P06.07 Immune evasion mediated by PD-L1 on glioblastoma derived extracellular vesicles

• Published in: Neuro-oncology (Charlottesville, Va.) Va.), 2017-05, Vol.19 (suppl_3), p.iii50-iii50
• Main Authors: Ricklefs, F. L., Speranza, M. C., Hayes, J., Balaj, L., Breakefield, X. O., Bronisz, A., Carter, B., Freeman, G., Lawler, S., Chiocca, E. A.
• Format: Article
• Language: English
• Subjects: POSTER PRESENTATIONS
• ISSN: 1522-8517; 1523-5866
• DOI: 10.1093/neuonc/nox036.181

Background: Extracellular Vesicles (EVs) shed by tumor cells have recently been demonstrated to act as a major conduit in cell-cell communication. Increasing knowledge of the effect of EVs on infiltrating lymphocytes within the tumor microenvironment shows the immune modulation capacity of EVs. In this study we investigated the potential role of PD-L1 on EVs in immune escape in heterogeneous glioblastoma (GBM). Methods: Peripheral blood mononuclear cells (PBMCs) (n=8) were stimulated by IL-2 or anti-CD3±anti-CD28±anti-PD1 treatment. Magnetic cell sorting was used to isolate CD3+ cells. Activation levels of CD3+CD4+, CD3+CD8+ and CD3-CD56+ were monitored by flow cytometry of the activation markers CD69, CD25, PD1 and TIM3. Concurrent eFluor staining was used to measure proliferation. EVs derived from four different glioblastoma stem-like cell lines (GSCs) from either the mesenchymal (M) or proneural (P) subtype were used in this study. PD-L1 expression was validated by immunoblotting and electron microscopy (EM). EV binding was visualized by PALM-tdTomato and PD-L1_RFP positive EVs and tested on plate-bound PD1. RNA-seq from TCGA and Immunohistochemistry (IHC) was used to correlate PD-L1 expression with CD3 infiltration. In addition, circulating EVs from GBM patients ( n =22) and controls ( n =5) were used to determine PD-L1 content via digital droplet-PCR. Results: PD-L1 was expressed on the surface of M GSC EVs. EVs were able to bind to the outer surface of CD3+ cells. In whole PBMCs EV treatment led to a significant, PD-L1 dependent, reduction of CD3+CD8+ and CD3+CD4+ T-cell activation as well as decreased proliferation in the M subtype. These effects were also observed in CD3+ sorted cells, indicating a direct effect of EVs on T cells. Furthermore PD-L1 on EVs was capable of binding directly to PD1. Using RNA-seq and IHC we show a correlation of CD3 infiltration and PD-L1 expression. In addition, circulating EVs from GBM patients show enriched PD-L1 DNA cargo that correlates with tumor volume. CONCLUSION: Our findings demonstrate the immunosuppressive potential of GSC-derived EVs in a PD-L1 dependent manner. Furthermore we show that PD-L1 on EVs can directly bind PD1 and is capable of hindering T cell activation locally and at distant sites and show how immune checkpoint blockade may systemically enhance immunity against GBM.