Enzymatic and cell factory approaches to the production of human milk oligosaccharides.

Infant formula milk companies try to develop fortified formula milk that mimics human milk as closely as possible, since it is well-known that breast milk has considerable implications in the development of the infant in the first years of life. Human milk is unique in terms of complex oligosaccharides content, known as human milk oligosaccharides (HMOs). Their role in the development of intestinal flora blocking the attachment of pathogens and modulating the immune system of the infant are currently recognized. Due to these biological effects, there is a great interest to introduce the main HMOs in the infant formula milk. Therefore, efficient synthetic strategies for HMOs production are required. Here we present a complete review of HMO production using either (chemo)enzymatic syntheses or cell factory approaches, focusing on the strategies that produce HMOs at least at the milligram scale. 42 HMO structures have already been produced as free sugars. Whereas short HMOs are well obtained by cell factory approaches, complex and branched HMOs are better produced by chemoenzymatic strategies. Inspite of the current advances, production strategies of some biologically relevant HMOs are still missing.

Structural-Functional Analysis Reveals a Specific Domain Organization in Family GH20 Hexosaminidases.

Hexosaminidases are involved in important biological processes catalyzing the hydrolysis of N-acetyl-hexosaminyl residues in glycosaminoglycans and glycoconjugates. The GH20 enzymes present diverse domain organizations for which we propose two minimal model architectures: Model A containing at least a non-catalytic GH20b domain and the catalytic one (GH20) always accompanied with an extra α-helix (GH20b-GH20-α), and Model B with only the catalytic GH20 domain. The large Bifidobacterium bifidum lacto-N-biosidase was used as a model protein to evaluate the minimal functional unit due to its interest and structural complexity. By expressing different truncated forms of this enzyme, we show that Model A architectures cannot be reduced to Model B. In particular, there are two structural requirements general to GH20 enzymes with Model A architecture. First, the non-catalytic domain GH20b at the N-terminus of the catalytic GH20 domain is required for expression and seems to stabilize it. Second, the substrate-binding cavity at the GH20 domain always involves a remote element provided by a long loop from the catalytic domain itself or, when this loop is short, by an element from another domain of the multidomain structure or from the dimeric partner. Particularly, the lacto-N-biosidase requires GH20b and the lectin-like domain at the N- and C-termini of the catalytic GH20 domain to be fully soluble and functional. The lectin domain provides this remote element to the active site. We demonstrate restoration of activity of the inactive GH20b-GH20-α construct (model A architecture) by a complementation assay with the lectin-like domain. The engineering of minimal functional units of multidomain GH20 enzymes must consider these structural requirements.