Shared binding modes of M3 muscarinic agonists and ...

Page 1/14

Shared binding modes of M3 muscarinic agonists and antagonists mayincentivize development of novel IOP-reducing drugs: �ndings from in-silicoassaysMinjae J. Kim 

Department of Ophthalmology, The Hamilton Eye Institute, The University of Tennessee Health Science Center, Memphis, TN, 38163Monica M. Jablonski  ( [email protected] )

Department of Ophthalmology, The Hamilton Eye Institute, The University of Tennessee Health Science Center, Memphis, TN, 38163

Article

Keywords:

Posted Date: June 1st, 2022

DOI: https://doi.org/10.21203/rs.3.rs-1662403/v1

License: This work is licensed under a Creative Commons Attribution 4.0 International License.   Read Full License

Page 2/14

AbstractM3 muscarinic acetylcholine receptors (M3R) are G-protein-coupled receptors expressed in the ciliary muscle. Pilocarpine, an M3R agonist, reduces intraocularpressure (IOP) by causing contraction of the ciliary muscle, which leads to widening of the spaces in the trabecular meshwork and facilitates aqueous humor�ow out of the eye. Despite their IOP-reducing potentials, there are only four M3R agonists that have been FDA-approved, and the chemical features thatdifferentiate agonists from antagonists are poorly understood. In this study, we created a homology model of human M3R (hM3R) to analyze the bindingmodes that differentiate M3R agonists from antagonists, using such in-silico assays as high-throughput virtual screening and molecular dynamics. Our in-silico results suggest that a nitrogen ring with a positive charge is crucial for forming cation-π exchange with hM3R, regardless of agonism or antagonism.However, we discovered π-π stacking exchange that is exclusively found among M3R antagonists. Structural analysis suggests that this π-π stackingexchange leads to structural changes in one of the α-helices and causes instability in the cytoplasmic domain, where G protein interactions occur. Together,these �nding may prove useful in developing drugs that not only target but also selectively agonize hM3R to reduce IOP in glaucoma patients.

IntroductionPrimary open angle glaucoma (POAG) is a leading cause of blindness worldwide1. POAG is characterized by progressive optic nerve damage from retinalganglion cell death, which is typically associated with elevated intraocular pressure (IOP). Steady-state IOP is generated by the balance of aqueous humor(AH) production by the ciliary body (CB) and its drainage through the trabecular meshwork and to a lesser degree the uveoscleral or nonconventional pathway.An imbalance between the in�ow and out�ow of AH leads to change in IOP, which could lead to persistent pressure in the eye and compromise optic nervehealth2–4. Hence, decreasing IOP is often considered as the �rst-line treatment for POAG.

Muscarinic acetylcholine receptors are G-protein-coupled receptors (GPCR) with �ve different subtypes, all of which are expressed in the anterior chamber ofthe eye5,6. In particular, M3 muscarinic acetylcholine receptors (M3R) accounts for nearly 60% of all muscarinic receptors that are expressed in the ciliarymuscle6. It has been experimentally validated that M3R activation causes contraction of the ciliary muscle, which leads to widening of the spaces in thetrabecular meshwork and facilitates aqueous humor �ow out of the eye7–9. M3R activation can be achieved by small molecules that resemble its endogenousligand, acetylcholine. For example, pilocarpine is an FDA-approved drug that has been used to reduce IOP by agonizing M3R9. On the other hand, M3Rantagonists, such as atropine, have little effect on IOP and are contraindicated in glaucoma patients prescribed with pilocarpine10–13. There is experimentalevidence suggesting that muscarinic agonists and antagonists share the same binding pocket14,15, yet the binding modes that differentiate each other arepoorly understood.

In drug discovery and lead optimization, the computational method is a highly effective technique and has been widely applied16–20. Speci�cally, moleculardynamics (MD) simulation considers the �exibility of protein structures based on Newtonian principles and is often used to analyze the protein-ligandinteraction or the binding free energy collected during MD simulation21–24. These results can then be utilized for compound optimization. Molecularmechanics and generalized born surface area (MM/GBSA) method is also used to predict the binding free energy of a compound binding tomacromolecules25–27. MM/GBSA based on MD simulation results can be used to study stability of a protein-ligand complex and analyze the variation in thesensitivity of a ligand caused by mutations in amino acid residues of a protein28. When the structure of the protein is unknown, which is the case in our study,it needs to �rst be predicted. Homology modeling is considered the most accurate among the computational structure prediction methods in this situation29.In homology modeling, a template of the target protein, which is the peptide sequence of another protein with high sequence identity, must �rst be found. Aprotein sequence with over 30% identity to a known structure can often be predicted with an accuracy equivalent to a low-resolution X-ray structure30. The twosequences are then aligned, and a homology model is constructed. Post-modi�cation of the model is often needed for more reliable results.

In this study, we developed a homology model of human M3R (hM3R) based on the crystal structure of rat M3R in complex with tiotropium (a known M3Rantagonist). We then docked tiotropium in our homology model and compared its binding mode with tiotropium in rat M3R crystal structure to test the �delityof our homology model. Subsequently, we performed high-throughput virtual screening (HTVS) of 21 compounds, composed of 6 M3R agonists and 15 M3Rantagonists, to analyze differences in their binding modes. Finally, we performed MD simulations with the top scoring agonist and antagonist to analyzevariations in the mode of interaction with hM3R. Our in-silico results suggest that a tertiary amine with a positive charge is crucial for formation of cation-πexchange with hM3R regardless of agonism or antagonism. However, the M3R antagonists formed π-π stacking exchange with hM3R, which is absent in M3Ragonists because they lack aromatic structures. Structural comparison of hM3R suggests that this π-π stacking exchange leads to structural changes in oneof the α-helices and causes instability in the cytoplasmic domain of hM3R, where G protein interactions occur.

ResultsHomology modeling of hM3R

Using crystal structure of rat M3R (PDB: 4DAJ), a homology model of hM3R was constructed, whose sequence identity was 62 %. The predicted structure ofhM3R contains three different domains: extracellular (EC), transmembrane (TM), and cytoplasmic (CT) domains (Fig. 1A). Like most GPCRs, G proteininteraction with hM3R is predicted to take place in the CT domain upon agonism31-33. Upon superimposition of rat M3R crystal structure to hM3R homologymodel, we observed conservation of TM and EC domains (Fig. 1B). Our homology model was further re�ned using Protein Preparation and Loop Re�nementfeatures in Maestro (Schrödinger Inc., USA). Binding site analysis of our homology model was done using Sitemap feature in Maestro (Fig. 1C), and the top-ranking binding site can be found in the EC domain (Fig. 1D). This is consistent with the crystal structure of rat M3R, where tiotropium is in complex within theEC domain. Upon superimposition, tiotropium from the crystal structure �ts tightly inside the top-ranking binding site of hM3R identi�ed by Sitemap (Fig. 2A).Upon molecular docking of tiotropium to hM3R homology model using Glide (Schrödinger Inc., USA), the predicted binding mode is nearly identical to the

Page 3/14

binding mode of tiotropium obtained from the crystal structure of rat M3R (Fig. 2B). Upon comparison of the binding forces of tiotropium in the rat crystalstructure to our hM3R homology model, it was determined that they were nearly identical, including the presence of π-π stacking exchange betweentiotropium’s aromatic ring and tryptophan residues (Fig. 2C and 2D). Together, the results indicate that molecular docking with our hM3R homology model canpredict ligand binding with high �delity.

High-throughput virtual screening (HTVS) of M3R muscarinic agonists and antagonists

The 3D chemical structures of M3R agonists and antagonists were obtained from Drug Banks database, and their structures were further re�ned usingLigPrep in Maestro. The binding modes of top three scoring M3R agonists revealed that cation-π exchange can be found between a positively chargednitrogen ring and residues with aromatic sidechains, like TYR149, TRP504, and TYR507 (Fig. 3A). Another notable force is hydrogen bonding between anoxygen group and ASN508. The binding modes of top three scoring M3R antagonists also contained cation-π exchange between a positively charged nitrogenring and residues with TYR149, TRP504, and TYR507 (Fig. 3B). ASN 508 can also be seen interacting with oxygen molecules. However, compared to M3Ragonists, the antagonists all shared presence of π -π stacking exchange between their aromatic rings and TRP504. In fact, this interaction is absent in all ofthe hM3R agonists available in the database likely because they lack aromatic side groups in their chemical structures. Table 1 displays the results of HTVSthat lists the docking scores and checks presence of π- π stacking exchange, which is found in 12 out of 15 M3R antagonists that we docked. It’s alsointeresting to note that M3R antagonists achieved higher docking scores than the agonists likely because of the added stability via π-π stacking exchange.

MD simulation for interaction analysis

To analyze additional interactions found in hM3R agonism and antagonism, we conducted MD simulations using docked complexes with the highest scoringagonist and antagonist, pilocarpine and tiotropium, respectively.  The protein-ligand complexes with the highest population during 100 ns simulation wereextracted for binding free energy calculation and structural comparisons using Trajectory Clustering tool in Maestro (Schrödinger Inc., USA). The videoanalysis of MD simulations for hM3R-pilocarpine and hM3R-tiotropium complexes can be found in Supplemental Fig. 1 and 2, respectively. The root meansquare distance (RMSD) analysis of hM3R in complex with pilocarpine was performed, where lower the RMSD indicates tighter binding (Fig. 4A). Pilocarpineachieved an RMSD of roughly 2.5 Å towards the end of the simulation, which indicates stable binding34. The protein RMSD equilibrated to roughly 7.0 Å,which implies the protein assumed a stable conformation towards the end of the simulation. On the other hand, tiotropium achieved an RMSD of roughly 3.0Å towards the end of the simulation (Fig. 4B). This value also indicates stable binding but not as tight as pilocarpine-hM3R complex likely because of theincreased molecular mass of tiotropium. The protein RMSD equilibrated to roughly 10 Å, which indicates the protein assumed a stable conformation towardsthe end of the simulation

Interaction fraction analyses of pilocarpine-hM3R and tiotropium-hM3R complexes were also performed. Interaction fraction value of 1.0 means the ligandmade contact with a binding residue during 100% time of the simulation. Values greater than 1.0 indicate more than one binding forces. For example, in theinteraction fraction analysis of pilocarpine-hM3R (Fig. 5A), TYR149 has an interaction fraction value of 1.498 and makes contact with pilocarpine viahydrogen bonds, hydrophobic forces (cation-π exchange), and water bridges.  The top three contributors to the binding interaction for pilocarpine wereTYR149, ASP148, and SER152 with scores of 1.498, 1.217, and 0.914, respectively. In the interaction fraction analysis of tiotropium-hM3R (Fig. 5B), the topthree contributors to the binding interaction were ASN508, TRP504, and TYR507 with scores of 1.755, 1.437, and 1.083, respectively. It’s interesting to note thatTRP504 has almost 3-times the interaction fraction value with tiotropium (1.477) compared to pilocarpine (0.591). Furthermore, the top three binding residueswith pilocarpine are upstream at positions 148-152, whereas they are downstream at positions 504-508 for tiotropium. Ligand-protein contact analysis ofpilocarpine-hM3R revealed ASP148, TYR149, and SER152 intimately associating with the nitrogen ring of pilocarpine (Fig. 5C). Meanwhile, the ligand-proteincontact analysis of tiotropium-hM3R displayed preservation of π-π stacking exchange via TRP504 during most of the simulation (Fig. 5D), which is consistentwith the results of HTVS.

To analyze amino acid side chain stability, root mean square �uctuations (RMSF) analysis was performed, where higher the RMSF values indicates higherinstability of amino acid side chain. In RMSF analysis of pilocarpine-hM3R complex (Fig. 6A), none of the amino acid side chains exceed RMSF value of 9.0 Å.On the other hand, in RMSF analysis of tiotropium-hM3R complex (Fig. 6B), there were three residues with RMSF values greater than 10.5 Å: LEU456 (11.03 Å),SER457 (12.16 Å), and LYS459 (11.89 Å). In fact, these three residues were much more stable in pilocarpine-hM3R complex: LEU456 (7.84 Å), SER457 (5.66 Å),and LYS459 (6.95 Å). It’s interesting to note that these residues span positions 456-459, which are near the top three binding residues (TRP504, TYR507, andASN508) from the interaction fraction analysis of tiotropium-hM3R complex. Table 2 displays the difference in stability among shared binding residues ofpilocarpine-hM3R and tiotropium-hM3R complexes. Overall, the binding residues of tiotropium have lower RMSF and hence higher stability. In fact, TRP504had the highest difference in RMSF of 0.27 Å. This suggests TRP504 is in a more stable and rigid conformation when interacting with tiotropium likely due toadded stability brought by π-π stacking exchange.

Through structural comparisons, we were able to locate TRP504 in helix 6 of hM3R (Fig. 7A and 7B). When we superimposed helix 6 of piloarpine-hM3R andtiotropium-hM3R (Fig. 7C), we identi�ed a 2.1 Å shift in the α-carbon of TRP504 when interacting with tiotropium for π-π stacking. This led to an angle changeof 18 ° upstream of TRP504, including the cytoplasmic domain, where the unstable residues LEU456, SER457, and LYS459 are directly located at. When wesuperimposed the cytoplasmic domain directly upstream of helix 6 (Fig. 7D), we could �nd LEU456, SER457, and LYS459 further away from the proteinbackbone in hM3R-tiotropium than in hM3R-pilocarpine, consistent with the high RMSF values we observed. It’s important to note that these three residueswere more stable in hM3R-pilocarpine complex. Perhaps the increased �uctuations of these residues in the cytoplasmic domain are preventing G proteininteractions and hence antagonism of hM3R.

Binding free energy analysis by MM/GBSA

Page 4/14

To compare the stability of protein-ligand complex between pilocarpine and tiotropium, we used the MM/GBSA method to obtain the binding free energy(ΔGbind). Table 3 lists the ΔGbind of pilocarpine-hM3R and tiotropium-hM3R complexes, whose overall predicted binding free energies were −46.81 kcal/moland -96.82 kcal/mol, respectively. This suggests that tiotropium forms a more stable complex with hM3R than pilocarpine does. Furthermore, near two-foldchange in binding free energy indicates that once tiotropium binds to hM3R, its binding mode is stable, and the TM domain that contains the binding pocketlikely remains in a stable, rigid conformation, which is consistent with the video analysis of tiotropium-hM3R complex in Supplemental Fig. 2.

DiscussionIn this study, we created a homology model of hM3R based on published structures of rat M3R since there were no crystal structure of hM3R published yet.There were minor differences in the cytoplasmic domains between hM3R and rat M3R, and this was expected because the cytoplasmic domain consists ofseveral loop structures that are not as rigid as α-helices or β-sheets. Meanwhile, the TM and EC domains were well-conserved. Since the tiotropium bindingpocket was found in TM and EC domains, we expected the binding site of hM3R to be conserved, which was con�rmed when we ran Sitemap analysis andsuccessfully superimposed tiotropium from rat M3R crystal structure to our homology model’s top-ranking predicted binding site. Molecular docking oftiotropium to hM3R revealed that the binding mode of tiotropium was nearly identical to that of rat M3R crystal structure, con�rming the �delity of ourhomology model for molecular docking.

When we ran HTVS of muscarinic agonists and antagonists of M3R, we discovered none of the agonists displayed π-π stacking exchange and lacked thecapacity to form one since they do not have any aromatic side groups. On the other hand, of the 15 antagonists we screened, 12 of them shared presence ofπ-π stacking exchange with TRP504. Among the three antagonists that did not share any π-π stacking exchange, dicyclomine was the only molecule that didnot have an aromatic side group. Furthermore, among the high-a�nity antagonists (docking score < -10 kcal/mol), oxyphencyclimine was the only molecule,whose aromatic side group did not form π-π stacking exchange with TRP504. MD simulation of the top-scoring agonist and antagonist revealed that the π-πstacking exchange causes TRP504 to shift and lead to angle changes in helix 6 that is directly associated with the cytoplasmic domain, where G proteininteractions occur. RMSF analysis revealed increased �uctuations of three residues found in the cytoplasmic domain that may interfere with G proteininteractions.

Together, our in-silico assays suggest that a tertiary amine with a positively charged nitrogen molecule is crucial for binding to M3R regardless of agonism orantagonism. This is consistent with the structure of acetylcholine, an endogenous ligand of M3R that also houses a positively charged tertiary amine.However, to promote agonism, it’s crucial the ligand does not contain aromatic side groups and has no capacity to form π-π stacking exchange with TRP504.Because there are only 6 muscarinic agonists that can be found in Drug Banks database (compared to 61 antagonists), our in-silico data may prove useful inmedicinal chemistry effort to develop drugs that not only target but also selectively agonize hM3R to reduce IOP in glaucoma patients.

Methods

Preparation for human M3R homology modelHuman M3R structure was constructed via homology modeling using Sequence viewer function in Maestro and were based on 4DAJ structure. Assignment ofbond orders and hydrogenation for the complex structures were performed using Protein Preparation in Maestro. The ionization state of the hM3R suitable forpH 7.0 ± 2.0 was predicted using Epik35. H-bond optimization was conducted using PROPKA36. Energy minimization was performed using the OPLS3e force�eld37. Finally, loop structures were re�ned using Loop Re�nement feature in Maestro. HTVS was performed using Glide in Maestro.

Md SimulationMD simulations for interaction analysis and binding free energy calculation were performed using Desmond (Schrödinger Inc., USA). All systems were set upusing “System Builder” software in Maestro. The pilocarpine-hM3R and tiotropium-hM3R complexes were placed in the orthorhombic box with a bufferdistance of 10 Å to create a hydration model using the SCP water model. The following parameters were set: cut-of radius for van der Waals, time step, initialtemperature, and pressure of the system set to 9 Å, 2.0 fs, 300 K, and 1.01325 bar, respectively. The sampling interval during the simulation was set to 100 ps.Finally, we performed MD simulations under the NPT ensemble for 100 ns.

Interaction Analysis And Trajectory Clustering For Mm/gbsaSimulation Interactions Diagram tool in Maestro was used to perform an interaction analysis of our MD simulation, where our ligand was given 100 ns toequilibrate to stable conformation in relation to hM3R. Trajectory Frame Clustering function in Maestro was used to estimate the most populatedrepresentative structure for each MD simulation.

Mm/gbsa Free Energy CalculationIn MD simulation, free energy calculations give quantitative production of protein–ligand binding energies. The binding energy (ΔGbind) was calculated byEquation [1]:

[1]ΔGbind = GR+L − (GR + GL)

Page 5/14

where GR +L represents the hM3R in complex with ligands, while GR and GL represent the optimized free hM3R and optimized free ligand Gibbs energies,

respectively38.

In the generalized born surface area (MM/GBSA) approach25, each free energy term in Equation [1] was calculated using Equation [2].

[2]:G = Gcoulomb + GvDW + Gcovalent + Gsolv + Gself−contact + GH −bond + Glipo + Gpacwhere Gcoulomb represents coulombic energy, GvDW Van der Waals energy, Gcovalent covalent binding energy, Gsolv Generalized Born electrostatic solvationenergy, Gself −contact self-contact energy, GH−bond hydrogen-bonding energy, Glipolipophilic energy, and Gpacking pi-pi packing energy.

The performance of the MM/GBSA algorithm is based on the speci�city of the force�eld and ligand partial charges, the speci�city of protein–inhibitorcomplex, MD simulation, inner dielectric constant, and the docking pose number based on top scoring39. The VSGB solvation model40 and OPLS3e force �eldwere set for the calculation via MM/GBSA feature in Maestro.

DeclarationsConsent for publication

All authors consent for publication.

Data Availability 

Data and materials will be available upon request.

Acknowledgement

This work was supported by NIH grants EY021200 (MMJ) and EY029950 (MMJ) and an unrestricted grant from Research to Prevent Blindness.

Author information

A�liation: Department of Ophthalmology, Hamilton Eye Institute, University of Tennessee Health Science Center, Memphis, TN 38163, USA

Names: Minjae J. Kim & Monica M. Jablonski

Contributions: Work conceived by M.J.K. and M.M.J. Experiments performed by M.J.K. Data graphed and analyzed by M.J.K. M.J.K. and M.M.J. wrote themanuscript. Project �nanced by M.M.J. All authors edited and approved �nal manuscript.

Corresponding author: Correspondence to Monica M. Jablonski

Ethics Declaration

Competing interests: The authors declare no competing interests.

References1. Kwon, Y. H., Fingert, J. H., Kuehn, M. H. & Alward, W. L. Primary open-angle glaucoma. N Engl J Med 360, 1113–1124, doi:10.1056/NEJMra0804630

(2009).

2. Goel, M., Picciani, R. G., Lee, R. K. & Bhattacharya, S. K. Aqueous humor dynamics: a review. Open Ophthalmol J 4, 52–59,doi:10.2174/1874364101004010052 (2010).

3. Llobet, A., Gasull, X. & Gual, A. Understanding trabecular meshwork physiology: a key to the control of intraocular pressure? News Physiol Sci 18, 205–209, doi:10.1152/nips.01443.2003 (2003).

4. Tamm, E. R. The trabecular meshwork out�ow pathways: structural and functional aspects. Exp Eye Res 88, 648–655, doi:10.1016/j.exer.2009.02.007(2009).

5. Gil, D. W., Krauss, H. A., Bogardus, A. M. & WoldeMussie, E. Muscarinic receptor subtypes in human iris-ciliary body measured by immunoprecipitation.Invest Ophthalmol Vis Sci 38, 1434–1442 (1997).

�. Zhang, X., Hernandez, M. R., Yang, H. & Erickson, K. Expression of muscarinic receptor subtype mRNA in the human ciliary muscle. Invest Ophthalmol VisSci 36, 1645–1657 (1995).

7. Lutjen-Drecoll, E. Structural factors in�uencing out�ow facility and its changeability under drugs. A study in Macaca arctoides. Invest Ophthalmol 12,280–294 (1973).

�. Lutjen-Drecoll, E., Wiendl, H. & Kaufman, P. L. Acute and chronic structural effects of pilocarpine on monkey out�ow tissues. Trans Am Ophthalmol Soc 96,171–191; discussion 192 – 175 (1998).

9. Skaat, A. et al. Effect of Pilocarpine Hydrochloride on the Schlemm Canal in Healthy Eyes and Eyes With Open-Angle Glaucoma. JAMA Ophthalmol 134,976–981, doi:10.1001/jamaophthalmol.2016.1881 (2016).

Page 6/14

10. Ho, M. C., Hsieh, Y. T., Shen, E. P., Hsu, W. C. & Cheng, H. C. Short-term refractive and ocular parameter changes after topical atropine. Taiwan J Ophthalmol10, 111–115, doi:10.4103/tjo.tjo_110_18 (2020).

11. Rengstorff, R. H. & Doughty, C. B. Mydriatic and cycloplegic drugs: a review of ocular and systemic complications. Am J Optom Physiol Opt 59, 162–177(1982).

12. Sanchez-Gonzalez, J. M., De-Hita-Cantalejo, C., Baustita-Llamas, M. J., Sanchez-Gonzalez, M. C. & Capote-Puente, R. The Combined Effect of Low-doseAtropine with Orthokeratology in Pediatric Myopia Control: Review of the Current Treatment Status for Myopia. J Clin Med 9, doi:10.3390/jcm9082371(2020).

13. Wu, T. E. et al. Evaluating the Effect of Topical Atropine Use for Myopia Control on Intraocular Pressure by Using Machine Learning. J Clin Med 10,doi:10.3390/jcm10010111 (2020).

14. Birdsall, N. J., Burgen, A. S., Hiley, C. R. & Hulme, E. C. Binding of agonists and antagonists to muscarinic receptors. J Supramol Struct 4, 367–371,doi:10.1002/jss.400040307 (1976).

15. Birdsall, N. J. et al. Muscarinic receptor subtypes and the selectivity of agonists and antagonists. Pharmacology 37 Suppl 1, 22–31,doi:10.1159/000138503 (1988).

1�. Chiba, S. et al. Identi�cation of potential inhibitors based on compound proposal contest: Tyrosine-protein kinase Yes as a target. Sci Rep 5, 17209,doi:10.1038/srep17209 (2015).

17. Wakui, N., Yoshino, R., Yasuo, N., Ohue, M. & Sekijima, M. Exploring the selectivity of inhibitor complexes with Bcl-2 and Bcl-XL: A molecular dynamicssimulation approach. J Mol Graph Model 79, 166–174, doi:10.1016/j.jmgm.2017.11.011 (2018).

1�. Yasuo, N. & Sekijima, M. Improved Method of Structure-Based Virtual Screening via Interaction-Energy-Based Learning. J Chem Inf Model 59, 1050–1061,doi:10.1021/acs.jcim.8b00673 (2019).

19. Yoshino, R. et al. Pharmacophore modeling for anti-Chagas drug design using the fragment molecular orbital method. PLoS One 10, e0125829,doi:10.1371/journal.pone.0125829 (2015).

20. Li, H., Ibrahim, M. M., Chen, H., Li, W. & Jablonski, M. M. In Silico Screening and In Vivo Evaluation of Potential CACNA2D1 Antagonists as IntraocularPressure-Reducing Agents in Glaucoma Therapy. Pharmaceuticals (Basel) 14, doi:10.3390/ph14090887 (2021).

21. Hayes, R. L. et al. Magnesium �uctuations modulate RNA dynamics in the SAM-I riboswitch. J Am Chem Soc 134, 12043–12053, doi:10.1021/ja301454u(2012).

22. Ingolfsson, H. I. et al. Lipid organization of the plasma membrane. J Am Chem Soc 136, 14554–14559, doi:10.1021/ja507832e (2014).

23. Sekijima, M., Motono, C., Yamasaki, S., Kaneko, K. & Akiyama, Y. Molecular dynamics simulation of dimeric and monomeric forms of human prion protein:insight into dynamics and properties. Biophys J 85, 1176–1185, doi:10.1016/S0006-3495(03)74553-6 (2003).

24. Yildirim, A., Sharma, M., Varner, B. M., Fang, L. & Feig, M. Conformational preferences of DNA in reduced dielectric environments. J Phys Chem B 118,10874–10881, doi:10.1021/jp505727w (2014).

25. Genheden, S. & Ryde, U. The MM/PBSA and MM/GBSA methods to estimate ligand-binding a�nities. Expert Opin Drug Discov 10, 449–461,doi:10.1517/17460441.2015.1032936 (2015).

2�. Kollman, P. A. et al. Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models. Acc ChemRes 33, 889–897, doi:10.1021/ar000033j (2000).

27. Srinivasan, J., Miller, J., Kollman, P. A. & Case, D. A. Continuum solvent studies of the stability of RNA hairpin loops and helices. J Biomol Struct Dyn 16,671–682, doi:10.1080/07391102.1998.10508279 (1998).

2�. Kammarabutr, J., Mahalapbutr, P., Nutho, B., Kungwan, N. & Rungrotmongkol, T. Low susceptibility of asunaprevir towards R155K and D168A pointmutations in HCV NS3/4A protease: A molecular dynamics simulation. J Mol Graph Model 89, 122–130, doi:10.1016/j.jmgm.2019.03.006 (2019).

29. Muhammed, M. T. & Aki-Yalcin, E. Homology modeling in drug discovery: Overview, current applications, and future perspectives. Chem Biol Drug Des 93,12–20, doi:10.1111/cbdd.13388 (2019).

30. Xiang, Z. Advances in homology protein structure modeling. Curr Protein Pept Sci 7, 217–227, doi:10.2174/138920306777452312 (2006).

31. De Lean, A., Stadel, J. M. & Lefkowitz, R. J. A ternary complex model explains the agonist-speci�c binding properties of the adenylate cyclase-coupledbeta-adrenergic receptor. J Biol Chem 255, 7108–7117 (1980).

32. Maguire, M. E., Van Arsdale, P. M. & Gilman, A. G. An agonist-speci�c effect of guanine nucleotides on binding to the beta adrenergic receptor. MolPharmacol 12, 335–339 (1976).

33. Weis, W. I. & Kobilka, B. K. The Molecular Basis of G Protein-Coupled Receptor Activation. Annu Rev Biochem 87, 897–919, doi:10.1146/annurev-biochem-060614-033910 (2018).

34. Ramirez, D. & Caballero, J. Is It Reliable to Take the Molecular Docking Top Scoring Position as the Best Solution without Considering Available StructuralData? Molecules 23, doi:10.3390/molecules23051038 (2018).

35. Shelley, J. C. et al. Epik: a software program for pK(a) prediction and protonation state generation for drug-like molecules. J Comput Aided Mol Des 21,681–691, doi:10.1007/s10822-007-9133-z (2007).

3�. Li, H., Robertson, A. D. & Jensen, J. H. Very fast empirical prediction and rationalization of protein pKa values. Proteins 61, 704–721,doi:10.1002/prot.20660 (2005).

37. Roos, K. et al. OPLS3e: Extending Force Field Coverage for Drug-Like Small Molecules. J Chem Theory Comput 15, 1863–1874,doi:10.1021/acs.jctc.8b01026 (2019).

Page 7/14

3�. Froloff, N., Windemuth, A. & Honig, B. On the calculation of binding free energies using continuum methods: application to MHC class I protein-peptideinteractions. Protein Sci 6, 1293–1301, doi:10.1002/pro.5560060617 (1997).

39. Halim, S. A., Waqas, M., Khan, A. & Al-Harrasi, A. In Silico Prediction of Novel Inhibitors of SARS-CoV-2 Main Protease through Structure-Based VirtualScreening and Molecular Dynamic Simulation. Pharmaceuticals (Basel) 14, doi:10.3390/ph14090896 (2021).

40. Li, J. et al. The VSGB 2.0 model: a next generation energy model for high resolution protein structure modeling. Proteins 79, 2794–2812,doi:10.1002/prot.23106 (2011).

TablesTable 1. HTVS of hM3R agonists and antagonists.

HM3R Agonists Docking Scores (kcal/mol) π-π stacking with TRP504?

Pilocarpine -8.993 No

NGX267 -8.692 No

Cevimeline -8.108 No

Bethanechol -6.539 No

Xanomeline -6.019 No

Methacholine -3.746 No

HM3R Antagonists Docking Scores (kcal/mol) π-π stacking with TRP504?

Tiotropium -11.036 Yes

Diphenidol -10.626 Yes

Tridihexethyl -10.606 Yes

Methanetheline -10.400 Yes

Solifenacin -10.378 Yes

Glycopyrronium -10.264 Yes

Oxybutin -10.159 Yes

Oxyphencycline -10.034 No

Promethazine -9.850 Yes

Homatropine methylbromide -9.79 No

Procyclidine -9.748 Yes

Dicycloamine -9.45 No

Olanzapine -9.443 Yes

Ipratropium -9.391 Yes

Tropicamide -9.064 Yes

Table 2. RMSF values (Å) of shared binding residues of hM3R for pilocarpine and tiotropium, from highest differences to lowest. 

Binding Residues RMSF (Å) with pilocarpine RMSF (Å) with tiotropium Difference (Å)

TRP504 1.18 0.91 0.27

Ser152 1.07 0.81 0.26

ALA236 1.95 1.77 0.18

CYS533 1.30 1.15 0.15

ASN508 1.60 1.58 0.02

ASP148 1.14 1.12 0.02

TYR534 0.73 0.73 0

TYR149 1.10 1.15 -0.05

TYR530 1.70 1.84 -0.14

Page 8/14

Table 3. MM/GBSA binding free energy calculation of hM3R-pilocarpine and hM3R-tiotropium complexes. All values are in kcal/mol.

Protein-Ligand Complex ΔGcoulomb ΔGvdW ΔGcovalent ΔGsolv ΔGself-contact ΔGH-bond ΔGLipo ΔGPacking ΔGBind

hM3R-

Pilocarpine

-0.99 -37.43 0.89 12.94 0 -1.82 -20.39 0 -46.81

hM3R-

Tiotropium

4.73 -60.93 2.72 -2.74 0 -1.22 -36.04 -3.34 -96.82

Figures

Figure 1

Homology Modeling and Sitemap analysis of human M3 muscarinic receptor (hM3R). (a) Predicted structure of hM3R, colorized by residue numbers from redto purple. Black arrows indicate spans of extracellular domain (EC), transmembrane domain (TM), and cytoplasmic domain (CP), where G protein interactionsoccur. (b) Superimposition of template structure (PDB: 5ZHP – rat M3 muscarinic acetylcholine receptor in complex with a selective antagonist) in purple andhomology model of hM3R in cyan (sequence identity: 62 %). Note that their main differences are found in the intracellular domain, while the transmembraneand extracellular domains are conserved. (c) Binding sites of our homology model identi�ed by SiteMap. The yellow mesh in the binding sites indicates thehydrophobic map, while the blue and red colors indicate the hydrogen-bond donor and hydrogen-bond receptor maps, respectively. (d) Top ranking binding siteidenti�ed by Sitemap in homology model of hM3R.

Page 9/14

Figure 2

Molecular Docking of tiotropium, a known M3 receptor antagonist, in human and rat M3 muscarinic receptors. (a) Superimposition of crystal structure of ratM3 receptor complex with tiotropium (purple) and homology model of hM3R (cyan). Upon superimposition, tiotropium from the crystal structure �ts tightlyinside the top-ranking binding site of hM3R identi�ed by Sitemap. (b) Binding mode of tiotropium obtained from crystal structure of rat M3 receptor (purple)and from docking with homology model of hM3R (cyan). (c) Ligand interaction diagram of tiotropium bound in rat M3 receptor complex. (d) Moleculardocking of tiotropium in homology model of hM3R with a docking score of -11.036 kcal/mol. Notice the similarity between the binding modes of tiotropium inrat and human M3 muscarinic receptors.

Page 10/14

Figure 3

Top 3 scoring agonists and antagonist upon high-throughput virtual screening (HTVS) of 6 agonists and 15 antagonists of hM3R, whose 3D structures wereobtained from Drug Banks database. (a) Binding mode of top 3 scoring agonists of hM3R upon Glide docking. Note that all three of them involve presence ofextensive π-cation and salt bridge interactions with the nitrogen ring via ASP148, TYR149, TRP504, TYR507, and TYR530. (b) Binding mode of top 3 scoringantagonists of hM3R upon Glide docking. In addition to the π-cation and salt bridge interactions found with the agonists, they all involved presence ofextensive π-π stacking with TRP504, which may explain the improved docking score among antagonists compared to agonists. None of the agonists availableon Drug Banks data base had any aromatic ring.

Page 11/14

Figure 4

Molecular Dynamics Root Mean Square Distance (RMSD) analysis of hM3R interacting with pilocarpine (a) and tiotropium (b). RMSD is calculated withaverage change in displacement of a selection of atoms for a particular frame with respect to a reference frame. Notice the ligand RMSD equilibrates toroughly 2.5-3.0 Å towards the end of the simulation, which indicates stable binding between our protein of interest and the ligand. 

Page 12/14

Figure 5

Protein-ligand contact analysis of hM3R-pilocarpine and hM3R-tiotropium complexes. (a) Interaction fraction summary of pilocarpine interacting with hM3R.The top three contributors to the binding interaction were TYR149, ASP148, and SER152. (b) Interaction fraction summary of pilocarpine interacting withhM3R. The top three contributors to the binding interaction were ASN508, TRP504, and TYR507. (c) Ligand-protein contact analysis of pilocarpine interactingwith hM3R. Notice the presence of extensive pi-cation and hydrogen bond networks between pilocarpine’s nitrogen ring and hM3R. (d) Ligand-protein contactanalysis of tiotropium interacting with hM3R. In addition to the pi-cation exchange with the nitrogen ring, notice the presence of π-π stacking with TRP504 andhydrogen bonding with ASN508.

Page 13/14

Figure 6

Protein side chain analysis of hM3R-pilocarpine and hM3R-tiotropium complexes. (a) Root-mean square �uctuation (RMSF) analysis of all the residues ofhM3R-pilocarpine complex. Dotted lines cut off the bar chart at 10.5 Å. Green lines mark positions of binding residues. (b) RMSF analysis of all the residues ofhM3R-tiotropium complex. Notice the presence of three residues with RMSF values greater than 10.5 Å in blue circles: LEU456 (11.03 Å), SER457 (12.16 Å),and LYS459 (11.89 Å). High RMSF values indicate instability or signi�cant structural changes in the corresponding protein domain. On the other hand, RMSFvalues of the same residues in hM3R-pilocarpine complex were much lower: LEU456 (7.84 Å), SER457 (5.66 Å), and LYS459 (6.95 Å).

Page 14/14

Figure 7

Structural comparison of helix 6 between hM3R-pilocarpine (green) and hM3R-tiotropium (orange). Black arrows indicate spans of different protein domains.(a) TRP504 interacting with pilocarpine in the binding pocket within transmembrane domain (TM) of hM3R. (b) TRP504 interacting with tiotropium. Notice thebent of helix 6 that starts at TRP504. (c) Superposition of the two complexes revealed a 2.1 Å shift in the α-carbon of TRP504 when interacting with tiotropiumfor π-π stacking. This led to an angle change of 18 ° downstream of TRP504, including the cytoplasmic domain, where G protein interactions occur. (d)Superimposition of two complexes at the cytoplasmic domain. Spheres display LEU456, SER457, and LYS459, which had RMSF values higher than 10.5 Å andhence predicted to be the most unstable residues in hM3R-tiotropium complex. (e, f) Positions of LEU456, SER457, and lYS459 in hM3R-pilocarpine andhM3R-tiotropium complexes, respectively. As discussed in Fig. 6, these three residues are more stable in hM3R-pilocarpine complex. Perhaps the increased�uctuations of these residues in the cytoplasmic domain are preventing G protein interactions.

Supplementary Files

This is a list of supplementary �les associated with this preprint. Click to download.

SupplementalVideo1.mpeg

SupplementalVideo2.mpeg