Effects of cannabinoid agonists and antagonists in male rats discriminating the synthetic cannabinoid AM2201

Abstract

The synthetic forms of delta-9-tetrahydrocannabinol (Δ⁹-THC), dronabinol or nabilone, have been approved to treat several indications. However, due to safety concerns their clinical utility remains limited. Consequently, there is a need for developing cannabinoid (CB) ligands that display behavioral pharmacological profiles better than Δ⁹-THC. Here, we utilized drug discrimination methods to compare the interoceptive effects of CB ligands that vary in potency, efficacy, and selectivity at the CB receptors, including two ligands, AM411 and AM4089, that show CB₁ partial agonist-like actions in vitro. Male rats were trained to discriminate 0.1 mg/kg AM2201 from saline under a fixed-ratio (FR) 10 response schedule of food reinforcement. After establishing AM2201’s discriminative-stimulus effects, pretreatment tests with the CB₁ antagonist/inverse agonist rimonabant blocked AM2201’s effects, whereas the peripherally-restricted antagonist AM6545 had no effect. Next, the generalization profiles of AM411 and AM4089 with CB₁ full (JWH-018, CP-55,940, AM8936), partial agonist (Δ⁹-THC) and non-cannabinoids (fentanyl, atropine) were compared. The CBs either fully (AM2201, CP-55,940, JWH-018, AM8936, Δ⁹-THC) or partially (AM411, AM4089) substituted for AM2201, whereas fentanyl and atropine did not produce AM2201-like effects. All CB drugs were more potent than Δ⁹-THC and correlation analysis confirmed the relative behavioral potencies of CBs corresponded strongly with their relative affinities at the CB₁ but not CB₂ receptors. Together, our results further demonstrate that AM411 and AM4089 exhibit a better pharmacological profile compared to Δ⁹-THC, in that, they are more potent and display in vivo partial agonist-like actions that are centrally mediated via CB₁ receptors.

1. Introduction

Increasing evidence suggests that the cannabinoid (CB) G-protein-coupled CB₁ and CB₂ receptors play an important role in a variety of biological functions (e.g., Lowe et al 2021). Consequently, considerable research has focused on developing drugs that target CB receptors to either modulate various physiological functions or as potential treatments for pathological conditions (e.g., Schurman et al. 2020; Finn et al. 2021; Lowe et al 2021). Currently, dronabinol or nabilone, the synthetic forms of delta-9-tetrahydrocannabinol (Δ⁹-THC) have been approved to treat nausea and vomiting following chemotherapy or as an appetite stimulant for AIDS-related anorexia (Coronado-Álvarez et al. 2021). Moreover, both Δ⁹-THC (combined with the noradrenergic agonist lofexidine) and nabilone have produced favorable results in cannabis withdrawal studies in humans (Haney et al., 2004), strongly supporting the potential therapeutic utility of partial CB₁ agonist-based medications for managing cannabis use disorders. Unfortunately, their clinical utility remains limited due to safety concerns related to the high lipophilic properties of Δ⁹-THC causing unpredictable pharmacokinetics, the production of active metabolites (Grotenhermen, 2003; Klumpers et al., 2012; Huestis 2007), and unwanted physiological/subjective effects that overlap with those of smoked cannabis (Hart et al., 2002; Sachs et al. 2015).

Additionally, Δ⁹-THC has a relatively low potency at CB receptors, thus higher concentration or repeated administration is needed to achieve a desirable therapeutic effect, which results in psychoactive and/or toxic effects (Coronado-Álvarez et al. 2021; Sachs et al. 2015). While in vitro studies have shown evidence of Δ⁹-THC’s partial CB₁ agonist activity (e.g., Dutta et al. 2022; Sim et al. 1996), such effects have not been consistently reported in in vivo laboratory animal studies. Unlike CB₁ full agonists, Δ⁹-THC only partially reduces body temperature in mice (McMahon and Koek 2007; Tai et al. 2015) and, in pretreatment studies, blocks the hypothermic effects of the full CB₁ agonist AM2389 in mice (Paronis et al. 2012). Studies have also suggested that a greater degree of pharmacological tolerance with Δ⁹-THC compared to the full CB₁ agonists may reflect the partial agonist-like effects of Δ⁹-THC (Singh et al. 2011, Hruba et al. 2012; Desai et al., 2013). However, in other studies Δ⁹-THC produces antinociceptive, cataleptic, and locomotor suppressant effects that are comparable to the bicyclic, cyclohexylphenol, aminoalkylindol, and naphthoylindoles full CB₁ agonists (e.g., Compton et al. 1992; Fan et al. 1994; Ginsburg et al. 2012; McMahon and Koek 2007; Wang et al. 2020). Moreover, in drug discrimination studies, full CB₁ agonists substitute for Δ⁹-THC’s discriminative-stimulus effects in rats or monkeys and in synthetic CB₁ agonists-trained subjects, Δ⁹-THC and other full CB₁ agonists generalize for their stimulus effects (e.g., Gatch and Forster 2014; Ginsburg et al. 2012; Hruba and McMahon 2017; Järbe et al. 2014; 2016a; 2011a; Wiley et al. 2012; 2014; Kangas et al. 2013; Rodriguez and McMahon 2014). Collectively, such observations have hindered our ability to identify CB₁-based pharmacotherapies with partial agonist-like activity.

We have previously synthesized and evaluated the structure activity relationship of a series of the 3-(1-adamantyl) substituted CB compounds that have a relatively similar chemical structure to Δ⁹-THC (Thakur et al., 2013). Unlike the full CB₁ agonist AM4054, two CBs, AM411 and AM4089, displayed promising in vitro (cyclic adenosine monophosphate assay cAMP) and in vivo (tetrad assays of antinociception and hypothermia) evidence of CB₁ partial agonists-like activity (Thakur et al. 2013). The present studies were conducted to further characterize the partial CB₁-agonist-like behavioral effects of AM411 and AM4089 in rats trained to discriminate the full synthetic CB₁ agonist AM2201. First, pretreatment studies with rimonabant (inverse agonist/antagonist) and AM6545 (peripherally-restricted) CB₁ antagonists were conducted to establish that AM2201’s discriminative-stimulus effects are centrally-mediated by CB₁ receptors. Next, to confirm that AM411 and AM4089 produce partial CB₁ agonist-like behavioral effects, we compared their substitution profiles with other standard CB (JWH-018, CP-55,940, AM8936, Δ⁹-THC) and non-cannabinoid (fentanyl and atropine) drugs. Finally, to further show that the discriminative-stimulus effects of CBs are mediated mainly through CB₁ receptors, correlation analysis was conducted to examine the correspondence between the behavioral effects of CBs in the present experiments and their affinities for the different CB receptors.

2. Materials and Methods:

2.1.1. Subjects.

Eight experimentally naive adult male Sprague-Dawley rats (Charles River Laboratories, Inc., Wilmington MA) weighing 275–350g were used. All rats were group-housed in a temperature and humidity-controlled room with a 12-h light/dark cycle (lights on 7:00 AM). Discrimination experiments were conducted five days/week during the light phase of the light/dark cycle between 9:00 AM and 5:00 PM. Rats were maintained at 85% of the free-feeding weight and were fed standard laboratory chows (~15 g) one hour after the behavioral session. Water was freely accessible except during experimental sessions. Rats were maintained per guidelines provided by the Guide for Care and Use of Laboratory Animals (National Academy of Sciences, 2011). Drug discrimination protocol (no. 20-0519R) was approved by the Institutional Animal Care and Use Committee at Northeastern University, Boston, MA, USA.

2.1.2. Apparatus.

Standard two-lever operant conditioning chambers that contain light and sound attenuating enclosures and supplied with continuous white noise and a fan to mask external noise were used (Med Associates, St. Albans, VT). The operant chamber was equipped with a feeder, and a food delivery tray centered on the front wall in between the two response levers set 17 cm apart in each chamber. A downward force of ~ 0.4 N was required to produce an audible click in the chamber and recorded as a response. Light-emitting diodes (LEDs) located directly above the response levers served as stimulus lights, and a house light located on the back panel of the chamber provided ambient illumination. During the behavioral sessions, experimental variables and data collection were controlled by computers with Med Associate interfacing equipment and operating software (Med Associates, St. Albans, VT).

2.1.3. AM2201 Discrimination.

The training procedure used to establish AM2201 discrimination has been described previously (Desai et al., 2003; Desai and Bergman 2010). Subjects were first trained to press both response levers under the fixed-ratio (FR) 10 response schedule of food reinforcement. Completing the FR-10 response requirement resulted in the delivery of one 45-mg pellet (BioServe, Frenchtown, NJ) into the food tray. Subsequently, subjects were trained to discriminate intraperitoneal (i.p.) injections of 0.1 mg/kg AM2201 from i.p. injections of saline. After AM2201 injection, responses only on one lever were reinforced; after saline injection, responses only on the other lever were reinforced. The assignment of AM2201-associated and saline-associated levers were counterbalanced across rats. Training sessions with AM2201 and saline were conducted five days per week and occurred in a double alternation sequence (i.e., saline, saline, 0.1 mg/kg AM2201, 0.1 mg/kg AM2201, saline, …etc.). Each experimental session was started by placing the rat inside the chamber immediately after the i.p. injection, all lights were extinguished, and responding had no scheduled consequences. After a 10-min timeout (TO), the session started and all lights in the experimental chamber were illuminated. Only completion of the FR-10 response requirement on the injection-associated (correct) lever was reinforced; responses on the other lever (incorrect) reset the lever response for food delivery. The presentation of each food pellet was followed by a 20-s TO period during which all lights were off, and lever responses had no scheduled consequences. Sessions ended after each rat obtained 20-reinforcements or after 15-min, whichever occurred first. Testing was initiated when the overall and the first FR-10 (1^st FR) injection-associated responding criteria of at least 80% during drug sessions and 20% or less during saline sessions were met over four successive training sessions. After subjects met this initial criterion, test sessions were only conducted over this course of these studies whenever rats met the above criteria for two consecutive training days in the sequence saline/0.1 mg/kg AM2201 or 0.1 mg/kg AM2201/saline. To determine the time course of action, the discriminative-stimulus effects of 0.1 mg/kg AM2201 were determined in single test session over multiple test days at 10, 30, 120, 360 min and 24 hours after drug administration. Following initial AM2201 dose-response and time course determinations, doses of each drug were tested in a random order in all subjects. Test sessions lasted 25-min (10-min timeout plus 15-min test session), during which the FR-10 schedule was in effect, and a single dose of each test drug was administered at the start of the 25-min cycle. The CB₁ antagonist’s interaction studies were conducted by administering 1.0 or 3.2 mg/kg of rimonabant or AM6545 20-min before injections of the training dose of 0.1 mg/kg AM2201. Doses of each antagonist were selected based on previous studies conducted to evaluate the effects of these CB₁ antagonists in rodents (e.g., Järbe et al. 2011b, Järbe et al. 2016a). All schedule parameters and contingencies during test sessions were identical to the training sessions, except that the 10 consecutive responses on either lever resulted in the delivery of a food reinforcer.

2.1.4. Drugs.

AM2201, JWH-018, AM8936, AM411, AM4089, rimonabant (SR141716A), and AM6545 were synthesized for these studies in the Center for Drug Discovery at Northeastern University, Boston, MA. Δ⁹-THC, CP-55,940, and fentanyl were provided by the National Institution of Drug Abuse drug supply program (Division of Therapeutics and Medical Consequences, NIDA). Atropine sulphate was purchased from Sigma-Aldrich (St. Louis, MO). AM2201, JWH-018, AM411, AM4089, and CP-55,940 were prepared in a 20:20:30:30 mixture of 95% ethanol, 100% dimethyl sulfoxide (DMSO), Alkamuls (EL-620), and saline. Δ⁹-THC and rimonabant were prepared in a 20:20:60 mixture of 95% ethanol, Tween-80, and saline. Fentanyl and atropine were dissolved in 0.9% saline. All drugs were administered via i.p. injections in volumes of 1.0 ml/kg or less. All tested drugs were given 30-minutes before the session except for AM2201, fentanyl and atropine which was given 10-minutes before the start of the session.

2.1.5. Data Analysis.

The percentage of AM2201-associated responses was calculated by dividing the number of responses on the drug lever by the total number of responses on both levers. The overall response rates were calculated for each session by dividing the total number of responses by the session duration without the timeout periods. Data for any rat that failed to emit at least 10 responses on either lever was excluded from the calculation of mean drug-appropriate responding at that dose. If fewer than three of the rats met the response rate requirement at a particular dose, no mean value was calculated for percentage of drug appropriate responding, but the subject’s data were included in the group mean for overall response rate analysis (Desai et al., 2003; Desai and Bergman 2010; Katz et al., 2004; Li et al., 2011; Wiley et al. 1998; 2014) Mean results for saline and each drug dose were calculated by averaging data from all subjects. Complete, partial, and no substitution for 0.1 mg/kg AM2201 was defined as ≥ 80%, 21%–79%, and ≤ 20% of total responses on the drug-associated lever, respectively. Repeated measures oneway ANOVA followed by Dunnett’s test was used to determine differences in overall response rates from saline values and the student’s paired t-test was used to determine the significant differences in responses in the antagonist studies. The functional in vivo half-life of AM2201 and its 95% confidence limits were determined using nonlinear regression (GraphPad Prism, version 9.1.3). The dose of drug needed to produce 50% AM2201-associated responding (ED₅₀ values and their 95% confidence limits) was determined using the linear portion of the dose-response curve (GraphPad Prism, version 9.1.3). The behavioral relative potency of CBs was determined using the ED₅₀ shift equation to evaluate the potency difference relative to AM2201 ED₅₀ (GraphPad Prism, version 9.1.3).

Correlation between the effects of drugs on behavior and competitive receptor binding (in vitro Ki values) studies were examined by comparing the calculated behavioral relative potency of each CB and their previously published relative in-vitro affinities for binding CB₁ and CB₂ receptors, i.e., Ki values of the tested drug divided by the Ki value for AM2201 alone (Table 1). Cannabinoid CB₁ receptor affinity values were obtained from previously published radioligand binding experiments in rat brains, whereas affinity values for cannabinoid CB₂ receptors were obtained from radioligand binding experiments in human CB₂ receptors due to the insufficient published data from rodents CB₂ receptors. Where appropriate, affinity values from multiple studies were averaged. Pearson’s product-moment correlation coefficient of these values was calculated using the correlation equation (GraphPad Prism, version 9.1.3) to determine the relationship between relative behavioral potency and relative binding affinity at cannabinoid CB₁ and CB₂ receptors.

Table.1:

ED₅₀ values (95% C.I) and relative behavioral potencies with which CBs produce increases in AM2201-associated responding in rats [AM2201 = 1] and their relative affinity at CB₁ and CB₂ receptors [AM2201 = 1].

Cannabinoid Ligand	%Drug-associated responding doses	%Drug-associated responding ED50 (95% C.I)	Relative Behavioral Potency (95% C.I)	rCB1 in-vitro affinity (Ki value)	Relative Affinity at CB1	hCB2 in-vitro affinity (Ki value)	Relative Affinity at CB2
	ms/kg	ms/kg	ED50 ratio	nM		nM
AM2201 (Start)	0.0032- 0.1	0.031 (0.017- 0.057)	1.0	1.0 a)	1.0	25 a)	1.0
AM2201 (End)	0.0032- 0.1	0.035 (0.02-0.06)	1.2 (0.7-1.8)	1.0 a)	1.0	25 a)	1.0
CP-55,940	0.0032- 0.32	0.042 (0.021- 0.084)	1.38 (0.68- 2.78)	0.97 b)	0.97	0.69h)	0.02
JWH-018	0.01- 0.56	0.11 (0.05- 0.24)	3.47 (1.76- 6.81)	4.5 c)	4.5	2.9 h)	0.12
AM8936	0.001- 0.1	0.012 (0.006- 0.025)	0.41 (0.2- 0.81)	0.55 d)	0.55	0.55 #	0.022
Δ9-THC	0.1- 5.6	0.95 (0.5- 1.8)	25.58 (12.46- 50.47)	39.5 e)	39.5	36.4 h)	1.45
AM411 *	0.032- 1.0	0.15 (0.04- 0.51)	10.52 (7.07- 13.98)	6.8 f)	6.8	52 f)	2.08
AM4089 *	0.0032- 0.1	0.016 (0.004- 0.06)	1.04 (0.74- 1.34)	2.1 g)	2.1	46.7 g)	1.7

^*Linear portion of the dose-response curves

^#Unpublished value

^a)(Järbe et al. 2016a)

^b)(Thomas et al. 1998), (Lambert and Fowler 2005)

^c)(Järbe et al. 2011a)

^d)(Papanastasiou et al. 2022)

^e)(Thakur et al. 2005)

^f)(Lu et al. 2005)

^g)(Thakur et al. 2013)

^h)(Han et al. 2013)

3. Results

3.1.1. AM2201 discrimination.

All the subjects learned to discriminate between i.p. injections of 0.1 mg/kg AM2201 from i.p. injections of saline within approximately 80 (SD=18.67) training sessions. Subjects started to acquire AM2201 discrimination at week five and control sessions after week seven show that injections of the training dose of 0.1 mg/kg AM2201 maintained an average of >80% responding on the drug-associated lever, whereas saline injections produced <20% responding on the drug-associated lever (Fig. 1, top panel). Administration of the training dose of AM2201 did not substantively alter response rates, with <10% difference in responding compared to saline values between sessions (Fig. 1, bottom panel). The mean group averages for rates of food-maintained responding during AM2201 and saline injections did not substantially differ during discrimination training sessions, i.e., 1.15 (± 0.14 S.E.M.) vs 1.27 (± 0.11 S.E.M.) responses per second after week 8, respectively (Fig. 1, bottom panel). Discriminative-stimulus control by the training dose of 0.1 mg/kg AM2201 was maintained throughout the study.

Figure 1.

Acquisition of discriminative-stimulus control in a group of male Sprague-Dawley rats trained to discriminate 0.1 mg/kg (i.p.) AM2201 from saline. Abscissae: Average of 5 training sessions from the first drug or saline injection. Ordinates: percentage of AM2201-associated responding (top) and response rates (bottom). Each data point represents the average (± S.E.M.) of effects in all subjects tested (at least six rats) at each dose. The horizontal dashed line at 80% and 20% drug lever responding (top) indicates the criterion for evidence of discriminative-stimulus control.

Injections of AM2201 (0.0032–0.1mg/kg) at the start and end of the study produced similar dose-dependent and full substitution for the training dose of 0.1 mg/kg AM2201, with a maximum of 86% (±11.9 S.E.M.) and 89% (±7.60 S.E.M.) responding on the drug associated-lever after administration of 0.1 mg/kg of AM2201 (Fig. 2, top left panel). Overall response rates were not significantly altered relative to control values following administration of any of the doses of AM2201 at the start and end of the study (Start: F_{(4, 35)} = 2.29, P > 0.05; End: F_{(4, 35)} = 1.97, P > 0.05; Fig. 2, bottom left panel). Table 1 shows that the calculated ED₅₀ the AM2201-associated lever responding at the start and end of the study did not substantially differ indicating no change in substitution potency of AM2201 over the course of the study (Start: 0.031 mg/kg; 95% C.I.: 0.017–0.057; End: 0.035 mg/kg; 95% C.I.: 0.02–0.06).

Figure 2.

Effects of the full cannabinoid agonist AM2201 determined at the start and end of the study in a group of male Sprague-Dawley rats trained to discriminate 0.1 mg/kg AM2201 from saline. Ordinates: percentage of AM2201-associated responding (top-right) and response rates (bottom-right) following i.p. injections of AM2201 or vehicle. The time course of action (top-left) and the response rates (bottom-left) of AM2201 training dose. Abscissae: S = saline; AM = AM2201 training dose in mg/kg (log scale). Data reported under S and AM are the average of 10 saline or 0.1 mg/kg AM2201 training session that were conducted before, during, and after testing of one of drugs in each panel. Each data point represents the average (± S.E.M.) in all subjects tested (at least six rats) at each dose or time point. The horizontal dashed line at 80% and 20% drug lever responding (top) indicates the criterion for evidence of discriminative-stimulus control.

The time course data shows that the discriminative-stimulus effects of the training dose of 0.1 mg/kg AM2201 was fully captured within 24-hrs with a maximum response of 89.04% on the AM2201-associated lever 10-min after treatment, ~45% from 30-min to 6-hrs, and <10% at 24-hrs post-treatment (F_{(4, 32)} = 5.089, P < 0.05; Fig. 2, top right panel). The estimated functional in vivo half-life of 0.1 mg/kg AM2201 was 111.4 mins (95% C.I.: 61.9 −205.8; Fig. 2, top right panel). The time course of the response rates produced by 0.1 mg/kg AM2201 did not significantly decrease or increase over the course of the 24-hr assessment (F_{(4, 30)} = 2.04, P > 0.05; Fig. 2, bottom right panel).

3.1.2. Pretreatment with CB antagonists.

Pretreatment with rimonabant (1.0 or 3.2 mg/kg) significantly blocked the discriminative-stimulus effects of 0.1 mg/kg AM2201 in a dose-dependent manner with a maximal reduction in AM2201-associated responding from ~90% to ~56 (t₍₁₄₎= 2.26; P < 0.05) and 33% (t₍₁₄₎= 3.78; P < 0.05), respectively (Fig. 3, top right panel). Both doses of rimonabant (1.0 or 3.2 mg/kg) did not significantly alter response rates (t₍₁₄₎= 0.57 and t₍₁₄₎= 0.701; P > 0.05, respectively; Fig. 3 bottom right panel). In contrast to rimonabant, pretreatment with AM6545 (1.0 or 3.2 mg/kg) did not block the discriminative-stimulus effects of training dose of 0.1 mg/kg AM2201 (t₍₁₂₎= 0.60 and t₍₁₂₎= 0.67; P > 0.05, respectively; Fig. 3, top left panel) or alter rates of food-maintained responding (t₍₁₂₎= 0.80 and t₍₁₂₎= 1.25, respectively; P > 0.05; Fig. 3, bottom left).

Figure 3.

Effects of pretreatment with CB₁ antagonists in a group of male Sprague-Dawley rats trained to discriminate 0.1 mg/kg AM2201 from saline. Ordinates: the percentage of AM2201-associated responding (top) and response rates (bottom) 30 mins following injections of rimonabant (left) or AM6545 (right) and the training dose of AM2201. Abscissae: Each bar graph represents the average (± S.E.M.) in all subjects tested (at least six rats) at each dose. The horizontal dashed line at 80% drug lever responding (top) indicates the criterion for evidence of discriminative-stimulus control. See Figure 2 caption for other details.

3.1.3. Effect of CB₁ full and partial agonists.

The CB₁ agonists AM2201 (0.0032–0.1 mg/kg; Fig. 2), CP-55,940 (0.0032–0.32 mg/kg), JWH-018 (0.01–0.56 mg/kg), and AM8936 (0.001–0.1mg/kg) (Fig. 4, top left panel) produced dose-dependent and full substitution for the training doses of 0.1 mg/kg AM2201, with maxima of approximately 86, 81, 84 and 86% AM2201-associated lever responding, respectively. There was no significant change in the overall response rates after administering JWH-018 (F_(6,41)= 1.44, P > 0.05), CP-55,940 (F_(6,41)= 1.22, P > 0.05) and AM8936 (F_(6,41)= 2.49, P > 0.05) compared to saline control values (Fig. 4, bottom left panel). The phytocannabinoid Δ⁹-THC (0.1–5.6 mg/kg) also fully substituted for the training dose of 0.1 mg/kg AM2201, with maxima of 97% responding on AM2201 associated-lever after administration of 5.6 mg/kg Δ⁹-THC (Fig. 4, top middle panel). Δ⁹-THC did not substantially alter overall rates of food-maintained responding (F_(6,48)= 1.76, P > 0.05; Fig. 4, bottom middle panel). In contrast, the partial CB₁ agonists AM411 (0.032–3.2 mg/kg) and AM4089 (0.0032–1.0 mg/kg) partially substituted for 0.1 mg/kg AM2201’s discriminative-stimulus effects as indicated by a plateau in the dose-response function and a maximal effect of approximately 55 and 50% on AM2201-associated lever after administration of 1.0–3.2 mg/kg AM411 and 0.1–1.0 mg/kg AM4089, respectively (Fig. 4, top middle panel). None of these partial agonists, AM411 or AM4089 showed a significant change in the overall response rates (F_(6,48)= 1.76, F_(5,40)= 1.62 and (F_(6,43)= 0.62, P > 0.05; Fig. 4, bottom middle panel). The rank order of potency for CB₁ ligands based on the calculated ED₅₀ values was: AM8936 > AM2201 ≈ AM4089 ≈ CP-55,940 > JWH-018 > AM411 > Δ⁹-THC (Table 1; Fig. 5).

Figure 4.

Effects of full CB agonists: CP-55,940, JWH-018 and AM8936 (left), partial CB agonists: Δ⁹-THC, AM411, and AM4089 (middle) and non-CB ligands: fentanyl and atropine (right) determined in a group of male Sprague-Dawley rats trained to discriminate 0.1 mg/kg AM2201 from saline. Ordinates: the percentage of AM2201-associated responding (top) and response rates (bottom) 30-min following injections of CB₁ agonists or 10-min following non-CB ligands. Abscissae: S = saline; AM = AM2201 training dose in mg/kg (log scale). Each data point represents the average (± S.E.M.) in all subjects tested (at least four rats) at each dose. The horizontal dashed line at 80% and 20% drug lever responding (top) indicates the criterion for evidence of discriminative-stimulus control. See Figure 2 caption for other details.

Figure 5.

The calculated ED₅₀ values of the cannabinoid ligands: AM2201, JWH-018, CP-55,940, AM8936, AM4089, AM411 and Δ⁹-THC and their 95% confidence limits (C.I). Ordinates: ED₅₀ values (mg/kg). Abscissae: Each bar graph represents the averaged ED₅₀ from each ligand with their upper and lower limits (95% C.I) calculated from all the tested subjects. Each point within the bar graph represents the ED₅₀ from a single subject (at least six rats). See Figure 2 caption for other details.

3.1.4. Effects of Non-cannabinoid Drugs.

The non-cannabinoid drugs fentanyl (0.0032–0.032 mg/kg) and atropine (0.032-1.0 mg/kg) did not generalize for the discriminative-stimulus effects of 0.1 mg/kg AM2201 at any of the tested doses (Fig. 4, top right panel). Doses of fentanyl tested in this study did not alter rates of food-maintained responding (F_{(4, 25)}= 0.57, P > 0.05), whereas atropine produced dose-dependent decreases in response rates (F_{(5, 33)}= 2.21, P < 0.05; Fig. 4, bottom right panel).

3.1.5. Pearson’s product-moment correlation coefficient.

The comparison between relative behavioral potencies of CB ligands that substituted for AM2201 with their previously published competitive receptor binding (in vitro Ki values) affinities at CB receptors are shown in Figure 6. Results show a strong association between the behavioral potencies and relative binding affinities (Ki values) at the CB₁ receptors in rat brain tissue (r = 0.96, P < 0.005; Fig. 6, right panel; Table 1). In contrast, no strong association between the behavioral potencies and relative binding affinities (Ki values) from human CB₂ receptors was found (r = 0.55, P > 0.05; Fig. 6, left panel; Table 1).

Figure 6.

Correlation between relative potencies of cannabinoid ligands (AM2201, AM8936, CP-55,940, JWH-018, Δ⁹-THC, AM411, and AM4089) in the present studies and their previously published relative affinities at CB₁ and CB₂ receptors in radioligand binding studies. Abscissae: affinity relative to AM2201 for inhibiting the binding of radioligand at rats CB₁ (left panels) and human CB₂ (right panels). Ordinates: behavioral potency relative to AM2201 based on ED₅₀ values for producing discriminative-stimulus effects (Fig. 4; Table. 1).

4. Discussion

The present studies were conducted to further characterize the discriminative-stimulus effects of various CB ligands in rats trained to discriminate the highly potent full CB₁ agonist AM2201. Results demonstrate that AM2201 serves as a robust discriminative-stimulus in male rats with no change in potency over the course of the study. The CB ligands (AM2201, CP-55,940, JWH-018, AM8936) and the phytocannabinoid Δ⁹-THC reliably and fully substituted for AM2201 in a dose-dependent manner without significantly altering the response rates. These findings are consistent with previous studies that have used other related CBs like JWH-018, AM5983, AM4054, and Δ⁹-THC as a discriminative-stimulus in rodents or nonhuman primates (Hruba and McMahon 2017; Järbe et al. 2016a; Järbe et al. 2016b; Järbe et al. 2012; Kangas et al. 2013; Wiley et al. 2014). Additionally, the time course data showed that intraperitoneal injections of the training dose of 0.1 mg/kg AM2201 displays a fast onset (>80 drug-associated responding at 10 min) and a short duration of in vivo action (<70 drug-associated responding after 60 min) with estimated functional in vivo half-life ~2.0 hrs. These findings are somewhat consistent with previous reports showing that AM2201 and other CBs like JWH-018 have a faster onset and shorter duration of action compared to Δ⁹-THC in Δ⁹-THC-trained rats (Gatch and Forster 2014; Järbe et al. 2011a; Järbe et al. 2016a) and produce effects on hypothermia, locomotor, and catalepsy that peak at ~2.0 hrs and lasts between 4.0-6.0 hrs after subcutaneous administration in rodents or monkeys (Carlier et al. 2018; Rodriguez and McMahon 2014). Moreover, the time course of AM2201’s effects are comparable with pharmacokinetic studies showing that administration of AM2201 produced dose-dependent effects with maximum plasma drug concentrations that last from 30 to120 mins in rats (Carlier et al. 2018).

Consistent with previous work in Δ⁹-THC or CB₁ agonists trained animals, pretreatment with the selective CB₁ inverse agonist/antagonists rimonabant blocked AM2201 discrimination (Ginsburg et al. 2012; Hruba and McMahon 2017; Järbe et al. 2016a; McMahon 2006; Wiley et al. 2014), whereas the peripherally restricted CB₁ antagonist AM6545 had no effect. These data confirm that AM2201’s discriminative-stimulus effects are centrally mediated via the CB₁ receptors. The lack of blockade by AM6545 is in line with the reported in vitro data showing this drug to function as a neutral CB₁ antagonist with limited brain permeability relative to the neutral CB₁ antagonist AM4113 and has rapid exertion from CNS via P-glycoprotein (Cluny et al. 2010; Tam et al. 2010). Likewise, other studies showed that unlike intracerebroventricular administration of CB₁ antagonists, systemic administration of AM251, AM4113, or AM6545 reduced feeding behavior in a dose-dependent manner, suggesting that the pharmacological effects of these drugs are probably mediated by actions at peripheral CB₁ receptors (Gómez et al. 2002; Randall et al. 2010; Sink et al. 2009). Also, intraperitoneal administration of AM6545 failed to generalize for rimonabants effects in rats further indicating that AM6545 is a peripherally restricted CB₁ antagonist (Järbe et al. 2011b, but see Kangas et al., 2020).

The full generalization produced by the CB₁ partial agonist Δ⁹-THC in AM2201-trained rats is consistent with previous studies in rats or non-human primates that have reported Δ⁹-THC to fully substitute for other higher efficacy CB₁ agonists, including JWH-018, AM5983, and AM4054 (Järbe et al. 2014; Järbe et al. 2012; Kangas et al. 2013; Wiley et al. 2014). Other in vivo studies have also generally reported Δ⁹-THC to produce hypothermic, antinociceptive, and cataleptic effects that are comparable to the full efficacy CB₁ agonists (Compton et al. 1992; Fan et al. 1994; Ginsburg et al. 2012; McMahon and Koek 2007; Wang et al. 2020). However, such observations are inconsistent with the in vitro characterization of Δ⁹-THC as a partial CB₁ agonist (e.g., Burkey et al. 1997; Dutta et al. 2022; Sim et al. 1996). It is noteworthy that except for some studies showing Δ⁹-THC only partially reduces body temperature (McMahon and Koek 2007; Tai et al. 2015) and antagonizing the hypothermic effects of the high efficacy full CB₁ agonist AM2389 in mice (Paronis et al. 2012), Δ⁹-THC’s CB₁ partial agonist-like actions in animal studies has not been reliably demonstrated. Analysis of pharmacological tolerance was used to characterize the in vivo effects of high vs. low efficacy CB₁ agonists by reporting a greater rightward shift on the dose-response curves after chronic administration of partial agonists (Desai et al. 2013; Paronis and Bergman 2011; Walker et al. 1995). Like these studies, Δ⁹-THC showed a greater tolerance compared to the higher efficacy CB₁ agonists in nonhuman primates (Hruba et al. 2012). The disconnect between the in vivo and in vitro actions of Δ⁹-THC remains unclear but might be related to the lower amount of receptor occupancy required for Δ⁹-THC to produce full CB₁-mediated discriminative-stimulus effects (Gifford et al. 1999). Also, the high lipophilic physicochemical properties of Δ⁹-THC likely lead to higher distribution and retention in the adipose tissue like the brain (e.g., Thomas et al. 1990), which might contribute to the observed full efficacy in the centrally mediated effects of Δ⁹-THC in drug discrimination studies.

The Δ⁸-THC analogue, AM411 partially substituted for AM2201. These data agree with previous reports suggesting that AM411 has CB₁ partial agonist-like effects on antinociception and hypothermia (Thakur et al. 2013). Also, a greater degree of tolerance was observed in monkeys after chronic AM411 administration compared to the CB₁ full agonists AM4054 and WIN 55212.2, suggesting that AM411 has a lower efficacy or partial effects at CB₁ receptors (Desai et al. 2013). However, the above is somewhat inconsistent with the in vitro efficacy profile of AM411 indicating that it is a full CB₁ agonist (Luk et al. 2004) and other studies showing that AM411 behaves as a full CB₁ agonist in rats discriminating Δ⁹-THC (Lu et al. 2005) or in open field behaviors (Järbe et al. 2004). Differences in AM411’s effects in AM2201- versus Δ⁹-THC-trained subjects could be related to the difference in effectiveness of the training dose of Δ⁹-THC and AM2201 to serve as a discriminative-stimulus (Järbe et al. 2014), i.e., partial agonists like AM411 fully substitute in rats trained to discriminate a lower dose of AM2201. Also, it might be due to AM411’s effects on G protein-coupled inwardly rectifying potassium channels (GIRK). For example, Luk et al. (2004) suggests the role of desensitization rates in the overall efficacy of full CB₁ agonists and demonstrates that AM411 has slower desensitization rates of GIRK channels like the partially effective agonist meth anandamide. Correspondingly, the 1-hydroxymethyl cannabinol analog AM4089 also partially substituted for AM2201’s discriminative-stimulus effects. This is consistent with its in vitro effects in the functional cAMP assay as well as hypothermia and antinociception studies in rodents (Thakur et al. 2013). To our knowledge, few studies have differentiated the in vivo behavioral pharmacology of CB₁ agonists like AM411 and AM4089 to produce partial generalization for a highly potent and full efficacy CB₁ agonist in discrimination studies. Evaluation of the ability of these drugs to antagonize the discriminative-stimulus effects of AM2201 or other full-efficacy CB₁ agonists are needed to further confirm the in vivo partial CB₁ agonist-like actions of these two drugs.

Previous work has suggested that the potency difference between CBs and the phytocannabinoid Δ⁹-THC might be a major contributing factor to the pronounced adverse effects of the psychoactive CBs. Here, the rank order of in vivo potencies of CB ligands versus Δ⁹-THC is consistent with the previously established in vitro CB₁ Ki values (Hoffman et al. 2017; Khajehali et al. 2015; Thakur et al. 2005; Vigolo et al. 2015) showing that all the CBs (AM8936, AM4089, AM2201, CP-55,940, JWH-018, AM411) were ~ 80, 60, 30, 23, 9, and 6-fold more potent than Δ⁹-THC, respectively (Table 1; Fig. 5). Also, in keeping with previous reports (Järbe et al. 2011a), it is notable that the 95% confidence limits of the calculated ED₅₀ value for Δ⁹-THC has a wider spread across individual subjects than the other CB₁ ligands reflecting individual variability across subjects (Fig. 6). Studies have generally reported that the discriminative-stimulus effects of Δ⁹-THC and CB agonists are primarily mediated through actions at the CB₁ receptors (e.g., Wiley et al. 2012); however, recent reports indicate that CB₂ receptors are also expressed in the CNS and therefore might contribute to the behavioral effects of CBs and Δ⁹-THC, including their antinociceptive, locomotor, cataleptic, and abuse-related effects (e.g., Jordan and Xi 2019; Kibret et al. 2022; Wang et al., 2020). Consequently, we examined the association between the behavioral potencies and relative binding affinities of CB ligands at both CB₁ and CB₂ receptors by correlating the previously published CB receptors binding affinities (Ki values) from radioligand binding assays and the relative behavioral potencies from the present AM2201 discrimination studies. The comparison suggests a strong association between the behavioral potencies and relative binding affinities at the CB₁ but not CB₂ receptors further confirming that the full and partial generalization by CB drugs for AM2201 discrimination is due to activation of CB₁ receptors.

5. Conclusion

Our results demonstrate that AM411 and AM4089 are more potent and display in vivo partial CB agonist-like actions that are centrally mediated via CB₁ receptors and as such can be reliably used as pharmacological tools to distinguish between full and partial CB₁ agonists.

Highlights.

• CB₁ agonists differentially substitute for AM2201’s discriminative stimulus effects

• Behavioral potency and affinities at CB₁ receptors strongly correlate

• Unlike Δ⁹-THC, AM411/AM4089 display in vivo partial CB₁ agonist-like actions

• AM411/AM4089 can be used as tools to distinguish full vs partial CB₁ agonists