How does claudin-2 protect against EPEC infection and colitis?

Why diarrhea alone cannot explain the protection

Citrobacter rodentium (Cra) is the mouse model for enteropathogenic E. coli (EPEC), an attaching-and-effacing pathogen that damages the intestinal epithelium and produces colitis resembling inflammatory bowel disease. Cra infection raises CLDN2, a tight-junction protein best known for forming paracellular Na+ channels, and an earlier explanation held that the resulting Na+ flux and diarrhea flush the pathogen out. The paper tests that explanation against a competing one: that CLDN2 acts inside the epithelium to drive crypt proliferation and regeneration.

The authors note two observations that weaken the flushing model. CLDN2 is described as dispensable for intestinal Na+ homeostasis, and infected Cldn2 knockout mice still develop diarrhea. In this study, Cldn2 knockout mice carried a higher Cra burden than wild-type mice at both day 7 and day 14 post-infection, and had significantly more diarrhea at day 14. More diarrhea alongside more bacteria is the opposite of what a flushing mechanism would predict, which is why the authors pursue an epithelial-intrinsic route.

CLDN2 rises alongside crypt proliferation

In wild-type mice infected with Cra, inflammation peaked at day 7, measured by body-weight loss, colon thickness and mucosal injury, and showed signs of recovery by day 14. Crypt height, however, kept increasing from day 7 to day 14. CLDN2 protein rose most at day 14, and co-staining showed CLDN2 extending toward the crypt top together with Ki67, a marker of dividing cells. The proportion of Ki67-positive and CLDN2-positive epithelial cells per crypt was higher than in uninfected controls at both time points and higher at day 14 than at day 7.

The same pattern appeared outside the whole animal: Cra infection of IEC-6 cells and of freshly isolated intestinal crypts produced time-dependent increases in CLDN2 and Cyclin D1, a proliferation marker. That makes the CLDN2 rise a crypt-intrinsic response rather than something imposed by circulating immune cells. The correlation is only a starting point, though; the loss- and gain-of-function mice below are what establish direction.

Loss and gain of CLDN2 flip disease severity

The causal test used two independently generated mouse lines: intestinal epithelial-specific Cldn2 knockout (Cldn2 KO) mice and Villin-Cldn2 transgenic (Cldn2 TG) mice overexpressing CLDN2 in the epithelium. Infected Cldn2 KO mice lost more weight, had thicker colons, worse mucosal injury scores and shorter crypts than wild-type littermates at days 7 and 14, with more crypt erosion. Their Cra colonization was significantly higher at both time points.

The transgenic line moved in the opposite direction. After Cra infection, Cldn2 TG mice were protected from weight loss, colon edema and mucosal injury, and their crypts were longer than those of wild-type mice. The molecular readouts matched the anatomy: Cldn2 KO colons kept elevated pSTAT3 and pNFκB at day 14, when wild-type inflammation was returning toward normal, and showed lower Cyclin D1 and higher cleaved caspase-3, while Cldn2 TG colons showed the reverse. Fewer Ki67-positive crypt cells in the knockout and more in the transgenic line tie the disease outcome to proliferation rather than to ion transport.

EGFR activation is required upstream of CLDN2

If CLDN2 is doing the protective work, something must induce it during infection. The paper places EGFR signaling upstream. In Cra-infected IEC-6 cells, the EGFR kinase inhibitor PD153035 prevented the rise in CLDN2. Exogenous IL-22, a cytokine already linked to CLDN2 induction during Cra infection, increased pEGFR and CLDN2 in IEC-6 cells and in cultured crypts, and co-treatment with PD153035 blunted both responses. That positions CLDN2 downstream of an IL-22/EGFR axis.

The in vivo test used gefitinib, an EGFR tyrosine kinase inhibitor, given orally to Cra-infected wild-type mice. Treated mice lost significantly more weight at day 8, had worse colon thickness and injury scores, shorter crypts, suppressed CLDN2 induction, lower Cyclin D1 and higher cleaved caspase-3. In other words, blocking EGFR reproduced the Cldn2 KO phenotype. Notably, pEGFR levels did not differ between wild-type and Cldn2 KO mice, which the authors read as CLDN2 acting downstream of EGFR rather than feeding back on it.

Crypt fetalization is the regenerative state CLDN2 enables

The mechanism the authors propose is crypt fetalization: epithelial cells reverting to a primitive, highly proliferative fetal-like state during infection or injury, a phenomenon previously reported in helminth infection and linked to tissue regeneration. Cra-infected wild-type mice strongly upregulated SCA1, a fetalization marker, along with other fetal colonic epithelial genes including Anxa1, Anxa5 and Tacstd2. Cldn2 KO mice failed to do so and instead showed elevated differentiation markers P21 and LRIG1.

Consistent with a requirement for this reprogramming, 3D-cultured Cldn2 KO crypts did not form spheroids and showed significant necrosis. The interpretation is that CLDN2 permits dedifferentiation toward the fetal-like state, and that this state, not luminal fluid flow, is what supports crypt turnover and clearance of infected cells. This is the authors' mechanistic reading of the data rather than a directly observed clearance event.

A β1-integrin–FAK–YAP axis carries the signal

The signaling link runs through FAK and YAP. FAK and YAP activation were evident in infected wild-type mice but not in Cldn2 KO mice. The FAK inhibitor PF-573228 suppressed YAP and Cyclin D1 while raising differentiation markers in Cra-infected IEC-6 cells and freshly isolated colon crypts, which supports directionality rather than mere association.

Co-immunoprecipitation then showed that CLDN2 physically associates with β1-integrin and pFAK Y397 in freshly isolated colonic epithelial cells from infected wild-type mice, and that this association was absent in Cldn2 KO cells. β1-integrin itself was upregulated by infection in both genotypes, so the difference is complex assembly, not ligand availability. The authors conclude that CLDN2 stabilizes the β1-integrin–FAK complex, allowing FAK recruitment and activation and downstream YAP-driven transcription. They frame this as a possible functional link between tight-junction remodeling and extracellular-matrix sensing, a model rather than a demonstrated structural mechanism.

What the mouse evidence does not settle

Several boundaries qualify how far these results travel. The work is entirely in mice and cell or organoid systems, using the Cra model of EPEC; no human tissue experiment tests whether CLDN2 drives fetalization in patients, and the authors only note that fetal-like crypts have been observed in human IBD. The genetic tools are epithelial-specific overexpression and deletion, so the findings describe epithelial CLDN2 and do not resolve contributions from other cell types.

The authors also flag a counterexample to a simple "more CLDN2 is better" rule: in other contexts, such as T-cell-mediated colitis, Cldn2 TG mice are not protected because enhanced translocation of gut microbiota into the mucosa can promote colitis. The protective direction shown here is therefore tied to the Cra infection setting and to epithelial regeneration, not to CLDN2 elevation in general.

The most useful distinction the paper draws is between two roles for the same protein. Diarrhea-driven flushing and epithelial fetalization both follow CLDN2 induction, but the loss- and gain-of-function data, the EGFR-inhibition phenocopy and the integrin–FAK–YAP complex point to the regenerative route as the one that tracks bacterial burden and injury severity. For readers weighing CLDN2 as a repair target, the relevant readout is crypt proliferation and fetalization status, not stool sodium or diarrhea severity.

Sources used in this answer