Ligands Based Drug Design for Covid 19 - A MultiFaceted Approach using Ligand Design, Molecular Dock by IJRASET

http://doi.org/10.22214/ijraset.2020.5134

May 2020

Ligands Based Drug Design for Covid 19 - A MultiFaceted Approach using Ligand Design, Molecular Docking and Binding Probability Calculation Mukundan Satyanarayanan1, Dr. V. Radhika2 1

Concepts 2 Consumer, Chennai, India Department of Mathematics, Easwari Engineering College, Chennai, India

Abstract: Covid 19 has been the most devastating pandemic of the recent years, affecting 3 million people in about 210 Countries. Entire World is working on inventing a drug for this pandemic. As a major breakthrough, the crystal structure of Covid 19 main protease 3CLPro or MPro , which plays a major role in mediating the replication and transcription of the virus , was derived by Jin, Z., Du, X., Xu, Y. et al., paving way for the drug design. The crystal structure of MPro with a computer aided design inhibitor N (6LU7) has been used as a potential target for drug design in this work. Three possible binding sites were identified for 3CLPro or MPro using DEEPSITE, a protein binding pocket predictor. Complimentary Ligand shapes were generated for the SARS CoV2 main protease MPro , making use of LIGANN, a structure based de novo drug design tool. They are purely structure based designs and do not have any previous history of synthesis or usage. Molecular docking of the new ligands with the target protein 6LU7 was done using iGEMdock. The binding free energy values were calculated. 10 best ligand designs , for each binding site, based on lowest free energy requirement for have been selected for further study. The binding probability for the 10 ligands were calculated using BINDSCOPE, a structure based protein ligand binding predictor. The identical structures for these ligands were identified using Drug bank database. The results were verified with Tanimoto Coefficient calculation. Based on various parameters like free binding energy, binding probability, structural identity and Tanimoto coefficient, top 5 ligand structures have been selected as potential leads for drug discovery. Keywords: Covid -19, SARS CoV2, Drug design, Natural Ligands, MPro , 3CL Main Protease, 6LU7, binding free energy, binding probability, Tanimoto coefficient I. INTRODUCTION Covid 19 has been the most devastating pandemic of the recent years, affecting 210 Countries so far. Close to 3 million people have been affected till date. The death toll is close to 0.2 million. This disease is caused due to a novel corona virus named SARS CoV2. Entire World is working on inventing a drug for this pandemic. No drug has proceeded till the commercial production stage, though a few drugs have entered into the clinical trial stage.

ÂŠIJRASET: All Rights are Reserved

844

International Journal for Research in Applied Science & Engineering Technology (IJRASET) ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429 Volume 8 Issue V May 2020- Available at www.ijraset.com II. SARS Co v2 INFECTION MECHANISM Majority of the Coronaviridae genome encodes pp1a and through ribosomal frameshifting during translation, pp1ab. These polyproteins are cleaved and transformed into mature non-structural proteins (NSPs) by 2 proteases 3CLPro (3CL like Protease) and PLPro. (Papain like protease) , encoded by open reading frame 1(3). The NSPs in turn play a fundamental role in the transcription/replication during infection [1] 3CLPro or MPro is a catalytically active dimer. Cleavage by MPro occurs at the Glutamine residue in the P1 position of the substrate through the protease CYS-HIS dyad. Cysteine thiol functions as the nucleophile in the proteolytic process. As a major breakthrough, the crystal structure of SARS CoV2 main protease 3CLPro (3CL like Protease) or MPro , which plays a major role in mediating the replication and transcription of the virus , was derived by Jin, Z., Du, X., Xu, Y. et al. [2], paving way for the drug design, considering this enzyme as a potential target molecule. The crystal structure of MPro with a computer aided design inhibitor N (6LU7) has been used as a potential target for drug design in this work. III. RESEARCH METHODOLOGY The possible binding sites of 3CLPro or MPro were identified using Deep Site protein binding site predictor [3]. The binding sites identified are mentioned in Table 1 Binding site

scores

centres

0.996279

[-34.87799835 16.13400078 52.64800072]

0.996786

[-16.87799835 34.13400078 58.64800072]

0.99318

[-10.87799835 20.13400078 68.64800072]

Complimentary Ligand shapes were generated for the Covid 19 main protease MPro , for binding site all the three binding sites making use of LIGANN [4], a structure based de novo drug design tool, based on generative neural-networks. Site 2 (-10.88, 20.13, 68.64) is considered as the most appropriate site for binding (presence of CIS-HYS diad) based on the study conducted by Marina Macchiagodena, et al (1). (a) SARS-CoV2 (orange) and SARS-CoV (green) main proteases. Violet spheres correspond to the alpha carbons of the 12 differing residues. Grey spheres indicate the CYS-HIS dyad (b) binding pocket with the main residues in bond representation (green and red for SARS-CoV2 and SARS-CoV, respectively). The shaded region marks the binding site for the substrate. The tool generated 88, 94 & 93 ligand models for binding sites 0,1,& 2 respectively. They are purely structure based designs and are not from the list of ligands which are already being used or have been discovered previously. In spite of this, a few of them closely resemble some of the ligands which have already been discovered. Molecular docking of the new ligands with the target protein 6LU7 was done using the graphical automatic drug design system for docking, iGEMdock [5], version 2.1. The application suggested 3 binding centres for the ligands. A validation trial was done for all the three centres using the first 5 ligands of the binding site 0 (-10.88, 20.13, 68.64). Binding centre PJE depicted the lowest docking fitness score for all the 5 ligands. The validation results are shown in Table 2. This was selected as the binding centre to measure the docking fitness score for all the 3 sets of ligands. The binding site radius was fixed at 10.0 Ă…. The following parameters were used for testing. Population Size: 150, Generations: 70, No of solutions: 2 The docking fitness was measured for all the 3 sets of new ligands. Out of this, the top 10 ligands, with lowest docking fitness score were selected for analysis from each of the binding site. The docking fitness score for all the three sets are shown in Table 3, 4 & 5. The binding probabilities for these selected ligands with the SARS CO V2 main protease MPro were calculated using the tool BINDSCOPE [6]. BINDSCOPE is a structure based protein ligand binding predictor. It calculates the binding probability based on the binding pocket and ligand pose. The binding probability was measured at the binding site suggested by DeepSite i.e. 0 (-34.88, 16.13, 52.65), 1(16.88, 34.13, 58.64) & 2 (-10.88, 20.13, 68.64). The results are shown in Table 3, 4 &5. New Ligands showing lowest binding fitness scores and highest binding probabilities were identified for all the three binding sites. They can function as potential ligand models for drug discovery. The structures of these ligands were checked for similarity using Drug bank database in Open Babel [7] platform. The results were verified with Tanimoto Coefficient calculation [8]. The results are shown in Table 3, 4 & 5.

ÂŠIJRASET: All Rights are Reserved

845

International Journal for Research in Applied Science & Engineering Technology (IJRASET) ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429 Volume 8 Issue V May 2020- Available at www.ijraset.com The research work finally lists out 5 new ligands based binding fitness score, binding probability, molecular similarity and Tanimoto coefficient. The details are listed in Table 6. Previous research has confirmed that the potential ligands for SARS Co V2 main protease MPro are all aromatic moieties with rotatable bonds of pseudo linear type [1]. The structural details of the potential 5 ligands have also been verified. The details are provided in the Table 7. The molecular structure of these 5 new ligands and their similar ligands are shown in Table 8. These molecules can be further studied. Their structure can be slightly modified to improve the above parameters, so that, they can function as potential drugs for Covid 19. These new ligands can efficiently bind with the SARS Co V2 main protease MPro , inhibiting the formation of NSPs and thereby the replication and transcription during infection. IV CONCLUSION Covid 19 has been the most devastating pandemic of the recent years, affecting 210 Countries so far. Close to 3 million people have been affected till date. This disease is caused due to a novel corona virus named SARS CoV2. The crystal structure of SARS CoV2 main protease 3CLPro (3CL like Protease) or MPro , which plays a major role in mediating the replication and transcription of the virus , was derived by Jin, Z., Du, X., Xu, Y. et al. (2), paving way for the drug design, considering this enzyme as a potential target molecule. The possible binding sites of 3CLPro or MPro were identified using Deep Site protein binding site predictor (3). Complimentary Ligand shapes were generated for the Covid 19 main protease MPro , for binding site all the three binding sites making use of LIGANN, a structure based de novo drug design tool, based on generative neural-networks. Molecular docking of the new ligands with the target protein 6LU7 was done using the graphical automatic drug design system for docking, iGEMdock, version 2.1. The docking fitness was measured for all the 3 sets of new ligands. Out of this, the top 10 ligands, with lowest docking fitness score were selected for analysis from each of the binding site. The binding probabilities for these selected ligands with the SARS CO V2 main protease MPro were calculated using the tool BINDSCOPE. New Ligands showing lowest binding fitness scores and highest binding probabilities were identified for all the three binding sites. The structures of these ligands were checked for similarity using Drug bank database in Open Babel platform. The results were verified with Tanimoto Coefficient calculation. New Ligand molecules showing some resemblance to existing drug molecules like Arlasetone, Ajmaline, Rufinamide, Nilotinib and Deferaserox can be taken up for further drug research. The molecular structure of these 5 new ligands and their similar ligands are shown in Table 8. REFERENCES [1] [2]

[3] [4] [5] [6] [7] [8]

Marina Macchiagodena, Marco Pagliai, and Piero Procacci, Identification of potential binders of the main protease 3CLpro of the COVID-19 via structure-based ligand design and molecular modelling, Chem Phys Lett. 2020 Jul; 750: 137489, Published online 2020 Apr 18. doi: 10.1016/j.cplett.2020.137489 Jin, Z., Du, X., Xu, Y., Deng, Y., Liu, M., Zhao, Y., Zhang, B., Li, X., Zhang, L., Peng, C., Duan, Y., Yu, J., Wang, L., Yang, K., Liu, F., Jiang, R., Yang, X., You, T., Liu, X., Yang, X., Bai, F., Liu, H., Liu, X., Guddat, L.W., Xu, W., Xiao, G., Qin, C., Shi, Z., Jiang, H., Rao, Z., Yang, H, Structure of Mpro from COVID-19 virus and discovery of its inhibitors in PubMed, 2020 Apr 9. Doi: 10.1038/s41586-020-2223-y. [Epub ahead of print] Jiménez, Doerr, Martinez-Rosell, Rose, De Fabritiis, DeepSite: Protein binding site predictor using 3D-convolutional neural networks, Bioinformatics, Volume 33, Issue 19, 01 October 2017, Pages 3036–3042. Skalic, M., Jiménez Luna, J., Sabbadin, D., & De Fabritiis, Shape-Based Generative Modeling for de-novo Drug Design. Journal of chemical information and modelling, 2019, Volume 59, Issue 3, Pages 1205-1214 Hsu, K., Chen, Y., Lin, S. et al. iGEMDOCK: a graphical environment of enhancing GEMDOCK using pharmacological interactions and post-screening analysis. BMC Bioinformatics 12, S33 (2011). https://doi.org/10.1186/1471-2105-12-S1-S33 Jiménez Luna, J., Skalic, M., Martinez-Rosell, G., & De Fabritiis, G, K DEEP : Protein-ligand absolute binding affinity prediction via 3D-convolutional neural networks, Journal of Chemical Information and Modeling, 2018, Volume 58, Issue 2, Pages 287-296. O'Boyle, N.M., Banck, M., James, C.A. et al. Open Babel: An open chemical toolbox. J Cheminform 3, 33 (2011). https://doi.org/10.1186/1758-2946-3-33 Bajusz, D., Rácz, A. & Héberger, K. Why is Tanimoto index an appropriate choice for fingerprint-based similarity calculations?. J Cheminform 7, 20 (2015). https://doi.org/10.1186/s13321-015-0069-3

©IJRASET: All Rights are Reserved

846

Protein

6LU7

S.No

Ligand SMILES

1 2 3 4 5

Binding site

Binding Site centre 02J PJE

CN(C)c1ccccc1CNc1cc(-c2cccs2)n[nH]1 CSc1ccccc1CC(=O)NC(c1nc2ccccc2o1)c1c(C)noc1C CSCC(C(=O)O)N(C(=O)c1nn(-c2ccccc2)c2c1CCC2)c1ccccc1 CCC(C)C(C(=O)NS(=O)(=O)C(C)C(F)(F)F)c1ccccc1 Cn1cnc2cccc(-c3nc(-c4ccsc4)no3)c21

-50.23 -55.37 -48.85 -67.34 -63.40

-67.52 -66.79 -75.18 -94.76 -80.78

010 -61.2 -56.54 -59.48 -70.21 -67.41

Table 3: Docking Fitness Score, Binding Probability, Molecular Similarity and Tanimoto coefficient for the top 10 new ligands for the binding site 0(-34.88, 16.13, 52.64)

S.No

SMILES

22 24 25 32 37 42 45 54 77 84

CN(Cc1ccccc1N1CCCC1=O)C(=O)C1(S(=O)(=O)c2ccccc2)CCCC1 Cc1cccc(-c2nc(-c3ccsc3)no2)c1F CC(c1ccccc1)n1cccc1C=NNc1cccc(F)c1 Cc1cccc(NCc2nc(-c3ccccc3)c[nH]2)c1Cl Cc1cccc(NC(=S)Nc2ccccc2-n2ccnc2)c1 CN(Cc1nc(N)c2ccccc2n1)Cc1ccccc1F Cc1cccc(C(C)C)c1NC(=O)COC(=O)c1cc2ccccc2[nH]1 Cn1c(-c2cnc3ccccc3n2)nc2c(F)cccc21 Cc1cc(Oc2ccc(-c3ccccc3)nn2)ccc1F O=Cc1sc(-c2ccccc2)nc1-c1ccncc1

Binding Affinity Fitness Score 0.0377 0.9283 0.0833 0.3861 0.9359 0.1185 0.4852 0.6428 0.478 0.8465

-96.88 -103.34 -99.12 -102.96 -102.18 -99.74 -102.9 -106.95 -100.48 -97.14

Similarity

Tanimoto coefficient

Irbesatran Rufinamide Antipyrine Thiabendazole Azathioprine Bretylium Nitazoxanide Nilotinib Papaverine Sulfamethaxazole

0.26 0.17 0.29 0.22 0.2 0.28 0.16 0.31 0.27 0.16

Table 4: Docking Fitness Score, Binding Probability, Molecular Similarity and Tanimoto coefficient for the top 10 new ligands for the binding site 1(-16.88, 34.13, 68.64) Binding Binding Free Tanimoto S.No SMILES Probability Energy Similar structure Coeffcient

1 3 5 8 9 22 24 34 44 88

O=C(c1coc(-c2ccccc2)n1)N1CCC(CO)c2ccccc21 C#CCOC(=O)c1cc(=O)n(-c2ccc(C)cc2)[nH]1 Fc1ccccc1-c1nnc(Sc2cccc(OC3CCCC3)c2)o1 O=C(O)c1cn(-c2ccc(Br)cc2)nn1 CCCC(=O)OCc1n[nH]c(-c2ccc(OC)cc2)n1 COc1cccc2oc(=O)n(C)c12 Cc1ccccc1N1CC(C(=O)O)C(F)(F)C1 CC(C)NC(=O)COC(=O)c1ccc(CCC#N)cc1 CCC(=O)OCc1nnc(-c2cc(C)cc(C)c2)[nH]1 CN(C)C(=O)c1cccc(-c2noc(Cc3ccccc3F)n2)c1

Â©IJRASET: All Rights are Reserved

0.0007 0.4166 0.106 0.8213 0.2641 0.9831 0.9937 0.0403 0.5155 0.3889

-93.08 -95.25 -105.01 -102.48 -90.3 -94.85 -90.34 -96.8 -91.39 -94.3

Alosetron Conivapton Rabeprazole Deferasirox Oxaprozin Tolnaftate Ajmaline Benzonatate Oxaprozin Estazolam

0.3 0.2 0.2 0.27 0.23 0.21 0.38 0.44 0.24 0.21

847

International Journal for Research in Applied Science & Engineering Technology (IJRASET) ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429 Volume 8 Issue V May 2020- Available at www.ijraset.com Table 5: Docking Fitness Score, Binding Probability, Molecular Similarity and Tanimoto coefficient for the top 10 new ligands for the binding site 2(-16.88, 34.13, 68.64) S.No

SMILES

Binding Probability

Binding free energy

Similar Molecule

Tanimoto Coefficient

14 20 51 55 58 61 63 90 91 93

Cc1cccc(CNc2nc(N)nc(C(F)(F)F)n2)c1 C#CCn1cc(-c2nc(N)nc(C)c2Br)nn1 Cc1cc(N(C)C)ccc1C(=O)Nc1nc2ccccc2[nH]1 COCC(=O)N1CCc2c([nH]c3c(F)cccc23)C1 Cc1cccc(C)c1OCc1nc(C(N)=O)no1 Cc1cccc(OCCNc2ccc(O)nn2)c1 c1cc(-c2cc(CN3CC4CCC(C3)N4)no2)no1 CCC(C)(C)c1cc(Nc2cc(C)on2)[nH]n1 Cc1[nH]nc(NS(=O)(=O)c2ccc3ccccc3c2)c1C(N)=O CC(=NNc1ccc(OC(F)F)cn1)c1cccc(O)c1

0.5852 0.06 0.0001 0.9808 0.0371 0.0004 0.0001 0.0001 0.6863 0.0576

-97.54 -100.42 -86.49 -89.76 -86.92 -86.29 -87.09 -86.53 -91.11 -93.9

Edrophonium Rufinamide Omeprazole Alosetron Nitazoxanide Erlotinib Alizapride Nitrofural Sulfamoxole Riluzole

0.13 0.24 0.25 0.52 0.22 0.35 0.23 0.12 0.35 0.32

Table 6: Top 5 ligands suitable for drug design based on docking fitness score, binding probability and Tanimoto coefficient. Binding S.No Binding site SMILES Probability Fitness Score Similar Molecule Tanimotto Coefficient

1 2 3 4 5

2 1 1 0 0

COCC(=O)N1CCc2c([nH]c3c(F)cccc23)C1 Cc1ccccc1N1CC(C(=O)O)C(F)(F)C1 O=C(O)c1cn(-c2ccc(Br)cc2)nn1 Cn1c(-c2cnc3ccccc3n2)nc2c(F)cccc21 Cc1cccc(-c2nc(-c3ccsc3)no2)c1F

0.9808 0.9937 0.8213 0.6428 0.9283

-89.76 -90.34 -102.48 -106.95 -103.34

Alosetron Ajmaline Deferasirox Nilotinib Rufinamide

0.52 0.38 0.27 0.31 0.17

Table 7: Molecular weight, rotatable bonds and aromatic rings in new ligands selected for drug design

S.No

SMILES

1 COCC(=O)N1CCc2c([nH]c3c(F)cccc23)C1 2 Cc1ccccc1N1CC(C(=O)O)C(F)(F)C1 3 O=C(O)c1cn(-c2ccc(Br)cc2)nn1 4 Cn1c(-c2cnc3ccccc3n2)nc2c(F)cccc21 5 Cc1cccc(-c2nc(-c3ccsc3)no2)c1F

Aromatic rings Molecular Weight

3 2 2 4 3

262.28 241.24 268.07 278.29 260.29

Table 8: Molecular structure and Structure of similar molecule for the 5 new ligands Ligand 1:

Alosetron

Ligand Name: 1-(8-fluoro-1, 3, 4, 9-tetrahydropyrido [3, 4-b] indol-2-yl)-2-methoxy-ethanone

848

Ajmaline

Ligand Name: 4, 4-difluoro-1-(o-tolyl) pyrrolidine-3-carboxylic acid Ligand 3:

Deferasirox

Ligand Name: 1-(4-bromophenyl) triazole-4-carboxylic acid Ligand 4:

Nilotinib

Ligand Name: 2-(4-fluoro-1-methyl-benzimidazol-2-yl)quinoxaline

ÂŠIJRASET: All Rights are Reserved

849

Rufinamide

Ligand name: 5-(2-fluoro-3-methyl-phenyl)-3-(3-thienyl)-1,2,4-oxadiazole

850