OTHR-04. Single-nucleus transcriptomic atlas of human hindbrain development identifies cellular origins of pediatric brainstem tumors

Abstract The hindbrain is a key central nervous system structure involved in communication between the brain and the spinal cord, integrating sensory information and coordinating motor responses. Dysregulation in hindbrain development is frequently associated with behavioral disorders and maybe associated tumor formation, stressing the importance of understanding the intricacies of human hindbrain development. Here we performed RNA-sequencing of over 300,000 single nuclei to construct a developing human hindbrain atlas, focusing on pons and medulla, ranging from embryonic to adult stages, profiling the entirety of the hindbrain’s neural and glial populations from progenitors to mature cell types. We describe the transcriptional identity of neuronal cells constituting the pontine and medullary nerve centers including precerebellar, respiratory, auditory and vestibular nuclei; and their differentiation trajectory from the progenitor domains in the embryonic rhombencephalon. To identify the origins of pediatric brain tumors, we mapped ~2,900 bulk and ~50 single-cell transcriptome tumor datasets to the developmental hindbrain atlas, focusing on medulloblastomas, and pediatric low- and high-grade gliomas, and determined cell populations and differentiation states associated with specific tumor groups. We confirmed migrating lower rhombic lip derivatives as the lineage of origin for WNT medulloblastoma. Further, we ascertained early ventral neural progenitor-derived oligodendrocyte lineage as the source of diffuse midline glioma H3K27-altered and postnatal oligodendrocyte progenitors as the origin of pilocytic astrocytoma. In addition, the ependymal lineage originating from glial progenitors was identified as the best match for ependymomas. We show that tumor development has a component that transcriptomically follows its cellular lineage of origin, and identify genes with preserved expression between the respective normal and malignant cell populations. Tumor to normal comparisons also identified tumor-specific genes, providing promising therapeutic candidates. We plan to make our data and analysis pipeline publicly available via user-friendly graphical interface at brain-match.org for unrestricted use of the generated results.