Multiple intramolecular triggers converge to preferential G protein coupling in the CB2R

Abstract

G protein-coupled receptors (GPCRs) are important therapeutic drug targets for a wide range of diseases. Upon activation, GPCRs can initiate several signaling pathways, each with unique therapeutic implications. Therefore, understanding how drugs selectively engage specific signaling pathways becomes paramount. However, achieving this selectivity remains highly challenging. To unravel the underlying multifaceted mechanisms, we integrate systematic mutagenesis of the CB₂R, comprehensive profiling of G_αi2 and β-arrestin1 engagements and computer simulations to track the effects of mutations on receptor dynamics. Our research reveals multiple triggers within a complex allosteric communication network (ACN) that converge to preferential CB₂R coupling by modulating evolutionarily conserved motifs. Utilizing network path analysis, we find that potent triggers are typically highly connected nodes and are located near regions of high information transmission within the ACN. Our insights highlight the complexity of GPCR signaling and provide a framework for the rational design of drug candidates tailored to evoke specific functional responses, ultimately enhancing the precision and efficacy of therapeutic interventions.

Fig. 1. Cell surface expression of 360 CB₂R mutants.

A CB₂R snake plot showing the distribution of mutants with low, WT-like, and high surface expression. These categories correspond to less than 80% (red), between 80% and 120% (gray), and more than 120% (yellow) of the wild-type expression level. B Barplot showing the distribution of the three expression categories: low, WT-like and high. C Relationship between evolutionary conservation of a residue and its impact on cell-surface expression upon mutation. The conservation score (Jensen-Shannon divergence) was computed for all residues within class A GPCRs. Residues with low conservation scores have the least effect on cell-surface expression, whereas high conservation scores can significantly impact cell-surface expression. Gray arrows indicate the increasing impact. WT surface expression level (100%) is shown with a solid horizontal gray line, while the two gray dashed lines delimit the boundaries of the WT-like expression level. Source data are provided as a Source Data file.

Fig. 2. Induced coupling profiles by CB₂R mutants.

A Location of mutations in the CB₂R classified on a snake plot as: Coup_{Gαi2_βarr1}: preserved G_αi2 and βarr₁ coupling (blue), PrefCoup_Gαi: preferential G_αi2 coupling through a loss of βarr₁ recruitment (orange), PrefCoup_βarr1: preferential βarr₁ recruitment through loss of G_αi2 coupling (red), and NoCoup_{Gαi2_βarr1}: G_αi2 and βarr₁ couplings are lost (green). Bold outlined positions indicate highly conserved residues/microswitches such as NPxxY, CWxP, etc. B Distribution of coupling profiles classes (see also scatter plot in Supplementary Fig. 1). C Emax distribution for CB₂R mutants with a preferential G_αi2 coupling profile. The Emax distribution resembles a normal distribution which is confirmed by a two-side D’Agostino and Pearson’s test (p-value = 4.89 E⁻⁰⁵_, sample size: n = 64 βarr₁ deficient mutants). D Distribution of coupling profiles for mutants in conserved regions. Percentages differ considerably from the ones of the full set of mutants in (Fig. 2B). E Kernel density estimation of the distribution of conservation scores among PrefCoup_Gαi2 mutants and Coup_{Gαi2_βarr1} mutants with a WT-like expression level (80% to 100%) showing a clear shift between both densities. The statistical significance of this shift is confirmed using a one-side Kolmogorov–Smirnov test (p-value of 0.07, sample size: n = 64 PrefCoup_Gαi2 and 251 Coup_{Gαi2_βarr1} mutants). A conservation score distribution for all coupling classes can be found in Supplementary Fig. 11. Source data are provided as a Source data file.

Fig. 3. The allosteric communication network in the WT CB₂R.

A Depiction of the entire allosteric communication network in the inactive WT CB₂R (WT ACN) (PDB ID: 6KPC). The WT ACN encompasses the ligand-stabilized ACN (LigACN, dark blue) and their top connections with high information transmission (LigACN_top, light blue). The thickness of the contacts is proportional to their information transmission between the ligand and the intracellular receptor site (see Methods). B Schematic illustration of information transmission and the connectivity of a specific node in the network. C PrefCoup_Gαi2 mutants are shown as orange spheres alongside the LigACN and its top connections (LigACN_top) on the inactive CB₂R structure (PDB ID: 6KPC). D Distance of PrefCoup_Gαi2 and Coup_{Gαi2_βarr1} mutants to the LigACN_top. The distance is represented as the number of edges from a mutation position to the LigACN_top (x-axis). The percentage of mutants at a given distance is provided (y-axis). Note that the plotted distance profile depends on the cutoff for the information transmission that defines the LigACN_top (cutoff for top connections: 0.146). To confirm that the described tendency holds true also for different cutoffs values, we provide distribution plots in Supplementary Fig. 14. E Connectivity distribution plot of PrefCoup_Gαi2 mutants (orange) and Coup_{Gαi2_βarr1} (blue). Connectivity for each residue position is computed as the number of connections in the network (see Fig. 3B). Vertical lines indicate the mean value of each distribution. Statistical analysis between PrefCoup_Gαi2 and Coup_{Gαi2_βarr1} mutants was performed using one-side Student’s t-test (*p < 0.001, sample size: n = 14 PrefCoup_Gαi2 and 20 Coup_{Gαi2_βarr1} mutants). Source data are provided as a Source Data file.

Fig. 4. MD-based classification of PrefCoup_Gαi2 and Coup_{Gαi2_βarr1} mutants.

A Location of simulated mutants and their coupling profiles shown on a snake plot (left) and the CB₂R structure (right). PrefCoup_Gαi2 mutants are shown in orange and Coup_{Gαi2_βarr1} in blue (PDB ID: 6KPC). B MD simulations scheme: the initial system was simulated in production conditions for 0.6 µs. After that, 5 replicates were respawned applying random perturbations to the atom speeds. These 5 replicates were evolved for further 0.4 µs reaching a total evolution time of 1 µs. This yields a total simulation time of 2.6 µs per mutant. The respawned replicates (5 × 0.4 µs) are used for data analysis. C Distribution plot of the network size of LigACNs (i.e., number of nodes) for PrefCoup_Gαi2 mutants compared to WT and Coup_{Gαi2_βarr1} mutants. The data for LigACNs including interatomic contacts in the ACN for all 34 mutants and the wild type is found in Supplementary Data 2. The center line in boxplots represents the median while the box boundaries extend from the first (25%) and third (75%) quartile, representing the interquartile range (IQR). Boxplot whiskers extend to 1.5*IQR and outliers are represented as points. Schematic representations illustrate the impact of a mutation on the LigACN_top size. D Principal Component Analysis (PCA) of MD data using contact frequencies for 10 different contact types: 3 types of hydrogen bonds, salt bridges, Pi-cation, Pi-stacking, T-stacking, Van der Waals including hydrophobic interactions, water bridges, extended water bridges. The best-performing PC plane for separating Coup_{Gαi2_βarr1} and PrefCoup_Gαi2 is PC1 versus PC10, selected using a Support Vector Classifier (SVC). The SVC model (blue contour line) encompasses the Coup_{Gαi2_βarr1} mutants in the middle of the plane and three different clusters for the PrefCoup_Gαi2 mutants (orange line) at peripheral regions. The WT control simulations are highlighted in red. Source data are provided as a Source data file.

Fig. 5. Structural characterization of PrefCoup_Gαi2 mutant clusters.

A Perturbation map of the ACN induced by PrefCoup_Gαi2 mutants. Top interaction features for PrefCoup_Gαi2 cluster 1 (dark brown), 2 (light brown) and 3 (orange). The interactions are plotted together with the computed ACNs for the WT CB₂R. For instance, such depiction highlights allosteric connections (brown to orange) to the ligand-stabilized ACN (light and dark blue) that are perturbed in preferential G_αi2 coupling of the CB₂R (PDB ID: 6KPC). B–D Mutant impact on highly conserved receptor regions for clusters 1 to 3. Non-conserved residues which are part of the modulation are represented as white spheres whereas the corresponding contact is highlighted as a cylinder indicating the change of contact stability compared to conserved coupling (Coup_{Gαi2_βarr1}): navy blue corresponds to increased interaction stability in the PrefCoup_Gαi2 mutants, yellow indicates decreased interaction stability. The thickness of the cylinders is proportional to the absolute value of the coefficient for that interaction in the logistic regression model (PDB ID: 6KPC). E Summary of mutant-perturbed domains of high conservation (i.e., CWxP, Na+ site, NPxxY, DRY) for PrefCoup_Gαi2 mutant clusters 1 to 3. For each cell, we indicate the implicated residues in the contact and the corresponding contact frequency for each pair of residues. Each cell is colored according to the prediction of a logistic regression model: more stable contacts (navy blue) and less stable contacts (yellow) of PrefCoup_Gαi2 compared to Coup_{Gαi2_βarr1} mutants. Further details about differences in contact frequencies between PrefCoup_Gαi2 and Coup_{Gαi2_βarr1} mutants are plotted in Supplementary Fig. 5. Source data are provided as a Source data file.