Effects of levodopa on endocannabinoid levels in rat basal ganglia: implications for the treatment of levodopa‐induced dyskinesias

The majority of Parkinson's disease patients undergoing levodopa therapy develop disabling motor complications (dyskinesias) within 10 years of treatment. Stimulation of cannabinoid receptors, the pharmacological target of Δ9‐tetrahydrocannabinol, is emerging as a promising therapy to alleviate levodopa‐associated dyskinesias. However, the mechanisms underlying this beneficial action remain elusive, as do the effects exerted by levodopa therapy on the endocannabinoid system. Although levodopa is known to cause changes in CB1 receptor expression in animal models of Parkinson's disease, we have no information on whether this drug alters the brain concentrations of the endocannabinoids anandamide and 2‐arachidonylglycerol. To address this question, we used an isotope dilution assay to measure endocannabinoid levels in the caudate–putamen, globus pallidus and substantia nigra of intact and unilaterally 6‐OHDA‐lesioned rats undergoing acute or chronic treatment with levodopa (50 mg/kg). In intact animals, systemic administration of levodopa increased anandamide concentrations throughout the basal ganglia via activation of dopamine D1/D2 receptors. In 6‐OHDA‐lesioned rats, anandamide levels were significantly reduced in the caudate–putamen ipsilateral to the lesion; however, neither acute nor chronic levodopa treatment affected endocannabinoid levels in these animals. In lesioned rats, chronic levodopa produced increasingly severe oro‐lingual involuntary movements which were attenuated by the cannabinoid agonist R(+)‐WIN55,212‐2 (1 mg/kg). This effect was reversed by the CB1 receptor antagonist rimonabant (SR141716A). These results indicate that a deficiency in endocannabinoid transmission may contribute to levodopa‐induced dyskinesias and that these complications may be alleviated by activation of CB1 receptors.


Introduction
L-3,4-dihydroxyphenylalanine (levodopa) represents the most commonly prescribed treatment for Parkinson's disease (PD). Although levodopa alleviates PD symptoms in the early stages of the disease, its long-term use results in disabling side-effects consisting of abnormal involuntary movements (dyskinesias) and psychiatric complications Bezard et al., 2001). The causes of levodopainduced dyskinesias are poorly understood. Among the possible mechanisms, a decreased ability of nigrostriatal neurons to store dopamine, and hypersensitivity of postsynaptic dopamine receptors in the striatum, have been suggested as playing a role (Gerfen, 1992;Graybiel et al., 1994;Graybiel et al., 2000). None of these factors alone, however, seem to account for the motor complications associate with levodopa therapy, and secondary alterations in the metabolism (Brotchie, 2000;Hirsch et al., 2000) and neurochemistry of basal ganglia circuitry are receiving increasing attention (Brooks, 2000;Chase & Oh, 2000a;Chase & Oh, 2000b;Gerfen, 2000).
Recent studies point to the endocannabinoid system as an important modulator of dopaminergic activity in the basal ganglia. This system consists of a family of naturally occurring lipids, the endocannabi-noids, of which anandamide (Devane et al., 1992;Di Marzo et al., 1994) and 2-arachidonylglycerol (2-AG) (Mechoulam et al., 1995;Sugiura et al., 1995;Stella et al., 1997) are the best-characterized examples, and their allied CB 1 receptors (Howlett et al., 2002). CB 1 receptors are highly expressed in brain areas involved in the control of motor functions, such as the basal ganglia, cerebellum and sensorimotor cortex (Herkenham et al., 1990;Glass et al., 1997;Hermann et al., 2002).
In rodents, pharmacological and neurochemical evidence indicates that stimulation of dopamine D 2 receptors is accompanied by an increase in anandamide, but not 2-AG, levels in dorsal striatum (Giuffrida et al., 1999); such selective release may serve as an inhibitory feedback signal countering dopamine activation of motor behaviour (Giuffrida et al., 1999;Masserano et al., 1999). Abnormalities in dopamine signalling, as reported after levodopa administration in animal models of PD (Abercrombie et al., 1990;Gerfen et al., 1990), may compromise this inhibitory feedback, resulting in a functional state characterized by motor disturbances. In keeping with this hypothesis, chronic levodopa has been shown to increase CB 1 mRNA levels in the striatum of rats unilaterally lesioned with the neurotoxin 6-hydroxydopamine (6-OHDA) (Zeng et al., 1999). However, there is no information on whether levodopa may also affect endocannabinoid levels in the brain. Here, we show that levodopa selectively elevates anandamide in different areas of the basal ganglia of normal rats via activation of dopamine D 1 /D 2 receptors. This response is disrupted after 6-OHDA lesion of the nigro-striatal pathway. The implications of endocannabinoid disturbances on levodopa-induced oro-lingual involuntary movements are also discussed.

Animals
Animal care and experiments were conducted in accordance with the NIH Guide for care and use of laboratory animals and approved by the UCI animal use committee.
Male Wistar rats (200±250 g; Charles Rivers Laboratories, Wilmington, MA, USA) were acclimatized to the laboratory conditions (12-h light±dark cycle, 22 AE 1 8C) 1 week before experimental use. Food and water were available ad libitum. Unilaterally 6-OHDAlesioned and sham-operated rats were anaesthetized with equithensin (30 mL/kg, i.p.) and stereotaxically injected with 6-OHDA hydrochloride (or saline, respectively) in the left substantia nigra at the following coordinates: from bregma, AP, À1.5; ML, 1.8; DV, À7.5 mm (incisor bar set at 2.5 mm). At 3 weeks after 6-OHDA injections, lesioned rats were screened for apomorphine-induced (0.5 mg/kg, one injection, s.c.) contralateral rotation (O'Dell & Marshall, 1996) to assess the ef®cacy of the lesions. The number of full turns contralateral to the lesioned hemisphere were counted using a rotometer (Raturn System, BAS; West Lafayette, IN, USA). Net contralateral turns were calculated by subtracting the number of ipsilateral from controlateral rotations. Only rats displaying >300 rotations per 30 min were included in the study. These rotation scores were found to correspond to >90% depletion of tyrosine hydroxylase (TH) staining in the substantia nigra, using a mouse monoclonal anti-TH antibody (Sigma; data not shown). Either acute or chronic levodopa treatments were started 48 h after apomorphine administration.

Brain tissue preparation and HPLC/MS analyses
Animals were killed with halothane (Halocarbon laboratories; River Edge, NJ, USA) and their brains were rapidly collected 1 h after the last injection of drugs or vehicle, snap-frozen in cold 2-methylbutane (À50 8C), placed on an ice-cold stainless steel mould and cut into 1mm coronal slices using razor blades. Tissue punches were excised from the caudate±putamen (S), lateral globus pallidus (GP), substantia nigra (SN) and motor cortex (C). The punches were thawed in 1 mL of methanol containing 25 pmol of [ 2 H 4 ]-anandamide and [ 2 H 8 ]-2-AG, and homogenized. Endocannabinoids were extracted with metha-nol±chloroform (1 : 2, v/v) and quanti®ed using high performance liquid chromatography/mass spectrometry (HPLC/MS) using an isotope dilution assay . MS analyses were performed with an electro-spray ion source set in the positive ionization mode.

Behavioural studies
Oral involuntary movements (vacuous chewing movements and tongue protrusions) were induced in 6-OHDA-lesioned animals (n 11) by daily intraperitoneal (i.p.) injections of levodopa (50 mg/kg) plus carbidopa (12.5 mg/kg) for up to 11 days. One and two hours after administration of levodopa or vehicle, rats were individually placed in plexiglas cages and observed for 10 min by an investigator blind to the study. Oral dyskinesias were scored on a severity scale from 0 to 4 (modi®ed from Lundblad et al., 2002): 0, absent; 1, chewing movements present for less than half of the observation time; 2, chewing movements present for more than half of the observation time; 3, chewing movements present for more than half of the observation time and accompanied by tongue protrusions; 4, oral movements present all the time and accompanied by self-biting. To investigate the effects on levodopa-induced oral dyskinesias, the cannabinoid agonist WIN (1 mg/kg, i.p.), its inactive enantiomer S-(À)WIN55,212-3 mesylate (1 mg/kg, i.p.), or vehicle (5% Tween 80, 5% polyethylene glycol (PEG) in saline; 1 mL/kg, i.p.) were administered daily (one injection/ day, from day 6 to day 11) 20 min before levodopa treatment. The CB 1 antagonist rimonabant (0.3 mg/kg, i.p.) was coadministered with WIN on day 11 only. All behavioural analyses were carried out in independent groups of lesioned rats that were distinct from those used for the quantitative analyses of endocannabinoids.

Statistical analyses
Data were analysed by using one-way ANOVA followed by Student's t-test with Bonferroni's correction, or by using Student's t-test where appropriate. Behavioural data were analysed using the Kruskal±Wallis test followed by Dunn's multiple comparison test.
A single injection of levodopa (50 mg/kg, i.p.) caused a signi-®cant elevation of anandamide in S, GP and SN 1 h after drug administration ( Fig. 1C) (one-way ANOVA, followed by t-test with Bonferroni's correction, P < 0.001), whereas it had no effect on 2-AG levels (Fig. 1D).
To investigate the pharmacological mechanisms of levodopainduced anandamide elevation, we studied the effects of systemic administration of D 1 -like (SCH-23390, 0.3 mg/kg, i.p., n 8) or D 2like (raclopride, 2 mg/kg, i.p., n 8) receptor antagonists on this response. When applied alone, neither SCH-23390 nor raclopride affected anandamide brain concentrations ( Fig. 2A and B). However, SCH-23390 completely blocked the stimulatory effect of levodopa on anandamide throughout the basal ganglia ( Fig. 2A), whereas raclopride prevented the levodopa-induced anandamide increase in S and GP but not in SN (Fig. 2B). These results indicate that levodopa exerts its stimulatory action on anandamide levels via activation of both D 1and D 2 -like receptors in S and GP and of D 1 -like receptors in SN. None of these treatments affected 2-AG levels (data not shown). On the other hand, chronic levodopa administration in intact animals (50 mg/kg, one injection/day for 11 days) produced only a modest, although not signi®cant, increase in anandamide in the basal ganglia (data not shown).
Together, these results show that levodopa acting on dopamine receptors selectively changes the levels of anandamide, and raise the possibility that disturbances of dopamine transmission might affect endocannabinoid signalling.

Effects of levodopa on endocannabinoid levels in 6-OHDA-lesioned rats
Loss of dopamine innervation in the striatum has been shown to induce both quantitative and qualitative changes in the response of basal ganglia neurons to acute application of dopamine agonists (Waszczak et al., 1984;Huang & Walters, 1994). To test whether striatal deafferentation may affect endocannabinoid levels, we measured anandamide and 2-AG in the basal ganglia of unilaterally 6-OHDA-lesioned rats. Sham-operated rats of similar age were used as controls.
Quantitative analyses of endocannabinoids in the lesioned (left) hemisphere showed signi®cant lower levels of anandamide (one-way ANOVA, followed by t-test with Bonferroni's correction, P < 0.05) in S (n 8) (Fig. 3A); no changes were observed in anandamide or 2-AG levels in any of the other areas examined, including those in the right (intact) hemisphere ( Fig. 3 and data not shown). The decrease in striatal anandamide levels in lesioned rats was not accompanied by changes in the activity of FAAH, the enzyme responsible for intracellular anandamide hydrolysis (Schmid et al., 1985;Cravatt et al., 1996), which was identical to that observed in control animals (in pmol per min per mg of protein: sham-operated rats, 417 AE 69; lesioned rats, 403 AE 76, n 5).
Acute levodopa administration (50 mg/kg, i.p., n 8) caused a signi®cant elevation of anandamide in S and GP of sham-operated animals (P < 0.05, Student's t-test; Fig. 3A), an elevation which was similar to that observed in intact rats. No signi®cant changes, however, were observed in the left SN, most probably as consequence of the tissue damage caused by the insertion of the infusion cannula (Fig. 3A). By contrast, neither acute nor chronic levodopa elevated anandamide in 6-OHDA-lesioned rats ( Fig. 3 and data not shown). Taken together, these results show that dopamine depletion induced by 6-OHDA lesions prevents levodopa from exerting its stimulatory effect on anandamide levels.

Effects of apomorphine on anandamide production in 6-OHDA-lesioned rats
To evaluate whether the inability of levodopa to increase anandamide levels in lesioned animals was related to the extensive degeneration of dopaminergic nigro-striatal neurons (which may represent a possible source of anandamide) or to altered dopamine receptor stimulation, we used the nonselective dopamine agonist apomorphine. This drug does not share transport or metabolic pathways with levodopa and directly stimulates dopamine D 1 /D 2 receptors. As shown in Fig. 3B, apomorphine (0.5 mg/kg, s.c., n 8) produced a signi®cant elevation of anandamide in S (Student's t-test, P < 0.01) and GP (Student's t-test, P < 0.05) but not in SN, suggesting that anandamide elevation in S and GP is triggered by dopamine receptor stimulation independently from the integrity of the nigro-striatal pathway. Similar results were obtained in the intact hemisphere (n 8): S, controls 10.7 AE 1.1 pmol/g, apomorphine 18.7 AE 1.7 pmol/g (Student's t-test, P < 0.05); GP, controls 8.6 AE 2.6 pmol/g, apomorphine 22.7 AE 8.8 pmol/g (P < 0.05); SN, controls 37.6 AE 19.2 pmol/g, apomorphine 37.9 AE 13.3 pmol/g (P > 0.05).

Effects of CB 1 receptor stimulation on levodopa-induced oro-lingual movements
In 6-OHDA-lesioned rats, chronic administration of levodopa (50 mg/ kg, one injection/day for 11 days) elicited a time-dependent sensitization of vacuous chewing movements and tongue protrusions which reached a plateau between days 7 and 11 (Fig. 4). By contrast, no behavioural sensitization was observed in sham-operated or intact animals (data not shown). Application of the cannabinoid agonist WIN from day 6 to day 11 (1 mg/kg, i.p., one injection/day, 20 min before levodopa) signi®cantly reduced levodopa-induced involuntary oral movements (Fig. 4, arrow) (Kruskal±Wallis test followed by Dunn's multiple comparison test, P < 0.001). By contrast, administration of its inactive enantiomer S(À)-WIN55,212-3 (1 mg/kg i.p. 20 min before levodopa) had no antidyskinetic effect (Fig. 4). Finally, administration of the CB 1 antagonist rimonabant (0.2 mg/kg, i.p., 20 min before WIN) on the last day of levodopa WIN treatment completely reversed the antidyskinetic effects of WIN (Fig. 4, arrowhead).

Discussion
Understanding the motor complications associated with levodopa use largely depends on our knowledge of how this drug affects the interplay of activities among neurotransmitter systems in the basal ganglia.
In this study, we show that acute levodopa administration causes a signi®cant and selective elevation of anandamide in the basal ganglia of normal rats. This elevation is prevented by pharmacological blockade of either D 1 -or D 2 -like receptors in S and GP, and by D 1 antagonists only in SN.
The participation of both D 1 -and D 2 -like receptors in the regulation of endocannabinoid levels con®rms our previous microdialysis study showing that activation of D 2 receptors is accompanied by anandamide release in rat striatum (Giuffrida et al., 1999). In the same study we found that SKF38393, which acts as partial agonist on D 1 receptors, was not able to increase anandamide output. This apparent incongruity is probably attributable to the use of a low dose of SKF38393, which may have not been suf®cient to lead to an increase in extracellular anandamide (Giuffrida et al., 1999). The D 1 /D 2 receptor-mediated anandamide elevation is also consistent with: (i) the high degree of colocalization of CB 1 and D 1 /D 2 -like receptors in the basal ganglia (Aizman et al., 2000;Hohmann & Herkenham, 2000;Meschler & Howlett, 2001;Hermann et al., 2002); (ii) the existence of functional interactions between CB 1 receptor stimulation and D 1 (Anderson et al., 1995;Miyamoto et al., 1996), and D 2 (San Äudo-Pen Beltramo et al., 2000;Rodrõ Âguez de Fonseca et al., 1995Glass & Felder, 1997 and D 1 /D 2 (Meschler et al., 2000;Meschler & Howlett, 2001) receptor-mediated responses in rodents.
In contrast to the responses observed in S and GP, levodopa-induced anandamide elevation in SN was dependent on activation of D 1 but not D 2 receptors. The higher expression of D 1 vs. D 2 receptors in SN (Savasta et al., 1986;Gerfen et al., 1990) may account for the predominant role of D 1 receptors in this region. Nonetheless, these data indicate that the mechanisms responsible for endocannabinoid production may differ throughout distinct areas of the brain.
Although the physiological signi®cance of levodopa-induced anandamide elevation remains to be elucidated, we hypothesize that this increase represents a negative feedback to enhanced dopaminergic transmission following levodopa administration. In keeping with this hypothesis, anandamide has been shown to counteract dopamineinduced motor activation (Giuffrida et al., 1999;Romero et al., 1995). Furthermore, unilateral intrapallidal and intranigral injections of the cannabinoid agonist CP55940 produce ipsilateral and controlateral turning behaviour in rodents (San Äudo-Pen Äa & Walker, 1998; San ; these behaviours are attenuated by the D 2 agonist quinpirole, thus suggesting reciprocal interactions between dopamine and cannabinoid receptor systems in the basal ganglia . Within this area, endocannabinoid signalling may also modulate responses mediated by other neurotransmitter systems. As an example, dopamine participates in the control of synaptic excitation in S through a complex interplay with NMDA receptors (Carlsson & Carlsson, 1990;Starr, 1995;Calabresi et al., 2000), and by favouring neuronal communication via gap junctions (Moore & Grace, 2002;Onn & Grace, 1994). Enhanced anandamide levels may play a role in these processes, as indicated by the ability of CB 1 receptor stimulation to reduce glutamate release in dorsal striatum (Gerdeman & Lovinger, 2001;Gubellini et al., 2002), and by the observation that anandamide modulates neuron±glial interactions by regulating gap-junction permeability (Venance et al., 1995).
Unlike acute administration, chronic exposure to levodopa did not signi®cantly affect anandamide levels in intact rats, most probably as consequence of an agonist-induced desensitization of dopamine receptor function (Gardner et al., 2001;Kim et al., 2001).
Our data show that extensive degeneration of dopaminergic nigrostriatal neurons causes a signi®cant decrease in anandamide levels in the S ipsilateral to the lesion. This decrease is most probably due to a de®ciency in anandamide production, as no changes in the activity of FAAH (the enzyme responsible for anandamide inactivation; Cravatt et al., 1996) were observed in the same region. These results differ from previous observations showing enhanced anandamide levels in the S of 6-OHDA-lesioned rats as consequence of decreased FAAH activity (Gubellini et al., 2002). Differences in the experimental setting, such as extension of the lesion and/or analytical procedures used, may explain this inconsistency. Indeed, in their paper, Gubellini et al. (2002) report that the basal levels of striatal anandamide in intact animals are in the range of 250 pmol/mg of protein. This value is extremely high if compared with previous studies carried out in the same area (Felder et al., 1996;Yang et al., 1999;Lastres-Becker et al., 2001;Gonzalez et al., 2002), which ®nd basal levels of striatal anandamide ranging between 20 and 96 pmol/g. This range is consistent with our own data (18.9 AE 4.4 pmol/g) which corresponds, when converted from pmol/g of tissue into pmol/mg of protein, to 0.55 AE 0.85 pmol/mg of protein. Therefore, the high endocannabinoid levels measured by Gubellini et al. (2002) may be due to ischemic insult of brain tissue during postmortem manipulation of samples, which is known to cause endocannabinoid release (Hansen et al., 2001;Schmid et al., 1995;Panikashvili et al., 2001). On the other hand, the decreased endocannabinoid tone observed by us in the S of lesioned rats is consistent with the higher spontaneous glumatergic transmission reported in these animals (Gubellini et al., 2002), which can be linked to the ability of endocannabinoids to negatively modulate corticostriatal glutamatergic transmission (Gerdeman & Lovinger, 2001).
In the course of our investigations on 6-OHDA-lesioned rats, we never found changes in the concentrations of the second endogenous cannabinoid 2-AG. These observations differ from previous studies carried out in the reserpine-treated rats, another model of PD. In these animals, elevated 2-AG levels were found in GP, and administration of dopaminergic agonists decreased both anandamide and 2-AG concentrations in the same area (Di Marzo et al., 2000). Neurochemical and electrophysiological differences in the basal ganglia circuitry between 6-OHDA-lesioned and reserpinized rats may account for these divergences. Indeed, electrophysiological responses recorded in the basal ganglia of 6-OHDA-lesioned animals challenged with D 1 dopamine agonists are opposite to those observed in intact or reserpinized animals (Huang & Walters, 1994).
No endocannabinoid changes were found either in the intact or in the lesioned side of 6-OHDA-lesioned rats after levodopa administration. However, systemic administration of the mixed D 1 /D 2 agonist apomorphine in these rats caused a signi®cant increase in anandamide in the S and GP of both hemispheres, indicating a normal function of postsynaptic dopamine receptors. This observation excludes the possibility that an uncoupling of D 1 /D 2 synergism, as reported in 6-OHDA-lesioned rats (Hu et al., 1990;La Hoste et al., 1993), may prevent levodopa from exerting its stimulatory action on anandamide production. Although we cannot exclude a modest presynaptic release of striatal anandamide from the nigro-striatal afferents spared by the lesion, the signi®cant anandamide elevation following apomorphine administration indicates that dopamine-responsive elements located postsynaptically in S most probably represent the major source of this signalling lipid. In agreement with our observations, biochemical (Di Marzo et al., 1994) and electrophysiological (Gerdeman et al., 2002) data support the existence of a postsynaptic release of endocannabinoids in the striatum. However, like levodopa, apomorphine failed to elevate anadamide in the SN of both hemispheres of lesioned rats, suggesting that unknown compensatory mechanisms may contribute to the unresponsiveness of the intact SN to dopaminergic agonists.
Taken together, our data indicate that a dysregulation of anandamide signalling occurs in 6-OHDA-lesioned rats as a consequence of striatal deafferentation. Given the ability of CB 1 receptor stimulation to counterbalance dopamine-induced motor activity (Giuffrida et al., 1999;Masserano et al., 1999;Meschler et al., 2000), this dysregulation may cause a de®cient modulation of dopamine transmission which, in turn, may result in motor disturbances. Indeed, the reduced endocannabinoid tone in S may contribute to increased D 1 receptor-mediated signalling (Bezard et al., 2001;Meschler & Howlett, 2001) and enhanced glutamatergic transmission (Gerdeman et al., 2002;Gubellini et al., 2002). These two events may, following levodopa treatment, produce overactivation of striatal efferents to medial GP and SN pars reticulata and contribute to the generation of dyskinesias (Brotchie, 2003). In keeping with this possibility, we found that the oro-lingual involuntary movements elicited in lesioned rats by repeated administration of levodopa were reduced by systemic application of the cannabinoid agonist WIN. This effect was obtained with a dose of WIN (1 mg/kg) which did not affect apomorphine-induced circling behaviour or induce catalepsy (Anderson et al., 1995), thus indicating that the reduction in oral dyskinesias was not due to a generalized motor depression. The implication of CB 1 receptor activation in this response was con®rmed by the lack of effect of the WIN enantiomer S(À)-WIN55,212-2, which is less active on CB 1 receptors (D'Ambra et al., 1992), as well as by the ability of the CB 1 antagonist rimonabant (SR141716A) to reverse the action of WIN. The effects of WIN may be predictive of an antidyskinetic action in humans, because levodopa-induced abnormal movements in rats are suppressed by compounds which have antidyskinetic ef®cacy in parkinsonian patients and nonhuman primates (Lundblad et al., 2002). In addition, CB 1 receptor stimulation has been shown to reduce levodopadependent motor complications in MPTP-lesioned monkeys (Fox et al., 2002) and PD patients (Sieradzan et al., 2001).
Based on our results, long-term treatment with levodopa is not suf®cient to cause abnormal oro-lingual movements by itself. Indeed, chronic levodopa failed to sensitize oro-lingual movements in intact rats. Thus, the presence of a lesion in the nigro-striatal pathway represents a prerequisite for the generation of levodopa-dependent behavioural sensitization (Bordet et al., 1997;Bordet et al., 2000) and dyskinesias (Cenci et al., 1998), most probably as consequence of the cascade of neurochemical and functional changes following dopamine depletion (Gerfen et al., 2002). In this context, disturbances in endocannabinoid signalling can be seen as a new component of the neurotransmitter unbalances associated with striatal denervation, which collectively contribute to the pathogenesis of levodopaassociated dyskinesias (for review see Bezard et al., 2001).
In conclusion, our ®ndings indicate that endocannabinoid transmission is dysfunctional in 6-OHDA-lesioned rats, and suggest that enhancement of the endocannabinoid system may represent a promising adjunctive therapy to suppress or prevent levodopa-induced motor complications. . Time course of the effects of i.p. administration of the CB 1 agonist WIN (^, 1.0 mg/kg, 1 injection/day, n 11) and its inactive enantiomer S(À)-WIN55,212-3 (*, 1.0 mg/kg, one injection/day, n 11) on involuntary oral movements induced by chronic levodopa (50 mg/kg, one injection/day for 11 days) in 6-OHDA-lesioned rats. &, lesioned rats treated with levodopa only (n 11). Each enantiomer was given to distinct groups of rats from day 6 to day 10, 20 min before levodopa (arrow), whereas rats treated with levodopa only (&) received an injection of vehicle (5% PEG 5% Tween-80 in saline) starting from day 6. On day 11, the CB 1 antagonist SR141617A (SR, 0.3 mg/ kg, i.p.; 15 min before WIN) was given to rats treated with WIN (arrowhead), whereas all other groups received an injection of vehicle. Values represent median score. Ã P < 0.05, ÃÃ P < 0.01 compared to lesioned rats treated with levodopa only (Kruskal±Wallis test followed by Dunn's multiple comparison test).