Stem cell morphology defines functional heterogeneity and therapeutic vulnerabilities in glioblastoma

Glioblastoma (GBM) is an aggressive brain tumor and an unmet clinical need due to its invasiveness and therapy resistance. These features are driven by glioblastoma stem cells (GSCs), which exhibit remarkable functional heterogeneity. However, GSC transcriptional profiling alone cannot predict clinically relevant behaviors. Here, we developed CellShape-seq, a spatial transcriptomics platform that integrates cell morphology with transcriptome. This identified three GSC morphoclasses corresponding to distinct transcriptomic states and functions: (1) nonpolar cells show differentiation and therapy sensitivity, (2) elongated cells are invasive, and (3) multipolar cells form intercellular networks. Importantly, chemoresistance is morphoclass-specific: elongated GSCs depend on YAP/TEAD1 signaling, while multipolar GSCs rely on gap junction-mediated networks. Targeting these vulnerabilities with specific inhibitors sensitized resistant GSC morphoclasses to temozolomide (TMZ) in patient-derived organoids. Our findings demonstrate that morphology provides critical insights into GSC behavior and establish a rationale for morphology-informed therapies to overcome resistance and improve outcomes in GBM.