Dimethylindanoylindole Isomers Revealed Different Pharmacological Profiles at CB1 and CB2 Cannabinoid Receptors

Abstract

Synthetic cannabinoid receptor agonists (SCRAs) exhibit higher efficacy and potency at CB1 and CB2 cannabinoid receptors compared to Δ⁹-tetrahydrocannabinol, contributing to adverse effects associated with their growing prevalence as novel psychoactive substances. The indole-based SCRAs GBD-002 and GBD-003, incorporating 2,2-dimethylindane, differ by a single substituent position yet display markedly different receptor binding profiles. We evaluated both the GBD-002/GBD-003 pair and an analogous structural isomer pair JWH-018/JWH-018 2′-naphthyl using bioluminescence resonance energy transfer (BRET) assays to assess key CB1 and CB2 transducer pathways, including Gα_i1 engagement, β-arrestin 2 recruitment, and cAMP inhibition. Site-directed mutagenesis characterized contributions of nonconserved residues between CB1 and CB2 as well as aromatic residues on CB1 transmembrane helix 2 (TM2). This study provided insights into the molecular determinants of cannabinoid receptor selectivity and efficacy.

Introduction

The endocannabinoid system (ECS) is a crucial neuromodulatory network that regulates diverse neurological and neuropsychiatric processes, including mood, movement, reward, appetite, and pain perception. − It has become an important target for therapeutic development in conditions such as obesity and chronic pain. ,, The ECS signals primarily through two G protein-coupled receptors (GPCRs): CB1 and CB2, among others. , Both receptors couple to inhibitory Gα_i/o proteins and undergo kinase-mediated phosphorylation, which recruits β-arrestin and leads to receptor desensitization and internalization.

CB1 is one of the most abundant GPCRs in the brain and plays a central role in modulating neural activity with its effects varying by neuronal subtype and brain region. − In contrast, although CB2 expression is also reported in the brain, it is predominantly expressed in the peripheral immune system, where it acts as a key regulator of immune and inflammatory responses. , CB1 and CB2 form the core signaling components of the ECS.

Synthetic cannabinoid receptor agonists (SCRAs), often conceptualized as comprising head, core, and tail moieties, were originally developed as research tools to probe ECS signaling and assess pharmacological potential. However, many early SCRAs were diverted for illicit use, likely due to their synthetic accessibility and initial lack of regulation. , These compounds bind to CB1 and CB2 as well as other receptors and are associated with adverse effects, particularly through CB1 activation. − For example, the naphthoylindole JWH-018 and various analogues emerged in smokable synthetic cannabinoid (SSC) products such as “K2” and “Spice”, prompting regulatory controls that render these compounds impractical for continued use in legitimate pharmacological research.

We recently developed a series of novel 1,3-disubstituted indole-based cannabinoids featuring a more structurally complex and synthetically challenging 2,2-dimethylindanoylindole scaffold with the goal of improving accessibility for academic research of the ECS. , Among these, regioisomers GBD-002 and GBD-003 display markedly different receptor-binding profiles. GBD-002 exhibits little selectivity between CB1 and CB2, while GBD-003 is CB2-selective, highlighting molecular interactions that may underlie receptor subtype discrimination.

CB1 and CB2 are seven-transmembrane (TM) receptors that share a high degree of sequence homology, especially within the TM regions that form the orthosteric binding site. This similarity may be why many cannabinoids, including Δ⁹-tetrahydrocannabinol (Δ⁹-THC), can bind both receptors with similar affinity. , While recent advances in cryo-electron microscopy have yielded high-resolution structures of both receptors in active and inactive states, these static snapshots offer limited insight into the dynamic mechanisms that underlie receptor selectivity and function. − Understanding the molecular determinants of receptor selectivity, particularly in cases such as GBD-002 and GBD-003, where regioisomers exhibit markedly different pharmacological profiles, can provide insight into how subtle chemical features are differentially recognized by CB1 and CB2.

We hypothesized that nonconserved residues within the orthosteric binding pocket may influence both functional potency and efficacy by differential interactions with ligand headgroup positioning. We investigated the structure–activity relationships (SARs) of GBD-002 and GBD-003 at CB1 and CB2 and used site-directed mutagenesis to characterize key binding site residues that contribute to the CB2 selectivity exhibited by GBD-003 relative to its structural isomer GBD-002.

Materials and Methods

Cell Culture

Human embryonic kidney 293T (HEK-293T) cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Corning) supplemented with 10% fetal bovine serum, 2 mM l-glutamine (Corning), and 1% penicillin–streptomycin (Corning). Cells were maintained in 10 cm plates at 37 °C and 5% CO₂. For experiments, 3 × 10⁶ cells were seeded in 10 cm plates and allowed to reach confluence over 24 h prior to transfection. Following overnight transfection (16 h), the medium was replaced with fresh DMEM. Experiments were performed 48 h post-transfection.

Bioluminescence Resonance Energy Transfer (BRET) Assays

HEK-293T cells were transiently transfected with cDNA (10–15 μg per assay) using linear polyethylenimine hydrochloride (PEI Max , Polysciences) at a 1:2 ratio (30 μg). The plasmid weight and transfection ratio were optimized and maintained across BRET assays. The transfection reagent–plasmid mixture was added dropwise to the medium and incubated overnight.

Five BRET assays were used in this experiment: Gα_i1 engagement, β-arrestin 2 recruitment, Gα_i1 activation, Gα_oA activation, and CAMYEL. For Gα_i1 engagement, Renilla luciferase 8 (RLuc) fused receptor and Venus fused Gα_i1 served as a resonance energy transfer (RET) pair. Gβ₁ and Gγ₂ constructs were cotransfected to avoid depletion of endogenous Gβγ. The β-arrestin 2 recruitment assay utilized an RLuc fused receptor and a β-arrestin 2 fused Venus as a RET pair. To enhance receptor phosphorylation and facilitate β-arrestin 2 recruitment, GRK2 constructs were cotransfected. The Gα_i1 and Gα_oA activation assay used RLuc fused Gα and Venus fused Gγ₂ as a RET pair. Receptor and Gβ₁ constructs were cotransfected. The CAMYEL (cAMP sensor using YFP-Epac-RLuc) biosensor is a single construct BRET sensor in which Rluc and a circularly permuted citrine flank an inactive Epac1 domain, serving as the RET pair. cAMP binding induces a conformational change that separates the donor and acceptor, decreasing the BRET ratio. Receptor constructs were cotransfected.

Bioluminescence Resonance Energy Transfer (BRET) Experiment

48 h post-transfection, cells were washed with PBS, lifted using 5 mM EDTA (Fisher Scientific), collected, and resuspended in PBS containing 0.1% glucose, 0.1% BSA, and 200 μM sodium bisulfite. Cells were distributed in white U-bottom 96-well plates, and ligands were added as agonists or antagonists. Agonists were transferred as 10-fold serial dilution in triplicate. Antagonists were preincubated with cells at 1 μM for 10 min before the addition of the agonist CP55940. Cells were incubated with 5 μM luciferase substrate (coelenterazine H) for 2.5 min before agonist transfer. For CAMYEL experiments, 10 μM forskolin and agonist were coadministered following the 2.5 min coelenterazine H incubation. BRET signals were recorded 10 min after ligand treatment using the PheraStar FSX plate reader. G protein engagement and dissociation were assessed by BRET ratios (530/480 nm) from RLuc and Venus-tagged proteins. Each condition was performed in four to five biological replicates.

Compounds and Plasmid Constructs

GBD-002 and GBD-003 compounds were prepared as described previously. All other cannabinoid ligands were obtained from Cayman Chemical (Ann Arbor, Michigan) or the NIDA Drug Supply Program (Rockville, Maryland) and dissolved at 10 mM in DMSO. DNA constructs were subcloned in pcDNA3 vectors. Site-directed mutagenesis was performed on plasmid constructs encoding CNR1 and CNR2 using the Quickchange Site Directed Mutagenesis Kit (Agilent) with the following primers: for CB1 mutagenesis, L193I (forward: 5′-TGTTT­CTGTT­CAAAA­TCGGTGG­GGTC­ACGG-3′, reverse: 5′-CCGTGA­CCCC­ACCG­ATTTT­GAACA­GAAA­CA-3′), I267L (forward: 5′-CTGT­TTGC­TCAGA­CCTTT­TCCCA­CACA­TTG-3′, reverse: 5′-CAATG­TGT­GGG­AAAAGG­TCTGA­GCA­AACAG-3′), L359V (forward: 5′-TCTGC­TGGGG­CCCT­GTGCTT­GCAAT­CATGG-3′, reverse: 5′-CCAT­GATT­GCAAG­CACA­GGGCCCC­AGCAGA-3′), F170A (forward: 5′-TGGG­GAGTGT­CATTGC­TGTCTA­CAGCT­TCAT-3′, reverse: 5′-ATGAA­GCTGT­AGAC­AGCAA­TGACACT­CCCCCA-3′), F174A (forward: 5′-ACAGC­TTCATT­GACGCCC­ACGTGTT­CCACC-3′, reverse: 5′-GGTGGA­ACACGT­GGGCGT­CAATG­AAGCTGT-3′), F177A (forward: 5′-ACAGC­TTCAT­TGACG­CCCACG­TGTTC­CACCG-3′, reverse: 5′-CGGT­GGAA­CACGT­GGGC­GTCAA­TGAAG­CTGT-3′), and H178A (forward: 5′-GCTT­CATTG­ACTTC­GCCGT­GTTCC­ACCGCA-3′, reverse: 5′-TGCG­GTGGAA­CACGGCG­AAGTCA­ATGAA­GC-3′); for CB2 mutagenesis, I110L (forward: 5′-TCT­TCCTG­CTGAAG­CTTGGCA­GCGTGA­CTA-3′, reverse: 5′-TAGTC­ACGC­TGCCA­AGCTT­CAGCAG­GAAGA-3′), L182I (forward: 5′-GGC­CCTGC­TCTGAGA­TTTTCCC­ACTGA­TCC-3′, reverse: 5′-GGAT­CAGTG­GGAAAA­TCTCAG­AGCAG­GGCC-3′), and V261L (forward: 5′-TCT­GTTGG­TTCCCAC­TGCTGG­CCCTC­ATGG-3′, reverse: 5′-CCATG­AGGG­CCAGCA­GTGGG­AACCAA­CAGA-3′). All mutations were confirmed by GENEWIZ and Azenta Life Sciences.

Flow Cytometry

Cell surface expression of wild-type and mutant CB1 and CB2 receptor constructs were assessed by flow cytometry. , HEK293T cells (∼600,000 cells) were plated in 6-well plates, transfected with 2 μg of myc-tagged RLuc-fused receptor construct, and collected 48 h after transfection. Cells were lifted using 5 mM EDTA, washed with PBS, and resuspended in cold FACS buffer (PBS containing 1% BSA). Samples were then stained with APC-conjugated anti-Myc antibody (Abcam, AB223896; 1:1000 dilution) in FACS buffer for 1 h, washed, and resuspended for acquisition. A total of 20,000 events per sample were recorded on a Beckman Coulter CytoFLEX S Cell Analyzer. Median fluorescence intensity (MFI) values were normalized to the corresponding wild-type receptor, and comparisons between mutant and wild-type constructs were performed using repeated measures one-way ANOVA with Dunnett’s post hoc test. All data were processed in FlowJo and GraphPad Prism v10.

Statistical Analysis

BRET ratios (530/480 nm) were used to generate dose–response curves and analyzed for ANOVA one-way statistical comparison in GraphPad Prism v10. All data were normalized to maximal and minimal BRET signal ratio of WT reference agonist CP55940. Specifically, E _max values for all ligands and mutant receptors were normalized to the maximal response of CP55940 at the wild-type receptor within each assay (e.g., Gα_i1 engagement, β-arrestin 2 recruitment). This normalization allows direct comparison of relative efficacy across ligands and mutant receptors within and between assays. Nonlinear regression was performed on compiled data to generate E _max and pEC₅₀.

Molecular Modeling

The binding modes of the GBD-002 and -003 within CB1 and CB2 were determined by a hybrid docking/MD simulation approach previously described [PMID: 32997485]. The molecular graphics were rendered from previously reported results [PMID: 33887913].

Results

Isomers with an Analogous Position in a Headgroup Display Similar Trends in Functional Efficacy across CB1 and CB2

We previously demonstrated that the two dimethyl-indane regioisomers, GBD-002 and GBD-003, exhibit distinct binding profiles. In binding assays, GBD-002 showed little to no selectivity between CB1 and CB2 receptors (CB1 K _i = 1.46 ± 0.49 nM; CB2 K _i = 1.36 ± 0.52 nM; SI = 1.1), whereas GBD-003 displayed CB2 selectivity (CB1 K _i = 38.8 ± 19.8 nM; CB2 K _i = 4.9 ± 0.2 nM; SI = 7.9). We then investigated how the orientation of GBD-003’s headgroup influences functional properties of G protein and β-arrestin recruitment.

To that end, we examined both the GBD-002/GBD-003 pair and an additional structural isomer pair with a different headgroup but analogous positional variation: JWH-018 and JWH-018 2′-naphthyl isomer (JWH-018 isomer) (Figure I). This allowed us to assess whether the general trend that headgroup attachment position influences receptor efficacy and selectivity, extending beyond the dimethylindane scaffold. While the JWH pair differs from the GBD pair in chemical composition and steric volume, its inclusion serves as a comparator pair to evaluate positional effects rather than to establish equivalent structure–activity relationships. The efficacy and potency of all four ligands were evaluated at CB1 and CB2 receptors using the bioluminescence resonance energy transfer assay (BRET)-based biosensors to monitor (1) receptor interaction with the Gα_i1 subunit (Gα_i1 engagement), (2) receptor recruitment of β-arrestin 2 (β-arrestin 2 recruitment), (3) Gα_i1 interaction with Gβγ (Gα_i1 activation), (4) Gα_oA interaction with Gβγ (Gα_oA activation), and (5) cAMP inhibition measurement via the CAMYEL biosensor. These modes of assays provide a direct measurement of Gα_i1 and β-arrestin 2 coupling and are not influenced by off target receptor activation.

1.

Dose response of dimethyl-indane regioisomers and naphthoylindole isomers at CB1 and CB2 receptors. BRET concentration–response curves for CP55940, GBD-002, GBD-003, JWH-018, and JWH-018 2′-naphthyl isomer measuring (A) CB1 Gα_i1 engagement and (B) preincubation with vehicle, 1 μM GBD-003, or JWH-018 isomer followed by addition of CP55940 for CB1 Gα_i1 engagement; (C) CB1 β-arrestin 2 recruitment and (D) corresponding preincubation; (E) CB2 Gα_i1 engagement and (F) corresponding preincubation; and (G) CB2 β-arrestin 2 recruitment and (H) corresponding preincubation. (I) Chemical structures of all tested compounds are shown, with head, core, and tail moieties labeled on GBD-002. Data presented as mean ± SEM of n = 4 experiments.

In the CB1 Gα_i1 engagement assay, GBD-002 demonstrated approximately 60% higher efficacy and an increase in potency compared to its isomer GBD-003 (Figure A, Table ). JWH-018 showed greater efficacy and potency than its isomer but with a smaller difference compared to the GBD isomer pair. These isomeric trends were consistent in the CB1 β-arrestin 2 recruitment assay (Figure C, Table ). Similar trends were observed in the CB1 Gα_i1 activation and Gα_oA activation assays (Figure S1, Table S1).

1. Efficacy and Potency of CB1 and CB2 Gα_i1 Engagement and β-Arrestin 2 Recruitment Assays ^,

	CB1		CB2
	Gαi1 Engagement	β-Arrestin 2 Recruitment	Gαi1 Engagement	β-Arrestin 2 Recruitment
	E max (%)
CP55940	99.4 ± 3.0	99.8 ± 2.1	99.8 ± 2.7	99.7 ± 2.7
GBD-002	116.4 ± 4.9****	163.3 ± 6.6****	78.7 ± 8.5	91.8 ± 7.3####
GBD-003	55.1 ± 5.4	53.7 ± 7.7	75.3 ± 5.2	80.8 ± 5.4##
JWH-018	139.6 ± 5.6**	167.8 ± 4.7****	73.5 ± 3.9####	75.1 ± 4.4####
JWH-018 isomer	110.7 ± 5.0	98.4 ± 5.3	84.5 ± 6.5#	91.8 ± 4.6
	pEC50 (M)
CP55940	8.03 ± 0.09	7.78 ± 0.06	8.76 ± 0.10#	8.13 ± 0.09
GBD-002	7.16 ± 0.10****	6.20 ± 0.08***	5.86 ± 0.20##	5.92 ± 0.14
GBD-003	5.89 ± 0.16	5.44 ± 0.19	6.35 ± 0.21	6.29 ± 0.14#
JWH-018	7.61 ± 0.12*	6.96 ± 0.07*	8.42 ± 0.19**	7.68 ± 0.16
JWH-018 isomer	6.99 ± 0.11	6.43 ± 0.12	7.35 ± 0.23	7.40 ± 0.15##

^aMean ± SEM values for E _max and pEC₅₀ are reported. E _max is normalized against CP55940.

^bGBD-002 and JWH-018 have *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 denoting differences relative to their respective isomers, GBD-003 and JWH-018 naphthyl, by one-way ANOVA followed by Šidák’s post hoc test.

^{c #} p < 0.05, ^## p < 0.01, and ^#### p < 0.0001 denote differences between CB1 and CB2 for each compound within a given assay determined by two-way ANOVA followed by Šidák’s post hoc test.

In contrast, the CB2 Gα_i1 engagement assay revealed no significant differences between GBD-002 and GBD-003, though GBD-003 trended toward 3-fold higher potency (Figure E, Table ). Compared to CB1, GBD-002 exhibited reduced efficacy and potency at CB2, whereas GBD-003 showed increased activity, supporting its CB2 selectivity. These findings were mirrored in the CB2 β-arrestin 2 recruitment assay, where both GBD compounds displayed similar efficacies and potencies with no significant differences (Figure G, Table ).

The JWH isomers showed a general trend of reduced efficacy but increased potency in both CB2 Gα_i1 engagement and β-arrestin 2 recruitment assays compared to the CB1 assays (Figure E, G, Table ). Notably, JWH-018 had significantly higher potency than the JWH-018 isomer in the CB2 Gα_i1 engagement assay but did not differ in potency in the β-arrestin 2 assay.

To further investigate the increased efficacy of GBD-003 in the CB2 assay compared to CB1, we performed a preincubation assay in which cells were first stimulated with vehicle, 1 μM, GBD-003, or 1 μM JWH-018 isomer followed by addition of the full agonist CP55940 (Figure B,D,F,H). This approach allowed us to assess the functional efficacy of partial agonists prior to being superseded by the high affinity and efficacy of CP55940. In this competitive context, the partial agonists (GBD-003 and JWH-018 isomer) act as antagonists, occupying the binding site and producing submaximal responses.

In both CB1 Gα_i1 engagement and β-arrestin 2 assays, preincubation of GBD-003 elicited minimal response and only reduced CP55940 maximal efficacy by ∼10%, consistent with its low efficacy at CB1. In contrast, the JWH-018 isomer significantly reduced CP55940 efficacy by 70% in both assays, reflecting higher efficacy. At CB2, however, GBD-003 reduced CP55940 efficacy by ∼50% in both Gα_i1 engagement and β-arrestin 2 assays, while JWH-018 isomer maintained an ∼60% reduction. These results corroborate results testing GBD-003 and JWH-018 isomer efficacy.

In addition, statistical comparisons between CB1 and CB2 revealed several significant differences (Table ). For Gα_i1 engagement efficacy, JWH-018 exhibited a reduction at CB2 relative to CB1 of ∼70%, while the JWH isomer showed a more modest but significant decrease of ∼30%. For β-arrestin 2 recruitment efficacy, GBD-002 (∼70%) and JWH-018 (∼90%) both showed a higher E _max at CB1 and CB2, whereas GBD-003 displayed the opposite pattern with a higher E_max at CB2 (∼30%). For potency, GBD-002 was more potent at CB1 than CB2 in the Gα_i1 engagement assay, while CP55940 and JWH-018 showed higher Gα_i1 engagement potency at CB2. In the β-arrestin 2 recruitment assay, GBD-003 and JWH-018 isomer exhibited higher potency at CB2.

Overall, our results demonstrate that the headgroup positioning influences receptor efficacy. GBD-003 displayed higher efficacy at CB2 compared to CB1, whereas GBD-002 showed the opposite trend. For the JWH isomers, both showed lower efficacy at CB2, but JWH-018 isomer exhibited higher efficacy than JWH-018, reversing their relative efficacies at CB1. Bias factor analysis with CP55940 as the reference agonist revealed no appreciable signaling bias between CB1 Gα_i1 engagement and β-arrestin 2 recruitment for any compound at either receptor (Table S3).

Reciprocal Substitutions of Nonconserved Residues between CB1 and CB2 Influence Efficacy but Not Selectivity

We next investigated how differences in binding between the isomeric pairs contribute to their potency and efficacy profiles by using site-directed mutagenesis in CB1 and CB2. Molecular dynamics simulations from a previous study revealed two key nonconserved residues in CB1 and CB2 that influence the binding of GBD-002 and GBD-003. In CB1, residues L193^3.29 and L359^6.51 stabilize GBD-002, which forms a hydrogen bond with S383^7.39. However, in the case of GBD-003, the positioning of the indane ring introduces steric strain, shifting the ligand toward extracellular loop 2 (ECL2) and disrupting the hydrogen bond with S383^7.39. In contrast, CB2 contains shorter residues at the corresponding positions, I110^3.29 and V261^6.51, which may allow both ligands greater rotational freedom within the binding site while potentially maintaining hydrogen bonding with S285^7.39. Furthermore, ECL2 is a divergent region between CB1 and CB2. It has been proposed that CB1 residue I267^ECL2, located within ECL2, contributes to ligand selectivity by creating additional space within the CB1 binding pocket, whereas the corresponding bulkier residue in CB2 may cause steric hindrance. , Docking poses of GBD-002 and GBD-003 within the CB1 binding pocket illustrate the spatial proximity of these nonconserved residues to the ligand headgroup (Figure E,F).

2.

Reciprocal substitution of nonconserved orthosteric residues in CB1 modulates ligand efficacy. Gα_i1 engagement BRET concentration–response curves for CP55940, GBD-002, GBD-003, JWH-018, and JWH-018 isomer at CB1 mutants (A) L193I, (B) I267L, (C) L359V, and triple mutant (D) L193I/I267L/L359V. Docking poses of (orange, E) GBD-002 and (yellow, F) GBD-003 in the CB1 binding pocket with nonconserved residues highlighted in pink (PDB: 5XR8). Data are presented as mean ± SEM (n = 4 experiments).

We then tested whether reciprocal substitutions of these nonconserved residues could recreate the opposite receptor’s ligand-binding characteristics. For CB1, we made the following mutations: L193I^3.29, I267L^ECL2, L359V^6.51, and a triple mutant containing all three substitutions (CB1 triple mutant). For CB2, we generated I110L^3.29, L182I^ECL2, V261L^6.51, and a corresponding triple mutant (CB2 triple mutant). Triple mutants would assess how the combined effects of the three residues influence ligand efficacy and potency. BRET ratios were normalized to the WT receptor response to CP55940 to enable comparison of each isomer’s efficacy relative to the wild-type baseline. This allowed us to assess how individual mutations altered the efficacy of each isomer and to compare those changes across mutations.

The CB1 L193I^3.29 mutation resulted in a significant reduction in efficacy for the CB1-selective agonists GBD-002 (∼30%) and JWH-018 (∼50%) compared to the wild-type (WT) receptor (Figure A, Table ). In contrast, GBD-003 and JWH-018 isomer showed minimal reductions in efficacy. The CB1 I267L^ECL2 mutation caused a general decrease in efficacy for all ligands, and this effect was particularly pronounced for GBD-002 (∼50%) and JWH-018 (∼65%) with nearly twice the decrease observed for their respective isomers GBD-003 (∼30%) and JWH-018 isomer (∼40%) (Figure B, Table ).

2. Reciprocal Substitutions of Nonconserved Residues in CB1 and CB2 Binding Pocket .

CB1 Mutant Receptors: Gαi1 Engagement
	WT	L193I	I267L	L359V	Triple Mutations
	E max (%)
CP55940	99.4 ± 3.0	98.2 ± 2.2	77.6 ± 2.5***	122.3 ± 3.3***	160.5 ± 1.6****
GBD-002	116.4 ± 4.9	83.6 ± 3.0**	67.5 ± 4.1****	124.4 ± 4.1	140.5 ± 2.4
GBD-003	55.1 ± 5.4	56.2 ± 4.3	26.9 ± 2.9*	40.3 ± 4.7	88.9 ± 1.9**
JWH-018	139.6 ± 5.6	91.8 ± 2.1****	75.4 ± 4.5****	126.7 ± 3.4	148.7 ± 1.5
JWH-018 isomer	110.7 ± 5.0	106.1 ± 2.8	72.6 ± 4.0****	102.4 ± 4.5	148.6 ± 2.6****
	pEC50 (M)
CP55940	8.03 ± 0.09	7.31 ± 0.06	7.92 ± 0.10	7.75 ± 0.08	7.76 ± 0.03
GBD-002	7.16 ± 0.10	7.36 ± 0.10	6.94 ± 0.15	7.19 ± 0.09	7.56 ± 0.05
GBD-003	5.89 ± 0.16	6.15 ± 0.14	6.78 ± 0.26	6.27 ± 0.23	6.55 ± 0.05
JWH-018	7.61 ± 0.12	7.88 ± 0.07	7.38 ± 0.17	7.71 ± 0.08	7.89 ± 0.03
JWH-018 isomer	6.99 ± 0.11	6.95 ± 0.06	6.87 ± 0.13	7.07 ± 0.11	7.09 ± 0.04

CB2 Mutant Receptors: Gαi1 Engagement
	WT	I110L	L182I	V261L	Triple Mutations
	E max (%)
CP55940	99.8 ± 2.7	83.1 ± 2.8***	32.2 ± 2.5****	33.8 ± 2.6****	3.0 ± 2.0****
GBD-002	78.7 ± 8.5	66.8 ± 6.9	17.6 ± 2.7****	24.7 ± 3.0****	–1.9 ± 2.0****
GBD-003	75.3 ± 5.2	74.2 ± 7.0	26.3 ± 3.6***	32.4 ± 12.5**	–5.4 ± 1.7****
JWH-018	73.5 ± 3.9	58.3 ± 5.9	15.1 ± 3.6****	29.9 ± 4.1****	–0.4 ± 2.3****
JWH-018 isomer	84.5 ± 6.5	66.5 ± 6.5	23.5 ± 3.9****	64.5 ± 9.7	10.8 ± 2.2****
	pEC50 (M)
CP55940	8.76 ± 0.10	8.81 ± 0.13	8.30 ± 0.27	9.32 ± 0.35	8.23 ± 9.48
GBD-002	5.86 ± 0.20	5.34 ± 0.40	9.35 ± 0.53	8.10 ± 0.46	8.73 ± 6.11
GBD-003	6.35 ± 0.21	5.16 ± 0.46	7.35 ± 0.42	5.55 ± 0.60	11.05 ± 2.40*
JWH-018	8.42 ± 0.19	7.96 ± 0.35	8.38 ± 0.84	9.23 ± 0.59	8.74 ± 3.29
JWH-018 isomer	7.35 ± 0.23	7.31 ± 0.27	7.37 ± 0.49	6.08 ± 0.30	8.74 ± 0.98

^a*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 denote differences in WT to mutant compound efficacies determined by one-way ANOVA with Dunnett’s post hoc test.

Interestingly, the CB1 L359V^6.51 mutation showed a trend toward increased efficacy for CP55940 and GBD-002 and a decrease for GBD-003, JWH-018, and the JWH-018 isomer (Figure C, Table ). In the CB1 triple mutant (L193I^3.29, I267L^ECL2, and L359V^6.51), all compounds displayed increased efficacies (Figure D, Table ). Notably, GBD-003 (∼35%) and JWH-018 isomer (∼40%) showed statistically higher efficacy compared to the WT receptor. Potency changes across all CB1 mutants were not statistically significant. The CB1 triple mutant also showed increased β-arrestin 2 recruitment efficacy for all tested compounds, consistent with the Gα_i1 engagement results (Figure S2A).

The reductions in efficacy for GBD-002 and JWH-018 at L193I^3.29 and I267L^ECL2, alongside the increased efficacy of GBD-003 and the JWH-018 isomer in the triple mutant, suggest that they partially mimic the CB2 receptor’s binding pocket in efficacy.

Docking poses of GBD-002 and GBD-003 within the CB2 binding pocket show the positions of the corresponding nonconserved residues relative to the ligand headgroup (Figure E,F). For the CB2 receptor mutants, the I110L^3.29 substitution resulted in a decrease in efficacy across all ligands (∼1% to 20%) (Figure A, Table ). The CB2 L182I^ECL2 mutation caused a significant reduction in efficacy for all compounds (∼50% to 70%) with GBD-003 exhibiting a slightly smaller decrease (Figure B, Table ). Similarly, the CB2 V261L^6.51 mutation significantly reduced efficacy for all compounds except the JWH-018 isomer, which showed only an ∼20% decrease (Figure C, Table ). Interestingly, the CB2 triple mutant (I110L^3.29, L182I^ECL2, and V261L^6.51) abolished agonist efficacy for all tested compounds (Figure D, Table ). No significant changes in potency were observed for any CB2 mutant relative to WT. In line with the Gα_i1 result, no drug-induced BRET was observed in the CB2 triple mutant β-arrestin 2 assay (Figure S2B).

3.

Reciprocal substitution of nonconserved orthosteric residues in CB2 reduces ligand efficacy. Gα_i1 engagement BRET concentration–response curves for CP55940, GBD-002, GBD-003, JWH-018, and JWH-018 isomer at CB2 mutants (A) I110L, (B) L182I, (C) V261L, and the triple mutant (D) I110L/L182I/V261L. Docking poses of (orange, E) GBD-002 and (yellow, F) GBD-003 in the CB2 binding pocket with nonconserved residues highlighted in green (PDB: 5ZTY). Data presented as mean ± SEM (n = 4 experiments).

Overall, CB2 mutants primarily exhibited substantial reductions in efficacy in contrast to the reciprocal mutations in CB1 with a partially CB2-like efficacy. A possible explanation may be the structural difference between the two receptors: the CB1 binding pocket is generally larger and more flexible, whereas the CB2 pocket is narrower and more rigid. As a result, amino acid substitutions in CB2 may cause effects beyond ligand interactions, disrupting receptor dynamics due to the compact binding site. ,

To confirm that the observed changes in efficacy were not attributable to differences in receptor expression levels, we assessed surface expression of all mutant constructs by flow cytometry. No significant differences in surface expression were observed between CB1 or CB2 mutant receptors and their respective wild-type controls (Figure S3). These results indicate that the efficacy changes observed in BRET assays reflect altered receptor activation rather than reduced receptor density. Furthermore, cAMP inhibition measured via the CAMYEL biosensor corroborated the Gα_i1 engagement results for key mutants exhibiting near complete loss of efficacy (Figures S4 and S5, Table S4).

Alanine Substitution of TM2 Aromatic Residues Reveals Their Significant Role in Modulating Ligand Efficacy

In comparison to CB2, CB1 receptor activation induces a structural rearrangement in transmembrane helix 2 (TM2), which has been proposed to play a key role in modulating ligand efficacy. ,, This shift repositions aromatic residues F170^2.57, F174^2.61, F177^2.64, and H178^2.65 toward the headgroup of MDMB-FUBINACA, a synthetic cannabinoid receptor agonist structurally similar to the GBD and JWH compounds. Given that efficacy was the primary differentiating factor between the isomeric ligand pairs, we examined whether variations in headgroup positioning influenced interactions with these TM2 residues. To test this, we performed alanine scanning mutagenesis and evaluated changes in ligand efficacy. Docking poses of GBD-002 and GBD-003 within the CB1 binding pocket highlight the spatial arrangement of TM2 aromatic residues relative to the ligand headgroup (Figure E,F).

4.

CB1 TM2 alanine mutations reduce ligand efficacy across all tested compounds. Gα_i1 engagement BRET concentration–response curves for CP55940, GBD-002, GBD-003, JWH-018, and JWH-018 isomer at CB1 TM2 mutants (A) F170A, (B) F174A, (C) F177A, and (D) H178A. Docking poses of (orange, E) GBD-002 and (yellow, F) GBD-003 in the CB1 binding pocket with TM2 residues highlighted in pink (PDB: 5XR8). Data are presented as mean ± SEM (n = 4 experiments).

The F170A^2.57 mutation caused a significant reduction in E _max for all compounds (Figure A, Table S2). JWH-018 and JWH-018 isomer were most affected, exhibiting ∼100% and ∼80% decreases, respectively, while GBD-002 and GBD-003 showed more moderate reductions (∼60% and ∼50%). Interestingly, the F174A^2.61 mutation led to a 32% increase in efficacy for GBD-002, while GBD-003 showed a downward trend (Figure B, Table S2). In contrast, both JWH compounds exhibited significant decreases in efficacy, indicating that F174^2.61 may play a more selective role in modulating efficacy for JWH ligands and GBD-003 but not GBD-002. The F177A^2.64 mutation resulted in a total reduction in efficacy across all compounds (Figure C, Table S2). Finally, the H178A^2.65 mutation caused a substantial decrease in efficacy for the JWH compounds (∼80%), while having a more moderate effect on the GBD compounds (∼35%) (Figure D, Table S2).

For TM2 mutations, alanine substitution reduced overall ligand efficacy while preserving the rank order of ligand responses, suggesting that TM2 may modulate efficacy independently of the headgroup chemical composition and spatial orientation. Together, these findings indicate that the TM2 helix plays a significant role in modulating the ligand efficacy.

Discussion

To investigate how the position of dimethyl-indane isomers GBD-002 and GBD-003 influences CB2 selectivity, we utilized BRET to monitor the direct coupling of wild-type and mutant CB1 and CB2 receptors with Gα_i/o and β-arrestin 2. Our findings suggest that the orientation of dimethyl-indane influences potency at CB1 but not CB2, while modulating efficacy at both receptors. Notably, GBD-002 and JWH-018 exhibited E _max values exceeding 100% relative to CP55940 at CB1, consistent with superagonism observed in BRET engagement assays in our recent study. However, BRET engagement assays measure receptor to G protein proximity, and we cannot exclude that signal amplification may contribute to the elevated E _max values observed. Single amino acid substitutions of nonconserved orthosteric residues between CB1 and CB2 revealed that introducing CB2 residues into CB1 partially recapitulated CB2-like efficacy but not potency, whereas the reverse CB1-to-CB2 substitutions did not recapitulate CB1-like efficacy. Additionally, mutagenesis of CB1 TM2 residues demonstrated their significant role in modulating ligand efficacy. Bias factor calculations revealed no significant signaling bias between Gα_i1 engagement and β-arrestin 2 recruitment for any compound at either receptor (Table S3), suggesting that the observed differences in efficacy between isomer pairs are not driven by pathway-selective signaling.

Contrary to our initial hypothesis that substituting nonconserved orthosteric residues between CB1 and CB2 would reproduce both efficacy and potency, only the CB1 triple mutant was able to restore CB2-like efficacy. A recent study using the CB2-selective agonist YL025 proposed that ligand recognition in CB1 and CB2 is driven predominantly by entropic factors, such as binding pocket flexibility, rather than enthalpic contributions from direct ligand–receptor interactions. In this model, the CB2 binding site is relatively rigid and undergoes minimal conformational change upon activation, favoring interactions with rigid ligands. In contrast, CB1 undergoes more extensive rearrangement upon activation, particularly involving inward movement of TM1 and TM2, making it more dynamic and better suited to accommodate flexible ligands. ,, This entropic explanation may account for why the reciprocal mutations did not result in significant potency shifts due to their conservative changes (e.g., leucine to isoleucine or valine).

Further, triple substitutions of nonconserved orthosteric residues led to striking divergence in receptor activity: introducing CB2 residues into CB1 (CB1 triple mutant) significantly enhanced efficacy, whereas introducing CB1 residues into CB2 (CB2 triple mutant) abolished it. This is consistent with computational studies showing that the antagonist-bound conformation of CB2 resembles the agonist-bound conformation of CB1 with similar pocket volumes and nearly identical residue arrangements. Our data suggest that, while CB1 and CB2 share overall structural homology in their binding pockets, a small number of nonconserved residues may function as molecular switches for receptor activation. The fact that subtle side-chain substitutions (e.g., leucine to isoleucine or valine) yielded such dramatic effects underscores the sensitivity of the residues of the orthosteric binding pocket to cannabinoid receptor activation and may explain how shifts in ligand headgroup orientation can differentially influence receptor activation.

This study provides insights into the molecular determinants of cannabinoid receptor selectivity and efficacy. The dimethyl-indane position in GBD compounds influenced both receptor selectivity and efficacy: GBD-002 showed high efficacy at both receptors without selectivity, whereas GBD-003 exhibited CB2 selectivity with enhanced efficacy. These distinct profiles demonstrate the unique utility of dimethylindane as a versatile headgroup for fine-tuning CB1 and CB2 receptor interactions. Reciprocal mutagenesis of nonconserved orthosteric residues revealed that introducing CB2 residues into CB1 partially recapitulated CB2-like efficacy, whereas the reverse substitutions abolished CB2 efficacy, while TM2 mutations demonstrated the significant role of these aromatic residues in modulating CB1 efficacy. Although point mutations may introduce broader conformational effects beyond ligand binding and the present study relies primarily on BRET-based readouts, several lines of evidence support the specificity of our findings. FACS-based expression analysis confirmed that all mutant receptors were expressed at levels comparable to wild-type, supporting the interpretation that efficacy changes reflect altered receptor activation rather than global structural disruption. Additionally, CAMYEL cAMP assays corroborated the BRET results for mutants exhibiting near-complete loss of efficacy (Figures S4 and S5 and Table S4).

Nevertheless, the mechanistic interpretations derived from mutagenesis and functional data remain inferential, and future studies incorporating molecular dynamics simulations, direct ligand receptor binding quantifications, and thermostability measurements will be important to further validate the proposed roles of individual residues in ligand recognition and receptor activation. Additionally, while the JWH isomer pair provided supporting evidence that headgroup attachment position broadly influences receptor efficacy and selectivity, conclusions regarding specific residue interactions remain limited to the GBD scaffold. These results offer a valuable framework for structure-guided drug design. This framework leverages dimethylindane’s modularity and contributes to a mechanistic understanding of how SCRA structural features govern cannabinoid receptor function.

Glossary

Abbreviations

APC

allophycocyanin

BRET

bioluminescence resonance energy transfer

BSA

bovine serum albumin

cAMP

cyclic adenosine monophosphate

CAMYEL

cAMP sensor using YFP-Epac-RLuc

CB1

cannabinoid receptor 1

CB2

cannabinoid receptor 2

CNR1

cannabinoid receptor 1 gene

CNR2

cannabinoid receptor 2 gene

DMEM

Dulbecco’s modified Eagle’s medium

DMSO

dimethyl sulfoxide

ECL2

extracellular loop 2

ECS

endocannabinoid system

EDTA

ethylenediaminetetraacetic acid

GPCR

G protein-coupled receptor

HEK

human embryonic kidney

MFI

median fluorescence intensity

NPS

novel psychoactive substances

PBS

phosphate-buffered saline

PEI

polyethylenimine

RET

resonance energy transfer

RLuc

Renilla luciferase

SAR

structure–activity relationships

SCRA

synthetic cannabinoid receptor agonist

SSC

smokable synthetic cannabinoid

THC

tetrahydrocannabinol

TM

transmembrane

TM2

transmembrane helix 2

WT

wild-type

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomedchemau.6c00090.

• BRET dose–response curves for CB1 and CB2 triple mutant receptors measuring Gα_i1 interaction, β-arrestin 2 recruitment, cAMP inhibition, and cell surface expression of mutant receptors detected by FACS (PDF)

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Soo Jung Oh: Data curation, Formal analysis, Investigation, Methodology, Visualization, Funding acquisition, Writing. Ruiru Guo: Investigation, Methodology. Christopher Lucaj: Investigation, Methodology, Funding acquisition. Neshira Gutierrez: Investigation, Methodology. Kh Tanvir Ahmed: Investigation. Alexander H. Williams: Investigation. Chang-Guo Zhan: Investigation. Gregory B. Dudley: Resources, Supervision, Writing. Hideaki Yano: Data curation, Formal analysis, Project administration, Resources, Funding acquisition, Supervision, Writing.

The authors declare no competing financial interest.