Extracellular Vesicles from High-Grade Glioma Exchange Diverse Pro-oncogenic Signals That Maintain Intratumoral Heterogeneity

Abstract

A lack of experimental models of tumor heterogeneity limits our knowledge of the complex subpopulation dynamics within the tumor ecosystem. In high-grade gliomas (HGG), distinct hierarchical cell populations arise from different glioma stem-like cell (GSC) subpopulations. Extracellular vesicles (EV) shed by cells may serve as conduits of genetic and signaling communications; however, little is known about how HGG heterogeneity may impact EV content and activity. In this study, we performed a proteomic analysis of EVs isolated from patient-derived GSC of either proneural or mesenchymal subtypes. EV signatures were heterogeneous, but reflected the molecular make-up of the GSC and consistently clustered into the two subtypes. EV-borne protein cargos transferred between proneural and mesenchymal GSC increased protumorigenic behaviors in vitro and in vivo Clinically, analyses of HGG patient data from the The Cancer Genome Atlas database revealed that proneural tumors with mesenchymal EV signatures or mesenchymal tumors with proneural EV signatures were both associated with worse outcomes, suggesting influences by the proportion of tumor cells of varying subtypes in tumors. Collectively, our findings illuminate the heterogeneity among tumor EVs and the complexity of HGG heterogeneity, which these EVs help to maintain. Cancer Res; 76(10); 2876-81. ©2016 AACR.

Figure 1

Heterogeneity of GSCs is mirrored by the diversity of EV protein composition. A. Workflow depicting isolation of GSCs from primary tumors for GSC culture and proteomic analysis of EVs (left). Representative micrographs of GSCs (middle) and GSC EVs (right). B. The GSC EV proteome profile distinguishes P (green) from M (red) GSC subtypes. Proteins sets that vary coherently between subtypes were identified by clustering. C. The GSC EV proteome profile separates P and M glioblastoma subtypes. Genes coding for proteins sets that vary coherently between GSC subtypes (classical – C, blue; mesenchymal – M, red; proneural – P, green; neural – N, magenta) from 89- proteins signature were retrieved from TCGA GBM dataset and identified by clustering: power of prediction of top 6 genes is shown. D. The GSC EV proteome content partially mimic cellular expression. Selected protein sets were validated by Western blotting using indicated antibodies, * indicate specific band on FASN blot.

Figure 2

Differential in vitro effects of GSC co-culture and exposure to EVs. A. Workflow depicting labelling and co-treatment of GSCs and GSC EVs (top). Representative micrographs of GSC spheroids releasing labelled EVs (bottom). Arrows indicate EVs shed by P and M GSCs. Scale bar 50μm. B. EVM promote growth of P GSCs. Representative micrographs of GSC spheroids co-cultured with EVs (left) and quantification of spheroid frequency (right). * P < 0.05. Scale bar 100μm. C. Workflow depicting co-culture of labelled GSCs (left). Representative micrographs of GSC spheroids and transfer of EVs between GSCs in the spheroid (right). Arrows indicates EV internalization. Scale bars 100μm (left) and 25μm (right). D. M GSCs promote growth of P GSCs. Quantification of cell growth in mono- and co-culture. * P < 0.05.

Figure 3

Heterogeneous GSCs exchange EVs in vivo and promote tumorigenicity. A. Workflow depicting in vivo model. B. EVM promote proliferation of P GSCs in vivo. Representative micrographs of P GSCs 4 days after co-implantation with EVM with Ki-67 immunostaining (upper left), higher magnification insets (bottom) and Ki-67 quantification (upper right). * P < 0.05. Scale bars 200μm (upper) and 50μm (bottom). C. EVs are transferred intra-tumorally in vivo. Representative micrographs of co-implanted tumors (top row). Distinct distribution of P and M GSCs (second row, arrow indicates infiltrating P GSCs). Intratumoral EV transfer between P and M GSCs (third and fourth row, arrows indicate internalized EVs). Scale bars from top to bottom: 1mm, 100μm, 25μm, 10 μm. D. Heterogenous GSC-originated tumors are associated with worsened survival. Kaplan-Meier curves are shown. N=4; P = 0.007. Average survival: G34 – 18.5 days, G146+G34 – 14 days.

Figure 4

GSC EV proteome predicts patients’ outcome. A-C. Survival analysis based on the impact of the prognostic index of multiple protein-coding genes from all GSC EVs (A), EVM (B) and EVP (C) based on retrospective data extrapolated from the TCGA. D. Summarizing cartoon: the heterogeneity of EV cargo contributes to the diverse complexity and enhanced progression of HGG.