Oncolytic viruses alter the biogenesis of tumor extracellular vesicles and influence their immunogenicity

Abstract

Extracellular vesicles (EVs) are mediators of intercellular communication in the tumor microenvironment. Tumor EVs are commonly associated with metastasis, immunosuppression or drug resistance. Viral infections usually increase EV secretion, but little is known about the effect of oncolytic viruses (OVs) on tumor EVs. Here, we investigated the impact of oncolytic vesicular stomatitis virus (VSV) and vaccinia virus on EVs secreted by human melanoma and thoracic cancer cells. We found that OV infection increases the production of EVs by tumor cells. These EVs contain proteins of viral origin, such as VSV-G, thus creating a continuum of particles sharing markers of both canonical EVs and viruses. As such, the presence of VSV-G on EVs improves the transfer of their protein content to cell types commonly found in the tumor microenvironment. A proteomic analysis also revealed that EVs-OV secreted during VSV infection are enriched in immunity-related proteins. Finally, CD8⁺ T cells incubated with EVs-OV from infected cells display slightly enhanced cytotoxic functions. Taken together, these data suggest that OVs enhance the communication mediated by tumor EVs, which could participate in the therapeutic efficacy of OVs. These results also provide rationale for engineering OVs to exploit EVs and disseminate therapeutic proteins within the tumor microenvironment.

Keywords: MT: Regular Issue; extracellular vesicles; immunogenicity; immunotherapy; intercellular communication; oncolytic viruses; t cell; vaccinia virus; vesicular stomatitis virus.

Graphical abstract

Figure 1

Tumor cells secrete more EVs upon OV infection (A and B) M113 melanoma cells were infected by VSV-GFP or VACV-GFP at a MOI of 0.1 or 1. Cell viability (A) and expression of the viral transgene (B) were measured over time. Dotted lines indicate when EVs were harvested in subsequent experiments. Data are presented as mean (SD). n = 3–4 biological replicates. (C) Representative transmission electron micrographs of EVs secreted by uninfected, VSV-infected or VACV-infected M113 melanoma cells. Scale bars, 100 nm. (D) Diameter of EV-shaped particles identified by TEM. Data are represented as mean (SD). n = 103–124 single EVs per condition. ∗∗p = 0.0032 (Kruskal-Wallis test). (E) Western blot analysis of EV (ALIX, CD81, and CD63) or cellular (calnexin) marker expression in cell lysates (C) or corresponding EV lysates (EVs) from uninfected, VSV-infected or VACV-infected M113 cells. Three μg of proteins were loaded in each lane. Representative of five independent experiments. (F and G) Single-EV flow cytometry analysis of EVs stained for tetraspanins (CD9/CD81/CD63). (F) EVs were gated based on the forward and side scatter parameters. (G) Relative quantification of CD9/CD63/CD81⁺ EVs secreted by M113, M6 melanoma, or ADCA153 lung adenocarcinoma cells. Data are presented as mean (SD). n ≥ 3 biological replicates. ∗∗∗∗p < 0.0001, ∗p = 0.0199 and 0.0189 for M6 and ADCA153, respectively (Kruskal-Wallis test). (H and I) Western blot analysis of EVs secreted by tumor cells, with equal volumes (10 μL) of EV lysates from uninfected or VSV-infected cells loaded in both lanes. Representative of 6 biological replicates. (I) Quantification of the CD63 signal on 6 biological replicates. Data are presented as mean (SD). ∗∗p = 0.0022 (Mann-Whitney test).

Figure 2

Tumor EVs spontaneously package proteins of viral origin (A) Western blot analysis of GFP and CD81 in EVs secreted by M113 cells infected by GFP-encoding VSV or VACV. Representative of three biological replicates. (B) Proteinase protection assay of EVs secreted by M113, Meso163 or ADCA153 cells infected by VSV-NLuc. EVs were incubated with either detergent, proteinase K or both before adding the NLuc substrate. Data are presented as mean (SD). n = 3 biological replicates. ∗p = 0.0125, 0.0372, and 0.0125 for M113, Meso163, and ADCA153, respectively (Kruskal-Wallis test). (C and D) Representative transmission electron micrographs of EVs secreted by uninfected or VSV-infected M113 cells and labeled with (C) an anti-VSV-G antibody coupled to 6-nm gold particles or (D) anti-CD63 (10-nm gold particles) and anti-VSV-G (6-nm gold particles) antibodies. Arrows indicate CD63 staining. (E) Single EV flow cytometry analysis (CD9/CD63/CD81 and VSV-G) of EVs secreted by uninfected, VSV-infected, or VACV-infected M113 cells. (F) Relative quantification of CD9/CD63/CD81/VSV-G⁺ EVs secreted by M113 cells. Data are presented as mean (SD). n = 3 biological replicates.

Figure 3

Viral material enhances EV-mediated intercellular protein transfer (A) EVs secreted by uninfected, VSV-infected or VACV-infected M113-NLuc cells were incubated for 4 h with parental M113 cells. EV internalization by recipient cells was measured by analyzing luminescence of target cells, normalized with input EV luminescence. Data are presented as mean (SD). n = 3 or 4 biological replicates. ∗p = 0.0115 (Kruskal-Wallis test). (B and C) EVs of Lenti-X 293T cells transfected to express Cre ± VSV-G were incubated with tumor cells transduced to express GFP upon Cre recombination. (B) Western blot analysis of Cre, VSV-G, ALIX and calnexin in purified EVs from transfected HEK cells. Three micrograms of proteins were loaded in each lane. Representative of two independent experiments. (C) Flow cytometry analysis of recipient cells incubated with Cre⁺ or Cre⁺/VSV-G⁺ EVs. The dotted lines indicate the background percentage of recipient cells expressing GFP. Data are presented as mean (SD). n = 2–4 biological replicates. ∗p = 0.05 for H441, H1975, Meso34, M113, HFF-2, and 0.0476 for THP-1 (Mann-Whitney test). (D) Confocal micrographs of tumor spheroids (loxP-dsRed/GFP-loxP H441 cells) incubated with mock, Cre⁺, or Cre⁺/VSV-G⁺ EVs. Images represent the maximum intensity Z-projections of all imaged slices. Scale bar, 100 μm. (E) Quantification of the GFP⁺ area in tumor spheroids (loxP-dsRed/GFP-loxP Meso34 or H441 cells) from (D). Data are represented as the ratio of GFP⁺ area over dsRed⁺ area within the same spheroid. n = 2–4 spheroids imaged per condition. ∗p = 0.0495 (Kruskal-Wallis test). (F) Confocal micrographs of MCTS (unlabeled Meso34 cells + loxP-dsRed/GFP-loxP THP-1 cells) incubated with mock, Cre⁺ or Cre⁺/VSV-G⁺ EVs. Images represent the maximum intensity Z-projections of all imaged slices. Scale bar, 100 μm. (G) Quantification of the GFP⁺ area in tumor spheroids (unlabeled Meso34 + loxP-dsRed/GFP-loxP THP-1 or HFF-2 cells) from (F). Data are represented as the ratio of GFP⁺ area over dsRed⁺ area within the same spheroid. n = 3–5 spheroids imaged per condition. ∗p = 0.0426 (Kruskal-Wallis test).

Figure 4

Exploratory MS screen identifies immunity-related proteins in EVs secreted by VSV-infected tumor cells (A–G) Liquid chromatography MS-based proteome analysis of EVs secreted by M113 cells. n = 2 biological replicates. (A and B) Relative quantification of proteins most commonly identified in Vesiclepedia (A) or annotated as ectosomes or exosomes markers (B). (C) Venn diagram of the identified proteins. (D) Volcano plot of proteins significantly enriched or depleted in EVs-VSV. Proteins encoded by IFN-stimulated genes are highlighted in red. (E) Significantly enriched Gene Ontology (GO) biological pathways in EVs-VSV. (F) Volcano plot of proteins significantly enriched or depleted in EVs-VACV. Proteins involved in endosomal transport are highlighted in blue. (G) Significantly enriched GO biological pathways in EVs-VACV. (H and I) Western blot validation of the enrichment of the melanoma antigen Melan-A in EVs-VSV secreted by M113 cells. (H) Representative experiment of 3 biological replicates. (I) Relative quantification of the Melan-A signal/CD81 signal in cell lysates (C) or EVs. n = 3 biological replicates. ∗p = 0.05. (J–L) Western blot validation of the enrichment of MHC class I molecules in EVs-VSV or EVs-VACV. (J) Detection of HLA-ABC and β2-microglobulin in EVs from uninfected and VSV-infected M113 cells. Three μg of proteins were loaded in each lane. Representative of 3 biological replicates. (K) Detection of HLA-ABC, CD63, CD81 and calnexin in EVs from uninfected and VSV-infected M6 or ADCA153 cells. Three micrograms of proteins were loaded in each lane. Representative of three biological replicates. (L) Relative quantification of the HLA-ABC signal in EVs-VSV and EVs-VACV from different cell lines compared with EVs from uninfected cells. n = 5–7 biological replicates. ∗∗p = 0.0039 and 0.084 for EVs-VSV and EVs-VACV, respectively (Kruskal-Wallis test).

Figure 5

EVs secreted by OV-infected tumor cells partly enhance the cytotoxicity of anti-tumor CD8⁺ T cell clones (A) EVs secreted by M113 cells were incubated with the WT4 or CTL03.1 CD8⁺ T cell clones. Supernatants from the coculture were collected, and T cells were incubated with M113 target cells to assess their functions. All data are represented as mean (SD). (B) Cytotoxicity of T cells (clones WT4 and CTL03.1) measured by NLuc release from target M113-NLuc cells. n = 3 biological replicates. ∗p = 0.05 (Mann-Whitney test). (C) ELISA analysis of granzyme-B secretion induced by EV incubation, normalized to the spontaneous granzyme-B secretion by T cells that were not incubated with EVs. n = 4–8 biological replicates. ∗p = 0.0176 (Mann-Whitney test).