Multi-omics analysis of arginine metabolism in ovarian cancer: A prognostic signature and GTF2F2-driven stromal remodeling.

BACKGROUND: Arginine metabolism shapes tumor growth, stromal activation, and antitumor immunity, yet its translational relevance in ovarian cancer (OV) remains incompletely defined. We sought to derive an arginine-metabolism-related signature (AMRS) for prognosis and therapy guidance and to nominate actionable regulators. MATERIALS AND METHODS: We analysed single-cell and harmonized bulk RNA sequencing datasets. Arginine-metabolism genes curated (77 candidates; 68 detected). scRNA-seq data was processed with Seurat, DoubletFinder and Harmony, and arginine-metabolism activity was scored with AddModuleScore and GSVA. Arginine-related subtypes were defined using unsupervised consensus clustering. The AMRS was constructed by combining ten machine learning algorithms as benchmarks. Random survival forest (RSF) was selected and explained by SHapley Additive exPlanations (SHAP). Immune contexture by ssGSEA, IOBR and ESTIMATE. Cell communication by CellChat. ICI-response surrogates by IPS and TIDE. The oncoPredict algorithm was utilized for prediction of drug sensitivity. Spatial deconvolution mapped gene-compartment associations. Functional validation included siRNA suppression of GTF2F2 in A2780 and HEY cells, qPCR, immunoblotting, CCK-8, colony formation, migration, Annexin V/PI flow cytometry, arginine deprivation, and dual-luciferase reporter assays. RESULTS: The AMRS distinguished the high risk group from the low risk group with a significant difference in survival. A nomogram with AMRS risk score, grade, stage and age was concordant and clinically useful. Active immune checkpoint and co-stimulatory signaling and higher infiltration of T, NK and dendritic cells characterized low-risk tumors. However, high-risk tumors have immune-cold stromal and myeloid characteristics. High-risk tumors were biologically more aggressive, but had lower mutational burden and polygenic co-occurrence. Drug-response prediction indicated that AMRS-high tumors will respond better than AMRS-low tumors to IGF-1R and PI3K inhibitors and taxanes, but not to mitochondrial complex I, S6K, survivin and TAF1-directed therapies. SHAP identified GTF2F2 as a top AMRS driver and spatial mapping identified hotspots colocalized with fibroblast and endothelial-rich areas. Knockdown of GTF2F2 inhibited proliferation, clonogenicity, migration and apoptosis through downregulation of ASS1, ASL, ARG2 and NOS2. GTF2F2 deletion also promoted growth-inhibitory and pro-apoptotic effects of arginine deprivation and increased susceptibility to stress. Dual-luciferase reporter assays showed that GTF2F2 knockdown decreased ASS1 and ARG2 promoter activity, indicating promoter regulation of arginine-metabolism gene transcription. CONCLUSION: The AMRS categorizes OV into immune-inflamed and stroma-dominant states with different therapeutic liabilities. GTF2F2 is identified as a tractable regulator of transcriptional control, arginine metabolism, stromal activation, and arginine-deprivation vulnerability, further supporting metabolism-informed therapeutic combinations and prospective validation.