Structure-Based Pipeline for Plant Enzymes: Pilot Study Identifying Novel Ginsenoside Biosynthetic UGTs.

Abstract

Models that predict the 3D structure of proteins enable us to easily analyze the structure of unknown proteins. Though many of these models have been found to be accurate, their application in plant proteins is not always entirely accurate. Thus, we aimed to develop a versatile yet simple pipeline that can predict novel proteins with a specific function. As an example, via benchmark studies, we sought to discover novel UDP-glycosyltransferases (UGTs) potentially involved in ginsenoside biosynthesis. Since the functionality of these UGTs has been shown to be determined by a few amino acids, a 3D-structure-based pipeline was required. Our pipeline includes four sequential steps: a sequence-based homology search, AlphaFold3-based 3D structure prediction, docking simulations with ginsenoside intermediates using SwissDock and CB-Dock2, and MPEK analysis to assess interaction stability. Through the application of this benchmark, we optimized the role of each module in the pipeline and successfully identified four novel UGT candidates. These candidates are predicted to catalyze the conversion of protopanaxadiol (PPD) to compound K (CK) or protopanaxatriol (PPT) to ginsenoside F1. This pilot study demonstrates how our pipeline can be used for the functional annotation of plant proteins and the discovery of enzymes involved in specialized pathways.