Integrating Single-Cell and Bulk RNA Sequencing Data to Explore Sphingolipid Metabolism Molecular Signatures in...

Background: Ovarian cancer (OC) is the deadliest malignant tumor in the female reproductive system. Sphingolipid metabolism (SM) is crucial for cellular function and has been linked to OC progression. Dysregulation of sphingolipid pathways contributes to tumor growth, chemoresistance, and metastasis in OC. Currently, investigations into the relationship between sphingolipid-related genes (SRGs) and OC prognosis in their initial stages. Our study aimed to develop a novel molecular subtyping based on SRGs and construct a signature to predict the prognosis of patients with OC, immune cell infiltration characteristics, and chemotherapy sensitivity. Methods: Bulk and single-cell RNA-sequencing data of OC was analyzed primarily from the TCGA and GEO databases. The gene set related to the sphingolipid pathway (hsa00600) was selected from the SM pathway, and the enrichment of SRGs was analyzed in the annotated single-cell sequencing data. The Scanpy function was used to score the gene features of each cell and further identify differentially expressed genes. By intersecting with the genes most closely related to SM activity identified through Weighted Gene Co-expression Network Analysis (WGCNA) based on bulk RNA sequencing data, and after performing univariate COX, multivariate COX and LASSO regression, three SRGs were identified. Subsequently, the SRGs-related prognostic signature was constructed. The analysis was further extended to clinical feature correlation, GSEA, tumor microenvironment (TME) analysis and chemotherapy sensitivity analysis. Finally, the expression and function of the key gene GBP5 in the model were validated through in vitro experiments. Results: Compared to other sites, SRG scores were highest in ascites, and among different cell types, SRG scores were highest in T cells. By integrating scRNA-seq and bulk RNA-seq analysis, three SRGs (C5AR1, GBP5, and MARCHF3) were ultimately selected to develop a prognostic model for SRGs. In this model, patients with higher risk scores had shorter overall survival, which was validated in the testing cohort. Immune infiltration analysis revealed that the risk score was negatively correlated with the abundance of CD8+ T cell infiltration and positively correlated with the abundance of M2 macrophage infiltration. Chemotherapy sensitivity analysis showed that the high-risk group exhibited increased resistance to Oxaliplatin, Gemcitabine, and Sorafenib. In vitro, we demonstrated that knockdown of the protective gene GBP5 in HEYA8 and SKOV3 cells enhanced cell viability, proliferation, and invasiveness, reduced apoptosis, and increased IC50 values for chemotherapy drugs. Conclusion: Our model effectively identifies high-risk patients and provides a reference for prognosis prediction using SRG signature. Moreover, hub gene GBP5 acts as a tumor inhibitory factor and regulates the chemosensitivity of oxaliplatin, gemcitabine, and sorafenib in OC.