Powered by: Volkamer lab

This is part of a community effort to rapidly find new hits to target the virus main protease.

The COVID-19 (coronavirus disease 2019) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has become a global health emergency. With no current evidence for specific antiviral treatment, there is an urgent need for effective anti-COVID drugs (more about potential drugs and clinical trials in the COVID-19 Science Report: Therapeutics). A promising target is the main protease M^pro of SARS-CoV-2 for which the first crystal structure has been determined in January 2020. UK’s Diamond Light Source performed a large crystal-based fragment screen on M^pro. In collaboration with PostEra and others, they encourage researchers from around the world to use their fragment hits as a starting point and contribute, amongst others, by suggesting potential inhibitors (effective and easy-to-make).

References:

• Jin, Zhenming, et al. "Structure-based drug design, virtual screening and high-throughput screening rapidly identify antiviral leads targeting COVID-19." bioRxiv (2020). DOI: 10.1101/2020.02.26.964882
• Zhang, Linlin, et al. "Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved α-ketoamide inhibitors." Science (2020). DOI: 10.1126/science.abb3405
• Lai, Chih-Cheng, et al. "Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and corona virus disease-2019 (COVID-19): the epidemic and the challenges." International journal of antimicrobial agents (2020): 105924. DOI: 10.1016/j.ijantimicag.2020.105924

In the following two strategies for finding new compounds fitting the needs of the challenge are proposed.

Note for setting up the pipeline, 5 known protease inhibitors were used as toy examples. Thus, the scripts and notebooks in code might still refer to them.

A rational structure-based approach to select screening compounds was investigated. Therefore, similar binding site for COVID-19 main protease were searched (structure-based, using ProBis), and the active compounds for the similar protein were collected from ChEMBL. Furthermore, due to time-constraints, the focused library was filtered with to keep the compounds having some resemblance with the 22 non-covalent fragments (using MCS).

A representative set of three structures, Mpro-x0387, Mpro-x0946, Mpro-x0967, were then chosen to run multiple dockings with the filtered library using the docking program, smina.

Based on the docking scores, toxicity criteria (trying to avoid respiratory toxicity using estimations from eMolTox webserver) and after visual inspection these compounds were selected (ChEMBL IDs)

• CHEMBL3937948
• CHEMBL2059095
• CHEMBL3690047
• CHEMBL1684519

Note that similar molecules are available in Enamine REAL, such a selection could be added here later (time restriction for now, pipeline available, see strategy B).

Here a structure-based approach was used for building on the complexes of different fragments bound against the virus main protease available from DiamondX.

As starting point in this example, fragment Mpro-x0967 was chosen based on its size, its match in our focused library (see strategy A) and its good estimated affinity using BioSolveIT's SeeSAR).

SeeSAR was then used to grow the fragment choosing a bond towards the bromide tail.

To guarantee synthetic accessibility, similar compounds within Enamine REALspace were searched using FTrees.The found compounds (no duplicates in current postera submissions, 31.03.2020) were cluster to find a diverse subset an the remaining compounds were redocked using SeeSAR.

Based on the fit and the estimated binding affinity the final molecules were selected.

• A-focused_library_docking_screening_pipeline/ contains the results for our strategy A.
• B-fragment_growing_pipeline/ contains the results for the strategy B.
• code/: scripts and notebooks developed to support obtaining the results in both strategies.
• data/: raw data that either can be taken as input in the strategies, or are a byproduct of the pipeline.
• devtools/: metadata files to reproduce the development environment the strategies needed.

1. Input data: Collected input molecules for screening pipeline
   • Approved protease inhibitors (taken from http://dx.doi.org/10.17179/excli2020-1189)
   • SARS-CoV-2 main protease focused ChEMBL subset using binding site comparison
   • Fragments from DiamondX
     • Non-covalent hits in the active site
     • Covalently-bound hits in the active site
   • Current postera submissions (24 march 2020 version)
2. Scripts and methods used for the pipelines
   • Main protease focused ChEMBL subset
     • Structure-based binding site comparison using ProBis
     • Querying ligands from ChEMBL known to bind to these similar proteins
   • Scripts for compound preprocessing and filtering
     • code/utils/: compound preprocessing
     • Notebooks to:
       • Generate 3D conformations
       • Filter by similarity to fragments
       • Select divers subset
   • Docking: Dock selected compounds to target ensemble using smina
   • DiamondX fragment growing
   • Ftrees: Explore available compounds in REAL space (Availability by Enamine)

Yonghui Chen, Dominique Sydow, Jaime Rodríguez-Guerra, Andrea Morger and Andrea Volkamer