The information exchange in complex biological systems relies on sophisticated mechanisms involving mediators at molecular and cellular levels. Recent literature has shown that different subcellular vesicles display potent regulatory functions, mediated both by their surface receptors and their content [1
]. Among these elements, the biological role of extracellular vesicles (EVs) has been strongly emphasized [2
Extracellular vesicle is the umbrella term for all types of cell-derived vesicles, including microvesicles and exosomes, therefore representing a heterogeneous population of small vesicles deriving from virtually all cell types (i.e., endothelial cells, platelets, leukocytes), and released during cell growth, proliferation, activation, apoptosis, or senescence processes [2
]. EVs are constantly present in the bloodstream and they have been implicated in the regulating functions of remote organs and tissues [2
]. EVs are also characterized by an integral plasma membrane, expressing the phenotype of the cells from which they originate. It has been demonstrated that EVs retain a broad enzymatic repertoire, being able to maintain a number of biological activities even after the budding from their parental cells [6
]. Moreover, EVs may represent the indices of cell activation and/or tissue degeneration, occurring during pathophysiological events in vivo.
It has been reported that EVs play a crucial role in a multitude of pathologies, including malignancies, cardiovascular, inflammatory, metabolic, and autoimmune diseases [3
]. As a consequence, circulating EVs have been proposed as reliable biomarkers, able to provide relevant information on pathogenic events and response to treatments in several clinical settings [3
]. However, due to their small size, state-of-the-art protocols for EVs detection requires a number of pre-analytical enrichment steps, such as centrifugation/ultracentrifugation, size exclusion chromatography, ultrafiltration, immunocapture, hydrostatic or hydrostatic filtration dialysis; therefore, their final characterization relies on material that differs from the original body fluid [16
In this context, evaluating to which extent the final measurement reflects the initial characteristics of the samples, and how these features have been influenced by pre-analytical processing enrichments may result difficult [18
Polychromatic flow cytometry (PFC) tends to emerge as a promising technique for EV characterization and enumeration. Due to its sensitivity, flexibility and ability to quickly analyze thousands of events and multiple parameters at the same time, it allows to simultaneously characterize and quantify EVs stemming from different parental cells [17
]. Even if typical applications of flow cytometry rely on forward scatter (FSC)/side scatter (SSC) measurements, some data suggest that flow cytometry approaches used to identify EVs, based solely on their scatter parameter detection, underestimate EVs counts [22
]. It has been also demonstrated that the careful choice of an EVs probe and of the staining conditions allows the application of a fluorescence triggering, which could have a profound impact on the amelioration of the sensitivity of EVs flow cytometry analyses [23
Therefore, the application of a simplified protocol, combined with the possibility of applying a fluorescence triggering, would represent a substantial step forward in the process of unequivocally identifying and enumerating EVs by PFC.
In the present study, we developed a straightforward procedure, which takes advantage of a lipophilic cationic dye (LCD), for the identification, enumeration, and separation of EVs from different origins (platelets, leukocytes or endothelial cells). The protocol was applied on fresh peripheral blood (PB) samples and the LCD staining of EVs allowed the use of a PFC analysis based on the application of a trigger threshold. We show that LCD identifies biological elements that display typical EVs features. Moreover, the application of this newly optimized PFC protocol allows reproducible EVs counts as well as the efficient purification of EVs by fluorescence-activated cell sorting. As demonstrated by proteomics analysis, sorted EVs preparations also carry a protein cargo linked to EVs specific bioactions (i.e., binding processes).
Altogether, these results show that this newly optimized method represents an essential step towards for the study of EVs clinical significance in different pathological settings.
EVs are circulating vesicles, generated as a cellular response to different stimuli that contribute to coagulation, inflammation, cellular homeostasis, survival, and waste material transport. EVs have also been shown to be effectors capable of delivering biological messages (mRNA, miRNA, proteins, surface molecules) to target cells [2
]. Their concentration, biochemical composition, and cellular origin may give relevant clinical information, and increased numbers of circulating EVs have been observed in a variety of diseases.
Identification and characterization of different types of EVs is challenging due to the lack of appropriate isolation and purification methods. Currently, differential ultracentrifugation, immunoaffinity capture, size-exclusion chromatography and microfluidics are the techniques recommended by the International Society for Extracellular Vesicles (ISEV) [18
]. Current protocols, however, require several pre-analytical blood manipulations, that may promote EVs release and/or may induce cell damage affecting the final measurement. For these reasons, the translation of EVs basic research into clinical practice is challenging.
Polychromatic flow cytometry (PFC) represents the method of choice for EVs identification and enumeration. However, the potential for EVs as biomarkers has been limited by the fact that even if a great number of PFC protocols were proposed to identify and enumerate EVs, results from previous studies showed a high degree of variability [31
Nevertheless, when a PFC method for EVs detection in whole blood is optimized, several other considerations about pre-analytical as well as analytical phases must be considered [18
]. Taking into account recent methodological guidelines for EVs studies [16
], we demonstrated that the flow cytometry analysis of EVs must be performed on freshly drawn whole blood samples (within 4 h from bleeding), confirming that prolonged storages generate artifacts [32
Concerning the flow cytometry analytical phase, we recommend referring to all established guidelines for setting a general PFC panel (i.e., use of the FMO controls, the appropriate gating strategy, the quality controls) [39
]. MISEV guidelines for analytical variables, as well as MIFlowCyt and MIFlowCyt-EV suggestions for general variables and experimental design related to flow cytometry EVs experiments were taken into account [18
Here we developed a simple procedure, which takes advantage of a lipophilic cationic dye (LCD) for the identification and separation of EVs. It must be underlined that the application of such a flow cytometry method allowed us to detect and enumerate EVs on freshly drawn whole blood samples, avoiding possible artifacts generated by the enrichment procedures (i.e., centrifugation/separation) that can activate or damage cells and artificially generate or disrupt EVs or induce their fusion [32
]. Therefore, the application of EVs analysis on fresh samples reduces the EVs loss problems, preserving the EVs characteristics [45
The here presented approach takes advantage from PFC and it is based on a fast (an hour from bleeding to acquisition) protocol that needs a small volume of whole blood allowing to discriminate intact from damaged EVs in an extensible way.
Furthermore, given that LCD stains the EVs compartment, we could apply the trigger threshold on the channel in which LCD emits, instead of using the triggering on SSC. We then demonstrated that, by this approach, it was possible to detect EVs. Finally, the use of the Rosetta Calibration allowed us to identify the EVs compartment based on size values, therefore distinguishing EVs from other blood elements. As matter of fact, by applying this newly optimized PFC protocol, which also combines the LCD staining to phalloidin for the identification of damaged membranes [47
], a high level of repeatability, in terms of EVs counts (CV~3–7%) was achieved [23
Circulating EVs (endothelium, platelet- and leukocyte-derived EVs) surrounded by an intact plasma membrane were also sub-typed, after optimizing an appropriate panel of antibodies. Data showed that the most abundant population among those that we detected was the one stemming from platelets, followed by the leukocyte-derived population and the one derived from endothelium [12
For the first time, by using such a method we were also able to isolate LCD+/Phalloidin- intact EVs by fluorescence-activated cell sorting.
The staining of EVs by LCD also resulted highly specific, given that it excludes HDL, LDL and chylomicron apolipoproteins, which are referred to be the most common EVs contaminates [16
]. Because the majority of plasma lipoprotein particles have a very small size, these results were confirmed by proteomic that appears as a more suitable method for this type of comparison. Fluorescence activated cell sorting, in fact, giving the possibility to obtain highly purified material, does not allow in any case to separate large numbers of EVs, compatible with some types of analyses (i.e., western blotting). Proteomics data showed that sorted preparations of EVs resulted significantly less contaminated from the soluble circulating components, such as abundant proteins and apolipoproteins, usually co-purified in UC protocols [16
]. Moreover, by applying this method we obtained similar results when some other biofluids (i.e., cerebrospinal fluid, tears) were analyzed [53
]. Furthermore, already published data confirm the nature of LCD+/Phalloidin- EVs, given that, as we have already demonstrated by proteomics, they carry a number of top proteins that were identified in EVs elsewhere [54
Therefore, these data demonstrated that, the application of the here presented method, allowed the separation of highly purified material from a low volume of samples, particularly suitable for further analyses and potentially useful for EVs detection in many other biofluids [53
]. As shown in the pie chart in Figure 4
C, which identifies the EVs protein cargo, these EVs resulted functionally organized, carrying binding proteins (possibly associated to their ability to interact with specific target cells), as well as catalytic and regulatory proteins (probably able to modulate a range of functions in their target cells). The identified EVs protein cargo resulted well organized even from a structural point of view, given that the 8.1 % of the identified proteins belong to the cytoskeleton, and are involved in the structure maintenance (Figure 4
D). Moreover, 10.8 % of those proteins resulted involved in the immune system-mediated activities and the remaining are functional proteins (enzyme modulators, oxidoreductase, receptors and signaling molecules). These data confirm, for EVs circulating in the blood, some other recently published results obtained for cerebrospinal fluid and tear EVs, indicating a general functional organization of the EVs cargo [53
4. Materials and Methods
The study was approved by the local ethics committee of Chieti-Pescara and University G. d’Annunzio, Chieti-Pescara (V. 1.0, 4 February 2016). All participants and activities were conformed to the current legislation and regulations in Italy, European Legislation, International Conventions and Declarations. In accordance with the Helsinki II Declaration, all involved subjects gave written informed consent before their inclusion in the study, and participants were identified by anonymized codes. Peripheral blood (PB) samples were obtained from 53 healthy Caucasian donors that gave written informed consent. They did not declare to be under chronic therapies, nor to be affected by chronic conditions. Samples from 31 healthy donors were used for protocol optimization, while the remaining 22 healthy volunteers, whose demographic characteristics are reported in Supplementary Table S2
, were analyzed for the assessment of EVs concentrations and phenotypes.
4.2. Flow Cytometry EVs Staining
Collection and staining of PB samples. PB was drawn (21 G needles) in two sodium citrate tubes (Becton Dickinson Biosciences-BD, San Jose, CA, USA, Ref 454387) and processed within 4 h from venepuncture. The first harvested tube was discarded to minimize venepuncture-induced vascular damage effects [55
]. To obtain a method for a rigorous EVs definition, different known EVs tracers were tested (Supplementary Table S3
). Among them, LCD resulted in the most promising marker, giving the best separation of the positive population with respect to the related internal negative one (Supplementary Figure S7
). The best combination of markers for EVs analysis was obtained after testing different reagent combinations (Supplementary Table S4
). In order to stain PB samples, a reagent mix was prepared by adding to 195 μL of PBS 1×, 0.5 μL of Fluorescein isothiocyanate (FITC)-conjugated phalloidin and LCD, and all reagents, as detailed in Supplementary Table S5
; then 5 μL of whole blood were added to the mix. The lipophilic cationic dye is a commercial compound, that we have validated and patented for its off label use to stain EVs for further flow cytometry analysis. The chemical structure of this molecule is not public. BD Biosciences produces the custom LCD kit on the basis of customer requests. Given that LCD kit is a custom product, it is not reported on standard catalogues, but related reference numbers are 626266 (antibodies, listed in Supplementary Table S5
) and 626267 (LCD and FITC-conjugated phalloidin, Supplementary Table S5
). To avoid immune complex formation and the unspecific background linked to the antibody aggregation, each reagent stock solution was centrifuged before its use (21,000× g
, 12 min). After 45 min of staining (RT, in the dark, or at 37 °C when Annexin V was not present in the reagent mix), 500 μL of PBS 1× were added to each tube and 1 × 106
events/sample were acquired by flow cytometry (FACSVerse, BD Biosciences, San Jose, CA, USA). In a subset of samples, Peridinin Chlorophyll Protein-Cyanin (PerCP-Cy) 5.5-conjugated Annexin V was also added (0.25 µL, BD Biosciences, Cat: 561431), and, in this case, Binding Buffer 1X (BD Biosciences) was used instead of PBS 1X. The dilution of the sample was optimized, and, at the used dilution (1:143), no swarm effects occur (Supplementary Figure S5A
All requirements imposed for polychromatic flow cytometry EVs analysis were taken into account [18
]. In detail, MISEV guidelines for analytical variables, as well as MIFlowCyt and MIFlowCyt-EV suggestions for general variables and experimental design related to FC EVs experiments were taken into account.
4.2.1. Analysis of Platelet Activation
The reagent mix was prepared by adding 3 μL of CD41a PerCP-Cy 5.5-conjugated (BD Biosciences, Cat: 333148), 20 μL CD62P R-phycoerythrin (PE, BD Biosciences, Cat: 555524) and LCD, phalloidin-FITC, CD31 PE-Cyanine 7 (PE-Cy7)-conjugated and CD45 Brilliant Violet 510 (BV510)-conjugated (Supplementary Table S5
); then 5 µL of whole blood were added to the mix for 45 min (37 °C, in the dark), then 500 µL of PBS 1X were added to each tube before the acquisition.
4.2.2. Staining of Apolipoproteins
The reagent mix was prepared by adding to 195 µL of PBS 1×, 0.5 μL of LCD and all reagents listed in Supplementary Table S6
; then 5 μL of PB were added to the mix and incubated (45 min, 37 °C, in the dark); 500 μL of PBS 1X were added to each tube before the acquisition.
4.3. Flow Cytometry Extracellular Vesicle Acquisition and Analysis
The trigger threshold was placed on the channel in which LCD emits (Allophycocyanin—APC—channel, threshold value = 200/262,144). MegaMix-Plus beads (Byocitex, Marseille, France) were measured in order to verify overtime the correct placement of the gating on the scattered dot-plots. For all used parameters the height (H) signals, as well as bi-exponential or logarithmic modes were selected. Rosetta Calibration (Exometry, Amsterdam, The Netherlands) was used, according to manufacturer’s specifications, to calibrate side scatter, relate side scatter in arbitrary units to standardized units of nm, as well as to the diameter and refractive index of particles [57
]. Throughout all other measurements, we used the same settings. The EVs/SSC relationship was automatically obtained by Mie theory, considering the optical configuration of the instrument (FACSVerse) and assuming a particle refractive index core of 1.40 (Figure 1
A, Supplementary Figure S4A
Instrument performances, data reproducibility, and fluorescence calibrations were sustained by the Cytometer Setup & Tracking Module (BD Biosciences). The evaluation of non-specific fluorescence was obtained by acquiring FMO combined with the respective isotype control [49
]. To ascertain LCD staining, Triton X-100 1% (n
= 3), buffer only and reagent only controls were acquired (Supplementary Figures S1D,E and Figure S2
]. Compensation was assessed using CompBeads (BD Biosciences) and single stained fluorescent samples. Data were analyzed using FACSuite v 188.8.131.5230 (BD Biosciences) and FlowJo X v 10.0.7 (BD Biosciences) software. EVs concentrations were obtained by the volumetric count function [59
]. The ERF for FITC, PE and APC channels were calculated using Ultra Rainbow Quantitative Particle Kit (Spherotech, Lake Forest, IL, USA, Cat. Number URQP-38-6K), following the manufacturer’s instructions.
4.4. Synthesis and Staining of Rhodamine-Liposomes
Thin layer evaporation and extrusion methods were used to synthesize non-fluorescent liposomes or Rhodamine-DHPE liposomes, as previously reported [23
]. Liposomes extruded by 100 nm membrane filters were measured by DLS (average size = 103 nm) and stained (100 µL) by adding LCD (BD Biosciences) at the same concentration used for EVs detection; then 500 µL of PBS 1X were added to the samples and analyzed by flow cytometry (FACSVerse, BD Biosciences).
4.5. Carbonyl Cyanide 3-Chlorophenylhydrazone Impact on LCD Staining of Extracellular Vesicles
To ascertain the possibility that LCD staining is also linked to the EVs trans-membrane potential, four PB samples were initially treated by 50 μM (CCCP; Sigma-Aldrich, Corporation, St. Louis, MO, USA) or its vehicle (0.1% DMSO) for 15 min at RT, and then stained by LCD and phalloidin-FITC, and finally acquired by flow cytometry.
4.6. EVs Separation by Fluorescence-Activated Cell Sorting
In order to separate EVs by fluorescence activated cell sorting, PB samples were stained by a reagent mix prepared as above described. Briefly, 0.5 µl of FITC-conjugated phalloidin and LCD (BD Biosciences–Catalogue, #626267, Custom Kit), and all reagents, as detailed in Table S5
, were added to 195 µL of PBS 1X; then 5 µL of whole blood were added to the mix. After 45 min of staining (RT, in the dark, or at 37 °C), at least 500 µL of PBS 1X was added to each tube. Such a dilution allowed us to maintain the correct event rate recommended for the nozzle that we have used (100 μm nozzle).
The total EVs fraction and/or PLT-derived CD41a+ EVs (gated as shown in Figure 1
) were separated (100 μm nozzle) from PB samples by a fluorescence-activated cell sorter (FACS, FACSAria III, BD Biosciences) [23
]. The instrument was set as already described for the analyzer. In detail, the trigger threshold was placed on the APC channel and, for all parameters, the height (H) signals, as well as bi-exponential or logarithmic modes were selected. The post-sorting purity (Supplementary Figure S1A–C
) was assessed by using the same instrument (FACSAria III) and the same setting applied for EVs separation [23
]. Instrument performances, data reproducibility, and fluorescence calibrations were sustained by the Cytometer Setup & Tracking Module (BD Biosciences).
4.7. Transmission Electron Microscopy Analysis
A Formvar/Carbon 300 Mesh Nickel grid (Agar Scientific, Stansted, UK, Cat: S162N3) was placed on the bottom of a polypropylene tube (14 × 89 mm, Beckman Coulter, Brea, CA, USA, Ref: 331372), filled by the suspension of FACS-purified EVs, and centrifuged (100,000× g, 70 min, 4 °C, max brake setting; Optima XL-100K ultracentrifuge, rotor = SW 41 Ti Swinging-Bucket Rotor, Beckman Coulter). Samples were then fixed by using 1% Glutaraldehyde in 0.1 M Cacodylate Buffer (pH 7.4). Images were acquired by a transmission electron microscope ZEISS 109 equipped with a Gatan-Orius SC200W-Model 830.10W TEM CCD Camera.
4.8. Dynamic Laser Light Scattering (DLS) Analysis of Purified EVs
Extracellular vesicles, purified as above described by FACS, were analyzed by DLS (90Plus/BI-MAS Zeta Plus, Brookhaven Instrument Corp, Holtsville, NY, USA). EVs size was obtained from the translational diffusion coefficient, by using the Stokes–Einstein equation. DLS data, referring to the intensity of scattering of the samples, were analyzed.
4.9. ImageStream Analysis of EVs
MegaMix-Plus beads SSC were used as reference material for dimensions on the brightfield channel. EVs were stained and separated by FACS, as aforementioned. EVs preparations were then immediately acquired on an ImageStream X Mk II (AMNIS Seattle, Seattle, WA, USA) imaging flow cytometer (using a 60 × objective), equipped with the Inspire software (V.200.1.620); data were analyzed by IDEAS software v6.2 (AMNIS). At least 5.5 × 103 sorted EVs/sample were recorded.
4.10. EVs Label-Free Proteomics
The number of purified EVs, established by FACS, was used to normalize proteomics analyses. In detail, 1 × 106
EVs separated by FACS were used for each proteomic detection as previously reported [53
]. EVs digested proteins were acquired in triplicate by LC-MS/MS using a Proxeon EASY-nLCII (Thermo Fisher Scientific, Milan, Italy) chromatographic system coupled to a Maxis HD UHR-TOF (Bruker Daltonics GmbH, Bremen, Germany) mass spectrometer. In-source reference lock mass (1221.9906 m/z) was acquired online throughout runs. Protein identification was carried out by the MASCOT search engine, assuming the carbamidomethylation and the methionine oxidation as fixed and variable modification, respectively. The Exponentially Modified Protein Abundance Index (emPAI) [29
] obtainable by MASCOT results from MS/MS data was used for approximate relative abundance of proteins in the mixture. Gene Ontology classification of identified proteins was carried out by PANTHER, which classified 58 identified EVs proteins. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [64
] partner repository with the dataset identifier PXD022807.
4.11. Statistical Analysis.
Flow cytometry and emPAI protein data were analyzed using the XLSTAT 2014 (Addinsoft, Paris, France) and GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA, USA). Two-sided Student’s t-test or paired t-test were used as indicated. Statistical significance was accepted for p < 0.05.