Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Cannabinoids include a family of compounds derived from Cannabis sativa, the most biologically active of which is ⁹-tetrahydrocannabinol (⁹-THC) (Gaoni and Mechoulam, 1964). In addition, a number of compounds have been developed as specific receptor ligands, including agonists and antagonists (Compton et al., 1993; Rinaldi-Carmona et al., 1994). CB₁ receptors, and splice variant CB_1A (Shire et al., 1995), represent the principle cannabinoid receptor type so far found in rat brain (Matsuda et al., 1990), and mediate the central nervous system actions of cannabinoid compounds (Compton et al., 1993). Their actions are transduced via the activation of G-proteins (Howlett et al., 1986) that results in the inhibition (Howlett, 1984) or stimulation of adenylyl cyclase (Glass and Felder, 1997; Maneuf and Brotchie, 1997), inhibition of Ca²⁺ conductance (Mackie and Hille, 1992; Mackie et al., 1995), stimulation of K⁺ conductance (Mackie et al., 1995), and stimulation of the mitogen-activated protein kinase pathway (Bouaboula et al., 1995). Cannabinoid receptors couple to at least six different G-subunits in brain membranes (Prather et al., 2000).

Receptor activation of G-proteins can be measured by agonist-stimulated binding of the hydrolysis-resistant GTP analog [³⁵S]guanosine-5'-O-(3-thio)triphosphate (GTPS) to G-protein -subunits in membranes (Hilf et al., 1989; Selley et al., 1996) or brain sections (Sim et al., 1995). This technique is sensitive to differences in agonist efficacy and potency for G-protein activation (Lorenzen et al., 1996; Selley et al., 1997, 1998). Previous results with agonist-stimulated [³⁵S]GTPS binding have demonstrated that ⁹-THC is a weak partial agonist (Sim et al., 1996), and anandamide is an intermediate efficacy partial agonist (Burkey et al., 1997) compared with WIN55212-2 or CP55940 in rodent brain membranes. Other studies have reported partial agonist activity of ⁹-THC and anandamide for inhibition of adenylyl cyclase (Howlett et al., 1986; Childers et al., 1994) and CP55940 for inhibition of Ca²⁺ currents (Shen et al., 1996). Previous work from our laboratory not only confirmed these differences in agonist efficacy (Breivogel et al., 1998) but also demonstrated that cannabinoid receptor activity varies across different regions of rat brain because receptors in each region exhibit different catalytic amplification factors, defined as the number G-proteins activated per agonist-occupied receptor (Breivogel et al., 1997).

Cannabinoid compounds inhibit adenylyl cyclase activity in cell lines (Howlett, 1984; Slipetz et al., 1995) and in brain membranes (Bidaut-Russell et al., 1990; Pacheco et al., 1991; Childers et al., 1994). In general terms, the pharmacology of cannabinoid-inhibited adenylyl cyclase matches that of cannabinoid receptor binding (Pacheco et al., 1991)_, including competitive antagonism by SR141716A (Rinaldi-Carmona et al., 1994). However, inhibition of adenylyl cyclase in membranes from different regions of rat brain was only detectable in cerebellum and striatum (Pacheco et al., 1991; Childers et al., 1994).

The present study compares the efficacies and potencies of several commonly used cannabinoid compounds for the stimulation of [³⁵S]GTPS binding and displacement of [³H]SR141716A binding in rat brain membranes from three brain regions under identical assay conditions. Inhibition of adenylyl cyclase by several of these agonists also is determined in rat cerebellar membranes to compare the efficacies and potencies of agonists for G-protein activation with those for a downstream effector system. Although the efficacies of these compounds have previously been determined in rat cerebellar membranes, the relationship of agonist receptor occupancy to G-protein activation was not determined, and it is not known whether these efficacy differences are maintained in membranes from different regions of rat brain. To test the hypothesis that efficacy is related to receptor density, two regions were chosen that contain either similar (hippocampus) or different (hypothalamus) levels of cannabinoid receptors and cannabinoid-activated G-proteins compared with cerebellum. Furthermore, direct comparison of receptor binding and G-protein activation under identical conditions allows determination of the receptor states that are involved in agonist activity. Finally, determination of efficacy for adenylyl cyclase inhibition by these agonists will test the hypothesis that cannabinoid receptors in cerebellum exhibit greater receptor reserve at this effector than at G-proteins.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. Male Sprague-Dawley rats were purchased from Zivic Miller Laboratories, Inc. (Zelienople, PA). [³⁵S]GTPS (1250 Ci/mmol), [-³²P]ATP (800 Ci/mmol), and [³H]cAMP (25 Ci/mmol) were purchased from New England Nuclear Corp. (Boston, MA). [³H]SR141716A (53-55 Ci/mmol) was obtained from Amersham Life Sciences (Arlington Heights, IL). CP55940 and levonantradol were obtained from Pfizer, Inc. (Groton, CT). ⁹-THC was provided by National Institute on Drug Abuse/Research Triangle Institute (Research Triangle Park, NC). WIN55212-2, anandamide, and R-(+)-methanandamide were purchased from Research Biochemicals International (Natick, MA). SR141716A was a generous gift from Dr. Francis Barth at Sanofi Recherché (Montpellier, France). GDP and GTPS were purchased from Boehringer Mannheim (New York, NY). All other reagent grade chemicals and enzymes were obtained from Sigma Chemical Co. (St. Louis, MO) or Fisher Scientific (Pittsburgh, PA).

Agonist-Stimulated [³⁵S]GTPS Binding and [³H]SR141716A Competition Assays. Cerebellum, hippocampus, and hypothalamus were dissected from fresh rat brains on ice and pooled. Each region was homogenized with a Tissumizer (Tekmar, Cincinnati, OH) in cold membrane buffer (50 mM Tris-HCl, pH 7.4, 3 mM MgCl₂, 0.2 mM EGTA, 100 mM NaCl, pH 7.7) and centrifuged at 48,000g for 10 min at 4°C. Pellets were resuspended in membrane buffer, and then centrifuged again at 48,000g for 10 min at 4°C. Pellets from the second centrifugation were homogenized in membrane buffer and stored at 80°C until use. Frozen membranes were thawed and diluted in membrane buffer, homogenized, and preincubated for 10 min at 30°C in 0.004 U/ml adenosine deaminase (240 U/mg of protein; Sigma Chemical Co.) to remove endogenous adenosine, and then assayed for protein content before addition to assay tubes. Assays were conducted at 30°C for 2 h in membrane buffer, including 8 to 10 µg (cerebellum and hippocampus) or 10 to 20 µg (hypothalamus) of membrane protein with 0.1% (w/v) BSA, 50 µM GDP, 0.5 nM SR141716A (³H-labeled for competition assays), and 0.05 nM GTPS (³⁵S-labeled in stimulation assays) in a final volume of 1 ml. Both assays were performed simultaneously by incubating membranes with various concentrations of each ligand. Each rack also included determination of [³⁵S]GTPS binding with 3 µM levonantradol to be able to normalize the amount of stimulation by each agonist to that obtained with a maximally effective concentration of levonantradol. Nonspecific binding was determined in the absence of agonists and the presence of 30 µM unlabeled GTPS ([³⁵S]GTPS assays) or 10 µM unlabeled SR141716A ([³H]SR141716A assays). Reactions were terminated by rapid filtration under vacuum through Whatman GF/B glass fiber filters, followed by three washes with cold Tris buffer, pH 7.4. For [³H]SR141716A binding, filters were presoaked for approximately 2 h in Tris buffer containing 0.5% (w/v) BSA, and cold Tris rinse buffer contained 0.05% (w/v) BSA. Bound radioactivity was determined by liquid scintillation spectrophotometry at 95% efficiency for ³⁵S or 45% efficiency for ³H after overnight extraction of the filters in 4 ml of ScintiSafe Econo 1 scintillation fluid (Fisher Scientific). Typical [³⁵S]GTPS binding results included 500 to 700 dpm for nonspecific binding, 200 to 450 dpm [³⁵S]GTPS bound/mg of protein basal binding (depending on the region), and 850 to 1100 dpm/mg bound in the presence of maximally effective concentrations of levonantradol.

Agonist Inhibition of Adenylyl Cyclase. Assays were performed according to the method of Salomon (1979) with some modifications. Fresh cerebella were dissected on ice and homogenized in membrane buffer with a ground glass homogenizer. Membrane suspensions were centrifuged at 48,000g for 10 min at 4°C, and then pellets were resuspended and homogenized in membrane buffer. Membranes were assayed for protein content before addition to assay tubes. Membranes (~15 µg) were incubated for 10 min at 30°C in membrane buffer in the presence of various concentrations of each agonist with 50 µM cAMP, 50 µM GTP, and 50 µM ATP plus 1.5 µCi [-³²P]ATP with 10 mM theophylline, 5 mM phosphocreatine, 20 U/ml creatine phosphokinase (250 U/mg of protein; Sigma Chemical Co.), and 0.1% (w/v) BSA in a final volume of 0.1 ml. Reactions were terminated by boiling for 3 min and addition of stopping solution (2% sodium lauryl sulfate, 45 mM ATP, and 1.3 mM cAMP in Tris buffer, pH 7.5). [³H]cAMP standard (50 µl; ~15,000 cpm) and 1 ml of deionized water were added to each tube before addition of samples to Dowex columns and processing according to a previously published method (Salomon, 1979). Recovery of ³²P and ³H were determined by liquid scintillation spectrophotometry at 99% efficiency for ³²P and 45% efficiency for ³H in 3.5 ml of 0.1 M imidazole buffer, pH 7.3, and 18 ml of scintillation fluid. The solvents used to dissolve the cannabinoid compounds (dimethyl sulfoxide for WIN55212-2 and 95% ethanol for all others) had no effect on cAMP formation at the highest concentration of each vehicle present in the assay (0.1%).

Data Analysis. Net agonist-stimulated [³⁵S]GTPS binding values were calculated by subtracting basal binding values (obtained in the absence of agonist) from agonist-stimulated values (obtained in the presence of agonist) and were normalized to the values obtained for a maximally effective concentration of levonantradol (3 µM) measured in the same assay rack. The amount of [³²P]cAMP formed was determined by normalizing ³²P cpm to the fraction of total ³H cpm recovered from the columns. Data analyses, including agonist concentration-effect curves and displacement curves, were conducted by iterative nonlinear regression by using JMP for Macintosh (SAS, Cary, NC) or Prism for Windows (GraphPad Software, San Diego, CA) to obtain EC₅₀, E_max, IC₅₀, I_max, and n_H (Hill slope) values. Determination of which model best fit the data was made by an F test with Prism to compare two models simultaneously. [³⁵S]GTPS binding and [³H]SR141716A displacement data were found to fit better to two- or three-component models than to a one-component model with variable Hill slope. Percentage of total [³H]SR141716A binding data was fit to two- or three-component models with specific binding constrained to between 0 and 100%. K_i, and K_s values were estimated from IC₅₀ and EC₅₀ values, respectively, by the Cheng-Prusoff equation and antagonist K_B values were determined by the equation K_B = [Ant]/(CR  1), where [Ant] is the concentration of antagonist and CR is the ratio of the agonist EC₅₀ values in the presence and absence of antagonist (Pratt and Taylor, 1990). Differences among agonists and regions were determined by ANOVA followed by the Tukey-Kramer test for multiple comparisons at P < .05. Unless otherwise indicated, all data presented are mean ± S.E. of at least three experiments performed in duplicate.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Stimulation of [³⁵S]GTPS Binding and Inhibition of [³H]SR141716A Binding by Cannabinoid Agonists. Concentration-effect curves were generated for both the stimulation of [³⁵S]GTPS binding and competition for [³H]SR141716A binding by the cannabinoid agonists WIN55212-2, levonantradol, CP55940, ⁹-THC, and methanandamide in rat cerebellar (Fig. 1), hippocampal, and hypothalamic membranes. Both assays were performed under the same conditions, with 0.5 nM SR141716A in [³⁵S]GTPS binding assays and 0.05 nM GTPS in [³H]SR141716A binding assays. Both assays included 50 µM GDP and 100 mM NaCl, both of which favor agonist stimulation of [³⁵S]GTPS binding and promote the low-affinity state of cannabinoid receptors for agonist binding. Agonist-stimulated [³⁵S]GTPS binding in cerebellar membranes was blocked by the CB₁-selective antagonist SR141716A (data not shown), and the K_B value of SR141716A in shifting WIN55212-2 concentration-effect curves to the right was 0.24 ± 0.08 nM, similar to the previously reported K_D value for [³H]SR141716A obtained by Scatchard analysis in cerebellar membranes (0.19 ± 0.01 nM; Breivogel et al., 1997), and consistent with competitive antagonism of CB₁ receptors.

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Concentration-effect curves for cannabinoid stimulation of [35S]GTPS binding (A) and displacement of [3H]SR141716A binding (B) in rat cerebellar membranes. Assays were performed under identical conditions as described under Experimental Procedures. Net agonist-stimulated [35S]GTPS binding is shown as percentage of that obtained with 3 µM levonantradol. Curve fits are to a one-site model with variable Hill slope. Data are mean ± S.E. from four assays performed in duplicate.

View this table: [in this window] [in a new window]   TABLE 1 Binding parameters for cannabinoid-stimulated [35S]GTPS binding and [3H]SR141716A binding in brain membranes Membranes from the indicated regions were analyzed for maximal stimulation of [35S]GTPS binding by 3 µM levonantradol and for [3H]SR141716A binding. Data for [35S]GTPS binding are presented as net levonantradol-stimulated binding (pmol/mg), and percentage of stimulation of [35S]GTPS binding over basal by levonantradol. Data for [3H]SR141716A binding are presented as specific binding determined with 0.5 nM [3H]SR141716A, and the estimated Bmax of [3H]SR141716A binding assuming a KD of 0.24 nM for SR141716A. Data are mean values ± S.E. from four separate experiments.

View this table: [in this window] [in a new window]   TABLE 2 Relative agonist efficacies for the stimulation of [35S]GTPS binding and inhibition of adenylyl cyclase ([32P]cAMP) determined using a one-site model Agonist concentration-effect curves were generated for stimulation of [35S]GTPS binding and inhibition of adenylyl cyclase ([32P]cAMP) in brain membranes. Emax values for [35S]GTPS binding and Imax values for adenylyl cyclase inhibition were normalized to the values obtained by maximally effective concentrations of levonantradol (3 µM) or WIN55212-2 (1 µM), respectively, before curve fitting. Normalized values for adenylyl cyclase inhibition are shown in parentheses. Data are mean values ± S.E. from four separate experiments.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   Comparison of levonantradol-stimulated [35S]GTPS binding (A) and inhibition of [3H]SR141716A binding (B) in rat cerebellar, hippocampal, and hypothalamic membranes. Net agonist-stimulated [35S]GTPS binding is shown as percentage of maximum stimulation obtained in each region. Curve fits are to a one-site model with variable Hill slope. Data are mean ± S.E. from four assays performed in duplicate.

View this table: [in this window] [in a new window]   TABLE 3 Agonist potencies for inhibition of [3H]SR141716A binding, stimulation of [35S]GTPS binding, and inhibition of adenylyl cyclase ([32P]cAMP) determined using a one-site model Agonist concentration-effect curves were generated under identical assay conditions for competition for [3H]SR141716A binding and stimulation of [35S]GTPS binding. Three representative compounds also were assayed in cerebellar membranes for inhibition of adenylyl cyclase ([32P]cAMP). Curves were fit by iterative nonlinear regression to one-site models with variable Hill slopes to obtain IC50 and EC50 values. Ki and Ks values for binding were calculated as described under Experimental Procedures. Data shown are mean ± S.E. from four assays performed in duplicate.

Agonist Inhibition of Adenylyl Cyclase. Three agonists that exhibited a wide range of efficacies for stimulating the binding of [³⁵S]GTPS (WIN55212-2, methanandamide, and ⁹-THC) were used to assess the concentration-effect relationship for inhibition of adenylyl cyclase in rat cerebellar membranes. All three agonists produced significant (ANOVA, P < .005 for each agonist) concentration-dependent inhibition of adenylyl cyclase (Fig. 3). The effects of each agonist on adenylyl cyclase were completely blocked by 100 nM SR141716A, which had no effect on adenylyl cyclase by itself (data not shown). IC₅₀ values for the three cannabinoid agonists in inhibiting adenylyl cyclase ranged from 32 to 155 nM, and they were significantly more potent than the corresponding EC₅₀ values of each agonist in stimulating [³⁵S]GTPS binding, but were similar to their K_i values in displacing [³H]SR141716A binding (Table 3). The cannabinoid agonists also exhibited different efficacies (I_max values) for inhibiting adenylyl cyclase (Table 2), with WIN55212-2 and methanandamide producing similar levels of inhibition, and ⁹-THC producing approximately 50% of the maximal inhibition produced by WIN55212-2 (P < .05, Tukey-Kramer test). The I_max values of methanandamide and ⁹-THC for adenylyl cyclase inhibition normalized to WIN55212-2 (85 ± 5 and 53 ± 4%, respectively) were significantly higher (P < .05, Mann-Whitney rank sum test) than those for [³⁵S]GTPS binding normalized to WIN55212-2 (61 ± 4 and 19 ± 1%, respectively).

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Cannabinoid inhibition of adenylyl cyclase in rat cerebellar membranes. Assays were performed by incubating membranes with various concentrations of each agonist as described under Experimental Procedures. Data are percentage of inhibition of basal adenylyl cyclase activity presented as mean ± S.E. from four assays performed in duplicate.

Multicomponent Analysis of Agonist-Stimulated [³⁵S]GTPS Binding and Agonist-Inhibited [³H]SR141716A Binding Curves. Inspection of the agonist concentration-effect curves for the stimulation of [³⁵S]GTPS binding and inhibition of [³H]SR141716A binding revealed relatively shallow curves. In fact, all data fit better to a one-site model with variable slope (and all Hill slope values were less than one) than to a one-site model with Hill slope constrained to one. Moreover, comparison of the binding and stimulation curves in each region showed that each agonist appeared to occupy cannabinoid receptors at lower concentrations than those that stimulated [³⁵S]GTPS binding (Fig. 4). In each region, approximately 10 to 30% of [³H]SR141716A binding was inhibited before any stimulation of [³⁵S]GTPS binding occurred. For example, in hippocampus, 1 nM levonantradol displaced 25% of [³H]SR141716A binding, but it did not stimulate [³⁵S]GTPS binding (Fig. 4). In contrast, complete receptor occupancy by each agonist, indicated by 100% inhibition of [³H]SR141716A binding, occurred at nearly the same concentration of agonist as maximal stimulation of [³⁵S]GTPS binding (e.g., 3 µM levonantradol; Fig. 4).

View larger version (35K): [in this window] [in a new window]   Fig. 4.   Multicomponent analysis of cannabinoid stimulation of [35S]GTPS binding (A) and displacement of [3H]SR141716A binding (B) in hippocampal membranes. Data are mean ± S.E. from seven assays performed in duplicate. Net agonist-stimulated [35S]GTPS binding is shown as percentage of that obtained with 3 µM levonantradol. Curve fits are to two- and three-site models as appropriate.

View this table: [in this window] [in a new window]   TABLE 4 Multicomponent analysis of agonist displacement of [3H]SR141716A and stimulation of [35S]GTPS binding in hippocampal membranes Agonist concentration-effect curves were generated under identical assay conditions for displacement of [3H]SR141716A binding and stimulation of [35S]GTPS binding. Data were normalized to percentage of total specific (or agonist-stimulated) binding. Pooled data from seven separate experiments were fit to multicomponent equations by nonlinear regression, as described under Experimental Procedures. The "%" column refers to the percentage of total binding sites that are of high, intermediate, or low affinity as fit by nonlinear regression. IC50 and EC50 values are presented as mean log values ± S.E. of each fit, with corresponding Ki and Ks values in nM listed in parentheses. Ki/Ks ratios are shown for receptor binding and [35S]GTPS-stimulating sites that are hypothesized to correspond to one another.

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Comparison of levonantradol (A) and methanandamide (B) stimulation of [35S]GTPS binding and cannabinoid receptor occupancy. Data are mean ± S.E. from seven assays performed in duplicate. Net agonist-stimulated [35S]GTPS binding is shown as percentage of maximum obtained with each agonist. Curve fits are to two- and three-site models as appropriate.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

For these studies, five agonists representing a wide range of efficacies and potencies for G-protein activation in rat cerebellar membranes were chosen to determine whether these differences for G-protein activation persisted in other brain regions, and whether they also were manifested at the level of a downstream effector system, adenylyl cyclase. Hippocampus and hypothalamus were compared with cerebellum because hippocampus exhibited similar levels of cannabinoid receptors and cannabinoid-activated G-proteins, whereas hypothalamus exhibited lower levels of cannabinoid receptors and higher levels of cannabinoid-activated G-proteins (Breivogel et al., 1997). Methanandamide, a stable analog, was used instead of the endogenous cannabinoid anandamide because previous results demonstrated that these agonists acted essentially identically for both [³⁵S]GTPS binding (Breivogel et al., 1998) and receptor binding (Abadji et al., 1994) (provided that the membranes had been pretreated with an esterase inhibitor).

It is clear that the responses measured in these studies were mediated by CB₁ receptors as assessed by using the CB₁-selective antagonist SR141716A. Agonist-receptor binding was measured by inhibition of [³H]SR141716A binding, stimulation of [³⁵S]GTPS binding was inhibited by SR141716A with a potency consistent with mediation by CB₁, and agonist inhibition of adenylyl cyclase was blocked potently by SR141716A. Furthermore, the relative affinities for each agonist agreed with previously published relative affinities at CB₁ cannabinoid receptors (Rinaldi-Carmona et al., 1996).

Results of this study indicated that relative agonist potency is determined by agonist receptor affinity because agonist potencies for functional responses (activation of G-proteins and the inhibition of cAMP accumulation) correlated with agonist receptor-binding affinities. Although there were no apparent differences in agonist affinities across regions, most agonists (except ⁹-THC where efficacy was low) were most potent for [³⁵S]GTPS binding in hippocampus and least potent in cerebellum. This may be explained by regional differences in receptor reserve, with reserve being highest in hippocampus and lowest in cerebellum. However, agonist receptor affinities did not vary across regions, and the K_i/K_s ratios in hippocampus were all close to one, indicating little, if any, cannabinoid receptor reserve for [³⁵S]GTPS binding. This agrees with the observation that near complete receptor occupancy is required to obtain maximal stimulation of [³⁵S]GTPS binding (Fig. 4).

Results (Tables 1 and 2) showed significant differences in relative agonist efficacy (E_max) and potency across regions, but the reasons for these differences are not clear. The numbers of cannabinoid-activated G-proteins and cannabinoid receptors and the ratios between them (catalytic amplification factors) in these brain regions were previously reported by our laboratory. These factors were very similar for cerebellum and hippocampus, but hypothalamus exhibited higher amplification factors (Breivogel et al., 1997). Neither differences in receptor density nor amplification factor correlated with differences in relative agonist E_max or potency across these regions because relative agonist E_max values and potencies were most similar in hippocampus and hypothalamus. However, it is possible that the types of G-proteins activated by cannabinoid receptors in these regions vary. This is supported by the observation that cannabinoid-inhibited adenylyl cyclase is measurable in cerebellar but not in hippocampal or hypothalamic membranes, indicating that perhaps there is a higher ratio of adenylyl cyclase-inhibiting G-proteins (G_i or certain G complexes) to other types of G-proteins coupling to cannabinoid receptors in cerebellum (Pacheco et al., 1991; Childers et al., 1994). However, data obtained in another study by using a photoaffinity GTP analog, azidoanilido [³²P]GTP, found no evidence for a regional difference in activation of G-protein -subunits that would account for this variation (Prather et al., 2000). Alternatively, it may be that these agonists acting at CB_1A or at undiscovered cannabinoid receptor subtypes display different efficacies, and that the multiple receptor subtypes are present in different ratios across these three regions.

In cerebellar membranes, it is clear that the potencies of each agonist, and the efficacies of the partial agonists relative to WIN55212-2, are higher for inhibition of adenylyl cyclase than for stimulation of [³⁵S]GTPS binding. This is consistent with the existence of receptor/G-protein reserve for adenylyl cyclase inhibition, or greater receptor reserve for adenylyl cyclase inhibition than for G-protein activation. Receptor/effector reserve for the adenylyl cyclase would be predicted to produce greater apparent potency for the agonists because lower levels of receptor occupancy would be required to obtain the maximal effect. Because full agonists would activate excess G-proteins over what is necessary for maximal inhibition of adenylyl cyclase, partial agonists would have greater efficacy relative to a full agonist (e.g., WIN55212-2) for adenylyl cyclase inhibition than for G-protein activation.

This study clearly shows that cannabinoid agonists exhibit a wide range of efficacies for G-protein activation, translating into efficacy differences for at least one downstream effector system, adenylyl cyclase. Some of these differences were reported previously by other measurements. For example, both anandamide and CP55940 were partial agonists for inhibiting Ca²⁺ currents (Mackie et al., 1993; Shen et al., 1996). Chronic treatments with agonists of different efficacies have shown that more tolerance develops to agonists of greater efficacy (Elliott et al., 1997). Cultured N18TG2 cells exhibited greater desensitization of cannabinoid-inhibited cAMP after chronic desacetyllevonantradol than ⁹-THC treatment (Dill and Howlett, 1988), and mice treated chronically with CP55940 showed greater behavioral tolerance than those treated with ⁹-THC (Fan et al., 1994). Although it is difficult to demonstrate acute efficacy differences behaviorally (Fan et al., 1994), these chronic data suggest in vivo efficacy differences. Thus, it appears that these differences in efficacy at different effector systems are produced at the level of G-protein activation. Differences in efficacy may have profound implications for both drug abuse and for the use of cannabinoids as medicinal agents, particularly with long-term use.

[³⁵S]GTPS binding assays produced multicomponent concentration-effect curves for all agonists. Some of this heterogeneity may be due to the simultaneous presence of G-protein-coupled and -uncoupled cannabinoid receptors because both guanine nucleotides and sodium decrease high-affinity agonist binding to cannabinoid receptors by decreasing receptor/G-protein coupling (Devane et al., 1988; Pacheco et al., 1994). Alternatively, the two apparent potencies observed for the stimulation of [³⁵S]GTPS binding may have been due to coupling to different subtypes of G-protein -subunits, as previously suggested (Prather et al., 2000).

The identity of the three apparent receptor-binding sites is not clear. The highest affinity sites probably represent precoupled cannabinoid receptors because these values were similar to those previously determined under high-affinity agonist binding conditions (Rinaldi-Carmona et al., 1996). Previous data indicating that coupling to different G-subtypes occurred with different potencies (Prather et al., 2000) suggest that the multiple agonist receptor-binding affinities observed for the agonists were due to this differential coupling. Coupling to different G-proteins has previously been proposed as the source of three receptor-binding affinities observed for muscarinic receptors in cardiac membranes (Green et al., 1997). Another possible explanation for the three apparent receptor-binding sites is that high-affinity receptor binding occurs when receptors are coupled to G-proteins that are not binding either GDP or GTP(S), and the remaining states arise from receptor coupling to different subtypes of G-proteins that have guanine nucleotide bound.

These data do not directly identify the receptor affinity state(s) responsible for G-protein activation, but they do provide evidence via a strong correlation. In this study, it appeared that highest affinity agonist-binding sites contributed to basal, and not to agonist-stimulated [³⁵S]GTPS binding. For example, a previous study suggested that the GDP affinity on receptor precoupled G-proteins is approximately 30-fold lower than on uncoupled G-proteins, and is decreased only 8-fold further by a cannabinoid full agonist (Breivogel et al., 1998). Thus, receptor/G-protein coupling may promote some [³⁵S]GTPS binding even in the absence of agonist. A mathematical modeling study by Shea and Linderman (1997) supports the suggestion that precoupling of receptors and G-proteins may lead to complex binding and activation curves as were observed experimentally in the present study. Intermediate- and low-affinity receptor-binding sites appeared to correspond to the high- and low-affinity [³⁵S]GTPS-stimulating sites. This interpretation is somewhat complicated by findings with WIN55212-2 (Table 4), where the percentage of intermediate receptor sites (59%) more closely corresponded to the percentage of low-affinity [³⁵S]GTPS sites (68%). Nevertheless, for the other agonists, the percentage of low-affinity receptor sites corresponds well with the percentage of low-affinity [³⁵S]GTPS sites. The lack of correspondence between the percentages of intermediate receptor sites and high-affinity [³⁵S]GTPS sites can be explained by the existence of high-affinity receptor-binding sites that reduce the percentage of intermediate receptor sites. Previous findings that the ratio between cannabinoid receptor number and cannabinoid-activated G-proteins is not constant across brain regions (Breivogel et al., 1997) would predict that the ratio between these individual sites also might not be constant.

In contrast to the three-site model that fit receptor-binding data for WIN55212-2, levonantradol, and CP55940, methanandamide appeared to recognize only two-receptor-binding sites, indicating that this ligand binds to two of the sites recognized by the other ligands with equal or very similar affinity. Perhaps the inability of methanandamide to recognize a high-affinity receptor-binding site (higher than those stimulating [³⁵S]GTPS binding) is related to its lower efficacy for stimulating [³⁵S]GTPS binding. Regardless of the interpretation, these data show that cannabinoid receptors exhibit multiple affinities for at least some agonists, and that agonist occupancy of cannabinoid receptors often occurs at lower concentrations than are able to stimulate [³⁵S]GTPS binding to G-proteins.