The Door a Gut Bacterium's Toxin Uses on Colon Cells Has Finally Been Found

Researchers led by Johns Hopkins have found how a toxin made by a common gut bacterium latches onto colon cells, filling in a missing step in a suspected route to colon cancer.

The protein it grabs is called claudin-4. A decoy version of it diverted the toxin and protected mice from damage, according to a study published in Nature on April 22, 2026 ^[2]. The team's own summary, released by Johns Hopkins Medicine and carried by ScienceDaily on October 31, puts it in plainer terms: a long-suspected culprit now has a known way in ^[1].

The finding is real and carefully done. It is also early. It does not show that the decoy prevents cancer, and it gives no one a reason to ask for a test or treatment today.

The suspect and the missing step

Bacteroides fragilis lives in the human gut, and most strains are harmless. Only the enterotoxigenic ones make a toxin called BFT. Earlier studies linked those strains to colorectal neoplasia, the abnormal growths that can turn into cancer, and showed they can drive tumours in susceptible mice ^[2].

Researchers also knew what BFT does when it arrives. It cuts E-cadherin, a protein that helps hold neighbouring colon cells together and keeps the gut barrier intact. Earlier work described in the study ties that barrier damage to inflammation and to increased cell proliferation ^[2].

What nobody knew was how the toxin got a grip on the cell in the first place. The Nature paper says the receptor, the surface molecule a toxin must bind before it can act, had never been identified ^[2].

"We've made several attempts over time to identify the receptor, so this is an exciting moment," said Cynthia Sears, the study's senior author ^[1].

How they found it

The team, with collaborators at Harvard Medical School and the Molecular Biology Institute of Barcelona, used a genome-wide CRISPR screen ^[2].

The idea is simple to state. Switch off every gene in a dish of cells, one gene per cell. Add the toxin.

Keep the cells that still have E-cadherin on their surface, then see which missing gene they have in common.

The cells were HT29, a line of colon cancer cells. Claudin-4 was the strongest hit ^[2]. Claudin-4 is a tight-junction protein, part of the seams that seal the gaps between cells lining the colon.

Removing it made the cells about 20 times more resistant to BFT cutting E-cadherin. Removing it together with its close relative claudin-3 produced about 500-fold resistance, which suggests claudin-3 can partly stand in when claudin-4 is gone ^[2].

A screen only points. So the researchers tested the link directly. In laboratory experiments, BFT and claudin-4 bound to each other and formed a one-to-one complex, and that binding helped the toxin cleave E-cadherin at the cell surface ^[2][3].

The decoy

The team then built a soluble claudin-4, a free-floating version meant to catch the toxin before it reached the colon wall. In mice, the decoy diverted BFT and protected against the damage it normally causes ^[1][3].

That result needs reading with care. The mice were protected from toxin-induced damage. The experiment did not test whether the decoy stops colorectal cancer, and it was not tried in people ^[1][2].

Study author Maxwell White said the team is looking at ways to improve the approach, including small molecules or other biologic drugs with better pharmacological properties ^[1][3]. The researchers also say the exact three-dimensional structure of the toxin sitting on claudin-4 has not yet been worked out ^[1][3]. Drug designers usually want that picture before they build anything precise.

What it means for your own gut

The evidence comes from cultured cells, laboratory binding experiments and mice. The reported work did not establish this receptor mechanism in samples from patients ^[2][3]. A chain of events that holds in a dish and in a mouse can still behave differently in a human colon.

Carrying the bacterium is not a diagnosis either. The researchers report that BFT-producing B. fragilis has been detected in up to 20 percent of healthy people, and that carrying it does not mean a person will develop cancer ^[1][2]. By that estimate, roughly one in five people may have the organism without trouble, so a test that merely finds it would sweep up many people who are never going to be ill.

That is why the screening, blocking and vaccine ideas the finding invites remain ideas. Nothing in the study establishes a test, a vaccine or a therapy for people.

What the study does is replace a correlation with a mechanism. Before, the case rested on toxin-producing bacteria turning up alongside neoplasia. Now there is a named receptor, a measurable effect of removing it, and an intervention that worked in an animal. For a field that has spent years suspecting this bacterium, that is how a suspicion starts to become a target.

Several questions are open. Can a decoy or a redesigned drug block BFT safely in people, and reach the parts of the gut where it matters? Would blocking it reduce tumours in an animal model, rather than just tissue damage? And can the mechanism be seen directly in patient tissue? The next papers will have to answer those before anyone talks about prevention.