Phenotypic Characterization of Acoustically Enriched Extracellular Vesicles from Pathogen-Activated Platelets

Abstract

Extracellular vesicles (EVs) are derived from the membrane of platelets and released into the circulation upon activation or injury. Analogous to the parent cell, platelet-derived EVs play an important role in hemostasis and immune responses by transfer of bioactive cargo from the parent cells. Platelet activation and release of EVs increase in several pathological inflammatory diseases, such as sepsis. We have previously reported that the M1 protein released from the bacterial pathogen Streptococcus pyogenes directly mediates platelet activation. In this study, EVs were isolated from these pathogen-activated platelets using acoustic trapping, and their inflammation phenotype was characterized using quantitative mass spectrometry-based proteomics and cell-based models of inflammation. We determined that M1 protein mediated release of platelet-derived EVs that contained the M1 protein. The isolated EVs derived from pathogen-activated platelets contained a similar protein cargo to those from physiologically activated platelets (thrombin) and included platelet membrane proteins, granule proteins, cytoskeletal proteins, coagulation factors, and immune mediators. Immunomodulatory cargo, complement proteins, and IgG3 were significantly enriched in EVs isolated from M1 protein-stimulated platelets. Acoustically enriched EVs were functionally intact and exhibited pro-inflammatory effects on addition to blood, including platelet-neutrophil complex formation, neutrophil activation, and cytokine release. Collectively, our findings reveal novel aspects of pathogen-mediated platelet activation during invasive streptococcal infection.

Introduction

Extracellular vesicles (EVs) are small vesicular bodies (30 nm–1 μm) that are derived from the cell membrane of various cells upon activation or injury, such as platelets, endothelial cells, erythrocytes, and leukocytes [1]. EVs can carry numerous bioactive molecules such as RNA, lipids, and proteins and typically present cell-specific surface antigens that reflect their cell of origin [2, 3]. EVs are continuously produced and found in all fluids of the human body but are most concentrated in the blood, where the majority (70–90%) of the circulating EVs are derived from platelets [4].

Platelets are the main regulators of hemostasis by prevention of bleeding and promotion of wound healing. This is achieved through the release of proteins and bioactive products from the platelet granules and upregulation of platelet surface receptors [5, 6]. Platelets have also been shown to become activated in response to infection and inflammation. Platelets interact with inflammation mediators, such as fibrinogen, IgG, and complement proteins [7], and directly modulate the function of other immune cells, including neutrophils and monocytes [5]. Platelet-derived EVs are also attributed an important role in maintaining hemostasis by provision of a binding surface for fibrinogen and coagulation factors [8]. The immunomodulatory effects of platelet-derived EVs are less investigated, but they can transfer bioactive cargo to immune cells, such as leukocytes, resulting in either activation or inhibition of leukocyte effector function [9]. Platelet-derived EVs also contribute to innate and adaptive immunity through antigen presentation via MHC-1 molecules and have access to the lymphoid organs and the bone marrow [10]. Circulating levels of EVs increase in several pathological states, such as thrombosis, rheumatoid arthritis, cancer, and sepsis [11, 12]. For example, platelet EV-derived mitochondrial damage-associated molecular patterns have been shown to enhance pro-inflammatory adverse reactions in transfusion recipients [13, 14]. The content and biological function of EVs may differ depending upon the agonist or conditions responsible for activation in distinct physiological or pathological scenarios [15, 16]; therefore, quantitative studies on the protein cargo of platelet-derived EVs may represent an important source of biomarkers of disease. The purpose of our study was to isolate and characterize EVs that may be released from pathogen-activated platelets during bacterial sepsis.

Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection [17]. The dual role of platelets in hemostasis and the immune response places them as key sentinels in the response to sepsis. S. pyogenes more commonly causes throat and skin infections; however, invasive disease such as sepsis is associated with distinct serotypes of the bacteria. The emm1 serotype of S. pyogenes is the most commonly associated with invasive disease [18]. The surface-associated M1 protein of the emm1 serotype contributes to evasion of the complement system and phagocytosis through interaction with several host plasma proteins: fibrinogen, albumin, the Fc-domain of IgG, and complement regulatory proteins [19]. The M1 protein can also be released from the bacterial surface to mediate activation of neutrophils, monocytes, and T cells [19]. We have previously demonstrated that M1 protein can activate platelets [20, 21] and stimulate platelet-leukocyte complex formation [22] as a result of the formation of immune complexes containing specific anti-M protein IgG and fibrinogen that engage with both the Fc- and the fibrinogen receptors on platelets incubated with fibrinogen-binding serotypes of M protein [20, 23–25].

A recently developed technology for isolating EVs is acoustic trapping, which utilizes an ultrasonic standing wave generated inside a microfluidic channel. Particles in the sound field that are denser and less compressible than the surrounding fluid will move to the pressure nodes in the field, where they can be retained against a flow [26]. In this way, acoustic trapping can be used for isolation, enrichment, and washing of EVs [27]. We recently reported on an acoustic trapping platform with significant improvements in throughput and capacity that enabled processing of milliliter sample volumes in minutes [28]. The decreased processing time and ability to work with small sample volumes, in combination with gentle forces, make acoustic trapping an attractive method for EV isolation as compared to conventional techniques such as ultracentrifugation.

In this work, we have applied an acoustic trapping platform to isolate platelet-derived EVs, with the aim of characterizing the EVs derived from plasma supernatants of pathogen (M1 protein)-stimulated platelets as compared with EVs from physiologically activated (resting or thrombin-activated) platelets. We applied quantitative mass spectrometry-based proteomics to characterize the protein cargo of enriched EV populations, and the functional integrity of acoustically enriched EVs was assessed in blood cell inflammation models.

Materials and Methods

Blood Collection and Preparation

Blood was collected from healthy donors into 0.1 m Na₃ citrate as an anticoagulant and centrifuged at 150 g for 15 min to obtain platelet-rich plasma (PRP). Informed consent was obtained from the donors prior to blood collection, and ethical approval was obtained from the Local Ethics Committee (approval 2015/801).

M1 Protein Purification

M1 protein was purified from the S. pyogenes mutant strain MC25, which lacks the cell wall-anchoring domain and secretes M1 protein into the supernatant, as previously described [20]. Briefly, MC25 bacteria were grown in Todd Hewitt broth with yeast, and proteins from the supernatant were precipitated with 80% (NH₄)₂SO₄. The precipitate was dialyzed against PBS and purified on Sepharose coupled with human fibrinogen. The M1 protein was eluted with 0.2 m glycine (pH 2.0) and dialyzed against PBS. Purified M1 protein was confirmed using SDS-PAGE.

Flow Cytometry for Determination of Platelet Activation

Flow cytometry was used to investigate platelet activation in response to M1 protein as previously described [21]. PRP from five healthy donors was diluted 2:3 in 1 mm HEPES buffer pH 7.4 and stimulated with M1 protein (2.5 μg/mL) for 15 min at room temperature. HEPES buffer alone was used to determine the background platelet activation, and the platelet agonist thrombin (1 U/mL, Triolab) was used as a positive control for platelet activation during hemostasis. The anticoagulant peptide Gly-Pro-Arg-Pro (1.25 mg/mL; Bachem) was added prior to stimulation with thrombin to prevent fibrin polymerization and coagulation. After stimulation, the samples were incubated with anti-CD62P-PE (1:10) (BD Biosciences, clone AC1.2) for 15 min at room temperature protected from light. Platelet counts were acquired on an Accuri C6 Plus flow cytometer (BD Biosciences), and the data were analyzed using C6 Plus Software. Platelets were gated based on size and granularity in logarithmic mode, and the CD62P intensity for the gated population was analyzed in histograms.

Isolation of EVs Using Acoustic Trapping

PRP from four healthy donors was stimulated with M1 protein (2.5 μg/mL), HEPES buffer alone, or thrombin (1 U/mL, Triolab), as described for flow cytometry. The samples were centrifuged at 1,600 g for 5 min, and the plasma supernatants, containing platelet-derived EVs, were processed on an acoustic trapping platform as previously described [28]. We have previously demonstrated that acoustic trapping generates equivalent populations of EVs as ultracentrifugation [29]. Briefly, an ultrasonic standing wave is generated in a glass capillary, and particles in the vicinity of the trap will move toward the pressure nodes of the standing wave, where they can be captured and retained against flow. By preloading the trap with large seed particles, submicron particles can be enriched through interaction with the seed particles [26]. This allows for isolation, enrichment, and washing of submicron particles, such as EVs. Prior to trapping, 400 μL of each plasma sample was diluted with D-PBS (Sigma-Aldrich) to 2,000 μL, yielding a concentration of 20% plasma. The trap was actuated at 12 V_pp (peak-to-peak) and loaded with 12 μm polystyrene seed particles (Sigma-Aldrich) to establish a seed particle cluster. The cluster was then washed with 1 mL PBS in order to remove excess seed particles. Sample was then run through the trap to isolate EVs in the seed particle cluster. While being held in the trap, the EVs were washed with 5 mL of PBS to remove background plasma proteins. Unless otherwise stated, the operating flowrate for all steps was 500 μL/min. Finally, the sound was turned off, and the cluster was allowed to sediment for 5 s to get closer to the exit, before it was eluted in a volume of 130 μL PBS at 5,000 μL/min. The isolated vesicles were then analyzed using flow cytometry and mass spectrometry. Additionally, the functional aspect of isolated EVs was investigated through addition of EVs to whole blood.

High-Sensitivity Flow Cytometry

Platelet-derived EVs in plasma or enriched by acoustic trapping were characterized using flow cytometry. PRP from the same four healthy donors as above was stimulated with M1 protein, thrombin, or HEPES buffer alone; plasma was collected; and EVs were isolated as described above. Platelet-derived EVs in plasma and acoustic trap-isolated EVs were incubated with PE-Cy™5 labeled mouse anti-human CD42b (0.25 mg/mL, BD Bioscience, Clone HIP1) for 30 min at room temperature protected from light. Samples were acquired on an Amnis CellStream Flow Cytometer (Luminex). Flow cytometer performance tracking was performed daily before all analyses using the Amnis CellStream Calibration Reagent (Luminex). The assigned voltage for forward scatter, side scatter, 488 (CD62p), and 642 (CD42b) was set to 100%. Acquisition of 10 μL per sample was performed at low speed (∼3 μL/min). Silica particles (Kisker Biotech GmbH & Co., Steinfurt, Germany) of known dimensions (100 nm, 200 nm, 300 nm, 500 nm, and 1,000 nm in diameter) were used for the instrument set up standardization (online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000531266). Fluorophores were chosen for distinct lasers to minimize compensation requirements.

Mass Spectrometry Sample Preparation

The proteome contents of isolated EVs from trapped samples and the non-trapped fraction of the original sample from the same four donors as above were analyzed using mass spectrometry. The vesicles were first lysed by addition of 200 μL of RIPA buffer (Sigma-Aldrich) for 10 min followed by mechanical disruption in a Bioruptor Plus (Diagenode) for 20 cycles (30 s on, 30 s off) using the low setting. Subsequently, 1,200 μL ice-cold acetone was added to each sample prior to incubation at −20°C overnight. The samples were then centrifuged at 18,200 g for 30 min at 4°C, and the supernatant was removed. Then, 500 μL of ice-cold 99.5% ethanol was added to each sample, and the samples were centrifuged again at 18,200 g for 30 min at 4°C and the supernatant removed. The samples were then dried in a SpeedVac (miVAC DUO) for 10 min and then resuspended in 100 μL of PBS.

The proteins were prepared for quantitative data-independent acquisition mass spectrometry (DIA-MS) using trypsin double digestion. A 4.6 μL solution containing 10 m urea and 50 mm ammonium bicarbonate, as well as 2 μL 0.5 μg/µL sequencing grade trypsin (Promega), for a final concentration of 432 mm urea, 2.16 mm ABC, and 9.4 ng/μL trypsin. The samples were incubated for 30 min at 37°C. Next, 45.4 μL of the urea-ABC solution were added to the samples and then incubated for 30 min at room temperature. The cysteine bonds were reduced by adding 0.5 μL of 500 mm tris (2-carboxyethyl) phosphine, resulting in final concentration of 1.64 mm tris (2-carboxyethyl) phosphine, for 60 min at 37°C. The cysteine bonds were then alkylated with 1 μL of 500 mm iodoacetamide, resulting in a final concentration of 3.26 mm iodoacetamide, for 30 min at room temperature in a dark environment. The samples were diluted using 250 μL of 100 mm ABC for a final urea concentration below 1.5 m, and, for protein digestion, 2 μL of trypsin were added, and the samples were incubated for 16 h at 37°C. The samples were acidified using 10% formic acid to a pH of 2–3, and the peptides were purified with SOLAµ HRP reverse phase columns (Thermo Scientific). The peptides were then dried in a SpeedVac and reconstituted in 2% acetonitrile and 0.2% formic acid. To ensure injection of an equal amount of peptide (1 μg) into the mass spectrometer, the peptide content in each sample was measured using a spectrophotometer (DeNovix, DS-11 FX+).

Liquid Chromatography Tandem Mass Spectrometry

The peptides from each sample were analyzed using DIA-MS analysis on a Q Exactive HF-X (Thermo Scientific) connected to an EASY-nLC 1200 (Thermo Scientific). The peptides were separated on a Thermo EASY-Spray column (Thermo Scientific 50 cm column) operated at 45°C and a maximum pressure of 800 bar. A linear gradient of 4%–45% acetonitrile in aqueous formic acid (0.1%) was run for 50 min. A full MS scan (resolution 60,000 for a mass range of 390–1,210 m/z) was followed by 32 full fragmentation MS/MS scans (resolution 30,000) with an isolation window of 26 m/z, including a 0.5 m/z overlap between windows. The precursor ions in each isolation window were fragmented using higher energy collisional-induced dissociation with normalized collision energy of 30. The automatic gain control was set to 3e6 for MS and 1e6 for MS/MS.

Mass Spectrometry Data Analysis

Raw MS data were gzipped and Numpressed mzML [30] using msconvert from ProteoWizard [31], v3.0.5930. Data were stored and managed with openBIS [32]. For DIA data analysis, a previously described spectral library containing the Homo sapiens and S. pyogenes serotype M1 reference proteomes (UniProt proteome IDs UP000005640 and UP000000750, respectively) was used [33]. The DIA-MS data were processed using the OpenSWATH pipeline [34]. For data analysis, raw data files were converted to mzXML using the tool msconvert and analyzed using OpenSWATH (version 2.0.1 revision:c23217). The retention time (RT) was calibrated using iRT peptides, and the RT extraction window was ±300 s. The m/z extraction tolerance was set at 0.05 Da. Peptide precursors were then identified by OpenSWATH (2.0.1) and PyProphet (2.0.1), with a false discovery rate of 1% at both the peptide precursor level and at the protein level. TRIC [35] was used to reduce identification error. The resulting datasets were analyzed using Jupyter Notebooks (3.1.1). Proteins identified with only one peptide, along with keratin proteins, were discarded. Total ion current (TIC) normalized intensities were calculated for the remaining proteins, and the proteins were assessed using Metascape [36].

Immunoblotting

Isolated vesicles were eluted with 25 μL PBS, incubated with reducing SDS-PAGE sample loading buffer (Thermo Fisher Scientific) for 10 min at 85°C, and loaded onto a stain-free protein gel (Bio-Rad). Plasma containing EVs from platelets stimulated with M1 protein, HEPES buffer alone, or thrombin was prepared as described above and subsequently centrifuged at 20,000 g for 90 min to isolate the EVs. The EVs that were isolated with centrifugation were loaded onto the gel and used as a positive control for the M1 protein. A Prestained Protein Ladder (Thermo Fisher Scientific) was used as a reference. SDS-PAGE was carried out, and the proteins were transferred to a membrane using a Trans-Blot Turbo System (Bio-Rad). The membrane was incubated with 5% skim milk (Millipore) for 1 h to reduce nonspecific binding and incubated with rabbit anti-M1 protein (1.2 μg/mL, BioGenes GmbH) at 4°C overnight. The membrane was washed 3 × 5 min with 0.05% Tween (Merck) in PBS and incubated with HRP-conjugated goat anti-rabbit (1:3,000, Invitrogen) for 1 h. The membrane was washed as described above and incubated with Western ECL Substrate (Bio-Rad) for 5 min. The membrane was imaged using a ChemiDoc system (Bio-Rad).

Transmission Electron Microscopy

Isolated EVs were fixed by addition of 65 μL of paraformaldehyde 6% to 130 μL of isolated EVs, yielding a concentration of 2% paraformaldehyde. The samples were incubated overnight at 4°C. The surface charge of the sample grids (TED PELLA Inc., Athene Hex Grids, Cu, 400M) was modified by glow discharge treatment. 30 µL of sample was added to the grid and incubated for 20 min. Following a wash with PBS for 5 min, the samples were treated with BSA (1%) for 10 min to reduce nonspecific binding. The EVs were immunogold labeled with a dual stain for both CD42 and M1 proteins. Samples were treated with mouse anti-CD42 primary antibody (1:30, BD Pharmingen) and incubated for 60 min. Following a wash with PBS for 5 min, the samples were treated with immunogold-conjugated secondary antibody (1:20, BBI Solutions, EM. Goat anti-mouse IgG, 10 nm) for 30 min and then washed with PBS for 5 min. The staining procedure was repeated with rabbit anti-M1 primary antibody (1:500, BioGenes GmbH) and immunogold-conjugated secondary antibody (1:20, BBI Solutions, Goat anti-rabbit, 15 nm). The samples were then fixed with 1% glutaraldehyde for 5 min and washed three times with Milli-Q for 5 min each. Finally, the samples were treated with 2% uranyl acetate for 5 min to improve contrast and washed with Milli-Q for 1 min. The samples were then imaged on a transmission electron microscope (FEI Tecnai Bio-Twin 120 kV).

Flow Cytometry of Platelet-Neutrophil Complex Formation and Neutrophil Activation

Whole blood from the same four healthy donors as above was stimulated for 15 min with 50 μL of EVs isolated from platelets from the same donor stimulated with HEPES, thrombin, or M1 protein, as described above. M1 protein (2.5 μg/mL) and thrombin (1U/mL) in combination with the anticoagulant peptide Gly-Pro-Arg-Pro (1.25 mg/mL) were used as positive controls for platelet activation. HEPES buffer alone was used to determine the background platelet and leukocyte activation. After stimulation, the samples were incubated for 15 min protected from light with anti-CD61-PE (1:10) (BD Biosciences, clone VI-PL2) to detect platelets associated with neutrophils and anti-CD11b-PerCP (1:10) (BD Biosciences, clone ICRF44) to determine neutrophil activation. The samples were acquired on an Accuri C6 Plus Flow Cytometer (BD Biosciences), and the data were analyzed using C6 Plus Software. The neutrophils were gated based on size and granularity, and the CD61 and CD11b intensity within the neutrophil gate was analyzed in histograms.

Whole Blood Stimulation and Cytokine Quantification

Cytokine release in response to isolated platelet-derived EVs was measured in whole blood using commercial ELISA kits. Whole blood from five healthy donors was diluted 1:5 in RPMI medium (Gibco, Life Technologies) and stimulated with 50 μL isolated EVs from platelets stimulated with HEPES, thrombin, or M1 protein for 24 h in a CO₂ incubator (37°C, 5% CO₂, >95% relative humidity). Heparin (0.6 μg/mL, Sigma-Aldrich) was added in combination with penicillin (100 mg/mL, Gibco, Life Technologies) and streptomycin (100 mg/mL, Gibco, Life Technologies) to prevent clot formation and contamination. Lipopolysaccharide from Escherichia coli O111:B4 (1 μg/mL, EMD Millipore Corp.) was used as a positive control for monocyte activation, and HEPES buffer alone was used to determine the background cytokine release. The cytokine release mediated by M1 protein alone (2.5 μg/mL) was also investigated. After incubation, the samples were centrifuged at 500 g for 5 min, and cytokine and chemokine levels in the supernatants were measured using commercial ELISA kits for IL-6 (Invitrogen, Thermo Fisher Scientific), human CCL2/MCP-1 (R and D Systems), and human CXCL8/IL-8 (R and D Systems), according to manufacturer instructions.

Statistical Analyses

A Mann-Whitney U test was chosen to compare sample distribution in two groups for nonparametric data. Sample distribution was compared to background levels, unless otherwise stated. Median and individual values were used to present nonparametric data for continuous variables. Results were considered statistically significant if p < 0.05. Data were analyzed using Prism 9 (GraphPad Software).

Results

M1 Protein from a Bacterial Pathogen Mediated Release of EVs That Were Acoustically Enriched

The purpose of this study was to use acoustic trapping to isolate and characterize EVs derived from pathogen (M1 protein)-stimulated platelets. A schematic overview of sample processing and analysis is shown in Figure 1. Platelet activation was investigated using flow cytometry of upregulation of CD62P to the surface of platelets in PRP from healthy donors. M1 protein and thrombin mediated significant platelet activation in all donors. A median of 77% and 93% of the platelet population became activated, respectively, compared to the background level of only 3% (Fig. 2a). EVs in plasma and isolated EVs captured by acoustic trapping (Fig. 1) from the healthy donors were characterized using flow cytometry. The platelet-specific marker CD42b was used to detect platelet-derived EVs in the size range 100–1,000 nm in plasma from platelets stimulated with thrombin and M1 protein, as well as in plasma from resting platelets incubated in HEPES buffer alone. The median number of EVs detected in plasma from M1 protein-stimulated platelets was 280,000/μL, compared to 250,000/μL in plasma from resting platelets and 300,000/μL in plasma from platelets stimulated with thrombin (Fig. 2b). Acoustic trapping generated isolated platelet-derived EVs from thrombin, M1 protein, and resting platelets. The median number of isolated EVs from M1 protein-stimulated platelets was 230,000/μL, compared to 240,000/μL from resting platelets and 230,000/μL from platelets stimulated with thrombin (Fig. 2c). Acoustic trapping has previously shown enrichment of EVs below 100 nm in diameter [28, 37]; however, the flow cytometer used is unable to resolve particles smaller than 100 nm. Therefore, only EVs in the size range 100–1,000 nm were included in the flow cytometry measurements, determined through the calibration particles. Size distributions of all particles found in the plasma samples, measured with nanoparticle tracking analysis, can be found in online supplementary Figure 2.

Fig. 1.

Schematic illustration of the sample processing. Whole blood was collected from five healthy donors, from which platelet-rich plasma (PRP) was prepared. Platelets in PRP were stimulated with one of three stimulants to release EVs. The platelets were then centrifuged, and EVs from the platelet-poor plasma were isolated using acoustic trapping. The isolated EVs were analyzed using mass spectrometry, high-sensitivity flow cytometry, transmission electron microscopy (TEM), immunoblots, and stimulation of whole blood.

Fig. 2.

M1 protein mediates platelet activation and release of EVs that are enriched by acoustic trapping. Platelet activation by the M1 protein was investigated using flow cytometry of platelet-rich plasma (PRP) from five healthy donors (n = 5) (a). Platelet activation is presented as percent platelets positive for CD62P. ** p < 0.01. Mann-Whitney test was used. Concentration of platelet-derived EVs in plasma from four healthy donors (b) and EVs after isolation with acoustic trapping (c) was determined using high-sensitivity flow cytometry. EV concentration is presented as EVs positive for CD42b/µL. Thrombin was used as a positive control for platelet activation and EV release, and HEPES buffer was used as a control for background platelet activation. Mann-Whitney test was used, sample distribution was compared to background platelet activation in resting platelets (HEPES). ns, not significant.

Proteome Profiling of Acoustically Enriched EVs

EVs derived from platelets stimulated with the three distinct agonists were isolated on the acoustic trapping platform and analyzed with quantitative mass spectrometry, in parallel with plasma samples of non-isolated EVs. In total, 557 proteins were identified. Proteins that were identified with only one peptide, along with all keratin proteins, were discarded, resulting in 344 remaining proteins. Protein intensities for the different samples were TIC normalized and plotted in a heatmap (Fig. 3a). Cluster I showed a weaker protein signal in the isolated EVs as compared to EVs in plasma, suggesting that these proteins were part of the background plasma signal that is washed away in the acoustic trap. Cluster II showed a stronger protein signal in the isolated EVs as compared to EVs in plasma, indicating proteins that were associated with vesicles and were enriched in the acoustic trap. It should be noted that since the protein intensities have been TIC normalized, they represent a fraction of the total signal and not an absolute value. A larger version of the heatmap where individual proteins are illustrated is available (online suppl. Fig. 3).

Fig. 3.

Establishing a pipeline for proteome profiling of isolated EVs. EVs derived from resting platelets (HEPES), and thrombin- or M1 protein-activated platelets were isolated on the acoustic trapping platform and analyzed with quantitative mass spectrometry in parallel with plasma samples of non-isolated EVs from four healthy donors (n = 4). a Heatmap of protein intensities for all samples. The intensities shown are the log 10 of the TIC normalized intensities. The proteins are row normalized, and the legend gives the z-score. Cluster I indicates proteins with lower intensity in isolated EVs, compared to EVs in plasma. Cluster II indicates proteins with higher signal in isolated EVs. b Gene enrichment analysis of proteins in cluster II, performed using Metascape. c Heatmap of protein intensities for isolated EV samples. Cluster III indicates proteins with higher intensity in EVs from platelets stimulated with thrombin or M1. Cluster IV shows higher intensity in EVs from resting platelets (HEPES). d Gene enrichment analysis of proteins in cluster III, performed using Metascape.

Metascape was used to perform gene enrichment analysis (Fig. 3b) on the 288 proteins found in cluster II. These proteins were associated with hemostasis, complement, and coagulation cascades. To further investigate the effects of thrombin or M1 protein activation on released platelet EVs compared to resting platelets, TIC normalized protein intensities for isolated EVs were plotted in a heatmap (Fig. 3c). Cluster III shows a stronger protein signal for EVs from thrombin- or M1 protein-activated platelets compared to the resting state (HEPES), suggesting these proteins are associated with platelet activation. Cluster IV shows proteins with a stronger signal for EVs from resting platelets, suggesting these proteins are normally found in platelet EVs. A gene enrichment analysis (Fig. 3d) was performed using Metascape software on the 183 proteins found in cluster III. The analysis showed a strong expression of proteins associated with complement, coagulation cascades, and platelet degranulation.

Distinct Proteins from Coagulation and Complement Systems Are Upregulated in EVs Derived from Pathogen-Stimulated Platelets

TIC normalized intensities of proteins associated with the groups that were revealed through the Metascape analysis were investigated further to assess differences in protein cargo of EVs derived from resting platelets, M1 protein- and thrombin-activated platelets. Several proteins associated with platelet activation and platelet aggregation were observed in platelet-derived EVs from all agonists (Fig. 4a, b). Integrin alpha-IIb (CD41), platelet glycoprotein V (GPV), integrin beta-3 (CD61), and P-selectin levels are slightly higher in EVs isolated from resting platelets compared to EVs isolated from platelets stimulated with thrombin and M1 protein. Platelet factor 4 (PF4) and platelet glycoprotein Ib (GPIb) were found at equivalent levels in all EV populations (Fig. 4a, b).

Fig. 4.

Distinct proteins associated with blood coagulation and complement system activation are upregulated in EVs derived from platelets stimulated with M1 protein. EVs derived from resting platelets (HEPES), and thrombin- or M1 protein-activated platelets from four healthy donors (n = 4) were isolated on the acoustic trapping platform and analyzed with quantitative mass spectrometry. Protein intensities are presented as TIC normalized intensities of proteins associated with platelet activation (a, b), blood coagulation (c), complement system activation (d), fibrinogen (e) and IgG (f). Mann-Whitney test was used; sample distribution was compared to EVs derived from resting platelets (HEPES). * p < 0.05; ** p < 0.01; ns, not significant.

Furthermore, proteins associated with blood coagulation were also present in platelet-derived EVs (Fig. 4c). Plasminogen was found at equivalent levels in all EV populations, whereas prothrombin and von Willebrand factor levels were slightly lower in EVs isolated from platelets stimulated with M1 protein compared to EVs isolated from platelets stimulated with thrombin and resting platelets (Fig. 4c).

Complement components and regulators of complement system activation were found in platelet-derived EVs (Fig. 4d). The majority of complement and complement regulatory proteins were significantly upregulated in EVs derived from platelets stimulated with M1 protein and thrombin, compared with resting platelets. Only complement component C5 was found at equivalent levels in all EV populations. Interestingly, the complement regulatory protein C4BP and the complement components C1q and C9 were significantly upregulated in EVs derived from platelets stimulated with M1 protein, as compared to both resting and thrombin-stimulated platelets.

Fibrinogen and all four subclasses of IgG were found in platelet-derived EVs (Fig. 4e, f). Fibrinogen was significantly upregulated in EVs derived from platelets stimulated with M1 protein and thrombin, compared to resting platelets (Fig. 4e). Furthermore, IgG1-2 and IgG4 were upregulated in EVs derived from platelets stimulated with M1 protein and thrombin, compared to resting platelets (Fig. 4f). Interestingly, IgG3 was significantly upregulated in EVs derived from platelets stimulated with M1 protein, as compared with both resting and thrombin-stimulated platelets (Fig. 4f). This demonstrates that the IgG3 required for platelet activation by M1 protein was also enriched in the vesicles released upon generation of EVs from activated platelets.

M1 Protein Was Packaged within EVs

In addition to human proteins associated with the groups that were revealed through the Metascape analysis, the TIC normalized intensity of M1 protein transported by EVs was investigated. The M1 protein was significantly enriched in EVs from platelets stimulated by M1 protein, as compared with non-trapped plasma samples stimulated with M1 protein (Fig. 5a). This suggests that M1 protein can be packaged into EVs during budding and transported with platelet-derived EVs in the circulation. Alternatively, M1 protein may associate directly with EVs released from activated platelets. The presence of M1 protein was confirmed using immunoblotting (Fig. 5b) and transmission electron microscopy (Fig. 5c). Both of these techniques determined that EVs from M1 protein-activated platelets contained the M1 protein (Fig. 5b, c).

Fig. 5.

M1 protein is transported in platelet-derived EVs after stimulation. EVs derived from resting platelets (HEPES), and thrombin- or M1 protein-activated platelets were isolated from four healthy donors (n = 4) on the acoustic trapping platform and analyzed with quantitative mass spectrometry (a), immunoblot (b), and transmission electron microscopy (c). a Protein intensity is presented as TIC normalized intensity fold increase of M1 protein in isolated EVs compared to non-trapped plasma controls. * p < 0.05; ns, not significant, Mann-Whitney test, sample distribution was compared to EVs derived from resting platelets (HEPES). b Immunoblot against the M1 protein. 1 = protein ladder, 2, 5, 8 = EVs derived from resting platelets (HEPES), 3, 6, 9 = EVs derived from thrombin-activated platelets, 4, 7, 10 = EVs derived from M1 protein-activated platelets. 2–4 = EVs isolated on the acoustic trapping platform and eluted in 25 μL PBS, 5–7 = supernatant of EVs isolated using ultracentrifugation, 8–10 = EVs isolated using ultracentrifugation. c Isolated EVs were stained with gold-labeled antibodies against the platelet-specific marker CD42b (10 nm) and against the M1 protein (15 nm).

Acoustically Enriched EVs Mediated Platelet-Neutrophil Complex Formation and Neutrophil Activation

Platelet activation results in platelet-neutrophil complex formation [22], which was investigated using flow cytometry. EVs were isolated from resting or stimulated platelets using acoustic trapping and added to whole blood from four healthy donors. The positive controls of thrombin and M1 protein alone mediated significant platelet-neutrophil complex formation, with a median level of 88 and 78% platelet-positive neutrophils, respectively (Fig. 6a). EVs from platelets stimulated with M1 protein and thrombin mediated significantly increased platelet-neutrophil complex formation with a median level of 55 and 51% platelet-positive neutrophils, as compared with the background level of 22%. Isolated EVs from resting platelets also mediated platelet-neutrophil complex formation, with a median level of 45% platelet-positive neutrophils. Interindividual variation was more pronounced in samples treated with platelet-derived EVs as compared with agonists alone. M1 protein alone is a more potent mediator of platelet activation and platelet-neutrophil complex formation than EVs isolated from M1 protein-activated platelets.

Fig. 6.

Platelet-derived EVs mediate platelet-neutrophil complex formation and neutrophil activation. Flow cytometry was used to assess platelet-neutrophil complex (PNC) formation (a) and neutrophil activation (b) by the isolated EVs from resting platelets (HEPES), and from platelets activated with thrombin or M1 protein in blood from four healthy donors (n = 5). Data are presented as % neutrophils that are CD61 positive (platelet associated) (a) and neutrophil CD11b median fluorescence intensity (b). Thrombin and M1 protein are used as positive controls for platelet-neutrophil complex formation and neutrophil activation. HEPES buffer is used as a control for background PNC formation and neutrophil activation. Mann-Whitney test was used; sample distribution was compared to background levels in resting platelets (HEPES). * p < 0.05; ** p < 0.01; ns, not significant.

In parallel analyses in blood samples from the same four donors as above, neutrophil activation was investigated. The positive controls of thrombin and M1 protein alone mediated the highest neutrophil activation with a median CD11b fluorescence intensity of 82,000 and 45,000, respectively (Fig. 6b). EVs isolated from platelets stimulated with M1 protein and thrombin mediated significantly increased neutrophil activation, with a median CD11b fluorescence intensity of 27,000 and 29,000, respectively, compared to the background level of 15,000. Isolated EVs from resting platelets also mediated neutrophil activation, with a median CD11b fluorescence intensity of 24,000. Collectively, this indicates that platelet EVs derived from all three conditions mediate neutrophil activation at equivalent levels.

Acoustically Enriched EVs Mediated Cytokine Release from Blood Cells

Cytokine release in response to isolated platelet-derived EVs was measured in whole blood using commercial ELISA kits for interleukin-8 (IL-8), interleukin-6 (IL-6), and monocyte chemoattractant protein-1 (MCP-1). These pro-inflammatory cytokines were selected as representative of early acute inflammation: IL-6 is a pleiotropic cytokine and critical mediator of the acute phase response, IL-8 is a predominant chemokine for neutrophils, and MCP-1 is a potent chemokine for monocytes. As expected, M1 protein alone significantly increased levels of IL-8, IL-6, and MCP-1 in whole blood, compared to background levels (Fig. 7a–c). Isolated EVs derived from both resting and M1 protein stimulation generated a statistically significant release of IL-8 when compared to background levels; however, the IL-8 levels generated by EVs derived from M1 protein-stimulated cells were relatively low as compared with EVs from resting platelets (Fig. 7a). Isolated EVs from resting platelets resulted in significantly increased levels of IL-6 and a tendency toward MCP-1 release in whole blood, while M1 protein- or thrombin-stimulated samples failed to mediate release of these cytokines (Fig. 7b). Collectively, our findings indicate that platelet-derived EVs mediate cytokine release in whole blood; however, the interindividual variation is high. EVs derived from M1 protein-stimulated platelets were generally weak mediators of pro-inflammatory cytokine release in whole blood, while the M1 protein alone was a potent mediator of cytokine release.

Fig. 7.

Platelet-derived EVs mediate cytokine release from monocytes in whole blood. Cytokine release mediated by the EVs from resting platelets (HEPES), and from platelets activated with thrombin or M1 protein in blood from five healthy donors, was investigated using commercial IL-8 (a), IL-6 (b), and MCP-1 (c) kits. Cytokine levels are presented as pg/mL. M1 protein was used as a positive control for cytokine release and HEPES buffer was used as a control for background cytokine release. Mann-Whitney test was used. * p < 0.05; ** p < 0.01; ns, not significant.

Discussion

In this study, we have determined that platelets release EVs upon stimulation with a virulence factor, M1 protein, from a significant bacterial pathogen. Intact EVs were isolated and enriched using a newly developed acoustic trapping platform, as confirmed with high-sensitivity flow cytometry, immunoblotting, and electron microscopy. Differential protein abundance profiling revealed distinct differences between intact plasma with EVs and EVs isolated from plasma using acoustic trapping, with one set of proteins washed away in the trap and one set of proteins enriched, indicating that they were specific to the EV proteome. It should be noted that the EV characterization with flow cytometry only detected EVs in the size range 100–1,000 nm, whereas the mass spectrometry measurements were derived from the entire EV population. Characterization of the protein cargo of EVs derived from pathogen-activated platelets, M1 protein, as compared with resting or thrombin-activated platelets, revealed that all three EV populations were enriched for platelet membrane proteins, granule proteins, and cytoskeletal proteins, in combination with coagulation factors and immune mediators. The functional integrity of EVs isolated using acoustic trapping was confirmed by their ability to stimulate platelet and neutrophil activation and cytokine release when incubated with blood cells.

The plasma levels of platelet-derived EVs did not increase significantly upon stimulation with thrombin or M1 protein under our experimental conditions. Previous studies have demonstrated that in vitro levels of EVs differ depending on the platelet agonist [15, 38, 39]. Thrombin is a powerful mediator of EV release, and we have determined that M1 protein generates EVs at equivalent levels in certain individuals. Under our experimental conditions, the resting platelet population also generated EVs which likely reflects in vitro activation of the platelets during centrifugation at 1,600 g in the absence of platelet inhibition. Platelets are highly reactive cells, and a heterogeneous interindividual response is to be expected upon in vitro stimulation of platelets from healthy individuals [2].

It is well known that platelets constantly scan the vasculature for areas of injury or inflammation or indirectly respond to pathogen invasion through interactions with leukocytes and the endothelium [40]. Previous studies have also shown increased levels of circulating platelet-derived EVs in several pathological states, in particular as biomarkers of cardiovascular disease [11]. Sepsis is a multifactorial syndrome involving immune and coagulation dysregulation, and platelet-derived vesicles are elevated in sepsis [41, 42]. It has been reported that EVs may have a protective role during sepsis and that platelet-derived EV levels are decreased in sepsis non-survivors. For example, EVs enhance the sensitivity of contraction of mouse aorta in response to serotonin and may protect against vascular hyporeactivity accounting for hypotension in sepsis patients [43, 44]. There were significant differences in the protein cargo of EVs isolated from resting platelets, physiologically stimulated (thrombin) platelets, and pathogen (M1 protein)-stimulated platelets. The observed differences were mainly quantitative, as the protein content within the distinct EV populations was very similar, but the protein enrichment levels differed significantly in some cases. Previous studies have shown that the proteins packaged within platelet-derived EVs differ depending on the platelet stimulus [15, 16]. Our findings demonstrate an increased level of blood coagulation proteins after stimulation with thrombin and M1 protein, as compared with resting platelets. This is in line with previous studies that reported increased procoagulant activity of EVs derived from activated platelets [45, 46]. Procoagulant EVs may contribute to the coagulopathy in sepsis [47], which may be an important attribute for M1 protein during invasive streptococcal disease. Furthermore, we have previously demonstrated that other M protein serotypes that can bind fibrinogen can mediate platelet activation [23]; therefore, the findings presented herein are likely applicable to other serotypes of S. pyogenes.

Increased levels of complement components were also observed after stimulation with both thrombin and M1 protein; however, this was further enhanced in the case of M1 protein. All four subclasses of IgG were also increased after stimulation; however, the level of IgG3 was specifically increased after stimulation with M1 protein. We have previously demonstrated that IgG3 is particularly enriched in immune complexes formed by M1 protein, resulting in C1q-mediated complement activation on M1 protein-activated platelets [20]. Fabris and colleagues [48] have shown that platelet-derived EVs in patients with sepsis contain platelet factor 4 and anti-bacterial IgG, which may also contribute to immune complex-mediated immune responses during infection.

Platelet-derived EVs can exhibit distinct immunomodulatory properties that are dependent on the cargo incorporated within the EVs, thereby encompassing both pro- and anti-inflammatory effects. The pro-inflammatory potential of platelet-derived EVs was investigated in our study. We determined that platelet-neutrophil complex formation, neutrophil activation, and cytokine release occurred in whole blood after incubation with platelet EVs. The relatively low number of donors (n = 5) investigated and the heterogenous inter-interindividual variation observed are, however, an important limitation to the study; therefore, we cannot conclude whether distinct pro-inflammatory profiles are generated for the three conditions investigated. The formation of platelet-neutrophil complexes has been previously observed in sepsis [49]. In addition, platelet-derived EVs have been reported to directly associate with leukocytes and facilitate chemotaxis of leukocytes [9, 50]. M protein stimulates platelet-neutrophil complex formation, resulting in functional impairment of these neutrophils with decreased chemotaxis and phagocytosis [22]. The present findings imply that platelet-derived EVs may also contribute to the immunomodulatory effects of M1 protein on neutrophils; however, M1 protein incorporated in EVs is a less potent pro-inflammatory mediator as compared with M1 protein alone under all conditions investigated.

Importantly, the M1 protein was itself associated with isolated EVs, which shows that M1 protein can be transported with platelet-derived EVs after stimulation. This mechanism of bacterial toxin transfer with host cell EVs has been previously described for Shiga toxin from enterohemorrhagic E. coli [51]. Interestingly, Shiga toxin is transported with host blood cell-derived EVs and taken up by renal cells, where it contributes to the pathogenesis of kidney failure. EVs are released from platelets stimulated with LPS from Gram-negative bacteria, and these EVs derived from LPS/TLR4 activation of platelets can stimulate further activation of endothelial cells [52], which may represent an important effector of vascular dysfunction that is a hallmark of sepsis.

Collectively, we have determined the phenotype of platelet-derived EVs that are isolated and enriched using acoustic trapping. Our subsequent proteomic profiling showed that blood coagulation proteins, complement components, and IgG were enriched in all EV populations irrespective of agonist. A specific increase in complement components and IgG3 was observed after pathogen-mediated platelet activation (M1 protein), and the bacterial virulence factor, M1 protein, was transported within platelet-derived EVs. Equivalent levels of pro-inflammatory responses were generated in blood cells exposed to the acoustically enriched EVs irrespective of agonist.

Acknowledgments

Lund University Bioimaging Centre (LBIC) is gratefully thanked for providing experimental resources. Sara Wettemark is thanked for excellent technical assistance with blood cell analyses. Figure 1 was created with BioRender.

Statement of Ethics

Blood was collected from healthy donors by trained personnel. Written informed consent was obtained from the donors prior to blood collection. This study protocol was reviewed and approved by the Regional Ethical Review Authority, Lund (approval number 2015/801).

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This research was supported by grants from Knut and Alice Wallenberg Foundation: 2016.0023 (O.S. and J.M.); WAF 2017.0271 (J.M.); the Swedish Research Council: 2018-05795 (O.S. and J.M.); 2019-00795 and 2018-03672 (T.L.); and 2021-01310 (J.W.S). G.M. is the recipient of a postdoctoral fellowship from Fonds de Recherche du Québec-Santé (FRQS).

Author Contributions

F.P. and A.B. performed experiments, analyzed the data, and wrote the manuscript. G.M. and J.W.S. designed, performed, and analyzed high-sensitivity flow cytometry of extracellular vesicles and edited the manuscript. O.S., T.L.L., and J.M. designed and supervised the study and edited the manuscript. All authors read and approved the manuscript for submission.

Funding Statement

This research was supported by grants from Knut and Alice Wallenberg Foundation: 2016.0023 (O.S. and J.M.); WAF 2017.0271 (J.M.); the Swedish Research Council: 2018-05795 (O.S. and J.M.); 2019-00795 and 2018-03672 (T.L.); and 2021-01310 (J.W.S). G.M. is the recipient of a postdoctoral fellowship from Fonds de Recherche du Québec-Santé (FRQS).

Data Availability Statement

All data generated or analyzed during this study are included in this article and its supplementary material files. Further inquiries can be directed to the corresponding author.