The global consumption of sugary foods is a significant health concern, linked to obesity, diabetes, and other diseases. Low-calorie sweeteners offer a potential solution, and steviol glycosides, extracted from *Stevia rebaudiana*, are promising candidates due to their intense sweetness and negligible caloric content. These glycosides consist of a steviol aglycone core with variable glycan attachments at the R1 (C13-hydroxyl) and R2 (C19-carboxylate) positions. The specific glycan structure dictates both sweetness intensity and taste quality (organoleptic properties). While stevioside (ST) and rebaudioside A (Reb A) are abundant but have an unpleasant aftertaste, rebaudioside D (Reb D) and M (Reb M) are intensely sweet and palatable but naturally present in trace amounts. The key difference lies in β(1→2) glucosylation at the R2 end, which is poorly catalyzed by the native *Stevia* enzyme UGT91D2. This research focuses on OsUGT91C1, a rice glycosyltransferase with homology to UGT91D2, to explore its potential in efficiently producing Reb D and Reb M through β(1→2) glucosylation at the R2 end. The study aims to characterize OsUGT91C1 biochemically and structurally to understand its catalytic mechanism and improve its efficiency for industrial applications.

Previous research has identified four key glycosyltransferases (UGTs) in *Stevia rebaudiana* responsible for steviol glycoside biosynthesis: UGT85C2 (adds glucose to R1), UGT74G1 (adds glucose to R2), UGT91D2 (β(1→2) glucosylation at R1), and UGT76G1 (β(1→3) glucosylation at R1 and R2). UGT91D2's limited activity at the R2 end limits Reb D and Reb M production. Studies have shown that the specific glycosylation patterns, particularly β(1→2) glucosylation at the R2 end, are crucial for the desirable taste profile. Other research has identified OsUGT91C1 (EUGT11) in rice, a homolog of UGT91D2, suggesting its potential for improved R2 glucosylation. Understanding the structure-function relationship of OsUGT91C1 is key to enhancing its catalytic activity for the large-scale production of high-value steviol glycosides.

This study employed a multi-pronged approach combining biochemical assays and structural biology techniques. OsUGT91C1 was expressed and purified from *E. coli*, and its activity was assessed using liquid chromatography-mass spectrometry (LC-MS) with various steviol glycoside substrates and UDP-glucose. Crystal structures of OsUGT91C1 were determined in different states (apo, complexed with UDP and different steviol glycosides, and a mutant complex) using X-ray diffraction. Site-directed mutagenesis was used to engineer OsUGT91C1 variants to improve β(1→2) glucosylation at the R2 end and eliminate β(1→6) glucosylation. Steady-state enzyme kinetic assays were performed using a UDP-Glo™ Glycosyltransferase Assay Kit to determine kinetic parameters (kcat, Km, kcat/Km) for the wild-type enzyme and its mutants with different substrates. The LC-MS data provided information on reaction products, while crystal structures revealed the enzyme's active site and substrate binding modes. Structural analysis and mutagenesis experiments were integrated to engineer improved enzyme variants.

OsUGT91C1 efficiently catalyzes β(1→2) glucosylation at both the R1 and R2 ends of various steviol glycoside substrates. It also exhibits β(1→6) glucosylation activity at the R1 end. Crystal structures revealed three distinct substrate binding modes, explaining the enzyme's promiscuity. The R2 end binding involves a catalytic dyad (His27 and Asp128) essential for β(1→2) glucosylation. Glu283 plays a crucial role in substrate recognition. The enzyme's flexibility allows for steviol aglycone rotation and glucose flipping, enabling access to different hydroxyl groups for glycosylation. Mutations like F208M enhanced β(1→2) glucosylation, while H93W and F379A eliminated β(1→6) glucosylation. The double mutant F379A/F208M showed a significant increase in kcat/Km for β(1→2) glucosylation, especially at the R2 end, representing a promising catalyst for Reb D and Reb M production. The study also confirmed that OsUGT91C1 can catalyze the reverse reaction, cleaving β(1→2) glycosidic bonds.

This study provides a detailed understanding of OsUGT91C1's catalytic mechanism and its ability to produce valuable steviol glycosides. The findings address the research question by demonstrating OsUGT91C1's potential to overcome the bottleneck in Reb D and Reb M production due to limited R2 β(1→2) glucosylation. The engineered OsUGT91C1 variants significantly improve the yield of these desirable sweeteners. The results are significant as they offer a biocatalytic route for producing palatable steviol glycosides at an industrial scale. The insights into the enzyme's flexibility and substrate recognition can guide the design of other improved glycosyltransferases for various applications.

This research successfully characterized the catalytic flexibility of OsUGT91C1 and engineered variants with improved activity for producing palatable steviol glycosides. The engineered enzyme F379A/F208M offers a promising biocatalytic approach for industrial-scale production of Reb D and Reb M. Future work could focus on further optimizing the enzyme for even higher yields and exploring its application in synthesizing other valuable glycosides.

The study primarily focused on the in vitro activity of OsUGT91C1. Further in vivo studies are needed to validate its effectiveness in a biological system. The crystal structures may not fully represent the dynamic nature of substrate binding in solution. While the engineered variants show improved activity, further optimization might be possible to achieve even higher yields and selectivity.