Back

Microbial induction of MHC-II expression in colon cancer cells overcomes immunotherapy resistance and limits metastasis

Chung, C.; Ozcelik, E.; Zhang, J.; Shen, Z.; Eskiocak, O.; Qin, Y.; Habel, J.; Ozler, K.; Subhash, S.; Aminzada, Z.; Dev, G.; Garcia, L.; Lyons, S. K.; Hand, T.; Rivadeneira, D. E.; Fox, J. G.; Westcott, P. M.; Rogava, M.; Beyaz, S.

2026-07-04 immunology
10.64898/2026.06.30.735621 bioRxiv
Show abstract

Colorectal cancer remains a major cause of cancer mortality, and most microsatellite stable tumors derive little benefit from immune checkpoint blockade. Here, we identify a microbiome-dependent mechanism that converts immune-refractory colorectal cancer into a more immunologically responsive state. Using orthotopic mouse models spanning distinct genetic and immunologic contexts, we show that a Helicobacter-containing microbiome suppresses primary tumor growth and limits metastasis. This protective state is associated with increased intratumoral lymphocyte infiltration and stronger effector programs. Mechanistically, microbial exposure induces MHC class II expression in colon cancer cells to promote anti-tumor immunity. Tumor-intrinsic loss of CIITA abrogates microbial protection, whereas enforced CIITA expression is sufficient to increase intratumoral T cell accumulation, restrict progression and metastasis, and sensitize microsatellite-stable tumors to PD-1 and CTLA-4 blockade. In human microsatellite-stable patient-derived organoids, increased cancer-cell MHC-II enhanced interactions with autologous immune cells and increased tumor cell apoptosis. Together, these findings define a microbiome-cancer cell antigen presentation axis that restrains metastasis and overcomes immunotherapy resistance in colorectal cancer.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.