TL;DR
Researchers have developed a comprehensive 3D database for the SARS-CoV-2 proteome, integrating experimentally determined structures with novel computational models. This resource aims to accelerate drug discovery by providing a user-friendly platform to explore viral protein structures, identify potential drug targets, and analyze the impact of mutations.
What Problem Does This Paper Address?
The rapid spread and high mortality rate of SARS-CoV-2 have spurred an urgent need for effective antiviral therapies. Developing targeted drugs requires a deep understanding of the virus's protein structures (proteome) and how these proteins interact with host cells. However, experimentally determined 3D structures are not available for all viral proteins, creating a gap in knowledge. This paper addresses this challenge by creating a comprehensive 3D database that combines existing structural data with high-quality computational models for the entire SARS-CoV-2 proteome, making it accessible for drug discovery efforts.
Methods
The researchers employed a multi-pronged approach to build their SARS-CoV-2 3D database:
- Proteome Modelling: They used the MODELLER software to build 3D models for proteins lacking experimental structures. This involved using PSI-BLAST and other sequence alignment methods to identify suitable templates from existing protein databases. They also developed a novel tool, ProtCHOIR, for building homo-oligomeric structures.
- Structure Validation: Models were rigorously assessed using tools like MolProbity and jsPISA to ensure quality and identify potential binding sites.
- Mutation Impact Prediction: Various computational tools (mCSM-Stability, DeepDDG, PROVEAN, etc.) were used to predict how single nucleotide mutations might affect protein stability and function.
- Protein-Protein Docking: They mapped known viral-human protein interactions using D3.js Force-Directed Graph for visualization and employed ClusPro for protein-protein docking.
- Ligand Virtual Screening: FDA-approved drug libraries were screened against key viral proteins (Nsps and S protein) using molecular docking (Glide) and MM/GBSA methods to identify potential drug candidates.
Datasets
The primary datasets used in this study include:
- GenBank: The SARS-CoV-2 genome sequence (MN908947.3) was obtained from GenBank.
- Protein Data Bank (PDB): Experimentally solved structures of SARS-CoV-2 and related coronaviruses were retrieved from the PDB.
- FDA-approved drug library: A library of 1930 compounds from the eDrug 3D web-resource was used for virtual screening.
- Viral-human protein-protein interactions: Data from Gordon et al. (2020) was used to map these interactions.
The paper introduces the SARS-CoV-2 3D database itself as a novel resource, which integrates these diverse data types into a user-friendly platform.
Key Insights and Ideas
- Comprehensive Proteome Coverage: The database achieves an impressive 97.5% sequence coverage for the SARS-CoV-2 proteome through a combination of experimental structures and high-quality models, filling crucial structural gaps.
- Focus on Oligomeric Structures: Recognizing that many viral proteins function as complexes, the database includes models of homo- and hetero-oligomers, providing a more biologically relevant view.
- Integrated Drug Discovery Platform: The database is not just a collection of structures but a functional tool that integrates structure, mutation analysis, protein-protein interactions, and ligand docking, streamlining the drug discovery pipeline.
- Identification of Potential Drug Repurposing Candidates: Virtual screening identified existing FDA-approved drugs like Mitoxantrone and Protokylol as potential hits against specific SARS-CoV-2 targets, offering a faster path to therapeutic development.
Key Results
- 97.5% Proteome Coverage: The database provides 3D structural information for nearly the entire SARS-CoV-2 proteome.
- Model Quality: The average MolProbity score for the modelled structures was 2.52, indicating good quality.
- Identification of Potential Drug Hits: Virtual screening identified several FDA-approved drugs as potential inhibitors for key viral proteins, including Mitoxantrone for Nsp3 and Protokylol for Nsp15 and Nsp16.
- Mutation Impact Analysis: The study identified specific residues in key proteins (S, Mpro, Nsp12) that are highly sensitive to mutations, providing insights into viral evolution and potential drug resistance.
Limitations and Future Work
- Model Accuracy: While efforts were made to ensure high-quality models, they are still computational predictions and may not perfectly reflect experimental reality, especially for regions with low sequence identity to known structures.
- Protein Stability Prediction Limitations: Some protein stability prediction tools require proteins to be in their apo forms, meaning the impact of glycation and cofactor binding on stability could not be fully evaluated.
- Future Updates: The authors plan to continuously update the database with new experimentally solved structures as they become available, incorporate pathway analysis, and model additional protein-protein interactions. They also aim to improve model accuracy as new templates emerge.
Summary updated