The crotonylation reader DPF2 promotes the development and progression of colon adenocarcinoma through cell-type-specific immune regulation and metabolic reprogramming

Background The crotonylation reader DPF2 has been implicated in tumor progression, but its role in colon adenocarcinoma (COAD), especially in cell-type-specific immune regulation and metabolic reprogramming, remains unclear. Methods We conducted an integrative multi-omics study combining genetic association analysis for COAD, public transcriptomic validation data, CRC-context single-cell and spatial transcriptomic datasets used as supportive expression evidence, circulating metabolite profiling, and clinical immunohistochemistry. Two-sample Mendelian randomization was used to evaluate the association between genetically predicted expression of crotonylation-related genes and COAD risk, followed by validation and mechanistic analyses centered on DPF2. Results Among 16 crotonylation-related candidate genes with available cis-eQTL evidence, only DPF2 showed a significant positive association with COAD risk. DPF2 was consistently upregulated in GEO datasets, TCGA-COAD, and clinical samples, and high DPF2 expression was associated with advanced disease and poor overall survival. Single-cell MR identified the strongest risk-associated signal in TCL1A + FCER2 + B cells, with a suggestive association in CD8 + S100B + T cells. Supportive analyses using public CRC-context single-cell and spatial transcriptomic datasets further indicated heterogeneous immune-cell expression and tumor-region enrichment of DPF2. In addition, DPF2 was significantly associated with 11 COAD-related circulating metabolites, and mediation analyses suggested partial mediation through Ximenoylcarnitine (C26:1), 16α-hydroxy DHEA 3-sulfate, and Andro steroid monosulfate C19H28O6S (1). Conclusion DPF2 is a crotonylation reader associated with COAD risk and progression. Its effects may involve cell-type-specific immune regulation and partial mediation through metabolic reprogramming. While the primary causal inference of this study focused on COAD, complementary single-cell and spatial transcriptomic findings from broader CRC datasets provided supportive expression-level evidence consistent with these associations.