Escherichia coli has a reputation problem. Pull up any bacteriology textbook and you’ll find it filed under pathogens, a microbe capable of hospitalising infants and contaminating water supplies. Yet in the gut of nearly every healthy breastfed baby, E. coli sits quietly at low abundance, outnumbered a hundredfold by Bifidobacterium, and has apparently been doing so for as long as humans have been nursing their young. Nobody could quite explain why. A new study published in Nature Communications suggests the answer lies in a chemical bargain brokered by breast milk itself.
The research, led by Professor Lindsay Hall at the University of Birmingham and involving a European team tracking 41 Dutch infants through their first year of life, reveals that E. coli and Bifidobacterium are locked in a genuinely mutualistic relationship, one in which each bacterium supplies something the other cannot make for itself. The intermediary, improbably enough, is the sugar in breast milk.
Human milk is replete with complex sugar molecules called oligosaccharides (HMOs), structures so elaborate that the infant’s own digestive system cannot break them down. They reach the large intestine essentially intact, where Bifidobacterium species snap them up and dismantle them using specialised enzymes. This has been understood for some time; it is, broadly speaking, why breastfed infants develop Bifidobacterium-rich gut microbiomes while formula-fed infants generally do not. What was not understood is how E. coli fits in. The bacterium lacks any machinery to attack HMOs directly. And yet there it is, present in 74% of infants at two months old and 96% at six months, comfortably co-existing with its more specialised neighbour.
The cross-feeding mechanism turns out to be rather elegant, and rather mutual. When B. bifidum dismantles the most abundant HMO, a molecule called 2′-fucosyllactose, it releases the component sugars, including fucose and galactose, into the gut environment. E. coli, a metabolic generalist with a high-affinity sugar transport system, scavenges these fragments. That part, cross-feeding between gut bacteria, was not entirely surprising. The twist is what E. coli gives back.
B. bifidum, it turns out, cannot synthesise cysteine, an amino acid essential for its metabolism. It is what biologists call an auxotroph, meaning it depends on an external source of the compound to survive. In a series of co-culture experiments, Hall’s team showed that E. coli, a cysteine prototroph capable of making its own supply, effectively donates the molecule to its auxotrophic partner. Remove the cysteine-producing capacity from E. coli using a targeted gene deletion and co-culture growth collapses significantly. Restore it, and both species thrive. As Hall puts it: “Our previous work, and that of others has already shown that human milk oligosaccharides feed Bifidobacterium. The exciting new development is the way that HMOs being eaten by Bifidobacterium also supports E. coli, something that has not been demonstrated before.”
Microbes from Different Worlds
The two species arrive in the infant gut by quite different routes, which makes their partnership all the more striking. Metagenomic sequencing of infant and maternal stool samples revealed that Bifidobacterium strains are commonly transmitted from mother to child, passed during birth or early contact, and they often persist through the first year as recognisably identical lineages. E. coli is different. The strains found in infants do not match those in their mothers and seem to originate from elsewhere, possibly the birth environment or early physical contact with others. Yet once established in an individual infant, specific E. coli strains can persist for months, suggesting the bacterium carves out a stable ecological niche despite entering from outside the family.
The sequencing data also showed higher genetic diversity within E. coli populations than within Bifidobacterium, consistent with a generalist strategy: multiple strains coexisting, competing, and turning over, versus the tight host-adapted lineages typical of B. bifidum in particular. “E. coli was first isolated from infant faeces in 1885 by Theodor Escherich,” noted Dr David Seki, first author of the paper. “Since then, this organism has shaped genetics and molecular biology like no other. However, fundamental gaps remain in our understanding of its ecology. Specifically, what factors determine whether E. coli adopts a commensal or pathological phenotype?”
To probe community dynamics more broadly, the team developed a novel computational pipeline called MAJIC, which assesses how strongly any given species constrains or expands the ecological space available to others. The analysis found no evidence of antagonism between Bifidobacterium and E. coli at any age point, which is perhaps counterintuitive given that bifidobacterial fermentation produces acids capable of suppressing E. coli growth in laboratory cultures at very low pH. The resolution, the team argues, is that the pH inside a healthy breastfed infant’s colon stays above roughly 5.5, well above the level at which acid becomes lethal to E. coli, and host bicarbonate secretion buffers the environment further. Strict exclusion by acidification appears not to operate in term-born infants; something more nuanced is going on.
Stability and What Could Destabilise It
What keeps this relationship stable rather than spiralling? Mutual dependence of the sort demonstrated in vitro, where neither partner can grow without the other’s contribution, can in principle generate boom-and-bust dynamics, the coupled populations either blooming or collapsing together. In the healthy infant gut, the team argues, several forces dampen that tendency: competition from other species, immunological regulation by the infant’s developing immune system, the spatial patchwork of the intestinal mucosal surface, and, perhaps most importantly, the presence of residual lactose that offers E. coli a food source independent of any Bifidobacterium activity. In premature infants, where lactase function is reduced, cysteine availability is limited, and the gut environment is more chemically extreme, those stabilising forces may be weaker, potentially tipping the interaction toward the stronger mutual dependence seen in the experimental cultures.
The findings also carry an implication that challenges a longstanding framing of early gut colonisation. Hall and her colleagues suggest that E. coli, at low levels, may not simply be a tolerated nuisance in the infant microbiome but rather a contributor: colonising niches that might otherwise be occupied by genuinely pathogenic Enterobacteriaceae, providing a cysteine supply to Bifidobacterium, and possibly helping shape immune calibration in early life. “This discovery sheds light on a mutualistic relationship in which each bacterium supports the other, and that both may be required for stable co-existence,” Hall said. “E. coli, rather than being viewed solely as harmful, may, at low levels, play a beneficial role in immune system maturation.”
Whether that framing holds across diverse infant populations, and in formula-fed infants who lack HMOs entirely, remains to be tested. The study cohort was exclusively Dutch and exclusively breastfed, a deliberate design choice that provided a clean dietary model but limits generalisation. Stable isotope tracing of carbon fluxes between species in the living gut, something not yet technically feasible at scale, would offer the most direct test of whether the cross-feeding loop operates in vivo as it does in the lab.
What the study does open, fairly clearly, is a design space for intervention. If stable, healthy gut microbiome development depends partly on a reciprocal chemical relationship between Bifidobacterium and E. coli, synbiotic supplements designed for preterm infants or those who cannot receive consistent breastfeeding might need to consider both partners rather than targeting Bifidobacterium alone. The field has spent decades thinking of E. coli as the bacterium to keep out. It may turn out to be one of the bacterium to keep.
Source: Seki et al., “Human milk oligosaccharide mediates mutualism between Escherichia coli and Bifidobacterium bifidum,” Nature Communications, 2026. DOI: 10.1038/s41467-026-71764-7
Frequently Asked Questions
Is E. coli actually beneficial in babies, or is it always dangerous?
At low levels, E. coli appears to play a genuinely useful role in the healthy infant gut, providing an amino acid called cysteine to Bifidobacterium and potentially occupying niches that might otherwise be filled by more harmful bacteria. The danger arises when E. coli overgrows, which the normal breastfed gut environment seems to prevent through a combination of competition, immune regulation, and the dominance of Bifidobacterium. So the answer, somewhat uncomfortably, is both: the same species can be beneficial or dangerous depending on abundance and context.
Why does breast milk help keep gut bacteria balanced in a way formula doesn’t?
Breast milk contains complex sugars called human milk oligosaccharides (HMOs) that the infant’s own gut cannot digest, so they pass intact to the large intestine where Bifidobacterium dismantles them. This gives Bifidobacterium a large, exclusive food source unavailable to most other bacteria, allowing it to dominate the early gut and indirectly regulate species like E. coli through the cross-feeding dynamics described in this research. Formula lacks HMOs, which is one reason formula-fed infants tend to have less Bifidobacterium-rich microbiomes and often higher E. coli abundance.
How does a bacterium like E. coli end up in a baby’s gut if it doesn’t come from the mother?
The study found that E. coli strains in infants rarely matched those in their own mothers, suggesting the bacterium arrives from the broader environment, possibly birth settings, skin contact, or the surrounding household. Despite this external origin, specific E. coli strains can persist within an individual infant for months, carving out a stable ecological position. This is quite different from Bifidobacterium, which is frequently transmitted directly from mother to child and maintains recognisable lineages over time.
Could this research change how probiotics or synbiotics are designed for preterm babies?
Potentially, yes. If healthy microbiome development depends partly on a mutualistic chemical exchange between Bifidobacterium and E. coli, supplements for infants who cannot receive consistent breastfeeding may need to account for both partners rather than targeting Bifidobacterium alone. Preterm infants are particularly vulnerable because the gut conditions that normally stabilise this relationship, adequate lactase function, buffered pH, and available cysteine, are often disrupted. The researchers suggest this opens a design space for rationally engineered interventions, though clinical testing would be required before any such product reaches use.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.