Beijing /PRNewswire/ - In a campaign to screen drugs for the treatment of COVID-19, the disease caused by SARS-CoV-2, StoneWise, a company specialized in applying artificial intelligence to innovative drug design solutions for innovative drug R&D enterprises and institutions, has conducted research and been sharing the electronic structure data of more than 1400 nucleoside inhibitors against various RNA polymerases from patents up to February 5, 2020. Using its proprietary AI-driven drug design platform, StoneWise has ranked dozens of nucleosides from these molecules as potential inhibitors against SARS-CoV-2 RNA polymerase.
StoneWise hereby makes all the related data freely available to the scientific community for further research and development. The download links are enclosed at the end of this article.
Furthermore, StoneWise and Institute of Materia Medica, Chinese Academy of Medical Sciences and Peking Union Medical College have reached a strategic research collaboration in the COVID-19 treatment. StoneWise will use its state-of-the-art AI-based drug discovery technology and graph-extraction algorithm in the collaboration to expedite the progress.
Sharing of structure data of more than 1400 RNA polymerase nucleoside inhibitors from patents
The National Medical Products Administration (NMPA) of China has swiftly approved several clinical trial applications for the treatment of COVID-19 recently, in which much hope is being rest on Remdesivir (GS-5734) due to its broad spectrum anti-RNA polymerases (RdRp) activities, the promising efficacy against SARS, MERS and Ebola viruses in pilot studies, and its safety profile in clinical trials so far [1-3]. The primary sequence alignment of SARS-CoV-2 nsp 12 (Gene Bank code:YP_009725307.1) and SARS-CoV nsp 12 (PDB code: 6NUR_A) by Blast test [4-5] indicates the identity of the two sequences is up to 96%, and is completely identical in the NTP-binding domains (the domains 612-626, 678-688 and 758-763 of SARS-CoV-2 nsp 12). This lays scientific foundation for the argument that RdRp-targeted drugs from other virus infectious diseases have the potential for COVID-19 treatment.
We expect Remdesivir (GS-5734) become a promising treatment against the COVID-19 in the near future. However, the SARS-CoV-2 is not exactly the same as SARS, MERS or Ebola viruses. Its micro-environment may demand different drug properties, and its drug resistance may also differ from other viruses. Therefore, RdRp inhibitors with distinct property profiles against SARS-CoV-2 are deemed necessary to develop. In the past decades, humans have synthesized many small-molecule nucleosides as RdRp inhibitors against various viruses, and the efficacy and toxicity for some of the molecules have been extensively studied in many laboratories. These molecules, including GS-5734, share similar fundamental structural features and pharmacophores against RdRp, and therefore they are most likely to become the potential high-value compound library to discover the better "fit" molecules specific to SARS-CoV-2. These molecules should be studied more deeply in a wider scope for SARS-CoV-2.
In this study, StoneWise retrieved more than 50 patents related to nucleoside-based RNA polymerase inhibitors (including the GS-5734 and its analogs developed by Gilead), and obtained more than 1400 compounds using StoneWise's AI-based graph-extraction algorithm in combination with manual annotation for structure extraction and processing.
Note: The structure data provided herein is derived from the keyword-based patent search, so mistakes and data missing are inevitable. The data is for reference of scientific research use only.
Potential application of RNA polymerase nucleoside inhibitors in inhibiting the SARS-CoV-2 RNA polymerase
After obtaining these embodiment compounds, StoneWise has conducted series of research and discovered dozens of nucleoside molecules including GS-573 parent drug structure to have potential inhibiting effect against the SARS-CoV-2 RNA polymerase. Details of small molecule processing, SARS-CoV-2 RNA polymerase homologous modeling, virtual screening, MD simulation and the AI-based proprietary compound screening technologies are in the experimental methods section at the end of this article. Limited by the size of this article, only some representative molecules are shown here.
The above molecules have been studied for anti-viruses, anti-flu and anti-HCV applications, including some basic research of druggability using prodrug strategy. Their synthetic accessibility has been validated and this will expedite the development phase if these molecules have been selected for further studies.
StoneWise is keen to make some contribution to the fundamental research for COVID-19 treatment via sharing our research progress and relevant data. Only with the fundamental research mature enough, can we have more robust and sufficient response strategy for the current and future epidemic outbreak.
Thanks for all the project members of their devotion to this research. Specifically we thank Prof. Liu Zhenming of Peking University for his support and assistance for this project.
Appendix: Introduction to the Experimental Methods
- Protein homologous modeling
Prior to protein homology modeling, HMM-HMM was used in the homologous multiple sequence alignment analysis based on SARS nsp 12 (PDB code: 6NUR_A) and Norwalk RdRp structure (3BSO). Taken 6NUR_A as a template, we constructed the initial SARS-CoV-2 RdRp 3D structure through single point mutation method referring to the primary sequence difference in the SARS nsp 12 (PDB code: 6NUR_A) and the SARS-CoV-2 RdRp (Gene Bank code: YP_009725307.1). In the initial SARS-CoV-2 RdRp structure's NTP binding domains 612-626, 678-688 and 758-763, we constrained the corresponding conformation at the CTP binding site of the 3BSO, and constrained the chelation configuration of Mn2+ ion with the key amino acid residues of SARS-CoV-2 RdRp (D618, Y619, D761 and D760). For the region of RNA model & primer binding domains of the COVID-19 RdRp, we constrained the conformation in the double-stranded RNA binding domains of the 3BSO. During the modeling, we retained the RNA, metal ions and key solvent molecules in the 3BSO. At the end of the modeling, the C515 of the mutated RNA chain was U, G and A respectively, and four models including C were obtained.
- Molecular docking
For the molecular docking, the triphosphoryl portion of GS-5734 was forced to maintain the correct chelation configuration with the Mn2+ ion, while the remaining portion was allowed to move freely. The SARS-CoV-2 RdRp (U) and GS-5734 complex obtained by docking is used for the molecular dynamics simulation in the next step.
- Molecular dynamics simulation
The SARS-CoV-2 RdRp (U) -GS-5734 complex is subject to parameterization, solvation and charge balance in the Charmm-GUI, and subsequently undergone a short NPT simulation at 600ps for complex energy optimization and hydrogen bond optimization. The resultant stable conformation from the simulation was selected for the subsequent docking/virtual screening operation.
- Virtual screening
The aforementioned 1400 compounds retrieved from patents were pre-processed by eliminating the prodrug functional moieties in the embodiment compounds and triphosphorylating the 5-CH2-OH at ribose to create final list of molecules for virtual screening. 450 molecules obtained after removal of duplicates are used for virtual screening in the SARS-CoV-2 RdRp (U) model. Ensemble docking was performed on the SARS-CoV-2 RdRp (A, C and G) models, coupled with StoneWise's proprietary AI-based prediction models to rank the activities of the molecules, and the screening results are summarized in electronic file for download.
 Sheahan, T.P., Sims, A.C., Leist, S.R. et al. Comparative therapeutic efficacy of remdesivir and combination lopinavir, ritonavir, and interferon beta against MERS-CoV. Nat Commun 11, 222 (2020).
 Tchesnokov, E.P.; Feng, J.Y.; Porter, D.P.; Götte, M. Mechanism of Inhibition of Ebola Virus RNA-Dependent RNA Polymerase by Remdesivir. Viruses 11, 326 (2019).
 Wang, M., Cao, R., Zhang, L. et al. Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (COVID-19) in vitro. Cell Res (2020).
 Kirchdoerfer, R.N., Ward, A.B. Structure of the SARS-CoV nsp12 polymerase bound to nsp7 and nsp8 co-factors. Nat Commun 10, 2342 (2019).
Data acquisition methods
The methods to acquire the related data and results generated in this study are as follows:
(1) Structured data of RNA polymerase nucleoside inhibitors
The downloaded data is available in the csv, xlsx and sdf formats.
(2) Virtual screening of SARS-CoV-2 polymerase nucleoside compounds and homologous modeling used in this study
The downloaded contents include:
1. SARS-CoV-2 and GS-5734 complex models
2. Docking results based on SARS-CoV-2-RdRp_GS-5734_complex.pdb
3. Ensemble docking results of other three base mutation models
4. GS-5734 and SARS-CoV-2 RdRp interaction diagram
Go to http://www.stonewise.cn/Report_en, and enter the structure data and docking result download page.
Thank you for your interest and support, and look forward to your valuable opinions. If you have any suggestion and/or collaboration requests, please send email to firstname.lastname@example.org. StoneWise will get in touch with you as soon as possible.
StoneWise, founded in 2018, is a technology company committed to accelerating novel drug discovery with artificial intelligence. We intend to use AI technology and innovation to lead a paradigm shift in Drug discovery for the pharmaceutical industry. Our goal is to replace the previous single-engine drug R&D focusing onexperts based model with the new twin-engine drug R&D model with advanced AI technology and expert system. Our platform is designed by cutting-edge technologies such as artificial intelligence and microfluidics, integrated with computational chemistry, computational biology, pharmacology and clinical medicine. The Platform provides pharmaceutical companies with integrated solutions. These solutions include a vast range of tools such as medical knowledge graph, the discovery of priviledged scaffolds, molecular generation and design, as well as predicting reaction products and planning path synthesis. Using this platform, StoneWise hopes to enhancing the efficiency and success rate of early stage drug discovery.