INTRODUCTION

The behavioral effects of psychostimulants such as amphetamine and cocaine are mediated by their indirect activation of dopamine receptors in the nucleus accumbens and caudate-putamen (Koob, 1996). Psychostimulants are thought to be addictive due to neuronal and molecular adaptations both within and outside the mesoaccumbens circuitry (Vanderschuren and Kalivas 2000). Behavioral sensitization, which is described as a progressive augmentation of responses to repeated drug administration, is an expression of neuroadaptations. In humans, sensitization to psychostimulant drugs may appear as craving behavior or paranoia (Robinson and Berridge 1993; Kalivas et al, 1998). Critical roles have been established for dopaminergic and subsequently glutamatergic transmission in psychostimulant-induced behavioral sensitization (Wolf, 1998). Recently, the adenosine A2A receptor (A2AR) has also emerged as a potential modulator of psychostimulant sensitization and as an attractive therapeutic target. A2ARs are ideally located to modulate neuronal pathways involved in psychostimulant sensitization, given that their brain expression is largely restricted to the nucleus accumbens, striatum, and olfactory tubercle (Rosin et al, 2003). Furthermore, A2A adenosinergic and D2 dopaminergic systems interact at the membrane (Ferre et al, 1994; Canals et al, 2003), intracellular (Morelli et al, 1995), and behavioral levels (Ferre et al, 1993; Fenu et al, 1997). In addition, activation of A2ARs enhances the release of several brain neurotransmitters including dopamine and glutamate, which contribute to the development of psychostimulant behavioral sensitization (Quarta et al, 2004).

Recent pharmacological data provide direct evidence for an important role of adenosine A2ARs in the long-term adaptive responses to repeated dopaminergic stimulation both in rats (Bove et al, 2002; Bibbiani et al, 2003) and non-human primates (Bibbiani et al, 2003). In addition, using A2AR knockout (A2A KO) mice, our laboratory has shown that behavioral sensitization to repeated treatment either with L-dopa in hemiparkinsonian mice or with amphetamine in unlesioned mice does not develop in the absence of the A2AR (Fredduzzi et al, 2002; Chen et al, 2003). However, a facilitative role of A2ARs in sensitization is controversial and other reports have shown that the A2A agonist CGS21680 attenuates the development of behavioral sensitization induced by methamphetamine (Shimazoe et al, 2000). This discrepancy between genetic and pharmacological studies of A2ARs functions may reflect particular limitations of either of these approaches, for example, developmental or chronic inactivation of the A2AR in the KO model, or exogenous vs endogenous modulation of the adenosinergic system using a pharmacological agonist.

In order to clarify the role for A2ARs in behavioral adaptations induced by repeated psychostimulant exposure, we used a brain-specific conditional KO of the A2AR in coordination with a classical pharmacological approach. The genetic model allows us to specifically explore the A2AR in forebrain areas and to avoid possible compensatory developmental responses of other genes, given the fact that the deletion of the A2AR is postnatal. Moreover, we took advantage of newly available A2A antagonists, which offer improved specificity despite persistent problems of solubility and stability, to test the role of the A2AR in the development of amphetamine sensitization and its discrete phases (induction and expression).

MATERIALS AND METHODS

Generation and Genotyping of Postnatal Forebrain-Specific A2A Conditional KO Mice (Cre/loxP System)

The L7ag13 line of CaMKIIα-cre transgenic mice (in a C57Bl/6 background) expresses the Cre recombinase under the direction of the CaMKIIα gene promoter in postnatal neurons of the striatum as well as other forebrain structures and in germline cells (Dragatsis et al, 2000; Dragatsis and Zeitlin 2000; Morozov et al, 2003) and was kindly provided by WT Dauer, A Morozov, R Hen, and ER Kandel. Mice with a ‘floxed’ adenosine A2AR gene were generated by insertion of loxP sequences within the introns flanking a critical exon (2) of the A2AR gene (YJ Day and J Linden, unpublished results). Homozygous floxed (A2A flox/flox) mice (F5 generation in a mixed 129-Steel and C57Bl/6 genetic background) were crossed with L7ag13 cre(+) mice, and female cre(+) (A2A flox/+) offspring were then crossed with A2A flox/+ males. Their cre(+) A2A flox/flox and cre(−) A2A flox/flox offspring were compared in an initial autoradiographic and pharmacological assessment (Figure 1a, c, and d). A2A genotypes for several brain regions and peripheral tissues from (4-month-old) adult A2A flox/+ (and cre(−) A2A flox/+) littermates were compared (Figure 1b) to test for tissue-specific expression of the cre transgene and consequent Cre-mediated recombination of the floxed A2A allele.

Figure 1
figure 1

Tissue-specific, functional disruption of the adenosine A2AR. (a) A2AR autoradiography (3H-CGS21680 binding) of coronal brain and sagittal kidney sections from a 3-month-old cre(−) A2A flox/flox control mouse and its cre(+) A2A flox/flox conditional KO littermate qualitatively shows striatum-specific depletion of A2ARs. st: striatum; ctx: cortex; med: medulla. (b) Assessing A2AR genotypes of CNS and peripheral tissues at 4 months in cre(+) A2A flox/+ mice using oligonucleotide primers designed (as schematized on the left) to produce distinct PCR products for WT [+], floxed, and recombined [−] A2A alleles. Frontal cortex (FCtx), striatum (Str), cerebellum (Cbel), spleen (Spn), kidney (Kid), and Tail. (c) Motor response to KW-6002 (3 mg/kg, i.p.) in cre(+) A2A flox/flox conditional KO and their littermate controls. *p<0.001, n=3. (d) Locomotor response to KW-6002 (3 mg/kg, i.p.) in cre(+) A2A flox/− conditional KO mice and their littermate controls. ***p<0.001, n=8.

Full size image

Initial efforts to expand the numbers of cre(+) A2A flox/flox and cre(−) A2A flox/flox littermates were complicated by unexpected germline cre transgene expression and recombination events in some female cre(+) A2A flox/+ breeder mice (in addition to those expected in male cre(+) A2A flox/+ mice; Dragatsis et al, 2000; Dragatsis and Zeitlin 2000; Zakharenko et al, 2003). This newly identified CaMKIIα promoter-driven recombination in female as well as male gametes may reflect differences in Cre-mediated recombination in germline cells seen with different floxed genes—presumably due to differences in local chromatin structure effects on loxP site accessibility (Morozov et al, 2003). Thus, the expected A2A flox/flox offspring resulting from crossing a female cre(+) A2A flox/+ with male cre(−) A2A flox/+ mice were found to have one recombined (−) as well as one floxed A2A allele (ie to be A2A flox/−). The cre(+) A2A flox/− mice among these offspring were then crossed with their cre(−) A2A flox/− littermates in multiple matings to produce the 16 male A2A flox/− (half that were cre(+) and half that were cre(−)) employed in this study of amphetamine sensitization (Figure 2).

Figure 2
figure 2

Absence of amphetamine-induced behavioral sensitization in postnatal brain-specific A2A KO mice. Conditional A2A KO (cre(+)) mice and littermate (cre(−)) controls were habituated to test cages and injected with saline (day 0, panel a), treated with amphetamine (2.5 mg/kg, i.p.) daily for 8 days, and then rechallenged with the same dose of amphetamine after a week-long washout period. Locomotor responses to amphetamine were recorded on the first day of treatment (panel b), the 8th day of daily treatment (panel c), and upon rechallenge on day 15 (panel d). ***p<0.001. Cumulative locomotion for the 60 min after amphetamine injection is compared between genotypes for days 1, 8, and 15 in panel e (†p<0.05, †††p<0.001 vs day 1 of cre(−); *p<0.05, **p<0.01 vs the same day of treatment of cre(−); ‡p<0.05 vs day 1 of cre(+); n=7–8).

Full size image

Genotyping for the presence of the cre transgene and separately for the presence of the wild-type (WT), floxed, or recombined (inactivated) alleles of the A2AR gene was conducted by polymerase chain reaction (PCR) analysis of tail DNA unless otherwise indicated. The three-probe PCR strategy employed for A2A genotyping is schematized in Figure 1b, and is based on the location of loxP inserts within the gene (Day and Linden, unpublished results).

Animals and Drug Treatments

All experiments were performed in accordance with the Massachusetts General Hospital and NIH guidelines on the ethical use of animals. Both colony and commercial (129-Steel and C57Bl/6) mice (Charles River Laboratories) were habituated to the testing environment for 120 min prior to behavioral testing. In the dose–response studies (Figure 3a), mice were treated intraperitoneally (i.p.) with the A2A antagonist SCH58261, KW-6002, or vehicle (10% DMSO, 15% ethoxylated castor oil, 75% distilled water). In the amphetamine-induced sensitization studies (Figure 3b–d), as schematized in Figure 4, mice received daily injections (in their test cage) of amphetamine (2.5 mg/kg, i.p.) combined with vehicle, SCH58261 (0.03 mg/kg, i.p.), or KW-6002 (0.03 mg/kg, i.p.) for 8 consecutive days. A2A antagonists were injected 1–2 min before amphetamine. Locomotor activity was recorded (using an automated open field system) for 120 min following drug injection on days 1 and 8 and 1 and 2 weeks after the cessation of the treatment (days 15 and 22). In the induction study of amphetamine sensitization (Figure 5a), drug treatments are identical to those of the amphetamine sensitization studies up to day 15. On day 22, the locomotor activity was monitored upon challenge with amphetamine (2.5 mg/kg, i.p.) alone in all mice. In the expression study (Figure 5b), all mice were treated for 8 consecutive days with amphetamine alone (2.5 mg/kg, i.p.). A week after the last treatment (day 15), mice were treated either with amphetamine (2.5 mg/kg, i.p.) alone or paired with SCH58261 (0.03 mg/kg, i.p.), and their locomotor activity was recorded on days 1, 8, and 15. When using the A2A conditional KO mice in the amphetamine sensitization study (Figure 2), animals were treated with amphetamine (2.5 mg/kg, i.p.) alone.

Figure 3
figure 3

A2A antagonists attenuate amphetamine-induced behavioral sensitization. (a) In 129-Steel mice, A2A antagonists enhance locomotor activity at concentrations ⩾0.3 mg/kg (n=8–10). *p<0.05, **p<0.01, and ***p<0.001. (b) 129-Steel mice (n=10–18) were treated with amphetamine (2.5 mg/kg or saline) and SCH58261 (0.03 mg/kg or vehicle) daily for 8 days. After a washout period of 1 week, they were rechallenged (with the same treatments) on day 15 and again on day 22. *p<0.01 vs day 1 of the vehicle-treated mice; †p<0.01 vs amphetamine day 1; ‡p<0.05 vs amphetamine at the same day of treatment. (c) C57Bl/6 mice were treated with amphetamine (2.5 mg/kg) and KW-6002 (0.03 mg/kg) daily for 8 days (n=8). **p<0.01 vs amphetamine at day 1. (d) C57Bl/6 mice (n=7–9) were treated with amphetamine (2.5 mg/kg) and SCH58261 (0.03 mg/kg) as in panel b. *p<0.05 vs amphetamine at day 1; †p<0.05, ††p<0.01 vs amphetamine treatment at the same day.

Full size image
Figure 4
figure 4

Schematics of paradigms for assessing A2AR involvement in different phases of amphetamine-induced locomotor sensitization.

Full size image
Figure 5
figure 5

A2ARs are required for the induction of sensitization by amphetamine. (a) C57Bl/6 mice (n=8) were treated for 8 consecutive days with amphetamine (2.5 mg/kg) alone or paired to SCH58261 (0.03 mg/kg). On day 22, all mice were challenged with amphetamine (2.5 mg/kg) alone. *p<0.05 vs amphetamine at day 1; †p<0.05, ††p<0.01 vs amphetamine treatment at the same day. (b) C57Bl/6 mice (n=8) were treated with amphetamine (2.5 mg/kg) daily for 8 days. On day 15, mice were challenged with amphetamine paired to SCH58261 (0.03 mg/kg) or vehicle. *p<0.05, ***p<0.001 vs amphetamine at day 1. Note that although several fold variability in acute motor response to amphetamine is routinely observed between experiments (eg a vs b for day 1), the amphetamine-induced sensitization phenomenon occurs consistently.

Full size image

Locomotor and Fine Motor Activity

Horizontal locomotor and fine motor activity were primarily assessed by an automated recording system (San Diego Instruments) in standard polypropylene cages (15 × 25 cm) placed into adjustable frames equipped with five infrared photocell beams that traverse each cage in a plane above the floor. Locomotor activity (‘ambulation’) was measured as number of sequential breaks in two adjacent beams, and fine motor activity was measured as number of sequential breaks in a single beam. For the functional characterization of cre(+) A2A flox/flox mice (Figure 1c), a manual blinded recording method for locomotor activity (time spent in horizontal motion during each 10 min period) was used.

Receptor Autoradiography

Qualitative receptor autoradiography for detecting A2ARs using the specific ligand 3H-CGS21680 (46.0 Ci/mmol; NEN, Boston, MA) was performed as described previously (Chen et al, 1999). Coronal brain and tissue sections were preincubated at room temperature with 50 mM Tris-HCl buffer, pH 7.5, and 1 U of adenosine deaminase for 30 min and then incubated with the Tris buffer containing 5 nM 3H-CGS21680 for 60 min. To define nonspecific binding, 2.5 μM of 2-chloroadenosine was coincubated in adjacent sections.

Statistical Analysis

All data are expressed as mean±SEM. Statistical analyses were performed using Prism3 software. The effects of genotype and chronic treatment (treatment days 1, 8, and 15) were analyzed by two-way ANOVA followed by post-test using the Bonferroni method. For all the other behavioral studies, one-way ANOVA followed by Dunnett's test was applied.

RESULTS

Complete, Specific Inactivation of Adult Brain A2ARs in Conditional A2A KO Mice

To clarify the neurobiology of A2ARs, we generated conditional A2A KO mice using a forebrain-specific Cre/loxP system (Morozov et al, 2003). Transgenic mice expressing the Cre recombinase under the direction of the CaMKIIα gene promoter in postnatal forebrain neurons (including those of the striatum) were crossed with mice whose A2AR gene contained loxP excision sequences inserted on either side of (ie flanking) a critical exon, yielding so-called ‘floxed’ A2AR alleles. Successful forebrain-specific recombination was confirmed by autoradiographic, genetic, and behavioral assessments.

Receptor autoradiography with the A2A agonist 3H-CGS21680 demonstrates characteristic ligand binding to A2ARs in the striatum of adult nontransgenic (ie cre(−)) A2A flox/flox mice but complete absence of detectable binding in the striatum of cre(+) A2A flox/flox littermates (Figure 1a). By contrast, A2AR binding sites in the kidneys (specifically in the renal medulla, where A2AR expression is known to be enriched; Weaver and Reppert 1992) of the same mice appear indistinguishable in the cre(+) and cre(−) (A2A flox/flox) mice (Figure 1a). Together with genetic evidence against Cre-mediated recombination in all other peripheral somatic tissues tested (heart, spleen, tail; Figure 1b and data not shown) in CaMKIIα-cre(+) mice, these anatomical findings demonstrate the brain specificity of this conditional A2A KO approach. It should be noted however that disruption of the A2AR gene also occurred in some gonadal cells (due to variable germline expression of cre as expected with the CaMKIIα promoter; Dragatsis et al, 2000; Morozov et al, 2003), complicating breeding strategies to generate cre(+) A2A flox/flox mice (see Materials and methods).

In addition, within the CNS, the predicted further restriction of A2AR gene recombination to the forebrain was confirmed by the genetic (PCR) analysis, showing prominent recombination of the floxed A2AR allele in the frontal cortex and striatum but no detectable recombination in the cerebellum (as well as peripheral tissues) (Figure 1b). Note that the relatively small amount of residual floxed (nonrecombined) A2AR allele likely reflects the lack of cre expression in glial cells, since forebrain neurons rather than glia are primarily targeted in the CaMKIIα-cre(+) mice used here (Dragatsis and Zeitlin 2000; Morozov et al, 2003). On the other hand, incomplete recombination in striatal neurons could be excluded as a contributor to the residual floxed A2AR gene in the adult striatum.

In initially assessing the functional effects of eliminating A2ARs in adult striatal neurons, we examined behavioral responses to the selective A2A antagonist KW-6002 (3 mg/kg, i.p.) in conditional A2A KO mice. Although basal locomotor activity did not differ between conditional A2A KO (cre(+) A2A flox/flox or cre(+) A2A flox/−) mice and their respective control (cre(−)) littermates, KW-6002 induced locomotion only in the cre(−) controls (Figure 1c and d). In control mice from the CaMKIIα-cre line (expressing the cre transgene in the forebrain without a floxed A2A gene target), KW-6002 stimulated locomotion to the same extent as in their nontransgenic littermates, ruling out the possibility that the absence of A2A antagonist-induced locomotion in the conditional KO is simply due to the expression of cre alone. It is to be noted that these locomotor data provide the strongest evidence yet that KW-6002 and other relatively specific A2A antagonists enhance movement in Parkinson's disease patients (Bara-Jimenez et al, 2003; Hauser et al, 2003) as well as laboratory animals (Jenner 2003) through blockade of neuronal CNS A2ARs rather than non-neuronal or peripheral A2ARs. Together, these behavioral, anatomical, and genetic features of the Cre/loxP conditional A2A KO approach confirm that it selectively eliminates the A2AR from the forebrain of adult mice.

Amphetamine-Induced Sensitization Requires Brain A2AR Activation

We compared the effects of brain A2AR inactivation on locomotor responses to daily treatment with a low dose of amphetamine (2.5 mg/kg) in cre(+) A2A flox/− (conditional KO) mice and their cre(−) A2A flox/− (control) littermates. Although forebrain-specific A2AR depletion had no effect on the locomotor response to a novel environment or to a habituating saline injection on day 0 (Figure 2a), locomotor activity after the first dose of amphetamine was slightly greater in control mice (on day 1; Figure 2b and e). This is consistent with a partial A2AR dependence of the acute stimulant action of amphetamine, which we had observed with a global A2AR KO line (Chen et al, 2000). Continued daily treatment with amphetamine markedly enhanced (sensitized) locomotor responses in control mice (p<0.05 day 8 vs 1), whereas no sensitization to amphetamine occurred in their conditional KO littermates on day 8 (Figure 2c and e). Similarly, 1 week after discontinuation of daily amphetamine exposure (day 15), robust locomotor sensitization persisted in control (p<0.001 vs day 1) but was not seen in conditional A2A KO mice (Figure 2d and e), although at this point a slightly enhanced locomotor activity appeared to be present in the conditional KO group compared to day 1 (p<0.05) (Figure 2e). Given this result and the fact that on day 1 the amphetamine response in the conditional A2A KO was lower compared to the WT mice, we cannot exclude the possibility that the absence of sensitization on day 8 in the conditional A2A KO is a reflection of subthreshold motor responses at the dose of amphetamine used in this study. We have however noted that the absence of sensitization we observed in global A2A KO mice (Chen et al, 2003) was independent of the amphetamine dose used and was not attributable to a threshold effect on sensitization.

The possibility that the absence of sensitization in conditional A2A KO mice is due to the expression of the cre gene rather than the selective deletion of the A2AR was excluded by the finding that (L7ag13) control (A2A WT) mice expressing cre (ie in the absence of a floxed A2AR gene) showed the same level of amphetamine sensitization as their nontransgenic (ie cre(−), fully WT) littermates (ambulations for 120 min after the eighth amphetamine dose: 4154±844 and 3568±854, respectively). Since amphetamine is known to induce stereotyped stationary behaviors as well as locomotion, we considered the possibility that the lack of locomotor sensitization in conditional KO mice could be due to immobility associated with increased stereotypes in these mice. However, an enhancement of amphetamine-induced fine movement behavior in the control mice on days 8 and 15 (p<0.05) was also blocked in their A2A conditional KO littermates—with increases from day 1 (182±40 repetitive single photobeam breaks) to day 8 (391±86) to day 15 (456±101) in cre(−) A2A flox/− controls, vs no significant change from day 1 (123±22) to day 8 (220±69) to day 15 (251±69) in cre(+) A2A flox/− mice. Together, these data demonstrate that brain A2ARs play an important role in amphetamine-induced sensitization of both locomotor and stereotyped behaviors.

A2A Antagonists Attenuate Amphetamine-Induced Locomotor Sensitization

To investigate whether pharmacological blockade of A2ARs can also attenuate amphetamine sensitization, we first determined the dose of A2A antagonist to be paired with amphetamine administration. To avoid the confound of the motor stimulant effects of adenosine A2A antagonists (Popoli et al, 1998; El Yacoubi et al, 2000; Halldner et al, 2000), and given the evidence that doses of an A2A antagonist below its threshold for motor stimulation are capable of modulating striatal physiology (Monopoli et al, 1998; Popoli et al, 2002), we tested several doses of two different A2A antagonists, SCH58261 and KW-6002. Both dose-dependently stimulated locomotor activity, with 0.3 mg/kg being the lowest dose that stimulated locomotion (Figure 3a). Accordingly, we selected 0.03 mg/kg as a subthreshold dose (ie that does not enhance locomotor activity) for each antagonist to be coadministered with amphetamine in sensitization experiments.

Pairing daily amphetamine doses with an A2A antagonist (SCH58261 or KW-6002) prevented locomotor sensitization on day 8 in both 129-Steel and C57Bl/6 strains of mice (p<0.05) (Figure 3b–d). Upon rechallenge with amphetamine plus SCH58261 on days 15 and 22, C57Bl/6 mice still did not show sensitization, while sensitization remained undiminished in the amphetamine plus vehicle group (Figure 3d). In contrast, 129-Steel mice developed a sensitized response upon delayed rechallenge with amphetamine plus SCH58261 on days 15 and 22 (Figure 3b). Fine movements were also monitored and they paralleled the results with locomotor activity (data not shown). These data suggest that A2AR blockade prevents or delays the development of locomotor sensitization to amphetamine, and that A2ARs facilitate neuroadaptive changes induced by repeated dopaminergic stimulation.

A2ARs are Required for the Induction (but not Expression) of Amphetamine Sensitization

We also explored the effects of adenosine A2ARs on discrete phases of sensitization. In particular, we challenged with amphetamine alone mice that have been daily treated with amphetamine or amphetamine plus the A2A antagonist SCH58261 (0.03 mg/kg). Upon amphetamine challenge, locomotor activity of amphetamine plus SCH58261 chronically treated mice was still not sensitized and remained significantly lower than that of the amphetamine chronically treated mice (p<0.05) (Figure 5a). We also addressed possible A2AR involvement in the expression phase of amphetamine sensitization. After having induced sensitization with daily administration of amphetamine alone, mice were challenged either with amphetamine or amphetamine paired with SCH58261 (0.03 mg/kg). Amphetamine-treated mice as well as SCH58261-treated mice still showed sensitization (p<0.001 vs amphetamine day 1) (Figure 5b). The data suggest that A2ARs play a role in the induction or maintenance of amphetamine-induced sensitization rather than its expression.

DISCUSSION

We have shown that postnatal inactivation of brain adenosine A2ARs dramatically attenuates sensitized behavioral responses to repeated amphetamine administration using a conditional gene depletion technique in combination with classical pharmacology. We previously showed that a global KO of the A2AR (ie in all cells and at all times from conception onward) prevents amphetamine sensitization. However, our earlier study could not distinguish between a developmental, chronic, or acute inactivation of the A2AR as the basis for this phenotype. Given that the A2AR is expressed in brain as early as E-15 in rats (Weaver, 1993), it was possible that altered development of dopaminergic, glutamatergic, GABAergic, or other CNS signaling systems in A2A KO mice could lead to an alteration in their amphetamine-induced behavioral sensitization. Similarly, the global KO study could not distinguish between the effects of A2AR inactivation in brain and its many effects in the periphery. The absence of amphetamine sensitization in the conditional KO in the present study argues strongly against compensatory developmental modifications by A2AR depletion as the cause of altered amphetamine-induced sensitization because the particular (L7ag13) line of CaMKIIa-cre mice used here has been shown to reduce forebrain ‘floxed’ gene expression somewhere between postnatal days 6 and 60 (Dragatsis and Zeitlin, 2000). In addition, the conditional KO phenotype also largely excludes the possibility of non-CNS (or non-neuronal) A2ARs contributing to amphetamine sensitization, as we (Figure 1) and others (Dragatsis et al, 2000; Dragatsis and Zeitlin 2000) have found no evidence of Cre recombinase activity outside the brain (or neurons) in male CaMKIIa-cre mice except for that in testes, which is unlikely to be a major contributor to psychostimulant-induced sensitization.

Although the postnatal conditional KO strategy helps eliminate a role for developmental actions of A2ARs, the absence of forebrain A2ARs for weeks to months prior to repeated amphetamine administration in the cre(+), floxed A2A mice precludes a distinction between effects of chronic receptor depletion and the effects of acute inactivation just at the times of the amphetamine exposure. To address A2AR involvement in amphetamine sensitization with an even greater temporal resolution than afforded by the conditional KO, we turned to complementary pharmacological antagonists of the A2AR that were administered acutely together with the individual amphetamine doses. Pairing of A2A antagonists with amphetamine also prevented locomotor sensitization after eight daily drug injections in both the 129-Steel and C57Bl/6 mouse strains, but prevented persistent sensitization weeks later only in C57Bl/6 mice. The different durations of sensitization blockade in the two mouse strains might be related to their differences in the metabolism of the drugs as well as to different drug sensitivities in the CNS. It also might be possible that the phases of sensitization (eg induction and maintenance) are affected differently in different mouse strains. In any event, the recapitulation of the global and conditional A2A KO phenotype of attenuated amphetamine sensitization in the antagonist-treated mice strengthens further the evidence against developmental or prolonged actions of the A2AR as the basis for its facilitative role in psychostimulant sensitization. Thus, from the present findings, we can now conclude that postnatal forebrain A2ARs—probably on neurons—play a critical role in behavioral sensitization to repeated amphetamine administration.

The absence of behavioral sensitization to repeated amphetamine treatment in A2A KO and antagonist-treated mice may reflect a broader phenotype of attenuated adaptive motor responses to intermittent dopaminergic stimulation. Fredduzzi et al (2002) showed that in unilaterally 6-OHDA-lesioned (global) A2A KO mice, daily treatment with L-dopa did not produce progressively sensitized behaviors (contralateral rotations and grooming) compared to their WT littermates. In analogous pharmacological studies of A2AR involvement in neuroplasticity induced by L-dopa in hemiparkinsonian rodents, Bibbiani et al (2003) have recently shown that oral KW-6002 coadministered with L-dopa daily prevented the characteristic shortening of motor response to acute L-dopa challenge. Together, these studies raise the possibility that the maladaptive involuntary movements (known as dyskinesias) that develop after chronic L-dopa treatment in Parkinson's disease may be reduced or prevented by antagonist coadministration. This hypothesis was strongly supported by a study of parkinsonian non-human primates in which chronic oral administration of KW-6002 with a dopaminergic agonist completely prevented the delayed development of dyskinesias (Bibbiani et al, 2003). Furthermore, A2AR involvement in neural adaptations may extend beyond those induced by direct dopaminergic stimulation. For example, El Yacoubi et al (2001) recently reported that classical genetic deletion of A2ARs also attenuates a withdrawal syndrome after chronic alcohol administration.

On the other hand, not all pharmacological studies have supported a facilitative role for A2ARs in the neural adaptations that underlie sensitization. Shimazoe et al (2000) found that the A2A agonist CGS21680 attenuates sensitization to repeated methamphetamine administration in rats. Their use of a different psychostimulant drug and paradigm of sensitization, as well as an A2A agonist (which may be less relevant to endogenous adenosine actions on CNS A2ARs than are A2A antagonists) could account for the difference in results. Moreover, although Lundblad et al (2003) have confirmed that treatment of 6-OHDA-lesioned rats with an adenosine A2A antagonist alone did not elicit any abnormal involuntary movements while relieving motor disabilities, they did not observe any effect of KW-6002 on the severity of dyskinesias when it was coadministered with L-dopa. Another study of unilaterally 6-OHDA-lesioned rats found that an A2A antagonist reversed but did not prevent L-dopa-induced motor alterations (Bove et al, 2002). In general, all these studies have suggested an A2AR role in behavioral sensitization despite some differences in the nature of its role. The present study greatly strengthens the evidence that in the case of the brain A2AR, its role in psychostimulant sensitization is facilitative.

Our finding that pharmacological blockade of A2ARs can be as effective as their genetic depletion in preventing amphetamine sensitization adds to the therapeutic potential of A2A antagonists for neuropsychiatric diseases. Several A2A antagonists (eg KW-6002) are already in various phases of clinical trials as a novel symptomatic treatment for Parkinson's disease. Our findings support the possibility that brain A2A blockade may help prevent or delay the development of maladaptive dyskinetic motor responses to chronic dopaminergic stimulation (Pinna et al, 2001; Fredduzzi et al, 2002; Bibbiani et al, 2003). Moreover, our data raise the possibility that A2A antagonists could provide a rational pharmacological intervention for the treatment of addictive disorders. In support of A2A antagonists as therapy in neuropsychiatric disorders is the efficacy of very low doses, which are subthreshold for stimulatory motor effects. The development of sensitization may result from a series of transient neural adaptations that occur with each drug administration, ultimately establishing enduring changes in the response of the brain to subsequent drug administration. Our results implicating CNS A2ARs in the development rather than the expression of amphetamine sensitization indicate not only that CNS A2ARs play a critical role in sensitized psychostimulant responses, but also that they could be targeted to prevent or delay the maladaptive neuroplasticity that contributes to the induction or maintenance phases of some addictive behaviors.

The neurochemical mechanisms by which basal ganglia A2ARs may facilitate behavioral sensitization are unknown. A2AR inactivation may prevent behavioral sensitization by modulating presynaptic dopamine release (Zetterstrom and Fillenz 1990; Okada et al, 1996; Dassesse et al, 2001). Since there is no evidence of A2ARs expression on nigrostriatal neurons (Rosin et al, 2003), it has been suggested that A2AR-mediated facilitation of dopamine release may arise indirectly, that is, through regulation of glutamate and GABA release (Sebastiao and Ribeiro, 1996; Wolf, 1998; Corsi et al, 1999). Alternatively, a direct postsynaptic interaction between A2A and dopamine D2 receptor may facilitate amphetamine sensitization. In addition, the interaction among A2A and mGluR5 metabotropic glutamate receptors in the basal ganglia could also modulate psychostimulant-induced sensitization (Chiamulera et al, 2001). Changes in the expression of presumably postsynaptic A2ARs after repeated dopaminergic exposures might also play a functional role in the nucleus accumbens or dorsal striatum (Zeng et al, 2000; Calon et al, 2004; Tomiyama et al, 2004). Potential downstream postsynaptic mediators of sensitization that are also known to be regulated by the A2AR include cytoplasmic signal transducers (eg dopamine- and cAMP-regulated phosphoprotein of 32 kDa or DARPP-32) and nuclear transcriptional targets (eg ΔFosB, enkephalin, and dynorphin gene expression in striatal neurons; Fienberg et al, 1998; Canals et al, 2003; Lundblad et al, 2003; Hakansson et al, 2004).

In summary, by complementing classical pharmacology with a powerful new conditional KO approach to brain receptor inactivation, we have demonstrated that antagonists of the A2AR and its genetic disruption in the postnatal forebrain markedly attenuate behavioral sensitization to repeated amphetamine exposure. Furthermore, the findings indicate a critical if not requisite role for brain A2ARs in an early phase of psychostimulant-induced neuroplasticity. Thus, targeting the A2AR in the basal ganglia may provide a novel therapeutic strategy to prevent or reduce maladaptive biochemical and behavioral responses to repeated drug administration in human psychostimulant addiction.