Proteomic comparison defines novel markers to characterize heterogeneous populations of extracellular vesicle subtypes

Abstract

Extracellular vesicles (EVs) have become the focus of rising interest because of their numerous functions in physiology and pathology. Cells release heterogeneous vesicles of different sizes and intracellular origins, including small EVs formed inside endosomal compartments (i.e., exosomes) and EVs of various sizes budding from the plasma membrane. Specific markers for the analysis and isolation of different EV populations are missing, imposing important limitations to understanding EV functions. Here, EVs from human dendritic cells were first separated by their sedimentation speed, and then either by their behavior upon upward floatation into iodixanol gradients or by immuno-isolation. Extensive quantitative proteomic analysis allowing comparison of the isolated populations showed that several classically used exosome markers, like major histocompatibility complex, flotillin, and heat-shock 70-kDa proteins, are similarly present in all EVs. We identified proteins specifically enriched in small EVs, and define a set of five protein categories displaying different relative abundance in distinct EV populations. We demonstrate the presence of exosomal and nonexosomal subpopulations within small EVs, and propose their differential separation by immuno-isolation using either CD63, CD81, or CD9. Our work thus provides guidelines to define subtypes of EVs for future functional studies.

Keywords: dendritic cells; ectosomes; exosomes; extracellular vesicles; microvesicles.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

DCs secrete heterogeneous EVs recovered in successive differential ultracentrifugation pellets. (A) DC stained with FM 4-64 FX dye and visualized with confocal microscopy. (Magnification, 100×.) Arrows point at vesicles budding at the cell surface. (B) Scheme of EV isolation by DUC from conditioned medium of human monocyte-derived DCs. (C) Protein content of each pellet (2K = 2,000 × g; 10K = 10,000 × g; 100K = 100,000 × g) is represented for individual donors (one symbol per donor), *P < 0.05; **P < 0.01 (Wilcoxon signed-rank test). (D and E) Whole-mount EM analysis of each pellet showing representative images (E) and size distribution (D) of the vesicles. Diameter of membrane-enclosed structures was determined with iTEM software, for two independent donors (2K n = 35 EVs, 10K n = 72, 100K n = 135). (Scale bars, 200 nm.) (F) The successive pellets (2K, 10K, 100K) of EVs isolated from DCs were analyzed by Western blot side-by-side with the lysate of producing cells (CL), using antibodies to proteins often considered as “exosome-markers.” Equivalent protein amounts of each sample were loaded on the gels. Representative images and quantifications in three to nine individual donors (Fig. S1C) are shown. As expected for sEV-markers, tetraspanins (CD63, CD9, CD81) are enriched in 100K, but CD9 and CD63 are also abundant in the 10K and 2K pellets respectively. Other putative sEVs markers (MHC II, HSC70, flotillin-1, actin) are ubiquitously present in the three pellets. The ER-retained protein GP96 is mainly found in cells and the material pelleting at the lowest speed, 2K.

Fig. S1.

(A) Protein content recovered in each pellet of DCs (micrograms per 106 secreting cells, y axis) as a function of the percentage of dead cells in the culture (x axis) does not show any correlation between the two parameters. Each symbol represents a unique donor. (B) 10K and 100K pellets were analyzed by Nanoparticle Tracking Analysis (NTA). (Upper) Representative example of NTA analysis: each line corresponds to one acquired video from a single sample. A wide range of particle sizes is observed in the 10K pellet, whereas the size range of particles in the 100K pellets is more restricted. (Lower) Particle number (Right) and mean size (Left), calculated from the mean of at least five videos per donor, are represented. Each symbol represents a unique donor. (C) Quantifications of the distribution of the different proteins in each successive centrifugation pellet, analyzed by Western blot (Fig. 1F) in three to nine independent experiments. For each pellet, AU = (SIpellet)/Sum(SI2K + SI10K + SI100K) (i.e., ratio of signal intensity in the given pellet to the total secreted protein). Each symbol represents a unique donor. *P < 0.05; **P < 0.01 (Wilcoxon signed-rank test). (D) Presence of HSP70 and MHC class I molecules in the pellets recovered at different centrifugation steps from DC-conditioned medium. Like MHC class II, HSC70, and flotillin-1 (Fig. 1F), these molecules are detected in all pellets, although at variable levels.

Fig. 2.

Floatation on iodixanol gradient separates four subfractions of EVs. (A) Pellets obtained after 10K and 100K centrifugations were allowed to float into an overlayed iodixanol gradient. (B) Ten fractions were collected and analyzed by Western blot (representative of nine experiments), showing the separation of two discrete fractions (F3 and F5+F6) of EVs from both the 10K and 100K pellets (Left). Densities of recovered fractions, as measured by refractometry, are displayed in the graph (mean ± SD of 36 independent gradients). (Right) Quantification of CD63 relative abundance in fractions F3–F5 of both pellets obtained from nine individual donors was performed as in Fig. S1C. Arbitrary Units (AU) = (SIfraction)/Sum(SIF3 + SIF4 + SIF5) where SI = signal intensity. Each symbol represents a single donor. **P < 0.01 (Wilcoxon signed-rank test). (C) Representative images of whole-mount EM of fractions three and five of 10K and 100K (Right), and size distribution of vesicles measured by ImageJ software (Left: F3 10K n = 98, F5 10K n = 99, F3100K n = 226, F5 100K n = 166, representative of two independent gradients). (Scale bar, 400 nm.) (D) Separation of 10K and 100K pellets on sucrose gradients analyzed by Western blot shows continuous distribution between 1.12 and 1.19 g/mL for both pellets, and a tendency for 10K to achieve equilibrium in denser fractions (1.17 and above) than 100K (1.15 + 1.17).

Fig. 3.

Qualitative and quantitative proteomic analyses of iodixanol fractions F3 and F5 from 10K and 100K pellets by LC-MS/MS suggest different intracellular origins of the four types of EVs, and identify potential specific proteins. (A) Venn diagram showing the distribution of proteins qualitatively identified in each fraction by at least three peptides in one of the three biological replicates. GO terms of protein families specifically enriched in a single fraction (or in two fractions for ribosomes), as determined by DAVID software, are shown. (B–D) Quantitative analysis of the amount of proteins in each fraction compared with F3-100K was performed (proteins displaying missing values among fractions were excluded from this analysis). (B) PCA analysis of the quantitative comparison shows a clear separation of fraction F3-100K from the three others. (C and D) Quantitative analysis of proteins present in F3-10K (C) or F5-100K (D) compared with F3-100K is shown as Volcano plot. x axis = log2(fold-change) (10K/100K), y axis = −log10(P value). The horizontal red line indicates P value = 0.05, vertical green lines indicate absolute fold-change = 2. Data represent results of three independent sets of donors pooled together. Position of proteins selected as potential specific markers of each fraction and analyzed further is shown.

Fig. S2.

Quantitative GO-term enrichment analysis was performed using myProMS Software for fractions F3-100K vs. F5-100K (Left) and F3-100K vs. F3-10K (Right). Red color highlights enriched families of proteins. Proteins equally present in both fractions are in the middle column: “cytoplasm” and “intracellular nonmembrane-bounded organelles” are enriched GO-terms in proteins present in all fractions. Proteins two to four times or above four times more abundant in F3-100K are displayed on the left hand-side of each diagram: “endosome membrane,” “plasma membrane,” and “intrinsic to plasma membrane” GO-terms are enriched in F3-100K. Finally, proteins enriched in F3-10K or F5-100K are displayed on the right hand-side of each diagram: the “mitochondrial part” GO term is enriched in both F3-10K and F5-100K.

Fig. 4.

Confirmation of selected proteins as specific markers of distinct vesicles subpopulation. (A) Western blotting of iodixanol fractions of 10K and 100K confirms the unique presence of TSG101 and syntenin-1 in fraction F3-100K. MW, molecular weight markers. (B and C) Western blot analysis, performed as in Fig. 1F, of the distribution, among the successive pellets of differential centrifugation (2K, 10K, 100K), of selected proteins identified by the quantitative proteomic analysis. Representative images are shown and, quantifications in two to five experiments are displayed in Fig. S3A. As expected from their enrichment in F3-100K, syntenin-1, TSG101, ADAM10, EHD4, and Annexin XI are strongly enriched in 100K and almost absent in the other pellets (B). Conversely, as expected from their enrichment in F3-10K, actinin-4, and mitofilin are enriched in 10K compared with 100K, but they are also present in 2K (C). Proteins which were less than twofold enriched in either fraction, such as LAMP2 or Annexin II did not show, upon Western blotting, differential expression in 10K vs. 100K pellets (C). (D) Western blot analysis of the secretion of various proteins in the different centrifugation pellets recovered from conditioned medium of tumor cell lines OV2008, MDA-MB-231 (in control or serum-starved conditions), compared with DCs. The 2K, 10K, and 100K pellets obtained from the same number of cells were loaded on the gel, side-by-side with CL from the indicated number of cells. Presence of syntenin-1 and ADAM10 and enrichment of CD9/CD63/CD81 was confirmed among all cell lines in 100K pellets. Presence of CD9, CD63, actinin-4 and HSC70 was confirmed in the 2K pellets of cell lines secreting large EVs [i.e., OV2008 and MDA-MB-231 under starvation (no serum) conditions].

Fig. S3.

(A) Quantifications of the distribution of the different proteins in each successive centrifugation pellet, analyzed by Western blot (Fig. 4 B and C) in three to five independent experiments. For each pellet, AU = (SIpellet)/Sum(SI2K + SI10K + SI100K); that is, ratio in the given pellet to the total secreted protein. Each symbol represents a unique donor *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (paired Student’s t test). (B) Western blot analysis of the secretion of various proteins in the different centrifugation pellets recovered from conditioned medium of tumor cell lines (IGROV-1, SHIN-3, HeLa-CIITA), or immortalized nontumor cells (HEK293T, RPE-1). The 2K, 10K, and 100K pellets obtained from the same number of cells (25 × 106) were loaded on the gel, side-by-side with CL from the indicated number of cells. Three of five cell lines secrete mainly small vesicles pelleting at 100K with only traces of signals in 2K pellets. Presence of syntenin-1 was confirmed among all cell lines in 100K pellets, whereas ADAM10 is not expressed by all cells. (C) SYPRO orange staining of total protein content in SDS-Page gel of EV recovered from MDA-MB-231 cultured with or without serum (= starvation conditions) before EV collection (Upper) reveals secretion of large EVs pelleting at 2K in starvation, but not in control conditions, without changes in metabolism analyzed by CellTiter-Blue assay (Promega) (Lower). (D) Western blot analysis of different proteins in pellets recovered from mouse bone marrow-derived DCs. Bone marrow cells from C57BL/6 Rab27a-flox/WT × CD11c-Cre mice (Rab27a-competent) were cultured for 8 d in IMDM supplemented with 10% FCS, GM-CSF (20 ng/mL), 50 μM 2-ME, 100 IU/mL penicillin and 100 μg/mL streptomycin (Gibco). Fresh medium depleted of FCS-derived EVs was fed to the cells 24 h before conditioned medium collection and EV isolation by DUC. EVs from 5 to 10 × 106 cells were loaded in each lane. Antibodies used to detect mouse proteins are: anti-CD9 (rat, KMC8, BD Biosciences), anti-CD63 (mouse, R5G2, MBL), anti-MHC II (rabbit polyclonal raised against the cytosolic domain of IA-α chain) (55), anti-TSG101 (rabbit, T5951, Sigma), anti-syntenin-1 (rabbit, GTX108470, Genetex), or same antibodies as for human samples (HSC70, GP96, and actinin-4).

Fig. 5.

Qualitative and quantitative proteomic analyses of sEVs immuno-isolated by CD9, CD63, or CD81-specific antibodies evidence additional sEV subpopulations. (A) The crude DC-derived 100K pellet was subjected to parallel immuno-isolation with beads coupled to irrelevant murine IgG, or antibodies against CD9, CD63, or CD81. PD vesicles and nonpulled down materials remaining in the FT were subjected to subsequent comparative analysis. (B) Equal volumes of materials from each PD and FT (the latter after concentration by ultracentrifugation) were loaded on a gel for Western blot analysis with antibodies specific for CD9, CD63, CD81, or MHC class II. All beads precipitated efficiently vesicles bearing the targeted protein. Note the remaining presence of CD9+ and CD81+ materials in the FT of CD63-beads, and of MHC II+ materials in the FT of all immuno-isolations, showing the existence of tetraspanin-negative sEVs. *Nonspecific signal from the immunoglobulins’ heavy (50 kDa, CD63 blot) or light (25 kDa, CD9 and CD81 blots) chains used for immuno-precipitation. (C) PD and FT obtained as shown in A and B were analyzed by label-free LC-MS/MS. Venn diagrams represent the number of proteins detected in each sample with minimum three peptides in each of three independent replicates, comparing PD and FT obtained from each antibody after exclusion of the 61 proteins present in the PD of irrelevant IgG. (D) Quantitative PCA shows that the CD63-PD are distinct from the CD81- and CD9-PD. (E) Venn diagram showing proteins identified in PD obtained from the three antitetraspanin antibodies (Upper), and distribution of the 241 proteins common to the CD9-, CD63-, and CD81-PD, compared with the FT of CD9 (Lower). Position of sEV-specific proteins is indicated. (F) Assignment of the proteins analyzed here, and previously described as canonical exosome markers, to the different types of EVs, as demonstrated by Western blotting (bold) or by the quantitative proteomic comparison. EVs are schematized as a lipid bilayer (thick brown circle) enclosing cytosol (light background). Brown: proteins shared by several types of EVs. Green: proteins specifically enriched in F3-100K (i.e., the light sEVs), including those specific of the tetraspanin-enriched endosome-derived exosomes (italic font), and those ubiquitously present in all sEVs. Gray: proteins coisolated with the small EVs (100K pellets) but in EVs of higher density (F5-100K). Blue: proteins specifically enriched in large and medium-sized EVs.