CD4 T Cell Help Powers Immunotherapy Response in Tumors Lacking MHC-I

Immune checkpoint blockade has transformed the treatment of mismatch repair-deficient colorectal cancer, yet one of its central puzzles has remained unsolved: many of these tumors lose the very molecules that cytotoxic T cells use to recognize them, and still respond to therapy. A new study published in EMBO Molecular Medicine offers a mechanistic explanation. Using genetically engineered mouse models and a syngeneic transplantable tumor system, researchers led by Marta Requesens, Luciana Gneo and Marco de Bruyn show that when the antigen-presenting machinery of the tumor is disabled, effective immunotherapy depends on a partnership between two T cell populations. Helper CD4+ T cells, acting through the CD40 signaling pathway, license CD8+ T cells to kill tumor cells by a route that does not require major histocompatibility class I, or MHC-I, the molecule whose loss would otherwise render the tumor invisible to conventional cytotoxic T cell recognition.

The team began by asking what happens to immune surveillance when DNA mismatch repair fails. Mismatch repair deficiency, which occurs in roughly 10 to 15 percent of colorectal cancers and can arise from germline mutations in Lynch syndrome, biallelic somatic mutations, or silencing of the MLH1 gene, causes tumors to accumulate mutations at microsatellite repeats and become hypermutated. That flood of neoantigens normally provokes a strong antitumor T cell response, which also creates intense selective pressure for immune escape. To track this process in living tissue, the investigators combined conditional knockout alleles of the mismatch repair gene Mlh1 with mutations in Apc or Braf under an intestinal Cre recombinase, generating mouse models that recapitulate both hereditary and sporadic forms of mismatch repair-deficient colorectal cancer.

The results were striking. Across every genetically engineered model examined, mismatch repair deficiency accelerated intestinal tumorigenesis, and the resulting tumors uniformly showed absent or markedly reduced expression of B2m, the light chain of the MHC-I complex. Alongside this loss of antigen presentation, the tumors excluded cytotoxic CD8+ T cells, while helper CD4+ T cells were retained. In the Apc-deficient models, which allowed direct comparison between mismatch repair-proficient and deficient tumors, gene expression analysis revealed that antigen processing and presentation pathways were among the most downregulated biological processes in tumors relative to normal intestine. Mismatch repair-deficient tumors showed greater depletion of CD8+ T cells, fewer infiltrating CD4+ T cells, and significantly more PD1-positive cells than their proficient counterparts.

High-dimensional spectral cytometry added detail to this picture of immune dysfunction. Within mismatch repair-deficient tumors, a distinct subset of CD8+ T cells co-expressed high levels of PD1 with LAG-3 and ICOS, a phenotype consistent with persistent antigen stimulation and exhaustion-associated dysfunction. Among CD4+ T cells, the researchers observed an expansion of proliferating Ki67-positive cells and of PD1-positive, CD25-high, FoxP3-positive regulatory T cells. Densities of natural killer cells, regulatory T cells marked by FoxP3, CD206-positive M2 macrophages, and CD86-positive costimulatory cells did not differ between mismatch repair-deficient and proficient tumors, pointing to T cell exhaustion and antigen presentation failure as the dominant features of tumor progression in these models.

Because universal MHC-I loss in the autochthonous models prevented direct testing of its consequences, the team turned to a complementary approach. Using CRISPR-Cas9, they deleted Mlh1, alone or together with B2m, in the CT26 colorectal cancer cell line, generating isogenic mismatch repair-deficient tumors with and without MHC-I expression. In immunocompetent Balb/c mice, mismatch repair deficiency slowed tumor growth relative to parental CT26 cells, plausibly because of denser CD8+ and CD4+ infiltration, and this growth defect disappeared in immunodeficient nude mice. Deleting MHC-I restored tumor growth to the rate of parental cells and reduced CD8+ infiltration, while CD4+ and gamma-delta T cell infiltrates were unchanged or increased. Notably, MHC-I loss did not increase infiltration by natural killer cells, which in principle should target MHC-I-negative cells, echoing earlier observations that NK cell function is impaired after tissue entry.

The therapeutic experiments delivered the study’s central surprise. Anti-PD-L1 treatment significantly slowed growth of mismatch repair-deficient tumors but had no effect on mismatch repair-deficient, MHC-I-knockout tumors, consistent with previous work. In contrast, combined anti-PD1 and anti-CTLA4 blockade suppressed tumor growth regardless of MHC-I status. In MHC-I-deficient tumors, neither anti-PD1 nor anti-CTLA4 monotherapy matched the efficacy of the combination, although anti-CTLA4 alone showed a modest effect. Histological analysis of combination-treated MHC-I-knockout tumors revealed necrosis and dense CD8+ infiltration, and bulk RNA sequencing showed that the CD8 cytotoxic T cell genes Cd8a and Cd8b1 were among the most strongly upregulated transcripts, with log2 fold-changes above 5, alongside markers of T cell activation, cytotoxicity, chemokine signaling and inflammatory pathways.

To determine which immune cells were truly required, the researchers depleted specific T cell populations before tumor inoculation and combination therapy. Depleting either CD8+ or CD4+ T cells abolished the therapeutic response, leaving tumors to grow as in vehicle-treated controls, whereas depleting gamma-delta T cells had no discernible effect. Because CT26 cells do not express functional MHC-II, direct CD4+ T cell cytotoxicity could be excluded, and the loss of response after CD8 depletion argued against a dominant role for CD4-reprogrammed myeloid effector cells. The investigators instead suspected a licensing mechanism: CD4+ T cells are known to enhance CD8+ T cell activation through CD40-CD40L interactions, either directly or, more potently, by licensing dendritic cells.

That hypothesis was tested with an elegant rescue experiment. When mice lacking CD4+ T cells were given a CD40 agonist antibody alongside combination checkpoint blockade, the antitumor response was restored, but CD40 agonism provided no benefit in mice depleted of CD8+ T cells. This confirmed that CD8+ T cells are the ultimate effectors and that CD4+ help is the limiting factor. Single-cell gene expression profiling of sorted CD8+ T cells showed that CD4+ depletion before therapy substantially downregulated the cytotoxicity marker Gzma, effector differentiation markers Ly6c and Irf4, the proliferation marker Hmgb3, and inflammatory signaling genes in tumor-draining lymph nodes, with most of these changes rescued by CD40 agonism. Flow cytometry confirmed that granzyme A production by CD8+ T cells, concentrated in a PD1-positive, Ly6C-high effector population, collapsed after CD4+ depletion and returned with CD40 agonism. CD4+ T cells themselves showed little granzyme expression and were characterized mainly by TNF-alpha production.

The authors are careful about the limits of their work. Universal MHC-I loss in the autochthonous models precluded comparison with MHC-I-retained controls, syngeneic tumors were implanted subcutaneously rather than orthotopically, and whether CD40 agonism acts directly on T cells or indirectly through other cells remains unresolved. The specific ligand-receptor interactions that allow CD8+ T cells to kill MHC-I-deficient tumor cells were outside the scope of the study. The findings also diverge from a recent report implicating gamma-delta T cells, and the researchers note that resolving this will require tissue-specific functional models and analysis of human tumors. Nevertheless, the implications are concrete. MHC-I loss does not preclude checkpoint blockade response in mismatch repair-deficient intestinal tumors, so MHC-I status alone should not exclude patients from immunotherapy. Instead, the authors suggest, MHC-I expression could help identify patients in whom the added toxicity of combination therapy over single-agent anti-PD1 is justified, and CD4+ T cell help and CD40 signaling may represent biomarkers or therapeutic targets for extending immunotherapy benefit to MHC-I-low and MHC-I-negative cancers.

Subject of Research: Mechanisms by which CD4+ T cell help enables CD8+ T cell-mediated response to combination checkpoint blockade in MHC-I-negative mismatch repair-deficient colorectal cancer

Article Title: CD4+ T cell-licensed CD8+ T cells are required for immunotherapy response in MHC-I negative MMRd cancer

Article References: Requesens, M., Gneo, L., Tomasevic, T., Rooij, N. V., Plat, A., Xi, M., O’Riordan-Moore, C., Kedzierska, K., Che, N., Glaire, M., Cai, Y., Madrova, M., Wang, L. M., Al-Obaidi, I., Miossec, M., Jansen, M., Lewis, A., Perujo Holland, N., Tomlinson, I., … de Bruyn, M. (2026). CD4+ T cell-licensed CD8+ T cells are required for immunotherapy response in MHC-I negative MMRd cancer. EMBO Molecular Medicine. https://doi.org/10.1038/s44321-026-00538-x

Image Credits: AI Generated

DOI: 10.1038/s44321-026-00538-x

Keywords: immunotherapy, colorectal cancer, mismatch repair deficiency, MHC-I loss, CD4 T cells, CD8 T cells, checkpoint blockade, CD40 signaling, tumor immune evasion, mouse models, anti-PD1, anti-CTLA4