the spontaneous activity of LC neurons in NR1^DbhCre mice can be explained by the fact that the preparations used for the recording contained functioning excitatory neurons (or their parts) that project to the LC. It is conceivable that CGP37849 administration led to reduced activity of those cells, which in turn resulted in a decreased excitatory input to LC neurons, and consequently a reduction in their activity. Finally, the presence of presynaptic NMDA receptors on afferents to the LC was reported (Van Bockstaele et al.2000), and their inhibition could also contribute to the genotype-independent decrease in activity observed after administration of the antagonist. Taken together, the data confirm loss of functional NMDA receptors in NA neurons upon NR1 deletion.

Catecholamine release in the NAc

We analysed the content and extracellular levels of catecholamines in the medial NAc including parts of both the core and shell subdivisions (Fig. 4a). There was a trend towards higher DA content (t₁₀=2.21, p=0.051), without differences in levels of DA metabolites (Fig. 4b). Then, we assessed extracellular levels of catecholamines after a single s.c. injection of 10 mg/kg morphine. Neurotransmitter levels were monitored using custom-made probes with 2-mm-long microdialysis windows and tips extending into the medial NAc (Fig. 4a; Supplementary Fig. S1). We found a robust morphine-induced increase of NA in the medial NAc/striatum of both control and NR1^DbhCre mice (Fig. 4c; time: F_8,80=3.3, p<0.01; genotype: F_1,10=3.4, p=0.096). The injection of morphine caused a slight increase of DA (n.s., Fig. 4d), accompanied by a robust ∼1.5-fold increase in the extracellular concentrations of DA metabolites (Fig. 4e, f; time: F_8,80=4.2 and 5.8, both p<0.001) without differences between control and mutant mice. There were no significant differences between NR1^DbhCre mice and littermate controls in terms of basal extracellular concentrations of catecholamines and related metabolites (Supplementary Table S1). In conclusion, neurotransmitter measurements demonstrated that morphine treatment robustly increases extracellular NA levels in the medial NAc/striatum.

Fig. 4

Content and extracellular levels of catecholamines in the nucleus accumbens. (a) Diagram of a coronal brain section shows NAc area used for catecholamine analysis. Grey circle indicates the piece punched out for tissue content measurements. The thick black line indicates approximate microdialysis probe placement, the length of the line corresponds to the ∼2-mm-long dialysis window extending from the tip. The number below the diagram (+1.18) is the distance from bregma (Paxinos & Franklin, 2001). (b) Catecholamine content (pg/mg of tissue). Values are mean±s.e.m. (n=6 mice per group). (c–f) Relative levels of extracellular catecholamines and their metabolites measured after s.c. injection of 10 mg/kg morphine (at minute 0). The lines on the graphs connect points representing mean values normalized to the median of the four measurements performed prior to morphine injection. Dotted line and open symbols (◯) correspond to controls, solid line and solid symbols (●) represent NR1^DbhCre animals. (c) Measured relative changes in extracellular NA levels, (d) corresponds to DA, (e) DOPAC and (f) HVA. Values are mean±s.e.m. (n=6 animals per group). Significant differences (Dunnett's post-hoc test p<0.05) vs. basal levels (minute 0) are indicated by ‘*’ for control and ‘#’ for NR1^DbhCre mice.

Psychomotor and reinforcing effects of morphine and naloxone

In order to test the role of NMDA receptor inputs in development of drug-conditioned behaviours we tested the development of psychomotor sensitization and CPP in NR1^DbhCre mice and control animals. A schematic diagram of the procedure is shown in Fig. 5a. We found that acute drug treatments caused an expected increase in activity in both mutant and control mice (Fig. 5, Supplementary Table S2). However, locomotor activity of NR1^DbhCre mice was not further increased upon intermittent treatment with morphine (10 mg/kg i.p.), while controls developed normal psychomotor sensitization (Fig. 5b; genotype: F_1,48=15.1, p<0.01; day: F_4,48=2.4, p=0.065; genotype×day: F_4,48=4.6, p<0.01; for complete statistics see Table S2). This observation was validated in an independent cohort of animals (Supplementary Table S2), confirming that the phenotype was stable in different generations of mutant mice. We reasoned that if NMDA receptor-dependent signalling was necessary for morphine-induced NA release, then pharmacological blockade of NA release should replicate the mutant phenotype. Indeed, when 30 min before morphine injections wild-type C57BL/6 mice were pretreated with clonidine (0.03 mg/kg i.p.), an α₂-adrenergic receptor agonist, psychomotor sensitization did not develop (Fig. 5c; treatment: F_1,72=12.6, p<0.01; day: F_4,72=4.3, p<0.01; day×treatment: F_4,72=2.1, p=0.08). Clonidine treatment at this dose significantly decreased basal locomotor activity (treatment: F_1,56=71.32, p<0.001).

Fig. 5

Drug-induced psychomotor sensitization. (a) Experimental design. The boxes represent sessions in the conditioned place preference (CPP) apparatus. Drugs or saline were administered on days marked by syringe symbols. On ‘test’ days the animals were allowed to explore the entire CPP cage. (b–e) Locomotor activity after saline injection (day 2), first drug injection (day 3), and after 5 d repeated treatment (day 12), (b) with 10 mg/kg morphine (n=7 mice per group), (c) 0.03 mg/kg clonidine 30 min prior to 10 mg/kg morphine (n=10 mice per group), (d) 0.5 mg/kg RX 821002 30 min prior to 10 mg/kg morphine (n=8 mutants, 11 controls) and (e) 0.5 mg/kg RX 821002 alone (n=6 mice per group). Data shown are mean±s.e.m., significant differences (Bonferroni's post-test) between mutant and control mice are indicated by: * p<0.05, ** p<0.01 and *** p<0.001. WT, Wild-type.

Conversely, a drug that activates NA release independently of NMDA receptor signalling rescued the sensitization in NR1^DbhCre mice. Psychomotor sensitization was rescued when, 30 min before morphine injections, NR1^DbhCre animals were pre-treated with RX821002 (i.p. 0.5 mg/kg), an α₂-adrenergic receptor antagonist (Fig. 5d; genotype: F_1,68<1, n.s.; treatment: F_4,68=9.2, p<0.001; genotype×treatment: F_4,68<1, n.s.). At the same time, pre-treatment with RX821002 had no further effect on ambulation in control mice, and injections of RX821002 (1 mg/kg) alone had no effect on activity of wild-type C57BL6 mice (Fig. 5e). Thus, morphine-induced psychomotor sensitization did not develop in NR1^DbhCre animals, but the phenotype could be rescued by facilitating NA release at the synapse.

To assess the consequences of the mutation with regard to behavioural reinforcement we measured morphine-induced CPP and naloxone-induced CPA. Both NR1^DbhCre and control mice acquired similar morphine CPP, extinction and reinstatement (Fig. 6a; genotype: F_1,12=1.7, n.s.; CPP stage: F_2,24=16.05, p<0.001). The trend towards stronger preference for the CS+ compartment observed in NR1^DbhCre mice after CPP and reinstatement was not significant. Naloxone-induced CPA was normal in NR1^DbhCre mice (Fig. 6b; genotype: F_{1, 10}<1, n.s.; CPP stage: F_2,20=5.24, p<0.05) Therefore, we conclude that the loss of functional NMDA receptors in NA neurons does not impair positive or negative reinforcement.

Fig. 6

Drug-induced conditioned place preference (CPP) or conditioned place aversion (CPA). (a) Injections of 10 mg/kg morphine induced CPP, while (b) treatment with 2 mg/kg naloxone induced CPA. The values (‘scores’) represent the difference in time spent in the drug-paired compartment vs. the saline-paired compartment, after conditioning (day 13), after 10 sessions of extinction (day 28) and after reinstatement of preference by drug injection (day 29). Data shown are mean±s.e.m.; n=7 mice per group in the CPP experiment and n=6 mice per group in the CPA experiment. Significant differences (Bonferroni's post-test) in time spent in the drug-paired compartment after conditioning compared to extinction is indicated by * p<0.05 and between extinction and reinstatement by ^#p<0.05 or ^##p<0.01.

Morphine tolerance, physical dependence and withdrawal in NR1^DbhCre mice

The activity of NA neurons regulates endogenous analgesic systems and is strongly affected during opiate physical dependence (Rasmussen et al.1990). Therefore, we examined the consequences of the mutation on pain sensitivity, opioid induced-analgesia and withdrawal. Sensitivity to pain, as measured by tail-flick and hot-plate tests, was similar in control and NR1^DbhCre mice (Supplementary Fig. S2). The analgesic effects of morphine [analgesic dose (AD₅₀) with 95% confidence intervals] and the development of tolerance to morphine analgesia after repeated treatment with increasing doses of morphine did not differ between genotypes (Supplementary Fig. S2). After treatment with morphine and injection of the opioid antagonist naloxone (4 mg/kg i.p.) to induce precipitated withdrawal (Fig. 7, Supplementary Table S3), mutant mice displayed less jumping (genotype: F_1,19=12.04, p<0.01; treatment: F_1,19=193.54, p<0.001; genotype×treatment: F_1,19=12.04, p<0.01), paw tremor (genotype: F_1,19=3.5, p=0.07; treatment: F_1,19=31.42, p<0.001; genotype×treatment: F_1,19=3.59, p=0.07) and teeth chattering (genotype: F_1,19=3.8, p=0.06; treatment: F_1,19=21.96, p<0.001; genotype×treatment: F_1,19=3.81, p=0.06). Thus, NMDA receptors on NA neurons act as an important modulator of behavioural symptoms of morphine withdrawal.

Fig. 7

Morphine withdrawal. (a) Morphine was administered s.c. at increasing doses for 5 d, and then withdrawal was precipitated by injection of 4 mg/kg naloxone (Nlx). Control groups of animals were challenged with saline. (b) Bar graphs summarize behavioural symptoms of morphine withdrawal: numbers of jumps, teeth chattering and paw tremors. Data are shown as mean±s.e.m., control (n=7) and mutant (n=6) mice were treated with naloxone, and five per genotype with saline. Significant differences (Bonferroni's post-hoc test) between naloxone-treated mutant and control mice are indicated by * p<0.05 and *** p<0.001.

Discussion

We show that NMDA-dependent input to NA cells controls their involvement in the development of specific morphine-induced behaviours. NR1^DbhCre mice do not develop psychomotor sensitization after intermittent treatment, and show attenuated withdrawal symptoms. However, the loss of functional NMDA receptors in noradrenergic cells had no apparent effects on the anxiety levels, learning ability or sensitivity to reinforcement in mutant mice.

The lack of morphine-induced psychomotor sensitization in NR1^DbhCre mice indicates a role for the noradrenergic system in incentive sensitization (Berridge & Robinson, 1998). This phenotype was stable in two independently bred cohorts of mutant mice, which excludes generation-dependent effects. Moreover, this phenotype could be reproduced using a pharmacological approach. When wild-type mice were pre-injected with clonidine, an α₂-adrenoceptor antagonist, which suppresses the brain's NA system, psychomotor sensitization to morphine did not develop. It should be noted that clonidine at the dose of 0.03 mg/kg may be causing sedation, and we observed a decrease in basal locomotor activity in clonidine-treated mice. Moreover, although α₂-adrenoceptor ligands act mainly at presynaptic sites, involvement of a post-synaptic NA mechanism or interaction with other neurotransmitter systems may not be completely excluded. Conversely, NR1^DbhCre mice had normal activity in the open-field apparatus, and furthermore, when they were treated with RX821002, a selective α₂-adrenoceptor antagonist which augments NA release, morphine psychomotor sensitization was rescued. Together, this strongly indicates that NMDA receptors are necessary for morphine-induced NA release, which in turn is essential for the development of psychomotor sensitization.

Here we report for the first time that the administration of morphine produced a robust, ∼3-fold increase of extracellular NA concentration in the medial NAc/striatum. Based on previously published data, this implies activation of NA inputs from the NTS and area A1, sole sources of NA in the NAc/striatum, chiefly projecting to the medial part of the NAc shell (Delfs et al., 1998, 2000; Olson et al.2006). It is not clear what the mechanism responsible for the NA increase could be, especially since direct administration of opioids inhibits activity of neurons (e.g. Williams et al.1982). However, previous microdialysis studies found an increase in extracellular NA in the prefrontal cortex after morphine injection (Ventura et al.2005). Thus, although direct action of morphine on NA cells is inhibitory, the net effect is an increase in their activity. In regard to our study, there is sparse data on origins of excitatory inputs on NA cells in the NTS or A1 with ascending projections to the striatum/NAc. Likely candidates are glutamatergic neurons from the brainstem and also excitatory afferents originating from the limbic brain areas or the hypothalamus (Baude et al.2009).

We found a mild increase in DA tissue content in the NAc of NR1^DbhCre mice, which was not associated with changes in its metabolite levels. Conversely, we observed that both in control and mutant mice morphine treatment caused a significant increase in extracellular levels of DA metabolites in the NAc (DOPAC and HVA), but levels of DA itself were only slightly increased (<30% compared to basal level). The change in extracellular DA levels is lower than expected based on previous reports (Chefer et al.2003; Murphy et al.2001). It should be noted, however, that the construction of the probe, its placement and the protocol used here were not identical to previous reports. Since a clear morphine-induced increase in extracellular levels of DA metabolites was observed, which implies a rise in extracellular DA levels, we believe that our data is in agreement with previous reports.

Interestingly, our data shows that the NMDA receptor-dependent inputs to NA neurons are either not involved or contribute weakly to the development of CPP or CPA. This is surprising because psychostimulant- or morphine-induced CPP was altered or abolished in Dbh and Adra1b knockout mice (Drouin et al.2002; Olson et al.2006). Moreover, there is ample evidence from pharmacological studies of NA system involvement in morphine-induced reinforcement (Hand et al.1989; Mantsch et al.2010; Tzschentke, 1998; Zarrindast et al.2002). In addition, morphine and other drugs induced NA release to the medial prefrontal cortex, and this increase was correlated to an increase in extracellular DA levels in the NAc as well as the development of CPP (Ventura et al.2005). Therefore, although NA signalling is involved in drug-conditioned responses, the NMDA receptor-dependent inputs are not essential in these mechanisms. Furthermore, based on previous reports and data reported here, we hypothesize that DA pathways have a dominant role in positive reinforcement, whereas NA signalling controls psychomotor sensitization (Airio & Ahtee, 1997; Tassin, 2008). Indeed, loss of functional NMDA receptors on DA neurons in NR1^DATCre mutants interferes with CPP, but has no effect on psychomotor sensitization (Engblom et al.2008). The complementarities between phenotypes observed previously in NR1^DATCre mice and reported here for NR1^DbhCre animals strongly indicate distinct behavioural roles of NMDA receptor-dependent plasticity in DA vs. NA neurons.

It should also be noted that the NA neurons play a central role in drug physical dependence, becoming strongly activated during opiate withdrawal (Aghajanian, 1978; Gold et al.1978; Rasmussen et al.1990). At the cellular level, the development of physical dependence was found to involve adaptive changes in intracellular cAMP-dependent signalling in LC NA neurons (Duman et al.1988). However, the exact role of this adaptation remains unclear, as it was reported that lesions of the LC do not attenuate symptoms of opiate withdrawal (Christie et al.1997). Furthermore, a specific deletion of cAMP-response element binding protein (CREB) in NA cells had minor effects on withdrawal symptoms (Parlato et al.2010). It was even suggested that the anatomical loci of withdrawal are non-NA neurons, and other brain areas, particularly the GABAergic neurons of the periaqueductal grey matter, are the actual substrates (Christie et al.1997). Our data indicate that the NMDA receptor-mediated signalling in NA neurons modulates opiate dependence, as the withdrawal symptoms were alleviated in NR1^DbhCre mice. This result is in agreement with the proposed role of the NA system and the glutamatergic afferents in mediating opioid withdrawal (Akaoka & Aston-Jones, 1991; Nestler et al.1999). We suggest a more significant role of NMDA receptors than previously attributed (Rasmussen, 1995), perhaps partially accounting for the observed alleviation of withdrawal symptoms following treatment with NMDA antagonists such as MK-801 (Trujillo & Akil, 1991).

Our results indicate that the importance of NA signalling in development of drug-conditioned behaviours has been underappreciated. Further efforts will focus on the significance of NMDA receptor-dependent plasticity of NA neurons in control of motivated behaviours.

Note

Supplementary material accompanies this paper on the Journal's website.

Acknowledgements

We thank Professor Günther Schütz for his support. J.R.P., W.S., J.K., S.G. and R.Prz. were supported by grants from the EU LSHM-CT-2004-005166 GENADDICT and LSHM-CT-2007-037669 PHECOMP, as well as the Polish Ministry of Science and Higher Education (MSHE) subsidiary grants 26/E-40/6.PR UE/DIE 305/2005-2008, 936/6.PR UE/2009/7 and 478/6.PR UE/2007/7. J.R.P. and R.Prz. were also supported by grant N405 143238 from the MSHE. W.S. received support from the Foundation for Polish Science individual grant ‘START’. K.G., K.T. and G.H. were supported by statutory funds from the MSHE awarded to the Institute of Pharmacology. R.S. was supported by the ‘Deutsche Forschungsgemeinschaft’ through the Collaborative Research Center SFB636/A4. M.N. and R.Pa. were supported by the Helmholtz Gemeinschaft Deutscher Forschungszentren through the Helmholtz Alliance for Mental Health in an Ageing Society (HelmA, HA-215). Language revision of one of the versions of the manuscript was performed by an external editor at ‘American Journal Experts’.

Statement of Interest

None.