Ex vivo dynamics of human glioblastoma cells in a microvasculature-on-a-chip system correlates with tumor heterogeneity and subtypes

Abstract

The perivascular niche (PVN) plays an essential role in brain tumor stem-like cell (BTSC) fate control, tumor invasion, and therapeutic resistance. Herein we report on the use of a microvasculature-on-a-chip system as a PVN model to evaluate the dynamics of BTSCs ex vivo from 10 glioblastoma patients. We observed that BTSCs preferentially localize in the perivascular zone. Live cell tracking revealed that the cells residing in the vicinity of microvessels had the lowest motility, while a fraction of cells on the microvessels unexpectedly possessed the highest motility and migrated over the longest distance. These results indicate that the perivascular zone is a niche for BTSCs, while the microvascular tracks are also a path for long-distance tumor cell migration and invasion. Additionally, the degree of co-localization between tumor cells and microvessels varied significantly across patients. To validate the results from our microvasculature-on-a-chip system, we used single-cell transcriptome sequencing (10 patients and 21,750 single cells in total) to identify the subtype of each tumor cell. The co-localization coefficient was found to correlate positively with proneural (stem-like) or mesenchymal (invasive) but not classical (proliferative) tumor cells. Furthermore, we found that a gene signature profile including PDGFRA correlated strongly with the \"homing\" of brain tumor cells to the PVN. Our findings demonstrated that ex vivo dynamics of human brain tumor cells in a microvasculature-on-a-chip model can recapitulate in vivo tumor cell dynamics, heterogeneity, and subtypes, representing a new route to the study of human tumor cell biology and uncover patient-specific tumor cell functions.