Single Cell Spatial Profiling Identifies Region-Specific Extracellular Matrix Adhesion and Signaling Networks in Glioblastoma

The human brain contains a rich milieu of extracellular matrix (ECM) components that are often dysregulated in pathologies including the malignant cancer glioblastoma (GBM). Here, we have used in situ single-cell spatial transcriptomic platforms to map the expression patterns of nearly 400 ECM genes in normal brain and GBM samples. Our analysis identifies at least four different GBM cell populations with unique ECM expression profiles that show spatial enrichment in distinct intratumor regions. Spatial mapping also demonstrates largely non-overlapping expression signatures of various ECM components in GBM stromal cell types, particularly in vascular endothelial cells and reactive microglia/macrophages. Comparisons of GBM (IDH1 wild type) versus lower-grade II and III astrocytoma samples (IDH1 R132H) identifies differential expression of key ECM components, including elevated levels of select ECM glycoproteins (IGFBP2 and MGP) and ECM-affiliated proteins (ANXA1 and ANXA2). In addition, we detect spatially enriched expression of COL8A1 (collagen), LUM (proteoglycan), and POSTN (ECM glycoprotein) in perivascular stromal cells in GBM but not in lower grade tumors. Computational analysis of putative ligand-receptor interactions reveals novel ECM communication networks between cancer cells and stromal components, particularly in regions of GBM microvascular proliferation and pseudopalisading necrosis. In summary, this comprehensive spatial map provides new insights into microenvironmental control of GBM initiation and progression and identifies potential therapeutic targets in the ECM.