Tumor edge architecture in glioblastoma is constructed by inter-cellular signals from vascular endothelial cells

2020 
Glioblastoma is a lethal brain cancer with active infiltration of tumor cells at the periphery where the vascular supply is enriched to create its own niche. Given that tumor recurrence is predominantly local, most of the seeds for lethal recurrence are hidden at the tumor edge and are surgically inaccessible. Here, we found that the spatial identity of the edge-located glioblastoma cells is, in the tested models, constructed by inter-cellular signals derived from the soluble factor endocan (protein product of Esm1) that is secreted by tumor-associated vascular endothelial (VE) cells. Injection of two distinct mouse glioblastoma models into Esm1 knockout (KO) mice resulted in tumors that failed to form typical edge lesions, resulting in the formation of compact core-only lesions. In sharp contrast, tumors derived in WT (Esm1-intact) mice harbored both tumor core and edge structures. Despite these obvious phenotypic differences, the aggressiveness of these two tumor types was indistinguishable in vivo. Surprisingly, regardless of the host system, co-injection of mixture of these two tumor subpopulations gave rise tumors with much worse survival, indicating that the accumulating tumor cell heterogeneity contributes to elevate malignancy. Mechanistically, endocan competes with PDGF to bind and activate PDGFRα in human glioblastoma cells, accompanied with the upregulation of the Myc transcriptional activity via alterations in its promoter chromatin structure. One of the mainstays of glioblastoma treatment, radiation, induces endocan secretion from VE cells, which promotes the radioresistance of the edge-located glioblastoma cells, allowing for the persistence of the edge structure. Collectively, these data suggest that intra-tumoral spatial heterogeneity is initiated by the VE cell-derived endocan signals through PDGFRα to construct tumor edge-to-core (E-to-C) architecture to cause lethal tumor recurrence.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    47
    References
    3
    Citations
    NaN
    KQI
    []