Anaerobic gut bacteria: why the good ones do not fit in a capsule

Healthy state

There is almost no oxygen in the colon. Measurements in the caecum of mice found values below 1 mmHg, compared with around 160 in room air. The little oxygen that seeps out of the gut wall forms a gradient: some remains right at the lining, practically none in the middle of the gut contents. Accordingly, oxygen-tolerant bacteria sit near the lining and strict anaerobes Anaerobes Bacteria that live without oxygen. For strict anaerobes oxygen is even toxic; they make up the large majority of the gut flora. live in the interior.

Strict anaerobes are bacteria for which oxygen is a poison. They lack the enzymes to neutralise its aggressive by-products. They make up the large majority of the gut flora Microbiome, gut microbiota The community of all microbes in the gut, mainly bacteria, plus viruses, fungi and archaea. Most live in the colon. More in the chapter “Building microbial diversity”: Faecalibacterium prausnitzii, Roseburia, Eubacterium rectale (now often called Agathobacter rectalis), Akkermansia muciniphila Akkermansia A gut bacterium that lives on the mucus of the gut wall and stimulates its renewal. In studies, high levels often go along with good metabolic health. More in the chapter “The mucus layer and additives” and Ruminococcus bromii. Bifidobacteria are not strict anaerobes — they tolerate some oxygen, though they thrive best without it. These are exactly the species that studies link with health.

The gut maintains its own low oxygen. Butyrate Butyrate A short-chain fatty acid and the main energy source for the cells of the colon wall. It strengthens the barrier and dampens inflammation. More in the chapter “Butyrate and other short-chain fatty acids”, which these bacteria make from fibre, is the main fuel of colon cells. They burn it with oxygen and use the oxygen up in the process. Mouse studies showed that the cells use a switch protein called PPAR-γ for this, and that without butyrate more oxygen reaches the gut interior. So the anaerobes use their own metabolic product to create the conditions they need.

They need three things to thrive. First food: indigestible carbohydrates that reach the colon. Second partners, because hardly any species manages the breakdown alone. Bifidobacteria break fibre down to acetate and lactate, Faecalibacterium turns that into butyrate. Ruminococcus bromii cracks resistant starch granules that others then feed on. Third an intact mucus layer Mucus layer A gel layer of mucins on the gut wall that keeps bacteria at a distance. In the colon, the inner layer is almost free of microbes. More in the chapter “The mucus layer and additives”: Akkermansia muciniphila lives on the mucus itself and releases sugars that others use in turn.

Faecalibacterium prausnitzii is the best studied example. It is reduced in people with Crohn’s disease, and after bowel surgery a low proportion predicted a higher risk of relapse. In the laboratory, the cell-free supernatant of the culture alone blocked a central inflammation switch, and in cell experiments the release of inflammatory messengers fell. One curiosity: the bacterium tolerates small amounts of oxygen somewhat better when riboflavin (vitamin B2) is present, because it can then pass electrons outwards.

Disrupted state: causes and effects

Which brings us to the heart of it: you cannot simply sell these bacteria in a capsule. They are highly sensitive to oxygen and easily lose their viability at every step from the fermenter through drying, filling and storage to being taken. Specialists call oxygen sensitivity the central obstacle for such products. So what sits on the shelf advertising “butyrate producers” or “rebuilding your gut flora” almost always contains something else: lactic acid bacteria, bifidobacteria or Bacillus spores, which survive air and stomach acid.

Research is working on it. In 2023 a group described isolating Faecalibacterium prausnitzii together with a partner bacterium and gradually adapting the culture to oxygen. The product was tolerated by mice and humans, but could only be detected in the gut of some participants. That is a feasibility study, not evidence of benefit.

Even if the bacteria arrive alive, they usually do not stay. A study that took samples directly from participants’ gut lining found that an eleven-strain probiotic Probiotics Live microorganisms that are meant to provide a benefit in sufficient amounts. The effect depends on the individual strain. More in the chapter “Using fermented foods well” settled in some people while others’ own flora fended it off. In stool that difference was invisible. Your gut is not an empty bed but a crowded one. If you already have a stable flora, foreign species get in only with difficulty.

The few real trials with anaerobes in humans are small. For Faecalibacterium prausnitzii, a first study in Crohn’s disease appeared in 2026: eight people, open label, no control group. It was tolerated, but endoscopy improved in nobody, and the study stopped early. The authors call it a proof of concept.

Akkermansia muciniphila shows the dilemma especially clearly. In the EU it is not the living bacterium that is authorised but the pasteurised, that is killed, form, as a novel food up to 5 × 10¹⁰ cells a day, not for pregnant or breastfeeding women. The authorisation says something about safety only. The efficacy study behind it had 40 participants, of whom 32 completed the three months; the pasteurised form improved insulin sensitivity, while the weight loss was not statistically meaningful. A multi-strain product with Akkermansia and butyrate producers only just reached its main goal, a flatter rise in blood sugar after a test meal, in 76 people with type 2 diabetes; the long-term value HbA1c fell by 0.6 percentage points as a secondary goal. The study came from the manufacturer, and all authors worked there.

Then there are the tests. Commercial stool analyses like to report “too little Akkermansia” or “too little Faecalibacterium” and sell the matching product alongside. A comparison found that the same sample gives very different results from different providers. A number from such a test is no basis for treatment.