Efficient fermentative production of lactodifucotetraose by controlling sequential glycosyltransferase reactions in <i>Escherichia coli</i>

Lactodifucotetraose (LDFT) is a human milk oligosaccharide (HMO) that might reduce inflammation in infants. In this study, we established a useful production process of LDFT by engineering two key enzymes, α1,2-fucosyltransferase (α1,2-FucT) and α1,3-fucosyltransferase (α1,3-FucT). First, we verified which of 2'-fucosyllactose (2'-FL) or 3-fucosyllactose (3-FL) (mostly unverified) was more useful. We searched for FucTs that functioned efficiently in vivo against the raw material lactose or the two intermediates 2'-FL or 3-FL by external substrate addition to culture medium. We found that α1,2- FucT (HMFT) from Helicobacter mustelae and the N-terminal truncated form of α1,3-FucT from Bacteroides fragilis (BfFucTΔN10) had high potential. 3-FL was not efficiently converted to LDFT, which might be attributed to the low reactivity of HMFT to 3-FL as well as the low uptake efficiency of 3-FL by LacY, as revealed by a growth test with exogenously added FL as the sole carbon source and heterologously expressed intracellular fucosidase. Furthermore, because 3-FL accumulation had a negative impact on cell growth, we avoided the route passing through 3-FL. By adjusting the copy numbers of HMFT and BffucTΔN10, we produced LDFT from lactose predominantly via 2'-FL. Finally, 17.5 g/L of LDFT (with 6.8 g/L 2'-FL and no 3-FL or residual lactose) accumulated in a 3-L fed-batch culture after 77 h. This study reports the detailed analysis of multiple pathways and shows the control of glycosyltransferases can improve the production efficiency of complex HMOs.