The chemical tag in question is N4-acetylcytidine, abbreviated ac4C, a modification installed on messenger RNA mainly by the enzyme NAT10. Of the more than 170 known RNA modifications, ac4C is the only acetylation event known to occur on eukaryotic messenger RNA. Previous work has shown that ac4C enhances mRNA stability and translational efficiency, thereby governing cell cycle dynamics, metabolic reprogramming, and oncogenic behavior. What remained conspicuously missing, the researchers note, was a systematic map of the global ac4C landscape in colorectal cancer and an understanding of how it shapes the tumor microenvironment, particularly the cancer-associated fibroblasts that act as its primary architects.

To fill that gap, the team quantified ac4C modification patterns across multiple independent cohorts drawn from the Gene Expression Omnibus and The Cancer Genome Atlas, using single-sample Gene Set Enrichment Analysis to compute a continuous ac4C score for each patient. Rather than relying on consensus clustering, this approach assigned every tumor an absolute enrichment value for the ac4C signature. Stratifying patients at the median score revealed two molecularly distinct subgroups, and principal component analysis and t-distributed stochastic neighbor embedding confirmed clear transcriptional separation between them. Crucially, the classification held up across four independent validation datasets, suggesting that ac4C-based subtyping is not an artifact of any single platform or population.

The clinical implications of this stratification were striking. Patients in the high-ac4C group were significantly enriched for aggressive features, including advanced nodal stage and distant metastasis, and the proportion of high-ac4C tumors rose steadily as pathological stage advanced. Pathway analysis showed that high ac4C scores correlated strongly with epithelial-mesenchymal transition and TGF-beta signaling, the canonical engines of invasion and spread, whereas low scores aligned with cell cycle regulation, fatty acid metabolism, and microsatellite-stable biology. Survival analysis reinforced the picture: the high-ac4C subgroup exhibited markedly worse overall survival, a finding reproduced in both training and external validation cohorts.

Profiling of the tumor microenvironment added a crucial dimension. Using CIBERSORT for immune cell deconvolution and the ESTIMATE algorithm to compute stromal and immune scores, the researchers found that high-ac4C tumors harbored dense, stroma-rich microenvironments. These tumors showed upregulated expression of key immune checkpoint and inhibitory molecules, including LAG3, CD14, LILRB2, SIRPA, and CD8A, alongside a correlation with innate immune components such as NK cells, monocytes, and macrophages. The authors interpret this combination as evidence that ac4C overactivity fosters an immune-excluded, fibrotic niche, one in which physical barriers of extracellular matrix impede immune infiltration while inhibitory signals exhaust the cytotoxic cells that do arrive.

To convert these population-level patterns into a clinically usable tool, the team intersected differentially expressed genes between the ac4C subgroups with core ac4C regulators and applied LASSO-Cox regression to derive a prognostic gene signature. The resulting ac4C score predicted poor overall survival with high area-under-the-curve values at one, three, and five years, and a nomogram integrating the score with standard clinicopathological features outperformed traditional TNM staging alone. Pharmacogenomic predictions added a therapeutic twist: high-ac4C tumors were predicted to resist conventional agents such as cisplatin, gemcitabine, and doxorubicin, yet to show heightened sensitivity to targeted kinase inhibitors including pazopanib and gefitinib. The authors caution that these are algorithmic, hypothesis-generating inferences rather than validated treatment guidelines.

The study’s most consequential discovery emerged from single-cell RNA sequencing. Mining the scCancer Explorer database, the researchers identified CERCAM, or Cerebral Endothelial Cell Adhesion Molecule, as the gene carrying the highest coefficient weight in their prognostic model. Single-cell mapping showed that CERCAM was expressed almost exclusively within the cancer-associated fibroblast cluster, with virtually no expression in epithelial tumor cells, T cells, or B cells. High-resolution subclustering restricted to 10X Genomics datasets to eliminate batch effects revealed further specificity: CERCAM was enriched in matrix-remodeling, myofibroblastic CAFs, the subpopulation responsible for collagen deposition and extracellular matrix crosslinking, while remaining low in inflammatory CAFs and resting normal fibroblasts.

Epigenetic and clinical evidence converged on the same target. DNA methylation analysis showed that CERCAM promoter methylation was highest in metastatic cells, intermediate in primary tumor cells, and lowest in normal cells, hinting at a coordinated dual layer of regulation involving both RNA modification and DNA methylation during tumor spread. In a cohort of 70 paired fresh tumor and adjacent normal tissue samples, 60 patients showed higher CERCAM expression in tumor tissue than in matched normal tissue. Quantitative PCR across cell lines confirmed that CERCAM expression in cancer-associated fibroblasts dramatically exceeded levels in normal fibroblasts and all epithelial or tumor cell lines tested.

Functional experiments then tested whether fibroblast-derived CERCAM actually drives tumor behavior. The researchers knocked down or overexpressed CERCAM in CAFs and co-cultured the modified fibroblasts with RKO and HCT116 colorectal cancer cells in Transwell systems that permit paracrine signaling without direct contact. When CERCAM was depleted from the fibroblasts, co-cultured tumor cells showed significantly impaired proliferation in CCK-8 assays, along with reduced migration in wound-healing assays and diminished invasion through Matrigel-coated inserts. The result demonstrates that the ac4C network does not merely act within cancer cells; it also reprograms stromal cells to emit paracrine signals and remodel the extracellular matrix in ways that push tumors toward metastasis.

The authors are careful to delineate the limits of their evidence. Direct acetylation of CERCAM messenger RNA by NAT10 has not yet been experimentally validated, and the mechanistic link between ac4C modification and the observed CERCAM DNA methylation remains hypothetical, pending techniques such as ac4C-RIP sequencing. Discrepancies between single-cell and qPCR sensitivity, an uneven clinical validation split of 60 high versus 10 low expressors, and the inherent limitations of bulk-transcriptome drug prediction algorithms all warrant caution. Even so, the study establishes a compelling new taxonomy for colorectal cancer, one in which an RNA chemical mark predicts prognosis, sculpts a fibrotic and immune-excluded microenvironment, and points to CAF-derived CERCAM as a promising target for dismantling the immunosuppressive stroma that shields these tumors from both drugs and immune attack.

Subject of Research: ac4C RNA modification profiling and CERCAM-expressing cancer-associated fibroblasts in colorectal cancer prognosis and tumor microenvironment regulation

Article Title: Comprehensive Profiling of ac4C RNA Modification Identifies CERCAM in Cancer‐Associated Fibroblasts as a Key Prognostic and Microenvironmental Regulator in Colorectal Cancer

Article References: Xia, Y., Dai, J., Zhou, X., & Zhang, R. (2026). Comprehensive Profiling of ac4C RNA Modification Identifies CERCAM in Cancer‐Associated Fibroblasts as a Key Prognostic and Microenvironmental Regulator in Colorectal Cancer. Cancer Reports, 9(9), Article e70677. https://doi.org/10.1002/cnr2.70677

Image Credits: AI Generated

DOI: 10.1002/cnr2.70677

Keywords: colorectal cancer, ac4C, RNA modification, epitranscriptomics, NAT10, CERCAM, cancer-associated fibroblasts, tumor microenvironment, prognostic signature, single-cell RNA sequencing, immune exclusion, drug sensitivity