Targeting HsDHODH: Shape and machine-learning guided discovery and structural validation of SARS-CoV-2 antivirals

Area of Science:

• Drug Discovery
• Computational Chemistry
• Virology

Background:

• Emerging RNA viruses and antiviral resistance necessitate new therapeutic strategies.
• Human dihydroorotate dehydrogenase (HsDHODH) is a validated target for antiviral drug development.

Purpose of the Study:

• To identify novel HsDHODH inhibitors using an integrated computational and experimental workflow.
• To evaluate the antiviral activity of identified inhibitors against SARS-CoV-2.

Main Methods:

• Integrated computational pipeline: shape-based models, machine learning (ML)-based models, and hotspot analysis.
• Virtual screening of the H3D chemical library followed by similarity-based hit expansion.
• Enzymatic assays, crystal structure determination, and in vitro antiviral assays (SARS-CoV-2 replication in Calu-3 cells).

Main Results:

• Identified four potent anthranilate-based HsDHODH inhibitors (IC50 < 5.0 μM).
• Crystal structures confirmed predicted binding modes and revealed conserved interactions.
• Compounds inhibited SARS-CoV-2 replication in vitro (EC50: 1.3–3.9 μM) with low cytotoxicity (CC50 ≥ 100 μM).

Conclusions:

• The integrated computational-experimental workflow is effective for discovering HsDHODH inhibitors.
• The identified scaffold is a promising starting point for developing novel antiviral agents, including against SARS-CoV-2.
• The described workflow is reproducible and adaptable for accelerating lead identification in medicinal chemistry.