Evaluating signaling bias for synthetic cannabinoid receptor agonists at the cannabinoid CB₂ receptor

Abstract

The rapid structural evolution and emergence of novel synthetic cannabinoid receptor agonists (SCRAs) in the recreational market remains a key public health concern. Despite representing one of the largest classes of new psychoactive substances, pharmacological data on new SCRAs is limited, particularly at the cannabinoid CB₂ receptor (CB₂). Hence, the current study aimed to characterize the molecular pharmacology of a structurally diverse panel of SCRAs at CB₂, including 4‐cyano MPP‐BUT7AICA, 4F‐MDMB‐BUTINACA, AMB‐FUBINACA, JWH‐018, MDMB‐4en‐PINACA, and XLR‐11. The activity of SCRAs was assessed in a battery of in vitro assays in CB₂‐expressing HEK 293 cells: G protein activation (Gα_i3 and Gα_oB), phosphorylation of ERK1/2, and β‐arrestin 1/2 translocation. The activity profiles of the ligands were further evaluated using the operational analysis to identify ligand bias. All SCRAs activated the CB₂ signaling pathways in a concentration‐dependent manner, although with varying potencies and efficacies. Despite the detection of numerous instances of statistically significant bias, compound activities generally appeared only subtly distinct in comparison with the reference ligand, CP55940. In contrast, the phytocannabinoid THC exhibited an activity profile distinct from the SCRAs; most notably in the translocation of β‐arrestins. These findings demonstrate that CB₂ is able to accommodate a structurally diverse array of SCRAs to generate canonical agonist activity. Further research is required to elucidate whether the activation of CB₂ contributes to the toxicity of these compounds.

Abbreviations

AUC

area‐under‐the‐curve

BRET

Bioluminescence resonance energy transfer

BSA

bovine serum albumin

CB₁

cannabinoid CB₁ receptor

CB₂

cannabinoid CB₂ receptor

ERK

extracellular signal‐regulated kinase

HA

haemagglutinin

PDL

poly‐D‐lysine

PEI

polyethylenimine

SCRA

synthetic cannabinoid receptor agonist

THC

(−)‐trans‐Δ⁹‐ tetrahydrocannabinol

1. INTRODUCTION

The application of natural or synthetically derived xenobiotics to manipulate specific physiological functions has revolutionized the discipline of medicine. However, the psychoactivity of many compounds, such as amphetamines, benzodiazepines, cannabinoids, and opioids, has encouraged their diversion to recreational use in order to evoke “pleasurable” effects. One of the most rapidly proliferating classes of New Psychoactive Substances (NPS) remains synthetic cannabinoid receptor agonists (SCRAs), which compromised 25% of the new compounds reported to the United Nations Office of Drugs and Crime in 2020. ¹ SCRAs initially presented as undetectable or unregulated alternatives to cannabis, mimicking the psychoactive effects of the phytocannabinoid Δ⁹‐tetrahydrocannabinol (THC). Concerns surrounding SCRA consumption mounted with increasing reports of severe adverse effects and fatalities, ² , ³ , ⁴ , ⁵ , ⁶ with over 60 deaths being associated with the SCRA AMB‐FUBINACA in New Zealand between May 2017 and February 2019. ⁷ Governments worldwide have implemented legislative restrictions and drug screening for various SCRAs, however, these are often thwarted by evolving structural diversification and the development of novel compounds.

The effects of SCRAs are predominantly attributed to activation of the cannabinoid CB₁ and CB₂ receptors, expressed primarily in the central nervous system and immune cells, respectively. ⁸ , ⁹ The cannabimimetic activity of numerous SCRAs have now been characterized, with the majority demonstrating potent and efficacious agonism at both CB₁ and CB₂ in vitro. ¹⁰ , ¹¹ Given the chemical heterogeneity of SCRAs, it has been posited that particular compounds may stabilize unique receptor conformations that elicit preferential activation of specific signaling pathways, referred to as biased agonism or functional selectivity. Recently, we have shown that a series of SCRAs were relatively balanced (unbiased) full agonists at CB₁ when compared to WIN55,212‐2, in contrast to the partial efficacy of THC. ¹² , ¹³ Other studies have corroborated these findings, with very few instances of compelling SCRA bias presented. ¹⁴ , ¹⁵ , ¹⁶

Despite this growing body of research, a clear mechanism of toxicity has not yet been established for SCRAs. Studies thus far have mainly focused on the pharmacology of SCRAs at CB₁, as this receptor is considered primarily responsible for the psychoactive effects of cannabinoids in humans. However, the extensive array of behavioral and physiological effects encountered upon SCRA abuse, including cardiac, gastro‐intestinal, and respiratory symptoms, suggests the involvement of other targets. ¹⁷ , ¹⁸ Many SCRAs possess an affinity for CB₂ that is equivalent or higher than for CB₁, questioning the involvement of this receptor in SCRA toxicity. ¹⁴ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ As novel SCRAs with a dearth of pharmacological data continue to emerge, it remains crucial to comprehensively characterize the in vitro activity of these compounds in order to identify potential molecular links to toxicity. Currently, only a handful of studies have explored SCRA bias at CB₂, ¹⁴ , ²⁵ which may also allude to the unique side‐effect profile of these compounds. This study therefore set out to examine the molecular pharmacology of a structurally diverse panel of SCRAs (Figure 1) at CB₂. The SCRAs selected were either newly emerging, possessed distinct chemical features, and/or had been previously associated with cases of human toxicity. The activity of the compounds was assessed using several in vitro assays, including G protein dissociation, ERK phosphorylation (pERK), and β‐arrestin translocation, with the resulting signaling profiles compared to evaluate ligand bias and potential structure–activity patterns.

FIGURE 1.

Chemical structures of THC and synthetic cannabinoid receptor agonists investigated.

2. MATERIALS AND METHODS

2.1. Drugs

4‐cyano CUMYL‐BUTINACA (4CN‐CUMYL‐BUTINACA), AMB‐FUBINACA, CP55940, HU‐308, phorbol 12‐myristate 13‐acetate (PMA), and XLR‐11 were purchased from Cayman Chemical Company; and (−)‐trans‐Δ⁹‐tetrahydrocannabinol (THC) was purchased from Toronto Research Chemicals Inc. Remaining SCRAs were synthesized and provided by Dr Samuel Banister (University of Sydney, Australia). Drugs were dissolved in absolute ethanol (CP55940, THC) or DMSO (HU‐308, other SCRAs) and stored as single‐use aliquots at −80°C to minimize solvent evaporation and chemical degradation from freeze–thaw cycles. Serial dilutions of drugs were vehicle‐controlled, which was maintained at a constant level (0.1%) for all experiments.

2.2. Cell culture

HEK 293 (RRID:CVCL_0045) wild‐type cells were transfected with human CB₂ (hCB₂) possessing the 63R single‐nucleotide polymorphism and chimerized with three haemagglutinin (HA) tags (3HA‐hCB₂ 63R). The sequence‐verified, linearized 3HA‐hCB₂ 63R pEF‐V4‐HisA (pEF4a) plasmid was transfected with Lipofectamine™ 2000. Following antibiotic selection, a mixed population of cells was isolated by flow cytometry. Briefly, cells were detached with Versene and labeled with mouse anti‐HA.11 IgG clone 16B12 (1:500; BioLegend; Cat# 901503; RRID:AB_2565005) and secondary Alexa Fluor^® 488 goat anti‐mouse IgG (1:300; Invitrogen, Thermo Fisher Scientific; Cat# A11029; RRID:AB_2534088). Following secondary antibody incubation, cells were washed and resuspended in HEPES‐buffered (10 mM) phenol red‐free DMEM (Gibco, Thermo Fisher Scientific) supplemented with 1 mg.mL⁻¹ bovine serum albumin (BSA; ICPBio), to achieve a concentration of approximately 2.5 × 10⁶ mL⁻¹. FACS was performed using a BD FACSAria™ Fusion Flow Cytometer (BD Biosciences) to obtain a pool of cells with high cell surface receptor expression. The resulting mixed culture is referred to as HEK 293 3HA‐hCB₂ 63R.

HEK 293 wild‐type and HEK 293 3HA‐hCB₂ 63R cells were cultured in high glucose DMEM (Gibco, Thermo Fisher Scientific) supplemented with 10% FBS (New Zealand‐origin, Moregate Biotech), and incubated at 37°C in a humidified atmosphere containing 5% CO₂.

2.3. G protein dissociation assay

Bioluminescence resonance energy transfer (BRET) assays to measure G protein dissociation were performed in accordance with Olsen et al. (2020), ²⁶ with minor modifications. In brief, HEK 293 wild‐type cells were seeded at an appropriate density in 10 cm dishes to achieve a confluency of 40%–50% for transfection and grown overnight. Culture medium was replaced, and transfection mixtures were prepared in Opti‐MEM reduced serum medium (ThermoFisher Scientific). Transfection mixtures contained 1 μg Gα‐RLuc8 pcDNA5/FRT/TO, 1 μg Gβ₃ pcDNA3.1, 1 μg Gγ‐GFP2 pcDNA3.1, and 1 μg pplss‐3HA‐hCB₂ 63R pEF4a, with a total mass of 4 μg (refer to Olsen et al. (2020) ²⁶ for G protein construct designs, Table 1 for TRUPATH plasmid combinations used and Patel et al. (2022) ²⁷ for CB₂ plasmid details). Plasmids were combined in a 1:9 ratio (DNA:PEI) with PEI MAX (1 μg.μL⁻¹; Polysciences), incubated at room temperature for 20 min, and then added dropwise to cells.

TABLE 1.

TRUPATH G protein plasmid combinations used for transfection.

Gα	Gβ	Gγ
Gαi1‐RLuc8 pcDNA5/FRT/TO	Gβ3 pcDNA3.1	Gγ9‐GFP2 pcDNA3.1
Gαi2‐RLuc8 pcDNA5/FRT/TO	Gβ3 pcDNA3.1	Gγ8‐GFP2 pcDNA3.1
Gαi3‐RLuc8 pcDNA5/FRT/TO	Gβ3 pcDNA3.1	Gγ9‐GFP2 pcDNA3.1
GαoA‐RLuc8 pcDNA5/FRT/TO	Gβ3 pcDNA3.1	Gγ8‐GFP2 pcDNA3.1
GαoB‐RLuc8 pcDNA5/FRT/TO	Gβ3 pcDNA3.1	Gγ8‐GFP2 pcDNA3.1

Following overnight culture, transfected cells were trypsinized and seeded at a density of 50 000 cells/well in poly‐D‐lysine (0.05 mg.mL⁻¹, PDL; Sigma Aldrich) coated, white 96‐well CulturPlate™ (PerkinElmer) plates. Cells were cultured overnight prior to detection. For assays, cells were washed with PBS and equilibrated in HEPES‐buffered (25 mM) phenol red‐free DMEM supplemented with 1 mg.mL⁻¹ fatty acid‐free BSA (“assay medium”) for approximately 30 min. Cells were then treated with 5 μM coelenterazine 400a (NanoLight Technologies) and luminescence at 410 nm and 515 nm was simultaneously detected for 5 min at 37°C in a LUMIstar^® Omega luminometer (BMG Labtech). Serial dilutions of drugs were then added, and luminescence was read for approximately 25 min at 37°C. BRET2 ratios (515 nm/ 410 nm) were exported from Omega MARS software and further analyzed in GraphPad Prism v9. For concentration‐response analysis, average BRET ratios for the vehicle conditions were subtracted from all drug responses and net area under the curve (AUC) obtained, with responses normalized to 10 μM CP55940.

2.4. ERK phosphorylation assay

Phosphorylation of extracellular signal‐related kinases (ERK) was assessed using an AlphaLISA^® SureFire^® Ultra™ p‐ERK1/2 (Thr202/Tyr204) kit (PerkinElmer). HEK 293 3HA‐hCB₂ 63R or HEK 293 wild‐type cells were plated at a density of 35 000 cells/well in PDL‐coated, Costar clear 96‐well culture plates (Corning^®) and incubated for 4–6 h. The culture medium was then replaced, and the cells were serum starved for at least 16 h in serum‐free DMEM supplemented with 1 mg.mL⁻¹ BSA. For experiments involving pertussis toxin (PTX; Sigma Aldrich) pretreatment to irreversibly inactivate Gα_i/o proteins, appropriate dilutions of PTX or PTX vehicle (50% glycerol, 50 nM Tris–HCl pH 7.5, 50 mM Na₂HPO₄, 500 mM NaCl) were prepared in serum‐free DMEM supplemented with 1 mg.mL⁻¹ BSA to achieve a final concentration of 100 ng.mL⁻¹ PTX. Cells were cultured in PTX or PTX vehicle overnight for at least 16 h prior to stimulation. Drugs, prepared in serum‐free medium supplemented with 1 mg.mL⁻¹ BSA, were applied to cells, with plates barely submerged in a 37°C water bath and stimulated for 4 min, which correlated to peak ERK phosphorylation (data not shown). At the conclusion of drug stimulation, plates were immediately placed on ice, medium aspirated, and cells lysed with 30 μL lysis buffer. AlphaLISA detection was subsequently performed in accordance with manufacturer instructions, and plates were read in a CLARIOstar^® plate reader (BMG Labtech). Data were normalized to the vehicle and matched 1 μM CP55940 treatments to remove day‐to‐day variability.

2.5. β‐arrestin translocation assay

β‐arrestin translocation BRET assays were performed as previously described in Finlay et al. (2019). ¹² HEK 293 wild‐type cells were seeded in 10 cm dishes to achieve a confluency of 40%–50% for transfection. Following overnight culture, culture medium was replaced and transfection mixtures were prepared, containing 2 μg mem‐Linker‐Citrine‐SH3 pcDNA3.1+, 50 ng Rluc8‐β‐arrestin pcDNA3.1+ (human β‐arrestin 1 or 2), 1.6 μg pplss‐3HA‐hCB₂ 63R pEF4a and 350 ng empty pcDNA3.1+, with a total mass of 4 μg. Plasmids were appropriately diluted in Opti‐MEM reduced serum medium (ThermoFisher Scientific) and combined in a 1:9 ratio (DNA:PEI) with PEI MAX (1 μg.μL⁻¹; Polysciences) and then incubated at room temperature for 20 min before the dropwise addition to cells. Following overnight incubation, transfected cells were lifted and plated at a density of 50 000–60 000 cells/well in PDL‐coated, white 96‐well CulturPlate™ plates (PerkinElmer), and cultured overnight. For assays, cells were washed with PBS and equilibrated in assay medium for approximately 30 min. Coelenterazine‐h (5 μM, NanoLight Technologies) was dispensed onto cells and incubated in a LUMIstar^® Omega luminometer (BMG Labtech, Ortenberg, Germany) for 5 min at 37°C, with luminescence at 475 nm and 535 nm simultaneously detected to establish a baseline BRET ratio. Serial dilutions of drugs were dispensed, and luminescence was detected for approximately 25 min. BRET ratios (535 nm/ 475 nm) were exported from Omega MARS software and further analyzed in GraphPad Prism v9. Raw BRET ratios were normalized to the average BRET values of the predrug (coelenterazine‐h) incubation and net AUC obtained, with data normalized to 10 μM CP55940 to generate concentration‐response curves.

2.6. Data and statistical analysis

Data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology. ²⁸ Blinding was unfeasible for all experimental and data analysis procedures as all assays were performed in 96‐well plates. For all assays, the ligand order and positions of the ligand concentrations used were randomized between experiments to avoid the potential influence of plate position.

Bias analysis was performed as described by Van Der Westhuizen et al. (2014), ²⁹ with modifications by Zhu et al. (2018). ³⁰ Briefly, LogR values for each independent (biological) replicate were obtained for each pathway in GraphPad Prism, with either CP55940 or a high efficacy SCRA designated the “pseudo” full agonist, and all other ligands considered partial agonists. CP55940 displayed full efficacy in the majority of pathways investigated and was therefore selected as the reference ligand. Average ΔLogR and ΔΔLogR values were determined following normalizations, in accordance with Van Der Westhuizen et al. (2014). ²⁹

To minimize unwanted sources of variability, data were normalized to matched responses produced by either CP55940 (1 μM or 10 μM) for SCRA experiments or PMA (100 nM) for pERK PTX experiments. Data presented are either representative data from a single experiment (performed in technical duplicate or triplicate, expressed as mean ± SD) or averaged (combined) data from at least five independent (biological) replicates (expressed as mean ± SEM). ³¹ Statistical analyses were performed only on collated data from independent biological replicates with n ≥ 5 using GraphPad v9. The Shapiro–Wilk test for normality and Brown‐Forsythe test for equality of variance were performed to ensure data sets were appropriate for analysis with parametric statistical tests. Statistical analyses were performed using one‐way ANOVA, unless otherwise specified. Data sets that indicated statistically significant differences (p < .05) were further analyzed with the Holm‐Šídák post‐hoc multiple comparisons test where F achieved statistical significance in ANOVA. All statistical analyses were performed using GraphPad Prism v9.

3. RESULTS

3.1. SCRAs induce submaximal activation of Gα_i3 and Gα_oB

To provide an initial assessment of agonist behavior, activation of the canonical G protein effectors of CB₂, Gα_i, and Gα_o, was characterized using real‐time BRET assays (Table 2). The selection of Gα_i3 and Gα_oB subtypes was based on relative assay performance within the Gα_i/o class (Figure S1), originally optimized by Olsen et al. (2020). ²⁶ The characterization of these two pathways sufficiently captures the canonical G protein activation profile of CB₂—an exhaustive G protein subtype characterization (including other Gα_i/o isoforms and Gα_z) was not the aim of this study.

TABLE 2.

Potencies and efficacies of G protein dissociation for Gα_i3 and Gα_oB at CB₂ for cannabinoid receptor agonists ^a .

	Gαi3		GαoB		Fold‐shift in EC50 (Gαi3/GαoB)
	pEC50 (M)	EMAX (%)	pEC50 (M)	EMAX (%)
CP55940	7.47 (0.02)	99.66 (0.51)	7.28 (0.05)	101.30 (0.47)	0.65
THC	6.88 (0.08)	47.95 (1.66)*	6.68 (0.08)	44.23 (0.41)*	0.63
JWH‐018	8.14 (0.05)	80.55 (1.22)*	7.81 (0.02)	82.16 (1.87)*	0.47
HU‐308	7.20 (0.05)	79.88 (0.63)*	6.75 (0.05)	77.06 (4.17)*	0.35
4CN‐CUMYL‐BUTINACA	8.34 (0.02)	78.19 (1.50)*	8.01 (0.02)	79.95 (1.85)*	0.47
4CN‐MPP‐BUT7AICA	7.46 (0.02)	80.72 (1.43)*	7.08 (0.04)	81.00 (2.11)*	0.42
4F‐MDMB‐BUTINACA	8.71 (0.07)	95.87 (1.00)*	8.67 (0.07)	96.72 (1.38)	0.91
5F‐AB‐P7AICA	6.69 (0.02)	79.69 (0.82)*	6.33 (0.02)	77.86 (1.52)*	0.44
5F‐ADB	8.32 (0.05)	92.33 (1.26)*	8.29 (0.05)	92.56 (1.12)*	0.93
5F‐CUMYL‐P7AICA	7.37 (0.03)	75.19 (1.95)*	7.01 (0.05)	75.62 (1.87)*	0.44
5F‐CUMYL‐PEGACLONE	8.29 (0.05)	81.11 (1.14)*	8.10 (0.06)	89.13 (2.69)*	0.65
5F‐MDMB‐PICA	8.47 (0.06)	91.94 (0.29)*	8.35 (0.04)	92.05 (1.76)*	0.76
AMB‐FUBINACA	8.34 (0.02)	85.76 (1.12)*	8.23 (0.02)	85.19 (1.12)*	0.78
MDMB‐4en‐PINACA	8.00 (0.04)	95.60 (0.75)*	7.95 (0.06)	97.52 (2.24)	0.89
MDMB‐CHMICA	8.15 (0.05)	92.08 (1.16)*	8.04 (0.07)	92.18 (2.52)*	0.78
XLR‐11	8.18 (0.10)	92.66 (1.23)*	8.00 (0.09)	92.29 (1.87)*	0.66

Note: Statistical significance for E_MAX determined using one‐way ANOVA, followed by Holm‐Šídák post‐hoc multiple comparisons test performed in GraphPad Prism, with p values indicated as * <.05 when compared to CP55940.

^a Data shown are means (± SEM) from five independent biological replicates performed in HEK 293 cells, with E_MAX presented as percent of 10 μM CP55940 (100%).

As expected, CP55940, THC, JWH‐018, and HU‐308 stimulated concentration‐dependent dissociation of Gα_i/o from the Gβγ subunits, signifying G protein activation. THC exhibited significantly lower efficacy in both pathways, producing only ~50% of the response generated by CP55940, congruent with the partial agonist classification of the ligand. ³² , ³³ Interestingly, all SCRAs also elicited significantly less Gα_i/o protein dissociation than CP55940 (between ~75% and 96%), with the exception of MDMB‐4en‐PINACA and 4F‐MDMB‐BUTINACA in the Gα_oB pathway. Variations in potency were observed among the SCRAs, with 4F‐MDMB‐BUTINACA the most potent (~2 nM) and 5F‐AB‐P7AICA the least potent (~0.3 μM) in both pathways. While there was a strong correlation in ligand potencies between Gα_i and Gα_o (R ² = .98), the potency was generally higher at Gα_i3 than Gα_oB (Figure S2).

3.2. SCRAs induce robust phosphorylation of ERK

To explore the downstream signaling consequences of G protein activation, phosphorylation of ERK1/2 was assessed. In constrast to G protein activation, the majority of the SCRAs induced phosphroylation of ERK to equivalent extents as CP55940 (Table 3). Notably, MDMB‐4en‐PINACA was significantly more efficacious than CP55940 in this pathway, whereas THC stimulated pERK with the least efficacy of the compounds tested. THC and MDMB‐4en‐PINACA were the least potent among the compounds investigated, with 4CN‐CUMYL‐BUTINACA possessing the highest potency. Cannabinoid‐induced phosphorylation of ERK was abolished by pretreatment with PTX (Figure S3; one‐way ANVOA compared with vehicle, p > .05), with no consistent non‐CB₂‐mediated effects observed (Figure S4; one‐way ANVOA compared to vehicle, p > .05).

TABLE 3.

Potencies and efficacies for phosphorylation of ERK at CB₂ for cannabinoid receptor agonists ^a .

	pEC50 (M)	EMAX (%)
CP55940	7.09 (0.09)	129.86 (1.74)
THC	6.60 (0.14)	72.70 (7.74)*
JWH‐018	7.71 (0.05)	99.06 (10.05)
HU‐308	7.01 (0.06)	144.92 (9.81)
4CN‐CUMYL‐BUTINACA	8.00 (0.12)	112.20 (6.40)
4CN‐MPP‐BUT7AICA	7.81 (0.05)	124.24 (8.16)
4F‐MDMB‐BUTINACA	7.50 (0.07)	150.34 (12.10)
5F‐AB‐P7AICA	7.22 (0.18)	121.75 (13.01)
5F‐ADB	7.08 (0.09)	169.88 (15.22)
5F‐CUMYL‐P7AICA	7.91 (0.09)	94.03 (12.46)
5F‐CUMYL‐PEGACLONE	7.66 (0.07)	102.26 (6.19)
5F‐MDMB‐PICA	7.45 (0.07)	119.55 (14.29)
AMB‐FUBINACA	7.40 (0.06)	122.97 (12.07)
MDMB‐4en‐PINACA	6.85 (0.07)	194.14 (21.60)*
MDMB‐CHMICA	7.04 (0.10)	128.13 (13.83)
XLR‐11	7.32 (0.08)	109.74 (7.61)

Note: Statistical significance for E_MAX was determined using one‐way ANOVA, followed by Holm‐Šídák post‐hoc multiple comparisons test performed in GraphPad Prism, with p values indicated as * < .05 when compared to CP55940.

^a Data shown are means (± SEM) from five independent biological replicates performed in HEK 293 3HA‐hCB₂ 63R cells, with E_MAX presented as percent of 1 μM CP55940 (100%).

3.3. SCRAs elicit differential translocation of β‐arrestins to CB₂

The influence of SCRAs on CB₂ signaling and regulation was further evaluated in human β‐arrestin 1 (arrestin 2) and 2 (arrestin 3) translocation BRET assays (Table 4). The synthetic cannbinoid ligands screened displayed highly variable agonist activity in the translocation of β‐arrestin 1 and 2 to CB₂. Most SCRAs were significantly less efficacious in both β‐arrestin pathways relative to CP55940, with the exception of MDMB‐4en‐PINACA, MDMB‐CHMICA, and XLR‐11. Potencies of the synthetic cannabinoids were largely comparable to CP55940, ranging from 7.59 nM to 372 nM for β‐arrestin 1 and 5.37 nM to 229 nM for β‐arrestin 2, with 4CN‐CUMYL‐BUTINACA and 5F‐AB‐P7AICA possessing the highest and lowest potencies in both pathways, respectively. For all ligands, translocation of β‐arrestin 2 was slightly more potent than β‐arrestin 1, although differences were less than three‐fold, with a strong correlation observed between the two pathways (R ² = .98, Figure S2), consistent with our findings at CB₁. ¹³ Notably, 4CN‐CUMYL‐BUTINACA revealed a distinct activity profile, attaining the highest potency and smallest efficacy in the translocation of both β‐arrestins among the SCRAs. Conversely, THC failed to stimulate the translocation of either β‐arrestin with sufficient potency and efficacy to reliably obtain E_MAX model parameters. Consistent non‐CB₂‐mediated effects were observed for >10 μM THC (Figure S5), which was likely non‐receptor‐mediated, in line with our previous findings. ¹³ , ³⁴

TABLE 4.

Potencies and efficacies for β‐arrestin 1 and 2 translocation to CB₂ for cannabinoid receptor agonists ^a .

	β‐arrestin 1		β‐arrestin 2		Fold‐shift in EC50 (β‐arrestin 1/β‐arrestin 2)
	pEC50 (M)	EMAX (%)	pEC50 (M)	EMAX (%)
CP55940	7.16 (0.06)	103.91 (1.75)	7.28 (0.04)	103.78 (0.90)	1.32
THC	–	2.34 (1.58) b	–	5.91 (0.76) b	–
JWH‐018	7.38 (0.06)	81.62 (5.08)*	7.53 (0.07)	85.45 (2.30)*	1.41
HU‐308	6.70 (0.04)	92.73 (3.05)*	6.97 (0.03)	91.36 (2.20)*	1.86
4CN‐CUMYL‐BUTINACA	8.12 (0.11)	43.33 (1.20)*	8.27 (0.02)	56.22 (1.23)*	1.41
4CN‐MPP‐BUT7AICA	7.03 (0.07)	70.79 (1.67)*	7.35 (0.02)	78.04 (1.95)*	2.09
4F‐MDMB‐BUTINACA	8.05 (0.03)	82.49 (1.16)*	8.16 (0.05)	92.82 (2.07)*	1.29
5F‐AB‐P7AICA	6.43 (0.06)	65.08 (3.18)*	6.64 (0.04)	77.25 (1.64)*	1.62
5F‐ADB	7.55 (0.05)	81.62 (3.57)*	7.62 (0.06)	93.14 (1.76)*	1.17
5F‐CUMYL‐P7AICA	7.02 (0.10)	46.22 (2.39)*	7.32 (0.05)	61.08 (2.05)*	1.20
5F‐CUMYL‐PEGACLONE	7.78 (0.04)	50.37 (2.28)*	8.03 (0.05)	62.15 (1.50)*	1.78
5F‐MDMB‐PICA	7.65 (0.06)	79.88 (2.57)*	7.91 (0.04)	88.49 (1.38)*	1.82
AMB‐FUBINACA	7.61 (0.02)	63.79 (4.15)*	7.82 (0.02)	75.21 (1.32)*	1.62
MDMB‐4en‐PINACA	7.16 (0.06)	109.15 (4.40)	7.37 (0.04)	104.40 (3.96)	1.62
MDMB‐CHMICA	7.32 (0.03)	98.70 (1.35)	7.50 (0.03)	100.16 (3.05)	1.51
XLR‐11	7.32 (0.06)	94.50 (3.99)	7.59 (0.05)	96.17 (2.70)*	1.86

Note: Statistical significance for E_MAX was determined using one‐way ANOVA, followed by Holm‐Šídák post‐hoc multiple comparisons test performed in GraphPad Prism, with p values indicated as * < .05 when compared to CP55940.

^a Data shown are means (± SEM) from at least five independent biological replicates performed in HEK 293 cells, with E_MAX presented as percent of 10 μM CP55940 (100%).

^b Response at 1 μM THC due to non‐CB₂‐mediated effects at high concentrations.

3.4. SCRAs display unbiased but submaximal activity at CB₂

To provide an intital qualitative assessment of ligand bias, normalized concentration–response curves were compared for each ligand (Figure 2). The rank order of potency was generally well conserved across the compounds, with Gα_i3 and Gα_oB protein dissociation more potent than the translocation of both β‐arrestin isoforms. Pathway potencies were relatively higher for 4CN‐CUMYL‐BUTINACA and 4F‐MDMB‐BUTINACA, and lower for 5F‐AB‐P7AICA across all pathways. Notably, the potency rank order for pERK differed, with THC, HU‐308, 4CN‐CUMYL‐BUTINACA, 4‐cyano MPP‐BUT7AICA (4CN‐MPP‐BUT7AICA), 5F‐AB‐P7AICA, and 5F‐CUMYL‐P7AICA possessing potencies for pERK that were greater than G protein dissociation, contrasting with the remaining compounds that demonstrated similar or reduced potency. Accordingly, the correlation between Gα_i3 protein dissociation and ERK phosphorylation was relatively weak (R ² = .11, Figure S2), which differed from the other pathways showing a strong correlation with Gα_i3 protein dissociation (R ² > .8). Overall, the SCRAs displayed a greater variation in pathway potencies, compared with the narrow range of potencies observed for CP55940. The majority of SCRAs elicited submaximal efficacy in the pathways examined when compared to CP55940. However, MDMB‐4en‐PINACA exhibited similar or higher efficacy in the Gα_oB, pERK, β‐arrestin 1, and β‐arrestin 2 pathways. Interestingly, THC produced a vastly different activity profile to the synthetic cannabinoids, inducing only partial responses for G protein dissociation and pERK, with no measurable β‐arrestin translocation, similar to our previous findings for THC at CB₁. ¹² , ¹³

FIGURE 2.

Concentration–response curves for five signaling pathways at CB₂ in response to CP55940, THC, and a series of synthetic cannabinoid receptor agonists. Data shown are mean ± SD from a representative experiment performed in technical duplicates or triplicates. All data normalized to either 10 μM (G protein and β‐arrestin) or 1 μM (pERK) CP55940 (100%).

To fully interrogate the divergences in activity between the SCRAs, formal quantitative analysis for functional selectivity was performed utilizing the operational model for bias, with methods described by Van Der Westhuizen et al. (2014) ²⁹ and incorporating later modifications ³⁰ (Table 5). CP55940 was selected as the reference ligand, demonstrating comparable potency and full efficacy in the signaling pathways examined. ΔLogR values (LogR relative to CP55940 in each signaling pathway) within each ligand were compared between the signaling pathways and evaluated for any statistically significant differences, which would be indicative of signaling bias (Figure 3). All synthetic cannabinoids showed bias relative to the pattern of signaling elicited by CP55940. JWH‐018, HU‐308, 4CN‐CUMYL‐BUTINACA, 4CN‐MPP‐BUT7AICA, 5F‐AB‐P7AICA, and 5F‐CUMYL‐P7AICA showed consistent bias towards pERK, and away from G protein dissociation and/or β‐arrestin translocation. Whereas 4F‐MDMB‐BUTINACA, 5F‐ADB, 5F‐MDMB‐PICA, AMB‐FUBINACA, MDMB‐4en‐PINACA, MDMB‐CHMICA, and XLR‐11 demonstrated bias towards G protein dissociation, and away from pERK and/or β‐arrestin translocation. Interestingly, 4CN‐CUMYL‐BUTINACA possessed the most balanced activity profile of the SCRAs, with significant bias only towards pERK and away from β‐arrestin 1 translocation.

TABLE 5.

Operational analysis results for cannabinoid agonists, displaying ΔΔLogR for all possible pathway comparisons at CB₂ ^a .

	Gαi3 − GαoB	Gαi3 − β‐arrestin 1	Gαi3 − β‐arrestin 2	GαoB − β‐arrestin 1	GαoB − β‐arrestin 2	β‐arrestin 1 − β‐arrestin 2	Gαi3 − pERK	GαoB − pERK	β‐arrestin 1 − pERK	β‐arrestin 2 − pERK
CP55940	0.00 ± 0.07	0.00 ± 0.09	0.00 ± 0.07	0.00 ± 0.11	0.00 ± 0.09	0.00 ± 0.10	0.00 ± 0.13	0.00 ± 0.14	0.00 ± 0.15	0.00 ± 0.14
THC	0.05 ± 0.13	–	–	–	–	–	−0.26 ± 0.21	−0.30 ± 0.21	–	–
JWH‐018	0.14 ± 0.07	0.59 ± 0.08	0.41 ± 0.10	0.45 ± 0.08	0.27 ± 0.10	−0.18 ± 0.10	−0.06 ± 0.10	−0.19 ± 0.11	−0.64 ± 0.11	−0.47 ± 0.12
HU‐308	0.29 ± 0.11	0.25 ± 0.09	−0.03 ± 0.07	−0.05 ± 0.12	−0.33 ± 0.10	−0.28 ± 0.09	−0.54 ± 0.15	−0.84 ± 0.17	−0.79 ± 0.16	−0.51 ± 0.15
4CN‐MPP‐BUT7AICA	0.18 ± 0.07	0.32 ± 0.09	−0.06 ± 0.06	0.14 ± 0.10	−0.23 ± 0.08	−0.37 ± 0.10	−0.94 ± 0.12	−1.12 ± 0.13	−1.26 ± 0.15	−0.89 ± 0.13
4CN‐CUMYL‐BUTINACA	0.14 ± 0.06	0.30 ± 0.14	0.04 ± 0.06	0.16 ± 0.14	−0.10 ± 0.08	−0.25 ± 0.14	−0.23 ± 0.16	−0.37 ± 0.17	−0.52 ± 0.21	−0.27 ± 0.17
4F‐MDMB‐BUTINACA	−0.15 ± 0.11	0.55 ± 0.10	0.39 ± 0.10	0.70 ± 0.10	0.54 ± 0.11	−0.16 ± 0.09	0.59 ± 0.13	0.74 ± 0.13	0.05 ± 0.12	0.21 ± 0.12
5F‐AB‐P7AICA	0.18 ± 0.06	0.19 ± 0.07	−0.11 ± 0.07	0.01 ± 0.09	−0.30 ± 0.08	−0.30 ± 0.09	−1.11 ± 0.17	−1.29 ± 0.18	−1.30 ± 0.18	−0.10 ± 0.18
5F‐ADB	−0.16 ± 0.09	0.64 ± 0.09	0.50 ± 0.10	0.80 ± 0.10	0.66 ± 0.11	−0.14 ± 0.10	0.57 ± 0.15	0.72 ± 0.15	−0.08 ± 0.15	0.06 ± 0.16
5F‐CUMYL‐P7AICA	0.17 ± 0.08	0.40 ± 0.12	−0.03 ± 0.08	0.23 ± 0.14	−0.20 ± 0.11	−0.43 ± 0.13	−1.04 ± 0.13	−1.22 ± 0.14	−1.45 ± 0.16	−1.01 ± 0.14
5F‐CUMYL‐PEGACLONE	−0.03 ± 0.09	0.56 ± 0.09	0.19 ± 0.08	0.59 ± 0.10	0.22 ± 0.09	−0.37 ± 0.09	0.10 ± 0.12	0.13 ± 0.14	−0.46 ± 0.13	−0.08 ± 0.13
5F‐MDMB‐PICA	−0.07 ± 0.09	0.72 ± 0.10	0.40 ± 0.09	0.79 ± 0.11	0.47 ± 0.09	−0.32 ± 0.10	0.50 ± 0.12	0.56 ± 0.12	−0.22 ± 0.13	0.09 ± 0.12
AMB‐FUBINACA	−0.07 ± 0.06	0.70 ± 0.06	0.40 ± 0.06	0.77 ± 0.08	0.47 ± 0.07	−0.30 ± 0.07	0.38 ± 0.11	0.46 ± 0.12	−0.31 ± 0.12	−0.01 ± 0.11
MDMB‐4en‐PINACA	−0.14 ± 0.09	0.64 ± 0.10	0.43 ± 0.08	0.78 ± 0.12	0.57 ± 0.09	−0.21 ± 0.11	0.43 ± 0.12	0.57 ± 0.14	−0.21 ± 0.14	0.00 ± 0.13
MDMB‐CHMICA	−0.08 ± 0.09	0.63 ± 0.08	0.44 ± 0.07	0.70 ± 0.10	0.52 ± 0.09	−0.19 ± 0.08	0.56 ± 0.12	0.64 ± 0.13	−0.07 ± 0.12	0.12 ± 0.12
XLR‐11	−0.01 ± 0.14	0.67 ± 0.11	0.40 ± 0.12	0.68 ± 0.12	0.40 ± 0.12	−0.27 ± 0.09	0.37 ± 0.15	0.37 ± 0.15	−0.30 ± 0.13	−0.03 ± 0.13

Note: ΔΔLogR values unable to be determined are denoted with “–” due to an insufficient response to accurately fit curves for at least one pathway.

^a Data shown are means ± SEM (n = 5) with CP55940 as the reference ligand. Positive ΔΔLogR values suggest bias toward the first pathway and away from the second pathway, whereas negative ΔΔLogR values suggest bias toward the second pathway and away from the first pathway.

FIGURE 3.

Comparison of ΔLogR values, demonstrating evidence of bias between signaling pathways relative to the reference ligand, CP55940. LogR values for five independent (biological) replicates were obtained for each pathway in GraphPad Prism. ΔLogR was determined by subtracting the average LogR value for each ligand from the average LogR of the reference ligand, with error determined from the average SEM of LogR for each ligand, in accordance with Van Der Westhuizen et al. (2014). ²⁹ Therefore, ΔLogR is a single value, and thus, cannot be presented as a scatterplot. Positive ΔLogR values indicate a greater propensity/efficiency for activating the specific pathway compared to CP55940, whereas negative ΔLogR values indicate a reduced propensity relative to CP55940. Statistical significance was determined using one‐way ANOVA, followed by Holm‐Šídák post‐hoc multiple comparisons test, with p values indicated as *<.05.

4. DISCUSSION AND CONCLUSIONS

The expansion of SCRAs over the last decade has prompted pharmacological evaluation of these novel compounds in an attempt to elucidate mechanisms underpinning their extensive toxicological profile. The cannabinoid receptors actively modulate several physio‐pathological processes, rendering CB₁ and CB₂ likely contributors to SCRA‐mediated toxicity. In the present study, a diverse panel of SCRAs were systematically characterized at CB₂ using live cell‐based assays, with all compounds demonstrating agonistic behavior in the pathways examined, albeit with varying potencies and efficacies.

To provide an initial pharmacological assessment of the SCRAs, activation of Gα_i/o, the canonical signaling phenotype of CB₂, was investigated. The SCRAs elicited concentration‐dependent activation of Gα_i3 and Gα_oB subtypes, indicative of agonistic properties at CB₂, congruent with previous findings on structurally related SCRAs. ¹⁴ , ²⁴ Notably, the majority of the SCRAs induced G protein activation to a slightly lesser extent than CP55940, suggesting partial agonism. The capacity to measure minute differences in efficacy highlights the improved sensitivity of the G protein dissociation assay compared with traditional second messenger assays like cAMP accumulation/inhibition, which are typically impeded by signal amplification and high levels of receptor reserve that confound the ability to detect partial agonism. ¹⁴ , ³⁵

CB₂‐mediated signaling further downstream of G protein activation was subsequently interrogated by assessment of ERK1/2 phosphorylation. Contrary to G protein activation, the SCRAs stimulated ERK phosphorylation to equivalent extents as CP55940, with THC exhibiting partial efficacy, in line with previous reports. ²⁵ The full agonist activity of the SCRAs, particularly the significantly larger response of MDMB‐4en‐PINACA, may reflect combined G protein‐ and β‐arrestin‐mediated activation of pERK. Nogueras‐Ortiz et al. (2017) ³⁶ used siRNA knockdown to conclude activation of the G protein‐independent pERK signal by internalized CB₂ was downstream of β‐arrestin 1. Similarly, β‐arrestin 1 putatively facilitates a late‐phase ERK1/2 response at CB₁. ³⁷ , ³⁸ , ³⁹ However, ERK phosphorylation was found to be entirely Gα_i/o dependent in this study (Figure S3), suggesting the higher efficacy may be the result of signal amplification of this pathway.

Potencies for ligands activating pERK were generally right‐shifted in comparison with G protein activation, implying the pathway is less efficiently activated, similar to the relationship observed at CB₁. ¹² This conflicts with the proposed greater receptor reserve and signal amplification in the ERK pathway, which is situated downstream of G protein activation. Perhaps, the overexpression of G proteins in the G protein dissociation assays may have contributed to the improved efficiency of this pathway compared with the pERK assay, which relied solely on endogenous G protein levels. However, 4CN‐MPP‐BUT7AICA, 5F‐AB‐P7AICA, and 5F‐CUMYL‐P7AICA demonstrated the opposite effect, with potencies greater for pERK than G protein activation, despite the efficacies for pERK remaining relatively comparable with the other SCRAs. This suggests the 7‐azaindole core of the aforementioned compounds may better stabilize a receptor conformation that is conducive to ERK phosphorylation. It should be noted that this purported confirmation failed to improve the efficiency of G protein activation, despite pERK being Gα_i/o‐mediated in this study, highlighting a more complex interplay between receptor conformation, G protein activation, and downstream signaling.

The effect of the SCRAs on receptor regulation was also evaluated using β‐arrestin translocation assays. The submaximal response observed for many of the SCRAs suggests a poor ability of these compounds to facilitate the translocation of β‐arrestin to CB₂. This conflicts with reports demonstrating full and supramaximal efficacy in CB₂ β‐arrestin 2 recruitment assays for structurally related SCRAs, specifically for 4F‐MDMB‐BUTINACA, 5F‐MDMB‐PICA, and AMB‐FUBINACA. ¹⁴ , ²³ , ⁴⁰ , ⁴¹ Discrepancies in assay design, cell line, and data normalization may contribute to these contrasting results. In particular, the β‐arrestin assay employed by other studies typically utilize modified receptors and β‐arrestins, which may alter the interaction between the proteins. The partial β‐arrestin efficacy in the current study also contrasts with our findings at CB₁, where SCRAs induced potent and efficacious translocation of both β‐arrestin isoforms under the same assay system, which suggests SCRAs are less efficient at recruiting β‐arrestin to CB₂. ¹² , ¹³ However, similar to CB₁, no measurable β‐arrestin translocation was detected for THC at CB₂, which may be attributed to diminished sensitivity of our assay compared with assays utilizing tagged receptors. ¹² , ¹³ , ³⁴ Taken together, our data support that the unique structural elements and configuration of SCRAs facilitate improved β‐arrestin recruitment at the cannabinoid receptors, in comparison with THC. ¹² , ¹³ , ³⁴

The highly variable activity profiles of the SCRAs at CB₂ were suggestive of biased signaling, which was quantitatively assessed using the operational analysis. The majority of significant biases detected involved bias toward G protein activation and away from ERK phosphorylation. Strikingly, HU‐308, 4CN‐MPP‐BUT7AICA, 5F‐AB‐P7AICA, and 5F‐CUMYL‐P7AICA showed a clear bias toward pERK, reflecting the relatively higher potency in this pathway for these compounds. In contrast, 4F‐MDMB‐BUTINACA, 5F‐ADB, and MDMB‐4en‐PINACA demonstrated no bias towards ERK phosphorylation, despite displaying greater efficacy than CP55940. This alludes to the relative insensitivity of the operational analysis model to changes in efficacy compared to potency. ⁴² In this study, many SCRAs retained similar relative potencies to CP55940, despite significant variations in efficacy, although the rank order of both metrics was largely preserved across the pathways. Together, this may have afforded the comparatively small ΔΔlogR values (most <1.0), suggesting the overall signaling profiles of the SCRAs were relatively balanced in comparison with CP55940, and it is unclear whether such small changes would manifest as physiologically relevant ligand bias.

A prominent finding in the current study is the stark difference in the activity profile of THC from the SCRAs, coinciding with our previous studies at CB₁. ¹² , ¹³ Specifically, the absence of measurable β‐arrestin translocation for THC may imply bias away from this pathway, which contrasted with the robust translocation obtained with the SCRAs. The β‐arrestin pathway has been associated with adverse effects, such as respiratory depression and constipation at the μ‐opioid receptor, ⁴³ , ⁴⁴ which has encouraged the development G protein‐biased ligands with improved safety profiles. ⁴⁵ , ⁴⁶ Consequently, β‐arrestin activation has been posited as a potential mechanism of SCRA‐induced toxicity at the cannabinoid receptors. ¹² , ¹³ However, recent reports have questioned the relationship between β‐arrestin recruitment and opioid‐mediated adverse effects, ⁴⁷ , ⁴⁸ with the low toxicity of novel, purportedly G protein‐biased opioid ligands being attributed to reduced intrinsic efficacy. ⁴⁹ , ⁵⁰ Indeed, the poor intrinsic efficacy of THC at both cannabinoid receptors (confirmed for CB₂ in this study), may account for the overall low toxicity of the compound. This supports structural studies on the cannabinoid receptors that propose the dibenzopyran ring of THC enables greater conformational variability, which weakens the interaction with key toggle switch residues important for receptor activation, thus affording the partial agonism of the ligand. ⁵¹ This differs from the rigid fused‐bicyclic heteroaromatic structures contained within SCRAs, which strongly stabilize toggle switch residues to form a Gα_i‐interacting cavity, potentially promoting sustained receptor activation and subsequent recruitment of β‐arrestins. ⁵¹ , ⁵² , ⁵³ However, SCRAs associated with incidences of clinical toxicity, such as XLR‐11 and 5F‐CUMYL‐P7AICA, possessed lower efficacy than other SCRAs at several CB₁‐mediated pathways, ¹³ suggesting intrinsic efficacy alone is insufficient to predict toxicity. However, inadequate data surrounding market availability, prevalence, dose, as well as the clinical reporting of intoxications and deaths, complicates the ability to make reliable inferences on the relative toxicity of SCRAs. ⁵⁴ Regardless of whether the adverse effects are attributed to bias or low efficacy, the contribution of G protein and β‐arrestin signaling to the toxicity of SCRAs should be further explored through studies utilizing knockout methodology to dissect the functional/physiological consequences of these pathways.

A unique structural element featured within the panel of SCRAs assessed was the pentenyl group of MDMB‐4en‐PINACA. The compound was consistently associated with higher efficacy across all novel SCRAs examined, particularly in ERK1/2 phosphorylation, even compared to its fluorinated analog 4F‐MDMB‐BUTINACA, indicating the pentenyl tail was well tolerated. Interestingly, the tail region has been suggested as a key determinant of ligand affinity, as opposed to receptor activation, by forming hydrophobic interactions within a conserved docking region or ‘narrow side pocket’ of CB₁ and CB₂. ⁵¹ , ⁵³ Previous studies support this hypothesis, with pentenyl‐substituted SCRAs possessing affinities and potencies similar to alternative tail moieties at both of the cannabinoid receptors. ²⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ The retention of reasonable cannabimimetic activity may partially explain the increasing prevalence of novel SCRAs incorporating an olefinic tail. ⁵⁸ , ⁵⁹ , ⁶⁰ , ⁶¹ While discussion of structural‐activity relationships is limited in this study, detailed characterization of a diverse array of SCRAs has highlighted the capacity of CB₂ to accommodate a diverse range of structural features, analogous to CB₁. ¹⁰ , ⁶²

The majority of studies on SCRAs thus far have focused on CB₁ due to the receptor's predominant location in the brain and consequent psychoactive effects. Accordingly, the physiological endpoints typically assessed in SCRA in vivo studies are common psychoactive/toxicological effects induced by cannabinoids, including hypothermia, catalepsy, convulsant activity and antinociception, which have been found to be non‐CB₂ mediated. ¹⁹ , ²⁰ , ²⁵ , ⁶³ , ⁶⁴ This highlights the need to expand the endpoints examined, as this study has confirmed the robust activity of SCRAs at CB₂, which most likely contributes, in some degree, to the extensive physiological effects observed in vivo. Increasing evidence demonstrates CB₂ expression in microglia and neurons, which may be upregulated in pathological conditions, including neurodegenerative diseases, neuroinflammation and drug addiction, ⁶⁵ and these sites may contribute to the on‐target toxicity of SCRAs. Of note, the majority of CB₁ agonists, such as SCRAs, have poor selectivity, enhancing the difficulty in probing the specific contributions of CB₁ and CB₂ to system/translational endpoints with this class of compounds. Despite this, the inducible nature of CB₂ during particular conditions renders it an attractive therapeutic target. Indeed, several CB₂‐selective agonists that have undergone clinical evaluation are well tolerated, suggesting activation of the receptor may have limited contribution to the more pronounced, severe adverse effects of SCRAs. ⁶² However, the translation of in vitro activity to in vivo effects remains a fundamental limitation to resolving the toxicity of SCRAs and potential therapeutic exploits of this class of drug. Systematic structural comparison of SCRAs for both in vitro and in vivo activity has thus far failed to reveal a clear link between molecular pharmacology and toxicity, even in cases of significant bias, ¹⁴ , ¹⁹ , ²⁰ , ²¹ , ²⁴ , ⁴¹ , ⁶³ , ⁶⁴ , ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ although this does not rule out the involvement of CB₁ and CB₂ in the toxicity of SCRAs. Off‐target activity is also being explored as contributors to the complex side‐effect profile of SCRAs, with confirmed interactions with the 5‐HT (namely subtypes 1A, 2B, 2C, and 6), GABA_A , GPR18, GPR55 and muscarinic acetylcholine receptors, ⁷¹ , ⁷² , ⁷³ , ⁷⁴ and T‐type calcium channels. ¹⁷ , ⁷⁵ Additional considerations in the enhanced toxicity of SCRAs include the formation of toxic thermal degradants ⁷⁶ , ⁷⁷ and active metabolites. ⁷⁸ , ⁷⁹ , ⁸⁰

In summary, this study has established a chemically diverse array of SCRAs are able to stimulate CB₂ with sufficient efficacy and potency in multiple signaling pathways to potentially elicit activity in vivo. The pharmacological profile of SCRAs differed from THC, most notably in the translocation of β‐arrestin, a finding consistent with our previous work at CB₁. Further study is required to delineate the physiological effects of signaling pathways regulated by CB₂, which is crucial to understanding their contribution to the effects of SCRAs. With increasing pharmacodynamic knowledge, the ability to predict toxicity based on the structural features of SCRAs may become possible.

AUTHOR CONTRIBUTIONS

MP designed and performed experiments, analyzed data, and wrote the paper. NG contributed to aspects of experimental design. SDB synthesized and supplied compounds used in the study. DBF and MG oversaw all elements of the project, obtained funding, and reviewed drafts of the paper. All authors reviewed and approved the final version of the paper.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

None.

Supporting information

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Patel M, Grimsey NL, Banister SD, Finlay DB, Glass M. Evaluating signaling bias for synthetic cannabinoid receptor agonists at the cannabinoid CB₂ receptor. Pharmacol Res Perspect. 2023;11:e01157. doi: 10.1002/prp2.1157

DATA AVAILABILITY STATEMENT

The data that supports the findings of this study are available from the corresponding author upon reasonable request.