EV-Elute: A universal platform for the enrichment of functional surface marker-defined extracellular vesicle subpopulations

Abstract

Intercellular communication via extracellular vesicles (EVs) has been identified as a vital component of a steadily expanding number of physiological and pathological processes. To accommodate these roles, EVs have highly heterogeneous molecular compositions. Given that surface molecules on EVs determine their interactions with their environment, EV functionality likely differs between subpopulations with varying surface compositions. However, it has been technically challenging to examine such functional heterogeneity due to a lack of non-destructive methods to separate EV subpopulations based on their surface markers. Here, we used the Design-of-Experiments (DoE) methodology to optimize a protocol, which we name 'EV-Elute', to elute intact EVs from commercially available Protein G-coated magnetic beads. We captured EVs from various cell types on these beads using antibodies against CD9, CD63, CD81 and a custom-made protein binding phosphatidylserine (PS). When applying EV-Elute, over 70% of bound EVs could be recovered from the beads in a pH- and incubation-time-dependent fashion. EV subpopulations showed intact integrity by electron microscopy and Proteinase K protection assays and showed uptake patterns similar to whole EV isolates in co-cultures of peripheral blood mononuclear cells (PBMCs) and endothelial cells. However, in Cas9/sgRNA delivery assays, CD63⁺ EVs showed a lower capacity to functionally deliver cargo as compared to CD9⁺, CD81⁺ and PS⁺ EVs. Taken together, we developed a novel, easy-to-use platform to isolate and functionally compare surface marker-defined EV subpopulations. This platform does not require specialized equipment or reagents and is universally applicable to any capturing antibody and EV source. Hence, EV-Elute can open new opportunities to study EV functionality at the subpopulation level.

Keywords: CRISPR/Cas9; extracellular vesicle subpopulations; extracellular vesicles; heterogeneity; immunoprecipitation.

FIGURE 1

Systematic optimization of EV elution protocol from magnetic beads using DoE methodology. (a) Overview of experimental setup to test elution of EVs captured on Protein G‐coated magnetic beads and EV permeability after exposure to a DoE‐based library of elution protocols. Elution protocols varied in incubation time, elution buffer pH and elution buffer TEA concentration. EV elution after each protocol was analyzed by on‐bead lipid fluorescent staining and flow cytometry. EV permeability was tested by exposing EVs to proteinase K during each elution protocol and western blotting for intraluminal EV proteins. (b) Elution efficiency of HEK293T EVs captured using antibodies against CD9 (red) or CD81 (blue) after exposure to PBS as controls or each elution protocol (Run1‐15). Mean ± SD of three technical replicates is shown. (c) HEK293T EV permeability during each elution protocol. Bars show mean ± SD of two technical replicates and indicate the average degradation of Alix, β‐actin, TSG‐101 and syntenin, expressed as a percentage of PK‐treated EVs exposed to PBS. EVs exposed to SDS at physiological pH, or SDS at high pH and TEA concentrations (TEA/SDS) served as positive controls for PK activity. (d, e) Prediction models of EV elution after capture with anti‐CD9 (d) or anti‐CD81 (e) antibodies, generated based on data in panel b. (f) Prediction model of EV permeability generated based on data in panel c. (g) Elution of HEK293T EVs captured on beads using antibodies against CD9, CD63, CD81 or PS, using elution buffers with pH ranging from 11.0 to 11.5 for 5 min. (h) Representative western blot of HEK293T EV permeability analysis based on Alix, TSG‐101, syntenin and β‐actin. EVs were exposed to elution buffers with pH ranging from 11.0 to 11.5 or SDS as a control in the presence of PK for 5 min. Alternatively, EVs were incubated with the pH 11.2 elution buffer for 5 min, neutralized to physiological pH and subsequently incubated with PK for 5 min (11.2*). The bar chart shows the average degradation of all four proteins based on band intensity analysis. DoE, Design‐of‐Experiments; EV, extracellular vesicles; PK, proteinase K; TEA, triethylamine.

FIGURE 2

EV subpopulations isolated by EV‐Elute show distinct protein profiles and morphology. (a) Western blot analysis of common EV proteins in HEK293T EVs separated into surface‐marker defined subpopulations using EV‐Elute. The first lane (Whole EV isolate) shows EVs used as input for capture on magnetic beads using antibodies against CD9, CD63, CD81 and PS. Non‐captured supernatant from beads was loaded in ‘Sup’ lanes, and EV subpopulations recovered with EV‐Elute were loaded in ‘PD’ lanes. (b) Representative NTA size distribution patterns of HEK293T EVs and their subpopulations. Lines show mean ± SD of 5 measurements in a single sample. (c, d) Representative TEM (c) and Cryo‐TEM (d) images of whole EV isolates and isolated EV subpopulations from HEK293T EVs. Scale bars represent 200 nm. EV, extracellular vesicles; NTA, nanoparticle tracking analysis; TEM, transmission electron microscopy.

FIGURE 3

EV‐Elute induces a rapid disruption of Protein G‐antibody binding. (a) HEK293T EVs were captured on Protein G‐coated magnetic beads using antibodies against CD9, CD63, CD81 and PS. Beads were exposed to EV‐Elute for 0–5 min, stained with PKH67 and EV elution was analyzed using flow cytometry. In 0‐minute incubation, elution buffer was added to beads and immediately removed. (b) Representative TEM pictures of whole HEK293T EV isolate mixed with capturing antibodies against CD9, CD63, CD81 and PS ('Pre‐capture’) and corresponding released EV subpopulations after applying EV‐Elute (Post‐elution’). EM grids were immunolabelled with rabbit‐anti‐mouse and protein A‐gold (10 nm). Scale bars represent 200 nm. (c) Semi‐automated quantification of EV‐membrane‐associated gold in pictures shown in (b). Bars show mean ± SD of 5–10 pictures. Statistical differences were determined using unpaired student's t‐test; * indicates p < 0.05, **p < 0.01, and ***p < 0.0001. EV, extracellular vesicles; TEM, transmission electron microscopy.

FIGURE 4

EV subpopulations are taken up efficiently and specifically by multiple cell types. (a) Schematic illustrating the experimental setup of EV subpopulation uptake assays. EVs were labelled with Memglow 590 and added to co‐cultures of human PBMCs and eGFP‐expressing endothelial cells. After 4 h, uptake by individual cell types was analyzed by multiplexed flow cytometry. (b) Representative uptake experiment showing absolute uptake of whole EV isolates and subpopulations by co‐cultures from a single PBMC donor. Uptake is expressed as the difference in MFI of Memglow 590 between EV‐treated cells and PBS‐treated controls (ΔMFI). Bars indicate mean ± SD from three technical replicates. (c) Cell type specificity of whole EV isolates and EV subpopulations. The uptake of EVs by each cell type was expressed as a percentage of total EV uptake within experiments. Bars show mean ± SD of four experiments with different PBMC donors. No significant differences in cell type specificity between EV subpopulations were detected using one‐way ANOVA with Tukey post‐hoc test. eGFP, enhanced green fluorescent protein; EV, extracellular vesicles; MFI, mean fluorescence intensity; PBMC, peripheral blood mononuclear cell

FIGURE 5

Capacity of EVs to deliver engineered cargo differs between EV subpopulations. (a) Schematic representation of engineered EVs derived from HEK293T cells transfected with constructs encoding CD9 or CD63 fused to MS2 coat protein via UV‐cleavable linkers, MS2 aptamer‐modified sgRNA, Cas9 and VSV‐G. (b) Schematic representation of the Cas9/sgRNA‐activatable stoplight reporter constructs stably expressed in HEK293T‐Stoplight cells. (c) Activation of HEK293T‐Stoplight reporter cells by escalating doses of whole isolates of engineered EVs (Cas9/sgRNA tethered to CD9) and derived surface‐marker defined EV subpopulations separated by EV‐Elute. EV doses were normalized based on Memglow 590 fluorescence and corresponded to 3E9 particles/well for the highest dose of whole EV isolate and 8E8 particles/well for the highest doses of derived subpopulations, as calculated from EV‐Elute recovery rates and applied fluorescence‐based dilutions. EVs were serially diluted twofold for medium and low doses. Reporter activation is expressed as a percentage of eGFP+ cells of mCherry+ cells, measured 4 days after EV addition by flow cytometry. The dotted red line indicates background reporter activation. (d) Activation of HEK293T‐Stoplight cells by non‐bound bead supernatants from EV subpopulations shown in (c). If no ‘active’ subpopulations are captured by the beads, a reporter activation similar to that induced by whole EVs would be expected. (e) Representative fluorescence microscopy pictures of HEK293T‐Stoplight reporter cells treated with high doses of EVs analyzed in (c). Activated reporter cells are indicated with white arrows. (f) Activation of HEK293T‐Stoplight reporter cells by two‐fold escalating doses of whole isolates of engineered EVs (Cas9/sgRNA tethered to CD63, calculated highest dose 2E9 particles/well) and derived surface‐marker defined EV subpopulations separated by EV‐Elute (calculated highest dose 8.5E8 particles/well). (g) Activation of HEK293T‐Stoplight cells by non‐bound bead supernatants from EV subpopulations shown in (f). All graphs are representative of at least 3 replicate experiments. Corresponding EV doses were compared with one‐way ANOVA with Tukey post‐hoc tests. *** indicates p < 0.001 compared to all other samples unless specific comparisons are indicated, **p < 0.01 and *p < 0.05. # indicates p < 0.05 compared to all samples except CD9+ EVs (ns) in (c), and p < 0.05 compared to all samples except CD9+ and CD63+ EVs (ns) in F. ns = not significant. eGFP, enhanced green fluorescent protein; EV, extracellular vesicles; VSV‐G, glycoprotein of vesicular stomatitis virus.