NK cell-based therapies have shown promise for hematological cancer forms, but their use against solid tumors is hampered by their poor ability to infiltrate the tumor. NK cells release extracellular vesicles (EVs) containing cytolytic proteins, indicating that NK-cell derived EVs may have therapeutic potential. In this study, we compared the tumor-targeting potential of EVs derived from either primary NK cells or the NK cell lines NK-92 and KHYG-1 cultured in IL-15 alone or in combination with IL-12 and IL-18. Primary NK cells were also stimulated through the activating receptor CD16. Tumor cell apoptosis was measured using a panel of human colon, melanoma, glioblastoma, prostate, breast, and ovarian tumor cell line spheroids. NK cells or NK-92 cells stimulated with IL-12, IL-15, and IL-18 generated EVs with higher efficiency than EVs from resting cells, although similar amounts of EVs were produced under both conditions. Proteomic analysis indicated similar distribution of cytolytic proteins in EVs from primary NK cells and NK-92, but lower levels in KHYG-1 EVs that translated into poor capacity for KHYG-1 EVs at targeting tumor cell lines. Further, we show that CD16-stimulated NK cells release low amounts of EVs devoid of cytolytic proteins. Importantly, EVs from cytokine-stimulated NK cells penetrate into the spheroid core, and tumor spheroid susceptibility to NK-cell derived EVs was linked to differential expression of the NKG2D ligands MICA/B, which was blocked with an anti-NKG2D antibody. We conclude that EVs from activated primary NK cells or NK-92 cells has the best potential to infiltrate and target solid tumors.

NK cells can broadly target and kill malignant cells via release of cytolytic proteins. NK cells also release extracellular vesicles (EVs) that contain cytolytic proteins, previously shown to induce apoptosis of a variety of cancer cells in vitro and in vivo. The EVs released by NK cells are likely very heterogeneous, as vesicles can be released from the plasma membrane or from different intracellular compartments. In this study, we undertook a fractionation scheme to enrich for cytolytic NK-EVs. NK-EVs were harvested from culture medium from the human NK-92 cell line or primary human NK cells grown in serum-free conditions. By combining ultracentrifugation with downstream density-gradient ultracentrifugation or size-exclusion chromatography, distinct EV populations were identified. Density-gradient ultracentrifugation led to separation of three subpopulations of EVs. The different EV isolates were characterized by label-free quantitative mass spectrometry and western blotting, and we found that one subpopulation was primarily enriched for plasma membrane proteins and tetraspanins CD37, CD82, and CD151, and likely represents microvesicles. The other major subpopulation was enriched in intracellularly derived markers with high expression of the endosomal tetraspanin CD63 and markers for intracellular organelles. The intracellularly derived EVs were highly enriched in cytolytic proteins, and possessed high apoptotic activity against HCT-116 colon cancer spheroids. To further enrich for cytolytic EVs, immunoaffinity pulldowns led to the isolation of a subset of EVs containing the cytolytic granule marker NKG7 and the majority of vesicular granzyme B content. We therefore propose that EVs containing cytolytic proteins may primarily be released via cytolytic granules.

MLL-rearranged leukemias are among the leukemic subtypes with poorest survival, and treatment options have barely improved over the last decades. Furthermore, despite increasing molecular understanding of the mechanisms behind these hematopoietic malignancies, this knowledge has had poor translation into the clinic. Identification of novel treatment methods is hampered by the lack of relevant in vivo models that allow for rapid identification of actionable drug targets and small molecule inhibitors. Here, we report a Drosophila melanogaster model system to explore the pathways affected in MLL-rearranged leukemia. We show that expression of the human leukemic oncogene MLL-AF4 in the Drosophila hematopoietic system resulted in increased levels of circulating hemocytes and an enlargement of the larval hematopoietic organ, the lymph gland. Strikingly, depletion of Drosophila orthologs of known interactors of MLL-AF4, such as DOT1L, rescued the leukemic phenotype. In agreement, treatment with small-molecule inhibitors of DOT1L also prevented the MLL-AF4-induced leukemia-like phenotype. Taken together, this model provides an in vivo system to unravel the genetic interactors involved in leukemogenesis and offers a strategy for a prompt identification of potential therapeutic options for treatment of MLL-rearranged leukemia.