Structural basis of THC analog activity at the Cannabinoid 1 receptor

Abstract

Tetrahydrocannabinol (THC) is the principal psychoactive compound derived from the cannabis plant Cannabis sativa and approved for emetic conditions, appetite stimulation and sleep apnea relief. THC's psychoactive actions are mediated primarily by the cannabinoid receptor CB₁. Here, we determine the cryo-EM structure of HU210, a THC analog and widely used tool compound, bound to CB₁ and its primary transducer, G_i1. We leverage this structure for docking and 1000 ns molecular dynamics simulations of THC and 10 structural analogs delineating their spatiotemporal interactions at the molecular level. Furthermore, we pharmacologically profile their recruitment of G_i and β-arrestins and reversibility of binding from an active complex. By combining detailed CB₁ structural information with molecular models and signaling data we uncover the differential spatiotemporal interactions these ligands make to receptors governing potency, efficacy, bias and kinetics. This may help explain the actions of abused substances, advance fundamental receptor activation studies and design better medicines.

Fig. 1. Active state CB₁ structures – completeness and complexes with G_i and THC analogs.

a Cryo-EM density map (sharpened) of the HU210/CB₁/G_i1 signaling complex. b CB₁-G_i complexes^,,, superposed based on the seven transmembrane helix backbone atoms. N-terminal (c) and TM5 (d) segments uniquely covered by the HU210/CB₁/G_i1 structure. The shown residues are the terminal residue in each structure. Extracellular (e) and (f) side views of THC analog/CB₁ structure complexes. g Structure similarity tree of active state CB₁ structures based on a conformational clustering in GPCRdb making a superposition-independent all-to-all residue distance comparisons across the transmembrane domain.

Fig. 2. G protein recruitment of THC analogs and endogenous agonists via CB₁.

Ligand concentration-response curves for (a) low-potency/high-efficacy agonists, (b) high-potency/high-efficacy agonists, (c) low-potency/low-efficacy agonists, (d) high-potency/low-efficacy agonist, and (e) inverse agonist, from mG_i recruitment. f Scatter plot of ligand pEC₅₀ (x-axis) and E_max (y-axis) in the mG_i experiments. HEK293-TR cells stably co-expressing a combination of either the CB₁-NlucC and NES-venus-mGsi or NES-venus-β-arrestin were used. Recruitment data, expressed as a % of the maximal response produced by the endogenous cannabinoid 2AG, is plotted versus log concentration for the indicated ligands. Data were fitted to a four-parameter logistic equation; mean ± S.E.M. of at least 3 independent experiments. N-values for the individual ligand CRCs are: 2AG (6), AEA (9), AM841 (6), AM11542 (6), CP55940 (17), HU210 (6), HU243 (6), Nabilone (6), Cannabinol (6), JWH133 (5), L759633 (3), THC (12) and THCv (6).

Fig. 3. Structure-activity relationships of studied THC analogs.

Scatter plot of ligand pEC₅₀ (x-axis) and E_max (y-axis) in the mG_i experiments. Dotted lines divide the plot into quadrants of low/high pEC₅₀/E_max. For THC analogs, the 2D structure is shown next to the data points.

Fig. 4. Spatiotemporal THC analog/CB₁ interactions.

a Left: Spatial distribution of the 25 CB₁ residues interacting with HU210 (pink), AM841¹⁷ (blue), AM11542¹⁶ (yellow), and CP55940¹⁸ (green). Right: The three subsites accommodating the tetrahydrocannabinol, alkyl branch, and alkyl tail, respectively. Pocket residues are shown as C_α spheres and labeled with generic residue numbers facilitating comparison of structurally corresponding positions across receptors. b Ligand-receptor interaction fingerprints denoting interactions, type, and CB₁ generic residue positions. Interaction labels indicate receptor-ligand contacts in pre-MD starting structure/model (see Methods). Grayscale illustrates frequencies of interactions (%) throughout simulations, averaged across three replicates. Five CB₁ residue positions marked “N-ter” and “ECL2” lack a generic residue number because they are not structurally conserved across receptors. c Differences in contact frequencies (in % averaged across three replicates) of THC analogs relative to THC (reference) at each receptor residue. Positive values (blue cells) indicate contact frequencies greater than those in THC, whereas negative values (red cells) indicate contact frequencies lesser than those in THC.

Fig. 5. Structural determinants of ligand activity observed in MD simulations.

Visual representations of ligand-receptor interactions from MD simulations that are key determinants of their structure-activity relationship. a Conformation of the elongated alkyl tail of AM841 in comparison with THC. b Binding mode of CP55940 depicting the unique hydrogen bond interaction between the 6a hydroxypropyl and the sidechain of S173^2x60. c–e Binding modes of low-potency/low-efficacy THC analogs (JWH133, L759633, and Cannabinol, respectively) in comparison with THC. f Conformational differences in THC analogs having different hybridization states at C9 carbon of the C-ring; HU210 (sp²) and AM841 (sp³) are shown as representative analogs. g Superposition of THCv conformations indicating binding flexibility.

Fig. 6. Structure-kinetics relationships of THC analogs studied for reversibility of agonist activity (mG_i recruitment) upon addition of the CB₁ antagonist rimonabant.

The plot shows the change in basal BRET upon addition of agonist (0 min, EC₈₀ concentration) and rimonabant (20 min, 10 μM). Data from HEK293-TR cells stably expressing CB₁-Nluc and NES-venus-mG_i. The number of observations (n) is 6 for each of the individual ligand dissociation curves except for THC (n = 5). Data plotted is mean ± S.E.M. THC as a partial agonist has a relatively small response; therefore, the visible error bars correspond to an “addition artifact” which is more apparent for this ligand.