Single-cell transcriptome sequencing reveals new epithelial-stromal associated mesenchymal-like subsets in recurrent gliomas

Gliomas, particularly glioblastomas, are highly malignant brain tumors with high recurrence rates and poor prognosis. Despite advances in treatment, recurrence remains a major challenge. Epithelial-mesenchymal transition (EMT) plays a key role in tumor invasion and recurrence. This study explores the transcriptional and regulatory mechanisms driving glioma recurrence, focusing on mesenchymal-like (MES-like) subpopulations. Single-nucleus RNA sequencing was performed on 52 IDH wild-type GBM specimens, including 26 primary and 26 recurrent tumors. Spatial transcriptomics data were also incorporated. Tumor subpopulations were identified through gene regulatory network analysis, copy number variation detection, and nonnegative matrix factorization. Functional validation was conducted using gene knockdown experiments, followed by xenograft studies. We discovered novel MES-like subpopulations in recurrent GBM enriched with EMT-related genes like EGR1 and SERPINE1. These subpopulations exhibited increased transcriptional activity and were associated with poor prognosis and invasiveness. Knockdown of SERPINE1 significantly reduced cell proliferation and migration. Spatial transcriptomics showed MES-like cells concentrated at the tumor margins, highlighting their role in invasion and recurrence. MES-like subpopulations, driven by EGR1 and SERPINE1, are critical in GBM. Targeting these regulators could offer new therapeutic strategies to reduce glioma recurrence and improve outcomes.