Interacting binding insights and conformational consequences of the differential activity of cannabidiol with two endocannabinoid-activated G-protein-coupled receptors

2.2 GPR55 receptor structures

Thus far, there is no experimentally-determined structural information regarding the GPR55 receptor. Hence, the structure of the rat GPR55 receptor was built using the homology modeling technique (Eswar et al., 2006). To identify possible adequate templates for GPR55, the Swiss-Model server was used (Waterhouse et al., 2018). The selected system was the lysophosphatidic acid receptor 6 (LPA6) which displays a sequence identity with GPR55 of about 30% (PDB ID code 5XSZ). The sequences of the LPA6 template and the rat GPR55 receptors were obtained from UNIPROT (with the codes Q08BG4 and F1MAK4, respectively). The sequences were aligned using BLASTp (Altschul et al., 1990) with subsequent manual adjustments. The pairwise sequence alignment is presented in Supplementary Figure S2. To generate 3D structural models of the rat GPR55 receptor, we used the Modeller 9.23 protocol (Šali and Blundell, 1993). 100 models of the R. norvegicus GPR55 receptor were generated and a representative structure was selected based on energetic considerations, that is, using the Modeller’s scoring function (see Supplementary Figure S3).

2.3 CB₁ receptor structures

To model the structures for the rat CB₁ receptor we used different templates for each receptor’s functional state. In the case of the CB₁ receptor that will form a complex with CBD (an tentative agonist for CB₁), the template utilized was the agonist-bound human CB₁ receptor (PDB ID code 5XRA). In the case of the CB₁ receptor that forms a complex with AM251 (an antagonist for CB₁), two structures of the antagonist-bound human CB₁ receptor were used as templates (PDB ID code 5U09 and code 5TGZ). In both cases, 100 models of the R. norvegicus CB₁ structures were modeled via homology modeling using the Modeller 9.23 protocol (Šali and Blundell, 1993). The sequences of the human and rat CB₁ receptors were obtained from UNIPROT (with the codes P21554 and P20272, respectively). Based on energetic considerations, a representative structure of the rat CB₁ receptor was selected in each case for further analysis, see Supplementary Figure S4.

2.4 CBD/GPR55 complex structures (inactive-like GPR55 system)

To identify possible molecular poses for the CBD/GPR55 complex, blind docking calculations were performed using the Autodock-Vina software (Trott and Olson, 2009). The selected molecular pose for the CBD/GPR55 complex is shown in Supplementary Figure S3. Chemical structure of the ligands are displayed in Supplementary Figure S5.

2.5 ML186/GPR55 complex structures (active-like GPR55 system)

Similar to the aforementioned case of the CBD/GPR55 complex, molecular poses for the ML186/GPR55 complex were identified via docking calculations using Autodock Vina (Trott and Olson, 2009). The selected molecular pose for the ML186/GPR55 complex is shown in Supplementary Figure S3.

2.6 CBD/CB₁ complex structures (active-like CB₁ system)

To predict possible molecular poses for the CBD ligand in the structural framework of the CB₁ receptor, docking studies were carried out using Autodock Vina (Trott and Olson, 2009). In addition to the energetic considerations (the scoring function of Autodock Vina), we used the structure of the AM11542/CB₁ complex (PDB ID code 5XRA), as a guide to select a representative structure of the CBD/CB₁ complex (see Supplementary Figure S4). The structure of the AM11542/CB₁ complex was selected based on the similarity of the CBD ligand and the AM11542.

2.7 AM251/CB₁ complex structures (inactive CB₁ system)

Similar docking calculations were utilized in the case of the AM251 antagonist. In this case, in addition to Autodock Vina’s scoring function, the structure of the AM6538/CB₁ complex (PDB ID code 5TGZ), was used as a guide due to the similarity between the AM251 ligand and the AM6538 ligand. The selected structure is indicated in Supplementary Figure S4.

2.8 All-atom molecular dynamics simulations

To provide the interacting ligand binding determinants of the protein systems (CBD/GPR55, ML186/GPR55, CBD/CB₁, and AM251/CB₁) and the concomitant conformational consequences, we performed all-atom molecular dynamics (MD) simulations in each of the four protein complexes. Similar general preparation for the systems has been previously used for GPCRs (Perez-Aguilar et al., 2014; Perez-Aguilar et al., 2019; Dror et al., 2011; Lee and Lyman, 2012; Méndez-Luna et al., 2015) and other membrane proteins (Rangel-Galván et al., 2021). Briefly, using the structural information from the docking calculations as the initial complex structure, hydrogen atoms were included representing the most probable protonation states for all the amino acids residues at neutral pH. The protein complexes were embedded in a hydrated symmetric POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) membrane, where sodium and chlorine ions were included so as to have a physiological salt concentration (0.15 M). Preparation steps that include the imposition of structural force restraints to avoid unrealistic conformation of the system (e.g., water penetration to the lipid membrane) were carried out (Perez-Aguilar et al., 2014; Perez-Aguilar et al., 2019; Rangel-Galván et al., 2021). Next, unbiased MD simulations were carried out for one microsecond (1.0 μs) at a constant temperature and pressure (37°C and 1 atm, respectively).

2.9 Class A GPCR position numbering

To define the position of each residue, we will be using the Ballesteros-Weinstein numbering as superscript. In this nomenclature, the first number indicates the transmembrane helix (from one to seven) whereas the second number indicates the residue position relative to the most conserved residue, which is assigned to the number 50 (Ballesteros and Weinstein, 1995).

3.1 Interacting determinants in the CBD/GPR55 and ML186/GPR55 complexes

3.2 Structural consequences observed in the GPR55 receptor by the presence of the ligands

In order to understand the structural consequence of the presence of CBD ligand in the receptor’s conformations, we analyze several structural motifs associated with GPCR activation. One particular locus linked with GPCR activation is position 6.48, the so-called toggle switch, which usually bears the aromatic residue tryptophan (W^6.48). Upon activation, there is a structural rearrangement in the vicinity of this residue, that seems to propagate conformational changes towards the intracellular side (Zhou et al., 2019). At this position, the GPR55 receptor contains another aromatic residue, the F239^6.48. In this context, we analyzed the conformations of this residue in the CBD-bound and ML186-bound GPR55 systems, particularly the rotameric states explored by the bulky F239^6.48 residue. As observed by the dihedral angles values of the angle defined by the N-CA-CB-CG atoms, in the case of the CBD/GPR55 complex, the distribution of the values explored by the receptor along the simulated trajectory are mainly distributed around 280°–320° (unimodal distribution, see Figure 2A). In the case of the agonist-bound GPR55 complex, the presence of the ML186 ligand provokes a change in the position of the F239^6.48 residue, as indicated by the bimodal distribution of the N-CA-CB-CG dihedral angle (Figure 2A). Experimental evidence indicates that a more dynamic and heterogeneous conformational behavior of the receptor is attributed to the binding of an agonist ligand (Manglik et al., 2015; Latorraca et al., 2017). It this case, the bimodal distribution observed in the ML186/GPR55 system at this position may be associated with active-like conformations while the unimodal distribution of this dihedral angles in the CBD/GPR55 complex are attributed to a GPCR’s inactive conformations. This behavior may be attributed to the thiophene-containing fused-rings moiety of the ML186 agonist, which directly contacts the side chain of the F239^6.48 residue (Figure 1A). These results are in good agreement with the current notion that CBD is an antagonist of the GPR55 receptor while ML186 exhibits agonistic function on the GPR55 (Ryberg et al., 2007; Ross, 2009; Sharir and Abood, 2010).

Open in a separate window

FIGURE 2

Conformation changes in the toggle switch motif distinguished the antagonist-bound versus the agonist-bound GPR55 complexes. (A) Distributions of the N-CA-CB-CD dihedral angle of the F239^6.48 residue in the CBD/GPR55 and ML186/GPR55 complexes. Position 6.48 is known as the toggle switch and has been implicated in GPCR activation. In the case of the GPR55, position 6.48 bears a phenylalanine residue, F239^6.48. (B) Another residue that seems to play a role in the changes of F239^6.48 rotameric state is the Y106^3.37, which also displays a distinctive dihedral angle distribution. (C) Superimposed final structures of the CBD/GPR55 and ML186/GPR55 simulations. The different position of residue F239^6.48 in the two GPR55 systems is illustrated (magenta arrow). The deeper molecular pose of the selective agonist ML186 (orange) may be responsible for the different position of the F239^6.48 sidechain.

Additionally, it has been suggested that conformational changes in position 6.48 with residues in TM3 and TM6 occur upon activation (Zhou et al., 2019; Hilger, 2021). Our simulations suggest that the conformational changes of the F239^6.48 residue are in close relation with those of two other aromatic residues, the bulky aromatic residues Y106^3.37 and F235^6.44. Analysis of the equivalent Y106^3.37 dihedral angle (N-CA-CB-CG atoms) indicates a drastic change in the rotameric states sampled by this residue. While in the CBD/GPR55 system, which displays the unimodal distribution of F239^6.48, the Y106^3.37 dihedral angle (N-CA-CB-CG atoms) also indicates that this aromatic residue samples a one-peak distribution conformations (Figure 2A). The more dynamic behavior of F239^6.48 causes a significant change in the distribution of rotameric states that explores the Y106^3.37 dresidue (Figure 2B).

Lastly, the changes in the GPR55 toggle switch causes F235^6.44 to follow similar trend, that is, the CBD/GPR55 displays an unimodal distribution of the dihedral angle define by the N-CA-CB-CG atoms while the ML186/GPR55 complex displays a more heterogenic distribution (Supplementary Figure S6).

The significant conformational response of the GPR55 receptor to the presence of the CBD antagonist and the ML186 agonist on structural motifs known to play a role on activation, seems to be in good agreement with the experimental evidence linked to these two ligands [Ryberg et al., 2007; Ross, 2009; Sharir and Abood, 2010; Kotsikorou et al., 2013; Anderson et al., 2021; Heynen-Genel et al., 2010 September 29 (updated 26 May 2011); Sharir. et al., 2012; Kotsikorou. et al., 2011]. Furthermore and regarding the known modulatory properties of the two ligands, CBD (antagonist) and ML186 (agonist), our computational approaches were able to distinguish the distinct GPR55 conformations elicited by the presence of each ligand (Figure 2C).

In addition to the aforementioned toggle switch residue, another structural feature that is a hallmark of class A GPCR activation is the presence of an charge-charge intracellular interaction, the so-called ionic lock, formed by R^3.50 and D/E^6.30 (Zhou et al., 2019). Upon activation, this ionic lock needs to be broken so that an outward shift of TM6 is possible (Zhou et al., 2019). Although there is an arginine residue at position 3.50 in GPR55 (R119^3.50), no acidic residue is present at position 6.30. Nevertheless, to have a sense of the relative displacement of the intracellular segment of TM6 from TM3, we calculate the distance between the CA atoms of R119^3.50 and I228^6.37. Position 6.37 was selected since its CA atom is located around the same plane as the CA atom from position 3.50. As indicated in Figure 3A, the R119^3.50 (CA) and I228^6.37 (CA) distance in the antagonist-bound receptor barely samples values longer than 9 Å. In contrast, in the agonist-bound receptor, most of the time this distance overpasses the 9 Å value even reaching 11 Å values. Our calculations reveal that R119^3.50 forms a salt bridge interaction with residue E294^7.59. Indicative of this interaction, the time evolution of the R119^3.50 (CZ)—E294^7.59 (CD) distance is shown in Figure 3B. In the CBD/GPR55 complex, the salt-bridge interaction remains very stable along the entire simulation, however, in the case of the ML186/GPR55 system, the break of this interaction is frequently observed. Again, the computational approaches capture distinctive conformations adopted by the receptor in the presence of different ligands, either antagonist or agonist (Figures 3C,D).

Hence, in the case of the GPR55, the structural consequences of the presence of CBD suggests that the ligand stabilizes the receptor’s inactive conformations, which is in good agreement with the suggested antagonistic activity of CBD on the GPR55 (Ryberg et al., 2007; Ross, 2009; Sharir and Abood, 2010). On the other hand, the presence of the selective ML186 agonist favors structural changes linked to GPCR’s active-like conformations, again, in good agreement with the functional properties of the ligand [Kotsikorou et al., 2013; Anderson et al., 2021; Heynen-Genel et al., 2010 September 29 (updated 26 May 2011); Sharir et al., 2012; Kotsikorou et al., 2011)

3.3 Interacting determinants of the CBD ligand in the CB₁ receptors

Since our computational approaches were able to distinguish the conformational consequences of each of the ligands, either agonist or inverse agonist, for the case of the GPR55 receptor, we decided to apply similar approaches to shed light on the yet-unclear functional activity of CBD on the prototypical CB₁ receptor. Based on the structural similarity of CBD with different cannabinoid ligands (Hua et al., 2017) including the structure of the phytocannabinoid delta-9-tetrahydrocannabinol (∆⁹-THC), which exhibits partial agonism at the CB₁ receptor (Pertwee, 2008; Dutta et al., 2022a), we decided to select molecular poses from our docking calculations that places the CBD compounds in the orthosteric binding site of CB₁.

As described in the Methods section, the initial model of the CBD/CB₁ was obtained from docking approaches. In this case, the information from the crystal structure of the CB₁ receptor in complex with an agonist ligand that shares structural similarities with the CBD ligand (5XRA.pdb accession code), was used as reference to evaluate the ligands poses (Supplementary Figure S5). The CBD/CB₁ system was investigated by unbiased all-atom MD simulations using protocols previously described for transmembrane proteins (see Methods). Figure 4A displays a representative final structure of both CBD complexes at the final stages of the 1 μs simulation. Residues that delineate the CBD binding site are also indicated.

In the case of the CBD/CB₁ complex, the main interactions that stabilized the ligand in the putative orthosteric binding site of CB₁ are indicated. The time evolution of distance representing two of the main interactions are indicated in Figure 4D. As observed, most of the interactions are aliphatic with a particularly relevant aromatic interactions between the aromatic ring of CBD and the side chain of residue F269^ECL2, which is located in the extracellular loop 2 (right inset Figure 4B) and the side chain of residue F380^7.35 in TM7 (left inset Figure 4B). The relevance of the latter interaction has been documented for CB₁ agonists, showing the mutation for alanine reduces agonist’s potency (Hua et al., 2017). Interestingly, the alkyl tail of CBD directly contacts the bulky side chain of residue W280^5.43 in TM5. Lastly, one of the OH groups in CBD forms a polar interaction, sometimes mediated by a water molecule, with residue S384^7.39 (insets Figure 4B). Notably, the presence of this polar residue, S384^7.39, has been found to be relevant in the interaction of CB₁ with cannabinoid-like agonists, since the removal of the hydroxyl group (mutation for alanine) reduced the potency of this type of ligands (Hua et al., 2017).

3.4 Structural consequence observed by the presence of CBD in the CB₁ receptor

The presence of CBD would elicit conformational changes in the CB₁ that stabilize either active-like or inactive conformations (Navarro et al., 2020; Grotenhermen, 2004; Batalla et al., 2021; Śmiarowska et al., 2022; Pertwee, 2008).

To see the consequences of the presence of CBD on the CB₁, first we analyzed the conformation of two bulky hydrophobic residues, which has been linked to CB₁ activation, that is, the F201^3.36 and W357^6.48 residues. These two “bulky” residues are sometimes denominated a “twin toggle switch,” since a conformational change in their rotameric state has been proposed to be essential for the receptor activation (Hua et al., 2017). The distribution of the dihedral angle values formed by the (N-CA-CB-CG atoms of each residue are plotted in Figure 5. We can see that, in the presence of CBD, the F201^3.36 residue explores two distinctive values (bimodal distribution). The first distribution peak overlaps with the value of the antagonist-bound CB₁ receptor [5TGZ.pdb; F^3.36(N-CA-CB-CG) dihedral angle of 182.18°] (Hua et al., 2016), however, the second distribution peak overlaps with the value observed in the agonist-bound CB₁ receptor [5XRA.pdb; F^3.36(N-CA-CB-CG) dihedral angle of 280.59°] (Hua et al., 2017). To put our results in context, we investigated the CB₁ receptor in complex with a well-known antagonist, the AM251 ligand using atomistic MD simulations (Figure 4C). In stark contrast with the CBD/CB₁ case, in the presence of the AM251 antagonist, only one conformation was observed (unimodal distribution) that overlaps with the antagonist-bound CB₁ receptor. This particular conformation of the F^3.36 has been associated with the inactive state of the receptor. Moreover, in the presence of CBD, the W357^6.48 residue also explores two distinct conformations (bimodal distribution) although the first peak is barely sampled by the system. Once again, in the presence of the AM251 antagonist, only one conformation was observed, which has been associated with the inactive state of the receptor (unimodal distribution). Finally, this conformational difference in the residues that constitute the twin toggle switch are related with the rotameric states explored by the bulky residue W280^5.35. As observed in Figure 4B, the alkyl tail of CBD contacts the side chain of W280^5.35 which could facilitate the changes identified in the twin toggle switch. Hence, the presence of CBD causes the receptor to adopt active-like conformations (twin toggle switch) that are associated with agonist-bound systems. In contrast, the results for the simulation of the AM251/CB₁ complex indicate that the conformations adopted by the receptor in the presence of the antagonist ligand are linked to inactive conformations of the receptors. Thus, our computational methods correctly differentiate the structural changes in CB₁ elicit by the different ligands, that is, CBD and AM251 (CB₁ antagonist). Moreover, our results suggest that CBD may act as an agonist on the CB₁ receptor.

Open in a separate window

FIGURE 5

Conformation changes in the toggle switch F201^3.36 and W357^6.48 residues displayed different distributions in the CBD/CB₁ and AM251/CB₁ complexes. (A) Distributions of the N-CA-CB-CD dihedral angle of the F201^3.36 residue in the CBD/CB₁ and AM231/CB₁ complexes. The values indicated in magenta correspond to the antagonist-bound CB₁ structure (5TGZ.pdb) while those colored in blue correspond to the agonist-bound CB₁ structure (5XRA.pdb). As seen here, only the CBD complex explored conformations of the twin toggle switch associated with the agonist-bound CB₁ structure. (B) Distributions of the N-CA-CB-CD dihedral angle of the W357^6.48 residue in the CBD/CB₁ and AM231/CB₁ complexes. Together these two residues are known as the twin toggle switch. (C) Superimposed final structures of the CBD/CB₁ and AM251/CB₁ complexes. The different rotameric states of the F239^3.36, W357^6.48, and W280^5.43 residues in the two CB₁ complexes are shown. As indicated in the main text and in Figure 6, the alkyl tail of CBD directly contacts the bulky side chain of residue W280^5.43.

As mentioned before, one of the hallmarks of GPCR activation is the polar interaction commonly formed by two charged residues located at the intracellular segment of TM3 and TM6. In the case of the CB₁ receptor, this so-called ionic lock is formed by residues R215^3.50 and D339^6.30 (see figure Figure 6). As indicated in Figure 6A, along the simulation time, we observed the disruption of the ionic lock, which has been established as a prerequisite for the interaction of the GPCR with its respective G-protein. Indicated by the distance between the R215^3.50(CZ)-D339^6.30(CG), our results indicate that the values for this distance remains larger that 8.8 A, which is the distance observed in the case of the agonist-bound CB₁ complex (AM11542/CB₁, PDB ID code 5XRA). This result indicates that in the presence of CBD, the ionic lock displays conformations associated with agonist-bound states. Similarly, we calculate the same distance, R215^3.50(CZ)-D339^6.30(CG), for the case of the AM251/CB₁ complex (the antagonist-bound system). As indicated in Figure 6A, the ionic lock remains close, around 4.5/4.7 Å (distances observed in antagonist-bound CB₁ complexes, 5TGZ.pdb/5U09.pdb) and almost never explores distances longer that 8.8 Å. Along the same lines, the R215^3.50(CA)-D339^6.30(CA) also indicates the formation of the ionic lock in the antagonist-bound system (AM251/CB₁ complex) but not in the case of the CBD/CB₁ complex, again, suggesting that CBD may act as an agonist on the CB₁ receptor (Figure 6B). Structures illustrating the ionic lock motif for the CBD/CB1 and AM231/CB1 complexes are shown in Figures 6C,D, respectively.

JP-A acknowledges the computing time granted by the LANCAD on the supercomputer xiuhcoatl at CGSTIC CINVESTAV, members of the CONACyT network of national laboratories. JP-A thanks the Laboratorio Nacional de Supercómputo del Sureste de México (LNS-BUAP) of the CONACyT network of national laboratories for the computer resources and support provided.