Background

In patients with inflammatory bowel disease (IBD), the colonic epithelium accumulates populations of mutated cells that expand and persist under chronic inflammation. These include mutations that weaken IL-17 and NF-κB signaling, or inactivation of the chromatin regulatory gene ARID1A. In contrast, colitis-associated cancer (CAC) is characterized by the selection of mutations that directly drive tumor growth, such as APC, KRAS, and TP53. It remains unclear what organizational features of the tissue microenvironment drive the divergence of these two cell populations.

Reconstructing the early stages of tumorigenesis using only patient tissue is challenging. Mutant clones are rare, the scope of tissue sampling is limited, and the effects of treatment on the cellular ecosystem are difficult to disentangle. In particular, it is unclear whether mucosal repair is simply a response to tissue damage or whether it provides an opportunity for the expansion of oncogenic clones. This has not been examined at the spatial level.

The researchers used mice lacking the mucin Muc2, which develop chronic colitis and dysplasia. They combined lineage tracing, targeted mutation analysis, spatial transcriptomics, and computational modeling to examine not only the types of mutations present but also the 'cellular neighborhood' in which these clones reside.

Key Findings

Confetti lineage tracing showed that in Muc2-deficient colons with persistent inflammation, large clones spanning multiple crypts were significantly more abundant at 7 months than at 2 months. In a computational model designed to allow crypts to repair damaged areas, the repair process itself promoted the neutral expansion of surrounding clones. When the crypt fission probability was set to 0.5 for normal clones and 0.95 for advantageous clones, the number of computational steps required for repair decreased from 115,315 to 80,000, and the advantageous clones accumulated more significantly in the damaged area.

The researchers administered the mutagen N-ethyl-N-nitrosourea (ENU) to 8-week-old Muc2-deficient mice and analyzed the colons 4 months later. Analysis of 16 tissue sections obtained from 5 mice using 10x Genomics Visium identified 16 distinct cellular neighborhoods, reflecting epithelial, immune, muscle, and tumor components. The 'repair neighborhood' with high expression of the fetal epithelial marker Trop2 and the neutrophil-rich neighborhood were positively correlated with the tumor area, with R = 0.81 and R = 0.66, respectively. The proportion of Trop2-positive cells in the tissue was also correlated with the number of tumors, with R = 0.68.

A notable feature of the study was the division of the same tissue into 288 biopsies with a diameter of 2 mm, followed by repeated sequencing of 22 genes related to IBD and CAC. Significant positive selection was observed for missense and nonsense mutations in Ctnnb1 and Apc, with the frequency of Ctnnb1 mutant alleles reaching up to 27%. The malignancy score, calculated as the proportion of tumor-associated mutations, was highest for Smad4 (41%) and Ctnnb1 (37%).

In contrast, an 'immune-resistant neighborhood' was observed. The malignancy scores for Pigr, Arid1a, Nfkbiz, Il17ra, and Il17rc were only 4-13%, and these mutant clones were mainly distributed outside the tumor. ARID1A-deficient clones increased along with the immune-resistant neighborhood but showed a negative correlation with the tumor and repair neighborhoods. This suggests that healthy epithelium diverges into repair and immune-resistant lineages, leading to different outcomes in tumorigenesis.

Significance and Outlook

This study demonstrates that cell fate cannot be predicted by mutation alone. The Trop2-positive repair environment that forms after damage increases crypt fission, providing a larger territory for oncogenic mutant clones. In contrast, non-tumorigenic clones that are insensitive to inflammatory signals may compete with oncogenic clones, potentially delaying tumor development. This is contrary to the common belief that mutations that are advantageous for surviving colitis are also advantageous for cancer.

However, the correlation between the repair environment and tumor burden alone cannot establish causality. It is also necessary to determine whether neutrophils directly increase genomic instability or whether pre-existing tumors induce a repair response through separate manipulation experiments. A limitation is that the key findings are based on a mouse model treated with ENU. Although the researchers confirmed the presence of ARID1A-deficient clones outside the tumor and increased ARID1A expression within the tumor in human CAC tissue, further validation is needed to link patient size and clinical course.