STAT3/TGFBI signaling promotes the temozolomide resistance of glioblastoma through upregulating glycolysis by inducing cellular senescence.

Glioblastoma (GBM) is the most lethal type of brain tumor. Recent studies have indicated that cellular senescence-targeted therapy is a promising approach for cancer treatment. However, the underlying mechanisms remain to be clarified. In this study, 101 unique combinations of 10 machine learning algorithms were used to construct prognostic models based on cellular senescence-related genes (CSRGs). We developed the CSRG signature (CSRGS) using machine learning models that exhibited optimal performance. GBM samples were stratified into high- and low-CSRGS groups based on CSRGS scores. Patients in the high-CSRGS group exhibited a worse prognosis, higher immune infiltration, and increased sensitivity to immune checkpoint blockade therapy. Furthermore, senescence-related pathways were significantly correlated with glycolysis, indicating upregulated glycolytic metabolism in senescent GBM cells. We identified TGFBI as a key regulator that played vital roles in both glycolysis and cellular senescence in GBM. TGFBI was overexpressed in GBM samples compared to normal brain tissues, and its knockdown via shRNA inhibited cellular senescence, glycolysis, and temozolomide resistance. Chromatin immunoprecipitation (ChIP) and luciferase reporter assays confirmed that TGFBI is a direct STAT3 target and is required for the STAT3-induced promotion of cellular senescence, glycolysis, and drug resistance. The STAT3-TGFBI axis could be a potential target for senescence-targeted GBM therapy.