Selective packaging of mitochondrial proteins into extracellular vesicles prevents the release of mitochondrial DAMPs

Abstract

Most cells constitutively secrete mitochondrial DNA and proteins in extracellular vesicles (EVs). While EVs are small vesicles that transfer material between cells, Mitochondria-Derived Vesicles (MDVs) carry material specifically between mitochondria and other organelles. Mitochondrial content can enhance inflammation under pro-inflammatory conditions, though its role in the absence of inflammation remains elusive. Here, we demonstrate that cells actively prevent the packaging of pro-inflammatory, oxidized mitochondrial proteins that would act as damage-associated molecular patterns (DAMPs) into EVs. Importantly, we find that the distinction between material to be included into EVs and damaged mitochondrial content to be excluded is dependent on selective targeting to one of two distinct MDV pathways. We show that Optic Atrophy 1 (OPA1) and sorting nexin 9 (Snx9)-dependent MDVs are required to target mitochondrial proteins to EVs, while the Parkinson's disease-related protein Parkin blocks this process by directing damaged mitochondrial content to lysosomes. Our results provide insight into the interplay between mitochondrial quality control mechanisms and mitochondria-driven immune responses.

Fig. 1. Mitochondria and EVs activate distinct pro-inflammatory cytokines.

a, b Extracellular mitochondria induce the secretion of pro-inflammatory cytokines. Mitochondria (Mito)(isolated from MEFs) or EVs (isolated by differential centrifugation from the media of MEFs grown for 24 h in EV-depleted media) were added to RAW cells in their culture media. The release of IL6 (a) and IP10 (b) into culture media was measured by ELISA 24 h. Individual points represent independent experiments (a, n = 3; b, n = 4). Bars show the average ± SD. One-way ANOVA. c Representative western blot showing the amount of the specified mitochondrial proteins in the indicated cell types (20 µg) vs their EVs (5 µg). TOM20, outer membrane protein; NDUFA9, Complex I subunit; mtHSP70 matrix protein associated with the IM; PDH, matrix protein. Actin is used as a control, as it has been shown to associate with EVs. d AA-treated MEFs mitochondria stimulate IP10 production. RAW cells were treated as above with EVs (from 10 × 10⁶ cells) and mitochondria (12 µg) isolated from Control or AA-treated MEFs, and IP10 release in culture media measured by ELISA. Individual points represent independent experiments (mitochondria, n = 4; EVs, n = 6). Bars show the average ± SD. One-way ANOVA. e AA treatment of WT MEFs causes the selective degradation of mitochondrial proteins. MEFs were treated with AA for 24 h and the indicated mitochondrial proteins analysed by western blot. f Enrichment of the indicated mitochondrial proteins in EVs was measured by western blot in Control (blue) and AA-treated (24 h, Red) WT MEFs. Individual points represent independent experiments (n = 4). Bars show the average ± SD. Two-sided t-test).

Fig. 2. Selective inclusion of mitochondrial proteins in EVs.

a EV markers were analysed by western blot in 20 µg cell extracts and 5 µg MEFs EVs isolated as in Fig. 1. b Quantification of protein inclusion in MEF EVs. The amount of the indicated protein present in EVs was normalised to its cellular content. Individual points represent independent experiments. Bars show the average ± SD. Alix, Exosome marker (blue); Lamin B1, nuclear protein; Rab9, Rab11, Snx9, endosomal proteins; Cyt c, IMS protein, OPA1, IM protein. Mitochondrial proteins (green), non-mitochondrial proteins (red), mtDNA (purple). c MEF EVs isolated as above were treated with Trypsin in the absence or the presence of detergent (TX100) and analysed by western blot for the presence of the indicated proteins. d–f MEF EV ultrastructure was analysed by EM and quantified based on structure (e) and size (f). Representative images are shown in (d). Scale bars, 200 nm. g, h Vesicles containing selective mitochondrial cargo are found in proximity to the plasma membrane (≤1 µm) but away from the main mitochondrial network (>1 µm). The number of TOM20-positive, mtHSP70-positive or PDH-positive vesicles are quantified in (g) with individual points representing the fraction of cells with plasma membrane-associated vesicles in four independent experiments. Each positive cell typically contained one vesicle. Points show independent experiments and bars show the average ± SD. A representative image is shown in (h) with GFP (blue) used as a cytosolic marker to identify cellular boundaries. The arrowhead denotes a TOM20-positive vesicle close to the plasma membrane. Scale bar, 2 µm.

Fig. 3. Snx9-dependent MDVs contribute to the inclusion of IM/matrix proteins into EVs.

a Schematic representation of the two distinct MDV pathways leading to EVs and lysosomes, respectively. b MEFs were treated for 24 h with a control siRNA (siCtrl, blue) or a siRNA against Snx9 (siSnx9, red). Mitochondrial proteins were then measured by western blot, with Actin serving as a loading control. c Representative images showing TOM20 and mtHSP70 MDVs in MEFs treated with a control siRNA (siCtrl) or a siRNA against Snx9 (siSnx9). Arrowheads denote MDVs (positive for one marker but negative for the other). Scale bars: top panels, 10 µm; enlargements in bottom panels, 2 µm. d MDV quantification from images as in (c). Each data point represents one cell. Bars represent the average of 40 cells in three independent experiments ± SD. Two-sided t-test. e Enrichment of the indicated mitochondrial proteins was measured by western blot as in Fig. 1. Individual points represent independent experiments (n = 4). Bars are shown as average ± SD. Two-sided t-test.

Fig. 4. OPA1 deletion inhibits the formation of IM/matrix MDVs.

Quantification of MDVs positive for PDH (a), mtHSP70 (b), NDUFA9 (TOM20-negative) (c), as well as TOM20-positive (mtHSP70-negative) (d) in immunofluorescence images of WT (blue) and OPA1 KO (red) MEFs. Each data point represents one cell. Bars represent the average of 23 cells (except for PDH where n = 20) in three independent experiments ± SD. Two-sided t-test.

Fig. 5. OPA1 regulates the inclusion of IM/matrix proteins and mtDNA into EVs.

a Quantification of EV protein yields relative to cellular protein content in WT and OPA1 KO EVs isolated as in Fig. 1. Each point represents one experiment (n = 7). Bars show the average ± SD. b OPA1 deletion does not alter overall EV size. WT and OPA1 KO EV size was analyzed by EM and binned as indicated. The average percent of EVs in each category ± SD is shown (n = 3). No EVs larger than 800 nm were detected. c, d The inclusion of cytoplasmic proteins into EVs is not altered following OPA1 deletion. Representative western blot (c) and quantification (d) as in Fig. 1. Individual points represent independent experiments (n = 3). Bars show the average ± SD. e–g The inclusion of IM/matrix proteins, but not TOM20, into EVs is prevented by OPA1 deletion. Representative western (e) and quantification as in Fig. 1 (f). mtDNA was quantified by PCR (g) Individual points represent independent experiments. Bars show the average ± SD. Two-sided t-test. WT (blue), OPA1 KO (red).

Fig. 6. OPA1 KO mitochondria but not EVs isolated from the same cells induce an IP10 inflammatory response.

a, b RAW cells were treated as in Fig. 1 with mitochondria (Mito, 12 µg) (a) or EVs (from 10 × 10⁶ cells) (b) isolated from WT or OPA1 KO MEFs and IP10 release in culture media measured by ELISA. Individual points represent actual concentrations (or normalised to WT for (b), right panel) in independent experiments (Mito, n = 6; EVs, n = 4). Bars show the average ± SD. One-way ANOVA, two-sided t-test for (b), right panel. c OPA1 KO EVs show reduced expression of IFN-dependent genes. RAW cells were treated as in (a–b) and mRNA levels of the IFN-dependent genes Rsad2 (left) and mIfit1 (right) measured by qPCR. Individual points represent independent experiments (Rsad2, n = 6; mIfit1, n = 4). Bars show the average ± SD. One-way ANOVA. d, e RAW cells were treated as in (a, b) and IL6 was measured in the culture media by ELISA. Individual points represent independent experiments (Mito, n = 4, EVs, n = 5). Bars show the average ± SD. One-way ANOVA.

Fig. 7. Parkin activation inhibits the inclusion of mitochondrial proteins within EVs.

a AA treatment induces MDV formation. WT MEFs grown on coverslips were incubated in the absence (blue) or presence of AA for 1 h (red). mtHSP70 and TOM20 MDVs were then quantified. Each data point represents one cell. Bars represent the average of at least 30 cells in a minimum of three independent experiments ± SD. Two-sided t-test. b, c Damage-induced MDVs are targeted to lysosomes. b WT MEFs were treated as in (a) in the presence of E64/Pepstatin and MDV associaton with the lysosomal marker LAMP1 quantified as the fraction of total MDVs associated with LAMP1. Alternatively (c) LAMP1-positive MDVs were quantified from OPA1 KO MEFs. Each data point represents one cell. Bars represent the average of at least 20 cells in a minimum of three independent experiments ± SD. Two-sided t-test). d Stable expression of GFP-Parkin increase MDV formation. U2OS cells stably expressing GFP (blue) or GFP-Parkin (red) were treated with AA for 4 h and MDVs quantified. Each data point represents one cell. Bars represent the average of at least 20 cells in a minimum of three independent experiments ±  SD one-way ANOVA. e mtHSP70 MDVs association with Parkin. U2OS cells stably transfected with GFP-Parkin were treated as in (d) and the association of mtHSP70 MDVs quantified by immunofluorescence. Individual points represent independent experiments (n = 4). Bars show the average percent of the total (blue) and Parkin-positive (green) mtHSP70 MDVs ± SD. Two-sided t-test. f GFP-Parkin expression impairs the inclusion of IM/matrix proteins into EVs. EVs were isolated from U2OS cells stably expressing GFP (blue) or GFP-Parkin (red) and the indicated mitochondrial proteins measured as in Fig. 1. Individual points represent independent experiments (n = 3, except for TOM20 where n = 4). Bars show the average ±SD. Two-sided t-test.