Computational prediction of potential inhibitors for SARS-COV-2 main protease based on machine learning, docking, MM-PBSA calculations, and metadynamics

Abstract

The development of new drugs is a very complex and time-consuming process, and for this reason, researchers have been resorting heavily to drug repurposing techniques as an alternative for the treatment of various diseases. This approach is especially interesting when it comes to emerging diseases with high rates of infection, because the lack of a quickly cure brings many human losses until the mitigation of the epidemic, as is the case of COVID-19. In this work, we combine an in-house developed machine learning strategy with docking, MM-PBSA calculations, and metadynamics to detect potential inhibitors for SARS-COV-2 main protease among FDA approved compounds. To assess the ability of our machine learning strategy to retrieve potential compounds we calculated the Enrichment Factor of compound datasets for three well known protein targets: HIV-1 reverse transcriptase (PDB 4B3P), 5-HT2A serotonin receptor (PDB 6A94), and H1 histamine receptor (PDB 3RZE). The Enrichment Factor for each target was, respectively, 102.5, 12.4, 10.6, which are considered significant values. Regarding the identification of molecules that can potentially inhibit the main protease of SARS-COV-2, compounds output by the machine learning step went through a docking experiment against SARS-COV-2 Mpro. The best scored poses were the input for MM-PBSA calculations and metadynamics using CHARMM and AMBER force fields to predict the binding energy for each complex. Our work points out six molecules, highlighting the strong interaction obtained for Mpro-mirabegron complex. Among these six, to the best of our knowledge, ambenonium has not yet been described in the literature as a candidate inhibitor for the SARS-COV-2 main protease in its active pocket.

Developing a SARS-CoV-2 main protease binding prediction random forest model for drug repurposing for COVID-19 treatment

“…[51][52][53] Regression models are not able to distinguish between active and inactive compounds, and thus are not suitable for assisting in drug repurposing. Gomes et al 54 used a pipeline consisting of molecular docking, metadynamics, and machine learning models for screening SARS-CoV-2 main protease inhibitors from compounds in DrugBank. Similar to our approach, they selected 74 SARS-CoV-2 main protease structures from PDB and generated 50 decoys as negatives for each positive.…”

Mpropred: A machine learning (ML) driven Web-App for bioactivity prediction of SARS-CoV-2 main protease (Mpro) antagonists

“…potentially suppress the enzyme activity of the M pro [27]. Nedra et al developed a machine learning approach by employing the support vector machine (SVM) classification model to categorize two hundred novel chemo-types as potentially active against the viral protease using a dataset of two million commercially accessible drugs [28].…”

A Tale of Two Proteases: MPro and TMPRSS2 as Targets for COVID-19 Therapies

“…Other examples of the application of computational tools in the development of M Pro inhibitors are: (i) the already mentioned work of Oerlemans et al who carried out molecular docking and resolved a co-crystal structure with boceprevir [ 53 ]; (ii) the thorough combination of computational tools used by Ngo et al over a database of ~4600 compounds comprising virtual screening, fast pulling of ligand (FPL), and free energy perturbation (FEP), identifying darunavir as potential SARS-CoV-2 M Pro inhibitor [ 105 ]; (iii) the approach followed by Semenov and Krivdin combining modelling of NMR chemical shifts and docking studies that found the natural compound berchemol to be a potential inhibitor [ 106 ]; (iv) the study of Yu et al applying docking, molecular dynamics (MD) simulations and MM-GBSA methods with four HIV protease inhibitors and ribavirin which provided information on blocking M Pro [ 107 ]; (v) Souza-Gomes et al also used a combined approach using machine learning, docking, MM-PBSA calculations, and metadynamics with FDA approved compounds, and they found mirabegron to form the strongest interaction with M Pro [ 108 ]; (vi) the study of Patel et al who applied docking, MD simulations, the free energy of binding, and DFT calculations on a set of 7809 natural compounds and identified theaflavin and ginkgetin as M Pro inhibitors [ 109 ]; (vii) the fluorinated tetraquinolines proposed by El Khoury et al in a computationally driven study using high resolution MD free energy binding calculations and machine learning predictions [ 110 ]; or (viii) the 28 drugs proposed by Piplani et al for repurposing as M Pro inhibitors which resulted from docking studies followed by high-throughput MD simulations of large set of natural products and licensed drugs [ 111 ].…”