P10.05.B RECAPITULATING PATIENT BIOLOGY IN EX VIVO MODELS: A SCRNASEQ ANALYSIS OF THE BLOOD-BRAIN-TUMOR BARRIER IN PEDIATRIC DIFFUSE MIDLINE GLIOMA

Abstract BACKGROUND Diffuse midline glioma (DMG) is a highly lethal brain tumor in children with limited treatment options. The blood-brain-barrier (BBB) remains a key challenge for therapeutics, preventing most drugs from entering brain tissue. Characterization of the blood-brain-tumor-barrier (BBTB) - the interaction between tumors and BBB vasculature - remains largely unexplored for DMG. We have established an ex vivo model of the DMG-BBTB, comprised of a co-culture of patient-derived DMG cells, primary human brain microvascular endothelial cells, primary human brain vascular pericytes, and primary human astrocytes, embedded in a fibrin matrix within a microfluidic chip. To probe the fidelity of our model and examine the impact of pediatric DMG on vasculature, we employed single cell RNA sequencing (scRNAseq). MATERIAL AND METHODS scRNAseq was performed on devices containing normal BBB and DMG-BBTB, as well as BBB devices dosed with tumor necrosis factor alpha (TNFa) as a positive inflammation control. Devices for each condition were seeded with cells and growth factors to form self-assembled microvascular networks within 7 days. On day 8, cells were extracted from the gel compartment to generate a single cell suspension, from which RNA was isolated to construct sequencing libraries. The prepared libraries were sequenced on the Illumina NovaSeq 6000 platform. After standard pre-processing and quality control, cells were clustered and annotated for downstream gene enrichment and pathway analyses as well projection onto reference datasets. RESULTS scRNAseq of our model revealed both expected and unexpected cell types, consistent with on-chip differentiation of a brain vascular model. We identified significant differential gene expression across all cell populations between BBB and BBTB conditions, with signatures distinct from TNFa-induced inflammation. When comparing our data to established scRNAseq atlases of vascular development, our model recapitulates key aspects of developmental biology such as heterogeneous vascular endothelial and perictye subtypes. For BBTB conditions, we classified the patient-derived DMG cells in our model according to published developmental hierarchies established from patient samples. CONCLUSION We developed a high fidelity ex vivo model to evaluate the microvasculature of the DMG-BBTB. Using scRNAseq, we demonstrate that our model recapitulates key features of the DMG tumor microenvironment. This representative model of the DMG-BBTB will provide an avenue for future preclinical testing of new therapies and exploring modalities for targeted drug delivery.