Integrative single-cell and bulk transcriptomics reveal cuproptosis subtypes, prognostic models, and COL5A1 as a key gene in glioblastoma.

Glioblastoma (GBM) remains one of the most aggressive primary brain tumors with limited therapeutic options. Cuproptosis, a recently identified copper-dependent form of regulated cell death, has emerged as a potential therapeutic vulnerability in cancer. Here, we systematically investigated the molecular and clinical significance of cuproptosis-related genes (CRGs) in GBM using integrated bulk and single-cell transcriptomic analyses. Based on the expression profiles of 31 differentially expressed CRGs, GBM patients were classified into two molecular subtypes with distinct prognostic and immune characteristics. Cluster 2 exhibited poorer survival and increased immune and stromal infiltration. To further assess the prognostic value of CRGs, we constructed a risk model using 110 subtype-associated genes and systematically compared 101 machine-learning algorithms. The SuperPC model achieved the best predictive performance and was successfully validated in two independent CGGA cohorts. The resulting risk score was closely associated with tumor microenvironment infiltration. Integrated single-cell analysis and experimental validation identified COL5A1 as a key GBM-associated gene. COL5A1 was specifically enriched in tumor cells across independent scRNA-seq datasets and positively correlated with the cuproptosis-related genes LOXL2 and SPARC. Functional assays demonstrated that COL5A1 knockdown suppressed GBM cell migration and invasion, reduced LOXL2 and SPARC expression, and enhanced sensitivity to cuproptosis induction. Collectively, our study establishes a robust cuproptosis-associated prognostic framework for GBM and identifies COL5A1 as a potential regulator linking extracellular matrix remodeling, malignant progression, and cuproptosis-related vulnerability, providing new insights into GBM biology and potential therapeutic targets.