In-Silico Approaches for the Screening and Discovery of Broad-Spectrum Marine Natural Product Antiviral Agents Against Coronaviruses

Abstract

The urgent need for SARS-CoV-2 controls has led to a reassessment of approaches to identify and develop natural product inhibitors of zoonotic, highly virulent, and rapidly emerging viruses. There are yet no clinically approved broad-spectrum antivirals available for beta-coronaviruses. Discovery pipelines for pan-virus medications against a broad range of betacoronaviruses are therefore a priority. A variety of marine natural product (MNP) small molecules have shown inhibitory activity against viral species. Access to large data caches of small molecule structural information is vital to finding new pharmaceuticals. Increasingly, molecular docking simulations are being used to narrow the space of possibilities and generate drug leads. Combining in-silico methods, augmented by metaheuristic optimization and machine learning (ML) allows the generation of hits from within a virtual MNP library to narrow screens for novel targets against coronaviruses. In this review article, we explore current insights and techniques that can be leveraged to generate broad-spectrum antivirals against betacoronaviruses using in-silico optimization and ML. ML approaches are capable of simultaneously evaluating different features for predicting inhibitory activity. Many also provide a semi-quantitative measure of feature relevance and can guide in selecting a subset of features relevant for inhibition of SARS-CoV-2.

Keywords: RNA dependent polymerases; SARS-CoV-2; betacoronavirus; genome replication; methyl transferases; natural products; protease; viral transcription.

Figure 1

Function of structural proteins S, E, M and N in the mode of entry of SARS-CoV-2. I The S protein binds to ACE2 on the host cell membrane initiating endocytosis. II: host acidifies endosome, causing TMPRSS2 or Cathepsin L mediated proteolysis of S protein. III: Proteolysis of S protein causes membrane fusion, facilitated by E and M proteins, and release of viral genome. Nucleocapsid disassembles. (Su et al, 2021). Figure made using BioRender© at Biorender.com.

Figure 2

Important structural features of PL^PRO are conserved across the betacoronavirus family. (A) ClustalOmega alignment of PL^PRO peptide sequence from MERS-CoV (first sequence, PDB: 4PT5), SARS-CoV (second sequence, PDB: 4MM3) and SARS-CoV-2 (third sequence, PDB: 6wx4), showing low sequence homology. (B) Protein 3D structure alignment with mTm-align (Dong et al, 2018) showing a common core with an RMSD of 1.48 (shown in magenta in both sequence view and 3D view) around the catalytic and peptide binding site. (C) Conserved Domain Database (NCBI) entry for betacoronavirus PL^PRO conserved catalytic domain, showing alignment of most diverse members. Conserved catalytic triad, Cys, His and Asp, are highlighted in yellow (Lu et al, 2020). (D) SARS-CoV-2PL^PRO (PDB: 6wx4) showing conserved catalytic triad. Figure made using Pymol© by Tubiana; Open-Source PyMOL is Copyright © Schrodinger, LLC. All Rights Reserved.

Figure 3

Ligand binding sites (yellow) of betacoronavirus proteins. (A) Superposition of PL^PRO of SARS-CoV-2 (PDB: 7cjm, blue), SARS-CoV (PDB: 3e9s, red), and MERS-CoV (PDB: 5w8u, green). (B) Superposition of M^PRO of SARS-CoV-2 (PDB: 6luv, blue), SARS-CoV (PDB: 3v3m, red), and MERS-CoV (PDB: 4rsp, green). (C) Superposition of 2’-O-MTase of SARS-CoV-2 (PDB: 6w4h, blue), SARS-CoV (PDB: 3r24, red), and MERS-CoV (PDB: 5ynb, green). (D) Superposition of RdRp of SARS-CoV-2 (PDB: 7bv2, blue) and RS-CoV (PDB: 6nur, red). Figure made by YMChoo using BIOVIA, Dassault Systèmes, Discovery Studio Visualizer, v21.1.0.20298 software program was used to perform the calculations and to generate the graphical results.

Figure 4

SARS-CoV-2 PL^PRO showing drugability hotspots in conserved domains. Fragment Hotspot Maps (Radoux et al, 2016) returned 4 binding hotspots (hydrophobic binding in yellow, hydrogen bond donor in red and hydrogen bond acceptor in blue) on top of the three active domains (blue circle – catalytic domain, grey circle – ubiquitin binding domain, green circle – zinc binding domain). Figure made using Pymol© by Tubiana; Open-Source PyMOL is Copyright © Schrodinger, LLC. All Rights Reserved.

Figure 5

A meta-heuristic strategy for searching a large chemical database for molecular docking. (A) Construction of the Network: For each pair of molecules, common structures are selected using a similarity score based on predetermined metrics. If the maximum similarity score is higher than a threshold, it is marked as a connection represented as an edge in the network constructed using a “push and pull” technique. (B) Swarm particle approach: A n number of nodes are selected to 'host' a particle, where the vector that the particle will take is defined by the adjacent neighbors of the node where the particle is a given momentum t based on an initial random velocity. Each node the particle travels through is a structure that is computed by molecular docking. (C) Swarm particle convergence: The swarm particle algorithm starts at time 0 with many random particles exploring the space through random vectors with their own magnitude, direction, and sense. With each iteration, the particle's vector will change with a velocity that is influenced by the optimal ligand that it has uncovered. Eventually, the particles will converge on an optimum at time. Figure made using BioRender© at Biorender.com.

Figure 6

Graphical representation of the relationship between Artificial Intelligence, Machine Learning and Deep Learning.

Figure 7

Compounds with associated activities specifically against 3CL^pro of SARS-CoV-2. ID: CHEMBL3927. Creative commons. Available from: https://www.ebi.ac.uk/chembl/g/#browse/activities/filter/target_chembl_id%3ACHEMBL3927.

Figure 8

Typical in-silico drug-design workflow. In green are highlighted ML approaches that can be exploited to improve speed, accuracy, and performances of the analyses.