RIG-I activation induces the release of extracellular vesicles with antitumor activity

Abstract

Activation of the innate immune receptor retinoic acid-inducible gene I (RIG-I) by its specific ligand 5'-triphosphate-RNA (3pRNA) triggers antitumor immunity predominantly via NK cell activation and direct apoptosis induction in tumor cells. However, how NK cells are mobilized to attack the tumor cells remains elusive. Here, we show that RIG-I activation induced the secretion of extracellular vesicles (EVs) from melanoma cells, which by themselves revealed antitumor activity in vitro and in vivo. RIG-I-induced EVs from melanoma cells exhibited an increased expression of the NKp30-ligand (BAG6, BAT3) on their surface triggering NK cell-mediated lysis of melanoma cells via activation of the cytotoxicity NK cell-receptor NKp30. Moreover, systemic administration of RIG-I-induced melanoma-EVs showed a potent antitumor activity in a melanoma mouse model in vivo. In conclusion, our data establish a new RIG-I-dependent pathway leading to NK cell-mediated tumor cell killing.

Keywords: BAG6; NK cells; NKp30; RIG-I; exosomes; extracellular vesicles; melanoma.

Figure 1.

RIG-I stimulation triggers the release of extracellular vesicles (EVs). (A) Melanoma cells D04mel or Ma-Mel-86c were analyzed for mRNA expression of RIG-I by quantitative real-time PCR in the presence or absence of type I Interferon (IFN) stimulation. Data are normalized to β-Actin. (n = 3) (B) Functionality of RIG-I signaling in D04mel and Ma-Mel-86c was determined by CXCL10 production 24 h after lipofection of cells with 1 µg/mL ctrl RNA or 3pRNA (n = 3) (C) Schematic overview of experimental procedures. Tumor cells (D04mel or Ma-Mel-86c) were transfected with 3pRNA (RIG-I ligand) or inert control RNA (non-RIG-I ligand) and EVs (RIG-I-EVs vs. ctrl-EVs) were purified using serial ultracentrifugation. Afterwards, EVs were analyzed regarding their effects on immune cells. (D) NTA analysis of purified vesicles (RIG-I-EVs vs. ctrl-EVs) derived from melanoma cells. (E) Cryo electron microscopy shows typical particles obtained by purification. One white or black scale bar indicates 100 nm. (F) Purified vesicles (RIG-I-EVs or ctrl-EVs) derived from D04mel cells were analyzed for expression of CD9, CD63, CD81 by western blot. (G) Purified vesicles (RIG-I-EVs vs. ctrl-EVs) derived from melanoma cells (D04mel, Ma-Mel- 86c) were analyzed for CD9 expression by flow cytometry (filled gray: isotype, dashed: ctrl-EVs, black line: RIG-I-EVs). (H) Amount of EVs after stimulation with 3pRNA (RIG-I-EVs) or ctrl RNA (ctrl-EVs) derived from melanoma cells (D04mel, Ma-Mel-86c) were estimated by quantification of proteins using Bradford Assay (n = 4–9). (I) Particle number of EVs after stimulation with 3pRNA (RIG-I EVs) or ctrl RNA (ctrl EVs) derived from melanoma cells (D04mel) was determined by NTA analysis (n = 6). All error bars reflect mean ± s.d. * indicates p < 0.05.

Figure 2.

Evs derived from RIG-I stimulated cells express enhanced levels of the NKp30-ligand BAG6. (A) CFSE labeled EVs (EV protein amount: 10 µg/mL) induced by 3pRNA (RIG-I-EVs) versus ctrl RNA (ctrl-EVs) were incubated with PBMCs and 24 h later CFSE staining of NK cells (CD3 negative, CD56 positive) or CD3 positive lymphocytes (CD3⁺ cells) were determined by flow cytometry (n = 3). (B) D04mel cells were transfected with 3pRNA (RIG-I-EVs) or ctrl RNA (ctrl-EVs) and the expression of MIC A/B, ULBP 1/2/3, Vimentin, B7-H6 and BAG6 on EVs was analyzed by flow cytometry (filled gray: isotype, dashed: ctrl-EVs, black line: RIG-I-EVs). One representative of four independent experiments is shown. (C) EVs induced by 3pRNA (RIG-I-EVs) vs. ctrl RNA (ctrl-EVs) were analyzed for binding of NKp30-fc by flow cytometry. Histogram shows one representative experiment (left, filled gray: isotype, dashed: ctrl-EVs, black line: RIG-I-EVs) and graph (right) shows quantification of x-fold induction of the geometric mean normalized to CD9 (n = 4). (D) Expression level of BAG6 on D04mel cells (left) or D04mel derived EVs (right) after transfection with 3pRNA or ctrl RNA was determined by flow cytometry (n = 5). (E) Purified EVs from melanoma (Ma-Mel-86c) cells induced by 3pRNA (RIG-I-EVs) versus ctrl RNA (ctrl-EVs) were analyzed for BAG6 expression on the surface by flow cytometry. Graphs show % induction of the geometric mean normalized to CD9 and s.e.m. of at least four independent experiments. (F) Exosomes from cells with siRNA mediated control knock down (ctrl k_d) or RIG-I knock down (RIG-I k_d) were analyzed for BAG6 expression on the surface by flow cytometry in response to 3pRNA (RIG-I-EVs) vs. ctrl RNA (ctrl-EVs). Graph shows geometric mean of BAG6 relative to CD9 (n = 4). All error bars reflect mean ± s.d. *, ** and *** indicates p < 0.05, p < 0.01 and p < 0.001.

Figure 3.

BAG6-positive RIG-I-EVs induce enhanced NK cell cytotoxicity in vitro. (A) PBMCs were incubated over night with RIG-I-EVs versus ctrl-EVs (EV protein amount: 10 µg/mL) and CD69 expression on NK cells (CD3 negative, CD56 positive) was determined by flow cytometry (n = 3). (B) Primary purified NK cells were left untreated (PBS) or incubated with 100 µg/mL (protein amount) RIG-I-EVs vs. ctrl-EVs for 36 h with (aNKp30) or without blocking (Isotype control antibody (AB)) of NKp30 (clone P30-15). Expression of CD69 (MFI) on NK cells after incubation with EVs was measured and normalized to PBS (n = 3). (C+D) Done as described in (B). (C) Cytotoxicity against untreated melanoma cells (D04mel) was assessed by europium release assay and results were normalized to 100% lysis by RIG-I-EVs in different E/T (n = 3). (D) Instead of NKp30 on NK cells, BAG6 was blocked on EVs with anti-BAG6 (aBAG6). Results were normalized to 100% lysis by RIG-I-EVs in an E/T-ratio of 5:1 (n = 3 for ctrl- and RIG-I-EVs + aBAG6, n = 2 for RIG-I-EVs + Isotype ctrl antibody). All error bars reflect mean ± s.d. *, ** and *** indicates p < 0.05, p < 0.01 and p < 0.001.

Figure 4.

RIG-I-EVs lead to activation of NK cells and inhibition of tumor growth in vivo. HCmel12 cells were transfected with 3pRNA (RIG-I-EVs) or ctrl RNA (ctrl-EVs) and the expression of BAG6 on EVs was analyzed by flow cytometry (n = 3). (B) Activation of mouse splenocytes with 10 µg/mL (protein amount) EVs ex vivo. Graph (left) shows CD69 expression on NK cells (NK1.1+CD3⁻) was determined by flow cytometry (n = 5). Right histogram shows one representative experiment (left, filled gray: isotype, dashed: ctrl-EVs, black line: RIG-I-EVs) (C–E) Application of RIG-I- or ctrl-EVs in vivo. (C) HCmel12 mouse melanoma cell derived EVs (20 µg EV protein amount per mouse) were injected intravenously. Graph (left) shows expression of CD69 on NK cells (NK1.1+CD3⁻) purified from the lymph node was determined by flow cytometry (n = 5). Histogram (right) shows one representative experiment (left, filled gray: isotype, dashed: ctrl-EVs, black line: RIG-I-EVs) (D) Treatment schema of in vivo experiment. C57BL/6 mice were injected with HCmel12 mouse melanoma cells subcutaneously in the flank at day 0 and treated with melanoma-derived EVs at day 6, 8, 10, 13. Melanoma bearing mice were treated with PBS (ctrl), 20 µg protein amount of 3pRNA-induced EVs (RIG-I-EVs) or EVs induced by control RNA (ctrl-EVs), both derived from HCmel12 cells. Mice were sacrificed at day 14. (E) Tumor size was measured in treated and untreated mice with or without depletion of NK cells using antibody directed against NK1.1 (NK-AB). Mean tumor size and s.d. of 5–9 animals are shown. Arrow indicates begin of treatment, filled square: PBS, filled triangle: ctrl-EVs, filled circle: RIG-I-EVs, empty triangle: ctrl-EVs+NK-AB, empty circle: RIG-I-EVs+NK-AB. *, ** and *** indicates p < 0.05, p < 0.01 and p < 0.001.