Human milk oligosaccharides

Human milk contains a large quantity of complex oligosaccharides — by mass, more than the protein. The infant cannot digest them: they pass through the small intestine essentially intact. A component present at that concentration, at real metabolic cost to the mother, that the infant does not absorb, demands an explanation, and there are two.

Structure

All HMOs are built on a lactose core, extended with N-acetyllactosamine units and decorated with fucose or sialic acid residues. From that simple grammar comes very large diversity — on the order of two hundred structures characterised — of which 2′-fucosyllactose is the most abundant in most women, alongside lacto-N-neotetraose and the sialyllactoses.

The decoration is not uniform across women. Fucosylation depends on the FUT2 gene: women who are non-secretors, carrying inactive alleles, produce little or no 2′-fucosyllactose while producing other HMOs in compensation. Roughly a fifth of many European populations are non-secretors, with the proportion varying substantially by ancestry. There is therefore no single human HMO profile, which matters when a formula ingredient is described as matching breast milk.

Mechanism one: selective feeding

HMOs reach the colon undigested and are fermented there — but only by organisms equipped to do it. Bifidobacterium longum subsp. infantis is the clearest case: it carries a dedicated gene cluster encoding transporters and intracellular glycosidases that import whole HMOs and break them down inside the cell, rather than degrading them extracellularly.

Importing before digesting is the key detail. It means the breakdown products are not released into the shared environment, so the organism does not feed its neighbours. This is a private nutrient channel, and it explains how a single substrate can shift the composition of an entire microbial community toward one genus. Fermentation then yields short-chain fatty acids and acidifies the colon, which further disfavours many pathogens.

Mechanism two: decoys

Many gut pathogens and viruses initiate infection by binding host cell-surface glycans. HMOs are soluble structural mimics of those glycans. A pathogen that binds a free HMO in the lumen has bound something that will be excreted, and adhesion is a prerequisite for colonisation and invasion.

This is a competitive-inhibition mechanism that requires no antimicrobial activity, and it is well demonstrated in vitro and in animal models for several pathogens.

Making them

Chemical synthesis of these structures is impractical at food scale, so production is by engineered E. coli or yeast fed lactose as the acceptor, with the fucose or sialic acid donor supplied by an engineered intracellular pathway. The characteristic strain-engineering step is negative: the host’s own lactose-utilisation genes are deleted, so it imports lactose but cannot consume it, forcing lactose to serve only as substrate.

Because the usual host is a Gram-negative bacterium, downstream purification must remove endotoxin and residual DNA to a standard appropriate for infants — a food-grade requirement stricter than most.

Clinical evidence is uneven: the mechanisms are solid, and outcome data for added 2′-FL and LNnT in formula shows modest effects on stool microbiota and some infection markers rather than large clinical benefits. Extension into adult supplements rests on weaker evidence still.