Mutability landscape guided engineering of a promiscuous microbial glycosyltransferase for regioselective synthesis of salidroside and icariside D2

Microbial glycosyltransferases efficiently synthesize glucosides and have garnered increasing interest. However, limited regioselectivity has impeded their broad application, particularly in the pharmaceutical industry. In this study, the UDP-glycosyltransferase YjiC from Bacillus licheniformis ( Bl YjiC) was engineered to achieve the bidirectional regioselective glycosylation of tyrosol and its derivatives. Initially, site-directed saturation mutagenesis was performed on two newly identified substrate-binding cavities in the acceptor pocket of Bl YjiC to provide a comprehensive blueprint of the interplay between mutations and function (mutability landscape). Iterative saturation mutagenesis was performed, guided by the mutability landscape. Two highly regioselective mutants M6 (M112L/I325Y/L70R/Q136E/I67E/M77R) and M2’ (M112D/I62L) were generated, exhibiting >99 % regioselectivity toward the alcoholic and phenolic hydroxyl of tyrosol, respectively, compared with the wild-type (product mixture: 51:49 %). Both mutants exhibited excellent regioselectivity toward several dihydroxy phenolic substrates, offering valuable biocatalysts for the regioselective synthesis of glucosides. Their application was confirmed in a short synthesis of salidroside (3.6 g/L) and icariside D2 (2.4 g/L), which exhibited near-perfect regioselectivity. This study provides valuable insights into future protein engineering of similar enzymes and opens new avenues for their practical applications.