CBF1 is clinically prognostic and serves as a target to block cellular invasion and chemoresistance of EMT-like glioblastoma cells.

BACKGROUND: Glioblastoma is the most common and most lethal primary brain cancer. CBF1 (also known as Recombination signal Binding Protein for immunoglobulin kappa J, RBPJ) is the cardinal transcriptional regulator of the Notch signalling network and has been shown to promote cancer stem-like cells (CSCs) in glioblastoma. Recent studies suggest that some of the malignant properties of CSCs are mediated through the activation of pro-invasive programme of epithelial-to-mesenchymal transition (EMT). Little is known whether CBF1 is involved in the EMT-like phenotype of glioma cells. METHODS: In a collection of GBM neurosphere lines, we genetically inhibited CBF1 and investigated the consequences on EMT-related properties, including in vitro invasiveness by Boyden chambers assay, chemoresistance using a clinical drug library screen and glycolytic metabolism assessing live-cell extracellular acidification rate. We also compared CBF1 expression in cells exposed to low and high oxygen tension. In silico analysis in large-scale Western and Eastern patient cohorts investigated the clinical prognostic value of CBF1 expression in low- and high-grade glioma as well as medulloblastoma. RESULTS: Mean CBF1 expression is significantly increased in isocitrate dehydrogenase 1 (IDH1) R132H mutant glioblastoma and serves as prognostic marker for prolonged overall survival in brain tumours, particularly after therapy with temozolomide. Hypoxic regions of glioblastoma have higher CBF1 activation and exposure to low oxygen can induce its expression in glioma cells in vitro. CBF1 inhibition blocks EMT activators such as zinc finger E-box-binding homeobox 1 (ZEB1) and significantly reduces cellular invasion and resistance to clinically approved anticancer drugs. Moreover, we indicate that CBF1 inhibition can impede cellular glycolysis. CONCLUSIONS: Mean CBF1 activation in bulk tumour samples serves as a clinical predictive biomarker in brain cancers but its intratumoral and intertumoral expression is highly heterogeneous. Microenvironmental changes such as hypoxia can stimulate the activation of CBF1 in glioblastoma. CBF1 blockade can suppress glioblastoma invasion in vitro in particular in cells undergone EMT such as those found in the hypoxic niche. Targeting CBF1 can be an effective anti-EMT therapy to impede invasive properties and chemosensitivity in those cells.

Longitudinal stability of molecular alterations and drug response profiles in tumor spheroid cell lines enables reproducible analyses.

The utility of patient-derived tumor cell lines as experimental models for glioblastoma has been challenged by limited representation of the in vivo tumor biology and low clinical translatability. Here, we report on longitudinal epigenetic and transcriptional profiling of seven glioblastoma spheroid cell line models cultured over an extended period. Molecular profiles were associated with drug response data obtained for 231 clinically used drugs. We show that the glioblastoma spheroid models remained molecularly stable and displayed reproducible drug responses over prolonged culture times of 30 in vitro passages. Integration of gene expression and drug response data identified predictive gene signatures linked to sensitivity to specific drugs, indicating the potential of gene expression-based prediction of glioblastoma therapy response. Our data thus empowers glioblastoma spheroid disease modeling as a useful preclinical assay that may uncover novel therapeutic vulnerabilities and associated molecular alterations.