INTRODUCTION

Repeated administration of psychostimulants (such as amphetamine or cocaine) induces an enhanced behavioral response to subsequent drug exposure, a phenomenon known as behavioral sensitization and capable of persisting for months (Robinson and Berridge, 1993; Pierce and Kalivas, 1997). The development of these maintained behavioral adaptations parallels the progressive and sustained enhancement of drug-craving and psychotic behaviors displayed by addicts only after repeated administration (Robinson and Berridge, 1993). Psychostimulant-induced behavioral sensitization in rodents provides a model of the addictive behaviors (such as those associated with craving and relapse) and psychotic complications of psychostimulant abuse (Robinson and Becker, 1986; Robinson and Berridge, 1993). Thus, understanding the neural adaptations associated with psychostimulant-induced behavioral sensitization may be relevant to the pathophysiology of psychostimulant-associated disorders such as drug addiction and psychosis.

The critical role of dopaminergic transmission in psychostimulant-induced behavioral sensitization has been inferred from evidence that psychostimulants enhance the level of dopamine in the synapse either by increasing presynaptic dopamine release (amphetamine) or blocking dopamine reuptake (cocaine) (Koob, 1992; Self and Nestler, 1995; Tan et al, 2000). Activation of both dopamine D1-like and D2-like receptors is involved in the behavioral response to amphetamine as antagonists at these receptors attenuate amphetamine-dependent behaviors (Hyman, 1996; Bardo, 1998; Wolf, 1998; Hyman and Malenka, 2001). Glutamatergic transmission has also been implicated in behavioral sensitization either through modulation of dopaminergic transmission or through an independent action (Wolf, 1998; Sripada et al, 2001). In spite of intensive studies of psychostimulant action, effective pharmacological strategies for treating chronic psychostimulant-associated disorders remain limited by side effects of glutamatergic and dopaminergic antagonists (Hyman, 1996; Hyman and Malenka, 2001). Moreover, the involvement of neurotransmitters other than dopamine and glutamate in psychostimulant-induced behavioral sensitization is largely unexplored, and may provide alternative therapeutic opportunities to modify psychostimulant effects.

The brain adenosine A2A receptor is emerging as a promising target site for the modulation of psychostimulant actions. Accumulating evidence supports functional interactions between the A2A adenosinergic and dopaminergic systems in the brain (Ferre et al, 1997; Svenningsson et al, 1999b). Brain A2A receptor mRNA is coexpressed with D2 receptor mRNA in the striatum and nucleus accumbens (Schiffmann et al, 1991; Fink et al, 1992) (Svenningsson et al, 1999a), a critical area involved in behavioral sensitization (Koob, 1992; Pierce and Kalivas, 1997). Activation of A2A receptors antagonizes D2 receptor-mediated behavioral and neurochemical effects, (eg on GABA and acetylcholine release and on the striatal expression of the neuropeptide enkephalin) (Ongini and Fredholm, 1996). Moreover, pharmacological studies show that A1 and A2A agonists inhibit acute amphetamine-induced locomotion in intact animals and amphetamine-induced rotation in rats unilaterally lesioned with 6-OHDA (Turgeon et al, 1996; Ferre et al, 1997; Rimondini et al, 1997; Chen et al, 2000). Similarly, A2A antagonists potentiate the acute motor effects of dopamine agonists, psychostimulants and L-dopa (Fenu et al, 1997; Fenu and Morelli, 1998; Poleszak and Malec, 2000). Thus, A2A receptors profoundly influence central dopaminergic mechanisms and psychostimulant action. This antagonistic interaction between A2A adenosine and dopamine receptors may be mediated by a direct A2A-D2 receptor–receptor interaction at an intramembrane level, as well as by an opposing, independent functional antagonism at the levels of postreceptor signaling pathways and of neural networks (Svenningsson et al, 1999a; Chen et al, 2001). In addition, activation of the A2A receptor has been shown to enhance the release of several neurotransmitters in brain including dopamine and glutamate, which contribute to the development of psychostimulant behavioral sensitization (Okada et al, 1996; Sebastiao and Ribeiro, 1996; Golembiowska and Zylewska, 1998). Together these studies suggest that the A2A receptor may represent an ideal site at which psychostimulant actions can be selectively modulated.

While pharmacological studies clearly demonstrate A2A adenosinergic modulation of acute psychostimulant action, relatively little is known about A2A receptor involvement in the development of behavioral sensitization by chronic psychostimulant treatment. Although recent reports showed that the A2A agonist CGS21680 attenuates the development of behavioral sensitization induced by methamphetamine or morphine (Weisberg and Kaplan, 1999; Shimazoe et al, 2000), there is no information available on the effect of A2A receptor blockade on chronic psychostimulant-induced behavioral sensitization. Furthermore, pharmacological investigation of A2A receptor involvement in behavioral sensitization is limited by the intrinsic partial specificity of A2A antagonists. A2A antagonists, generally also have poor solubility and poor CNS penetration. In addition, most A2A antagonists are unstable in solution and undergo rapid isoform conversion in light, resulting in reduced affinity for the receptor (Dionisotti et al, 1994; Nonaka et al, 1994; Ongini and Fredholm, 1996). All of these limitations make A2A antagonists difficult to use for in vivo pharmacological studies. In this study, we employed an A2A receptor knockout (A2A KO) model that provides specific and complete inactivation of A2A receptors in order to evaluate their role in the adaptive behavioral and neurochemical responses to repeated exposure to dopaminergic stimulation. Specifically, we examined whether the A2A receptor is required for behavioral sensitization in response to repeated treatments with amphetamine or dopaminergic agonists.

MATERIALS AND METHODS

Breeding and Genotyping of A2A KO Mice

The A2A KO mouse line was generated and genotyped as previously described (Chen et al, 1999). Mutant mice in a pure 129-Steel genetic background were used in this study to avoid potential confounding effects of commonly employed mixed genetic backgrounds. A2A KO mice in a pure 129-steel genetic background were produced by crossbreeding chimeric A2A KO mice (derived from embryonic stem cells of 129-steel background) to 129-steel mice (Taconic), and were maintained through breeding by heterozygote intercrosses (Chen et al, 1999). For each of the present experiments WT and KO littermates (both male and female) of the F2–F4 generations were matched for gender, age (2–5 months), and body weight.

Animals and Drug Treatments

All experiments were performed in accordance with Massachusetts General Hospital and NIH guidelines on the ethical use of animals. The homozygous KO (−/−) and WT (+/+) mice were housed in plastic cages and provided free access to food and water. The animals were maintained in temperature and humidity-controlled rooms with a 12-h light–dark cycle (light from 7:00 am to 7:00 pm). Prior to behavioral testing (which was conducted during the light phase of the light–dark cycle), all mice were habituated to the testing environment for 120 min. The mice were evaluated for spontaneous locomotion, as well as for locomotor activity in response to dopamine agonists. The mice were treated intraperitoneally with amphetamine (2.5 or 5.0 mg/kg), quinpirole (1 mg/kg), and SKF81297 (2.0 mg/kg) for 8–9 days. Locomotor activity was recorded at the first and eighth or ninth days of the treatment, or 2 weeks after the cessation of the treatment (24th day).

Locomotor and Fine Motor Activity

Horizontal locomotor and fine motor activities were assessed in standard polypropylene cages (15×25 cm2) placed into adjustable frames equipped with five infrared photocell beams (San Diego Instruments) that traverse each cage in a plane above its floor. Ambulation (sequential breaks in two adjacent beam), fine movement (sequential breaks in a single beam), and total locomotion (ambulation plus fine movement) were recorded and analyzed on a computer. On the first and eighth treatment day mice were weighed and placed in the test cages for a 120-min habituation period before drug injection. On the second through seventh treatment days mice remained in their home cages before and after injections.

6-Hydroxydopamine Lesions and Rotational Behavioral Analysis

Mice were pretreated with desipramine hydrocloride (25 mg/kg, Sigma), to minimize damage to noradrenergic neurons. Under Avertin (2% 2,2,2-tribromoethanol and 1% amyl alcohol) anesthesia (20 ml/kg, i.p.), 10 μg of 6-OHDA (2.5 μg/μl in normal saline containing 0.05% of ascorbic acid) was delivered by a microinfusion pump (1 μl/min) into the left dorsal striatum at the following coordinates (from Bregma point: 1.1 mm anterior, 1.5 mm lateral, 2.0 mm ventral). Seven days following 6-OHDA lesioning, rotational behavior was evaluated by an observer who was blind to the genotype of the animals. WT and A2A KO mice (n=9–12) were treated daily with quinpirole (1.0 mg/kg) for 3 weeks. Contralateral rotation behavior was evaluated in a test cage daily. The intensity and kinetic profile of L-dopa-induced contralateral behavior was established by monitoring the number of complete (360°) rotations ipsilateral and contralateral to the lesion in a 30-min test period immediately after the injection of quinpirole.

In Situ Hybridization Histochemistry

In situ hybridization histochemistry with cRNA probes was performed according to protocols described previously (Moratalla et al, 1996a). Mouse brain sections (10 μm thick) were cut in a cryostat and then sequentially postfixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), washed for three times in PBS, acetylated in acetic anhydride and dehydrated in graded ethanol. Sections were hybridized overnight at 55°C in a humid chamber with 35S-labeled cRNA probes (150.000 cpm/μl buffer) specific for the rat prodynorphin cDNA (provided by Dr J Douglass) (Civelli et al, 1985). Hybridization buffer contained 50% formamide, 4× SSC, 1× Denhardt's solution, 100 μg/ml salmon sperm DNA, 250 μg/ml yeast tRNA, 10% w/v dextran sulfate, and 100 mM dithiothreitol. After hybridization, sections were washed and then treated with RNase A (100 μg/ml), and washed again to final stringency in 0.1× SSC at 70°C for 30 min. The slides were rinsed, dried, and exposed to BioMax MR films (Amersham) for 10–15 days.

Dynorphin mRNA levels were determined by optical density using a computing densitometer equipped with an image analysis system (Model 300A, Molecular Dynamics). Five to six sections through the striatum (both hemispheres) were analyzed for each mouse.

Statistical Analysis

Statistical analyses of the behavioral data were performed using SAS (8.0) or SPSS (11.0) software. The effects of genotype (WT and KO) and chronic treatment (treatment days 1, 8, or rechallenge at day 24) and postinjection time (0–90 min) were analyzed by three-way ANOVA for repeated measurements. This is followed by post hoc comparisons between different treatment days at various postinjection times using Tukey's honestly significant difference test. For behavioral analysis of experiments using amphetamine at 5.0 mg/kg, we adapted a two-way ANOVA, with an unbalanced design for repeated measurements because of an unequal number of animals in the groups for different treatment days. Following two-way ANOVA, the LSD test was used for post hoc preplanned comparison.

RESULTS

A2A Receptor Inactivation Prevents Amphetamine-Induced Behavioral Sensitization in Mice

First we examined the effects of A2A receptor inactivation on locomotor responses to repeated amphetamine treatment in WT and A2A KO mice. Mice were treated with amphetamine (2.5 mg/kg, i.p.) daily for 8 days and locomotor response to amphetamine was monitored on the first and eighth day (Figure 1a). Acute treatment with amphetamine induced significant locomotion in both WT and A2A KO mice. Following daily injection of amphetamine for 7 days, WT mice displayed significantly enhanced locomotor response to the same dose of amphetamine on day 8 compared to their response on day 1 (Figure 1a, left panel, n=7–8, p<0.05, two-way ANOVA followed by Tukey's test). By contrast, A2A KO mice did not show any enhancement of locomotion on day 8 compared to day 1 (Figure 1a, right panel, n=8). Three-way ANOVA analysis shows that there were significant effects of treatment day (F(17,261)=3.95, p=0.031) and genotype (F(1,277)=16.6, p=0.002), and an interaction between genotype and treatment (F(17,251)=2.94, p=0.057).

Figure 1
figure 1

Absence of repeated psychostimulant-induced behavioral sensitization in A2A KO mice. WT and A2A KO mice (3- to 8-week-old, male and female) were treated with amphetamine (a) 2.5 mg/kg, (b) 5.0 mg/kg) daily for 8 or 24days. Ambulation was recorded for 120 min after amphetamine treatment on day 1 and either immediately following (day 8, (a) and (b)) or 2 weeks after (day 24, (b)) 1 week of daily amphetamine injections. WT=wild-type; KO=A2A receptor knockout. All mice are offspring of heterozygote crosses in a pure 129-steel background. *p<0.05, three-way ANOVA with repeated treatments for two factors, followed by Tukey's post hoc test, n=7 and 8 (a); n=11, 8, and 6 for days 1, 8, and 24, respectively (b).

Full size image

Furthermore, although the number of fine movements recorded in WT mice after the initial amphetamine injection was significantly higher than that in KO mice (WT vs KO, n=8, F(1,31)=13.4, p=0.003, two-way ANOVA followed by the LSD test), repeated treatments with amphetamine did not induce a significant increase in fine movements (comparing day 8 vs day 1, n=8, F(1,31)=0.61, p=0.447, two-way ANOVA followed by the LSD test) in either WT (mean±SE are 1570±151 and 1793±154 for days 1 and 8, respectively, over a 120 min period) or A2A KO mice (1017±154 and 940±148 for days 1 and 8, respectively). This suggests that the lack of amphetamine-induced locomotor sensitization in A2A KO mice is not owing to sensitized stereotyped behavior of the type reflected in horizontal fine movements, which could have masked the appearance of enhanced locomotor activity in these mice.

To distinguish between stress-induced and amphetamine-induced behavioral sensitization in WT mice we also compared amphetamine-stimulated motor activity following 1 week of daily saline injections with amphetamine-stimulated motor activity in naïve mice. Locomotor activity simulated by amphetamine (2.5 mg/kg) in WT mice measured on day 8 (290±42 ambulations per 120 min) following seven daily saline injections was significantly increased over saline-stimulated locomotion on day 1 (62±17, n=8, p<0.001, two-way ANOVA followed by Tukey's test), but was indistinguishable from amphetamine-stimulated locomotor activity on day 1 (202±13; p>0.05, n=8, two-way ANOVA followed by Tukey's test). Thus, the A2A receptor-dependent sensitization of the motor response to amphetamine primarily reflects amphetamine- (rather than injection stress-) induced behavioral sensitization.

The Absence of Behavioral Sensitization in A2A KO Mice is not Attributable to a Threshold Effect or Delayed Sensitization

The process of sensitization may require that individual locomotor responses to each amphetamine injection achieve a threshold magnitude in order to generate enhanced responses upon repeated administration. Indeed, we found that the small motor stimulation produced by at a very low dose (1.25 mg/kg) amphetamine did not augment with repeated daily administration under our standard 1-week sensitization paradigm (data not shown). Since the locomotion induced by a single dose of amphetamine was found to be partially attenuated in A2A KO compared to WT mice (Chen et al, 2000); and also in Figure 1a, by ∼25% although not significantly, the absence of sensitization observed in KO mice using amphetamine at 2.5 mg/kg (Figure 1a) could have simply reflected subthreshold motor responses at that dose. We addressed this possibility by testing the effects of a higher dose of amphetamine. In order to exceed any such motor response threshold in A2A KO mice, we treated the mice daily for 8 days with the amphetamine dose doubled (to 5 mg/kg), and with locomotor responses again recorded in activity cages on days 1 and 8 (Figure 1b).

As expected, on day 1 the higher amphetamine dose induced a greater cumulative locomotor response in KO mice (1405 total ambulations on average over the 2 h after 5 mg/kg in A2A KO mice; Figure 1b, upper right panel) than the lower dose had induced in WT mice (994 after 2.5 mg/kg in WT mice; Figure 1a, left panel). Analysis by two-way ANOVA for repeated measurements shows that there were significant effects of treatment day (F(1,569)=7.70, p=0.010) and the interaction between treatment day and genotype (F(1,569)=6.06, p=0.021). If the lack of sensitization observed in A2A KO mice at 2.5 mg/kg amphetamine were in fact because of a subthreshold motor response, we would now expect a robust sensitization in KO mice treated with double the dose (since their initial motor response of 1405 ambulations would now be well above a possible motor response threshold of ⩽994). However, the sensitized locomotor response to 5 mg/kg amphetamine observed on day 8 in WT mice (p<0.05, n=13–15, two-way ANOVA followed by Tukey's test) was again completely prevented by A2A receptor inactivation in the KO mice (Figure 1b, upper panels, n=11 and 8 at days 1 and 8, respectively). These data argue strongly against a nonspecific motor threshold effect as the basis for attenuated locomotor sensitization in A2A KO mice.

In addition, the locomotor sensitization in WT mice and its absence in A2A KO mice persisted for at least 2 weeks after discontinuation of daily amphetamine injections (Figure 1b, lower panels, n=11 and 6 for days 1 and 24). Two-way ANOVA analysis with an unbalanced design shows that there was a borderline significant effect of treatment day (F(1,519)=4.05, p=0.050). While the extent of sensitization in WT mice appeared to be long lasting (Figure 1b, n=10–12, p<0.05, comparing the day 24 vs day 1, two-way ANOVA followed by Tukey's test) with at least as much expression on day 24 as on day 8 (Figure 1b, lower vs upper panel, n=10–12), delayed rechallenge with amphetamine still did not produce an enhanced motor response in KO mice. These data suggest that A2AR inactivation prevents (rather than delays) locomotor sensitization to amphetamine.

Repeated D1 or D2 Agonist Treatment Produced Indistinguishable Behavioral Sensitization in WT and A2A KO Mice

In contrast to amphetamine, the full D1 receptor agonist SKF81297 (2.5 mg/kg) produced an identical locomotor sensitization in WT and A2A KO mice, following the same 8-day treatment schedule used for amphetamine (Figure 2a). The peak locomotion induced by SKF81297 in WT mice was (mean±SE) 176±18 and 254±26 ambulations per 10 min at days 1 and 8, respectively (n=8, p<0.05, three-way ANOVA followed by Tukey's test). Similarly, peak locomotion by SKF81297 in A2A KO mice were 131±22 and 292±21 ambulations per 10 min at days 1 and 8, respectively (n=7, p<0.05, three-way ANOVA followed by Tukey's test). However, there was no significant difference in motor responses by SKF81297 treatment between WT and A2A KO littermates on either day (Figure 2a, n=7). Three-way ANOVA shows that there were significant effects of treatment day (day 1 vs day 8) (F(1,215)=11.15, p=0.075) and postinjection time (F(1,215)=23.76, p<0.001), and an interaction between treatment day and postinjection time (F(1,215)=3.78, p=0.001), but that there was no significant effect of genotype (F(1,215)=0.00, p=0.970).

Figure 2
figure 2

Dopamine D1 and D2 agonists induced indistinguishable behavioral sensitization in WT and A2A KO mice. (a) Mice were treated with D1 agonist SKF81297 (2.0 mg/kg, i.p.) daily for 8 days, and ambulation was recorded for 120 min after the injection on days 1 and 8 as described in the Methods. Repeated treatments with SKF81297-induced sensitized locomotor responses in both WT and A2A KO mice (n=8, *p<0.05, three-way ANOVA followed by Tukey's post hoc test). However, the locomotor sensitization was indistinguishable between WT and A2A KO mice (n=7–8, F(1,215)=0.02, p=0.96, three-way ANOVA comparing WT and A2A KO mice at days 1 and 8, respectively). (b) To study behavioral sensitization by D2 agonists, WT, and A2A KO mice were first unilaterally lesioned by intrastriatal injection of 6-OHDA. Seven days after lesioning, mice were treated with daily injections of quinpirole (1.0 mg/kg, i.p.). Contralateral rotations were recorded for 30 min on days 1 and 13. Repeated treatments with quinpirole apparently increased rotational responses in both WT and A2A KO mice (n=7).

Full size image

Owing to the specific antagonistic interaction between adenosine A2A and dopamine D2 receptors, we also addressed the role of D2 receptor involvement in A2A receptor modulation of amphetamine-induced behavioral sensitization by comparing the effects of repeated treatments with D2 agonists in WT and A2A KO littermates. Using the same 1-week daily treatment paradigm, we failed to demonstrate locomotor stimulation by several D2 agonists, including quinpirole, bromocriptine and pramipexole at various doses (from 0.1 to 5 mg/kg), and under both context-dependent conditions (ie in which the mice were placed in the test cage after each injection regardless of whether locomotion was measured) and our standard context-independent conditions (data not shown). This prompted us to adopt a parallel paradigm of repeated D2 agonist treatment in which rotational responses to daily i.p. quinpirole administration were measured daily in unilaterally 6-OHDA-lesioned mice for 2 weeks. Acute treatment with quinpirole produced typical stereotyped behavior (freezing, and fixed and stretched postures for several seconds to minutes) without any appreciable locomotion in WT and A2A KO mice. Although repeated treatment with quinpirole for 8 days did not induce locomotor sensitization in either WT or A2A KO mice, continued daily treatment apparently increased rotational behavior by day 13 in both WT and A2A KO mice (Figure 2b), which remained enhanced for the duration of the treatment period (data not shown). This sensitized rotational behavior was observed in both WT and A2A KO mice, (n=7), although there was considerable variation in the rotational response to quinpirole. These results suggest that individual D1- and D2-mediated signaling pathways leading to motor sensitization were not affected by the A2A receptor.

Repeated Combined Administration of D1 and D2 Receptor Agonists also Produced Indistinguishable Sensitization in WT and A2A KO Mice

Since D1–D2 receptor interactions are critical to dopaminergic function, the A2A receptor might contribute to amphetamine-induced sensitization by altering the interplay between its D1 and D2 components (ie without actually altering sensitization produced by these individual components). To address this possibility we also examined the effect of A2A receptor inactivation on locomotor sensitization induced by coactivation of D1 and D2 receptors. Mice received intraperitoneal injections of SKF81297 (2.5 mg/kg) and quinpirole (1 mg/kg) 1 min apart, daily for 8 days. Locomotor responses were recorded at days 1 and 8. Acute treatment with the combination of SKF81297 and quinpirole produced significantly less motor stimulation (compared to SKF81297 alone) in both WT and A2A KO mice (at day 1; see Figure 3, compared to Figure 2a; F(8,127)=4.96, p=0.001, two-way ANOVA). Nevertheless, following 1 week of daily combined treatments, SKF81297 plus quinpirole produced locomotor sensitization in both WT (Figure 3, n=7–8,) and A2A KO mice (Figure 3, n=7–8,). However, there was no difference in locomotor sensitization between WT and A2A KO mice (Figure 3, n=6–8, p=0.85 and 0.32 for days 1 and 8, respectively, LSD test). Three-way ANOVA shows that there were significant effects of treatment day (day 1 vs day 8 or 9) (F(1,251)=33.15, p=0.001), but clearly there were no effects of genotype (F(1,251)=1.68, p=0.219). There were also significant effects of postinjection time (F(1,251)=23.31, p<0.001) and a significant interaction between treatment day and postinjection time (F(1,251)=29.3, p<0.01). Thus, inactivation of A2A receptors did not affect locomotor sensitization induced by direct D1 and D2 agonists, either alone or in combination.

Figure 3
figure 3

Coinjection of D1 and D2 agonists induced indistinguishable locomotor behavioral sensitization in WT and A2A KO mice. WT (+/+) and A2A KO (−/−) mice were injected with SKF81297 (2.0 mg/kg, i.p.) and quinpirole (1.0 mg/kg, i.p.) one min apart, daily for 8 days. Locomotor responses to the coinjection of SKF81297 and quinpirole were recorded for 120 min after the treatment on days 1 and 8. Daily coadministration of SKF81297 and quinpirole produced enhanced locomotor activity (n=7–8, *p<0.05, three-way ANOVA followed by Tukey's test, comparing day 8 to day 1 for WT or A2A KO mice), but the locomotor sensitization was indistinguishable between WT and A2A KO mice (n=7–8, F(1,251)=1.68, p=0.22, three-way ANOVA comparing the WT with A2A KO mice at day 8).

Full size image

Repeated Amphetamine Induces Striatal Dynorphin mRNA in WT but not A2A KO Mice

Finally, we also examined the effect of A2A receptor inactivation on the induction of dynorphin mRNA, a cellular readout of striatal function. Dynorphin mRNA is predominantly coexpressed with dopamine D1 receptor mRNA in striatonigral neurons (of the ‘direct’ striatal output pathway) (Gerfen et al, 1990; Le Moine et al, 1991), and has been implicated in neuroadaptative responses to psychostimulants (Xu et al, 1994a,1994b; Moratalla et al, 1996b). Consistent with previous studies, repeated treatments with amphetamine significantly enhanced dynorphin mRNA expression in the striatum of WT mice (Figure 4, p=0.044 compared to saline-treated mice, two-way ANOVA followed by the LSD test). However in the absence of the A2A receptor, repeated amphetamine administration had no effect on dynorphin mRNA expression (Figure 4, p=0.641 comparing amphetamine to saline treatments in A2A KO mice by two-way ANOVA followed by the LSD test).

Figure 4
figure 4

Repeated treatments with amphetamine increased striatal dynorphin mRNA in WT but not in A2A KO mice. WT and A2A KO mice were treated with amphetamine (2.5 mg/kg, i.p.) daily for 8 days. At 120 min after the last injection of amphetamine, mice were killed and whole brains were removed and sectioned through the striatum. Mouse brain sections were hybridized to radiolabeled dynorphin RNA probe as described in the Methods. Striatal dynorphin mRNA levels were quantified by densitometry and expressed in arbitrary optical density units. *p<0.05, t-test, compared to the saline-treated group. The numbers inside the bar represent animal numbers for the corresponding groups.

Full size image

DISCUSSION

Amphetamine-Induced Behavioral Sensitization Requires Adenosine A2A Receptors

The colocalization of A2A and D2 receptor mRNA in the dopamine receptive areas, the antagonistic interaction between adenosine A2A and dopamine receptor systems, and the A2A receptor-mediated facilitation of glutamate and dopamine release, all point to the A2A receptor as an important target for regulating psychostimulant-induced behavioral sensitization. Using the A2A KO model we have unambiguously demonstrated that inactivation of A2A receptors can abolish amphetamine-induced behavioral sensitization. The absence of behavioral sensitization in A2A KO mice was demonstrated following repeated treatments with amphetamine for 1 week as well as after a 2-week washout period (ie upon rechallenge at day 24), indicating a long-lasting effect of A2A receptors in amphetamine sensitization. Furthermore, the absence of amphetamine-induced behavioral sensitization cannot be attributed to a nonspecific threshold effect since a higher dose of amphetamine still failed to produce any locomotor sensitization in the A2A KO mice (Figure 1b). Finally, at the cellular level, repeated amphetamine induced an increase in dynorphin mRNA in WT but not A2A KO mice. These results strongly suggest that the activation of the adenosine A2A receptor is required for behavioral sensitization induced by repeated psychostimulant treatment.

The absence of behavioral sensitization to repeated amphetamine treatment in A2A KO mice reflects their broader phenotype of attenuated adaptive motor responses to intermittent dopaminergic stimulation. Our recent studies demonstrate that in unilaterally 6-OHDA-lesioned animals, daily L-dopa treatment produced progressively sensitized behaviors (contralateral rotation and grooming) in WT mice, but not in A2A KO mice. Similarly, repeated L-dopa treatment reversed the 6-OHDA-induced reduction of striatal dynorphin mRNA in WT but not in A2A KO mice (Fredduzzi et al, 2002). Furthermore, El Yacoubi et al (2001) recently reported that genetic depletion of A2A receptors also attenuates a withdrawal syndrome after chronic treatment with alcohol. Together these results demonstrate that the A2A receptor plays a critical role in facilitating the neural adaptations that underlie behavioral sensitization.

Although the few available pharmacological studies of adenosine receptors in sensitization phenomena have suggested an important role for adenosine, its effects may differ depending on the particular adenosine receptor subtype and sensitizing stimulus involved (Weisberg and Kaplan, 1999; Shimazoe et al, 2000; El Yacoubi et al, 2001). Our complementary transgenic approach has provided strong evidence that the A2A receptor can in fact play a critical role in sensitization to certain dopaminergic stimuli. Nevertheless, our specific conclusion that the A2A receptor may facilitate the development of sensitization after repeated amphetamine administration seems to differ from that suggested by Shimazoe et al (2000), who found that the A2A agonist CGS21680 attenuates sensitization to repeated methamphetamine administration. The apparent discrepancy between our A2A KO study and their A2A agonist study may result from the intrinsic limitations of adenosinergic pharmacology (particularly when relying on agonists that do not assess the more physiological role of endogenous ligands as do antagonists).

Alternatively, limitations of the standard KO technology employed here may also have contributed to the dissociation between transgenic and pharmacological data on the role of the A2A receptor. The A2A receptor is completely inactivated throughout development as well as during adult life in the KO mice studied here. Given that A2A receptors can be detected at E16 and that the effect of chronic A2A receptor inactivation in adulthood is not known, it is possible that the A2A receptor deficiency may alter the development or maintenance of dopaminergic, glutamatergic, cholinergic, and GABAergic systems, and in turn lead to altered amphetamine-induced behavioral sensitization. The findings that A2A KO mice produced normal behavioral sensitization in response to direct D1 and D2 agonists are, however, not consistent with this generalized, nonspecific alteration in brain, and argue for selective impairment of behavioral sensitization via an apparent presynaptic site. Despite indistinguishable D1 and D2 agonist-induced behavioral sensitization, and despite our neuroanatomical characterization of the striatum in A2A KO mice showing no evidence for such a developmental effect of A2A receptor deficiency (Chen et al, 1999,2000,2001), such a possibility has not been excluded. Future studies of mutant mice with an inducible gene depletion of the A2A receptor may be needed to clarify this potential confounding effect.

Adenosine A2A Receptors may Modulate Psychostimulant Action through Interaction with Dopamine and Glutamate Systems

Extensive adenosine–dopamine interactions in the brain suggest that adenosinergic modulation of psychostimulant effects may result from the regulation of dopaminergic activity. At the postsynaptic level, specific changes in G proteins, adenylyl cyclase, and transcription factors (such as c-Fos, c-Jun and CREB) have also been documented to accompany chronic amphetamine-induced behavioral sensitization (Moratalla et al, 1996a; Nestler and Aghajanian, 1997). Since A2A receptors are known to be positively coupled by Gs to adenylyl cyclase and PKA activity (Svenningsson et al, 1999a), inactivation of A2A receptors may affect psychostimulant sensitization by influencing the cAMP signaling pathway, which in turn affects long-term gene expression associated with psychostimulant-induced behavioral sensitization.

Since A2A and D2 receptors are colocalized in striatopallidal neurons and have antagonistic interactions in brain, we attempted to evaluate D2 agonist-induced behavioral sensitization, and its modulation by the A2A receptor. However, instead of a motor stimulation, we consistently observed a motor depressant effect of D2 agonists, which appears to be stronger and more consistent in mice than that in rats. Similar motor depression by D2 agonists in mice at the doses used in the current study have been reported by other groups (Xu et al, 1997; Wang et al, 2000). This precluded the direct comparison between amphetamine and D2 agonist-induced sensitization in identical paradigms, thus limiting the strength of the evidence that A2A receptor involvement in amphetamine sensitization does not involve postsynaptic D2 receptors. Nevertheless, the demonstration that behavioral sensitization to direct D1 and D2 agonists (alone or in combination) is similar in A2A KO and WT littermates argues against significant alteration in postsynaptic D1 or D2 receptor-mediated signaling pathways as the basis of attenuated sensitization in the A2A KO mice. The selective absence of amphetamine-induced behavioral sensitization suggests that A2A receptor inactivation may prevent behavioral sensitization by impairing a presynaptic mechanism. A prominent presynaptic contribution to psychostimulant-induced sensitization has long been supported by the findings of enhanced dopamine release in the striatum and nucleus accumbens after chronic amphetamine treatment (Robinson and Becker, 1986; Kalivas and Stewart, 1991; Pierce and Kalivas, 1997). In this context, the findings that A2A agonists generally increase and antagonists decrease dopamine accumulation in striatal microdialysates (Zetterstrom and Fillenz, 1990; Okada et al, 1996; Okada et al, 1997) suggest a presynaptic mechanism through which A2A receptor inactivation could prevent psychostimulant-induced sensitization. In agreement, our preliminary study showed that depolarization-elicited dopamine release was significantly attenuated in striatal synaptosomes from A2A KO mice (Chen et al, 1998). This notion is further supported by recent in vivo data demonstrating that basal dopamine efflux in the striatum, measured by microdialysis, is significantly lower in A2A KO mice compared to their WT littermates (Dassesse et al, 2001). Together these studies suggest that A2A receptors can facilitate dopamine release in the striatum, and inactivation of A2A receptors may attenuate psychostimulant sensitization by attenuating dopamine release.

The notion of A2A receptor-mediated facilitation of dopamine release suggests that A2A receptor regulation of dopaminergic activity may depend on a fine balance between pre- and postsynaptic functions of the A2A receptor. Thus, A2A receptor-mediated presynaptic facilitation of dopamine release may offset the A2A receptor-mediated postsynaptic inhibition of D2 receptor function to ‘fine tune’ the final cellular and behavioral output from central dopaminergic pathways.

In addition, A2A receptors may modulate psychostimulant-induced behavioral sensitization through interactions with neurotransmitter systems other than dopamine. For example, both muscarinic (Heidbreder and Shippenberg, 1996) and nicotinic (Schoffelmeer et al, 2002) cholinergic receptors have been implicated in psychostimulant behavioral sensitization, and A2A receptor stimulation is capable of facilitating the release of acetylcholine in the striatum (Kurokawa et al, 1994). Thus, A2A antagonists could attenuate amphetamine behavioral sensitization in part by reducing acetylcholine release in the striatum. Moreover, increasing evidence suggests that activation of A2A receptors enhances glutamate release in the striatum and cortex (Sebastiao and Ribeiro, 1996). Activation of A2A receptors can also modulate NMDA receptor conductance (Norenberg et al, 1998; Wirkner et al, 2000). Both NMDA and non-NMDA glutamate receptor antagonists have been shown to interfere with the development of behavioral sensitization to amphetamine (Wolf, 1998). Thus, inactivation of A2A receptors could block behavioral sensitization by diminishing glutamate release presynaptically or affecting NMDA receptors in neurons directly. Moreover, recent studies have suggested that metabotropic glutamate receptors also regulate psychostimulant-induced neurochemical and behavioral effects (Chiamulera et al, 2001; Swanson et al, 2001). Interestingly, Chiamulera et al (2001) reported that reinforcing and locomotor stimulant effect of cocaine are absent in mice lacking mGluR5. Several studies have demonstrated a close interaction between A2A receptor and mGluR5 receptors in modulating quinpirole-induced turning, D2 receptor binding, and GABA release (Kearney and Albin, 1995; Kearney et al, 1997,1998; Popoli et al, 2001; Diaz-Cabiale et al, 2002). Therefore, A2A receptors could either directly or indirectly (eg through glutamatergic pathways) exert a modulatory effect on psychostimulant-induced sensitization.

Finally, d'Alcantara et al (2001) recently reported that genetic depletion or pharmacological blockade of A2A receptors attenuates long-term potentiation in the nucleus accumbens without affecting basal synaptic transmission. Thus, inactivation of A2A receptors may impair LTP and learning processes to influence psychostimulant-induced behavioral sensitization.

In summary, we demonstrated that genetic inactivation of A2A receptors abolishes locomotor behavioral sensitization and dynorphin mRNA expression induced by repeated treatments with amphetamine. The absence of locomotor behavioral sensitization appears selective for amphetamine since D1 and D2 agonists alone or in combination produce identical locomotor sensitization in WT and A2A KO mice, pointing to a presynaptic action of the receptor. These results indicate that behavioral sensitization to repeated psychostimulant administration requires the activation of the adenosine A2A receptor. Thus, targeting the brain A2A receptor may provide a pharmacologically specific strategy to prevent or reverse maladaptive biochemical and behavioral responses to repeated drug administration in human psychostimulant addiction.