Single-cell and spatial transcriptomics reveal a JUN–MIF program associated with secretory epithelial plasticity and CD74+ myeloid niches in colorectal cancer liver metastasis

Colorectal cancer liver metastasis (CRLM) is shaped by interactions between malignant epithelial plasticity and the immune microenvironment, yet the epithelial programs involved remain poorly defined. Single-cell RNA sequencing data from primary colorectal tumors and liver metastases were integrated to characterize epithelial and immune heterogeneity. Epithelial subclustering, pseudotime analysis, functional enrichment, CellChat-based communication inference, transcription factor screening and scTenifoldKnk virtual knockout analysis were performed to identify regulatory circuits. Spatial transcriptomics data from GSE225857 were used to assess the tissue-level organization of the JUN-MIF-CD74 axis. TCGA-COAD and GSE39582 cohorts, in vitro JUN knockdown and overexpression experiments, ChIP-qPCR, dual-luciferase reporter assays, ELISA, functional assays, immunohistochemistry of clinical CRLM specimens and in vivo MC38-based syngeneic subcutaneous tumor and intrasplenic liver metastasis models were used for validation. We identified a secretory epithelial state, Epi_4, characterized by SPINK4, TFF3, FCGBP and REG4 expression and enriched in oxygen-response and mitochondrial programs. JUN emerged as an Epi_4-associated regulator, with MIF identified as a major downstream candidate. Cell–cell communication analysis identified predicted MIF–CD74-related interactions between Epi_4 and myeloid populations, while exploratory spatial transcriptomic analysis suggested more prominent JUN–MIF–CD74 co-enrichment in the analyzed primary colorectal tumor sections than in liver metastatic sections. JUN bound the MIF promoter and increased MIF expression and secretion. JUN overexpression enhanced wound closure, cell viability and clonogenicity, whereas JUN knockdown produced opposite effects. In vivo, Jun knockdown suppressed syngeneic tumor growth and experimental liver metastasis and reduced Mif expression. Clinical analyses further supported a positive association between JUN and MIF. These findings support an Epi_4-associated JUN-dependent MIF transcriptional regulatory program and predicted MIF–CD74-related epithelial–myeloid interactions in colorectal cancer. Although JUN perturbation affected tumor-associated phenotypes and MIF expression, whether MIF functionally mediates these phenotypes remains unresolved, and the predicted MIF–CD74-related epithelial–myeloid interactions require further functional validation. The associated spatial signals were more prominent in primary colorectal tumors than in established liver metastases, suggesting a potential association with metastatic competence rather than liver-metastasis-specific activation.