Identification of vascular cues contributing to cancer cell stemness and function.

Glioblastoma stem cells (GSCs) reside close to blood vessels (BVs) but vascular cues contributing to GSC stemness and the nature of GSC-BVs cross talk are not fully understood. Here, we dissected vascular cues influencing GSC gene expression and function to perfusion-based vascular cues, as well as to those requiring direct GSC-endothelial cell (EC) contacts. In light of our previous finding that perivascular tumor cells are metabolically different from tumor cells residing further downstream, cancer cells residing within a narrow, < 60 µm wide perivascular niche were isolated and confirmed to possess a superior tumor-initiation potential compared with those residing further downstream. To circumvent reliance on marker expression, perivascular GSCs were isolated from the respective locales based on their relative state of quiescence. Combined use of these procedures uncovered a large number of previously unrecognized differentially expressed GSC genes. We show that the unique metabolic milieu of the perivascular niche dominated by the highly restricted zone of mTOR activity is conducive for acquisition of GSC properties, primarily in the regulation of genes implicated in cell cycle control. A complementary role of vascular cues including those requiring direct glioma/EC contacts was revealed using glioma/EC co-cultures. Outstanding in the group of glioma cells impacted by nearby ECs were multiple genes responsible for maintaining GSCs in an undifferentiated state, a large fraction of which also relied on Notch-mediated signaling. Glioma-EC communication was found to be bidirectional, evidenced by extensive Notch-mediated EC reprogramming by contacting tumor cells, primarily metabolic EC reprogramming.

Isolation of Tumor Cells Based on Their Distance from Blood Vessels.

Differential exposure of tumor cells to microenvironmental cues greatly impacts cell phenotypes, raising a need for position based sorting of tumor cells amenable to multiple OMICs and functional analyses. One such key determinant of tumor heterogeneity in solid tumors is its vasculature. Proximity to blood vessels (BVs) profoundly affects tumor cell phenotypes due to differential availability of oxygen, gradient exposure to blood-borne substances and inputs by angiocrine factors. To unravel the whole spectrum of genes, pathways and phenotypes impacted by BVs and to determine spatial domains of vascular influences, we developed a methodology for sorting tumor cells according to their relative distance from BVs. The procedure exemplified here using glioblastoma (GBM) model is based on differential uptake of intra-venously injected, freely-diffusing fluorescent dye that allows separation of stroma-free tumor cells residing in different, successive microenvironments amenable for subsequent OMICs and functional analyses. This reliable, easy to use, cost effective strategy can be extended to all solid tumors to study the impact of vasculature or the lack of it.

Intra-Tumoral Metabolic Zonation and Resultant Phenotypic Diversification Are Dictated by Blood Vessel Proximity.

Differential exposure of tumor cells to blood-borne and angiocrine factors results in diverse metabolic microenvironments conducive for non-genetic tumor cell diversification. Here, we harnessed a methodology for retrospective sorting of fully functional, stroma-free cancer cells solely on the basis of their relative distance from blood vessels (BVs) to unveil the whole spectrum of genes, metabolites, and biological traits impacted by BV proximity. In both grafted mouse tumors and natural human glioblastoma (GBM), mTOR activity was confined to few cell layers from the nearest perfused vessel. Cancer cells within this perivascular tier are distinguished by intense anabolic metabolism and defy the Warburg principle through exercising extensive oxidative phosphorylation. Functional traits acquired by perivascular cancer cells, namely, enhanced tumorigenicity, superior migratory or invasive capabilities, and, unexpectedly, exceptional chemo- and radioresistance, are all mTOR dependent. Taken together, the study revealed a previously unappreciated graded metabolic zonation directly impacting the acquisition of multiple aggressive tumor traits.