Consensus docking and MM-PBSA computations identify putative furin protease inhibitors for developing potential therapeutics against COVID-19

Abstract

The coronavirus disease 2019 (COVID-19) is a pandemic that has severely posed substantial health challenges and claimed millions of lives. Though vaccines have been produced to stem the spread of this disease, the death rate remains high since drugs used for treatment have therapeutic challenges. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes the disease, has a slew of potential therapeutic targets. Among them is the furin protease, which has a cleavage site on the virus's spike protein. The cleavage site facilitates the entry of the virus into human cells via cell-cell fusion. This critical involvement of furin in the disease pathogenicity has made it a viable therapeutic strategy against the virus. This study employs the consensus docking approach using HYBRID and AutoDock Vina to virtually screen a pre-filtered library of 3942 natural product compounds of African origin against the human furin protease (PDB: 4RYD). Twenty of these compounds were selected as hits after meeting molecular docking cut-off of - 7 kcal.mol^-1, pose alignment inspection, and having favorable furin-ligand interactions. An area under the curve (AUC) value of 0.72 was computed from the receiver operator characteristic (ROC) curve, and Boltzmann-enhanced discrimination of the ROC curve (BEDROC) value of 0.65 showed that AutoDock Vina was a reasonable tool for selecting actives for this target. Seven of these hits were proposed as potential leads having had bonding interactions with catalytic triad residues Ser368, His194, and Asp153, and other essential residues in the active site with plausible binding free energies between - 189 and - 95 kJ/mol from the Molecular Mechanics Poisson-Boltzmann Surface Area (MM-PBSA) calculations as well as favorable ADME/Tox properties. The molecules were also predicted as antiviral, anti-inflammatory, membrane permeability inhibitors, RNA synthesis inhibitors, cytoprotective, and hepatoprotective with probable activity (Pa) above 0.5 and probable inactivity values below 0.1. Some of them also have anti-influenza activity. Influenza virus has many similarities with SARS-CoV-2 in their mode of entry into human cells as both are facilitated by the furin protease. Pinobanksin 3-(E)-caffeate, one of the potential leads is a propolis compound. Propolis compounds have shown inhibitory effects against ACE2, TMPRSS2, and PAK1 signaling pathways of SARS-CoV-2 in previous studies. Likewise, quercitrin is structurally similar to isoquercetin, which is currently in clinical trials as possible medication for COVID-19.

Supplementary information: The online version contains supplementary material available at 10.1007/s11224-022-02056-1.

Keywords: COVID-19; Consensus docking; Furin protease; MM-PBSA; Natural products; SARS-CoV-2.

Fig. 1

A graphical representation of the step-by-step approach employed in this study. The work involved screening of natural product library against furin protease using molecular docking, molecular dynamics (MD) simulations, elucidation of molecular interactions, and prediction of biological activity. ANPDB, African Natural Products Database; MM-PBSA, Molecular Mechanics Poisson-Boltzmann Surface Area; PASS, Prediction of Activity Spectra for Substances; ADMET, absorption, distribution, metabolism, excretion and toxicity

Fig. 2

ROC AUC assessment of AutoDock Vina after screening 52 active compounds and 224 decoys against the furin protease. A reasonable AUC value of 0.72 was achieved

Fig. 3

Chemical structure of the bound ligand, para-guanidinomethyl-Phac-R-Tle-R-Amba (MI-1148)

Fig. 4

Furin protease binding pocket. (A) Active site region of furin protease. Labels S1, S2, S3, S4, and S5 are regions within the active site where active ligands are supposed to occupy. Regions S1 and S2 are areas where catalytic triad Ser368 (green), His194 (yellow), and Asp153 (purple) are found, and these were areas most occupied by potential inhibitors. (B) The active site catalytic triad are rendered in sticks representation

Fig. 5

Chemical structures of the potential lead compounds predicted against furin protease. (A) Quercitrin. (B) Teucrol. (C) Malvinidin-3-arabinoside. (D) N-E-caffeoyl tyramine. (E) Pinobanksin 3-(E)-caffeate. (F) Abyssinone IV. (G) ZINC000085967772

Fig. 6

Poses of selected potential lead compounds. Almost all the compounds were found to bind in S1 and S2 regions where catalytic residues are located. (A) Abyssinone IV and (B) Quercitrin had hydrogen or hydrophobic interactions with catalytic residues His194 and Ser368. Residues involved in hydrogen bond interactions with ligands are labelled in black while residues involved in hydrophobic interactions are labelled red. Images were generated with PyMOL (v 2.0.6)

Fig. 7

Root mean square deviation (RMSD) plot of furin-ligand complexes over 100-ns simulation. The backbone after least-squares (lsq) fit to the backbone

Fig. 8

Radius of gyration (Rg) plot of furin-ligand complexes over 100-ns simulation

Fig. 9

Binding free energies of protein–ligand complexes of potential lead compounds from MM-PSA calculations

Fig. 10

Predicted ADME properties of potential lead compounds via SwissADME. (A) Quercitrin. (B) Teucrol. (C) Malvinidin-3-arabinoside. (D) N-E-caffeoyl tyramine. (E) ZINC000085967772. (F) Pinobanksin 3-(E)-caffeate. (G) Abyssinone IV