Selective microbial production of lacto-N-fucopentaose I in Escherichia coli using engineered α-1,2-fucosyltransferases

Lacto-N-fucopentaose I (LNFP I) is the second most abundant fucosylated human milk oligosaccharide (HMO) in breast milk after 2'-fucosyllactose (2'-FL). Studies have reported that LNFP I exhibits antimicrobial activity against group B Streptococcus and antiviral effects against Enterovirus and Norovirus. Microbial production of HMOs by engineered Escherichia coli is an attractive, low-cost process, but few studies have investigated production of long-chain HMOs, including the pentasaccharide LNFP I. LNFP I is synthesized by α1,2-fucosyltransfer reaction to the N-acetylglucosamine moiety of the lacto-N-tetraose skeleton, which is catalyzed by α1,2-fucosyltransferase (α1,2-FucT). However, α1,2-FucTs competitively transfer fucose to lactose, resulting in formation of the byproduct 2'-FL. In this study, we constructed LNFP I-producing strains of E. coli with various α1,2-fucTs, and observed undesired 2'-FL accumulation during fed-batch fermentation, although, in test tube assays, some strains produced LNFP I without 2'-FL. We hypothesized that promiscuous substrate selectivity of α1,2-FucT was responsible for 2'-FL production. Therefore, to decrease the formation of byproduct 2'-FL, we designed 15 variants of FsFucT from Francisella sp. FSC1006 by rational and semi-rational design approaches. Five of these variants of FsFucT surpassed a twofold reduction in 2'-FL production compared with wild-type FsFucT while maintaining comparable levels of LNFP I production. These designs encompassed substitutions in either a loop region of the enzyme (residues 154-171), or in specific residues (Q7, H162, and L164) that influence substrate binding either directly or indirectly. In particular, the E. coli strain that expressed FsFucT_S3 variants, with a substituted loop region (residues 154-171) forming an α-helix structure, achieved an accumulation of 19.6 g/L of LNFP I and 0.04 g/L of 2'-FL, while the E. coli strain expressing the wild-type FsFucT accumulated 12.2 g/L of LNFP I and 5.85 g/L of 2'-FL during Fed-bach fermentation. Therefore, we have successfully demonstrated the selective and efficient production of the pentasaccharide LNFP I without the byproduct 2'-FL by combining protein engineering of α1,2-FucT designed through in silico structural modeling of an α1,2-FucT and docking simulation with various ligands, with metabolic engineering of the host cell.