Multi-strategy Engineering of a Glycosyltransferase for More Sustainable and Efficient Biosynthesis of Mogroside IE

Mogroside IE is a valuable natural sweetener precursor, but its industrial production is hampered by inefficient plant extraction and nongreen chemical synthesis. Herein, we report a computational-experimental engineering strategy that transformed a putative glycosyltransferase, UGTX1, with negligible initial activity (2.4% conversion) into a highly efficient biocatalyst. This was achieved by synergistically combining machine learning (ESM-2) for zero-shot prediction, dynamic cross-correlation network analysis (DCCNA) for allosteric control, and structure-guided tunnel engineering to rationally design a focused mutant library. Screening identified beneficial single-point mutations, which were mapped onto the enzyme structure to define three functional zones (core, tunnel, and shell). Guided by this spatial insight, a focused rational iterative site-specific mutagenesis (FRISM)-based combinatorial assembly yielded the quintuple mutant UGTX1-M5, which exhibits a 54-fold increase in catalytic efficiency (kcat/Km = 1.13 × 104 M–1 s–1). The engineered enzyme enabled the scalable synthesis of mogroside IE in a simple, green aqueous-phase system, achieving >98.5% conversion of 1.5 mM mogrol within 14 h at 35 °C without exogenous metal ions. Molecular dynamics simulations revealed that the mutations collectively stabilize a catalytically competent conformation, enlarge the substrate-binding pocket, and optimize interaction networks. This work establishes an efficient and sustainable route to mogroside IE and provides a generalizable multi-strategy engineering paradigm for unlocking the potential of unexplored enzymatic resources.