How do intestinal epithelial cells program barrier immune memory?

One epitope, four bacterial addresses

The study asks a narrow question with a clean design: if the same peptide is displayed by the same pathogen but rides on different bacterial proteins, does the T cell response change? The authors built isogenic Citrobacter rodentium strains expressing the gp66 CD4 epitope fused to intimin, tir, nleA or espZ. These carrier proteins differ in where they sit relative to the infected intestinal epithelial cell: intimin and tir are membrane-associated, whereas nleA and espZ are delivered into the epithelial cytosol by the type III secretion system.

Holding the epitope constant controls for peptide affinity for MHC class II, a common confounder when comparing responses to different antigens. The readout was the clonal CD4 response of adoptively transferred SMARTA T cells, tracked in the colon and draining lymph nodes after infection. The design therefore isolates antigen localization as the variable rather than antigen identity.

Abundance does not predict immunogenicity

The four carrier proteins were not equally abundant in bacterial–epithelial conjugates. Intimin and espZ were detected at significantly lower levels than tir and nleA, which were the most abundant of those assessed. Yet nleA and espZ induced similar antigen-specific responses despite espZ's lower abundance, and tir, the most abundant, induced a minimal response.

That mismatch is the paper's first key result: compartmentalization, not abundance, was the more critical determinant of immunogenicity. The authors acknowledge that the four carriers differ in structure, stability and turnover, so they built a second comparison in which only subcellular location varied.

Translocation alone drives a 100-fold split

The cleaner test used full-length espZ fused to gp66 and a truncated espZ lacking its N-terminal 21 amino acids, which keeps the protein and epitope identical but removes access to the host cytosol. Both constructs were inserted at the xylE pseudogene locus, and immunoblotting of isolated bacteria-bound colonocytes confirmed comparable abundance, excluding epitope dose as a confounder.

The response bifurcated sharply. At the peak on day 9, full-length espZ induced more than a 100-fold increase in mucosal SMARTA T cells over the truncated variant. The difference was not merely retention in the mucosa, because lymph nodes draining the distal colon showed comparable differences across the time course. With protein, epitope and abundance held constant, cytosolic access alone accounted for the largest difference in the clonal CD4 response.

Priming begins in the cecum, not the colon

Because Citrobacter rodentium first colonizes the cecum before spreading to the distal colon, the authors asked whether the muted response to noncytosolic antigen reflected a different site of priming. Fluorescence in situ hybridization showed focal bacterial colonization at the cecal apex, near but not directly on the cecal patch, a lymphoid structure analogous to Peyer's patches.

Tracking SMARTA cells across regions showed that the cecum-draining distal mesenteric lymph node was the dominant priming site: its response preceded that of distal colon-draining lymph nodes and exceeded it by more than fivefold. The cecal patch response peaked earlier but was roughly tenfold lower than the draining lymph node and about 50-fold lower than the cecal mucosa. This geography matters because it places the relevant antigen presentation in the tissue where the bacteria first establish themselves.

Infected epithelial cells must present directly

The paper then tests whether the response depends on presentation by the infected epithelial cells themselves rather than by professional antigen-presenting cells that might pick up antigen secondhand. Mice with MHC class II deleted specifically in intestinal epithelial cells were compared with floxed controls.

In the absence of epithelial MHC class II, the clonal response was blunted and memory development defaulted to a nonresident fate, generating few memory cells in the intestinal epithelium. The authors interpret this as evidence that direct presentation by infected epithelial cells, which are nonprofessional antigen-presenting cells, is required for the cytosolic antigen to program residency.

Two-way epithelial–T cell talk shapes fate

Single-cell transcriptomic analyses showed sustained, bidirectional communication between epithelial cells and T cells. This communication was needed both for barrier-protective functions and for transcriptional programs that favor tissue residency over central memory differentiation. When epithelial MHC class II was absent, memory differentiation was biased away from epithelial residency and toward the gut-associated lymphoid tissue, as circulating or secondary lymphoid resident memory cells.

The authors propose that the antigen-presenting epithelial cell may not supply indispensable signals for memory subset development. Instead, sustained epithelial interaction may increase T cell exposure to a microenvironment enriched for residency-promoting signals such as TGFβ. That is the authors' model rather than a directly demonstrated causal step, and they note that the transcriptional players they identify await functional characterization.

What the findings do not settle

The loading route for cytosol-delivered antigen remains unresolved. Because injected antigens bypass the classical endolysosomal MHC class II pathway, the authors favor macroautophagy, supported by the absence of KFERQ-like motifs in espZ and nleA and by the greater immunogenicity of cytosolic versus membrane-associated antigens. They explicitly state this will require further study.

The work is also confined to a mouse model and to an attaching-and-effacing pathogen that uses a type III secretion system. The authors note that invasive bacteria such as Salmonella or Listeria have direct entry mechanisms that Citrobacter rodentium lacks, so the localization hierarchy described here may not transfer to pathogens that penetrate epithelial cells intact. The weak response to bacterium-associated intimin was unexpected given that intimin is a target of an IgG response important for clearing infection, and the authors offer uptake of epithelial debris or a cellular intermediate as possible explanations rather than settled mechanisms.

Why cytosolic delivery takes priority

The practical priority the paper establishes is a ranking among antigen classes, not a general rule about antigen dose. For this pathogen, injected effectors were the dominant antigen class programming tissue-resident memory, focusing those responses on epithelium-associated antigens. Linking antigen recognition on epithelial cells to local residency makes biological sense: it deploys memory cells to the site where they will meet antigen again upon reinfection.

For mucosal vaccine design, the implication the authors draw is that targeting antigens to the cytosol of epithelial cells may enhance protective mucosal immunity. The comparison that should guide readers is the one the study actually made: same epitope, same abundance, different subcellular address, more than 100-fold difference in mucosal T cell expansion at peak. That contrast, rather than any single carrier protein, is the paper's central evidence.

Sources used in this answer