A 3D topographical model of parenchymal infiltration and perivascular invasion in glioblastoma

Glioblastoma (GBM) is the most common and invasive primary brain cancer. GBM tumors are characterized by diffuse infiltration, with tumor cells invading slowly through the hyaluronic acid (HA)-rich parenchyma toward vascular beds and then migrating rapidly along microvasculature. Progress in understanding local infiltration, vascular homing, and perivascular invasion is limited by the absence of culture models that recapitulate these hallmark processes. Here, we introduce a platform for GBM invasion consisting of a tumor-like cell reservoir and a parallel open channel "vessel" embedded in the 3D HA-RGD matrix. We show that this simple paradigm is sufficient to capture multi-step invasion and transitions in cell morphology and speed reminiscent of those seen in GBM. Specifically, seeded tumor cells grow into multicellular masses that expand and invade the surrounding HA-RGD matrices while extending long (10-100 μm), thin protrusions resembling those observed for GBM in vivo. Upon encountering the channel, cells orient along the channel wall, adopt a 2D-like morphology, and migrate rapidly along the channel. Structured illumination microscopy reveals distinct cytoskeletal architectures for cells invading through the HA matrix versus those migrating along the vascular channel. Substitution of collagen I in place of HA-RGD supports the same sequence of events but with faster local invasion and a more mesenchymal morphology. These results indicate that topographical effects are generalizable across matrix formulations, but the mechanisms underlying invasion are matrix-dependent. We anticipate that our reductionist paradigm should speed the development of mechanistic hypotheses that could be tested in more complex tumor models.