Gut Bacterial Toxin Binds Claudin-4, Helping Cleave Cell-Surface E-cadherin
Background: The first step in toxin-mediated intestinal epithelial injury
A receptor through which the toxin from enterotoxigenic Bacteroides fragilis (ETBF) acts on intestinal epithelial cells has been identified. A study published in Nature presents experimental evidence that claudin-4 binds the toxin directly and helps it cleave the cell-surface adhesion protein E-cadherin.
The study concerns toxin-producing strains, not B. fragilis as a whole. These strains have been reported to be epidemiologically associated with neoplastic lesions in the human colon and induce tumor formation in susceptible mice. The key factor, B. fragilis toxin (BFT), is a metalloprotease that cuts the extracellular portion of E-cadherin, damaging the epithelial barrier and causing inflammation. However, the receptor it binds and how it promotes cleavage were unclear. Earlier biochemical experiments could not reproduce direct cleavage, prompting proposals that host-cell signaling or another enzyme might be involved.
Key findings: Receptor identified through gene knockout and binding experiments
The researchers conducted a CRISPR knockout screen targeting 18,675 protein-coding genes in HT29 colon cancer cells. Because BFT does not kill cells, they selected the top 10% of cells that retained the most surface E-cadherin after toxin exposure, rather than selecting for survival. After two rounds of selection, claudin-4 emerged as the strongest candidate.
Removing claudin-4 from HT29 cells increased their resistance to BFT, measured by E-cadherin cleavage, to about 20 times that of wild-type control cells. Reintroducing claudin-4 restored toxin susceptibility. Removing claudin-3 alone had no clear effect, but removing both proteins increased resistance to an average of about 500 times that of wild-type controls. This supports the idea that claudin-3 partly compensates for claudin-4. These fold changes describe resistance to cleavage in cell experiments; they do not represent a reduction in cancer risk.
Separate binding experiments confirmed a direct interaction between BFT and claudin-4. BFT-mediated cleavage was not detected in 293T cells expressing E-cadherin alone, but it occurred when claudin-4 was co-expressed. In T84 cell monolayers designed to model intestinal epithelial architecture, claudin-4 removal also mitigated the BFT-induced drop in transepithelial electrical resistance. These results support a model in which BFT acts at the cell surface and receptor binding helps it directly cleave E-cadherin.
Implications and outlook: Potential for toxin-trapping strategies and limits of the evidence
The researchers also tested whether CLN4sol, a soluble claudin-4 analogue, could block toxin activity. They injected 10 μg of BFT together with 300 μg of the analogue into the ceca of female mice and assessed the animals four hours later. Epithelial cell shedding and edema were reduced, and membrane E-cadherin was largely preserved. A control analogue in which the extracellular portion was replaced with a claudin-2 sequence did not show the same protective effect.
However, this was an acute-injury model in which the toxin and candidate compound were mixed before being administered together. The findings do not establish treatment efficacy after natural infection has progressed, long-term tumor suppression, or prevention of human colorectal cancer.
Gaps also remain in the detailed mechanism. The researchers propose that claudin-4 concentrates BFT near the cell membrane, helping it reach the E-cadherin cleavage site. An AlphaFold 3 complex prediction supports this model, but confidence in the predicted extracellular region of claudin-4 that contacts the toxin was low, so the molecular binding mode remains unresolved. How toxin in the intestinal lumen crosses tight junctions to reach E-cadherin, and whether additional receptors exist, are also questions for future work. The findings define a more specific target for antitoxin development, but clinical application requires further validation.
The study design and results can be reviewed in the provided open full text on Europe PMC of the Nature paper.
Nature, Published online: 22 April 2026; doi:10.1038/s41586-026-10375-0. CRISPR screening shows that claudin-4 is a receptor for the pro-carcinogenic B. fragilis toxin on colonic epithelial cells, and that this interaction promotes cleavage of E-cadherin, leading to epithelial barrier disruption and inflammation.
This study proposes an antitoxin development strategy that blocks binding to a host receptor instead of inhibiting BFT’s enzymatic activity itself. Soluble receptor analogues could trap the toxin, while candidate compounds could be assessed by measuring cell-surface E-cadherin preservation and barrier function. The researchers note that this approach might avoid off-target inhibition of similar human metalloproteases. However, it is not a treatment with established selectivity and safety. The effect observed in mice was also limited to conditions in which the toxin and candidate compound were injected together. The potential for treating infection or preventing cancer in people cannot be judged until the effects of post-infection dosing, intestinal delivery and stability, long-term safety, and whether the approach suppresses tumors are evaluated.