Hydrophobic Ligand Entry and Exit Pathways of the CB1 Cannabinoid

Subscriber access provided by Binghamton University | Libraries

Article

Hydrophobic ligands entry and exit pathways of CB1 cannabinoid receptor Jakub Jakowiecki, and Slawomir H. Filipek J. Chem. Inf. Model., Just Accepted Manuscript • DOI: 10.1021/acs.jcim.6b00499 • Publication Date (Web): 28 Nov 2016 Downloaded from http://pubs.acs.org on December 5, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Chemical Information and Modeling is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Chemical Information and Modeling

Hydrophobic Ligands Entry and Exit Pathways of CB1 Cannabinoid Receptor

Jakub Jakowiecki1, Slawomir Filipek1 * 1

Faculty of Chemistry, Biological and Chemical Research Centre, University of Warsaw, ul. Pasteura 1, 02-093 Warsaw, Poland

* Corresponding author: Slawomir Filipek e-mail: [email protected]

The authors declare no competing financial interest.

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT It has been reported, that some hydrophobic ligands of G-protein-coupled receptors (GPCRs) access the receptor’s binding site from the membrane rather than from bulk water. In order to identify the most probable ligand entrance pathway into CB1 receptor we performed several Steered Molecular Dynamics (SMD) simulations of two CB1 agonists, THC and anandamide, pulling them from the receptor’s binding site with constant velocity. The four main directions of ligand pulling where probed: between helices TM4-TM5, TM5-TM6, TM7-TM1/TM2 and towards the bulk water. The smallest forces were measured during pulling between TM7TM1/TM2 helices. We have also performed Supervised Molecular Dynamics (SuMD) simulations for both anandamide and THC entering CB1 receptor’s binding site and found the same pathway as in pulling simulations. The residues F1742.61 and F1772.64 (both on TM2 helix) were involved in gating mechanism and, by forming π-π interactions with ligand molecules, facilitated ligand orientation required for passage. Using SuMD we also found alternative binding site for THC. The results of mutagenesis studies evidencing that residues F1742.61 and F1772.64 are important for CB1 ligand binding are in agreement with our observations.


Page 2 of 30

Page 3 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION The cannabinoid receptor subtype CB1 is expressed at high levels in central nervous system (CNS) and constitute a part of the endocannabinoid system, which is responsible for regulation of various physiological processes including pain-sensation, cognition, memory, appetite, mood, and metabolic functions.1 In physiological processes CB1 receptor is activated by its endogenous ligands - anandamide (AEA) and 2-arachidonoylglycerol (2-AG). CB1 is also sensitive to various compounds present in cannabis plant (ie. phytocannabinoids) of which some are its agonists (e.g. THC - tetrahydrocannabinol), some are antagonists (e.g. THCV tetrahydrocannabivarin) and some are allosteric modulators (e.g. CBD - cannabidiol, negative allosteric modulator) and therefore CB1 mediates the majority of cannabis medical effects and is responsible for its psychoactive properties. Finally, there is a group of synthetic compounds capable of binding to CB1 receptor and inducing its activation. Such compounds are called synthetic cannabinoids. Some of them are structurally related to THC and are classified as classical cannabinoids (e.g. HU-210 and CP 55,940) while others are structurally distinct from THC (e.g. JWH-018 and WIN 55,212-2) but all of them are highly hydrophobic compounds (Figure 1). The crystal structures of cannabinoid receptors are not available so the computational studies are based on homology models or the structure–activity relationship (SAR) studies. Liu et al. modeled CB1 based on rhodopsin template and performed enrichment study using over 3000 compounds.2 They tested various models of CB1 binding site and selected the best one for usage in virtual screening. Nguyen et al. performed SAR studies on CB1 receptor allosteric modulators.3 Such studies may provide the basis for further optimization of such modulators. Since there is no structure of N-terminus of any GPCR apart from rhodopsin, currently there is no possibility of such studies apart from SAR. The recent study on CB2 receptor complexes (homology modeled on adrenergic receptor β2AR) with agonist and inverse agonist were done by Hu et al.4 They employed 100 ns molecular dynamics (MD) simulations and found a breakdown of the “ionic lock” between helices TM3 and TM6, as well as the outward/inward movements of transmembrane domains of the active CB2 that bind with G protein. Their results also indicated that W6.48 in TM6 and residues in TM4 (V4.56-L4.61) contributed greatly to the binding of the agonist. In another study by Alqarni et al., the molecular modeling and ligand docking combined with site-directed mutagenesis experiments allowed to find residues V3.32 and L5.41 to be important for ligand binding and downstream signaling in CB2 receptor.5 Due to high hydrophobicity of CB1 ligands and the fact that CB1 receptor has a large N-terminal domain which possibly shields the receptor from the extracellular environment, the ligand approach from the extracellular milieu seems unlikely. For this reason an alternative ligand entrance pathway via lipid bilayer has been proposed for cannabinoid receptors and for other GPCRs which bind hydrophobic ligands.6 In this study we revealed the possible entry and exit pathways of two CB1 agonists, anandamide and THC, using a combination of Supervised 3 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molecular Dynamics (SuMD) with classical Molecular Dynamics (MD) as well as Steered Molecular Dynamics (SMD).

Figure 1. CB1 receptor ligands. (A) Anandamide (AEA), (B) 2-arachidonoylglycerol (2-AG), (C) THC ((-)-∆9-tetrahydrocannabinol), (D) THCV (tetrahydrocannabivarin) (E) CBD (cannabidiol) (F) HU-210 (G) CP55,940, (H) JWH-018, (I) WIN55,212-2. The alkyl chain in ligand molecules is relevant for its properties. The longer the chain is, the stronger agonistic properties of the compound are. HU-210 having alkyl chain 7-carbon atoms long is 100-800 times more potent CB1 receptor agonist than THC (alkyl chain 5-carbons long), while THCV (3carbons chain) is CB1 antagonist.

Methods CB1 homology model building. Since the crystallographic structure of CB1 receptor has not been determined we constructed a homology model encompassing residues 40-412. For homology model construction we used S1PR1 (sphingosine-1-phosphate receptor type 1, PDB ID: 3V2Y)7 and LPAR1 (lysophosphatydylic acid receptor type 1, PDB ID: 4Z34)8 crystallographic structures as templates. We applied Modeller 9.14 program9 for homology modeling using multiple templates method implemented in GPCRM service.10 Smaller loops were refined with YASARA v.16.211 and longer with Rosetta in ab initio approach.12 N-terminus (starting from residue K40) was refined with Rosetta. Obtained CB1 model was placed in


Page 4 of 30

Page 5 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


hydrated POPC bilayer and equilibrated for 50 ns. Details about CB1 homology model generation can be found in Supporting Information. Ligand docking. To build ligands and perform docking the Maestro and Glide programs from the Schrödinger Suite release 2015-3 were employed. In all docking experiments we used Glide, also from Schrödinger suite, using flexible ligands and a rigid receptor grid which is a default method in Glide. For the preliminary docking experiments the CB1 receptor binding site was defined as a cubic box (size: 20Å), a centroid of the following residues: F174, F177, F189, M363. We chose those residues since they were crucial for ligand binding according to mutagenesis studies.13, 14 In the later docking experiments we used the ligand position obtained from MD simulations as a reference for defining the ligand binding site. Preparation for molecular dynamics simulations. Quantum-mechanical charges for molecular dynamics were calculated using Jaguar (Schrödinger suite) with functional B3LYP and 6-311G** basis set. The parameter files for ligands were obtained using ParamChem server employing CGenFF (CHARMM General Force Field) for small molecules.15 All subsequent simulations were performed in a hydrated 1-palmitoyl-2-oleoyl-3-phosphocholine (POPC) bilayer using the NAMD v.2.11 simulation package with CUDA support and CHARMM36 force field. Charmm-gui service (http://www.charmm-gui.org/)16-18 was used for NAMD input files generation. MD simulation of THC in lipid/water environment. A single THC molecule was equilibrated (50 ns MD simulation) in water/POPC bilayer system. At the beginning of equilibration THC was placed in the water layer. After 23 ns THC migrated from water phase to POPC bilayer, and remained there for the rest of simulation, however, it remained close to water/bilayer interface with polar OH group oriented towards bulk water. Then, we equilibrated ten THC molecules in hydrated POPC bilayer for another 50 ns. Starting positions of all THC molecules were in POPC bilayer. None of THC molecules migrated from lipid bilayer to water phase during the simulation. Pulling of ligands from CB1 receptor binding site. We performed simulations of ligand pulling from CB1 receptor for anandamide and THC using SMD approach in NAMD program. We explored four directions of pulling for each ligand and for each direction we performed three simulations. For all SMD simulation a constant velocity (v = 0.31 m/s) and a virtual spring with force constant k = 69.5 pN/Å were applied. The starting poses for SMD simulations were generated by performing the unbiased (with no restraints or external forces) MD simulation (50 ns for each ligand) of ligand-receptor complexes obtained from docking. Supervised Molecular Dynamics. To simulate the ligand’s entrance we used modified Supervised Molecular Dynamics (SuMD) approach described by Sabbadin et al.19 This is a special supervision tabu-like algorithm, which monitors distance between the ligand and the receptor binding site and restart simulation from the checkpoint if that distance increased too much (Scheme 1). The supervision algorithm was implemented by us in NAMD program. 5 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scheme 1. Supervision algorithm for MD simulations of ligand entrance to the receptor binding site. First, we performed a short 40 ns unsupervised MD simulation for THC located close to TM7-TM1/TM2 interface to find out whether the position of ligand is stable in that location and to obtain the starting pose for SuMD simulation of THC entrance. For SuMD simulation of anandamide entrance to CB1 receptor a membrane lipid molecule (POPC) located in TM7-TM1 crevice was replaced with anandamide and the system was equilibrated. In the same way we also simulated THC and anandamide entrance between helices TM4-TM5 and between TM5TM6, however those attempts were unsuccessful. In this paper the Ballesteros and Weinstein numbering scheme was used.20

Results Simulation of THC behavior in water/lipid bilayer interface. Before making attempt to identify the ligand’s entrance pathway into the CB1 receptor we tried to determine in which part of the water/membrane system the concentration of THC is the highest. Results of simulations confirmed that THC prefers lipid bilayer than bulk water. Those results are in agreement with low water solubility of THC in water (0.0028 mg/ml, T=23 °C).21 This is also in agreement with postulated lipid pathway for THC entrance to the receptor binding site and implies, that the entrance of the ligand is most probably between transmembrane helices, however, in close proximity of water/membrane interface (Figure 2).


Page 6 of 30

Page 7 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. Ten THC molecules (yellow) in membrane/water environment – a frame from MD trajectory after 50 ns of simulation. Membrane lipids (POPC) were not shown to improve image clarity. All THC molecules are dissolved in lipid bilayer with their polar oxygen atoms forming contacts with the water layer, therefore in the membrane part exposed to water the THC concentration is the highest. Pulling anandamide and THC out of CB1 receptor. Details on stabilizing interactions of the extracellular side of CB1 receptor can be found in Supporting Information Figure S1. Details on ligand docking are described in Methods. For finding the putative THC exit pathway from the receptor it was crucial to identify the transmembrane helices between which the ligand passage would be the most favorable. For this purpose SMD simulations were performed for pulling THC and anandamide from their binding sites in CB1 receptor with constant velocity. The four distinct directions for ligand pulling were probed: between TM4-TM5, TM5-TM6, TM7TM1/TM2 and between ECH1-ECH2, towards extracellular side. We have chosen those four main directions after analyzing interfaces between transmembrane helices of CB1. We identified all residues with short side chains (like GLY and ALA) to find all spots between transmembrane helices where the ligand could potentially crawl out. Although helix pairs TM4-TM5, TM5TM6, and TM7-TM1 were selected in this process, we have also tried some other directions of ligand puling (TM6-TM7 and TM1-TM2) but we were unable to observe ligand’s exit along those pathways. To include the possibility of ligand’s exit to the bulk water we added one additional direction to our study - parallel to the membrane normal between N-terminal helices ECH1-ECH2 (Table 1 and Supporting Information Figure S2). Selection of a group of ligand atoms as a pulling group was based on the orientation of the ligand in the binding site: ligand atoms located closer to the exit via particular direction were pulled. For each direction the three SMD simulations were performed. Table 1. SMD simulations of ligand pulling from CB1 receptor in different directions pulling orientations. The ligands orientations during particular SMD simulations are specified. 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Pulling Direction Between TM5-TM6 Between TM4-TM5 Between ECH1-ECH2

Anandamide Hydrophobic tail first Hydrophobic tail first Terminal -OH group first

Between TM7-TM1/TM2 Terminal -OH group first

THC Hydrophobic tail first Hydrophobic tail first Methyl group (CH3-C=C) first Cyclohexenyl ring first

During SMD simulations of both ligands the lowest force values were measured when pulling between TM7 and TM1/TM2 (Figure 3, yellow line). This is especially seen for anandamide pulling where the force was nearly stable via this route. In contrast, the pulling of THC generated larger forces and the initial increase of force was independent of the pulling direction. However, for TM7-TM2 (and then TM7-TM1) direction the highest force peaks were lower compared to force peaks obtained for pulling in other directions. Additionally, anandamide departure along TM7-TM2 pathway did not cause significant receptor structure distortion and the only structural change observed was the rearrangement of residues F1742.61 and F1772.64 resembling the gate opening. In the stable position of anandamide after equilibration (Figure 4A) these two residues stabilize anandamide in the binding site due to π-π interactions they form with ligand’s double bonds (F1742.61 stacks with double bond C8=C9 of the ligand and F1772.64 with C5=C6). When anandamide is being pulled between TM7 and TM2, the side chains of F1742.61 and F1772.64 move apart allowing the ligand to pass between them (Figure 4B).


Page 8 of 30

Page 9 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Force vs. time exemplary plots of ligand pulling out between TM4-TM5 (red), TM5TM6 (purple), ECH1-ECH2 (blue) and TM7-TM2 (yellow). (A) THC; (B) anandamide.

Figure 4. Role of residues F174 and F177 for anandamide binding and pulling out. (A) Anandamide in CB1 binding site after 50 ns of equilibration. Residues F174 and F177 stabilize the ligand’s pose by π-π interactions; (B) A frame from SMD simulation - anandamide crawls between aromatic rings of F174 and F177. Red lines indicate π-π interactions. 9 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 30

Supervised ligand entrance to the receptor. To reveal which path is the optimal for the ligand’s entrance we performed supervised MD simulation of the entry of anandamide and THC into CB1 receptor. The concept of a similar supervision algorithm for simulating ligand binding events during its entry into A2A adenosine receptor has been employed recently by Sabbadin et al.19 Authors simulated the ligand recognition mechanisms for the following A2A antagonists: T4G, T4E, ZM241385 and caffeine, and were able to reproduce the crystallographic pose for each ligand with high accuracy. Our attempts to simulate ligand entrance between TM4 and TM5 as well as between TM5 and TM6 proved fruitless although we employed different initial orientations of ligands toward the receptor. During a simulation of entrance between TM7-TM1/TM2 we determined that the 'alkyl chain first' orientation is favored for both ligands. For alternative orientations (Table 2) we observed only partial ligand entrance or no entrance at all. Based on our simulations the ligand entrance can be divided into two steps: 1) ligand entrance to the crevice between TM7 and TM1, and 2) ligand passage between TM7 and TM2, which is the actual rate limiting step. Table 2. Comparison of simulation times for ligand entrance in different orientation towards TM7-TM1/TM2 gate. Simulation Anandamide entrance hydrophilic head first

Time elapseda 145 ns

Comment Although the ligand approached the receptor cavity it had tendency to leave it each time the SuMD algorithm was deactivated.

Anandamide entrance - alkyl chain first

25 ns

Alkyl chain entered receptor between residues F174 and F177 just after the simulation started and the rest of molecule followed soon afterwards displacing water molecules from the binding site (Supporting Information Movie S1).

THC entrance - dimethylpiran ring first

n/a (>180ns)

No ligand entrance during 180 ns of simulation (THC molecule entered only partially).

THC entrance - pentyl chain first

40 ns

Alkyl chain entered receptor between residues F174 and F177 just after the simulation started and the rest of molecule followed some time later (Supporting Information Movie S2).

a

time of simulation elapsed till the whole ligand entered the receptor, but not necessarily occupying the binding site - residues F174 and F177 were arbitrarily defined as the receptor border.


Page 11 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


During the anandamide entrance the ligand’s polar head was oriented towards extracellular water for most of the simulation time, while the hydrophobic alkyl chain crawled between residues F1742.61 and F1772.64 (Figure 5). The π-π interactions forming between these residues and the ligand molecule (similar to those shown on Figure 4B) facilitated ligand’s orientation required for passing between TM7 and TM2. The rearrangement of residues W3566.48 and F2003.36 after 40 ns of SuMD simulation was the first activation event observed during the simulation. Although the values of χ1 and χ2 of W3566.48 and F2003.36 did not change significantly, the stacking between their aromatic rings was broken.

Figure 5. Frames from SuMD simulation trajectory of anandamide (yellow) entering CB1 receptor between TM7-TM1/TM2 gate. 0 ns means end of equilibration. Anandamide enters the receptor with its hydrophobic terminus, while its polar head stays in contact with extracellular solution for the most of the simulation time. The residues F1742.61 and F1772.64 are involved in gating mechanism – ligand’s entrance can occur only when their side chains are in specific orientation (13 ns). Starting from 25 ns of SuMD simulation the whole ligand molecule is inside the receptor (behind residues F1742.61 and F1772.64). After 40 ns of SuMD simulation the ligands hydrophobic chain is located deep inside a binding pocket, contacting residues W3566.48 and



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 30

F2003.36 which constitute the rotamer toggle switch. Breaking of the π-π aromatic stacking between these residues was the first activation event observed. The SuMD simulation of THC entrance into CB1 receptor was preceded by 40 ns of unbiased MD simulation (equilibration) of THC located in the crevice between TM7 and TM1/TM2. During this simulation the ligand rearranged residues F1742.61 and F1772.64 and induced movement of their aromatic side chains, resembling opening a gateway (Supporting Information Figure S3). The phenolic hydroxyl group of THC formed a hydrogen bond with T377 hydroxyl group (Supporting Information Figure S3B). The pentyl chain of THC crawled between F174 and F177 aromatic rings at the beginning of SuMD trajectory and the rest of ligand molecule followed soon afterwards (Figure 6).

Figure 6. Supervised Molecular Dynamics (SuMD) simulation, starting from the end of MD simulation of THC entering the TM7-TM1/TM2 crevice of CB1 receptor. (A) THC enter the CB1 receptor binding site in similar manner to anandamide, with ‘hydrophobic tail first orientation', while the polar -OH group maintains contact with extracellular solution. The residues F1742.61 and F1772.64 are involved in gating mechanism. (B) After passing beyond these gating residues the THC molecule reaches the “alternative” binding site formed by residues L86, F89, F174, F177, F181, R182, K376, F379, M384 (C) and remains there for over 100 ns of unbiased (unsupervised) MD simulation. (D) Receptor-ligand interactions stabilizing THC in the “alternative” binding site - VdW interactions in green and π-π interactions in red.


Page 13 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Surprisingly, after crawling beyond residues F1742.61 and F1772.64 the THC molecule stopped instead of going deeper into the binding pocket to the position found in docking studies. After several unsuccessful attempts of moving the ligand further with SuMD algorithm we performed three parallel unsupervised MD simulations to check how THC will behave in this location. During the first simulation the ligand’s pose was maintained for the whole 100 ns simulation and no activation events were observed. In the second one, after 40 ns MD simulation, THC molecule left the receptor, in a similar fashion as it entered. In the third 100 ns simulation the ligand changed its orientation significantly but during next 5 ns it returned to its initial pose which was maintained for the rest of simulation. It indicates that THC molecule in that “alternative” binding site (Figure 6 C-D) is stable. After 36 ns of the third simulation we observed a change of W3566.48 rotamer from g(-) to trans (Figure 7 and 8A) as well as an increased water influx into the receptor (Figure 8B) which are indicators of the receptor activation.22, 23

Figure 7. Rearrangement of the rotamer toggle switch during the third MD simulation. Residues W3566.48 and F2003.36 are shown in red, gating phenylalanines in yellow, and water oxygen atoms as blue balls. (A) Position before a switch change, W3566.48 in g(-) conformation; (B) after a switch change, W3566.48 in trans conformation, number of water molecules increased.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 30

Figure 8. Activation events observed during one of three unsupervised MD simulations performed for THC located in “alternative” binding site of CB1 receptor; (A) changes of dihedral angles, χ1 (red) and χ2 (blue) for residue W3566.48 observed during 3rd MD simulation; (B) number of water molecules penetrating the receptor's interior (within 6 Å from residues D163, F200, S203, W356, N389, and N393) measured during three MD simulations - 1st simulation (in green), 2nd simulation (in red), 3rd simulation (in blue). Increased water influx (in blue) was observed only for the simulation with action of a rotamer toggle switch (panel A). Red line in 2nd simulation is ended at 40 ns since THC exited the receptor.

DISCUSSION Most of the published CB1 receptor homology models used human β2 adrenergic receptor (β2AR) as the template (26% identity and 47% similarity of the sequence), some used bovine rhodopsin, turkey β1 adrenergic receptor and human adenosine A2A receptor, and such models are still used for computational studies24 even though new crystallographic structures with higher sequence similarity to CB1 are available these days. All those models had the N-terminus truncated in most cases covering only residues between P113 and F412. Since we constructed CB1 receptor homology model based on two templates, human S1PR17 (34% identity, 58% similarity) and human LPAR18 (34% identity, 57% similarity) and covering sequence from K40 14 ACS Paragon Plus Environment

Page 15 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


to F412, we believe our model is more accurate and reliable than the previous ones. We tested the obtained CB1 model by docking of classical cannabinoids and the poses we obtained were very similar to those reported before.14, 25 The presence of long N-terminus, containing EC0 loop and ECH1 and ECH2 helices reduces receptor’s accessibility from the extracellular side. This finding is in agreement with hypothesis, that cannabinoid receptor’s ligands enter the binding site via the lipid bilayer, rather than from extracellular side. Such ligand entrance pathway is similar to other GPCRs which bind hydrophobic ligands.6 Another argument supporting the lipid entrance pathway of CB1 ligands is that anandamide can be synthesized on demand from the membrane components by N-acetyltransferases (NATs)26-28 and when no longer needed is metabolized by fatty acid amide hydrolase (FAAH), the enzyme capable of penetrating the membrane,28 so the endocannabinoid system can function fast and efficiently without its signaling molecules ever leaving the membrane environment. The ligand entrance pathway between TM7 and TM1 has been proposed for S1P1 receptor7 and for rhodopsin.29 The other pathway was suggested for 2-arachidonoylglycerol (2-AG) entering cannabinoid CB2 receptor model based on rhodopsin. The ligand was entering CB2 receptor between helices TM6-TM7 and with its polar head first.30 During nearly 2 µs MD simulation 2-AG was not able to enter the binding site and still 2/3 of ligand’s length was outside. Here, we propose not only another pathway, via TM7-TM1/TM2, but also the opposite way of entering the receptor with the hydrophobic tail first. In these circumstances it was possible to obtain entrance of the whole anandamide into the CB1 binding site. The ligand entrance between TM7-TM1/TM2 transmembrane helices containing a gate composed of residues F1742.61 and F1772.64 is located close to the water/membrane interface where the probability of finding the ligand is the highest. Although most of CB1 ligands are highly hydrophobic they also contain a hydrophilic moiety. The residues F1742.61 and F1772.64 are not only important for ligand’s entry and exit but they can also stabilize ligands in the binding site. We found that they directly interact with anandamide in the binding site but not with THC in CB1 binding site (the pose resulting from docking). However, mutagenesis results indicate that those gating residues are relevant not only for anandamide but also for THC binding to CB1 receptor.14 It can be explained by the role these residues play during ligand entrance, or, on the other hand, by proposing an “alternative” binding site for THC which was found by us during SuMD simulation of THC entrance. In favor of the second hypothesis, two activation events occurred in the “alternative” position of THC: action of rotamer toggle switch and influx of water molecules into the receptor interior (Figures 7 and 8). In our simulations we did not observe breaking of the ionic lock, as found by Hu et al.4 in 100 ns simulations of CB2 with agonists, neither a significant outward/inward movements of transmembrane domains in any of CB1 simulations, even those longer than 100 ns. It is because in our simulations agonists were in constant movement entering or exiting the receptor binding site so we observed only initial signs of activation. Also there were no direct interactions 15 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 30

between residues I2474.56-L2524.61 of CB1 receptor and both ligands investigated by us. In general, there were no interactions with any of residues of helix TM4. However, the interactions with residues I267 and F268 from extracellular loop 2 (EC2) were identified for both studied CB1 ligands. The interactions with W3566.48 was observed only for ligands containing a long alkyl chain, such as anandamide, HU-210 or CP55940 (results not shown) penetrating deep inside the receptor. We also compared our results with those of Alqarni et al.5 done on CB2, concluding that residue V3.32 is also present in CB1 receptor (V1963.32) and, similarly to CB2, it also participates in CB1 ligand binding by formation of a hydrophobic interactions with ligands alkyl chain stabilizing both anandamide and THC in their binding pocket. Two alternative approaches, SMD and SuMD, used for ligand exit and entry, provided the same pathway through TM7-TM1/TM2 crevice, and also orientation of the ligand was the same during its exit and entry. The hydrophobic tail of the ligand first penetrated the receptor during entry and then the rest of molecule passed through the gate with polar head at the end, while during the exit a pulling of the polar head generated the lowest forces. The fact, that THC required longer time (40ns) to enter to the CB1 receptor than anandamide (25 ns) and, additionally, THC did not achieved its final binding pose predicted by docking while anandamide achieved its final binding mode, stays in agreement with relative pharmacological properties of those compounds: anandamide has shorter onset time than THC.31 The longer simulation times may be required to elucidate the complete THC binding process. Most likely the THC requires twisting inside a receptor which is difficult for such rigid and large molecule, or THC had difficulty with displacing all water molecules initially present in the binding site. It is also possible that the “alternative binding site” we identified for THC is the final and the only binding site for this ligand (Figure 6 C,D).

CONCLUSIONS Hydrophobic ligands attain an access to CB1 receptor via the lipid bilayer. Anandamide and THC enter the receptor by binding to a broad crevice formed between helices TM7 and TM1 and then crawling between TM7 and TM2. A passage between the latter is the rate limiting step. Both ligands enter CB1 receptor with 'alkyl chain first' orientation. Residues F1742.61 and F1772.64 form π-π interactions with ligand molecule and are involved in gating mechanism, allowing only molecules with certain structural properties to get into the receptor. The importance of residues F1742.61 and F1772.64 for ligand binding has been confirmed by mutagenesis studies.14 Based on our results the residue T377 forms H-bond with hydroxyl group of THC molecule located on TM7-TM1/TM2 pathway and, additionally with hydrophobic residues F174, F177, F381 and M384, induce a distinct position of THC molecule. Further mutagenesis studies are required to confirm the hypothesis regarding the presence of a transient


Page 17 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


H-bond between residue T377 and the ligand’s hydroxyl group and the role of the “alternative” binding site. Although we concluded that anandamide and THC (and probably most of classical cannabinoids) attain access to CB1 binding site from the membrane by passing between residues F1742.61 and F1772.64, there are also larger CB1 ligands containing more polar groups (like Rimonabant or WIN-55,212-2) for which such pathway seems less probable. However, it should be noted that a large part of ligand entrance channel between TM7-TM1/TM2 helices is located on extracellular side of the receptor and therefore the bigger and more hydrophilic CB1 ligands could still enter the receptor through the same channel but in slightly different manner. We believe that the long N-terminus of CB1 receptor plays an important role for the ligand binding process. It has been reported that N-terminus of CB1 receptor contains the cannabidiol (CBD) allosteric binding site, associated with residues C98 and C107.32 For this reason a structure determination/modeling of the whole CB1 N-terminus is of a paramount importance. Considering that the ligand effectiveness is determined not only by its interactions within the binding site but also during its way into the receptor’s binding site, the knowledge about the mechanism of ligand entrance is indispensable for new ligands design.

SUPPORTING INFORMATION Movies S1 and S2 illustrating entrance of anandamide and THC into CB1 receptor; Description of CB1 homology model construction; Figures S1-S3; and supporting references.

ACKNOWLEDGEMENTS Part of the work was done at the Center and Interdisciplinary Centre for Mathematical and Computational Modeling in Warsaw as grant G07-13. S.F. received funding from National Center of Science, Poland, grant no. 2011/03/B/NZ1/03204. We also thank Przemysław Miszta for his participation in SuMD algorithm implementation. J.J. and S.F. participate in the European COST Action CM1207 (GLISTEN). REFERENCES (1) Howlett, A. C.; Barth, F.; Bonner, T. I.; Cabral, G.; Casellas, P.; Devane, W. A.; Felder, C. C.; Herkenham, M.; Mackie, K.; Martin, B. R.; Mechoulam, R.; Pertwee, R. G., International Union of Pharmacology. XXVII. Classification of Cannabinoid Receptors. Pharmacol. Rev. 2002, 54, 161-202. (2) Liu, H.; Patel, R. Y.; Doerksen, R. J., Structure of the Cannabinoid Receptor 1: Homology Modeling of its Inactive State and Enrichment Study Based on CB1 Antagonist Docking. MedChemComm 2014, 5, 1297-1302. 17 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 30

(3) Nguyen, T.; German, N.; Decker, A. M.; Li, J. X.; Wiley, J. L.; Thomas, B. F.; Kenakin, T. P.; Zhang, Y., Structure-Activity Relationships of Substituted 1H-Indole-2-Carboxamides as CB1 Receptor Allosteric Modulators. Bioorg. Med. Chem. 2015, 23, 2195-2203. (4) Hu, J.; Feng, Z.; Ma, S.; Zhang, Y.; Tong, Q.; Alqarni, M. H.; Gou, X.; Xie, X. Q., Difference and Influence of Inactive and Active States of Cannabinoid Receptor Subtype CB2: From Conformation to Drug Discovery. J. Chem. Inf. Model. 2016, 56, 1152-1163. (5) Alqarni, M.; Myint, K. Z.; Tong, Q.; Yang, P.; Bartlow, P.; Wang, L.; Feng, R.; Xie, X. Q., Examining the Critical Roles of Human CB2 Receptor Residues Valine 3.32 (113) and Leucine 5.41 (192) in Ligand Recognition and Downstream Signaling Activities. Biochem. Biophys. Res. Commun. 2014, 452, 334-339. (6) Hurst, D. P.; Schmeisser, M.; Reggio, P. H., Endogenous Lipid Activated G Protein-Coupled Receptors: Emerging Structural Features from Crystallography and Molecular Dynamics Simulations. Chem. Phys. Lipids 2013, 169, 46-56. (7) Hanson, M. A.; Roth, C. B.; Jo, E.; Griffith, M. T.; Scott, F. L.; Reinhart, G.; Desale, H.; Clemons, B.; Cahalan, S. M.; Schuerer, S. C.; Sanna, M. G.; Han, G. W.; Kuhn, P.; Rosen, H.; Stevens, R. C., Crystal Structure of a Lipid G Protein-Coupled Receptor. Science 2012, 335, 851-855. (8) Chrencik, J. E.; Roth, C. B.; Terakado, M.; Kurata, H.; Omi, R.; Kihara, Y.; Warshaviak, D.; Nakade, S.; Asmar-Rovira, G.; Mileni, M.; Mizuno, H.; Griffith, M. T.; Rodgers, C.; Han, G. W.; Velasquez, J.; Chun, J.; Stevens, R. C.; Hanson, M. A., Crystal Structure of Antagonist Bound Human Lysophosphatidic Acid Receptor 1. Cell 2015, 161, 1633-1643. (9) Webb, B.; Sali, A., Comparative Protein Structure Modeling Using Modeller. Curr. Protoc. Bioinformatics. 2014, 47, 5 6 1-32. (10) Latek, D.; Pasznik, P.; Carlomagno, T.; Filipek, S., Towards Improved Quality of GPCR Models by Usage of Multiple Templates and Profile-Profile Comparison. PLOS ONE 2013, 8, e56742. (11) Krieger, E.; Vriend, G., New Ways to Boost Molecular Dynamics Simulations. J. Comput. Chem. 2015, 36, 996-1007. (12) Yarov-Yarovoy, V.; Schonbrun, J.; Baker, D., Multipass Membrane Protein Structure Prediction Using Rosetta. Proteins 2006, 62, 1010-1025. (13) Shim, J. Y., Understanding Functional Residues of the Cannabinoid CB1. Curr. Top. Med. Chem. 2010, 10, 779-798. (14) Shim, J. Y.; Bertalovitz, A. C.; Kendall, D. A., Identification of Essential Cannabinoid-Binding Domains: Structural Insights into Early Dynamic Events in Receptor Activation. J. Biol. Chem. 2011, 286, 33422-33435. (15) Vanommeslaeghe, K.; Raman, E. P.; MacKerell, A. D., Jr., Automation of the CHARMM General Force Field (CGenFF) II: Assignment of Bonded Parameters and Partial Atomic Charges. J. Chem. Inf. Model. 2012, 52, 3155-3168. (16) Brooks, B. R.; Brooks, C. L., 3rd; Mackerell, A. D., Jr.; Nilsson, L.; Petrella, R. J.; Roux, B.; Won, Y.; Archontis, G.; Bartels, C.; Boresch, S.; Caflisch, A.; Caves, L.; Cui, Q.; Dinner, A. R.; Feig, M.; Fischer, S.; Gao, J.; Hodoscek, M.; Im, W.; Kuczera, K.; Lazaridis, T.; Ma, J.; Ovchinnikov, V.; Paci, E.; Pastor, R. W.; Post, C. B.; Pu, J. Z.; Schaefer, M.; Tidor, B.; Venable, R. M.; Woodcock, H. L.; Wu, X.; Yang, W.; York, D. M.; Karplus, M., CHARMM: The Biomolecular Simulation Program. J. Comput. Chem. 2009, 30, 1545-1614. (17) Jo, S.; Kim, T.; Iyer, V. G.; Im, W., CHARMM-GUI: A Web-Based Graphical User Interface for Charmm. J. Comput. Chem. 2008, 29, 1859-1865. (18) Lee, J.; Cheng, X.; Swails, J. M.; Yeom, M. S.; Eastman, P. K.; Lemkul, J. A.; Wei, S.; Buckner, J.; Jeong, J. C.; Qi, Y.; Jo, S.; Pande, V. S.; Case, D. A.; Brooks, C. L., 3rd; MacKerell, A. D., Jr.; Klauda, J. B.; Im, W., CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM Simulations Using the CHARMM36 Additive Force Field. J. Chem. Theory Comput. 2016, 12, 405-413.


Page 19 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(19) Sabbadin, D.; Moro, S., Supervised Molecular Dynamics (SuMD) as a Helpful Tool to Depict GPCR-Ligand Recognition Pathway in a Nanosecond Time Scale. J. Chem. Inf. Model. 2014, 54, 372376. (20) Ballesteros, J. A.; Weinstein, H., Integrated Methods for the Construction of Three-Dimensional Models and Computational Probing of Structure-Function Relations in G Protein-Coupled Receptors. Methods Neurosci. 1995, 25, 366-428. (21) Garrett, E. R.; Hunt, C. A., Physiochemical Properties, Solubility, and Protein Binding of Delta9Tetrahydrocannabinol. J. Pharm. Sci. 1974, 63, 1056-1064. (22) Latek, D.; Kolinski, M.; Ghoshdastider, U.; Debinski, A.; Bombolewski, R.; Plazinska, A.; Jozwiak, K.; Filipek, S., Modeling of Ligand Binding to G Protein Coupled Receptors: Cannabinoid CB1, CB2 and Adrenergic beta2AR. J. Mol. Model. 2011, 17, 2353-2366. (23) Trzaskowski, B.; Latek, D.; Yuan, S.; Ghoshdastider, U.; Debinski, A.; Filipek, S., Action of Molecular Switches in GPCRs - Theoretical and Experimental Studies. Curr. Med. Chem. 2012, 19, 1090-1109. (24) Scott, C. E.; Ahn, K. H.; Graf, S. T.; Goddard, W. A., 3rd; Kendall, D. A.; Abrol, R., Computational Prediction and Biochemical Analyses of New Inverse Agonists for the CB1 Receptor. J. Chem. Inf. Model. 2016, 56, 201-212. (25) Shim, J. Y.; Padgett, L., Functional Residues Essential for the Activation of the CB1 Cannabinoid Receptor. Methods Enzymol. 2013, 520, 337-355. (26) Di Marzo, V., Targeting the Endocannabinoid System: To Enhance or Reduce? Nat. Rev. Drug Discov. 2008, 7, 438-455. (27) Piomelli, D., The Molecular Logic of Endocannabinoid Signalling. Nat. Rev. Neurosci. 2003, 4, 873-884. (28) Bracey, M. H.; Hanson, M. A.; Masuda, K. R.; Stevens, R. C.; Cravatt, B. F., Structural Adaptations in a Membrane Enzyme that Terminates Endocannabinoid Signaling. Science 2002, 298, 1793-1796. (29) Hildebrand, P. W.; Scheerer, P.; Park, J. H.; Choe, H. W.; Piechnick, R.; Ernst, O. P.; Hofmann, K. P.; Heck, M., A Ligand Channel through the G Protein Coupled Receptor Opsin. PloS one 2009, 4, e4382. (30) Hurst, D. P.; Grossfield, A.; Lynch, D. L.; Feller, S.; Romo, T. D.; Gawrisch, K.; Pitman, M. C.; Reggio, P. H., A Lipid Pathway for Ligand Binding is Necessary for a Cannabinoid G Protein-Coupled Receptor. J. Biol. Chem. 2010, 285, 17954-17964. (31) Razdan, R. K., Structure-Activity Relationships in Cannabinoids. Pharmacol. Rev. 1986, 38, 75149. (32) Laprairie, R. B.; Bagher, A. M.; Kelly, M. E.; Denovan-Wright, E. M., Cannabidiol is a Negative Allosteric Modulator of the Cannabinoid CB1 Receptor. Br. J. Pharmacol. 2015, 172, 4790-4805.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For Table of Contents use only

Hydrophobic ligands entry and exit pathways of CB1 cannabinoid receptor Jakub Jakowiecki, Slawomir Filipek


Page 20 of 30

Page 21 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TOC 260x105mm (200 x 200 DPI)

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1 236x164mm (200 x 200 DPI)


Page 22 of 30

Page 23 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2 158x120mm (200 x 200 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3 172x169mm (200 x 200 DPI)


Page 24 of 30

Page 25 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4 246x78mm (200 x 200 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 5 253x200mm (200 x 200 DPI)


Page 26 of 30

Page 27 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6 247x177mm (200 x 200 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7 239x162mm (200 x 200 DPI)


Page 28 of 30

Page 29 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 8 198x185mm (200 x 200 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scheme 1 422x278mm (72 x 72 DPI)


Page 30 of 30

Hydrophobic Ligand Entry and Exit Pathways of the CB1 Cannabinoid

Recommend Documents