An accurate, precise method for general labeling of extracellular vesicles

Graphical abstract

incubated at 37 8C for 20 min to allow for intra-vesicular esterase activity to render the calcein AM fluorescent. The labeled EV suspensions were diluted with 200 mL PBS before data acquisition by flow cytometry (calcein excitation max 495 nm/emission max 516 nm, common fluorescein isothiocyanate channel) (Becton Dickinson, LSRII). A FSC/SSC dot plot compares gated events (EV candidates) with size calibration beads (Bangs Laboratories, Inc., cat. 832 and 833) (Fig. 1A). Gated events (Fig. 1A) were then selected for fluorescence above the background autofluorescence of non-stained EV candidates. The conversion of calcein AM in plasma MVs was apparent after 30 s of incubation (Fig. 1C). We used 20 min incubation with 10 mM calcein AM as standard conditions for reaching maximum fluorescence of different EVs (Fig. 1D).
Filtered PBS (200 mL) was then added to the suspended fluorescent EVs (300 mL final volume) before analysis by flow cytometry. To confirm that fluorescence was due to intra-vesicular calcein rather than free calcein, labeled EVs were subjected to multiple washes with PBS. The fluorescence spectrum did not shift appreciably after washing, indicating that fluorescence was not due to free activated calcein, which could lead to false detection of non-vesicular events ( Fig. 2A). In fact, washing EVs caused significant loss of events (Fig. 2B) so the numbers of washing cycles were minimized, and EVs were labeled with calcein AM directly before flow cytometry processing. Concentrations of EVs were calculated by ratiometric comparison with Flow-Check Fluorospheres (Beckman Coulter, Inc., cat. 6605359).

Validation Approach 1 -Sizing of isolated EVs
To confirm that the isolated events were EV-sized, we acquired events in the flow cytometry forward scatter channel for the plasma, HAEC, and RBC EVs, as well as 0.76, 0.99, and 2.53 mm size calibration beads. Comparison to the size calibration beads indicated that the gated events were within the size range of EVs (Fig. 3).

Validation Approach 2 -Permeabilization and general membrane staining
To demonstrate that calcein AM labels intact EVs-but not membrane fragments or other debris-we permeabilized the plasma, RBC, and HAEC EVs with two different agents: Triton X-100 (Sigma-Aldrich, cat. T-8787) and saponin (Sigma-Aldrich, cat. 47036-50G-F). Triton X-100 is a nonionic surfactant that non-specifically permeabilizes lipid membrane bilayers and lyses cells. Saponin is a plant-derived amphipathic glycoside that complexes with cholesterol to permeabilize membrane bilayers. Permeabilization would allow for esterase and calcein AM leakage from the EVs, and we hypothesized that we would see an inverse relationship between the concentration of the permeabilizing agent and calcein fluorescence intensity. To track the permeabilization and lysis of the EVs, we co-stained with the non-specific membrane dye, PKH26 (excitation max 551 nm/emission max 567 nm, common phycoerythrin (PE) channel) (Sigma-Aldrich, cat. MINI26-1KT).
After isolation, EVs were resuspended in varying concentrations of 0.2 mm-filtered Triton-X 100 (0%, 0.001%, 0.01%, 0.1%, and 1%) or saponin (0 mg/mL, 1 mg/mL, 10 mg/mL, 100 mg/mL, or 1 mg/mL) for 10 min or 20 min, respectively. The permeabilized EV suspensions were then 2Â diluted in filtered PBS and centrifuged at 16,100 Â g for 20 min to remove residual agents. The pellets were then resuspended in 100 mL of 10mM calcein AM as before for 20 min at 37 8C. To prevent PKH26 micelle formation, physiologic salts in the PBS were removed by pelleting the calcein[ 4 _ T D $ D I F F ] AM-labeled EVs (16,100 Â g, 20 min) and resuspending the pellet in Diluent C (Sigma-Aldrich, provided with PKH26 dye) before addition of [ ( F i g . _ 5 ) T D $ F I G ] the PKH26 dye (2 mM final concentration). After 5 min incubation at room temperature, filtered PBS was added to the EV suspensions (300 mL final volume) before flow cytometry processing. Single stains of calcein AM and PKH26 on non-permeabilized EVs, along with an non-stained control, were prepared for compensation controls.
Co-staining of non-permeabilized EVs resulted in events that were double positive for calcein AM and PKH26 (Figs. 4 and 5). As expected, with increasing levels of permeabilizing agents, the EVs remained positive for PKH[ 5 _ T D $ D I F F ] 26, indicating that the detected events were lipid membranes[ 2 _ T D $ D I F F ] , but the calcein fluorescence decreased substantially with increasing permeabilizing agent concentration. This demonstrated that calcein AM labels intact EVs rather than debris.

Validation Approach 3 -Permeabilization and specific membrane labeling
To further demonstrate that calcein AM labels intact EVs, we modified our first validation strategy by labeling RBC EVs with fluorescently-labeled antibody against the RBC-specific antigen CD235a instead of staining with the general dye, PKH26.
Red blood cell EV pellets were resuspended in 100 mL of filtered PBS before the addition of either anti-CD235a conjugated antibody (PE/Cy7 anti-human CD235a (Glycophorin A); BioLegend, cat. 349112) or the isotype control (PE/Cy7 Mouse IgG2a, k isotype Ctrl; BioLegend, cat. 400232). After 20 min of incubation at room temperature, the EV suspension was diluted with 1.4 mL of filtered PBS before centrifugation at 16,100 Â g for 20 min to remove non-bound antibodies. The pellets to be permeabilized were resuspended in 500 mL filtered saponin solution (1 mg/mL) for 20 min at room temperature. All RBC EV pellet suspensions were raised to 1.5 mL with filtered PBS before centrifugation (16,100 Â g, 20 min). The pellets were resuspended in 100 mL of 10mM calcein AM and incubated at 37 8C for 20 min. Filtered PBS (200 mL) was added to each of the suspensions before flow cytometry processing (PE/Cy7 excitation max 496 nm/emission max 785 nm). Single labeled and nonlabeled compensation controls of non-permeabilized EVs were also prepared.
In summary, we found that non-permeabilized EVs were positive for both CD235a and calcein AM labeling (Fig. 6) Upon permeabilization, however, the EVs lacked calcein fluorescence while retaining CD235a labeling. The isotype control-which had not been permeabilized-exhibited calcein positivity but CD235a negativity. These results confirm the previous findings that calcein AM labels intact EVs, rather than lysed EVs or debris.
In our statistical measurement of calcein AM as an indicator for RBC EVs, the sensitivity was 99.1 AE 0.15%, the specificity was 54.3 AE 6.2%, and the calcein CV was 188 AE 2.8% (n = 6, mean AE SEM). Furthermore, the method was largely free of systematic or random errors, exhibiting 98.8 AE 0.17% accuracy and 99.8 AE 0.066% precision.
The activation and retention of non-fluorescent calcein AM to the fluorescent calcein in intact EVsbut not permeabilized EVs -indicates that labeling with calcein AM is a suitable method for flow cytometry-based detection of intact EVs released from multiple cells types.
[ ( F i g . _ 6 ) T D $ F I G ] Fig. 6. Red blood cell EVs labeled with calcein and anti-CD235a antibody were double positive for both, but lost calcein fluorescence when pre-permeabilized with saponin. Dot plot (A) and quantified events in quadrants (B).