If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Imaging flow cytometry challenges the usefulness of classically used extracellular vesicle labeling dyes and qualifies the novel dye Exoria for the labeling of mesenchymal stromal cell–extracellular vesicle preparations
Correspondence: Bernd Giebel, PhD, Institute for Transfusion Medicine, University Hospital Essen, University of Duisburg-Essen, Virchowstr. 179, Essen 45147, Germany.
Extracellular vesicles (EVs) are involved in mediating intercellular communication processes. An important goal within the EV field is the study of the biodistribution of EVs and the identification of their target cells. Considering that EV uptake is assumed to be important for EVs in mediating intercellular communication processes, labeling with fluorescent dyes has emerged as a broadly distributed strategy for the identification of EV target cells and tissues. However, the accuracy and specificity of commonly utilized labeling dyes have not been sufficiently analyzed.
Methods
By combining recent advances in imaging flow cytometry for the phenotypic analysis of single EVs and aiming to identify target cells for EVs within therapeutically relevant mesenchymal stromal cell (MSC)-EV preparations, the authors explored the EV labeling efficacy of various fluorescent dyes, specifically carboxyfluorescein diacetate succinimidyl ester, calcein AM, PKH67, BODIPY TR ceramide (Thermo Fisher Scientific, Darmstadt, Germany) and a novel lipid dye called Exoria (Exopharm Limited, Melbourne, Australia).
Results
The authors’ analyses qualified Exoria as the only dye that specifically labeled EVs within the MSC-EV preparations. Furthermore, the authors demonstrated that Exoria labeling did not interfere with the immunomodulatory properties of the MSC-EV preparations as tested in a multi-donor mixed lymphocyte reaction assay. Within this assay, labeled EVs were differentially taken up by different immune cell types.
Conclusions
Overall, the results qualify Exoria as an appropriate dye for the labeling of EVs derived from the authors’ MSC-EV preparations. This study also demonstrates the need for the development of next-generation EV characterization tools that are able to localize and confirm the specificity of EV labeling.
Extracellular vesicles (EVs) are membrane-enclosed particles in the nano- and micrometer range that are secreted into their extracellular environment by virtually all cells. According to their origin, EVs are classified into different groups. The most prominent groups are exosomes, derivatives of the endosomal system with sizes ranging from 70 to 150 nm; microvesicles, shed-offs of the plasma membrane measuring 100–1000 nm; and apoptotic vesicles, which can be as small as exosomes and, as apoptotic bodies, reach sizes up to several micrometers [
Despite these classes, EVs of each given subtype are very heterogeneous as well. Depending on the cell source they originate from, they provide specific molecular compositions, qualifying them as a new class of biomarkers. Specifically, EVs residing in the plasma are increasingly used as biomarkers for different diseases [
Elevated levels of extracellular vesicles are associated with therapy failure and disease progression in breast cancer patients undergoing neoadjuvant chemotherapy.
]. In addition to conventional methods, such as cytokine assays or cellular analyses, the prevalence of selected EV subpopulations can provide important new information on the course of respective diseases.
It has become evident that EVs are of physiological relevance and mediate complex intercellular interactions at local and remote sites under both healthy and pathological conditions [
]. Thus, it is a goal of many EV researchers to dissect such intercellular communication processes in a number of different biological processes. In this context, it is a relevant task to identify EV target cells. In addressing this challenge, a common strategy evolved using fluorescent dyes considered to specifically label EVs and to apply labeled EV fractions to assumed target cells/tissues in vitro or in vivo. Among the commonly used dyes are dyes that immediately integrate into membranes, such as PKH dyes; dyes that become fluorescent after enzymatic reactions, including non-fluorescent carboxyfluorescein diacetate succinimidyl ester (CFDA-SE), which is processed by esterase into carboxyfluorescein succinimidyl ester (CFSE); calcein AM, which binds calcium cations to become fluorescent or 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY TR ceramide)-conjugated fatty acids, including ceramide [
]. Because of their micelle-forming capabilities, use of several of these dyes is challenging for the EV field. Sophisticated preparation and washing procedures need to be followed to efficiently deplete dye aggregates, very often resulting in minor recovery rates [
]. Furthermore, some dyes, including CFDA-SE, require specific enzymatic activities to become fluorescent, in the case of CFDA-SE an esterase, to bind EV-associated proteins and thus efficiently label EVs.
For the quality control of dye-labeled EV fractions, particle quantification methods are commonly used and often performed by nanoparticle tracking analysis (NTA) or resistive pulse sensing. In 2011, the authors’ group, in addition to Dragovic et al., introduced NTA as an “exosome” quantification method [
]. However, the detailed comparison of data recorded using NTA and imaging flow cytometry (IFCM) (a more advanced EV characterization method) indicated that particle quantification methods are not appropriate for calculating EV concentrations in EV samples—unless they are ultra-pure. This cannot be achieved with conventional EV preparation techniques such as differential centrifugation, polymer precipitation or simple size exclusion technologies [
Single extracellular vesicle analysis performed by imaging flow cytometry and nanoparticle tracking analysis evaluate the accuracy of urinary extracellular vesicle preparation techniques differently.
]. In their traditional form, particle quantification methods cannot distinguish prepared particles such as protein precipitates, salt crystals and lipoprotein agglomerates from EVs. In addition, because of their limited sensitivity, light scattering-based methods can detect only some particles smaller than 100 nm [
Particle size distribution of exosomes and microvesicles determined by transmission electron microscopy, flow cytometry, nanoparticle tracking analysis, and resistive pulse sensing.
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
]. These protocols allowed the authors to investigate whether tetraspanins—specifically, CD9, CD63 and CD81—whose expression within EV samples has been confirmed by western blot are co-localized on individual EVs or recovered on distinct EV subsets. With this technology, the authors analyzed EV preparations from mesenchymal stromal cell (MSC)-conditioned, human platelet lysate-supplemented media, whose therapeutic activities were studied in different animal models, and confirmed their therapeutic potential in a graft-versus-host disease patient [
Small extracellular vesicles obtained from hypoxic mesenchymal stromal cells have unique characteristics that promote cerebral angiogenesis, brain remodeling and neurological recovery after focal cerebral ischemia in mice.
]. The authors demonstrated that in MSC-EV preparations that contain a high concentration of platelet-derived CD9+CD81− EVs, CD9 and CD81 reside on different EV subpopulations, all of which are in the exosomal size range [
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
Considering the IFCM method very informative for EV characterization and intending to qualify a pan-EV labeling dye, the authors thus decided to evaluate the EV labeling efficacy of different dyes used for EV marking. In addition to conventionally used BODIPY TR ceramide, calcein AM, CFSE and PKH67 dyes, the authors included a novel dye called Exoria in our studies. Exoria, developed at Exopharm Limited (Melbourne, Australia), was designed to be a pH-stable fluorescence dye with reduced micelle-forming propensity that could incorporate into EVs. Specifically, Exoria is based on Rhodamine being covalently linked to octadecane (see supplementary Figure 1). Since the structure of Exoria is similar to other lipid dyes, intercalation into the lipid membrane is likely to occur during staining.
In this study, the authors investigated the labeling efficiency of MSC-EV preparations with the aforementioned dyes. Counterstaining of PKH67- and Exoria-labeled objects was performed with anti-tetraspanin antibodies. The impacts of Exoria labeling on the immunomodulatory capabilities of the MSC-EV preparation were investigated in a multi-donor mixed lymphocyte reaction (mdMLR) assay [
]. Furthermore, the authors documented the uptake of Exoria-labeled objects by different immune cells within the mdMLR assay.
Methods
Preparation of EVs from MSC-conditioned cell culture media
MSC-EVs were prepared from human platelet lysate containing MSC-conditioned media by polyethylene glycol 6000 precipitation followed by ultracentrifugation as described previously [
Precipitation with polyethylene glycol followed by washing and pelleting by ultracentrifugation enriches extracellular vesicles from tissue culture supernatants in small and large scales.
]. Conditioned media were harvested every 48 h. Obtained MSC-EV preparations were diluted in sodium chloride (NaCl)–4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany) such that 1 mL of the final samples contained a preparation yield of the conditioned media of approximately 4.0 × 107 cells, which was then stored at –80°C until usage. In total, six different MSC-EV preparations were used in this study.
Characterization of the EV preparations
Obtained EV preparations were characterized according to minimal information for studies of extracellular vesicles criteria [
Minimal information for studies of extracellular vesicles 2018 (MISEV2018): a position statement of the International Society for Extracellular Vesicles and update of the MISEV2014 guidelines.
]. Briefly, average particle concentrations were determined by NTA on a ZetaView PMX-120 platform equipped with analysis software (version 8.03.08.02) (Particle Metrix GmbH, Meerbusch, Germany) as described previously [
Precipitation with polyethylene glycol followed by washing and pelleting by ultracentrifugation enriches extracellular vesicles from tissue culture supernatants in small and large scales.
]. Protein concentration was determined by bicinchoninic acid assay (Pierce, Rockford, IL, USA) in 96-well plates according to the manufacturer's recommendations. The presence of EV-specific proteins (CD9, CD63, CD81 and syntenin) and absence of impurities (calnexin) were confirmed in western blots performed as described previously [
Precipitation with polyethylene glycol followed by washing and pelleting by ultracentrifugation enriches extracellular vesicles from tissue culture supernatants in small and large scales.
Aliquots of three independent MSC-EV preparations were stained with the different dyes. Staining with CFDA-SE (known as CFSE) (Thermo Fisher Scientific, Darmstadt, Germany) was based on the manufacturer's protocol. Slight modifications were required to reduce the signal from unbound CFSE. Briefly, the CFSE stock solution was diluted to a working solution of 10 µM CFSE. The solution was centrifuged three times for 10 min at 17 000 × g (see supplementary Figure 2). Subsequently, 25 µL of the MSC-EV preparations, corresponding to the amount of EVs derived from 1 × 106 MSCs, was incubated with the centrifuged CFSE solution for 20 min at 37°C. The sample was diluted 1:20 to a final volume of 1 mL prior to analysis.
Staining of MSC-EV preparations with calcein AM (Thermo Fisher Scientific) followed the manufacturer's instructions. Briefly, 25 µL of the MSC-EV preparations was incubated with 25 µL of a 20-µM solution of calcein AM for 40 min at 37°C. The sample was diluted 1:20 to a final volume of 1 mL to reduce background noise, avoiding the requirement of a washing step.
Staining of MSC-EV preparations with BODIPY TR ceramide (Thermo Fisher Scientific) followed the manufacturer's instructions. Briefly, 25 µL of the MSC-EV preparations, corresponding to EVs purified from 4 × 106 MSCs, was incubated with 25 µL of a 20-µM solution of BODIPY TR ceramide for 20 min at 37°C. A total of 450 µL of 0.9% NaCl (B. Braun Melsungen AG, Melsungen, Germany) with 10 mM HEPES (Thermo Fisher Scientific) buffer was added, and the EVs were washed using a Vivaspin 500 centrifugal concentrator (Sartorius, Göttingen, Germany) at 100 000 molecular weight cutoff. The retentate was adjusted to 500 µL prior to analysis.
Staining of MSC-EV preparations with PKH67 (Thermo Fisher Scientific) followed the manufacturer's instructions for labeling EVs. Briefly, using 200 µL of the given MSC-EV preparation, corresponding to EVs purified from 8 × 106 cells, the solution was adjusted with Diluent C to a final volume of 1 mL. A total of 6 µL of PKH67 dye was added to each tube and mixed continuously for 30 seconds. After 5 min at room temperature, the solution was quenched by adding 2 mL of 10% (w/v) bovine serum albumin fraction V (Carl Roth GmbH + Co KG, Karlsruhe, Germany). Serum-free medium—low-glucose Dulbecco's Modified Eagle's Medium (PAN-Biotech GmbH, Aidenbach, Germany) supplemented with 100 U/mL penicillin–streptomycin–glutamine (Thermo Fisher Scientific)—was used to adjust the volume to 8.5 mL. A total of 1.5 mL of a 0.971-M sucrose solution (Carl Roth GmbH + Co KG) was added to the bottom of the tube, and the tube was centrifuged for 2 h at 190 000 × g in an SW 40 Ti swing-out rotor (Beckman Coulter, Krefeld, Germany) at 4°C and a k-factor of 137. The supernatant was discarded and the pellet resuspended in NaCl-HEPES buffer. After resuspension, the volume was adjusted to 5 mL and transferred to a Vivaspin 6 centrifugal concentrator (Sartorius). The retentate was adjusted to 120 µL prior to analysis.
The MSC-EV preparations were stained with Exoria following the protocol provided by Exopharm Limited Exoria was provided as a lyophilized powder. A total of 1 mg was resuspended with 1 mL buffer to a final concentration of 0.2 µM. Like CFDA-SE, the Exoria solution was centrifuged for 10 min at 17 000 × g to reduce background noise to a minimum. Briefly, for EV labeling, 25 µL of the MSC-EV preparations was incubated with 25 µL of a prepared, centrifuged Exoria solution (0.2 µM) for 1 h at 37°C. The sample was diluted 1:20 to a final volume of 1 mL prior to analysis. For EV uptake experiments, Exoria-labeled MSC-EV preparations were cleared from EV-unbound Exoria by ultrafiltration. Briefly, after labeling with Exoria, 50 µL of the given stained and undiluted MSC-EV preparation was washed for 10 min by centrifugation at 12 000 × g through Vivaspin 500 filters (Sartorius). The retentates (approximately 10–30 µL) were collected as labeled EV samples and adjusted with NaCl-HEPES to final volumes of 50 µL.
Antibody labeling of prepared EVs
After dye labeling, 5 µL of Exoria-stained MSC-EV samples was mixed with 20 µL of 10-nM anti-human CD9 fluorescein isothiocyanate (EXBIO Praha, a.s., Vestec, Czech Republic), 12 nM anti-human CD63 AF488 (EXBIO Praha, a.s.) or 13 nM anti-human CD81 fluorescein isothiocyanate (Beckman Coulter) antibody solution and incubated for 2 h at room temperature. PKH67-stained MSC-EV samples were incubated with 10 nM anti-human CD9 phycoerythrin (PE) (EXBIO Praha, a.s.), 12 nM anti-human CD63 PE (EXBIO Praha, a.s.) or 13 nM anti-human CD81 PE (Beckman Coulter) for 2 h at room temperature as described previously [
]. Accordingly, isotype controls were performed (see supplementary Table 1). For Exoria, final preparations were diluted with phosphate-buffered saline to 500 µL for CD9 (end dilution factor of 1:100) and 200 µL for CD63 and CD81 analyses (end dilution factor of 1:40). PKH67 preparations were diluted to 100 µL for all three analyses (1:20 dilution).
Detergent control
To test for the EV nature of labeled objects, detergent controls were performed by adding 2% (w/v) NP-40 solution (Calbiochem, San Diego, CA, USA) to the samples.
IFCM analyses
All samples were measured using the built-in autosampler from Falcon 96-well U-bottom plates (category 353077; Corning, Kaiserslautern, Germany), with 5-min acquisition time per well, on the Amnis ImageStreamX Mark II flow cytometer (Luminex Corporation, Seattle, WA, USA). All data were acquired at ×60 magnification at low flow rate (0.3795 ± 0.0003 μL/min, directly determined by the system) and with the removed beads option deactivated as described previously [
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
]. Additional settings can be found in supplementary Tables 2 and 3.
Multi-donor mixed lymphocyte reaction
The immunomodulatory potential of Exoria-labeled and non-labeled MSC-EV preparations was compared using an mdMLR assay as described previously. Briefly, the Ficoll (Bio&Sell, Nürnberg, Germany)-prepared peripheral blood mononuclear cells (PBMCs) of 12 healthy donors that were routinely collected in our blood donation center were mixed in equal proportions, aliquoted and stored in the vapor phase of liquid nitrogen until use. After thawing, 600 000 cells were seeded per well of 96-well U-bottom plates and cultured in 10% human AB serum (produced in-house) and Roswell Park Memorial Institute 1640 medium (Thermo Fisher Scientific) supplemented with 100 U/mL penicillin and 100 µg/mL streptomycin (Thermo Fisher Scientific) at a final volume of 200 µL per well in either the presence or absence of the MSC-EV preparations to be tested. Aliquots of six well-characterized MSC-EV preparations (three with confirmed immunomodulatory activities and three lacking these activities) were labeled with Exoria, and unbound dye was removed by ultrafiltration as described earlier. As a control, Exoria was mixed with the buffer and processed in a manner analogous to that used for the MSC-EV-containing samples. After 5 days, cells were harvested, stained with a collection of different fluorescence-labeled antibodies—specifically, CD4-BV785 (BioLegend, San Diego, CA, USA), CD25-PE-Cy5.5 (BD Biosciences, Franklin Lakes, NJ, USA) and CD54-AF700 (EXBIO Praha, a.s.)—and analyzed on a CytoFLEX flow cytometer with CytExpert 2.3 software (Beckman Coulter). Activated and non-activated CD4+ T cells were discriminated by means of their CD25 and CD54 expression, respectively. Typically, 5 µL of the MSC-EV preparations to be tested was applied to the respective wells. The following antibodies were used to further discriminate subpopulations: CD8-BV650 (BioLegend), CD14-PO (EXBIO Praha, a.s.), CD19-ECD (Beckman Coulter) and CD56-APC (BioLegend). Evaluation of the data was carried out with Kaluza 2.1 software (Beckman Coulter).
Statistical analysis
Statistical analysis and graphical representation were performed with Prism 8.4.3 (GraphPad Software, San Diego, CA, USA). Mean values ± standard deviations are provided.
Results
CFSE, calcein AM and BODIPY TR ceramide do not label MSC-EVs
With the goal of identifying a dye that allows specific labeling of EVs in therapeutically active MSC-EV preparations, the authors decided to evaluate the accuracy of conventionally used EV labeling dyes—specifically, CFSE, calcein AM, PKH67 and BODIPY TR ceramide—and a novel lipid dye called Exoria. MSC-EV preparations that have been extensively explored in various animal models were obtained from supernatants of MSCs raised in 10% human platelet lysate-supplemented media by the authors’ well-established polyethylene glycol ultracentrifugation protocol [
Precipitation with polyethylene glycol followed by washing and pelleting by ultracentrifugation enriches extracellular vesicles from tissue culture supernatants in small and large scales.
]. Since micelle formation of some of the dyes has already been reported, the authors followed Minimum Information about a Flow Cytometry Experiment EV recommendations [
] and initially added all of the labeling dyes without the EV sample to the NaCl-HEPES buffer—the buffer MSC-EVs were suspended in. Samples were processed according to the manufacturers’ recommendations and analyzed by IFCM with protocols the authors have successfully established for the characterization of antibody-labeled MSC-EVs [
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
Notably, depending on the specific manufacturer's protocol, different amounts of MSC-EV preparation were required. For CFSE, calcein AM, BODIPY TR ceramide and Exoria labeling, the authors started with an MSC-EV preparation volume of 25 µL; for PKH67, the starting volume was 200 µL. Initially, the authors analyzed all recorded objects. Based on the authors’ prior experience, small EVs (sEVs) appear as fluorescently labeled objects with minimal side scatter (SSC) signals. Upon comparing the dye-only solutions, CFSE, calcein AM and Exoria did not contain labeled objects. By contrast, upon analyzing the BODIPY TR ceramide and PKH67 solutions, solid populations of labeled objects with minimal SSC signals were identified (Figure 1). The data imply micelle or aggregate formation of BODIPY TR ceramide and PKH dyes.
Figure 1PKH67 and Exoria specifically label objects in MSC-EV preparations with sEV-like SSC properties. Fluorescence labeling procedures were performed for CFSE, calcein AM, BODIPY TR ceramide, PKH67 and Exoria in the absence of any EV preparation (top row), in the presence of MSC-EVs alone (middle row) or in the presence of MSC-EVs and the detergent NP-40 (bottom row). Fluorescence intensities of dye-labeled objects (x-axis) are plotted against the intensity of their size-reflecting SSC signals (y-axis).
Subsequently, the authors analyzed MSC-EV preparations labeled using the same procedure. Unlike all other dyes, the PKH67 protocol provided by the manufacturer includes a density gradient centrifugation step for the removal of unbound dye. Since density gradient centrifugation is associated with increased material loss during preparation, the PKH67 labeling procedure was initiated with 200 µL of the original MSC-EV preparation, whereas only 25 µL of the original MSC-EV sample was used for all other labeling procedures. In contrast to the buffer-only solutions, solid populations of labeled objects were observed after BODIPY TR ceramide, PKH67 and Exoria labeling, and some objects were observed following CFSE labeling (Figure 1). Calcein AM failed to label any detectable objects. BODIPY TR ceramide+ objects that were not recovered in the buffer–BODIPY TR ceramide solution control revealed SSC signals that were much higher than those typically seen for sEVs. By contrast, the light scattering properties of objects specifically labeled with PKH67 or Exoria reflected those of sEVs. Notably, in good agreement with published reports that PKH dyes increase the size of labeled EVs [
], the PKH67+ objects specifically labeled in the MSC-EV preparation indicated higher SSC signals than Exoria+ objects (Figure 1).
To determine whether the specifically labeled objects were detergent-sensitive, the dye-labeled MSC-EV samples were treated with NP-40. Although BODIPY TR ceramide+ objects were specifically detected in MSC-EV preparations, those with higher light scattering properties and all PKH67+ and Exoria+ objects disappeared following NP-40 treatment. By contrast, the population of CFSE+ and BODIPY TR ceramide+ objects with sEV light scattering properties was hardly affected by NP-40 treatment. Therefore, the authors determined that neither detergent-resistant CFSE+ nor BODIPY TR ceramide+ objects with low light scattering properties were sEVs. Considering that BODIPY TR ceramide may label lipoproteins contained in the MSC-EV samples, the authors labeled a mixture of commercially available high-density lipoprotein and low-density lipoprotein with BODIPY TR ceramide and observed a population of objects similar to that seen in BODIPY TR ceramide-labeled MSC-EV samples (see supplementary Figure 3). Accordingly, the authors concluded that BODIPY TR ceramide bound to the lipoproteins rather than to the EVs contained in the samples. Based on this and the failure of calcein AM to label any specific objects, the authors excluded CFSE, calcein AM and BODIPY TR ceramide from all later analyses and focused on exploring the accuracy of PKH67 and Exoria as MSC-EV labeling dyes.
PKH67 fails to effectively label CD9+, CD63+ and CD81+ sEVs
Next, the authors investigated the potential co-localization of PKH67 with known EV markers. To this end, the authors continued with the well-characterized MSC-EV preparations. These were stained by either PKH67 alone or in combination with anti-CD9, anti-CD81 or anti-CD63 antibodies. To reduce the background noise and exclude coincident events—specifically, the simultaneous detection of two or more independent objects (coincidences with high object numbers)—the authors applied an optimized gating strategy. Briefly, the authors focused on objects recognized as singlets in the PKH67 channel without a simultaneous antibody signal or as singlets in the antibody channel without a simultaneous PKH67 signal as well as on events appearing in both channels as singlets not providing two individual objects (Figure 2A).
Figure 2pPKH67 fails to effectively label small MSC-EVs. (A) Gating strategy of the detected objects. The gating strategy is presented for MSC-EV preparations counterstained with PKH67 and anti-CD9 antibodies. Both fluorescence channels (Ch02 and Ch03) were initially plotted against the SSC intensities (SSC) of all recorded objects. The number of coincident objects per channel is depicted (2nd plots in row 1 and 2). Of all recorded objects, the only objects considered in subsequent analyses were those that showed either single signals in the PKH67 or antibody channel or single signals in both channels (singlets). Within the Ch02 SSC singlet plots, three different gates were defined with singlets in R1 and R2 (low SSC signals) and R3 (concrete SSC signals). Objects in R1 revealed no PKH67 signal and those in R2 and R3 demonstrated concrete PKH67 signals. Ch02 signals plotted against SSC signals of the singlets are shown in either the same size plot as that shown in the left column before gating or in the zoomed in versions of the same plots (right column). (B) Distribution of recorded singlets in R1, R2 and R3 without antibody labeling or following anti-CD9, anti-CD63 or anti-CD81 labeling, respectively. Plotting of the fluorescence intensities of singlets in the PKH67 (Ch02) or antibody (Ch03) channel against SSC intensities of the singlets. Columns three to five show fluorescence intensities of R1–3 gated singlets. (C) Number of events in gates R1–3 for the respective measurements, as exemplified by one of three biological triplicates. Mean ± standard deviation of a technical triplicate is indicated.
Upon plotting SSC against PKH67 intensities, many more objects were recovered in samples that had been counterstained by anti-CD9 antibodies than in the PKH67-labeled buffer and MSC-EV-containing controls. Most of these objects were negative for PKH67 and showed low SSC signals. The region containing these objects was defined as R1. For anti-CD9 staining, 6672 ± 1170 objects were recovered in R1. Notably, hardly any objects were recovered in R1 in the PKH67-labeled MSC-EV samples that were not counterstained by antibodies. A slight increase in object numbers was recorded when PKH67-labeled MSC-EV samples were counterstained with anti-CD63 (221 ± 42 objects) or anti-CD81 (96 ± 32 objects) antibodies. Most of the objects that were positive for PKH67 revealed low SSC signals. The region that included these objects was defined as R3. A smaller number of PKH67-labeled objects with low SSC signals were clustered in the region defined as R2. In contrast to the number of objects in R1, the numbers of objects in R2 and R3 were only slightly affected by the antibody labeling procedures (Figure 2B,C). Within the control not labeled with antibodies, 2040 ± 344 objects were recovered in R2. After anti-CD9 staining, the number of objects was 2927 ± 466; after anti-CD63 staining, the number of objects was 1747 ± 141; and after anti-CD81 staining, the number of objects was 1734 ± 200. In all antibody-labeled MSC-EV preparations, more objects were found in R3 (anti-CD9, 6625 ± 803 objects, anti-CD63, 5915 ± 271 objects, anti-CD81, 5777 ± 675 objects) than in the control not labeled with antibodies (2046 ± 157 objects). To analyze objects within the three different regions in more detail, their antibody labeling intensities were plotted against PHK67 labeling intensities. The results clearly confirmed that a huge proportion of the objects in R1 were effectively labeled by anti-CD9 antibodies. Although the R1 object populations were much smaller after anti-CD63 and anti-CD81 antibody staining than after anti-CD9 antibody staining, a proportion of these objects were clearly recognized as CD63+ and CD81+ (Figure 2B). By contrast, all objects in R2 and R3 appeared as CD63− and CD81− objects, most of which were able to be labeled by anti-CD9 antibodies. Notably, the frequency of CD9+, CD63+ and CD81+ objects recovered in R1 was congruent with the authors’ previous observations that MSC-EV preparations contain a dominating CD9+CD81− sEV population and a minor CD9−CD81+ sEV population [
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
]. Overall, the authors determined that most antibody-stained objects in R1 were sEVs that were not labeled by PKH67 and that most of the sEVs could be detected only if they were succesfully stained with any of the three antibodies. Thus, the authors’ data question the efficiency of PKH67 as an MSC-sEV-labeling dye.
Exoria effectively labels CD9+, CD63+ and CD81+ EVs in MSC-EV preparations
Next, the reliability of Exoria as an EV labeling dye was investigated in a manner comparable to that used for PKH67. To this end, the MSC-EV preparations (n = 3) were stained with Exoria alone or in combination with anti-CD9, anti-CD63 or anti-CD81 antibodies. Without defining R1–3 subgates, gating strategies were applied as depicted in Figure 2A. In contrast to the PKH67 labeling experiments, many more objects with lower SSC signal intensities were labeled by Exoria, even in the absence of any of the three antibodies. No clear increase in the numbers of detected objects was observed between MSC-EV samples that were labeled solely by Exoria and those that included the addition of anti-CD9 antibodies (Figure 3). Thus, in contrast to PKH67 labeling, Exoria labeling was sufficient to label most of the sEVs within the authors’ MSC-EV preparations. Interestingly, upon plotting the Exoria labeling intensities against those of the different antibodies, it appeared that CD81+ objects were more intensively labeled with Exoria than CD63+ objects. Furthermore, the results implied that more than 90% of the CD9+ and CD81+ objects had been labeled with Exoria, whereas only 60% of the CD63+ EVs had been. All labeled objects were confirmed to be detergent-sensitive (see supplementary Figure 2). Even though the authors do not have an explanation for the weaker Exoria stainability of CD63+ objects compared with CD9+ and CD81+ objects, overall, the data demonstrate that Exoria successfully labeled most of the sEVs in the MSC-EV preparations.
Figure 3Exoria stains tetraspanin-containing EVs. MSC-EV preparations were counterstained with Exoria and anti-CD9, anti-CD63 or anti-CD81 antibodies. The same gating strategy as that described for Figure 2 was applied. Fluorescence intensities of singlets are plotted against SSC intensities for either the antibody (Ch02) or Exoria (Ch03) channel. In the third column, Exoria signals are plotted against signals of the respective antibodies. NP-40 lysis controls are presented in supplementary Figure 4. FITC, fluorescein isothiocyanate.
Exoria staining does not affect the immunomodulatory capacity of MSC-EV preparations
Next, the authors investigated whether Exoria affected the immunomodulatory capability of the MSC-EV preparations. To this end, the authors performed an mdMLR assay to compare the activity of Exoria-stained MSC-EV preparations with the corresponding non-labeled MSC-EV preparations. Upon pooling the PBMCs of 12 healthy donors, allogeneic immune reactions were induced that could be monitored by the activation status of CD4+ T cells. After 5 days in culture, approximately 25% of all CD4+ T cells expressed CD25 and CD54, indicating T-cell activation (Figure 4). As previously described, MSC-EV preparations with immunomodulatory capabilities effectively reduce the content of activated CD4+ T cells [
]. Consistently, in the presence of the non-labeled MSC-EV preparations, only 16% of the monitored CD4+ T cells were found to display activation cell surface markers (Figure 4B). In the presence of Exoria-labeled MSC-EV preparations (n = 3), the authors observed a comparable reduction in CD4+ T-cell activation. Notably, Exoria itself did not influence the activation status of CD4+ T cells. Thus, Exoria did not recognizably affect the immunomodulatory capability of the applied MSC-EV preparations.
Figure 4Exoria staining does not affect the immunomodulatory capability of MSC-EV preparations. Mixtures of PBMCs of 12 different donors were cultured in the presence or absence of non-labeled or Exoria-labeled MSC-EV preparations or in the presence of Exoria dye for 5 days. Thereafter, cells were harvested and stained with DAPI and fluorescently labeled anti-CD4, anti-CD25 and anti-CD54 antibodies and analyzed by conventional flow cytometry. (A) Gating strategy for CD4 T cells. Living cells were identified according to their FSC and SSC features as singlets and DAPI– cells. CD4 T cells were gated as CD4+ living cells. (B) Fluorescence intensities of CD25 and CD54 gated living CD4+ cells of mdMLR assay cultured in the absence of any additives (stim), in the presence of non-labeled MSC-EVs (MSC-EVs) or Exoria-labeled MSC-EVs (MSC-EVs+Exoria) or in the presence of buffer-solved Exoria (Buffer+Exoria). DAPI, 4′,6-diamidino-2-phenylindole; FSC, forward scatter; FSC-A, forward scatter area; SSC-A, SSC area; SSC-H, SSC height; stim, stimulated.
Exoria-stained MSC-EVs are taken up differently by various immune cell types
To test whether Exoria EV labeling allowed the identification of EV-uptaking cells, the authors examined the labeled EV uptake of different immune cells using the mdMLR assay. To this end, a pool of PBMCs derived from 12 healthy donors was cultured for 5 days in the presence of Exoria-labeled MSC-EVs (n = 3) that had been cleared of excess Exoria dye by ultrafiltration. Thereafter, cells were harvested, labeled with antibodies and analyzed by flow cytometry. The content of Exoria-labeled cells within different PBMC subtypes was then determined. Almost all monocytes (CD14+ cells, 99%) revealed Exoria signals. By contrast, only proportions of the different lymphocytes appeared as Exoria+ cells—specifically, 71% of all CD4+ T cells (CD4+ cells), 34% of all CD8+ T cells (CD8+ cells), 72% of all B cells (CD19+ cells) and 15% of all natural killer cells (CD56+ cells). In addition, IFCM was used to visualize the subcellular staining of Exoria+ cells (Figure 5C). Obtained images revealed labeled structures that, according to the authors’ experience, were located subcellularly. Thus, Exoria-labeled EVs within the authors’ MSC-EV preparations were taken up at various rates by different immune cell types within the assay.
Figure 5Analyses of various immune cell types in mdMLR assay reveal differences in uptake of Exoria-stained MSC-EVs. Immune cells in mdMLR assay were examined for uptake of Exoria-stained MSC-EV preparations. (A) Discrimination of different subpopulations was performed using antibodies against CD4 and CD8 (T cells), CD19 (B cells), CD56 (NK cells) and CD14 (monocytes). (B) Different immune cells were examined for the presence of a signal for Exoria. Unstained MSC-EVs and a buffer control with Exoria were used as controls. (C) Analysis of subcellular staining following uptake of Exoria-stained EVs via imaging flow cytometry. Light (bright-field), fluorescence microscopy (PE) and merged (merge) images are shown. NK, natural killer.
In this study, the authors evaluated the accuracy of dye-mediated EV labeling as an example of well-studied MSC-EV preparations. Upon analyzing labeled MSC-EV preparations by IFCM, the authors demonstrated that none of the conventionally used dyes (BODIPY TR ceramide, calcein AM, CFSE and PKH67) allowed accurate labeling of MSC-EVs. By contrast, as determined using an mdMLR assay, the novel dye Exoria allowed the quantitative labeling of EVs in the authors’ MSC-EV preparations without interfering with their immunomodulatory properties. Furthermore, upon removing unbound dye by ultrafiltration, the authors identified EV uptaking cells. Notably, CD81+ EVs were stained more intensively with Exoria than CD63+ EVs, indicating potential differences in the membrane composition of both EV subtypes, which significantly influences their stainability. In this context, it is worth mentioning that upon comparing the intensity of Exoria-labeled HEK293T and THP-1 EVs in an ongoing study, almost all HEK293T EVs were able to be efficiently labeled, whereas most THP-1 EVs remained unstained (data not shown). Thus, even though Exoria appeared to be a very useful dye for EVs in the authors’ MSC-EV preparations, it should not be considered a pan-EV labeling dye. This could also be related to the saturated hydrocarbon chain, which is instrumental for possible intercalation into the lipid membrane. It has already been shown by others that intercalation is better or worse depending on where unsaturated compounds are located in a lipid dye [
For now, the EV field has just started exploring the heterogeneity within given EV preparations, with pioneering work performed using bead-capturing approaches, mainly against tetraspanins or other EV surface proteins [
]. Another critical component of EVs affecting dye incorporation is the lipid composition of EV subtypes. Many studies have demonstrated that EV preparations from different cell sources vary in their lipid composition [
Su H, Rustam YH, Masters CL, Makalic E, McLean C, Hill AF, et al. Characterization of brain-derived extracellular vesicle lipids in alzheimer's disease, bioRxiv (2020) 2020.08.20.260356.
]. Furthermore, many groups have utilized such differences to selectively capture EVs by utilizing different lipid-binding molecules (i.e., cholera toxin B chain, Shiga toxin B subunit and annexin V) [
Although the authors were unable to identify the cause of the difference in EV subtype labeling, this study clearly demonstrates that experimenters need to critically (re)evaluate the appropriateness of EV labeling dyes for their purposes. The specificity of EV labeling dyes for the EVs of interest cannot be investigated using conventional technologies such as differential centrifugation protocols for EV preparations or particle quantification devices [
]. Therefore, novel analysis devices that allow analysis of dye-stained and antibody-labeled EVs are required. In addition to IFCM, other devices that allow the co-localization of at least two different fluorescent labels on a single EV-sized object have entered the field, including a novel generation of flow cytometers for nanoparticles, such as the NanoFCM device [
]; and NTA devices in fluorescence modes (Particle Metrix). Indeed, elaborate analysis of EV labeling results performed on an NTA platform in fluorescence mode has revealed discrepancies in the labeling of EVs with PKH. These results imply that PKH uptake might not be connected to EVs [
]. We do not yet have any idea of the nature of PKH-labeled particles. Perhaps PKH dyes stain components other than EVs within given EV preparations. The authors’ results suggest that BODIPY TR ceramide efficiently labels high-density lipoprotein and low-density lipoprotein but not EVs. However, the potential to stain EVs might also depend on their origin, as was observed by the authors for Exoria staining in this study. In this respect, it is worth mentioning that although the authors were unable to label EVs with CFSE in the MSC-EV preparations, CFSE is qualified as an appropriate labeling dye for the staining of EVs prepared from conditioned medium of immature dendritic cells [
]. This discrepancy might be connected to varying esterase content in EVs of different cell types, which may eventually reduce the utility of CFSE for the labeling of EVs to EVs of selected cell types.
Overall, the results obtained in this and other studies question the reliability of broadly used “EV labeling” dyes, challenging the interpretations of many EV studies that use dye-labeled EV preparations for the identification of potential EV target cells. For now, it is a common strategy in the field to label EV preparations with “EV dyes” and perform uptake experiments with the dye-labeled EV preparations. In the authors’ opinion, researchers need to reevaluate whether their EVs are indeed specifically labeled and whether the cells that take up labeled particles are indeed the target cells of the EVs. Thus, EV uptake experiments remain extremely challenging. To the best of the authors’ understanding, the EVs in the MSC-EV preparations used in this study were accurately labeled, and the authors are confident that the Exoria-labeled EVs were specifically taken up by the various immune cell types. However, which labeled EV subtypes (CD9+, CD63+ or CD81+) were taken up most efficiently by the cells remains unknown. The authors also recognize that different immune cell types may have preferences for different EV subtypes. Although the current study fails to provide answers to these questions, the authors hope that it helps sensitize EV researchers around the globe to the challenges associated with the identification of EV target cells. Indeed, issues are further complicated by the fact that EV target cells do not necessarily need to take up EVs for transducing their signal. Similar to what has already been shown for synaptic vesicles [
], at least a proportion of EV-mediated intercellular interactions might follow the kiss-and-run principle, in which EVs bind to receptor platforms on cells and activate these platforms and are then shed off.
Declaration of Competing Interest
BG is a scientific advisory board member of Innovex Therapeutics SL and Mursla Ltd and a founding director of Exosla Ltd. MS and PFJ are employees and shareholders of Exopharm Limited.
Funding
This study was supported by funds from the European Union (ERA-NET EuroTransbio 11: EVTrust [031B0332B] and the European Union's Horizon 2020 research and innovation programme (EVPRO under grant agreement no. 814495, and AutoCRAT under grant agreement no. 874671). The materials presented and views expressed here are the responsibility of the authors only. The EU Commission takes no responsibility for any use made of the information given.
Author Contributions
Conception and design of the study: TT, MS, PFJ and BG. Acquisition of data: TT, OS and AA. Analysis and interpretation of data: TT, MS, PFJ and BG. Drafting or revising the manuscript: TT and BG. All authors have approved the final article.
Acknowledgments
The authors thank the Westdeutsche SpenderZentrale for providing bone marrow samples of healthy donors. The authors are grateful to the healthy blood donors whose cells were used in the mdMLR assay. The authors also thank Advanced Molecular Technologies Pty Ltd., Scoresby, Australia, for the synthesis and supply of Exoria.
Elevated levels of extracellular vesicles are associated with therapy failure and disease progression in breast cancer patients undergoing neoadjuvant chemotherapy.
Single extracellular vesicle analysis performed by imaging flow cytometry and nanoparticle tracking analysis evaluate the accuracy of urinary extracellular vesicle preparation techniques differently.
Particle size distribution of exosomes and microvesicles determined by transmission electron microscopy, flow cytometry, nanoparticle tracking analysis, and resistive pulse sensing.
Optimisation of imaging flow cytometry for the analysis of single extracellular vesicles by using fluorescence-tagged vesicles as biological reference material.
Small extracellular vesicles obtained from hypoxic mesenchymal stromal cells have unique characteristics that promote cerebral angiogenesis, brain remodeling and neurological recovery after focal cerebral ischemia in mice.
Precipitation with polyethylene glycol followed by washing and pelleting by ultracentrifugation enriches extracellular vesicles from tissue culture supernatants in small and large scales.
Minimal information for studies of extracellular vesicles 2018 (MISEV2018): a position statement of the International Society for Extracellular Vesicles and update of the MISEV2014 guidelines.
Su H, Rustam YH, Masters CL, Makalic E, McLean C, Hill AF, et al. Characterization of brain-derived extracellular vesicle lipids in alzheimer's disease, bioRxiv (2020) 2020.08.20.260356.
00049-4&id=gr1.jpg){kind=link}
00049-4&id=gr2.jpg){kind=link}
00049-4&id=gr3.jpg){kind=link}
00049-4&id=gr4.jpg){kind=link}
00049-4&id=gr5.jpg){kind=link}