Comprehensive analysis of glycolysis and glycolysis-related mRNA signature in survival and immune response in bladder cancer

Objective Glycolysis provides the metabolic requirements for proliferation of bladder cancer (BCa). Biosynthesis and energy demand of immune cells in the tumor microenvironment (TME) rely on the reprogramming of glucose metabolism. However, how glycolysis functions in the TME and responds to immunotherapy in BCa remains unclear. Methods Non-negative Matrix Factorization was used to analyze 58 513 single cells from single-cell RNA-seq data of eight BCa patients and various TME clusters in different cell subtypes were identified. A machine learning-based procedure was then applied to develop a glycolysis-related mRNA signature (GRmRS), which was compared with a public database. In vitro and in vivo experiments were conducted to explore the potential function of glycolysis-related gene COL5A1 . Results A Non-negative Matrix Factorization algorithm was applied to identify different subclusters according to unique marker genes. A further analysis revealed that these glycolysis-related TME clusters were significantly associated with biological characteristics and prognostic outcomes of BCa. GRmRS was constructed based on the top 18 prospective glycolysis-related genes screened by 10 machine learning algorithms and 101 algorithm combinations, which predicted oncological survival and immune response of molecular subtypes in BCa. The meta-analysis showed GRmRS performed better than other 133 prognostic signatures. In vitro and in vivo experiments demonstrated that COL5A1 in cancer-associated fibroblasts promoted BCa proliferation and reduced infiltration of CD8 + cells. Conclusions We found that glycolysis in the TME promotes the connection to cancer cells and improves tumor immune evasion. GRmRS is a promising tool for survival prediction and individualized management of BCa patients. Meanwhile, COL5A1 in cancer-associated fibroblasts promotes BCa development and immune escape.