SINGLE-NUCLEI TRANSCRIPTOMICS RELATES GLIOBLASTOMA INFILTRATION TO DISTINCT GLIAL PROGENITOR STATES

Abstract Our understanding of glioblastoma (GBM) intratumoral heterogeneity, particularly in the context of neurodevelopment, has thus far been primarily focused on the more surgically accessible tumor core niche. In contrast, the biology of GBM cells at the infiltrative edge, which evade surgical resection and drive tumor recurrence, remains poorly characterized. To this end, we microdissected and performed single-nuclei RNA sequencing (snRNA-seq) on approximately 62,000 nuclei taken from the tumor core and from the infiltrative edge of six GBM tumors with diverse genomic drivers, including IDH1, EGFR, PDGFRA, FGFR3, and NF1. Unbiased clustering reveals distinct neoplastic and non-neoplastic populations, further distinguished using copy number variation analysis. After projecting previously defined signatures taken from snRNA-seq analysis of human adult neocortex/subventricular zone and prenatal germinal matrix, we find that approximately 90% of tumor cells recapitulate a neurodevelopment-like molecular phenotype, reprising gene expression signatures of prenatal astrocytes and of a distinct glial intermediate progenitor cell population (g-IPC) that precedes both astrocyte and oligodendrocyte lineage differentiation. Examining the infiltrative edge of samples with the most confident microdissection (n=4), we see that while distinct populations of tumor cells in this niche express proneural and classical signatures, these cells are overall enriched for a g-IPC-like phenotype, relative to the tumor core, irrespective of the tumors’ genomic alterations. A subset of cells at the infiltrative edge, in particular, recapitulates the signature of an uncommitted g-IPC subtype, expressing both astroglial and oligodendroglial markers. Trajectory analyses also reveal distinct branches of core and edge tumor cells, which are predominantly astrocyte- and g-IPC-like, respectively. Differential gene expression analysis of GBM cells at the infiltrative edge vs. tumor core reveals a migration signature, dominated by EGFR, ERBB4, PCDH9, and PCDH15. Ultimately, this high resolution analysis of heterogeneity at the infiltrative edge allows us to uncover potentially targetable drivers of invasion in GBM.