Centrosome amplification mediates small extracellular vesicle secretion via lysosome disruption

Abstract

Bidirectional communication between cells and their surrounding environment is critical in both normal and pathological settings. Extracellular vesicles (EVs), which facilitate the horizontal transfer of molecules between cells, are recognized as an important constituent of cell-cell communication. In cancer, alterations in EV secretion contribute to the growth and metastasis of tumor cells. However, the mechanisms underlying these changes remain largely unknown. Here, we show that centrosome amplification is associated with and sufficient to promote small extracellular vesicle (_SEV) secretion in pancreatic cancer cells. This is a direct result of lysosomal dysfunction, caused by increased reactive oxygen species (ROS) downstream of extra centrosomes. We propose that defects in lysosome function could promote multivesicular body fusion with the plasma membrane, thereby enhancing _SEV secretion. Furthermore, we find that _SEVs secreted in response to amplified centrosomes are functionally distinct and activate pancreatic stellate cells (PSCs). These activated PSCs promote the invasion of pancreatic cancer cells in heterotypic 3D cultures. We propose that _SEVs secreted by cancer cells with amplified centrosomes influence the bidirectional communication between the tumor cells and the surrounding stroma to promote malignancy.

Keywords: PDAC; ROS; cancer; centrosome amplification; exosomes; extracellular vesicles; invasion; lysosomes; multivesicular bodies; stellate cells.

Figure 1

Centrosome amplification promotes the secretion of _SEVs in PADC cells (A) Representative confocal images of mitotic cells with normal and amplified centrosomes. Cells were stained for α-tubulin (magenta), centrin2 (green), and DNA (cyan). Scale bar, 10 μm. (B) Quantification of _SEVs and _LEVs secreted by PDAC cell lines. Average of the percentage of centrosome amplification (CA) per cell line is highlighted in orange. (C) Linear regression of the data presented in (B) and Spearman correlation coefficients for _SEVs and _LEVs. (D) Quantification of secreted _SEVs and _LEVs in PaTu-S.iPLK4 and HPAF-II.iPLK4 cell lines upon induction of centrosome amplification (+DOX), before and after depletion of Sas-6 by small interfering RNA (siRNA). Average percentage of CA per condition is highlighted in orange. (E) Western blot analyses of proteins associated with _SEVs in extracts from cells and _SEVs collected by UC. (F) Top: representative images of IEM of _SEVs collected from HPAF.iPLK4 cells. Dark beads represent immunogold labeling with anti-CD63. Scale bar, 200 nm. Bottom: quantification of the percentage of positive CD63 _SEVs is shown. (G) Quantification of _SEVs diameter by cryoelectron microscopy (cryo-EM). PaTu-S.iPLK4 _SEVs n_(−DOX) = 232 and n_(+DOX) = 216; HPAF-II.iPLK4 n_(−DOX) = 541 and n_(+DOX) = 493. For all graphics, error bars represent mean ± SD from three independent experiments. ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗∗p < 0.0001. The following statistics were applied: for graphs in (D), two-way ANOVA with Tukey’s post hoc test was applied, and for graphs in (G), unpaired t test was applied. See also Figure S1 and Table S1.

Figure 2

Proteomic analyses of _SEVs secreted by cells with extra centrosomes support their endocytic origin (A) Experimental flowchart. (B) Venn diagram comparing the _SEVs proteomes of SEC fractions 7–9. (C) Venn diagram comparing the _SEVs proteome of SEC fractions 7–9 with the Vesiclepedia database. (D) Dot plot representation of the enrichment analyses performed for the common proteins in all SEC fractions. Only proteins that were identified in both forward and reverse labeling experiments were considered for this analysis. See also Figure S2, Data S1, and Tables S2, S3, and S4.

Figure 3

ROS promote lysosome dysfunction and _SEV secretion in cells with extra centrosomes (A) Schematic representation of intraluminal vesicle formation (ILV) and multivesicular bodies (MVBs) fate and how ROS could affect this process. (B) Levels of intracellular ROS quantified by the ratio of GSH/GSSG in PaTu-S.iPLK4 and HPAF-II.iPLK4 cell lines. Decrease in the GSH/GSSG ratio indicates higher ROS levels. 5 mM of NAC and 100 μM H₂O₂ were used. (C) Representative confocal images of cells stained with Magic Red (magenta), as a proxy for lysosome function, and for DNA (cyan). MAX projection images shown (see Figure S3F for SUM intensity images). Scale bar, 10 μm. (D) Quantification of intracellular Magic Red fluorescence intensity normalized for cell area in PaTu-S.iPLK4 cells. AU, arbitrary units. 5 mM of NAC and 100 μM H₂O₂ were used. n_(−DOX) = 158, n_(+DOX) = 189, n_(+DOX+NAC) = 221, and n_{(−DOX+H2O2)} = 175. (E) Quantification of secreted _SEVs and _LEVs in PaTu-S.iPLK4 and HPAF-II.iPLK4 cell lines. For all graphics, error bars represent mean ± SD from three independent experiments. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001, and n.s., not significant (p > 0.05). The following statistics were applied: for graphs in (B), one-way ANOVA with Tukey’s post hoc test; for (D), one-way ANOVA with a Kruskal-Wallis post hoc test; and for (E) two-way ANOVA with Tukey’s post hoc test. See also Figure S3.

Figure 4

Centrosome amplification decreases MVBs-lysosome co-localization in a ROS-dependent manner (A) Representative confocal images of cells stained for acidic lysosomes (lysotracker, magenta), late endosomes/MVBs (anti-LBPA, green), and DNA (gray). Insets show higher magnification of lysotracker and LBPA-labeled vesicles. Scale bar, 10 μm. (B) Quantification of the number of lysotracker-labeled lysosomes per cell. 5 mM of NAC and 100 μM H₂O₂ were used. n_(−DOX) = 166, n_(+DOX) = 182, n_(+DOX+NAC) = 245, and n_{(−DOX+H2O2)} = 187. (C) Quantification of LBPA-labeled late endosomes/MVBs per cell. 5 mM of NAC and 100 μM H₂O₂ were used. n_(−DOX) = 88, n_(+DOX) = 102, n_(+DOX+NAC) = 129, and n_{(−DOX+H2O2)} = x99. (D) Quantification of the percentage of lysotracker and LBPA-labeled intracellular vesicles co-localization normalized to LBPA numbers. 5 mM of NAC and 100 μM H₂O₂ were used. n_(−DOX) = 86, n_(+DOX) = 102, n_(+DOX+NAC) = 129, and n_{(−DOX+H2O2)} = 98. (E) Quantification of the percentage of lysotracker and LBPA-labeled intracellular vesicles co-localization normalized to lysotracker number. 5 mM of NAC and 100 μM H₂O₂ were used. n_(−DOX) = 86, n_(+DOX) = 102, n_(+DOX+NAC) = 129, and n_{(−DOX+H2O2)} = 98. (F) Representative image depicting method for quantifying LPBA distance from the nucleus center. Cells stained for LBPA (green) and DNA (cyan) are shown. Yellow arrows depict distance measured, d. Scale bar, 10 μm. (G) Quantification of the average LBPA-nucleus center distance per cell. n_(−DOX) = 62, n_(+DOX) = 68, n_(+DOX+NAC) = 61, and n_{(−DOX+H2O2)} = 57. (H) Quantification of all LBPA-nucleus center distance. For all graphics, error bars represent mean ± SD from three independent experiments. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗∗p < 0.0001, and n.s. (p > 0.05). For all graphs, a one-way ANOVA with a Kruskal-Wallis post hoc test was applied. See also Figure S4.

Figure 5

_SEVs secreted by PDAC cells with amplified centrosomes activate pancreatic stellate cells (A) Experimental flowchart. (B) Representative confocal images of PSCs stained for αSMA (green) and DNA (cyan). Scale bar, 10 μm. (C) Quantification of the percentage of PSCs activated upon treatment with _SEVs collected by UC from PaTu-S.iPLK4 (left) and HPAF-II.iPLK4 (right), with (+DOX) and without (−DOX) extra centrosomes. PaTu-S.iPLK4 isolated _SEV: PSCs n_{(−DOX SEVs)} = 398; n_{(+DOX SEVs)} = 373; and n_(ctr) = 475. HPAF-II.iPLK4 isolated _SEV: PSCs n_{(−DOX SEVs)} = 914; n_{(+DOX SEVs)} = 1,057; and n_(ctr) = 718. (D) Representative confocal image of PSCs incubated with _SEVs. Cells were stained for f-actin (phalloidin, gray) and DNA (cyan). Isolated _SEVs were labeled with BODIPY (green). Inset depicts higher magnification of sEVs associated with PSCs. Scale bar, 10 μm. (E) Representative images of cells acquired with the ImageStream. Cells (gray, bright field) and internalized _SEV labeled with CellVue (red) are shown. (F) Quantification of the percentage of PS1 cells positive for CellVue labeling. n_(unstained) = 6,280, n_{(cond. medium)} = 6,417, n_{(−DOX sEVs)} = 7,066, and n_{(+DOX sEVs)} = 7,230. (G) Quantification of the percentage of PSCs activated upon treatment with _SEVs collected by UC followed by SEC from PaTu-S.iPLK4 (left) and HPAF-II.iPLK4 (right), with (+DOX) and without (−DOX) extra centrosomes. PaTu-S.iPLK4 isolated _SEV: PSCs n_{(−DOX SEVs SEC7)} = 161; n_{(+DOX SEVs SEC7)} = 154; PSCs n_{(−DOX SEVs SEC8)} = 490; n_{(+DOX SEVs SEC8)} = 387; PSCs n_{(−DOX SEVs SEC9)} = 463; and n_{(+DOX SEVs SEC7)} = 454. HPAF-II.iPLK4 isolated _SEV: PSCs n_{(−DOX SEVs SEC7)} = 499; n_{(+DOX SEVs SEC7)} = 410; PSCs n_{(−DOX SEVs SEC8)} = 541; n_{(+DOX SEVs SEC8)} = 713; PSCs n_{(−DOX SEVs SEC9)} = 1,035; and n_{(+DOX SEVs SEC7)} = 914. For all graphics, error bars represent mean ± SD from three independent experiments. ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001, and n.s. (p > 0.05). For all, graphs were analyzed using two-way ANOVA with Tukey’s post hoc test. See also Figure S5.

Figure 6

_SEVs secreted by cells with extra centrosomes can promote PDAC invasion (A) Experimental flowchart. (B) Representative bright-field images of heterotypic spheroids. Black arrows: invasive protrusions. Scale bar, 100 μm. (C) Quantification of the percentage of invasion in 3D spheroids. 5 ng/mL TGF-β was used as positive control. Spheroids n_(+PSCs) = 40, n_{(+PSCs TGF-β)} = 34, n_{(+PSCs −DOX SEVs)} = 31, and n_{(+PSCs +DOX SEVs)} = 31. (D) Confocal images of spheroids composed of cancer cells (expressing H₂B-RFP; magenta) and PSCs (expressing H₂B-GFP; green). Scale bar, 100 μm. Inset depicts higher magnification of invasive protrusion. Scale bar, 20 μm. For all graphics, error bars represent mean ± SD from three independent experiments. ^∗∗∗∗p < 0.0001, n.s. (p > 0.05). Graph was analyzed using one-way ANOVA with a Kruskal-Wallis post hoc test.

Figure 7

Schematic representation of working model Increased ROS levels in cells with extra centrosomes compromise lysosomal function. We propose that this changes MVBs fate toward fusing with the plasma membrane, resulting in increased secretion of _SEVs. _SEVs secreted by cancer cells with extra centrosomes are functionally distinct and can induce PSCs activation to promote cell invasion.