Heterogeneity in extracellular vesicle secretion by single human macrophages revealed by super-resolution microscopy

Abstract

The diverse origins, nanometre-scale and invasive isolation procedures associated with extracellular vesicles (EVs) mean they are usually studied in bulk and disconnected from their parental cell. Here, we used super-resolution microscopy to directly compare EVs secreted by individual human monocyte-derived macrophages (MDMs). MDMs were differentiated to be M0-, M1- or M2-like, with all three secreting EVs at similar densities following activation. However, M0-like cells secreted larger EVs than M1- and M2-like macrophages. Proteomic analysis revealed variations in the contents of differently sized EVs as well as between EVs secreted by different MDM phenotypes. Super resolution microscopy of single-cell secretions identified that the class II MHC protein, HLA-DR, was expressed on ∼40% of EVs secreted from M1-like MDMs, which was double the frequency observed for M0-like and M2-like EVs. Strikingly, human macrophages, isolated from the resected lungs of cancer patients, secreted EVs that expressed HLA-DR at double the frequency and with greater intensity than M1-like EVs. Quantitative analysis of single-cell EV profiles from all four macrophage phenotypes revealed distinct secretion types, five of which were consistent across multiple sample cohorts. A sub-population of M1-like MDMs secreted EVs similar to lung macrophages, suggesting an expansion or recruitment of cells with a specific EV secretion profile within the lungs of cancer patients. Thus, quantitative analysis of EV heterogeneity can be used for single cell profiling and to reveal novel macrophage biology.

Keywords: class II MHC protein; extracellular vesicles; macrophages; single-cell analysis; super-resolution microscopy.

FIGURE 1

CD81 is enriched in EVs secreted at the surface membrane of activated macrophages. (a,b) Macrophages were incubated on glass slides coated with PLL and indicated concentrations of IgG from human serum for 20 h and supernatants were analysed for TNFα (a) and IL‐10 (b) by ELISA. n = 3 individual donors and experiments, mean ± SD. (c,d) Macrophages were incubated for 15 min on glass slides coated with 0.01% PLL only (0 μg/ml IgG) or 0.01% PLL and IgG (10 μg/ml IgG) and fixed. (c) Representative interference reflection microscopy (IRM) images (Scale bars; 10 μm). (d) Cell area distributions. Representative of three individual donors and experiments. (e–h) Macrophages were incubated on glass slides coated with 0.01% PLL or 0.01% PLL + 10 μg/ml IgG from human serum for 15 min, fixed, blocked and stained with anti‐CD81‐AF647 mAb, then phagocytic synapses were imaged using STORM. (e) Representative images, with and without activation are shown for M0‐, M1‐ and M2‐like macrophages (Scale bar; 10 μm, White box; STORM Zoom). (f) Density of EVs secreted from macrophages. (g) Relative frequency of EV diameter (10 nm binning). (h) Distribution of EV diameters. n = 3 individual donors and experiments. *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001. Statistical significance assessed by Mann‐Whitney test (A,B) or Kruskal‐Wallis test (D, F–H)

FIGURE 2

Macrophages activated on lipid bilayers secrete CD81‐enriched EV. M0‐, M1‐ and M2‐like macrophages were plated on planar lipid bilayers containing 10 μg/ml IgG for 20 min, fixed, blocked and stained with anti‐CD81‐AF647 mAb before imaging by STORM. (a) Representative brightfield, TIRF, STORM and Zoom images (Scale bar; 5 μm, Zoom; 1 μm, Yellow box; TIRF crop for STORM, White box; STORM Zoom). (b) Representative zoom images of EVs with line profiles. Scale bar; 100 nm. (c–e) Quantitative analysis of EVs including density, defined as the number of EVs per μm² (c), a histogram of detected diameters normalised to the mode of each histogram (d) and violin plots of the detected EV diameters (e). n = 3 individual donors and experiments. **, p ≤ 0.01; ****, p ≤ 0.0001; Statistical significance assessed by Kruskal‐Wallis test

FIGURE 3

Macrophage secreted EVs are retained in situ after cellular removal. (a) Macrophages were incubated on IgG‐containing lipid bilayers for 20 min, pulse stained with streptavidin‐AF488, detached, fixed and stained with anti‐CD81 mAb conjugated to AF647. (b) Representative Brightfield images, images of the bilayer stained with AF488 showing shadows where cells have been (Shadow‐AF488), TIRF and STORM images of CD81 stained with AF647 (TIRF CD81‐647 and STORM CD81‐647) and zoomed‐in regions of the STORM image (Zoom; 3 × 3 μm). Scale bars: Brightfield and TIRF; 10 μm, STORM; 5 μm, Zoom; 1 μm. Yellow boxes indicate TIRF crop for STORM. White boxes indicate STORM Zoom. C‐E) Quantitative analysis of EVs including the vesicle density (number of vesicles per μm²) (c), a normalised diameter histogram (d) and violin plots of the detected diameters (e). (f) Diameters of individual EVs from each cell analysed. One dot represents one cell or one EV, n = 4 individual donors and experiments. Mean ± SD. **, p ≤ 0.01; ****, p ≤ 0.0001; Statistical significance assessed by Kruskal‐Wallis test

FIGURE 4

Extracellular vesicles colocalise with holes in the actin mesh and can be inhibited by cambinol. (a–d) Macrophages were treated for 120 min with either 2 μM Cambinol (+) or DMSO (−), incubated for 15 min on glass slides coated with 0.01% PLL and 10 μg/ml IgG, fixed, blocked and stained with anti‐CD81‐AF647 mAb. (a) Area of spread cells. (b) Representative STORM and Zoom (5 × 5 μm) images (Scale bar; 5 μm, Zoom; 1 μm, White boxes; STORM Zoom). (c) Vesicle density defined as the number of vesicles per μm². (d) Vesicle density relative to control condition from matched donors. Open dots represent single cells and closed dots represent mean donor values. n = 4 individual donors and experiments, mean ± SD. (e–h) Macrophages were plated on glass slides coated with 0.01% PLL only (0 μg/ml IgG) or 0.01% PLL and IgG (10 μg/ml IgG) for 15 min, fixed, permeabilised, blocked and stained with anti‐CD81‐AF647 mAb and phalloidin‐AF488. The actin mesh was imaged with STED and CD81 using confocal microscopy. Shown are representative images and composites with two indicated zoom regions (3 × 3 μm) with matched line profiles for M0‐like (e), M1‐like (f) and M2‐like (g) macrophages (Scale bars; 10 μm, Zoom; 1 μm, White boxes; Zoomed region). (h) Pearson correlation co‐efficient for staining of actin and CD81. n = 3 individual donors and experiments. Each dot represents one cell. Mean ± SD. *, p ≤ 0.05; ***, p ≤ 0.001; ****, p ≤ 0.0001; Statistical significance assessed by unpaired t test (A, C, H)

FIGURE 5

Proteomic analysis of extracellular vesicles reveals distinct contents for different macrophages. (a) Monocyte‐derived macrophages were stimulated in solution with IgG‐coated beads for 1 h to stimulate EV secretion. Large and small EVs were isolated via differential ultracentrifugation, digested, analysed by mass spectrometry and label free quantification performed. (b,c) Proportional Venn diagrams showing the number of shared proteins between the different macrophage types (b) or different vesicle types (c). (d) Principal component analysis of all identified proteins in each sample. (e) Heatmap of selected significantly regulated pathways, identified using a weighted query in G:Profiler, from large and small EV, colour coded for their adjusted p‐value as indicated. (f–i) Heatmaps of all significantly regulated proteins with tables highlighting selected proteins for large EVs (f–g) and small EVs (h–i). n = 3 individual donors and experiments for all panels

FIGURE 6

EVs secreted by M1‐like macrophages display higher levels of MHC class II protein. (a) Relative expression of HLA‐DRα and HLA‐DRB1*15 across macrophage subtypes in small and large EVs as measured by proteomic analysis. n = 3 individual donors and experiments, mean ± SD. (b–c) Differentiated macrophages were assessed by flow cytometry for cell surface expression of HLA‐DRα, with the relative gMFI indicated (b) and percentage of positive cells (c). n = 3 individual donors and experiments, mean ± SD. D‐H) M0‐, M1‐ and M2‐like macrophages were incubated on planar lipid bilayers containing IgG from human serum for 20 min, detached, fixed, blocked and stained with anti‐CD9‐AF488 mAb, anti‐CD63‐AF488 mAb, anti‐CD81‐AF488 mAb and anti‐HLA‐DR‐AF647 mAb. (d) Representative images of secreted EVs visualised using STORM (Scale bars; 5 μm, Zoom; 1 μm, Yellow boxes; TIRF crop for STORM, white boxes; STORM Zoom). (e) Density of vesicles secreted from each cell analysed. (f) Percentage of HLA‐DRα‐positive vesicles per individual cell. (g) Mean pixel fluorescence intensity of HLA‐DRα on individual EV, summarised for each cell. (h) Box and whisker plots displaying the mean pixel intensity of HLA‐DR on each individual vesicle, separated as profiles for individual cells with median and quartiles indicated. (e–g) Open symbols represent single cells, with filled symbols showing each donor's mean, n = 5 individual donors and experiments, mean ± SD. **, p ≤ 0.01; Statistical significance assessed by one‐way ANOVA (E–G)

FIGURE 7

Human lung macrophages secrete EVs upon activation through their FcγRI. Lung macrophages were isolated from human lung resections and selected for by adherence. (a) Representative widefield images of lung macrophages (Scale bars; 100 μm, Zoom; 30 μm, Yellow box; Zoomed region). (b–d) Macrophages were incubated on glass slides coated with 0.01% PLL and indicated concentrations of IgG from human serum for 20 h and supernatants were analysed for TNFα (b) and IL‐10 (c) by ELISA. n = 4 individual donors and experiments, mean ± SD. (d) Representative IRM images and corresponding cell contact area quantification. (e–i) Macrophages were incubated on IgG‐containing lipid bilayers for 20 min, pulse stained with streptavidin‐AF488, detached, fixed and stained with anti‐CD81‐AF647. n = 4 individual donors and experiments, mean ± SD. (e) Representative Brightfield, TIRF AF488 (Shadows), TIRF AF647, STORM and Zoom (3 × 3 μm) images (Scale bars: Brightfield and TIRF; 10 μm, STORM; 5 μm, Zoom; 1 μm, Yellow boxes; TIRF crop for STORM, white boxes; STORM Zoom). (f) Secreted EV density. (g) EV diameter violin plot. (h) Normalised EV diameter histograms. (i) Diameters of every EV for individual lung macrophages, displayed as box and whisker plots with median and quartiles indicated. **** **, p ≤ 0.0001; Statistical significance assessed by unpaired t test

FIGURE 8

Comparing the EV profiles of individual macrophages. (a–d) Lung macrophages were incubated on IgG containing planar lipid bilayers for 20 min, detached, blocked and stained with anti‐CD9‐AF488 mAb, anti‐CD63‐AF488 mAb, anti‐CD81‐AF488 mAb and anti‐HLA‐DR‐AF647 mAb. n = 4 individual donors and experiments, mean ± SD. (a) Representative STORM images of secreted EVs (Scale bars; 5 μm, Zoom; 1 μm, Yellow boxes; TIRF crop for STORM, White boxes; STORM Zoom). (b) Density of EVs secreted from each cell. (c) Percentage of secreted EVs expressing HLA‐DR per cell. (d) Mean fluorescence intensity of HLA‐DRα on individual EVs, summarised for each cell. (e) Histogram with EV diameter binned by 10 nm, showing the effect of EV size on HLA‐DR intensity and the percentage of HLA‐DR positive EVs, comparing different macrophage subtypes. (f) Principal components analysis resolving the heterogeneity of individual EVs for each individual cell with macrophage subtype indicated. (g) A heatmap showing individual cell expression of HLA‐DRα on EVs of different sizes with unbiased hierarchical clustering of individual cells and EV diameter bins (representative of 168 cells from four (lung) or five individual donors)