Dopamine D2 receptors (Rs) and adenosine A2ARs are coexpressed on striatopallidal neurons, where they mediate opposing actions. In agreement with the idea that D2Rs tonically inhibit GABA release from these neurons, stimulation-evoked GABA release was significantly greater from striatal/pallidal slices from D2R null mutant (D2R−/−) than from wild-type (D2R+/+) mice. Release from heterozygous (D2R+/−) slices was intermediate. However, contrary to predictions that A2AR effects would be enhanced in D2R-deficient mice, the A2AR agonist CGS 21680 significantly increased GABA release only from D2R+/+ slices. CGS 21680 modulation was observed when D2Rs were antagonized by raclopride, suggesting that an acute absence of D2Rs cannot explain the results. The lack of CGS 21680 modulation in the D2R-deficient mice was also not caused by a compensatory downregulation of A2ARs in the striatum or globus pallidus. However, CGS 21680 significantly stimulated cAMP production only in D2R+/+ striatal/pallidal slices. This functional uncoupling of A2ARs in the D2R-deficient mice was not explained by reduced expression of Gs, Golf, or type VI adenylyl cyclase. Locomotor activity induced by the adenosine receptor antagonist caffeine was significantly less pronounced in D2R−/− mice than in D2R+/+ and D2R+/− mice, further supporting the idea that D2Rs are required for caffeine activation. Caffeine increased c-fos only in D2R−/− globus pallidus. The present results show that a targeted disruption of the D2R reduces coupling of A2ARs on striatopallidal neurons and thereby responses to drugs that act on adenosine receptors. They also reinforce the ideas that D2Rs and A2ARs are functionally opposed and that D2R-mediated effects normally predominate.
- adenosine A2A receptor
- dopamine D2 receptor
- D2 receptor knock-out mouse
- CGS 21680
- [3H]SCH 58261
- [3H]CGS 21680
- striatopallidal pathway
- GABA release
- cAMP stimulation
- type VI adenylyl cyclase
- locomotor activity
Efferent neurons from rodent striatum project either directly to the substantia nigra or indirectly via the globus pallidus (for review, see Gerfen, 1992). Both of these striatal projections are GABAergic. However, striatonigral and striatopallidal neurons express different combinations of peptides and receptors. Striatopallidal neurons are distinguished from striatonigral neurons by expression of the preproenkephalin gene and a high density of dopamine D2 and adenosine A2A receptors (Rs) (Schiffmann et al., 1991; Fink et al., 1992; Schiffmann and Vanderhaeghen, 1993;Svenningsson et al., 1997b).
The coexpression of A2ARs and D2Rs on striatopallidal neurons provides an anatomical basis for the opposing interaction that exists between these receptors. Opposing A2AR/D2R effects have been shown at several different levels, including behavior, neurotransmitter release, receptor binding, and gene expression (for review, seeFerré et al., 1992, 1997). For example, A2AR stimulation reduces D2R-mediated locomotor activity (Ferré et al., 1991). Likewise, A2AR activation antagonizes the D2R agonist-induced decrease in GABA release from globus pallidus (Ferré et al., 1993; Mayfield et al., 1996).
Two lines of D2R null mutant (D2R−/−) mice have been generated (Baik et al., 1995; Kelly et al., 1997). The striatal D2R density in heterozygous (D2R+/−) mice is ∼50% of that in wild-type (D2R+/+) mice, whereas no specific striatal D2R binding is detectable in D2R−/− mice. As anticipated, D2R−/− mice show some impairments in spontaneous locomotor activity (Baik et al., 1995;Kelly et al., 1998). D2Rs are inhibitory modulators of neurotransmitter release in the striatum. D2 autoreceptor inhibition of dopamine release is abolished in striatal synaptosomes from D2R−/− mice (L'hirondel et al., 1998). Interestingly, however, basal and potassium-evoked extracellular dopamine levels, measured within vivo microdialysis, are similar in D2R+/+ and D2R−/− mouse striata (Dickinson et al., 1999).
D2R deficiency may alter release of other striatal neurotransmitters, e.g., GABA. Indeed, Baik et al. (1995)observed an increase in mRNA expression of the GABA synthetic enzyme glutamic acid decarboxylase (GAD) in the striatum of D2R−/− mice, suggesting that GABA release from striatal projection neurons is increased. Enhanced GABA release from striatopallidal neurons in D2R−/− mice would be predicted whether GABA release is tonically inhibited by D2Rs and/or the action of A2ARs is unopposed by D2Rs. The adenosine receptor antagonist caffeine induces locomotor activation, and this behavioral hyperactivity has been shown to involve A2ARs, D2Rs, and GABA (Mukhopadhyay and Poddar, 1995; Svenningsson et al., 1997a; Khisti et al., 2000). A better understanding of the interactions between A2ARs and D2Rs may lead to novel therapies to treat basal ganglia movement disorders, such as Parkinson's disease and Huntington's disease (for review, seeOngini and Fredholm, 1996; Sebastião and Ribeiro, 1996;Ferré et al., 1997; Moreau and Huber, 1999). Therefore, we have further examined the significance of A2AR/D2R interactions using D2R-deficient animals. We determined whether the ability of the selective A2AR agonist CGS 21680 to modulate GABA release and induce cAMP production, as well as the ability of the adenosine receptor antagonist caffeine to induce locomotor activity and c-fos expression, differed among D2R+/+, D2R+/−, and D2R−/− mice.
MATERIALS AND METHODS
Subjects. Male and female N5congenic mice (20–35 gm; Vollum Institute, Portland, OR) were used in most of the experiments. Detailed methods by which these mice were produced have been reported (Kelly et al., 1997, 1998). Briefly, a D2R-genomic clone from a 129/SvEv library was isolated and used for the construction of a replacement-type targeting vector. Homologous recombination in D3 embryonic stem cells produced a mutant D2R allele, which has a deletion of exon 8 sequences encoding the sixth and seventh putative transmembrane domains, the third extracellular loop, and the cytoplasmic C-terminal tail. Germ-line transmitting chimeras derived from targeted embryonic stem cells were mated to wild-type females to generate F1 heterozygous mice on the mixed 129S2/SvPas × C57BL/6J background. The mutated D2R allele was then backcrossed an additional five generations by successive matings of heterozygous mice to wild-type C57BL/6J mice. The congenic N5 mice used here were derived from intercrosses of heterozygous mice that yielded all three possible genotypes in normal Mendelian proportions: D2R+/+, D2R+/−, and D2R−/−. The genotypes of all mice were confirmed by Southern blotting. Male C57BL/6J mice (20–30 gm) obtained from The Jackson Laboratory (Bar Harbor, ME) were used only in the experiments testing the effects of raclopride alone and in combination with CGS 21680 on GABA release. Mice were housed in groups of two to five under a 12 hr light/dark cycle with food and water available ad libitum. All animal-use procedures were in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Animal Care and Use Committee at the University of Colorado Health Sciences Center.
In vitro GABA release. Mice were killed by cervical dislocation, and 400 μm coronal brain slices were cut before dissecting out the region containing the striatum/globus pallidus (Franklin and Paxinos, 1997). During preparation the tissue was maintained in ice-cold modified Krebs' buffer (118 mmNaCl, 4.7 mm KCl, 11.1 mm d-glucose, 25 mm NaHCO3, 1.2 mm MgCl2, 1.0 mmNaH2PO4, 2.6 mmCaCl2, and 4.0 μmNa2-EDTA, saturated with 95% O2/5% CO2, pH 7.4, at 34°C). The slices were equilibrated in a metabolic shaker at 34°C for 30 min before being transferred to the superfusion chambers.
The GABA release method has been described previously (Mayfield and Zahniser, 1993). Briefly, striatal/pallidal slices were superfused at a rate of 0.25 ml/min with modified Krebs' buffer (34°C) containing 1-(2- (((diphenylmethylene)imino)oxy)ethyl)-1,2,5,6-tetrahydro-3-pyridine-carboxylic acid hydrochloride (NO-711) (10 μm) to inhibit GABA uptake. Three trains of monophasic rectangular pulses (12 Hz, 2 msec; 30 mA, 1 min) were applied at t = 72, 104, and 136 min after the start of superfusion, with the collection of 24 consecutive 1 ml superfusate fractions beginning att = 60 min (Fig. 1). CGS 21680 (100 nm; stock solution made up in DMSO), raclopride (1 μm), or vehicle was included in the buffer at t = 120 min, 16 min before the third stimulation.
GABA was quantified by HPLC with electrochemical detection after precolumn derivatization with o-pthaldialdehyde (Mayfield and Zahniser, 1993). The mobile phase was 1 mdisodium phosphate, pH 6.4 (0.5 m final concentration), 0.86 mm NaCl, and 37.5% acetonitrile. The lower sensitivity limit was < 25 pg of GABA per 50 μl injection. Spontaneous GABA outflow, designated PS1, PS2, or PS3, was defined as the mean concentration of GABA in the three fractions immediately preceding each of the three periods of stimulation, respectively. Stimulation-evoked GABA overflow, designated S1, S2, or S3, was determined from the summed amount of GABA release that exceeded PS1, PS2, or PS3, respectively.
cAMP levels. Striatal/pallidal slices were prepared as outlined above and incubated at 34°C in multiwell plates in a metabolic shaker, with a change in modified Krebs' buffer at 30 and 45 min during a 1 hr equilibration period. Slices were then incubated in 0.5 ml of fresh buffer containing either no addition, 1 μm CGS 21680, 10 μm CGS 21680, or 1 μm CGS 21680 + 150 μm8-(p-sulfophenyl)theophylline (8-p-SPT) at 34°C for 15 min. The incubation was terminated by the addition of 0.25 ml of 2.4% perchloric acid, sonication, and centrifugation at 30,000 × g for 15 min (Lu and Ordway, 1997). Pellets were dissolved in 0.2 ml of 0.1N NaOH for protein determination (Bradford, 1976) using bovine serum albumin as the standard. The supernatant was neutralized with excess (∼20 mg) CaCO3 (Thion et al., 1977). After centrifugation, the supernatant was evaporated to dryness and reconstituted in 0.25 ml of 50 mm sodium acetate buffer, pH 6.2, and cAMP was measured in duplicate 0.1 ml samples by radioimmunoassay (NEN Life Science Products, Boston, MA).
Locomotor activity. Mice were allowed to acclimate in individual transfer cages to the behavioral testing room for 30 min. Saline or caffeine (15 mg/kg) was then injected intraperitoneally in a volume of 1 ml/100 gm. Each mouse received only a single injection. Immediately after injection, mice were placed into individual activity chambers (San Diego Instruments, San Diego, CA). The room lights were turned off, and locomotor activity was recorded as the total distance traveled during 5 min periods for 120 min. Caffeine-induced locomotor activity was determined at each time point as a “difference score” by subtracting the mean activity of each genotype after saline injection from the activity of each mouse of that same genotype after caffeine injection.
Quantitative receptor autoradiography. Mice were killed 4 hr after saline or caffeine injection. Brains were frozen in powdered dry ice and stored at −70°C. Coronal sections (10 μm) were cut at −20°C at the levels of the rostral striatum/nucleus accumbens and the caudal striatum/globus pallidus. Binding of [3H]SCH 58261 and [3H]CGS 21680 was assayed using the published methods of Fredholm et al. (1998). Briefly, specific binding of 0.3 nm [3H]SCH 58261, defined with 50 μm5′-N-ethylcarboxamidoadenosine (NECA), was measured in 170 mm Tris-HCl buffer, pH 7.4, containing 2 U/ml adenosine deaminase. For [3H]CGS 21680, the assay buffer also contained 10 mmMgCl2. Indirect saturation curves were generated with [3H]CGS 21680 (2.5 nm) by incubating slide-mounted brain sections with either no drug (total binding), one of nine concentrations of unlabeled CGS 21680 (1 nm–10 μm), or 2-chloroadenosine (20 μm; nonspecific binding). [3H]raclopride binding was measured as described by Johansson et al. (1997). Specific binding of 2 nm [3H]raclopride, defined with 1 μm (+)-butaclamol, was measured in 170 mm Tris-HCl buffer, pH 7.6, containing 120 mm NaCl, 5 mm KCl, 2 mm CaCl2, 1 mm MgCl2, and 0.001% ascorbic acid. Slides and tritium standards were apposed to tritium-sensitive film for either 3 weeks ([3H]SCH 58261), 6 weeks ([3H]CGS 21680), or 8 weeks ([3H]raclopride).
Films were analyzed using computer-based imaging systems (Imaging Research, St. Catherines, Ontario, Canada). Inplot software (Graph Pad, San Diego, CA) was used to fit the indirect saturation curves. The affinity (Kd ) and number of receptors (B max) were determined from the equations published by DeBlasi et al. (1989). For statistical analyses, data from saline- and caffeine-treated mice of the same genotype were pooled on the basis of the observations that (1) caffeine does not interfere in binding assays because it is easily dissociated from A2ARs by washing (Johansson et al., 1996) and (2) no differences between the saline- and caffeine-treated groups were detected when these data sets were analyzed separately.
In situ hybridization. Mice were killed 4 hr after saline or caffeine injection. Previously published methods were used to measure mRNA for A2ARs and c-fos(Svenningsson et al., 1997a,b, 1999); these papers also describe the specificity of the probes. Briefly, consecutive coronal brain sections (14 μm) were cut with a cryostat and thaw-mounted onto poly-l-lysine-coated slides. The A2AR probe was a riboprobe, 208 bases long, encoding amino acids 1196–1405 of the rat A2AR protein (Svenningsson et al., 1997b) and was35S labeled by in vitrotranscription using 35S-labeled UTP. The c-fos probe was an oligodeoxynucleotide probe, 48 bases long, encoding amino acids 137–152 of the rat c-Fos protein. The oligodeoxynucleotide probe (Scandinavian Gene Synthesis AB, Köping, Sweden) was labeled using terminal deoxynucleotidyl transferase (Pharmacia, Uppsala, Sweden) and35S-labeled α-dATP (NEN Life Science Products, Stockholm, Sweden) to a specific activity of ∼109 cpm/μg. Sections were hybridized in 50% deionized formamide (Fluka, Buchs, Switzerland), 4× standard sodium citrate, 1× Denhardt's solution, 1% sarcosyl, 0.02m NaPO4, pH 7.0, 10% dextran sulfate, 0.5 mg/ml yeast tRNA (Sigma Labkemi, Stockholm, Sweden), 0.06 m dithiothreitol, 0.1 mg/ml sheared salmon sperm DNA, and 107 cpm/ml probe. After hybridization for 15 hr at 42°C, the sections were washed four times, for 15 min each, in 1× standard sodium citrate at 55°C; dipped briefly in water and 70, 95, and 99.5% ethanol; and air-dried. The sections were exposed to tritium-sensitive film for 2–5 weeks. The films were analyzed as in the autoradiographic studies above.
The expression of mRNA for the α subunits of Gsand Golf, as well as for type VI adenylyl cyclase (AC VI), was examined using cRNA35S-riboprobes essentially as described previously (Le Moine et al., 1997; Svenningsson et al., 1997b) using sense and antisense probes for the corresponding rat proteins (Jones and Reed, 1987, 1989; Glatt and Snyder, 1993). The transcription was performed using MAXI-script in vitro transcription kits according to the manufacturer's protocol (Ambion, Austin, TX). The probes were separated from unincorporated ribonucleotides using Sephadex G-50 chromatography. No signals were detected with the sense probes.
Drugs. [3H]SCH 58261 was a gift from Dr. Ennio Ongini (Schering-Plough, Milan, Italy); [3H]raclopride and [3H]CGS 21680 were obtained from NEN Life Science Products (Boston, MA, or Stockholm, Sweden); and NO-711, CGS 21680, raclopride, 8-p-SPT, caffeine, NECA, 2-chloroadenosine, and (+)-butaclamol were obtained from Sigma/RBI (St. Louis, MO).
Endogenous GABA release was evoked from striatal/pallidal slices by three periods of electrical stimulation (Fig. 1). The GABA uptake inhibitor NO-711 (10 μm) was present throughout the entire experiment. Control release was assessed in response to the first two stimuli. Spontaneous GABA outflow was measured before each of these stimuli and equaled ∼0.18 ng of GABA per mg wet weight of tissue per ml in the D2R+/+ and D2R+/− mice (Fig.2 A). In comparison with both the D2R+/+ and D2R+/− mice, spontaneous GABA release was significantly elevated by ∼40% in the D2R−/− mice (Fig.2 A). In response to each of the two stimuli, GABA overflow was significantly increased in all three genotypes. However, as seen in rat striatal/pallidal slices (Mayfield and Zahniser, 1993;Mayfield et al., 1996), the second stimulation evoked consistently lower GABA release than did the first stimulation in all of the mice (Fig. 2 B). This is not caused by depletion of releasable pools of GABA but rather involves activation of GABABRs (Mayfield and Zahniser, 1993). Nonetheless, both the first and second stimuli increased GABA overflow to a significantly greater extent, by 71 and 63%, respectively, in the D2R−/− versus the D2R+/+ mice (Fig.2 B). Stimulated GABA overflow in the D2R+/− mice was intermediate and did not differ statistically from that of either of the other two genotypes (Fig. 2 B). Because the S2/S1 ratios were similar in all three genotypes, the data indicate that GABABR-mediated autoinhibition is unaltered by elimination of D2Rs.
The A2AR agonist CGS 21680 was introduced before the third period of stimulation (Fig. 1) because preliminary experiments showed a more consistent A2AR-mediated modulation of GABA release when the effects of CGS 21680 were tested during the third, rather than the second, stimulation. The more consistent results could reflect the fact that the second and third stimuli released similar amounts of GABA (Fig. 3; controlS3/S2 ratios = ∼1) whereas the second stimulus evoked less release than did the first stimulus (Fig.2 B; S2/S1 ratios < 1). The inclusion of 100 nm CGS 21680 in the superfusion buffer 16 min before the third stimulation did not alter spontaneous GABA outflow from striatal/pallidal slices from any of the three genotypes (data not shown). However, in the D2R+/+ mice, exposure to 100 nm CGS 21680 significantly increased stimulation-evoked GABA overflow by 90% (Fig. 3;S3/S2 ratio). In contrast, stimulated release was not significantly altered in the presence of CGS 21680 in either the D2R+/− or D2R−/− mice (Fig.3).
To test the effect of acute blockade of D2Rs on CGS 21680 modulation during the GABA release assay, we measured release in C57 mouse striatal/pallidal slices in the presence of maximally effective concentrations of CGS 21680 and/or the D2R antagonist raclopride. Neither drug altered spontaneous GABA outflow (data not shown). In agreement with the results of the D2R+/+ mice, CGS 21680 (100 nm) significantly increased stimulation-evoked GABA overflow from the C57 mouse slices (Fig.4; 154% above control). Raclopride (1 μm) also significantly increased evoked release by 119% (Fig. 4). When CGS 21680 and raclopride were combined, overflow was increased to a significantly greater extent (256%) than when raclopride was present alone (Fig. 4). These results demonstrate that in the presence of D2R blockade, CGS 21680 increased GABA release from mouse striatal/pallidal slices and suggest that an acute absence of D2Rs cannot explain the lack of A2AR modulation in the D2R-deficient mice.
A2AR mRNA, A2ARs, and D2Rs
Downregulation of A2ARs would be one explanation for the lack of CGS 21680-modulated GABA release in the D2R-deficient mice. A2AR mRNA could be quantitated in the striata and nucleus accumbens, areas containing cell bodies of neurons expressing A2ARs, but was not detectable in the globus pallidus, the area containing terminals of neurons expressing A2ARs (Fig. 5). In all three genotypes, lower levels of A2AR mRNA were expressed in the nucleus accumbens than in the striatum (Table1). However, within each brain region, no significant differences among the genotypes were observed in A2AR mRNA expression.
[3H]SCH 58261 is a relatively new antagonist that is highly selective for A2ARs versus A1Rs (∼800-fold) (Fredholm et al., 1998). Its binding was quantitated by autoradiographic analysis in the striatum, nucleus accumbens, and globus pallidus of the D2R+/+, D2R+/−, and D2R−/− mice. An approximate Kd concentration of [3H]SCH 58261 was used (0.3 nm) (Fredholm et al., 1998). In contrast with A2AR mRNA, specific A2AR antagonist-binding sites were observed in both the striatum and globus pallidus (Fig. 5). Levels of specific binding were approximately threefold higher in the striatum than in the globus pallidus, with that in the nucleus accumbens being intermediate (Table 1). However, in agreement with the A2AR mRNA determinations, similar levels of binding were observed within each brain region of the three genotypes (Table 1). Binding of the D2R antagonist [3H]raclopride was also measured in the striatum and nucleus accumbens of these mice. In comparison with the D2R+/+ mice, levels of D2R binding were 50% lower in both brain regions of the D2R+/− mice and were not detectable in the D2R−/− mice (Table1). These results agree with those reported for the F2 generation of these mice (Kelly et al., 1997).
Indirect saturation curves were generated using the A2AR agonist [3H]CGS 21680 and quantitative autoradiographic analysis in the striatum, nucleus accumbens, and globus pallidus in the D2R+/+, D2R+/−, and D2R−/− mice. Assays contained 10 mm Mg2+ to induce the high-affinity agonist-binding state, and the curves were best fit by a single-site model. The affinities, ranging from 23 to 29 nm, were similar in all brain regions of the three genotypes (Table 2). In agreement with the [3H]SCH 58261 results, the receptor densities measured with [3H]CGS 21680 were approximately threefold higher in the striatum than in the globus pallidus, with that in the nucleus accumbens being intermediate (Table2). Likewise, there were no differences among genotypes in the densities of A2AR agonist-binding sites within a single brain region.
An alternative explanation to receptor downregulation that could underlie the lack of CGS 21680 modulation of GABA release in the D2R+/− and D2R−/− mice would be an uncoupling of A2ARs from adenylyl cyclase. To test this hypothesis, basal and CGS 21680-stimulated cAMP levels were measured in striatal/pallidal slices from the three genotypes. Preliminary experiments confirmed that, similar to previous results in rats (Lupica et al., 1990), maximal stimulation was produced by 1 and 10 μm CGS 21680. Furthermore, preliminary experiments showed that the adenosine receptor antagonist 8-p-SPT (150 μm) blocked the increases in cAMP induced by 1 μm CGS 21680, confirming that this is an A2AR-mediated response. Basal levels of cAMP were similar in the three genotypes (Fig. 6; legend). CGS 21680 induced a significant 64% increase in cAMP formation in the D2R+/+ mice, whereas cAMP levels in the D2R−/− mice were not altered by CGS 21680 (Fig. 6). Although there was a trend for CGS 21680 to increase cAMP (34% above basal) in the D2R+/− mice, this change was not statistically significant (Fig. 6). These results suggest that normal A2AR signaling via increased cAMP production is disrupted in the striatopallidal neurons in the D2R-deficient mice.
Levels of mRNA for the stimulatory G-proteins Golf and Gs and for AC VI were measured by in situ hybridization in the D2R+/+, D2R+/−, and D2R−/− mice. These experiments were conducted as an initial investigation into putative downstream-signaling mechanisms that might explain the compromised ability of A2ARs to stimulate cAMP accumulation in the striatum/globus pallidus of D2R-deficient mice. The isoform(s) of adenylyl cyclase to which striatopallidal A2ARs couple is unknown. However, we focused on AC VI because its mRNA is present in both the striatum and globus pallidus (Liu et al., 1998) and A2ARs are known to activate AC VI in pheochromocytoma 12 cells (Chern et al., 1995). Interestingly, Golf and AC VI mRNAs, but not Gs mRNA, were readily detected in the striatum (Fig. 7). Thus, in this respect the mouse appears similar to the rat (Hervé et al., 1993). However, among the three genotypes, no differences were observed in the levels of expression of any of these mRNAs (Fig. 7).
Caffeine-induced locomotor activity
Caffeine is known to stimulate locomotion secondarily to blockade of A2ARs (Ledent et al., 1997; Svenningsson et al., 1997a; Hauber et al., 1998). Caffeine-induced locomotor activity was therefore measured as an additional test of A2AR function in the D2R+/+, D2R+/−, and D2R−/− mice. After acclimatization to the behavioral testing room, the mice were injected with either saline or caffeine (15 mg/kg, i.p.), and locomotor activity was measured for 2 hr (Fig.8 A,B). During the first 45 min after saline injection, the activity of the D2R−/− mice was significantly lower than that of the D2R+/+ and/or D2R+/− mice (Fig.8 A). To factor out baseline activity differences, caffeine-induced difference scores were generated (Fig.8 C). Mice of all three genotypes injected with caffeine were more active than were the respective controls injected with saline (Fig. 8 C). Over the 2 hr period after caffeine injection, the total caffeine-induced distance traveled by the D2R+/+ mice was 42100 ± 9750 cm (n = 5), that traveled by the D2R+/− mice was 47200 ± 6970 cm (n = 5), and that traveled by the D2R−/− mice was 24200 ± 3100 cm (n = 9). Statistical analysis revealed that the less-pronounced caffeine-induced activation of the D2R−/− mice reflected the fact that their activity was significantly lower than that of the D2R+/+and/or D2R+/− mice from 45 to 100 min after injection (Fig. 8 C). Direct observation of D2R−/− mice injected with caffeine revealed that the lower caffeine-induced locomotor activation was not caused by an increase in stereotypic behaviors (data not shown).
Caffeine-induced c-fos expression
The differences in caffeine-induced behavioral activation among the three genotypes might reflect differences in the D2R-mediated activity of intrinsic pallidal neurons. To investigate this possibility, we quantitated the levels of c-fos mRNA by in situ hybridization 4 hr after injection of either saline or caffeine (15 mg/kg). Expression of c-fos in the cerebral cortex, measured as a control, was similar in saline- and caffeine-treated mice of all three genotypes (data not shown). Expression of c-fos also did not differ significantly in the globus pallidus of saline- or caffeine-treated D2R+/+ and D2R+/− mice (Fig.9). In contrast, c-fosexpression was significantly increased by 94% in the globus pallidus of the caffeine-treated, versus the saline-treated, D2R−/− mice (Fig.9). Caffeine-induced c-fos expression in the D2R−/− mice was also significantly higher than caffeine-induced c-fosexpression in either the D2R+/+ or D2R+/− mice (Fig.9).
Evidence of tonic D2R inhibition of striatopallidal GABA release
Activation of D2Rs inhibits stimulation-evoked GABA release from striatopallidal neurons. Conversely, in striatal/pallidal slices from control C57 mice, we observed that the D2R antagonist raclopride increased GABA release. We also hypothesized that GABA release from these striatal projection neurons would be elevated in mice lacking D2Rs. In agreement with this hypothesis, we observed significantly greater (40–70%) spontaneous and electrically evoked GABA release from D2R−/− than from D2R+/+ mouse striatal/pallidal slices. GABA release from D2R+/− mice was intermediate. Our observations show the essential correctness of the surmise of Baik et al. (1995). They found higher levels of striatal GAD mRNA in a different line of D2R−/− mice and, on the basis of this observation, suggested that GABA release might be increased. Together, these observations strengthen the idea that D2Rs tonically inhibit GABA release from striatopallidal neurons.
The reduced baseline locomotor activity observed here in the N5 congenic D2R−/− mice is compatible with disinhibited pallidal GABA release. During the initial 40 min after saline injection, the D2R−/− mice were significantly less active than were the other two genotypes. Subtle differences in the initiation of movement were also observed between D2R+/+ and D2R−/− mice in the F2 generation (Kelly et al., 1998). The lower level of locomotor activity in the D2R−/− mice is consistent with pharmacological studies showing that increased GABAA receptor activation, as well as decreased dopamine receptor activation, reduces locomotor activity (Mukhopadhyay and Poddar, 1995).
Lack of A2AR agonist effects in D2R-deficient mice
An antagonistic interaction between D2Rs and A2ARs in striatopallidal neurons is well established (for review, see Ferré et al., 1992, 1997). This antagonistic interaction impacts GABA release. The A2AR agonist CGS 21680 not only increases stimulation-evoked GABA release from rat striatal/pallidal slices but also abolishes the D2R agonist-mediated inhibition of this release (Mayfield et al., 1993, 1996). Likewise,Ferré et al. (1993) observed that CGS 21680 antagonizes the D2R-mediated reduction in extracellular GABAin vivo in rat globus pallidus. Here we observed that CGS 21680 also markedly increased stimulation-evoked GABA release from D2R+/+ and C57 mouse striatal/pallidal slices. Thus, we hypothesized that without opposing D2Rs, the stimulation of GABA release by A2ARs would be potentiated. Unexpectedly, however, CGS 21680 did not affect GABA release in either D2R+/− or D2R−/− mice. It is unlikely that the lack of CGS 21680 potentiation in the D2R-deficient mice was caused by an acute absence of D2Rs during the assay. Stimulation-evoked GABA release from control mouse striatal/pallidal slices was still augmented by CGS 21680 when the D2R antagonist raclopride was included in the assay. It is also unlikely that a ceiling effect, i.e., that GABA release was already maximal, explains the lack of CGS 21680 potentiation because stimulated GABA release from D2R+/− slices was not statistically higher than that from D2R+/+ slices (Fig.2 B).
Uncoupled A2ARs in D2R-deficient mice
Loss of A2AR effects in the D2R+/− and D2R−/− mice likely reflects changes, which may have occurred during development, to compensate for the reduced number of D2Rs. Previous studies (Baik et al., 1995; Kelly et al., 1998) suggest that striatonigral activity is not increased in D2R-deficient animals. Downregulation of A2AR expression was the first potential change that we investigated. Both A2ARs and D2Rs are expressed at very low levels during early striatal development and reach adult levels only at the end of the second postnatal week (Johansson et al., 1997). Nonetheless, levels of A2AR mRNA in the striatum or globus pallidus were similar in the D2R+/+, D2R+/−, and D2R−/− mice. The antagonist [3H]SCH 58261 has the same high affinity for both G-protein-coupled and -uncoupled states of the A2AR (Fredholm et al., 1998). Thus, its binding measures the total complement of A2ARs. In contrast, in the presence of 10 mmMg2+, [3H]CGS 21680 detects primarily the G-protein-coupled state with high affinity for agonists (Johansson et al., 1992). However, we observed similar numbers of A2ARs among the three genotypes in the striatum or globus pallidus with both [3H]SCH 58261 and [3H]CGS 21680. These results indicated that changes in A2AR expression are not the basis for differences in A2AR modulation of GABA release.
A second potential mechanism for reduced A2AR function is the uncoupling of A2ARs from their effector molecule adenylyl cyclase. Whereas CGS 21680 significantly increased cAMP production in striatal/pallidal slices from D2R+/+ mice, it was ineffective in D2R+/− and D2R−/− mice. Thus, despite unchanged agonist binding, the cAMP experiments suggested that A2ARs are functionally uncoupled in the D2R-deficient mice. Furthermore, the ability of CGS 21680 to increase GABA release likely depends on increases in cAMP (Wang and Johnson, 1995). The uncoupling does not appear to be caused by lower expression of the α subunits of stimulatory G-proteins (Golf or Gs) or the catalytic subunit of AC VI. Alternatively, the A2AR and G-protein could be kinetically uncoupled, the levels of G-protein and/or AC VI protein could be reduced, and/or the subcellular localization of the A2AR relative to the G-protein and/or AC could be altered in the D2R-deficient mice.
Altered effects of caffeine in D2R−/− mice
Dopamine receptors are involved in mediating caffeine-induced locomotion, and the differential results in the D2R+/+ and D2R−/− mice further support a role for D2Rs. Characteristic of the D1R/D2R synergy required for many dopamine-mediated behaviors, locomotor stimulation induced by caffeine requires activation of both receptor subtypes and is blocked by either selective D1R or D2R antagonists (Garrett and Holtzman, 1994). Moreover, Fenu and Morelli (1998) have demonstrated that caffeine produces motor stimulation in 6-hydroxydopamine-lesioned rats, which have had prolonged compromised D1R/D2R stimulation, only when the rats have been primed with a dopamine receptor agonist before caffeine testing. Therefore, differential behavioral effects of caffeine were expected in the D2R+/+ and D2R−/− mice. We observed an inability of the D2R−/− mice to sustain caffeine-induced activation whereas the D2R+/+ and D2R+/− mice were activated to similar extents.
Caffeine has similar affinities for A1Rs and A2ARs (Daly, 1993; Fredholm, 1995; Fredholm et al., 1999). However, it is A2AR antagonism that is required for locomotor activation (Ledent et al., 1997; Svenningsson et al., 1997a; Hauber et al., 1998). Our observation of uncoupled A2ARs in the D2R-deficient mice predicts that caffeine activation should have been precluded. This clearly was not the case with the D2R+/− mice and suggests that A2ARs in these mice must be at least partially functional. We did observe a trend for CGS 21680 to increase cAMP in the D2R+/−striatum/pallidum (Fig. 6). In any case, it is clear that only half the normal complement of D2Rs is sufficient to allow normal caffeine activation.
Although locomotor stimulation by caffeine was markedly reduced in the D2R−/− mice, it was not totally absent. It is possible that blockade of A1Rs could have also contributed. Indeed, from several rodent studies, combined blockade of A1Rs and A2ARs is known to produce larger locomotor responses than is blockade of only one receptor subtype (see Daly, 1993; Fredholm et al., 1999). Because the D2R−/− mice are functionally devoid of A2AR signaling, blockade of A1Rs may produce greater effects than in intact animals. A1Rs can negatively influence D1R signaling by blocking dopamine release in the striatum (Jin et al., 1993; Harvey and Lacey, 1997), by increasing the firing of dopaminergic neurons (Stoner et al., 1988), and by direct interactions at striatonigral neurons (Ferré et al., 1998). Thus, blockade of A1Rs, by antagonizing any or all of these mechanisms, can increase activity in striatonigral neurons (Ferré et al., 1996). Furthermore, activation of D1Rs leads to a larger locomotor response in D2R−/− mice than in D2R+/+ or D2R+/− mice (Kelly et al., 1998).
In rodents, c-fos expression in the globus pallidus is increased when higher doses of caffeine, which induce locomotor depression, are administered (Svenningsson et al., 1995; Svenningsson and Fredholm, 1997; Bennett and Semba, 1998). The induction of c-fos in the globus pallidus appears to require simultaneous inhibition of striatopallidal neurons and activation of striatonigral neurons (Le Moine et al., 1997). Indeed, inhibition of A2ARs, combined with activation of D1Rs, is very effective. Hence, our data showing pallidal c-fos induction after caffeine in the D2R−/− animals might be explained if in these animals striatonigral activity is increased by caffeine, even though basal activity is not elevated.
The present results show that a targeted disruption of the D2R influences responses to drugs that act on adenosine receptors. Thus, enhancement of GABA release induced by an A2AR agonist was completely eliminated when D2R expression was reduced. This likely involves some adaptive change(s), because acute treatment with a D2R antagonist does not produce the same result. The adaptation is not simply a change in A2AR expression or a loss of any key signaling component. However, the exact mechanism remains unclear. The motor stimulatory effect of caffeine, which depends on A2AR blockade, was substantially reduced in mice lacking D2Rs. These findings support the idea that some level of D2R activity is required for the action of caffeine, as has been postulated previously on the basis of results with antagonists. The results also show that animals with a targeted disruption of the D2R have a functional uncoupling of the A2AR. This underscores that not all of the phenotypic changes reported for such mice can be necessarily attributed to the D2R loss.
This work was supported by National Institutes of Health Grants NS 26851 and DA 12062 and by the Swedish Medical Research Council Project No. 2553. We thank Drs. Tom Dunwiddie and Per Svenningsson for their helpful discussions.
Correspondence should be addressed to Dr. Nancy R. Zahniser, Department of Pharmacology C-236, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Denver, CO 80262. E-mail:.