In vitro and in vivo pharmacological activity of minor cannabinoids isolated from Cannabis sativa

The Cannabis sativa plant contains more than 120 cannabinoids. With the exceptions of ∆9-tetrahydrocannabinol (∆9-THC) and cannabidiol (CBD), comparatively little is known about the pharmacology of the less-abundant plant-derived (phyto) cannabinoids. The best-studied transducers of cannabinoid-dependent effects are type 1 and type 2 cannabinoid receptors (CB1R, CB2R). Partial agonism of CB1R by ∆9-THC is known to bring about the ‘high’ associated with Cannabis use, as well as the pain-, appetite-, and anxiety-modulating effects that are potentially therapeutic. CB2R activation by certain cannabinoids has been associated with anti-inflammatory activities. We assessed the activity of 8 phytocannabinoids at human CB1R, and CB2R in Chinese hamster ovary (CHO) cells stably expressing these receptors and in C57BL/6 mice in an attempt to better understand their pharmacodynamics. Specifically, ∆9-THC, ∆9-tetrahydrocannabinolic acid (∆9-THCa), ∆9-tetrahydrocannabivarin (THCV), CBD, cannabidiolic acid (CBDa), cannabidivarin (CBDV), cannabigerol (CBG), and cannabichromene (CBC) were evaluated. Compounds were assessed for their affinity to receptors, ability to inhibit cAMP accumulation, βarrestin2 recruitment, receptor selectivity, and ligand bias in cell culture; and cataleptic, hypothermic, anti-nociceptive, hypolocomotive, and anxiolytic effects in mice. Our data reveal partial agonist activity for many phytocannabinoids tested at CB1R and/or CB2R, as well as in vivo responses often associated with activation of CB1R. These data build on the growing body of literature showing cannabinoid receptor-dependent pharmacology for these less-abundant phytocannabinoids and are critical in understanding the complex and interactive pharmacology of Cannabis-derived molecules.


Results
Type 1 cannabinoid receptor (CB1R). Radioligand Table 1). The estimated binding affinities of ∆ 9 -THCa and CBG were significantly lower than that of the reference compound, CP55,940, and ∆ 9 -THC (Table 1). Some compounds, such as THCV and CBC, displayed comparatively wide confidence intervals around the estimated K i values (Table 1). These values may be the result of complex pharmacology for these compounds and their incomplete competition with [ 3 H]CP55,940 resulting in an inherently 'noisy' system. These data stress the importance of multiplicity of approach (here inhibition of cAMP accumulation, βarrestin2 recruitment, and in vivo studies) for a more comprehensive understanding of a compound's activity. The estimated binding affinities of several compounds were not estimated if the upper and lower bounds of the competition binding curve could not be seen within the concentration range used, these compounds K i are listed as > 10,000 in Table 1.
Inhibition of forskolin-stimulated cAMP. The majority of compounds tested for their ability to inhibit the accumulation of forskolin (FSK)-stimulated cAMP were partial agonists of this effect at hCB1R. All compounds displayed an E max in the cAMP inhibition assay that was less than that of CP55,940 ( Fig. 3; Table 2). ∆ 9 -THCa, CBD, and CBDa were less efficacious than ∆ 9 -THC ( Fig. 3; Table 2). ∆ 9 -THC, THCV, and CBC were less potent than the reference agonist CP55,940 (Table 2).
βarrestin2 recruitment. All phytocannabinoids tested displayed little-to-no activity in the βarrestin2 recruitment assay relative to the reference agonist CP55,940 and within the concentration range used (Fig. 4). Among the phytocannabinoids, ∆ 9 -THC was the most-efficacious displaying an E max value of 37 ± 7.5% relative to CP55,940 ( Fig. 4; Table 3).  5; Table 1). The estimated binding affinity of CBDV, CBG, and CBC was significantly lower than that of the reference compound, CP55,940, and ∆ 9 -THC ( Table 1).
βarrestin2 recruitment. ∆ 9 -THC and CBD displayed modest agonist activity at hCB2R in the βarrestin2 recruitment assay relative to the reference agonist CP55,940 and within the concentration range used (Fig. 7). THCV and CBDV treatment produced demonstrable efficacy at hCB2R in the βarrestin2 recruitment assay, but with very low potency ( Fig. 7; Table 3).   Table 3.
Scientific Reports | (2020) 10:20405 | https://doi.org/10.1038/s41598-020-77175-y www.nature.com/scientificreports/ Ligand bias. Phytocannabinoid ligand bias at hCB1R and hCB2R was assessed between inhibition of FSKstimulated cAMP accumulation and βarrestin2 recruitment using the operational model of Black and Leff and as described previously 11,25 . Analyses were conducted such that ∆∆LogR values > 0 represented bias for inhibition of cAMP and ∆∆LogR values < 0 represented bias for the recruitment of βarrestin2 (Fig. 8). From these analyses, we observed that, at hCB1R, ∆ 9 -THCa, CBDa, CBG, and CBC displayed a bias for the inhibition of cAMP relative to βarrestin2 (Fig. 8a). No phytocannabinoids tested displayed βarrestin2 recruitment bias at hCB1R. At hCB2R, ∆ 9 -THC, CBDa, CBDV, CBG, and CBC all displayed a bias for inhibition of cAMP relative to βarrestin2; and no phytocannabinoids displayed a bias for βarrestin2 recruitment (Fig. 8b). Of note, CBDa, CBG, and CBC displayed a bias for inhibition of cAMP at both hCB1R and hCB2R (Fig. 8). Other phytocannabinoids tested did not display a bias for either inhibition of cAMP or βarrestin2 recruitment at hCB1R or hCB2R.
Selectivity between hCB1R and hCB2R. We estimated the selectivity of the phytocannabinoids tested by comparing between results obtained with hCB1R or hCB2R within each assay.

Discussion
In general, we observed that all phytocannabinoids tested displayed some degree of activity at either CB1R or CB2R in cell culture assays. The incomplete competition of these phytocannabinoids with [ 3 H]CP55,940 suggests that the occupied binding site of these ligands differs slightly from that of CP55,940. It is possible that the phytocannabinoids tested bound only a subset of amino acids in the CB1R ligand binding site(s) compared to CP55,940, as has been shown for CBD, Org27569, rimonabant, and anandamide 10,18,27,28 . ∆ 9 -THC that is commonly consumed in cannabis products is a decarboxylated derivative of the naturally present ∆ 9 -THCa 19 . In this study, ∆ 9 -THCa partially displaced [ 3 H]CP55,940 from CB1R and CB2R, with greater affinity for CB2R; and was a weak partial agonist of CB1R-dependent inhibition of cAMP accumulation and a weakly potent agonist of CB2R-dependent cAMP accumulation. ∆ 9 -THCa produced anti-nociceptive and hypolocomotive effects at 3 and 10 mg/kg, and anxiolytic-like effects at 10 mg/kg. ∆ 9 -THCa has previously been shown to bind murine cannabinoid receptors weakly 21 . Similar to our findings, Palomarés et al. 29 recently reported ∆ 9 -THCa binding to CB1R and CB2R, with potential orthosteric and allosteric activity at CB1R. In vivo, the anti-inflammatory activity of ∆ 9 -THCa in a rodent model of arthritis was CB1R-and peroxisome proliferatoractivated receptor γ (PPARγ)-dependent 29 . Similar neuroprotective and PPAR-dependent effects have been observed in rodent models of Huntington's disease 30 . Here, ∆ 9 -THCa displayed similar efficacy to ∆ 9 -THC in the cAMP inhibition assay but a lower affinity for CB1R in [ 3 H]CP55,940 competition experiments. These data corroborate the findings of Palomarés et al. 29 and indicate the antinociceptive effect of ∆ 9 -THCa observed here is likely in part due to CB1R agonism. Our data, together with previous studies, indicate ∆ 9 -THCa may mediate neuroprotective and anti-inflammatory actions via CB1R, CB2R, and PPARγ when administered at sufficiently high concentrations and in the absence of other cannabinoids.
THCV has been posited to act at both CB1R and CB2R, and is considered by some to behave as a CB1R antagonist and CB2R agonist [31][32][33] ; reviewed in 19 . Here, we found that THCV was able to displace [ 3 H]CP55,940 from both CB1R and CB2R. Unlike previous studies suggesting THCV is a CB1R antagonist, THCV did produce a weak partial agonist response in the CB1R cAMP inhibition assay. THCV was an agonist of CB2R-dependent cAMP inhibition, as described previously 34 and βarrestin2 recruitment. In vivo, THCV produced cataleptic (10 mg/kg), hypothermic (10 mg/kg), anti-nociceptive (3 and 10 mg/kg), hypolocomotive (3 and 10 mg/kg), and anxiolytic (3 and 10 mg/kg) effects. These effects are consistent with others' observations of anti-epileptic, hypolocomotive, neuroprotective effects in the range of 0.25-2.5 mg/kg 32,33 ; reviewed in 19 . Bolognini et al. 34 demonstrated that THCV is able to reduce hyperalgesia in mice via both CB1R and CB2R, because anti-hyperalgesic effects were limited by both the CB1R antagonist SR141716A and the CB2R antagonist SR144528. Given the in vitro efficacy of THCV at both CB1R and CB2R, it is likely that the in vivo effects of THCV observed here are dependent on both cannabinoid receptors, as has been described previously 34 .
CBD has been much more-extensively studied than the other phytocannabinoids. CBD has been described as a CB1R negative allosteric modulator 10  www.nature.com/scientificreports/ 1,36 ; 5HT1A, 5HT2A, adenosine 1A, and PPARγ partial agonist 16,[37][38][39] ; and an allosteric modulator of the µ-and δ-opioid receptors 4,9 . In this study, we included CBD in an assay format that would detect agonism at CB1R and CB2R. Not surprisingly, CBD displayed low affinity at both CB1R and CB2R and displayed minimal activity at both receptors. Similarly, we observed no effects of CBD between 1 and 10 mg/kg in the tetrad of assays, as expected. CBD has been shown in animal models of disease to have anti-convulsant, anti-inflammatory, and anti-nociceptive effects; all of which appear to be independent of CB1R and CB2R 19 . CBDa was inactive at CB1R and displayed weak partial agonism at CB2R in our cell culture assays. Navarro et al. 12 reported higher than anticipated affinity and activity of CBDa in both [ 3 H]CP55,940 binding for CB1R and CB2R in CHO cells and signaling in HEK-293T cells, which were similar to our findings. In vivo, CBDa produced significant hypolocomotive and anxiolytic effects in the OFT. Beyond CB1R and CB2R, CBDa has shown efficacy in reducing inflammatory pain and nausea 40,41 , seizure incidence 42 , and Parkinsonian signs in rodent models 13 . Accumulating data suggest the in vivo effects of CBDa are 5HT1A-mediated 40 . Therefore, the effects observed in this study in the tetrad maybe 5HT1A-mediated, rather than cannabinoid receptor-dependent.
CBDV displayed little to no activity at CB1R but did display demonstrable affinity and activity at CB2R. Similarly, Navarro et al. 12 observed that CBDV was nearly inactive at CB1R with greater activity at CB2R. However, the binding affinity of CBDV to CB2R was greater in our experiments compared to those of Navarro et al. 12 . This difference may be due to differing radioligand concentrations and/or differential expression of the receptor in our cell models. In our assays, CBDV displayed a bias toward the inhibition of cAMP relative to βarrestin2 recruitment; whereas previous reports in HEK-293T cells have described the opposite, although these bias analyses were conducted by different methods 12 . CBDV produced no significant in vivo responses in our assays. Earlier reports have shown anti-convulsant effects associated with CBDV treatment in rodent models and occur at 200-400 mg/kg p.o., doses higher than were tested here (reviewed in 19 ).
In CHO cells stably expressing human receptors, CBG was a weak partial agonist of both CB1R-and CB2Rdependent inhibition of cAMP that displaying a low affinity for CB1R and comparatively higher affinity for CB2R. Previous reports have similarly described CBG as a weak partial agonist of these two receptors, with low (> 1 µM) affinity for both [6][7][8]43,44 . Interestingly, previous observations have also indicated that CBG-dependent recruitment of βarrestin via the cannabinoid receptors is less potent and efficacious than other signal transduction pathways 12,44 , similar to our findings. Although our data indicate a higher affinity for CBG at CB2R than previous reports (500 nM vs. > 1 µM), all of our other observations are in concordance with previous findings 12, 44 . Whereas Navarro et al. 12 observed CBG to be a balanced, non-biased, ligand at CB1R and CB2R in HEK-293T cells; CBG displayed a bias for the inhibition of cAMP in CHO cells stably expressing CB1R or CB2R. In vivo, CBG produced a small but statistically significant anti-nociceptive effect at 3 mg/kg and an anxiolytic-like effect in the OFT at 10 mg/kg. CBG has previously been reported to have disease-ameliorating anti-inflammatory effects in mouse models of Huntington's disease, multiple sclerosis, Parkinson's disease, and amyotrophic lateral sclerosis [45][46][47][48][49] . These effects were likely mediated by PPARγ, and not cannabinoid receptors 43,48 49 . CBG has also been shown to have in vitro activity at GPR55, 5HT1A, the α2 adrenoceptor, and several transient receptor potential (TRP) channels [6][7][8] . Therefore, it is possible the in vivo anti-nociceptive and anxiolytic effects observed in our study occurred via these, cannabinoid receptor-independent, mechanisms.
Similar to CBG, CBC was a partial agonist of both CB1R-and CB2R-dependent signaling, with great selectivity and potency at CB2R relative to CB1R in the assays utilized here. Previous studies have observed weak partial agonism of CBC at both CB1R and CB2R 8 . More recently, Udoh et al. 50 reported that CBC produced CB2R-dependent membrane hyperpolarization in AtT20 cells. This effect was absent in CB1R-expressing AtT20 cells, indicating CB2R-specificity for CBC 50 . In vivo, CBC is able to increase neuronal viability via ERK phosphorylation in nestin-positive neural stem cells, but the receptor-specific mechanism has not been described for this effect 51 . CBC has also been shown to be anti-inflammatory and reduce hypermobility in a mouse model of gut inflammation, although these effects occur via TRPA1 and not cannabinoid receptors 52 . Similar to the findings of Izzo et al. 52 , CBC did produce a small, but statistically significant anti-nociceptive effect in the tail withdrawal assay and a hypolocomotive effect in the OFT; both of these effects may have occurred independently of cannabinoid receptors.
Based on these data, it is unclear whether the cannabinoids tested here bind an identical orthosteric site to that of ∆ 9 -THC or CP55,940. Future experiments utilizing site-directed mutagenesis will need to assess this question directly. Moreover, the partial agonist effects displayed by these ligands suggests they may be functionally antagonistic in the presence of higher agonist concentrations and in vivo. This functional antagonism has been previously demonstrated for ∆ 9 -THC itself when administered alongside full and potent CB1R agonists 10,53-56 . A growing body of literature also supports the notion that phytocannabinoids such as ∆ 9 -THCa, CBD, and others may be able to occupy both orthosteric and allosteric sites with varying affinity, further complicating our understanding of the cannabinoid receptors 10,15,29 . Future work assessing the potential antagonist activity of these compounds in the presence of a full agonist such as CP55,940 will be able to better-classify the mechanisms of action for these compounds beyond what has been done here.
This work represents an initial step into the assessment of the pharmacology for a subset of Cannabis-derived phytomolecules at the most-thoroughly studied cannabinoid receptors, CB1R and CB2R. It is possible that if the phytocannabinoids tested here that displayed CB1R agonist and in vivo activity were present at sufficiently high concentrations in cannabis products; they may produce intoxicating effects similar to those of ∆ 9 -THC. However, given the typically low content of these phytomolecules in cannabis products and their weak displacement of [ 3 H]CP55,940, these compounds are all probably more likely to diminish ∆ 9 -THC's effects in whole organisms. Beyond CB1R, ∆ 9 -THC itself is known to modulate the signaling of several proteins, including the orphan GPCR GPR55, and the TRP vanilloid 1 Ca 2+ channel (TRPV1) 17 . Other cannabinoids, such as CBD, modulate the activity of a wide array of cannabinoid and non-cannabinoid receptors, including CB1R as a negative allosteric modulator, CB2R and 5HT1A as a partial agonist, the µ-opioid receptor, and PPARs 17 . Our in vivo observations Scientific Reports | (2020) 10:20405 | https://doi.org/10.1038/s41598-020-77175-y www.nature.com/scientificreports/ that CBDa, CBG, and CBC (among other compounds tested) mediated changes in locomotion and time in the centre quadrant despite having little in vitro activity at CB1R, warrant further study to determine what other receptors are utilized by these ligands. Therefore, in order to comprehensively understand the poly-pharmacology of cannabinoid receptors in vivo, other receptor targets must be considered. Finally, cannabis products contain many phytomolecules that are co-administered when Cannabis is consumed. This study represents an initial foray into potential between-and among-phytomolecule interactions that can be built upon gradually. Pharmacology is a reductionist approach to biochemical interactions that cannot always model the complex interactions occurring in nature. As initial characterizations of single ligands are made, more complex combinatorial testing of pharmacology can be carried out. Eventually, we hope to assess potential interactive effects of complex cannabinoid mixtures and assess pharmacodynamic-and pharmacokinetic-differences that stem from chemically distinct ligands.
CHO cell membrane preparation and radioligand displacement assay. CHO-K1 hCB1R and hCB2R cells were disrupted by cavitation in a pressure cell, and membranes were sedimented by ultracentrifugation, as described by Bolognini et al. 34,58 . The pellet was resuspended in TME buffer (50 mM Tris-HCl, 5 mM MgCl 2 , 1 mM EDTA, pH 7.4) and membrane proteins were quantified with a Bradford dye-binding method (Bio-Rad Laboratories, Mississauga, ON).
Assays were carried out with [ 3 H]CP55,940 and Tris binding buffer (50 mM Tris-HCl, 50 mM Tris-base, 0.1% BSA, pH 7.4), total assay volume 2 mL, using the filtration procedure described previously by Baillie et al. 18 . The binding was initiated by the addition of transfected human CHO-K1 hCB1R and hCB2R cell membranes (50 µg protein per well). All assays were performed at 37 °C for 60 min before termination by the addition of ice-cold Tris binding buffer, followed by vacuum filtration using a 24-well sampling manifold (Brandel Cell Harvester; Brandel Inc, Gaithersburg, MD, USA) and Brandel GF/B filters that had been soaked in wash buffer at 4 °C for at least 24 h. Each reaction well was washed 6 times with 1.2 mL aliquots of Tris-binding buffer. The filters were air-dried overnight and then placed in 5 mL of scintillation fluid (Ultima Gold XR, PerkinElmer). Radioactivity was quantified by liquid scintillation spectrometry. Specific binding was defined as the difference between the binding that occurred in the presence and absence of 1 µM unlabelled CP55,940. The concentration of [ 3 H] CP55940 used in our displacement assays was 0.7 nM.
HitHunter cAMP assay. Inhibition of FSK-stimulated cAMP was determined using the DiscoveRx HitHunter assay in CHO-K1 hCB1R and hCB2R cells as we have described previously 15,23 . Briefly, cells (20,000 cells/well in low-volume 96 well plates) were incubated overnight in Opti-MEM containing 1% FBS at 37 °C and 5% CO 2 . Following this, Opti-MEM media was removed and replaced with cell assay buffer (DiscoveRx) and cells were co-treated at 37 °C with 10 µM FSK and ligands for 90 min. Following this, cAMP antibody solution and cAMP working detection solutions were added to cells according to the manufacturer's directions (Discov-eRx), and cells were incubated for 60 min at room temperature 15,23 . cAMP solution A was added according to the manufacturer's directions (DiscoveRx), and cells were incubated for an additional 60 min at room temperature before chemiluminescence was measured on a Cytation5 plate reader (top read, gain 200, integration time 10,000 ms).
PathHunter βarrestin2 assay. βarrestin2 recruitment was determined using the DiscoveRx PathHunter assay in CHO-K1 hCB1R and hCB2R cells as we have described previously 15,23 . Briefly, cells (20,000 cells/well in low-volume 96 well plates) were incubated overnight in Opti-MEM containing 1% FBS at 37 °C and 5% CO 2 . Following this, cells were treated at 37 °C with ligands for 90 min. Following this, the detection solution was added to cells according to the manufacturer's directions (DiscoveRx), and cells were incubated for 60 min at Animals and tetrad testing. Adult male C57BL/6 mice aged 6-12 weeks (mean weight 22 ± 0.3 g) were purchased from Charles River Labs (Senneville, QC). Animals were group housed (3 per cage) with ad libitum access to food, water, and environmental enrichment and maintained on a 12 h light/dark cycle. Mice were randomly assigned to receive i.p. injections of vehicle (1:1:18 ethanol:emulphor:saline) or 0.1-100 mg/kg cannabinoid (n ≥ 6 per group). All protocols were in accordance with the guidelines detailed by the Canadian Council on Animal Care 59,60 and approved by the Animal Research Ethics Board and the Scientific Merit Review Committee for Animal Behaviour at the University of Saskatchewan. In keeping with the ARRIVE guidelines, power analyses were conducted to determine the minimum number of animals required for the study and animals were purchased-rather than bred-to limit animal waste, and all assessments of animal behaviour were made by individuals blinded to treatment group 60 . Catalepsy was assessed in the ring holding assay 10 min following injection. The mice were placed such that their forepaws clasped a 5 mm ring positioned 5 cm above the surface of the testing space. The length of time the ring was held was recorded (seconds). The trial was ended if the mouse turned its head or body, made 3 consecutive escape attempts, or at 60 s of immobility (i.e. MPE = 60 s).
Internal body temperature was measured via rectal thermometer 12 min following injection. Anti-nociception was determined by assessing tail flick latency 15 min following injection. Mice were restrained with their tails placed ~ 1 cm into 52 °C water and the time until the tail was removed was recorded as tail flick latency (sec  11 . In order to estimate bias for compounds with an EC 50 > 10,000 nM, EC 50 was set to 10,000 nM and E max was set to the maximum observed response. In vivo data are presented as % MPE for catalepsy (MPE = 60 s) and anti-nociception (MPE = 20 s), °C for body temperature, and distance travelled (m) and % time in the centre quadrant in 5 min for OFT. Statistical analyses were conducted by one-way analysis of variance (ANOVA), as indicated in the figure legends, using GraphPad. Post-hoc analyses were performed using Tukey's (one-way ANOVA) test. Homogeneity of variance was confirmed using Bartlett's test. All data were evaluated for possible outliers using Grubb's test in GraphPad. No outliers were removed. All results are reported as the mean ± the standard error of the mean (SEM) or 95% confidence interval (CI), as indicated. p values < 0.05 were considered to be significant.