Abstract
As the majority of diffuse intrinsic pontine glioma (DIPG) have H3K27M mutations, epigenetic-targeting agents have been studied, though evaluations have been limited by their model systems, untranslatable drug concentrations, and/or evasive mechanisms of action. To develop a more translational model, we used biopsy samples from newly diagnosed DIPG patients to create treatment-naïve in vitro and in vivo models (molecular aberrations in parentheses), including PBT-09FH (H3FA3, PI3KCA), PBT-22FH (H3F3A, TP53), PBT-24FH (PMS2), and PBT-27FH (HIST1H3B, TP53, NTRK2). Models demonstrated radiation-resistance similar to the patient from whom the culture was generated, supporting the models’ relevance (e.g. cell viability after 8 Gy was 36%, 81%, 71%, and 61% in PBT-09FH, -22FH, -24FH, and -27FH, respectively, compared to 7% in the medulloblastoma model MED-411FH). We evaluated cell viability and apoptosis following treatment with a panel of HDAC inhibitors, identifying the low nanomolar IC50 of quisinostat (~50 nM) and romidepsin (~5 nM). While RNA expression changes induced by 100 nM panobinostat and quisinostat included shared overexpression of the top 20/25 genes (e.g. FSTL5, ITIH5) and shared downregulation of the top 22/25 (e.g. GPR37L1, HEPACAM), only 9/25 were downregulated by panobinostat, quisinostat, and romidepsin (e.g. C21orf62, IFIT2), identifying these as potential vulnerabilities or biomarkers of lethal HDAC inhibition. Mass-spectrometry (LC-MS) demonstrated panobinostat as the greatest acetylator of cortactin, potentially related to thrombocytopenia. While PBT-09 flank models demonstrated quisinostat’s on-target acetylation and efficacy, orthotopic xenograft models did not, supporting our model’s intact blood-brain barrier and emphasizing the need for CNS penetrant versions of potentially efficacious agents.