Dissecting the role of CB₁ and CB₂ receptors in cannabinoid reward versus aversion using transgenic CB₁- and CB₂-knockout mice

Abstract

Cannabinoids produce both rewarding and aversive effects in humans and experimental animals. However, the mechanisms underlying these conflicting findings are unclear. Here we examined the potential involvement of CB₁ and CB₂ receptors in cannabinoid action using transgenic CB₁-knockout (CB₁-KO) and CB₂-knockout (CB₂-KO) mice. We found that Δ⁹-tetrahydrocannabinol (Δ⁹-THC) induced conditioned place preference at a low dose (1 mg/kg) in WT mice that was attenuated by deletion of the CB₁ receptor. At 5 mg/kg, no subjective effects of Δ⁹-THC were detected in WT mice, but CB₁-KO mice exhibited a trend towards place aversion and CB₂-KO mice developed significant place preferences. This data suggests that activation of the CB₁ receptor is rewarding, while CB₂R activation is aversive. We then examined the nucleus accumbens (NAc) dopamine (DA) response to Δ⁹-THC using in vivo microdialysis. Unexpectedly, Δ⁹-THC produced a dose-dependent decrease in extracellular DA in WT mice, that was potentiated in CB₁-KO mice. However, in CB₂-KO mice Δ⁹-THC produced a dose-dependent increase in extracellular DA, suggesting that activation of the CB₂R inhibits DA release in the NAc. In contrast, Δ⁹-THC, when administered systemically or locally into the NAc, failed to alter extracellular DA in rats. Lastly, we examined the locomotor response to Δ⁹-THC. Both CB₁ and CB₂ receptor mechanisms were shown to underlie Δ⁹-THC-induced hypolocomotion. These findings indicate that Δ⁹-THC’s variable subjective effects reflect differential activation of cannabinoid receptors. Specifically, the opposing actions of CB₁ and CB₂ receptors regulate cannabis reward and aversion, with CB₂-mediated effects predominant in mice.

1. INTRODUCTION

The marijuana plant (cannabis sativa) has been cultivated and utilized for recreational, religious, and medicinal purposes for over 4,500 years (Chopra, 1969; Kuddus et al., 2013). The changing political and social landscape in the United States has opened doors to the budding cannabis industry leading to a diverse array of marijuana products with the potential for greater use and abuse. In 2018, an average of 8,400 Americans age 12 or older initiated marijuana use each day (SAMHSA, 2019). Some users describe marijuana’s subjective effects as positive, reporting elation, relaxation, creativity, and friendliness (Ewusi Boisvert et al., 2020; Matheson et al., 2020; Sexton et al., 2019), while others recount a negative experience including anxiety, depression, nausea, and irritability (Ewusi Boisvert et al., 2020; Schlienz et al., 2020; Spindle et al., 2018). Interestingly, a single marijuana user may experience a positive affective state in one smoking session and a negative state in another (Pang et al., 2017). The neurobiological underpinnings that differentiate a positive versus a negative experience with cannabis are not clear.

Pre-clinical data have also demonstrated variation in marijuana’s hedonic profile. Δ⁹-tetrahydrocannabinol (Δ⁹-THC), the main psychoactive component of cannabis, typically produces a place aversion to a location paired with drug delivery, indicative of a negative or aversive subjective effect (Cheer et al., 2000; DeVuono et al., 2017). However, Δ⁹-THC-induced place preferences have also been observed in this model at similar doses (Braida et al., 2004; Lepore et al., 1995). In intracranial self-stimulation (ICSS) experiments, in which animals are trained to make an operant response to receive electrical stimulation of the medial forebrain bundle, Δ⁹-THC has been shown to both facilitate (Gardner et al., 1988; Katsidoni et al., 2013; Lepore et al., 1996; Spiller et al., 2019) and attenuate brain-stimulation reward (BSR; Vlachou et al., 2007; Wiebelhaus et al., 2015). Similarly, squirrel monkeys can be trained to respond on a manipulandum for Δ⁹-THC infusions (Justinova et al., 2003; Tanda et al., 2000), but attempts with other primate species (rhesus, cynomolgus) have failed (Harris et al., 1974; John et al., 2017; Mansbach et al., 1994). Conflicting findings in brain microdialysis studies have also been reported such that Δ⁹-THC produced an increase (Chen et al., 1991; Tanda et al., 1997) or no change (Castañeda et al., 1991) in extracellular dopamine (DA) levels in the nucleus accumbens (NAc) in rats. The neurobiological mechanisms underlying such paradoxical effects are poorly understood.

Δ⁹-THC is a partial agonist at the cannabinoid CB₁ (K_i =5.05 nM) and CB₂ (K_i = 3.13 nM) receptors (Pertwee, 2008). The majority of investigations into the mechanism underlying Δ⁹-THC’s rewarding vs. aversive effects have focused on the CB₁ receptor, given its high expression in brain tissues (Howlett and Abood, 2017; MacKie, 2006). Electrophysiological evidence in brain slices indicates that Δ⁹-THC may disinhibit DA neurons in the ventral tegmental area (VTA) by reducing presynaptic GABA inputs (Melis et al., 2000). Therefore, it was proposed that cannabinoid reward might be mediated by CB₁R activation of VTA GABA neurons (Lupica et al., 2004; Tanda, 2016; Tanda and Goldberg, 2003). However, there is lack of direct behavioral evidence supporting this hypothesis in vivo. This hypothesis also cannot explain how cannabinoids produce aversive effects as described above.

With recent advances in transgenic, RNAscope ISH, and optogenetic techniques, we have attempted to dissect the roles of CB₁ and CB₂ receptors in cannabis reward vs. aversion. Using conditional CB₁-knockout (CB₁-KO) mice as controls, we have recently reported that CB₁ receptors are expressed not only in VTA GABA neurons, but also in VTA glutamate neurons (Han et al., 2017). Optogenetic activation of VTA glutamatergic neurons was rewarding as assessed by intracranial self-stimulation (ICSS) and inhibited by Δ⁹-THC, indicating that Δ⁹-THC is reward-attenuating or aversive. Importantly, selective deletion of CB₁Rs on glutamate neurons blocked Δ⁹-THC action in optical ICSS, implicating CB₁Rs on glutamate neurons in cannabinoid aversion (Han et al., 2017). In addition, Δ⁹-THC or WIN55,212-2 produced biphasic effects on classical ICSS in that low doses enhanced, whereas high doses inhibited electrical BSR (Spiller et al., 2019). Strikingly, these effects were blocked by CB₁ or CB₂ receptor antagonists, respectively, and mimicked by CB₁ or CB₂ receptor agonists. Taken together, these findings suggest that both CB₁ and CB₂ receptor mechanisms mediate cannabinoid reward us. aversion (Spiller et al., 2019).

CB₂R expression was previously thought to be limited to the periphery particularly immune cells (Atwood and MacKie, 2010; Howlett and Abood, 2017). However, RNAscope ISH and fluorescent IHC assays enabled the detection of low levels of CB₂R expression in the brain (Zhang et al., 2014; Zhang et al., 2017). Indeed, CB₂Rs have been identified on dopaminergic neurons in the VTA (Zhang et al., 2014; Zhang et al., 2017; Zhang et al., 2015; Zhang et al., 2019) as well as the NAc (Bystrowska et al., 2018; Foster et al., 2016). Further, activation of CB₂Rs inhibits DA neuronal firing in the VTA and decreases DA release in the NAc (Ma et al., 2019; Xi et al., 2011; Zhang et al., 2014; Zhang et al., 2017). Cell type-specific knockout mice with CB₂ deleted from DA neurons show enhanced cocaine place preferences (Canseco-Alba et al., 2019). These findings suggest that CB₂R activation suppresses BSR and drug reward (Galaj and Xi, 2019; Jordan and Xi, 2019). However, it is unknown whether genetic deletion of CB₁ or CB₂ receptors alters the affective and neurochemical responses to Δ⁹-THC.

Therefore, in the present study, we utilized two mutant mouse lines to differentiate the relative contributions of CB₁ and CB₂ receptors to Δ⁹-THC reward and aversion as assessed by conditioned place preference (CPP) and brain DA response to Δ⁹-THC, given the critically important role of DA in drug reward (Galaj et al., 2020; Jordan et al., 2019; Volkow and Morales, 2015). An additional set of microdialysis experiments in rats was included to evaluate possible species differences in NAc DA response to Δ⁹-THC or SR141716A (a CB₁R antagonist or inverse agonist) and compare our findings with previous reports in rats (Chen et al., 1991; Chen et al., 1990; Malone and Taylor, 1999; Tanda et al., 1997).

We also observed the locomotor response to Δ⁹-THC since Δ⁹-THC-induced hypoactivity is a classical pharmacological effect of cannabinoids (Ledent et al., 1999; Wang et al., 2020; Zimmer et al., 1999) and locomotor behavior is largely DA-dependent (Beninger, 1983; Li et al., 2009; Xi et al., 2011). We have previously reported that both CB₁ and CB₂ receptors modulate locomotor activity (Li et al., 2009; Xi et al., 2011; Wang et al., 2020). Specifically, deletion of CB₁Rs lowers basal locomotor activity, attenuates cocaine hyperlocomotion, and mediates Δ⁹-THC-induced catalepsy and immobility (Li et al., 2009; Wang et al., 2020). However, CB₂R agonism decreases cocaine’s locomotor activating effects in wild-type, but not CB₂-KO mice (Xi et al., 2011). Therefore, we were interested in determining whether deletion of CB₁Rs and/or CB₂Rs would alter Δ⁹-THC-induced hypoactivity. We hypothesized that activation of CB₁Rs is rewarding and locomotor-stimulating, while activation of CB₂Rs is aversive and locomotor-suppressing. Accordingly, genetic deletion of CB₁Rs should block Δ⁹-THC reward, whereas deletion of CB₂Rs should block Δ⁹-THC aversion and lead to an increase in Δ⁹-THC reward.

2. EXPERIMENTAL PROCEDURES

2.1. Subjects

Male and female wildtype (WT) and mutant mice (CB₁^−/− and CB₂^−/−) were bred at the National Institute on Drug Abuse Intramural Research Program on a C57BL/6J genetic background. Initial CB₁^+/− breeding pairs (3) were kindly provided by Dr. Andreas Zimmer during his tenure at the National Institute of Mental Health (Bethesda, MD; Zimmer et al., 1999). Dr. George Kunos of the National Institute on Alcohol Abuse and Alcoholism (Rockville, MD) donated 3 CB₂^+/− breeding pairs (Buckley et al., 2000). Mutant mice from each heterozygous genotype were bred to produce homozygous offspring and genotyped by Charles River Laboratories. In addition, male Long–Evans rats (Charles River Laboratories, Raleigh, NC, USA), weighing 250–300 g, were used for in vivo microdialysis. Animals were maintained on a reverse 12 hr light-dark cycle (lights on at 7am, off at 7pm) with ad lib food and water provided. The vivarium was set to a temperature of 21-23° C with 40-50% humidity. All experimental procedures were subject to the Guide for the Care and Use of Laboratory Animals, 8^thedition. The study protocols (#13-BNRB-48 and #13-BNRB-84) were approved by the Institutional Care and Use Committee at the National Institute on Drug Abuse.

2.2. Drugs

Δ⁹-THC was provisioned by the Pharmacy of the National Institute on Drug Abuse Intramural Research Program (Baltimore, MD, USA). The stock Δ⁹-THC (50 mg/ml) was dissolved in 100% ethanol. Therefore, we used 0.5% Tween-80 (Sigma-Aldrich, St. Louis, MO, USA) and saline to dilute it by 50× to the final working solution containing 1 mg/ml Δ⁹-THC, 0.5% Tween-80 and ~2% ethanol. Accordingly, the vehicle for Δ⁹-THC also contained 0.5% Tween-80 and ~2% ethanol. SR141716A (free base form) was provided by Research Triangle Institute (Research Triangle Park, NC, USA) and dissolved in vehicle containing 0.5% Tween 80 and saline (without ethanol).

2.3. General Procedures

2.3.1. Apparatus

Six CPP boxes (MED-CPP-3013, Med Associates, St. Albans, VT, USA) were utilized for place conditioning testing. Each apparatus contained two large chambers (17.4 × 12.7 × 12.7 cm³) conjoined by a smaller middle compartment (11.7 × 12.7 × 12.7 cm³). Distinct visual and tactile cues were present in each compartment. The left chamber was black with a stainless-steel floor composed of 3.2 mm diameter rods. The right chamber was white with stainless steel mesh flooring. The middle compartment consisted of gray walls and a gray polyvinyl chloride floor. Each chamber housed a clear plexiglass top with an LED light affixed. Photobeams in each apparatus detected movement, which was recorded using Med Associates software (SOF-700RA-4).

2.3.2. CPP procedure

The full CPP procedure is illustrated in Figure 1. During pre-conditioning, naïve mice were given two sessions (1 a day) with full access to the three CPP chambers for 30 minutes. Time spent in each chamber was monitored and no preference was observed for the white or black side (AVG 358.9 s white, AVG 312.3 s black), indicating an unbiased CPP apparatus. Individual subjects with a preference for either compartment (defined as > 600 s) were removed from all analyses. Subjects were randomly assigned to Δ⁹-THC dose groups such that 48 mice received 1 mg/kg (n = 18 WT, 17 CB₁^−/−, and 13 CB₂^−/−) and 48 received 5 mg/kg (n = 20 WT, 12 CB₁^−/−, and 16 CB₂^−/−). On conditioning day 1, Δ⁹-THC was administered i.p. followed by placement in the white or black compartment for 45 minutes. The compartment paired with drug injections was randomly assigned (i.e. a counterbalanced or unbiased procedure). The following day, vehicle (0.5% Tween-80) was injected, and subjects were restricted to the opposite compartment for 45 minutes. This procedure was repeated for a total of 4 drug pairings and 3 vehicle pairings. 24 hours and 48 hours after the final Δ⁹-THC injection, a test was run under identical parameters to the pre-conditioning sessions. CPP scores were calculated based on the average time spent on the drug paired side - vehicle paired side over the two test days.

Figure 1.

Δ⁹-THC place conditioning in wild type, CB₁^−/− and CB₂^−/−. (A) Experimental timeline of events for each day of CPP conditioning and testing. (B) WT mice displayed a significant increase in CPP score at 1 mg/kg (p<0.05). but not at 5 mg/kg (p>0.05). (C) CB₁^−/− mutants displayed a dose-dependent trend toward CPA after Δ⁹-THC administration. (D) Δ⁹-THC produced a robust dose-dependent place preference in CB₂^−/− mice. *p<0.05, ***p< 0.001, relative to preconditioning. #p< 0.05 compared to WT mice.

2.3.3. Open-field locomotion

Locomotor activity was next examined in naïve groups of mice to determine the impact of CB1 or CB2 receptor deletion on basal activity levels as well as activity following acute Δ⁹-THC administration. During the week before locomotor testing, mice were habituated to locomotor detection chambers (Accuscan Instruments, Columbus, OH, USA) during 3 h daily sessions. On the test day, mice were placed in the chamber for 1 h of habituation (baseline), after which the program was paused, and Δ⁹-THC was injected. Next, mice were placed back in the open-field apparatus and locomotor activity was measured over a 2 h period. Each Δ⁹-THC dose (0, 1, 3, 10 mg/kg, i.p.) was tested in a balanced manner in 3 genotype groups of mice (n = 8 WT, 8 CB₁^−/−, and 7 CB₂^−/−). The time intervals between the drug tests were 2-4 days. The distance traveled before and after injections was collected in 10-min intervals using the VersaMax data analysis system (Accuscan Instruments).

2.3.4. In vivo microdialysis with HPLC assays

Naïve mice were anesthetized using a cocktail of ketamine hydrochloride (80 mg/ml) and xylazine hydrochloride (12 mg/ml) prior to insertion of an intracranial guide cannulae (MAB 4.15.IC, SciPro Inc., Sanborn, NY, USA) into the NAc (stereotaxic coordinates: AP +1.4 mm, ML ±1.5 mm, DV −3.8 mm with an angle of 8° from vertical). Dental acrylic was applied to secure the guide cannulae to the skull. Standard aseptic surgical and stereotaxic procedures were followed. Subjects were given 7 days to recover preceding the experimental procedure. A probe (MAB 4.15.2.PES, SciPro Inc.) was inserted into the NAc 12 hours before sample collection to reduce the occurrence of damage induced neurotransmitter release. A syringe pump (Bioanalytical Systems, West Lafayette, IN, USA) infused dialysate buffer at least 2 hrs prior to sample collection. Baseline microdialysate samples were taken in 20 minute intervals for 1 hr, followed by an i.p. Δ⁹-THC injection (1 or 3 mg/kg) and further sampling for 3 hrs. Mice were randomly assigned to receive 1 mg/kg (n = 7 WT, 8 CB₁^−/−, and 6 CB₂^−/−) or 3 mg/kg (n = 15 WT, 6 CB₁^−/−, and 11 CB₂^−/−) Δ⁹-THC.

The procedures for in vivo brain microdialysis in rats were the same as in mice. The stereotaxic coordinates for the NAc in rats were AP +1.6 mm, ML ±1.8 mm, DV −4.3 mm with an angle of 6° from vertical. Rats (n=20) were randomly divided into 3 dose groups for i.p. injections of Δ⁹-THC or SR141716A (6-8 per group), and an additional 10 rats were divided into 2 groups for intra-NAc local perfusion of Δ⁹-THC (n=5) or SR141716A (n=5). Samples were frozen at −80 °C until analysis could be performed using an ESA electrochemical detection system (Chelmsford, MA, USA) for a detailed description see (Li et al., 2009; Xi et al., 2006).

2.5. Statistical Analysis

Data are expressed as means ±SEM. Three-way repeated measures analyses of variance (ANOVA) were utilized to analyze the CPP and locomotor data with the factors of dose, genotype, and test or time. A two-way repeated measures ANOVA assessed differences during baseline microdialysis sampling across genotype and time. Two-way ANOVAs were used to analyze the microdialysis data for each genotype and to directly compare each mutant strain and WT mice. Area under the curve (AUC) was calculated for the normalized locomotor data as the sum of all time points following drug injections. AUC data was analyzed using a two-factor (dose × genotype) ANOVA. Significant interactions were followed by post hoc tests using Fishers Least Significant Difference. Statistical significance was set to p<0.05.

3. RESULTS

3.1. Δ⁹-THC place conditioning in mutant mice

The general experimental procedures are shown in Figure 1A. Systemic administration of Δ⁹-THC produced a significant place preference at 1 mg/kg, but not 5 mg/kg, in WT mice (Fig. 1B) , a dose-dependent trend toward conditioned place aversion (CPA) in CB₁-KO mice (Fig. 1C), and a dose-dependent THC place preference in CB₂-KO mice (Fig. 1D). The three-way repeated measures ANOVA revealed a significant main effect of Genotype (F_2,90 =6.256, p<0.01) and Test (F_1,90 = 18.52, p<0.001) as well as a significant interaction between Genotype, Test and Dose (F_2,90 =3.292, p<0.05). Subsequent post-hoc comparisons indicated that CPP scores were significantly higher during the test than pre-conditioning for WT subjects at 1 mg/kg and CB₂-KO mice at 1 and 5 mg/kg. WT mice also showed significantly lower CPP scores at 5 mg/kg relative to CB₂-KO mutants and significantly higher scores than CB₁-KO mice at both doses.

3.2. Effects of Δ⁹-THC on extracellular NAc DA in mutant mice

We then compared basal levels of extracellular DA in the NAc between different genotypes of mice. We found a significant decrease in basal levels of extracellular DA in CB₁-KO mice (average value across three baseline samples: 0.179 ± 0.046 nM, n=14) and a trend towards a decrease in CB₂-KO mice (0.249 ± 0.063 nM, n=17), relative to WT mice (0.433 ± 0.066 nM, n=20). A one-way ANOVA revealed a significant main effect of Genotype (F_2,48 =4.651, p<0.05). Post-hoc group comparisons showed a significant reduction in CB₁-KO mice (p<0.05), but not CB₂-KO mice (p=0.065), compared to WT mice. These findings are consistent with our previous reports (Li et al., 2009; Xi et al., 2011).

Given the differences observed at baseline, Δ⁹-THC-induced changes in extracellular DA are presented as a percent of baseline to enable data comparisons among the three mouse strains. The results shown in Figure 2 indicate that Δ⁹-THC produced a dose-dependent reduction in extracellular DA in WT mice (Fig. 2A) and CB₁-KO mice (Fig. 2B). In contrast, Δ⁹-THC elicited a potent and dose-dependent increase in extracellular NAc DA in CB₂-KO mice (Fig. 2C). A two-way repeated measures ANOVA on WT mice (Fig. 2A) revealed a significant main effect of Time (F_5,86 = 4.704, p<0.01) and Dose (F_1,18 = 7.111, p<0.05) as well as a Dose × Time interaction (F_5,86 = 3.363, p<0.01). Post-hoc comparisons showed significantly lower DA release relative to the final baseline timepoint beginning 20 minutes post 3 mg/kg Δ⁹-THC. In CB₁^−/− mice (Fig. 2B), a main effect of Dose (F_1,12 = 8.531, p<0.05), Time (F_4,50 = 3.348, p<0.05) and a Dose × Time interaction (F_4,50 = 3.346, p<0.05) were uncovered. Post-hoc tests demonstrated decreased DA release relative to baseline 40-60 and 120-180 minutes following Δ⁹-THC administration (3 mg/kg). The two-way ANOVA on CB₂^−/− mice (Fig. 2C) revealed a main effect of Time (F_1,15 = 5.81, p<0.05), but not Dose (F_1,15 = 0.461, p>0.05) and no interaction between Time and Dose (F_1,15 = 0.974, p>0.05). Post-hoc comparisons revealed a significant increase in accumbal DA concentrations 40 minutes post Δ⁹-THC relative to baseline. We also compared Δ⁹-THC-induced changes in extracellular DA between WT and CB₁ - or CB₂-KO mice (Fig. 2D). The two-way ANOVA analyzing the effects of 3 mg/kg Δ⁹-THC across three genotypes of mice revealed a main effect of Genotype (F_2,27 = 13.28, p0.001) and a significant Time × Genotype interaction (F_22,36 = 1.902, p<0.05). Post-hoc comparisons revealed that DA levels were significantly higher in CB₂-KO mice than in WT mice during all timepoints following Δ⁹-THC administration. An additional two-way ANOVA was run directly comparing CB₁-KO and WT mice at 3 mg/kg Δ⁹-THC (Fig. 2D). There was a main effect of Genotype (F_1,19 = 5.109, p<0.05) and Time (F_11,209 = 16.853, p<0.001) and a significant interaction between these factors (F_11,209 = 3.445, p<0.05). CB₁-KO mice exhibited a greater decrease in NAc release 40-80 and 120-160 min post Δ⁹-THC relative to WT mice (p<0.05). Figure 2E shows a representative image illustrating the location of a microdialysis probe in the mouse brain. Most of the probes (membrane portions) penetrated both the shell and core of the NAc (Fig. 2F).

Figure 2:

Effects of Δ⁹-THC on extracellular DA in the NAc. Δ⁹-THC produced a dose-dependent decrease in extracellular DA in WT (A) and CB₁^−/− mice (B), while CB₂^−/− mice demonstrated a dose-dependent increase in extracellular DA relative to baseline (C). (D) Δ⁹-THC, at 3 mg/kg, reduced extracellular DA in CB₁^−/− mice and elevated DA in CB₂^−/− mice compared to WT controls. (E, F) A representative image and histological verification showing that microdialysis probes are located in the shell and core of the NAc. *p<0.05, **p<0.01, ***p<0.001.

3.3. Effects of Δ⁹-THC and SR141716A on extracellular NAc DA in rats

Given that the above findings in mice contrast with previous findings in rats (Chen et al., 1991; Tanda et al., 1997), the effects of Δ⁹-THC and SR141716A (a CB₁R antagonist or inverse agonist) were evaluated on extracellular NAc DA in rats. Systemic administration of 1, 3 mg/kg Δ⁹-THC did not produce significant changes in extracellular DA levels compared to vehicle (Fig. 3A). The two-way repeated measures ANOVA did not reveal a significant main effect of Dose (F_2,17 = 1.34, p>0.05) or a Dose × Time interaction (F_22,187 = 0.91, p>0.05), but revealed a significant Time effect (F_11,178 = 4.70, p<0.001). Although extracellular DA displayed a trend to increase at 40 and 60 min compared to baseline, there was no difference observed between vehicle and Δ⁹-THC groups at any time point. In addition, the effects of systemic administration of SR141716A were evaluated on extracellular DA. We found that 2 and 10 mg/kg SR141617A failed to alter extracellular DA in rats (Fig. 3B). The two-way ANOVA did not reveal a significant main effect of treatment (F_2,15 = 0.86, p>0.05), time (F_11,165 = 1.02, p>0.05), or a treatment × time interaction (F_11,165 = 0.78, p>0.05).

Figure 3:

Effects of Δ⁹-THC and SR141716A on extracellular DA in the NAc in rats. (A) Systemic administration of Δ⁹-THC or vehicle produced a small decrease in extracellular DA, but no difference was found between the vehicle and THC-treatment groups. (B) Systemic administration of SR141716A failed to alter extracellular DA in the NAc. (C) Local perfusion of Δ⁹-THC failed to alter extracellular DA, while SR141716A elevated extracellular DA in a concentration-dependent manner. *p<0.05, ***p<0.001, compared to baseline.

To confirm the above finding and also determine whether the 2% alcohol present in the vehicle or Δ⁹-THC solution may underlie the non-Δ⁹-THC-specific increase in extracellular DA, we further diluted the Δ⁹-THC solution from 1 mM to 0.1 μM using 0.5% Tween-80 to minimize ethanol levels in the lower concentrations of Δ⁹-THC solution, and then locally perfused them into the NAc in a sequence from 0.1 μM to 100 μM. We found that Δ⁹-THC also failed to alter extracellular DA (Fig. 3C). Unexpectedly, SR141716A produced a dose-dependent increase in extracellular DA in the NAc (Fig. 3C), suggesting that endocannabinoids may tonically inhibit the mesolimbic DA system. Two-way ANOVA for repeated measures over time revealed a marginally significant treatment main effect (F_1,8 = 4.91, p=0.057), time main effect (F_14,112 = 9.25, p<0.001) and treatment × time interaction (F_14,112 = 2.59, p<0.01). Post-hoc individual group comparisons revealed significant increases in DA at 100 μM SR141716A, when compared with baseline (Fig. 3C).

3.4. Δ⁹-THC-induced changes in open-field locomotion

Lastly, locomotor activity was assessed in WT and mutant mice following Δ⁹-THC administration. Relative to WT mice, CB₁-KO mice displayed a significant reduction, while CB₂-KO mice showed a moderate increase in basal levels of locomotion before and/or after vehicle injections (Fig. 4A), suggesting that the endocannabinoids can differentially modulate CB₁ and CB₂ receptors. A one-way ANOVA for the area under curve (AUC) data revealed a genotype main effect on basal locomotion in the 1 h period before the vehicle injection (Fig. 4B, F_2,20=24.72, p<0.001) and for 2 hrs after vehicle injections (Fig. 4C, F_2,20=26.12, p<0.001). Post-hoc individual group comparisons revealed a significant reduction in CB₁-KO mice before and after vehicle and a significant increase in CB₂-KO mice after vehicle compared to WT mice.

Figure 4:

Genetic deletion of CB₁ and CB₂ receptors differentially alters basal levels of open-field locomotor activity. (A) Time course of locomotion before and after vehicle injection in WT, CB₁^−/− and CB₂^−/− mice. (B) Area under curve (AUC) data during the initial 1-h habituation (before vehicle injection) in open-field locomotion chambers. (C) AUC data during 2-h maintenance (after vehicle) in locomotion chambers in three different genotypes of mice. *p<0.05, ***p<0.001, relative to WT mice.

Consistent with the above microdialysis data, systemic administration of Δ⁹-THC (1,3 mg/kg, i.p.) produced a dose-dependent reduction in open-field locomotor activity in WT mice (Fig. 5A), but not CB₂-KO mice (Fig. 5G). The time course in locomotor response to Δ⁹-THC in WT mice appears to parallel the time course of DA response to Δ⁹-THC, suggesting a possible correlation between Δ⁹-THC-induced changes in DA and locomotor activity. CB₁-KO mice displayed a robust reduction in basal levels of locomotion (Fig. 4) and the amplitude of locomotor response to Δ⁹-THC (Fig. 5D). The three-way repeated measures ANOVA showed a significant Genotype × Time × Dose interaction (F_102,408 =1.56, p<0.05). Since the basal levels of locomotion differ between WT and mutant mice, Δ⁹-THC-induced locomotor depression was normalized to percent changes over baseline (Fig. 5–B, E, H). A two-way ANOVA on the ΔAUC data produced a significant Dose Δ Genotype interaction (F_6,78 = 2.61, p<0.05). Post-hoc group comparisons revealed that 3 and 10 mg/kg Δ⁹-THC significantly decreased open-field activity in WT mice, but not CB₂-KO mice. CB₁-KO mice demonstrated hypolocomotion at 1 and 3 mg/kg Δ⁹-THC, but not 10 mg/kg.

Figure 5:

Open field locomotion following Δ⁹-THC administration in mice. (A, B, C) Original locomotor activity (A), normalized data (% baseline) (B), and ΔAUC (C) in WT mice, illustrating that Δ⁹-THC dose-dependently decreased locomotor activity (n = 8). (D, E, F) Original locomotor activity (D), normalized data over baseline (% baseline) (E), and ΔAUC (F) in CB₁-KO mice, illustrating that CB₁-KO mice displayed significantly lower basal level of locomotion and lower locomotor response (amplitude) to Δ⁹-THC (D) (n = 8). Unexpectedly, the normalized data (E, F) also revealed dose-dependent locomotor depression after Δ⁹-THC administration in CB₁-KO mice. (G, H, I) Original locomotor activity (G), normalized data over baseline (% baseline) (H), and ΔAUC (I) in CB₂-KO mice, illustrating that Δ⁹-THC had no effect locomotor activity (n = 7). *p<0.05, **p<0.01 relative to vehicle.

4. DISCUSSION

In the present experiment we investigated how genetic deletion of the CB₁ and CB₂ receptors impacted the rewarding versus aversive effects of Δ⁹-THC as assessed by CPP, NAc DA release, and open-field locomotor activity. In WT mice, Δ⁹-THC produced a CPP at a low dose (1 mg/kg), which was absent in CB₁-KO mice, suggesting a CB₁-mediated effect. At 5 mg/kg Δ⁹-THC, CPP scores were significantly reduced in CB₁-KO mice relative to WT. In contrast, CB₂-KO mice developed robust dose-dependent CPP, indicating that CB₂R activation underlies the aversive effects produced by Δ⁹-THC. Neurochemical assessments in WT mice revealed a dose-dependent reduction in extracellular DA following Δ⁹-THC injections. Genetic deletion of CB₁Rs potentiated Δ⁹-THC-induced decreases in extracellular DA, whereas in CB₂-KO mice, Δ⁹-THC produced a significant increase in extracellular DA at both doses. Similarly, deletion of CB₁ and CB₂ receptors attenuated Δ⁹-THC-induced locomotor depression. These findings suggest that both CB₁ and CB₂ receptors likely mediate the hedonic spectrum of Δ⁹-THC such that CB₁R activation underlies cannabis reward and CB₂R activation mediates its aversive effects.

The paradoxical effects of Δ⁹-THC in animal models of reward have been well documented (Panagis et al., 2008; Tanda, 2016). Repeated attempts to train rodents and primates on Δ⁹-THC self-administration have been unsuccessful (Harris et al., 1974; John et al., 2017; Lefever et al., 2014; Mansbach et al., 1994), although one research group demonstrated that squirrel monkeys can be trained to self-administer cannabinoids (Justinova et al., 2008; Justinova et al., 2003; Tanda et al., 2000). Δ⁹-THC has been reported to produce biphasic effects in CPP and ICSS experiments, such that low doses are rewarding, while high doses are aversive (Kubilius et al., 2018; Spiller et al., 2019). These effects tend to vary across rat strains, species, and procedures such that rewarding and aversive effects have been described at the same Δ⁹-THC dose (DeVuono et al., 2017; Valjent and Maldonado, 2000). Discrepancies in the affective properties of cannabinoids have been noted in the literature for decades; however, it is unknown why such variability exists. One theory advanced is that the enjoyment reported by human users is based on negative reinforcement i.e. the removal of an aversive state such as anxiety or pain (Rubino et al., 2007; Rutkowska et al., 2006), which is not measured in typical CPP and ICSS experiments. However, this theory has not been supported by work evaluating the rewarding profile of Δ⁹-THC following a stressful experience (DeVuono et al., 2017; Fokos and Panagis, 2010).

An alternative explanation is that the affective experience of a cannabinoid is dependent on the underlying neurocircuitry and varying distributions of CB₁ and CB₂ receptor populations. Prior work from our laboratory has attempted to dissect the role of different cannabinoid receptors in cannabis reward versus aversion (Galaj and Xi, 2019; Han et al., 2017; Jordan and Xi, 2019; Spiller et al., 2019). By using highly sensitive RNAscope ISH assays with CB₂-KO mice as controls, we have recently identified functional CB₂R expression in VTA DA neurons (Xi et al., 2011; Zhang et al., 2014; Zhang et al., 2017). Given that activation of CB₂Rs inhibits firing rates of VTA DA neurons and decreases DA release in the NAc (Foster et al., 2016; Ma et al., 2019; Xi et al., 2011; Zhang et al., 2014; Zhang et al., 2017), we proposed that CB₂R activation may in part underlie cannabinoid aversion (Galaj and Xi, 2019; Humburg et al., under review; Jordan and Xi, 2019; Spiller et al., 2019). This hypothesis was supported by work demonstrating that Δ⁹-THC or WIN55,212-2 attenuates brain-stimulation reward (BSR), maintained by electrical stimulation of the medal forebrain bundle in rats or by optical stimulation of VTA DA neurons in DAT-cre mice, an effect that can be blocked by CB₂R antagonism (Humburg et al., under review; Spiller et al., 2019). Furthermore, beta-caryophyllene, a CB₂R agonist, inhibits optical BSR and cocaine-enhanced BSR (Galaj et al., in press; He et al., 2020).

As described in the introduction, selective deletion of glutamatergic CB₁Rs can block Δ⁹-THC’s inhibitory action in optical ICSS, suggesting that a CB₁R mechanism in glutamate neurons may also contribute to cannabinoid aversion (Han et al., 2017). Thus, stimulation of CB₁Rs on glutamate neurons as well as CB₂Rs on DA neurons likely mediates cannabinoid aversion and activation of CB₁Rs on GABA neurons has been implicated in cannabinoid reward (Figure 6). The final outcome of a given cannabinoid likely depends on the balance of these opposing actions (Galaj and Xi, 2019; Han et al., 2017; Jordan and Xi, 2019; Spiller et al., 2019). This new hypothesis may partially explain why the affective properties of cannabinoids are so inconsistent in the literature, since different species, rodent strains or even individual subjects may have unique cellular distributions of CB₁Rs and CB₂Rs in the brain.

Figure 6:

Graphic diagram illustrating how Δ⁹-THC modulates the mesolimbic DA system and cannabinoid action via CB₁ and CB₂ receptors. Δ⁹-THC may produce rewarding effects by binding to CB₁ receptors on GABAergic neurons or afferents in the VTA, thereby reducing GABA-mediated inhibition of VTA DA neurons and increasing DA release in the NAc. Conversely, Δ⁹-THC may produce aversive effects by activating CB₁ receptors on glutamatergic neurons or afferents in the VTA as well as CB₂ receptors on DA neurons, thereby inhibiting VTA DA release to the NAc. The subjective effects of Δ⁹-THC may thus depend on the balance of opposing effects produced by activation of CB₁ and CB₂ receptors in different phenotypes of neurons. VTA: Ventral tegmental area. NAc: Nucleus accumbens, DA: Dopamine.

In the present study, we provide additional evidence supporting this CB₁-CB₂ hypothesis. Given the wide range of CPP scores often observed with cannabinoids, we used large group sizes (n = 12-20 per group) to reduce variability and found that Δ⁹-THC was rewarding at a low dose (1 mg/kg) in WT subjects, consistent with prior work in mice (Ghozland et al., 2002; Soria et al., 2004; Valjent and Maldonado, 2000). CB₁R deletion blocked Δ⁹-THC-induced CPP, supporting the conjectured role of these receptors in cannabinoid reward. This is also consistent with earlier studies in which pharmacological antagonism of CB₁Rs reduced Δ⁹-THC-induced CPP, self-administration, and BSR (Braida et al., 2004; Justinova et al., 2008; Spiller et al., 2019). Perhaps the most important finding in this study is that that a high dose of Δ⁹-THC activates the CB₂R, producing an aversive effect that antagonizes CB₁-mediated reward. Specifically, 5 mg/kg Δ⁹-THC produced no effect in WT mice, lowered CPP scores in CB₁-KO mice, but induced a robust place preference in CB₂-KO mice, implicating a CB₂R mechanism in Δ⁹-THC’s aversive effects. These findings are consistent with our previous report demonstrating that electrical BSR was increased by ACEA, a selective CB₁R agonist, but suppressed by JWH133, a selective CB₂R agonist (Spiller et al., 2019). Moreover, in prior work JWH133 administration or over-expression of CB₂Rs in the brain attenuated cocaine CPP, self-administration and locomotor sensitization (Aracil-Fernandez et al., 2012; Delis et al., 2016; Lopes et al., 2020; Xi et al., 2011), while local infusions of JWH133 into the NAc and VTA inhibited cocaine self-administration in WT and CB₁-KO mice, but not CB₂-KO mice (Xi et al., 2011; Zhang et al., 2014; Zhang et al., 2017). Further, genetic deletion of CB₂Rs from midbrain DA neurons enhanced cocaine-induced hyperactivity, locomotor sensitization, and CPP (Canseco-Alba et al., 2019). Growing evidence indicates that CB₂ receptors are expressed in VTA DA neurons and regulate DA neuronal activity (Foster et al., 2016; Xi et al., 2011; Zhang et al., 2014), which may explain how CB₂ receptors modulate the mesolimbic DA reward system and cannabinoid aversion (Galaj and Xi, 2019; Jordan and Xi, 2019).

Surprisingly, no increase in extracellular DA was observed following Δ⁹-THC administration at a dose that induced a place preference in WT mice (1 mg/kg). One explanation for these disparate behavioral and neurochemical results is the technical limitations of microdialysis. Specifically, this approach may not be able to detect subtle changes in extracellular DA produced by Δ⁹-THC at low doses due to its poor temporal and spatial resolution. Indeed, another report in mice showed that a subthreshold Δ⁹-THC dose (0.3 mg/kg), increased extracellular DA, but did not induce a place preference (Robledo et al., 2007), suggesting a possible decoupling between the behavioral and neurochemical changes. A second possibility is that Δ⁹-THC action in DA signaling may display some subregion-specificity within the NAc. Tanda and colleagues demonstrated that Δ⁹-THC preferentially increases DA in the NAc core over the NAc shell in rats (1997). In the present study, we did not restrict probe placement to the NAc shell or core due to the practical difficulty in separating these subregions in mice.

It is important to note that in previous studies acute Δ⁹-THC injections increased extracellular DA in the NAc within rats (Chen et al., 1991; Chen et al., 1990; Malone and Taylor, 1999; Tanda et al., 1997). This work conflicts with our own findings in mice. To further address this potential species difference in the DA response to cannabinoids, we performed a microdialysis experiment in rats. Systemic administration of Δ⁹-THC trended towards an increase in NAc DA. However, vehicle infusions produced an analogous surge in DA, suggesting a non-specific effect. To further confirm this finding and determine whether the ethanol (~2%) present in the vehicle and Δ⁹-THC solution may underlie the non-specific DA response observed, we diluted the Δ⁹-THC solution using 0.5% Tween-80 and saline to vary the ethanol level in the Δ⁹-THC solution. Local perfusion of varying Δ⁹-THC concentrations (0.1~100 μM) into the NAc also failed to alter extracellular DA. This is consistent with a previous report in rats demonstrating no change in NAc DA following acute Δ⁹-THC administration (Castañeda et al., 1991). However, a trend towards an increase in DA was observed at 100 μM Δ⁹-THC, which contained a higher concentration of alcohol (0.625%). These data demonstrate that systemic administration or intra-NAc local perfusion of Δ⁹-THC has no effect on extracellular DA in rats. The small DA response observed is likely elicited by the presence of ethanol in the vehicle and Δ⁹-THC solution.

In addition to Δ⁹-THC, systemic SR141716A (rimonabant) also failed to alter extracellular DA. This is consistent with our previous findings with AM251, another CB₁R antagonist or inverse agonist, which failed to alter extracellular NAc DA (Xi et al., 2006). However, intra-NAc local perfusion of SR141716A elevated extracellular DA in a concentration-dependent manner, suggesting that endocannabinoids may tonically inhibit DA release via activation of CB₁Rs in the NAc. Although our theory of the affective properties of cannabinoids suggests that activation of CB₁Rs is rewarding, this effect is thought to be mediated specifically by GABAergic CB₁Rs. These receptors are not limited to GABA neurons but are also expressed on VTA glutamate and a subpopulation of DA neurons (Han et al., 2017; Humburg, 2020). As such, activation of CB₁Rs on VTA DA and glutamate neurons may lead to a reduction in DA release. These findings provide additional evidence supporting the presence of a species difference in brain DA response to Δ⁹-THC between rats and mice (Zhang et al., 2015). Specifically, systemic administration of cannabinoids tends to inhibit NAc DA release in mice but has no significant effect on NAc DA in rats. The CB₁-CB₂ hypothesis proposed above (Fig. 6) may well explain such species-specific effects in cannabinoid action (Liu et al., 2019; Liu et al., 2009; Zhang et al., 2015).

In relation to the locomotor data, genetic deletion of the CB₁ receptor robustly decreased basal locomotor activity, while genetic deletion of the CB₂Rs produced a moderate increase, suggesting that both receptors regulate basal locomotor activity. CB₁-KO mice also failed to demonstrate Δ⁹-THC-induced hypolocomotion at 10 mg/kg. This data nicely parallels prior work demonstrating the involvement of CB₁Rs in the motor effects of cannabinoids. Our lab has shown that CB₁-KO mice display lower basal and cocaine-enhanced locomotion relative to WT mice (Li et al., 2009) and previous reports indicate that CB₁R deletion blocks high doses of Δ⁹-THC-induced catalepsy and immobility (Compton et al., 1996; Grim et al., 2017; Ledent et al., 1999; Varvel et al., 2005; Wang et al., 2020). Surprisingly, when Δ⁹-THC-induced changes in locomotion were normalized over baseline (% baseline), CB₁-KO mice showed no differences from WT at 3 mg/kg, implicating other receptors in the motor effects of cannabinoids. Indeed, genetic deletion of CB₂Rs was able to block Δ⁹-THC-induced locomotor depression at this dose, suggesting that CB₂Rs mediate some of Δ⁹-THC’s locomotor suppressing effects. This finding is somewhat unexpected as CB₂-KO mice displayed a robust dose-dependent increase in DA response to Δ⁹-THC. As such, a corresponding rise in locomotor activity might be predicted. These findings denote a possible dissociation between DA and locomotor activity to cannabinoids in CB₂-KO mice. It may be that deletion of CB₂RS also modulates neurotransmitter release in the brain, which could counteract DA-induced hyperactivity. Another possibility is that neuroadaptations may occur in CB₂-KO mice, which compromise the locomotor activating effects of DA in this context.

In conclusion, the present findings demonstrate that CB₁R and CB₂R mechanisms differentially underlie cannabinoid reward versus aversion. The final outcome of cannabinoid action likely depends on the relative balance of CB₁R and CB₂R activation on different phenotypes of neurons. These findings provide insight into the paradoxical hedonic effects described by cannabis users as well as conflicting preclinical reports with Δ⁹-THC and other cannabinoids.