A structure-guided computational screening approach for predicting plant enzyme-metabolite interactions.

Area of Science:

• Plant biochemistry and molecular biology
• Computational chemistry and bioinformatics
• Metabolomics and enzyme discovery

Background:

• Plants synthesize diverse metabolites using evolved enzymes, but many enzymes remain unidentified.
• Current gene-based computational methods for enzyme discovery have limitations.
• AI-driven protein structure prediction enables new protein-based screening strategies.

Purpose of the Study:

• To present a rapid, user-friendly, open-source virtual screening method for plant metabolic enzymes.
• To demonstrate the application of this method using Arabidopsis thaliana UGT74F2.
• To facilitate the identification of enzyme substrates, products, or inhibitors.

Main Methods:

• Utilized AI-based protein structure prediction for high-quality protein models.
• Employed virtual screening with AutoDock Vina against a curated library of metabolites and herbicides.
• Analyzed compounds based on relative binding affinities and evaluated binding modes using molecular visualization (PyMOL).

Main Results:

• Presented a validated computational workflow for virtual screening of plant metabolic enzymes.
• Successfully applied the method to Arabidopsis thaliana UGT74F2 as a proof of concept.
• Generated ranked lists of potential substrates or inhibitors for the target enzyme.

Conclusions:

• Virtual screening offers a powerful approach to identify substrates for enzymes with unknown functions.
• This method can aid in revisiting enzyme substrate selectivity and discovering novel inhibitors.
• Computational findings require validation through in vitro or in vivo assays.