Epigenetic Memory from Inflammation May Prime the Gut for Colon Cancer

When colon tissue heals, it looks fine. Under the surface, it may still be listening.

A team led by researchers at the Broad Institute has shown in mice that colonic stem cells retain a lasting epigenetic memory of past inflammation — a chromatin-level scar that persists for more than 100 days after tissue appears fully healed. When those primed cells later acquire a cancer-driving mutation, tumors grow faster than they would in cells with no inflammatory history. Block the central player — a family of transcription factors called AP-1 — and the cancer-promoting effect vanishes entirely.

The paper, published in Nature in March 2026 (first posted as a preprint on bioRxiv in February 2025), comes from the Cancer Grand Challenges PROSPECT team. First author is Surya Nagaraja; corresponding author is Jason D. Buenrostro, a Broad core member and associate professor at Harvard in the Department of Stem Cell and Regenerative Biology. The work was funded through Cancer Grand Challenges, a partnership backed by the National Cancer Institute and Cancer Research UK.

The biological model the team used was colitis — gut inflammation that is the hallmark of inflammatory bowel disease. Patients with ulcerative colitis carry a two- to five-fold elevated risk of colorectal cancer compared to the general population, with those diagnosed in childhood or with pancolitis facing the steepest increases. That elevated risk has long been clinically documented. What has been less clear is the mechanism: why does inflammation that comes and goes leave behind a cancer predisposition even after it stops being visible?

"The missing piece," as Buenrostro put it, "is epigenetic changes."

The PROSPECT team developed a technique called SHARE-TRACE to simultaneously track gene expression, chromatin accessibility, and clonal lineage in individual stem cells — allowing them to watch how inflammatory memory is laid down and passed along. They profiled 52,540 single cells across 23 animals spanning control, acute injury, chronic injury, and recovery groups. The data showed that the memory is not a passive residue but an active, cumulative gain in AP-1 transcription factor accessibility — a progressive opening of chromatin regions that regulate cell growth and stress response. Stem cells carrying the strongest inflammatory memory passed those epigenetic changes to their daughter cells, spawning entire families of primed cells.

The therapeutic implication is direct: if AP-1 activity is what makes the inflammatory memory malignant, then interrupting AP-1 could in principle break the memorys pro-cancer effect. The researchers showed exactly that in mice. Whether AP-1 is a druggable target in humans — and at what point in a patients disease course intervention would actually help — remains open. Transcription factors are notoriously difficult to drug, and the window between inflammation resolving and a cancer mutation appearing may be hard to identify clinically.

The more near-term impact is likely to show up in surveillance strategy. If some UC patients accumulate stronger epigenetic memory than others, a test that detects those chromatin changes could help identify who needs more aggressive monitoring. That is a biomarker story, not yet a therapeutic one, but it is a real and tractable near-term question that follows directly from this work.

There is also a larger epidemiological context the paper lands inside. People born around 1990 have roughly double the risk of colon cancer and quadruple the risk of rectal cancer compared to those born around 1950 — a shift too rapid to be explained by inherited genetics alone. Early-onset colorectal cancer has been one of the most puzzling trends in oncology for the past decade. The PROSPECT finding offers one mechanistic thread: the colons stem cell population may be accumulating inflammatory insults that compound over time, creating a different risk landscape than the one that guided current screening guidelines.

Buenrostro holds patents related to ATAC-seq, a foundational chromatin-accessibility assay, and serves on the scientific advisory boards of Camp4 and seqWell. He is also a consultant at the Treehouse Family Foundation. These relationships are standard in functional genomics and do not constitute a conflict for this work, but they belong in any disclosure.

The paper is a mouse study. The jump to human colorectal cancer is biologically plausible and clinically urgent, but not yet proven. The PROSPECT team has built the case for what to look for in human tissue next. What they have not done is show it there yet. That distinction matters — especially because the surveillance and therapeutic angles depend entirely on whether human colonic stem cells behave similarly.

The PROSPECT project itself is one of the more ambitious efforts in cancer Grand Challenges: a multinational team assembled to study how chronic inflammation drives cancer, with funding deliberately structured to outlast any single labs grant cycle. The breadth of the collaboration shows in the author list and in the scale of the single-cell dataset. The rigor of the claims — the 100-day window, the AP-1 dependence, the clonal transmission — rests on that scale.

For now, the takeaway is this: the body keeps a record of its own inflammation, and that record may be part of why some colons become cancer factories while others do not. Figuring out how to read that record — and whether to act on it — is the work that follows.