Spatial multi-omics defines a shared glioblastoma infiltrative signature at the resection margin

Abstract

Glioblastoma (GBM) remains an untreatable disease. Understanding GBM's infiltrative biology at the resection margin is limited, despite causing disease recurrence and progression. To address this, we generated a high-throughput single-nucleus (sn)RNA-seq and snATAC-seq multi-omic dataset from six tumors with distinct genomic drivers and combined it with spatial transcriptomics to characterize the unique molecular phenotype of GBM near the margin. By contrasting GBM-specific biology in matching "Core" vs. "Margin" dissections, we define a unique, shared "GBM infiltration" signature, epigenetically co-opted at the margin. We prioritize EGFR as a top differentially expressed and accessible "Margin" marker across GBM subtypes, show its dynamic expression along a core-to-margin infiltration trajectory, and validate its role in migration through CRISPR/Cas9 deletion in two patient-derived models. ChIP-seq occupancy studies furthermore corroborate preferential TEAD1 binding at EGFR"s accessible regulatory elements. This validated multi-omic dataset enables further studies into tumor and microenvironment biology in the context of residual GBM disease.