Epigenetic Landscapes Identify Functional Therapeutic Vulnerabilities in Glioblastoma

Glioblastoma is a highly lethal primary intrinsic brain tumor for which no targeted therapeutic agents are effective. Despite a decade of massive efforts to comprehensively profile the genomic architecture of glioblastomas, the current standard-of-care approaches rely on non-specific mechanisms, including surgical resection, ionizing radiation, and the alkylating agent temozolomide. However, even maximal therapy affords patients a median survival of around 15 months. While genomic alterations, including the presence of IDH mutations and MGMT promoter mutations inform patient prognosis and response to therapy respectively, this knowledge has not translated into targeted therapeutics that extend survival. Here, we interrogated the epigenetic landscapes of glioblastomas and identified novel tumor dependencies. First, we investigated the active enhancer profiles of primary glioblastoma surgical resection specimens and cell models to identify glioblastoma-stem-cell specific super-enhancers. ELOVL2 is one of these super-enhancer associated genes that acts to promote glioblastoma growth and survival by maintaining cell membrane composition and allowing for efficient EGFR signaling. These findings suggest that epigenetic upregulation of a metabolic enzyme is essential for powering signaling through a critical glioblastoma growth factor receptor. Second, we identified a novel form of DNA methylation (N6-mA) in glioblastoma that occurs on adenine bases. N6-mA is highly enriched in glioblastoma tissues compared to normal astrocytes and marks silenced heterochromatin regions throughout the genome. The DNA demethylase, ALKBH1, shapes the genomic N6-mA landscape and contributes to the functional control of a number of oncogenic gene pathways that are critical for glioma stem cell survival, especially the hypoxia response pathway. These findings reveal a new class of DNA modification in human disease, describe how glioblastomas co-opt early embryogenesis epigenetic mechanisms for growth advantage, and identify a potential novel therapeutic vulnerability. Finally, we developed a digital-mirror-device-based three-dimensional bioprinting technique to construct physiologically and clinically relevant biomimetic glioblastoma tissue environments to improve modeling of cellular interactions, mechanical properties, physiologic gradients, and transcriptional states found in the actual disease in a tractable controlled model. Together, these findings advance our understanding of epigenetic regulation in glioblastoma, provide new therapeutic opportunities, and deliver advanced platforms to improve modeling of this lethal disease.