Cannabinoid Receptor Localization in Brain: Relationship to Motor and Reward Systems

Marijuana (Omnabri satipa) is one of the oldest and most widely used drugs in the world, with a history of use dating back over 4,000 years.l.2 It was not until about twenty years ago that the principal psychoactive ingredient of the marijuana plant was isolated and found to be A9-tetrahydrocannabinol (A9-THC).3-5 A9-THC and other natural and synthetic cannabinoids produce characteristic behavioral and cognitive effect^,^.^ most of which can be attributed to actions on the central nervous system8 Marijuana use in this country is widespread, with approximately 40 million Americans having at least tried the drug9 Compulsive use is associated with social and psychological problems in individuals. lo There is little evidence of adverse health side effects or t~xicity.~Jl-lj Until recently, very little was known about the cellular mechanisms through which cannabinoids act. The unique spectrum of cannabinoid effects and the stereoselectivity (enantioselectivity) of action of cannabinoid isomers in behavioral studies (see below) strongly suggested the existence of a specific cannabinoid receptor in blain, but early attempts to identify and characterize such a recognition site were not successful (discussed in re&. 14-16). Without evidence that cannabinoids act through a specific receptor coupled to a functional effector system, researchers were prone to study the effects of cannabinoids on membrane properties, membrane-bound enzymes, eicosanoid production, metabolism, and other neurotransmitter systems in ~im.8J7-1~ As pointed out before,20 most of the biochemical studies employed concentrations of A9-THC that were in excess of physiologically meaninfl concentrations that mlght be found in brain (for review, see re&. 8, 18). In addition, the criterion of structure-activity relationship was not met-that is, the potencies of various cannabinoids in the in pim assays did not correlate with their relative potencies in eliciting characteristic behavioral effects.8.20 Particularly d a m a p g to the relevance of these in vim studies was the absence of enantioselectivity.20 However, several groups have reported enantioselectivity ofTHC isomers in various behavioral tests in vim. Martin's group bund that the potencies of (-) and (+) forms of each of the crj and tnzm isomers of A9-THC d S e r by 10to 100-bld in producing


INTRODUCI'ION
Marijuana (Omnabri satipa) is one of the oldest and most widely used drugs in the world, with a history of use dating back over 4,000 years.l.2 It was not until about twenty years ago that the principal psychoactive ingredient of the marijuana plant was isolated and found to be A9-tetrahydrocannabinol (A9-THC).3-5 A9-THC and other natural and synthetic cannabinoids produce characteristic behavioral and cognitive effect^,^.^ most of which can be attributed to actions on the central nervous system8 Marijuana use in this country is widespread, with approximately 40 million Americans having at least tried the drug9 Compulsive use is associated with social and psychological problems in individuals. lo There is little evidence of adverse health side effects or t~xicity.~Jl-lj Until recently, very little was known about the cellular mechanisms through which cannabinoids act. The unique spectrum of cannabinoid effects and the stereoselectivity (enantioselectivity) of action of cannabinoid isomers in behavioral studies (see below) strongly suggested the existence of a specific cannabinoid receptor in blain, but early attempts to identify and characterize such a recognition site were not successful (discussed in re&. 14-16).
Without evidence that cannabinoids act through a specific receptor coupled to a functional effector system, researchers were prone to study the effects of cannabinoids on membrane properties, membrane-bound enzymes, eicosanoid production, metabolism, and other neurotransmitter systems in ~im.8J7-1~ As pointed out before,20 most of the biochemical studies employed concentrations of A9-THC that were in excess of physiologically meaninfl concentrations that mlght be found in brain (for review, see re&. 8, 18). In addition, the criterion of structure-activity relationship was not met-that is, the potencies of various cannabinoids in the in pim assays did not correlate with their relative potencies in eliciting characteristic behavioral effects. 8.20 Particularly d a m a p g to the relevance of these in vim studies was the absence of enantioselectivity.20 However, several groups have reported enantioselectivity ofTHC isomers in various behavioral tests in vim. M a r t i n ' s group b u n d that the potencies of (-) and (+) forms of each of the crj and tnzm isomers of A9-THC d S e r by 10to 100-bld in producing static ataxia in dogs, depressing schedule-controlled responding in monkeys, and in producing hypothermia and inhibiting spontaneous activity in mice.21 Hollister ct ul. 22 showed cannabinoid enantioselectivity in human studies using indices of the subjective experience, or "high." May's group fbund enantiosclectivity of a series of s pthetic cannabinoids in tests of motor depression and a n a l g e~i a .~~-~~ One of May's compounds, (-)-9-nor-9~-hydroxyhexahydmcannabinol @-HHC), was used as a lead compound by Johnson and Melvinx fbr the synthesis of a rather

-
series of structurally novel, dassical and nondassical, cannabinoids fbr studies of their potential use as analgesics (FIG. 1). The synthetic cannabinoids share physicochemical properties with the natural cannabinoids and produce many behaviod and physiological effects characteristic of A9-THC, but are 5-1000 times more potent and show high enantioselectivity.  The availability of the nonclassical compounds revolutionized the study of the biochemical basis of cannabinoid activity. Howlett's group used them in neuroblastoma cell lines to show inhibition of adenylate c y k activity.27 Such inhibition is enantioselective, and the pharmacological profile correlates well with that observed by Martin's group in tests of mouse spontaneous activity, catalepsy, body temperature, and analgesia .28 One of the nonclassical compounds, CP 55,940, was tritiated and used by Howlett's group to identify and M y characterize a unique cannabinoid receptor in membranes from rat brain.l'The results from the centrifugation assay showed that [3H]CP 55,940 receptor binding is saturable, has high afEnity and enantioselectivity, and exhibits characteristics expected fbr a neuromodulator receptor associated with a guanine nucleotide regulatory (G) protein.
Recently, we characterized and validated the binding of [3H]CP 55,940 in slidemounted brain sections and described assay conditions to autoradiographically visualize the CNS distribution of cannabinoid receptors in a number of mammals, including humans.29 Autoradiography revealed a unique distribution that is similar in all mammalian species examined; binding is most dense in outflow nuclei of the basal gangha-the substantia nigra pars rcticulata and globus pallidus-and in the hippocampus and cerebellum.
The localization of dense receptors in the outflow nuclei of the basal gangha may account fbr some of the actions of cannabinoids. Dense binding localized in the globus pahdus, entopeduncular nudeus, and substantia nigra pars reticulata suggests an association of cannabinoid receptors with striatal efferent projections to these nuclei and, therefbre, a role fbr cannabinoids in motor control. In addition, binding may be localized on mesostriatal dopaminergic neurons, which would implicate a role fbr cannabinoids in direct control of dopamine release and, therefore, brain reward mechanisms. This report summarizes several key htures of our cannabinoid receptor localization studies: 1) validation that the in binding in brain sections is the same binding that mediates the e&cts of cannabinoids in pipg; 2) general htures of brain distribution in several species, including human; and 3) neuronal localization of cannabinoid receptors to motor and/or limbic components of the basal gangha, assessed by making selective chemical lesions of either the striatal GABAergic efferent or dopaminergic afferent pathways interconnecting the caudate-putamen (CPu) and the substantia q r a .

BkdingAswrys Cannabinoid kzptw Bind&
The procedures fbr obtaining cryostat-cut sections of fiesh, frozen brain mounted on "subbed" microscope slides were described previ~usly.'~,~ Assay conditions yielding 8690% specific binding are: incubation of slide-mounted sections at 37OC fbr 2 h in 50 mM Tris-HCl buflkr, pH 7.4, with 5% bovine serum albumin (BSA) and 1-10 nM [3H]CP 55,940 (sp. act. 76 Ci/mmole). Slides are washed at O°C fbr 4 h in the same buffer with 1% BSA and then dried. For use in competition studies to characterize and validate binding, natural and synthetic cannabinoid hgands were obtained from the National Institute ofDrug Abuse and Pfizer, Inc. Names and stereochemical configurations of some of the cannabmoids are shown in FIGURE 1.
Autoradiography was pertbrmed on 15-25 pm-thick brain sections of rat (Sprague-Dawley), guinea pig (Hartley), dog (beagle), rhesus monkey and human (dying of nonneurological disorders). Sections were incubated in 10 nM [3H]CP 55,940 using op timizcd conditions, then washed, dried, and exposed to tritium-sensitive film (LKB or Amersham) for 3 to 4 weeks before developing. Developed films were digitized with a solid-state video camera and Macintosh I1 computer-based system for densitometry. Receptor densities were quantified using IMAGEe s o h a r e (Wayne Rasband, Research Services Branch, NIMH).
Both the D1 and Dz receptor assays were carried out as previously described.30-32 For DI receptor binding, slides were warmed to mom temperature and incubated at 25OC for 2.5 h in 25 mM Tris-HC1 bufir, pH 7.5, with 100 mM NaCI, 1 mM MgCIz, 0.001% ascorbate, and 0.55 nM [3H]SCH 23390 (sp. act. 74.8 Ci/mmole). Slides were washed fbr 10 min in the same bufir at 4OC, dipped in deionized water, and dried. Nonspecific binding was determined by addition of 2 pM cis-flupenthixol and was typically <5% of total binding. Sections were exposed to film for 2 weeks.

Dopamine uptakc sin!
Assay conditions were described p r e v i~u s l y .~~~~ Slides were incubated at 2OC for 30 h in 50 mM sodium phosphate buffer, pH 7.5, with 120 mM NaCI, 0.01% BSA, 0.001% ascorbate, 500 nM zmns-flupenthixol, and 0.25 nM [JHIGBR 12935 (sp. act. 53.1 Ci/mmole). They were washed b r 2 h in same buffer at 2OC. Nonspecific binding, determined by addition of 20 pM mazindol, was typically <15% of total binding. Sections were exposed to film for 8 weeks.
Male rats (Sprague Dawley) were anesthetized and placed in a stereotaxic frame. A cannula was lowered into the caudate-putamen. Via an infusion pump and tubing, 1.5 pl(7.5 Icg) of ibotenate dissolved in normal saline was infused over 8 min. Animals survived for 2 or 4 weeks befbre sacrifice by decapitation.

6-OHDA Lesbu
Rats were prepared as above but were injected i.p. with desmethylimipramine (15 mg/kg) 30 min befbre infusion. The cannula was lowered into the medial fbrebrain bundle (mfb). Four pl(8 pg) of 6-OHDA dissolved in normal saline with 0.1% ascorbate added was infused over 8 min. Animals survived fbr 4 weeks befbre sacrifice; at 2 weeks post-lesion they were tested fbr rotational behavior 40 min after adminismtion of 5 mgkg of &amphetamine sulfate (Sigma). Only those rats showing greater than 10 rotations per min during a 5-min test were used in the binding experiment.

RESULTS
A large series of cannabinoid and non-cannabinoid drugs was assayed to test fbr competitive displacement of [3H]CP 55,940 (TABLE 1). The competition curves and derived inhibition constants (Ki's) fbr the natural and synthetic cannabinoids provided a test for validation of binding. We found that highly significant (p < 0.OOOl) cornlations exist between the Ki's and potencies of the drugs in tests of dog ataxia and human subjective experience, the two most reliable markers of cannabinoid acdvity.6,' The K:s also correlate very closely with relative potencies in tests of motor function (ataxia, hypokinesia, catalepsy), analgesia, and inhibition of contractions of guinea pig ileum and adenylate cyclase in neuroblastoma cell lines in piho.29 Enantioselectivity is striking; the (-) and (+) forms of CP 55,244 differ by more than 10,000-fold in vim, a separation predicted by the rigid structure of the molecule  (TABLE 1).
Autoradiography showed that in all species very dense binding is fbund in the globus pallidus, substantia n i p pars reticulata (SNR.), and the molecular layers of the cerebellum and hippocampal dentate gyrus (FIGS. 2 and 3). Dense binding is also found in the cerebral cortex, other parts of the hippocampal formation, and striatum. In rat, rhesus monkey and human, the SNR contains the highest level of binding (FIG. 3). In dog, the cerebellar molecular layer is most dense (FIG. 2H). In guinea pig and dog, the hippocampal fbrmation has selectively dense binding (FIG. 2E, F)  The single injection of ibotenate into the caudate-putamen resulted in a small central site of nonspecific destruction marked by gliosis and a much l a q p (appmximately 3 x 4 mm) surrounding area of selective neuronal degeneration, in accordance with previous descriptions of toxicity in the dose mge and location used.% In the &cted striatal territory, the losses on the lesion relative to conrsponding territory on the control side were the most profound fbr the dopamine D 1 receptors, showing a 96% reduction in binding at 4 weeks WLE 2). Cannabinoid and D 2 receptors were reduced FIGURE 2. Autoradiography of 10 nM [JHICP 55,940 binding in brain. Tritium-sensitive f i l m exposed for 4 weeks, developed and computer digtized. Images were photographed directly from the computer monitor. Gray levels represent relative levels of meptor densities. Sapttal section of rat brain in A. Comnal brain sections of human in B, D, and G; rhesus monkey in C and I; dog in P and H; and rat in J. Horizontal section of guinea pig brain in E. Inatr  specific binding in adjacent sections (miniaturized images are shown). A&-FIGURE 3. Rclativc densities of cannabinoid reccpton a c m brain structures in rat, rhesus monkey, and human. Autoradiographic images were dqytized by a solid-state camera and Macintosh I1 computer-based system for quantitaavc densitomcay using I m q S s o h (Wayne Rasband, Rcsmrh Services Branch, NIMH). ?iansmittance lcvels were converted to holcs/mg tissue using tritium standards, then normalized to the densest s t~~c t u~ in each animal (SNr h r all three). For every section incubated for total binding, an adjacent section was incubated in the presence ofCP 55,244 to permit subtraction of nonspeafic binding on a regional basis. Structure abbreviations not given in by 78-8096 in the a&cted temtory, and the dopaminc uptake site, which mides on afferent dopaminergic axons, was slightly reduced at 4 weeks (not shown).
Both cannabinoid and D1 dopamine receptors am lost in similar patterns (FIG. 4) and amounts (GUILE 2) in projection zones of lesioncd striad eflkrent neurons. In agreement with known medial-lateral topography of striad projections,39 receptor losses in both GP and SNR were p t e s t medially, with sparing of binding in lateral parts receiving projections from unltsioned parts of the lateral and posterior caudateputamen (FIG. 4). For cannabinoid receptor binding, losses in the labeled striato@ pathway were also evident (not shown).

Mfb 6-OHDA Lcrionr
Nissl-stained sections of the substantia nigra showed unilateral loss of neumns in the SNC (FIG. 4e). Autoradiography showed no change in cannabinoid receptor binding in either the striatum (FIG. 4d) or the nigra (FIG. 4e), whereas dopamine u p take sites were lost throughout the striatum and n i p on the lesioned side (FIG. 4f).
Quantitative densitometry showed major losses of dopamine uptake sites in the caudate-putamen, accumbens nucleus (ACb), and substantia n i p pars compacta (SNC), but no loss of cannabinoid receptors (TABLE 3).

DISCUSSION
The section binding assay is easy to perfbrm, is reliable, and shows high sensitivity to manipulations of binding conditions, such as the addition of guanine nucleo-  tidesz9 BSA appears to act as a carrier to keep cannabinoids in solution without a p preciably aflkcting binding kinetics. The low nonspecific binding and absence of binding in white matter indicates that the autoradiographic patterns are not affected by &and lipophilia. The inclusion of BSA in the incubation medium may actually mimic the disposition of cannabinoids administered in piw, as they would quickly complex with serum albumin or other carriers in the blood.
The structure-activity profile suggests that the receptor defined by the binding of [3H]CP 55,940 is the same receptor that mediates many of the behavioral and pharmacological effects of cannabinoids (TABLE l), including the subjective experience termed the human "high." All other tested psychoactive drugs, neurotransmitters, steroids, and eicosanoids at 10 pM concentrations Med to bind to this receptor. There was no compelling evidence fbr receptor subtypes from that analysis.
Autoradiography of cannabinoid receptors rev& a heterogeneous distribution pattern that conforms to cytoarchitectural and hnctional subdivisions in the brain. The distribution is unique-no other pattern of receptors is similar-and it is similar across several mammalian species, including human, suggesting that cannabinoid receptors are phylogenetically stable and conserved in evolution. The distribution appears to be similar to the distribution of the mRNA probe hybridized to a rat brain cannabinoid receptor gene.40 The locations of cannabinoid receptors help to understand cannabinoid pharmacology. High densities in the hippocampus and cerebral neocortex implicate roles h r cannabinoids in cognitive functions. High densities in axons and terminals of the GABAergic striatal neurons of the basal p g h a and of glutamatergic granule cells of the cerebellum suggest a modulatory role in movement systems. Sparse densities in lower brainstem areas controlling cardiovascular and respiratory functions may explain why high doses of A9-THC are not lethal.
The results of the 6-OHDA lesions indicate that cannabinoid receptors do not reside on mesencephalic dopamine neurons projecting to either the caudate-putamen or the ACb. Systemically administered A9-THC has been shown to elevate extracel-Mar levels of dopamine in the caudate-putamen41 and ACb.42 The mechanism of action appears to be indirect, as the effects are attenuated by n a l~x o n e .~~ Nevertheless, it has been proposed that drugs which elevate dopamine levels in the striaturn are those that are known to have abuse liability in humans.43~~ In humans, cannabinoids can produce a fkeling of euphoria as part of the subjective experience known as the marijuana "high," but dysphoria, dizziness, thought disturbances, and sleepiness are also r e p~r t e d .~.~.~~ Animals generally will not self-administer A9-THC.4s*M Cannabinoids did not lower the threshold fbr electrical self-stimulation in one study." In another study they did,a but apparently both this phenomenon and the enhancement of basal dopamine efflux from the ACb by A9-THC are strain-specific, occuning only in Lewis rats.49 Thus, the effects of cannabinoids on dopamine circuits thought to be common mediators of reward are indirect and different from those of drugs such as cocaine and morphine which directly affect extracellular dopamine levels and produce craving and powem drugseeking behavior. Accounts of cannabis use in humans stlrss the loosening of associations, fiagmentation of thought, and confusion on attempting to remember recent accurrences.7.50 These cognitive effects may be mediated by receptors in the cerebral cortex, especially the receptordense hippocampal cortex. The hippocampus "gates" infbrmation during memory consolidation and codes spatial and temporal relations among stimuli and res p o n s e~.~~.~~ A9-THC causes memory "intrusions,"53 impairs temporal aspects of performance," and suppresses hippocampal electrical activity.5s The localization of cannabinoid receptors in motor areas suggests therapeutic a p plications. Cannabinoids exacerbate hypokinesia in Parkinson's disease but are beneficial for some forms of dystonia, txtmor, and spasticity.6.7~~~ The association of cannabinoid receptors with GABAergic striatal projection neurons suggests roles fbr cannabinoids in control of movement, perhaps therapeutic roles in hyperkinesis and dystonia. Cannabinoids have been shown to be beneficial h r some hrms of dystonia, mmor, and spasticity.6J~~S9 Lack of association of cannabinoid receptors with dopamine neurons indicates that cannabmoids do not directly a&ct dopamine release associated with rrward and *-seeking behavior. Further work may show the basis for the reported usefulness in conmlling nausea and stimulating appetite in patients receiving chemothenpy h r cancer or AIDS. Finally, the development of an antagonist could lead to additional therapeutic applications. The section binding assay can be used to screen the potencies of novel drugs and serve to idenrify cannabinoid receptor subtypes, which could lead to renewed intemt in developing cannabinoid drugs without unwanted side &cts.