Deep learning- and biochemical activity-based screening strategy for rapid discovery of COVID19 drug candidates: inhibition for both SARS-CoV-2 cell-entry and 3CL protease replication

Authors

• D Rajapakse Todd
• J Fan
• D Daily
• L Dong
• K Peterson
• R Fariss
• G Wistow

Abstract

To enhance COVID19 treatment development and gain experience for similar pandemic response, a systematic yet comprehensive elucidation of biological mechanism involving the inhibition of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is highly needed. To propose an efficient yet reliable drug discovery strategy to overcome this terrible disease, both cell-entry and intra-cell inhibitions should be considered. The role played by the cell surface biopolymer, heparan sulfate (HS), is crucial for viral entry bio-process. Inside the host cell, the compounds with high-performance inhibition to viral replication should be preferred. Thus, the ideal candidate compound should be competitive in both aspects, instead of being one-sidedly emphasized. In this study, we proposed a combinational drug discovery strategy for anti-SARS-CoV-2 inhibitors screening. The virtual screening part was based on a double-layer deep learning algorithm that consists of both molecular graph processing and descriptors correction. The experimental validations of the screened compounds were crossly conducted upon 3-chymotrypsin-like (3CL) based, pseudotyped particle (PP), ${\alpha}$-Synuclein fibrils uptake and cell-based live SARS-CoV-2 (CPE) assays. Based on this strategy, we successfully identified the lead compounds that display high inhibitions for both viral entry and intra-cell replication; and their efficacy was also verified by PRNT and iPSC-derived 3D lung model. To further elucidate the reaction pathways of these inhibitors, quantum mechanics (QM) calculations were conducted for microscopic binding observations under biological environment. This novel strategy is instructive for future biological investigations; and the proposed virtual screening workflow is also a robust complementary for qualitative high-throughput screening (qHTS) based drug discovery.

Scientific Focus Area: Computational Biology

This page was last updated on Monday, September 25, 2023