Extracellular Vesicles from Ocular Melanoma Have Pro-Fibrotic and Pro-Angiogenic Properties on the Tumor Microenvironment

Abstract

Uveal melanoma (UM) is the most common primary intraocular tumor and often spreads to the liver. Intercellular communication though extracellular vesicles (EVs) plays an important role in several oncogenic processes, including metastasis, therapeutic resistance, and immune escape. This study examines how EVs released by UM cells modify stellate and endothelial cells in the tumor microenvironment. The surface markers, and the concentration and size of EVs derived from UM cells or choroidal melanocytes were characterized by high-resolution flow cytometry, electron microscopy, and Western blotting. The selective biodistribution of EVs was studied in mice by fluorescence imaging. The activation/contractility of stellate cells and the tubular organization of endothelial cells after exposure to melanomic EVs were determined by traction force microscopy, collagen gel contraction, or endothelial tube formation assays. We showed that large EVs from UM cells and healthy melanocytes are heterogenous in size, as well as their expression of phosphatidylserine, tetraspanins, and Tsg101. Melanomic EVs mainly accumulated in the liver and lungs of mice. Hepatic stellate cells with internalized melanomic EVs had increased contractility, whereas EV-treated endothelial cells developed more capillary-like networks. Our study demonstrates that the transfer of EVs from UM cells leads to a pro-fibrotic and pro-angiogenic phenotype in hepatic stellate and endothelial cells.

Keywords: endothelial cells; extracellular vesicles; hepatic stellate cells; metastatic uveal melanoma.

Figure A1

Dynasore and proteinase K pretreatments prevented EV internalization in stellate cells. (A) Fluorescence of HSteCs without addition of EVs by flow cytometry. Fluorescence of HSteCs after the addition of DiI-labeled large EVs derived from UM cell lines (Mµ2F, TJU-UM001) or NCMs for 24 h (A,B). (B) Fluorescence of HSteCs pretreated with Dynasore or after the addition of TJU-UM001-EVs pretreated with proteinase K are shown as controls. The gates indicate the percentage of fluorescent HSteCs.

Figure A2

Dynasore and proteinase K pretreatments prevented UM-derived EV internalization in endothelial cells and inhibit the formation of capillary-like tubes. Untreated GFP-HUVECs were cultured for 12 h on Matrigel^® with medium without angiogenic factors (negative control; black bars). Large EVs isolated from metastatic UM cell line TJU-UM001 were added to GFP-HUVECs for 24 h before their seeding on Matrigel^® for 12 h (white bars). GFP-HUVECs were also pretreated with Dynasore (dark grey bars), or TJU-UM001-EVs were pretreated with proteinase K (light grey bars) as controls. The number of junctions, master junctions, nodes, and meshes were quantified in function of time (hours) for each condition (n = 1).

Figure 1

UM cells release more EVs than melanocytes. The size distribution of EVs derived from NCMs, non-metastatic UM cell line Mµ2, and metastatic cell lines Mµ2F and TJU-UM001 was determined by HR-FC. (A) EV size was determined with forward and side scatter, and EVs between 100–1000 nm were included in the analysis. (B) The NCM- and UM-derived EVs were stained with CMFDA to identify intact EVs and exclude debris. CMFDA-positive events were analyzed for Anx5 (PS)/CD63 expression. (C) The concentration of EVs (particles/mL) for the various CMFDA+/PS/CD63 subpopulations was determined by normalizing the number of particles to the total volume acquired for each sample.

Figure 2

Morphology of UM-derived EVs. (A) Large EVs isolated from metastatic UM cell lines TJU-UM001 and Mµ2F or HSteCs (healthy liver cell controls) were labeled with 10-nm CD63-gold-NPs or Anx5-gold-NPs (dark particles) for cryo-TEM analysis. Scale bars, 100 nm. (B) The percentage of EVs positive for CD63 or PS was quantified for each cell type.

Figure 3

Exosomal and melanocytic markers present in UM-derived EVs. Cell pellets and EV fractions isolated from UM cell lines TJU-UM001 (metastatic), Mµ2 (non-metastatic), and Mµ2F (metastatic), or NCMs, were analyzed for their expression of exosomal (CD9, CD63, CD81, Tsg101) or melanocytic (Melan-A, TYRP1) markers by Western blotting. The anti-α-tubulin was used as marker of cytosolic proteins. Ponceau S staining was used as a loading control. Arrows indicate the specific band for each marker.

Figure 4

Accumulation of melanomic EVs in the liver and lungs of immunodeficient mice. (A) Inoculation of PBS served as control without EVs. (B) Large EVs derived from HSteCs, (C) HEK293, or (D–F) UM cell lines (Mµ2, Mµ2F, TJU-UM001) were labeled with Exoglow before being inoculated in the retro-orbital sinus of NCG mice (n = 4/condition). Fluorescence of the liver, lungs, kidneys, heart, and spleen was analyzed with the IVIS after 24 h.

Figure 5

Activation of stellate cells after the internalization of UM-derived EVs. (A) The expression of F-actin (Phalloidin 647; grey) and α-SMA (green) was assessed in HSteCs without addition of EVs by confocal microscopy. Nuclei were counterstained with DAPI (blue). (B) Large EVs isolated from NCMs (normal cell control) or (C) metastatic UM cell lines Mµ2F and (D–F) TJU-UM001 were labeled with DiI (red) and added to HSteCs for 24 h to allow for their uptake before observing the expression of F-actin (grey) and α-SMA (green). (E) HSteCs were pretreated with Dynasore, or (F) TJU-UM001-EVs were pretreated with proteinase as controls. Scale bar, 50 µm.

Figure 6

Increased contractility of stellate cells after the internalization of UM-derived EVs. HSteCs were grown on 5 kPa polyacrylamide gels containing fluorescent beads and exposed for 24 h to large EVs from metastatic UM cells TJU-UM001. Then, bead displacements between stressed (+EVs) and null (CTRL) states were measured using TFM. (A) Representative traction stress maps (Pa) and (B) traction force (nN) are shown for untreated (CTRL) or treated (+EVs) HSteCs. Scale bar, 50 µm. Plots are median ± the minimum and maximum values. * p < 0.05 (Welch’s t-test).

Figure 7

UM-derived EVs increase collagen gel contraction by stellate cells. (A) Representative images of collagen gels (1 mg/mL) without embedded HSteCs, with embedded HSteCs not treated with EVs, and with embedded HSteCs treated during 24 h with large EVs derived from NCMs or metastatic UM cell line TJU-UM001 taken with a stereomicroscope. (B) Corresponding quantification of the collagen gel area (in percentage) for each condition during 48 h. Error bars indicate ±SEM (n = 3); ** p < 0.01 and **** p < 0.0001 (two-way ANOVA with Sidak’s multiple comparisons test).

Figure 8

UM-derived EVs increased the formation of capillary-like tubes by endothelial cells. (A) Untreated GFP-HUVECs (green) were cultured for 12 h on Matrigel^® with medium without VEGF (negative control; top panels) or medium supplemented with VEGF (positive control; middle panels). Images are from the 6 h time point. Nuclei were counterstained with DAPI (blue). Scale bars: Merge 5X, 200 µm; Merge 20X, 50 µm. (A) Large EVs isolated from metastatic UM cell line TJU-UM001 were labeled with DiI (red) and added to GFP-HUVECs (green) for 24 h before their seeding on Matrigel^® for 6 h (bottom panels). (B) The number of junctions, master junctions, nodes, and meshes were quantified in function of time (hours) for each condition. Error bars indicate SEM (n = 3); * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001 (two-way ANOVA with Sidak’s multiple comparisons test).