ABSTRACT: Background: Single-cell RNA sequencing (scRNA-seq) contributes to discovering intrinsic heterogeneity of pancreatic adenosquamous carcinoma (PASC), a lethal human malignancy. However, the gene-pattern variations within ductal and stromal cells remain unclear. Methods: We fulfilled scRNA-seq profiling with one healthy donor pancreas, one intraductal papillary mucinous neoplasm (IPMN), and one PASC sample tissue. Multiple bioinformatics analyses were used to investigate the transcriptomics abnormalities from normal stage to tumor formation. Results: We identified ten subpopulations from 10,346 single cells, including myeloid, immune, ductal, fibroblast, acinar, stellate, endothelial, and malignant cells. Within our analysis, pancreatic cancer cells were further divided into five clusters; notably, cluster 1 exhibited stem-like phenotypes, such as UBE2C, ASPM, and TOP2A. We also found that S100A2 could be a specific biomarker for tumor cells, relating to PASC progression. LGALS1, NPM1, RACK1, and PERP were upregulated gradually from ductal to malignant cells. Furthermore, the level of copy number variation in ductal and cancer cells was significantly greater than reference cells. The expression of EREG, FCGR2A, FCGR2A, and CCL4L2 increased in myeloid cells along with tumor development, while cancer-associated fibroblast could induce the immunosuppressive tumor microenvironment, promoting pancreatic cancerogenesis. Finally, we demonstrated that EGFR-associated ligand-receptor pairs were activated in stromal to malignant cell communication. Conclusion: This study revealed the variations of ductal and stromal cells, defined cancer-associated signaling pathways, and deciphered cellular interactions following PASC progression. Thus, we provided valuable information to further understand the crucial regulatory mechanisms that embrace the potential therapeutic targets for PASC.
