Diffuse intrinsic high grade glioma (DIPG) is an almost universally fatal tumor of childhood characterized by epigenetic dysregulation driven by somatic H3.3/H3.1 K27M mutations observed in >80% of tumors. We conducted a chromatin-focused CRISPR screen and identified a novel strategy to inhibit the growth of patient-derived DIPG cells by co-targeting lysine specific demethylase 1 (LSD1) and histone deacetylases (HDACs). Consistent with the genetic data, we demonstrate that a bifunctional inhibitor of HDACs and LSD1, Corin, inhibits the growth of DIPG cells both in vitro and in orthotopic xenografts. Mechanistically, co-targeting LSD1 and HDACs with Corin synergistically alters the levels of histone modifications in DIPG, rescuing H3K27me3 levels suppressed by the dominant negative effects of K27M mutant histones and inducing simultaneous increases in both HDAC-targeted H3K27ac and LSD1-targeted H3K4me1 at thousands of genomic locations. Coincident with these chromatin changes, we observe robust transcriptional changes in DIPG cells, including the repression of cell cycle-related genes and the activation of neuronal differentiation genes. Consistently, phenotypic assays reveal that Corin reduces S-phase and Ki67+ proliferating cells, induces a neuronal-like morphology, alters the expression of stem and differentiation markers, and increases DIPG cell death. Finally, analyses of patient DIPG expression datasets indicate that Corin-dependent transcriptional signatures are overrepresented in normal brain compared to DIPG tumors and correlate with increased patient survival time. Together, these data reveal that co-inhibiting LSD1 and HDACs synergistically reduces DIPG growth by re-programming the chromatin landscape to activate a latent differentiation response, and suggest a strategy of co-inhibiting LSD1 and HDACs with Corin to treat DIPG.
Analysis of Chromatin Pathways as Regulators of High Grade Glioma Gene Expression Patterns
We plan to identify genes that are up- and down-regulated in brain tumors compared to normal tissue samples and genes that are up- and down-regulated in brain tumors with different mutations. We will compare these patient-derived gene expression profiles to data we have generated using human brai