Single-Cell Transcriptome Analysis Reveals IRF1-Driven Epithelial States and Glycosaminoglycan–Glycolysis Coupling in Cisplatin-Resistant HGSOC
Zhaoyang Jia, Wenjing Pan, Xibo Zhao, Lianxiu Li, Wenhua TanObjectives
Cisplatin resistance is the principal cause of relapse in high-grade serous ovarian cancer (HGSOC), but bulk-expression signatures cannot localize resistant malignant states or the tumor–microenvironment (TME) interactions that sustain them. This study aimed to define cisplatin-resistant epithelial cell states and their regulatory and metabolic circuits by integrating multi-cohort single-cell transcriptomes with pharmacogenomic and clinical data.
Methods
We assembled 159,419 cells from 32 HGSOC tumors across six public single-cell RNA-sequencing cohorts and performed harmonized integration, clustering, and lineage annotation. Cisplatin response was mapped to single cells by coupling scRNA-seq data to Genomics of Drug Sensitivity in Cancer predicted cisplatin response score values using Scissor, with AUCell-based validation. We then applied receptor–ligand–based cell–cell communication analysis (CellChat), transcription-factor (TF) activity inference (SCENIC and NetAct with TRRUST), pseudotime trajectory reconstruction (Monocle3), and pathway-level metabolic scoring. Associations with drug sensitivity and patient outcome were evaluated in ovarian cancer cell lines and The Cancer Genome Atlas (TCGA) HGSOC cohort.
Results
Among 34 epithelial subclusters, 14 were significantly enriched for a cisplatin-resistant phenotype and collectively accounted for most predicted resistant cells. These states were transcriptionally characterized by stress, interferon, and apoptotic programs and formed dense communication hubs with endothelial cells, fibroblasts, myeloid cells, and T/NK cells via extracellular-matrix and adhesion pathways (for example, LAMA3–CD44, COL6A1/2–CD44, NECTIN3–NECTIN2, and CD99–CD99 interactions). TF-activity modeling converged on an IRF1-centered regulatory program that increased along an epithelial trajectory and coordinated inflammatory and apoptotic gene expression. Metabolically, resistance-enriched epithelial states showed selective up-regulation of glycosaminoglycan biosynthesis, particularly keratan sulfate, coupled to heightened glycolysis; glycolytic activity correlated with predicted cisplatin response score in cell lines and an IRF1/STAT1 readout (GBP3) stratified survival in TCGA HGSOC.
Conclusion
Cisplatin resistance in HGSOC is encoded in discrete IRF1-driven epithelial states that are supported by specific TME communication networks and a glycosaminoglycan–glycolysis metabolic axis. This integrative single-cell informatics framework yields testable biomarkers and therapeutic targets for overcoming platinum resistance in ovarian cancer.