PD-1 nanoclustering as a mechanism potentiating EV-mediated suppression of T-cells

Introduction: : Solid tumor cancers develop within protective microenvironments that pose significant challenges to immunotherapeutic strategies due to their highly immunosuppressive nature. Tumor-derived extracellular vesicles (TEVs) potently mediate such immunosuppression through the bioactive cargo they carry. The immune checkpoint molecule PD-L1 is one such cargo that potently suppresses recipient cytotoxic T-cells. However, only a small fraction of the TEV population is enriched for PD-L1, and thus the mechanism driving their potent immunosuppression remains unclear. Previously, we found that PD-L1+ TEVs can alter the localization of PD-1 molecules on the T-cell membrane in silico. Thus, we hypothesize that TEVs induce PD-1 “nanoclusters” on the T-cell membrane and that these nanoclusters are productive, inhibitory signaling complexes that drive T-cell suppression. To test this, we assess the interactions between MDA-MB-231 triple-negative breast cancer cell TEVs and primary human cytotoxic T-cells using dSTORM superresolution microscopy and ELISA cytokine profiling. Using these methods, we find that TEVs do alter PD-1 localization on recipient T-cells and that TEV-treated T-cells exhibit potent suppression. These results provide novel evidence for PD-1 nanoclusters as a mechanism potentiating TEV suppression of cytotoxic T-cells.

Materials and

Methods: : TEVs were isolated via differential ultracentrifugation from the supernatant of cultured MDA-MB-231 cells that were stimulated with 10ng/mL of IFNg. PD-L1 expression on harvested TEVs was rigorously characterized using Nanoparticle Tracking Analysis, Western Blots, and nanoscale flow cytometry. Primary human cytotoxic T-cells were isolated from the blood of healthy donors at the University of Pennsylvania’s Human Immunology Core. These T-cells were stimulated with CD3/CD28 antibodies for 24 hours before treatment with TEVs to induce antigen-independent activation. TEVs used for treating T-cells were pre-treated with either saline or a PD-L1 blocker antibody to disrupt PD-L1/PD-1 interactions between the TEVs and T-cells. Direct STochastic Optical Reconstruction Microscopy (dSTORM) was then performed on TEV-treated T-cells using ONI’s Nanoimager machine. dSTORM images were analyzed using ONI’s CODI cloud platform to identify and characterize TEV clusters. These clustering analyses were supplemented by co-localization studies to quantify the difference in clustering behavior between different conditions. Specifically, Kolmogorov-Smirnov tests were performed to quantify the difference between density profiles between conditions (ie. saline vs PD-L1 antibody pre-treatment of TEVs). Parallel to these dSTORM experiments, we used ELISA assays to profile the T-cells’ responses to the TEVs. T-cells and TEVs were stimulated and pre-treated, respectively, as detailed above. Following TEV treatment, T-cell supernatant was profiled for the inflammatory cytokines IFNg and TNFa. ELISA assays were performed in triplicate, and Welch’s statistical t-tests were performed to determine the significance of observed differences between conditions.

Results, Conclusions, and Discussions:: Using Western Blot and NanoFCM analyses, we characterize PD-L1 expression on TEVs derived from MDA-MB-231 (MDA) triple-negative breast cancer cells (Fig.1A,B). From these results, we find that while PD-L1 is enriched on MDA TEVs compared to TEVs derived from other cell lines (data not shown), only one percent of the MDA TEV population is detectably positive for PD-L1. Additionally, we previously simulated TEV/T-cell interactions to characterize the density profiles of PD-1 and the integrin LFA-1 on the T-cell membrane. From this, we found that TEVs induced more PD-1 clustering within a 150nm radius (the size of a TEV) compared to LFA-1 clustering (Fig.1C). Furthermore, we found that TEV-levels of PD-L1 were capable of potent reduction in T-cell activation when such nanoclustering behavior was accounted for (Fig.1D). To verify these in silico findings, we use dSTORM superresolution microscopy to visualize the spatial organization of PD-1 molecules on the membranes of stimulated primary human cytotoxic T-cells treated with MDA TEVs. We use these dSTORM images to quantify the difference in observed nanocluster patterns induced by TEVs pre-treated with either saline or a PD-L1 blocking antibody. An example of our analysis pipeline is shown using tetraspanins on our MDA TEVs (Fig.2). From this pipeline, we further characterize heterogeneity in marker expression in our sample (Fig.2B) and use localization data from our dSTORM images (Fig.2A) to generate density profiles for each marker (Fig.2C). We then quantify differences in density profiles between conditions or markers using Kolmogorov-Smirnov tests (Fig.2D). From these tests, we find that PD-L1+ TEVs alter PD-1 nanoclustering behavior on the membranes of recipient T-cells. Furthermore, we profile the supernatant of TEV-treated T-cells for IFNg and TNFa using ELISA assays and find that PD-L1+ TEVs significantly reduce expression of these cytokines. Taken together, these results provide novel, mechanistic evidence for how MDA TEVs use PD-1 nanoclusters to mediate potent suppression of recipient cytotoxic T-cells.

Acknowledgements and/or References (Optional):: We acknowledge funding from NCI under the Physical Sciences in Oncology Network and NHGRI under the T32 fellowship program.