In response to bacteria, neutrophils release extracellular vesicles capable of initiating thrombin generation through DNA-dependent and independent pathways

Abstract

Neutrophils release extracellular vesicles, and some subsets of neutrophil-derived extracellular vesicles are procoagulant. In response to Staphylococcus aureus, neutrophils produce extracellular vesicles that associate electrostatically with neutrophil extracellular traps. DNA in neutrophil extracellular traps is procoagulant, but whether neutrophil extracellular vesicles produced during bacterial challenge have similar activity is unknown. Given that extracellular vesicle activity is agonist and cell-type dependent and coagulation contributes to sepsis, we hypothesized that sepsis-causing bacteria increase production of neutrophil-derived extracellular vesicles, as well as extracellular vesicle–associated DNA, and intact extracellular vesicles and DNA cause coagulation. We recovered extracellular vesicles from neutrophils challenged with S. aureus, Staphylococcus epidermidis, Escherichia coli, and Pseudomonas aeruginosa and measured associated DNA and procoagulant activity. Extracellular vesicles from S. aureus–challenged neutrophils, which were previously characterized, displayed dose-dependent procoagulant activity as measured by thrombin generation in platelet-poor plasma. Extracellular vesicle lysis and DNase treatment reduced thrombin generation by 90% and 37%, respectively. S. epidermidis, E. coli, and P. aeruginosa also increased extracellular vesicle production and extracellular vesicle–associated extracellular DNA, and these extracellular vesicles were also procoagulant. Compared to spontaneously released extracellular vesicles, which demonstrated some ability to amplify factor XII–dependent coagulation in the presence of an activator, only extracellular vesicles produced in response to bacteria could initiate the pathway. S. aureus and S. epidermidis extracellular vesicles had more surface-associated DNA than E. coli and P. aeruginosa extracellular vesicles, and S. aureus and S. epidermidis extracellular vesicles contributed to initiation and amplification of thrombin generation in a DNA-dependent manner. However, DNA on E. coli or P. aeruginosa extracellular vesicles played no role, suggesting that neutrophils release procoagulant extracellular vesicles, which can activate the coagulation cascade through both DNA-dependent and independent mechanisms.

Neutrophil extracellular vesicles associate with DNA and contribute to coagulation through DNA-dependent and independent mechanisms.

1. Introduction

Sepsis is usually caused by bacterial infections and is defined as a dysregulated host response to infection or injury, and it results in hyperinflammation, immunosuppression, coagulopathy, and organ dysfunction.^1–5 Anticoagulants, including activated protein C, have been used as a treatment but were discontinued due to bleeding complications.⁶ Thus, identifying the initial causes of coagulation is essential to develop safe and effective treatments. Based on this, we focused on 2 neutrophil products that are elevated in sepsis, are involved in coagulation, and correlate with disease severity: extracellular vesicles (EVs) and neutrophil extracellular traps (NETs).^7–10 EVs are a heterogeneous group of small, membrane-bound, cell-derived vesicles that carry all types of macromolecules. EVs differ in their molecular composition based on the releasing cell type and the environment in which EV generation occurs.

Depending on these factors, EVs have been shown to activate 1 of 2 coagulation pathways: the tissue factor (TF) pathway or contact-dependent pathway.^11–15 The tissue factor pathway is initiated when TF, a membrane-bound protein, is exposed on the surface of endothelial cells, monocytes, or macrophages.^16–18 Once exposed, the serine protease factor VII anchors itself to TF and initiates the coagulation cascade. In contrast, the contact-dependent pathway is activated through the activation of circulating coagulation factor XII. This process requires high molecular weight kininogen and plasma kallikrein. Factor XII normally circulates in an inactive form, but when it comes into contact with a negatively charged surface, it undergoes a conformational change, gains proteolytic activity, and is cleaved into factor XIIa.¹⁹ Negatively charged proteins on the cell surface, extracellular nucleic acids, inorganic polyphosphates, glycosaminoglycans, and bacterial surface proteins have all been shown to activate factor XII.^20–26 These 2 pathways converge into a common pathway, which involves the cleavage of prothrombin into thrombin by factor Xa and its cofactors. Thrombin, in turn, transforms soluble fibrinogen into insoluble fibrin, creating a mesh that solidifies a blood clot.¹⁹

EVs from monocytes treated with lipopolysaccharide possess TF and drive coagulation through the TF pathway, whereas the procoagulant activity of platelet-derived EVs is tissue factor independent.¹¹ However, understanding role of neutrophil EVs in coagulation is more nuanced. First, EVs associate with extracellular DNA in vitro and in vivo.^15,27–29 In mice, DNA-containing NETs interact with myeloid-derived EVs, and in humans, extracellular DNA has been found on EVs isolated from plasma and urine samples.^30,31 NETs are composed of DNA coiled around histones and studded with enzymes derived from neutrophil granules, and their production has been implicated in dysregulated coagulation.^32,33 Mechanistically, histones trigger platelet activation, whereas in platelet-poor plasma (PPP), negatively charged DNA activates the contact-dependent coagulation pathway.^34–36 Although questions persist regarding whether DNA may be obscured by proteins in intact NETs,³⁶ NETs on EVs released from phorbol 12-myristate 13-acetate (PMA)-treated murine bone marrow neutrophils cause DNA-dependent coagulation.¹⁵

Second, there is also evidence that the procoagulant activity of neutrophil EVs is unrelated to NETosis. For example, Kolonics et al.³⁷ demonstrated that EVs produced in response to zymosan failed to trigger coagulation, whereas EVs released from resting and apoptotic neutrophils could. The mechanism for this activity remains undefined, but these data suggest a role for EVs themselves rather than expelled DNA in coagulation. Since EVs and NETs are released concurrently from neutrophils and both elements have thrombogenic properties, there is a need to identify whether bacterial-induced EVs have the capacity to independently support thrombin generation (TG) or if this function is attributed to NET-DNA.^15,28,38

Neutrophils release EVs in response to Staphylococcus aureus, the most common cause of Gram-positive sepsis, and these EVs associate with NET-DNA through electrostatic interactions.^1,27,28,39 Therefore, we hypothesized that neutrophil EVs produced in response to sepsis-causing bacteria would associate with DNA on their surface and promote contact- and DNA-dependent coagulation. To test this, we isolated and characterized EVs produced following neutrophil challenge with S. aureus, Staphylococcus epidermidis, Escherichia coli, or Pseudomonas aeruginosa; quantified EV-associated DNA; and measured their ability to initiate and/or amplify the coagulation pathway, as measured by TG. All neutrophil EVs (including spontaneously released EVs) amplified factor XII–dependent coagulation in PPP containing an activator. However, only neutrophil EVs produced during bacterial challenge had the capacity to initiate TG, as well. Although more extracellular DNA associated with EVs generated during bacterial challenge, only EV subsets isolated following challenge with S. aureus and S. epidermidis relied on the presence of DNA for procoagulant activity. These data demonstrate that EVs cause coagulation through multiple mechanisms, including pathways that do not involve extracellular DNA. Overall, this work highlights the importance of studying EV heterogeneity, since discovery of differential mechanisms will provide better insights into designing therapeutics for etiologically diverse conditions, such as sepsis.⁴⁰

2. Methods

2.1. Ethics statement

Written consent was obtained from volunteers in accordance with a protocol approved by the Institutional Review Board and Institutional Biosafety Committee at Central Michigan University (Mount Pleasant, MI, USA) and California Polytechnic University San Luis Obispo (San Luis Obispo, CA, USA).

2.2. Bacterial culture

S. aureus USA300 LAC (gifted from Dr. DeLeo, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases) and S. epidermidis 1457 (gifted from Dr. Alex Horswill, University of Colorado Anschutz Medical Campus) were grown on tryptic soy agar and in tryptic soy broth. E. coli 25922 (ATCC) was grown on Lenox L agar and in Luria broth. P. aeruginosa 19660 (ATCC) was grown on Todd–Hewitt agar supplemented with 5% yeast and in Brain Heart Infusion Broth. Single colonies were inoculated into broth and grown overnight at 37°C with shaking (180 rpm). Overnight cultures were diluted in broth to an OD550 of 0.05 and incubated for 3 h with shaking to reach mid-logarithmic growth. Bacteria were pelleted at 2,800 × g for 5 min and resuspended in 20 mM HEPES-buffered HBSS^+/+. The bacteria were then opsonized in 10% human serum while tumbling for 20 min at 37°C. Opsonized bacteria were pelleted and washed twice with RPMI 1640 without phenol red and resuspended at 2 × 10⁸ colony-forming units (CFU)/mL.

2.3. Neutrophil isolation

Neutrophils from venous blood of healthy human donors were isolated as previously described.⁴¹ Briefly, heparinized blood was subject to dextran sedimentation (Pharmacosmos), followed by density gradient separation using Ficoll Paque PLUS (Cytiva) and hypotonic lysis of red blood cells. Neutrophil purity was determined to be >95% by Hema-3 staining in all experiments, and for EV isolation, neutrophils were resuspended at 2 × 10⁷ cells/mL in RPMI 1640 without phenol red.

2.4. EV isolation from neutrophils

Bacteria were cultured to mid-logarithmic phase, opsonized in 10% human serum, washed twice with RPMI 1640 without phenol red, and resuspended at 2 × 10⁸ CFU/mL as described in supplemental methods. Neutrophils (2.2 to 6.6 × 10⁷) were either challenged with preopsonized bacteria at a multiplicity of infection (MOI) of 10 (10 bacteria per 1 neutrophil) or diluted in buffer to achieve a cell density of 1 × 10⁷ cells/mL. Cells were incubated for 20 min at 37°C with gentle inversion every 5 min. EV production was halted by placing cells on ice for 10 min. After stimulation, lactate dehydrogenase (LDH) and extracellular DNA release from neutrophils was measured using the Pierce cytotoxicity kit and Sytox Green staining, respectively, as previously described.³² For EV isolation, neutrophils and bacteria were pelleted twice at 4,000 × g for 20 min at 4°C, and the EV-containing supernatant was collected. EVs were pelleted at 160,000 × g for 51 min in a TH-641 swinging bucket rotor (ThermoFisher) at 4°C, pooled, and pelleted again. Isolated EVs were resuspended in PBS^−/− and stored at −80°C. In some experiments, EV-containing supernatants were treated with 0.05% Triton X-100 (TX-100) and vortexed for 30 s to lyse EVs or were passed through a 0.2-µm polyethersulfone (PES) filter that removes large EVs, surface DNA, and bacteria.²⁷ In other experiments, neutrophils were omitted during the process of generating EVs, allowing recovery of any nanosized bacteria-derived products. Protein in preparations was quantified with the Pierce BCA protein assay (ThermoFisher), and EV-associated bacteria were enumerated from serial dilutions incubated overnight on agar plates at 37°C.

2.5. GW4869 treatment

GW4869 was used to inhibit sphingomyelinase, an enzyme involved in ceramide production that contributes to exosome formation, a subpopulation of extracellular vesicles.^42,43 First, primary neutrophils were pretreated with 10 µM GW4869 or vehicle for 2 h at 37°C prior to bacterial challenge. EVs were isolated as described above. Alternatively, HL-60 cells (1 × 10⁶ cells/mL) were differentiated (dHL-60) with 1% dimethylsulfoxide for 5 d. To isolate EVs, 8 × 10⁷ dHL-60s were collected by centrifugation at 200 × g for 5 min, resuspended in RPMI 1640 without phenol red to achieve a density of 2 × 10⁷ cells/mL, and pretreated with 20 µM GW4869 or vehicle and incubated for 2 h at 37°C. Posttreatment, dHL-60s were cocultured at a cell density of 1 × 10⁷ cells/mL for 20 min at 37°C at an MOI of 5 with opsonized S. aureus (SA). dHL-60s and SA were pelleted at 4,000 × g for 20 min at 4°C. EV-containing supernatants were passed through a 0.2-µm PES filter and pelleted at 160,000 × g for 51 min in a SW 55 Ti swing bucket rotor (Beckman Coulter). EVs were resuspended in 200 µL PBS^−/− and snap frozen for future use.

2.6. EV-DNA analysis

EV-DNA was isolated using the Qiagen DNeasy Kit and visualized on a 0.7% agarose gel stained with GelRed. Exact copy number of nuclear, mitochondrial, and bacterial DNA was quantified with droplet digital polymerase chain reaction as described previously.²⁸ In some experiments, EVs were incubated with 0.5 U/µL DNase I or vehicle for 1 h at 37°C. Following treatment, 10 mM EDTA was added to stop DNase activity. Primers used to amplify EV-DNA are listed in Supplementary Table 1. The concentration of EV surface DNA was quantified using the Quant-iT PicoGreen fluorescent staining kit based on manufacturer instructions.

2.7. Thrombin generation assay

To obtain PPP, venous blood was collected from healthy human donors and diluted into acid citrate dextrose for a final concentration of 10%. Platelets were removed by double centrifugation at 1,560 × g for 20 min⁴⁴ and filtration through a 0.2-µM PES filter.³⁶ Thrombin generation in PPP was determined according to methods previously described with some modifications.⁴⁵ Briefly, a 10-µL EV suspension (1.5 or 3 µg) or PBS was added to 30 µL PPP, followed by the addition of TF-phospholipid reagent (10 µL; Technothrombin RCL). To ascertain whether contaminating bacteria were accountable for EV-mediated TG, bacteria were cultured to mid-logarithmic growth, opsonized, resuspended at 10⁷ CFU/mL, and flash frozen. After thawing the bacterial suspensions, 10 µL of the suspension was used to assess TG. For some experiments, lyophilized human factor VII–deficient, factor XII–deficient, and control plasma (Technoclone) were reconstituted in sterile water, centrifuged at 1,560 × g, and filtered prior to use. In other experiments, EVs were modified postisolation. EV-associated DNA was removed by incubating with 0.5 U/µL DNase I for 1 h at 37°C. DNase activity was halted with 10 mM EDTA, and samples were used in the thrombin generation assay (TGA). Phosphatidylserine (PS) was blocked by incubating EVs with recombinant Annexin V (BioLegend) for 60 min at room temperature. Following incubation, EVs were washed at 160,000 × g in an AH-650 swinging bucket rotor (ThermoFisher) for 25 min, resuspend in PBS, and used in the TGA. As described previously,⁴⁶ anti-CD66b-conjugated Dynabeads (Invitrogen) were used to deplete CD66b⁺ EVs from EVs isolated from dHL-60 cells that had been challenged with S. aureus. Following elution from beads, both CD66b⁺ and CD66b⁻ fractions were pelleted at 160,000 × g in an AH-650 swinging bucket rotor (ThermoFisher) for 25 min, resuspend in PBS, and used in the TGA.

2.8. Transmission electron microscopy

Negative-positive staining was performed as previously described.⁴⁷ EVs preserved in 2% paraformaldehyde at −80°C were thawed and fixed with 1% glutaraldehyde. Fixed EVs were placed on formvar-coated copper grids for 20 min. After 5 washes, grids were negatively stained with methylcellulose (2%)/uranyl acetate (4%) in a 9:1 ratio mixed just prior to use. This was followed by positive staining with uranyl oxalate. Following staining, grids were air-dried for at least 2 h before imaging on a Hitachi 7700 transmission electron microscope. Three EV preparations were imaged per group, and representative images were selected from each imaged EV subset.

2.9. EV characterization

EVs were characterized according to the guidelines set by the Minimal Information for the Studies of Extracellular Vesicles (MISEV).⁴⁸ All flow cytometry was performed on a Beckman Coulter CytoFLEX Flow Cytometer. EVs (5 µg) were conjugated to aldehyde/sulfate latex beads³⁵ and labeled with 3 µM CellVue Claret Far Red (Sigma), washed, and stained with either Annexin V–FITC (BioLegend) or propidium iodide (PI) (Sigma) for 15 min at room temperature or with anti–CD63-PE (BioLegend, H5C6) for 30 min on ice. In some experiments, EVs were incubated with recombinant Annexin V (BioLegend) prior to bead conjugation. For immunoblotting, whole-cell lysates (WCLs) were prepared by incubating neutrophils in RIPA buffer²⁸ or by sonication⁴¹ and then removing insoluble material by centrifugation. Proteins in EVs and WCL were resolved by gel electrophoresis on a 4% to 15% polyacrylamide gel and transferred onto a PVDF membrane using a Trans-Blot Turbo Transfer System (Bio-Rad). The following antibodies were used for blotting: anti–flotillin 1 (1:20,000, EPR6041; Abcam), anti–glucose-regulated protein 94 (GRP94) (1:1,000, PA5-27866; ThermoFisher), anti-myeloperoxidase (1:1,000, E1E7I; Cell Signaling), anti–neutrophil elastase (1:20,000, JF098-6; ThermoFisher), and anti–histone H3 (1:1,000, 07-690; Sigma).

2.10. Nanoparticle tracking analysis

Nanoparticle tracking analysis was performed by Alpha Nano Tech. Analysis was performed on a Zetaview Quatt using the 488-nm laser and sCMOS camera.

2.11. Statistical analysis

Statistical analyses were performed between 3 or more biological replicates (n) using Prism 10.0.2 (GraphPad Software). The statistical tests used for each figure are indicated in the corresponding legend.

3. Results

3.1. Neutrophil EVs produced in response to S. aureus enhance thrombin generation

Whether neutrophil EVs produced during bacterial challenge are procoagulant or contribute to thrombosis is unknown. When neutrophils ingest opsonized S. aureus, increased EV production coincides with vital NETosis, a process monitored by the presence of extracellular DNA, and the absence of LDH in the supernatant.^28,39 Hereafter, these EVs are referred to as SA-EVs to denote the type of bacterial challenge. Given our prior data demonstrating the presence of citrullinated histone 3 and neutrophil granule proteins on EVs, the electrostatically attached DNA are referred to as NETs.^27,28 To test whether SA-EVs and/or associated NETs were procoagulant, we treated PPP with SA-EVs and measured TG over time. SA-EVs elevated the amount of TG in a dose-dependent manner (Fig. 1A) and 10 and 30 µg/mL SA-EVs significantly increased peak and total thrombin (Fig. 1B and C). Thus, SA-EVs are procoagulant and therefore may contribute to coagulation abnormalities during sepsis.

Fig. 1.

EVs from S. aureus–challenged neutrophils induce coagulation in a dose-dependent manner. PPP was treated with either buffer or SA-EVs at the concentrations indicated, and TG was subsequently measured. Data are depicted as a representative TG curve over time (A), the mean peak thrombin concentration ± SD (B), and the mean total net thrombin ± SD (C) (n = 5). Statistical significance was assessed using a repeated-measures 1-way analysis of variance and Dunnett's posttest (*P < 0.05, ***P < 0.001).

3.2. The procoagulant activity of SA-EVs requires intact EVs and is partially mediated by DNA

Next, we investigated whether EV-mediated TG was dependent on associated DNA and/or intact EVs. Surface DNA was eliminated with DNase I, and compared to vehicle-treated SA-EVs, DNase treatment reduced TG by 37%, suggesting that the DNA increases thrombin production (Fig. 2A). However, DNase treatment did not completely abolish thrombin production in PPP. Thus, other factors contribute to this procoagulant activity as well. To evaluate whether intact EVs were required, we attempted to eliminate EVs in 2 ways: by lysing EVs with TX-100 prior to ultracentrifugation or by preventing EV biogenesis. Compared to SA-EVs, TX-100–treated SA-EVs did not enhance TG. Additionally, to confirm TX-100 itself did not interfere with TG, we treated PPP with or without 0.05% TX-100 (the same concentration used to treat EVs before ultracentrifugation) followed by the addition of PBS or DNA (Fig. 2B). Treatment of PPP with TX-100 had no impact on the ability of DNA to enhance TG in PPP.

Fig. 2.

NET-DNA binds to the outer surface of SA-EVs via phosphatidylserine and influences EV-mediated coagulation. SA-EV preparations were processed as follows: EVs were isolated by ultracentrifugation and treated with DNase I or vehicle, or EVs were lysed with TX-100 prior to ultracentrifugation and then treated with DNase I or vehicle. EVs (3 µg or an equivalent volume of lysed EVs) were added to PPP and TG was measured (A, n = 3). P values were determined using a 2-way analysis of variance and Tukey's posttest (^aP < 0.05 vs buffer, ^bP < 0.05 vs vehicle EVs, ^cP < 0.05 vs DNase-treated EVs). PPP was treated with or without TX-100, followed by the addition of PBS or neutrophil DNA (25 µg/mL) to PPP (B, n = 3). P values were determined using a repeated-measures 2-way analysis of variance and Tukey's posttest. SA-EVs were conjugated to beads, stained with CellVue, and following washes, PS or DNA was measured with Annexin-V–FITC (AV) (C) or PI (D), respectively. EVs were identified by gating on CellVue⁺ beads and shown is a representative histogram of AV staining (C, n ≥ 3). EVs were treated with recombinant AV for 1 h at room temperature, and DNA (D) and TG (E) were analyzed. DNA was measured using PI and flow cytometry as before. Data are expressed as the percentage of PI⁺ events normalized to EVs without AV pretreatment (D, n = 5 ± SD). For TG, data are expressed as the percentage of peak thrombin normalized to EVs without AV pretreatment (E, n = 5 ± SD). P values were determined using a repeated-measures 1-way analysis of variance and Dunnett's posttest (*P < 0.05, **P < 0.01).

Given EVs were required for procoagulant activity, we also attempted to decrease EV production using GW4869, a neutral sphingomyelinase inhibitor. Attempts to restrict SA-EV production from neutrophils were unsuccessful as GW4869 failed to reduce flotillin 1, an abundant protein in EVs (Supplementary Fig. 1). We extended GW4869 treatment to differentiated HL-60 cells (dHL-60) to exclude the short life span of primary cells as a confounding variable. By nanoparticle tracking analysis, particle concentrations were similar in both experimental groups. CD81 was lower in EVs recovered following GW4869 treatment, suggesting that CD81 is not a suitable EV marker for the production of EVs from neutrophil-like cells (Supplementary Fig. 1). Nonetheless, treatment with DNase and detergent showed that the presence of intact, bioactive EVs was required for procoagulant activity and that the DNA only partially contributes to this activity.

Given these data, we evaluated the contribution of PS to the procoagulant activity of EVs. We hypothesized that exofacial PS on SA-EVs captured DNA since PS can bind to histones, which are present in NETs, and can also directly contribute to platelet EV-induced TG.^11,49 First, we established that bead-associated CellVue⁺ EVs were PS⁺ (Fig. 2C). Next, to investigate the role of PS in EV-DNA association, SA-EVs were incubated with recombinant Annexin V (AV) and washed, and external DNA was assessed by staining with PI. Compared to the vehicle control, treatment of EVs with 10 µg/mL AV reduced PI staining by 20.6% (Fig. 2D), indicating surface DNA associates with PS to some extent. To investigate the involvement of PS in coagulation, we treated SA-EVs with recombinant AV and measured TG. Compared to SA-EVs, TG was reduced by 26.2% when EVs were preincubated with 10 µg/mL AV (Fig. 2E). Taken together, these data suggest that PS partially mediates procoagulant activity and may do so though binding to DNA.

3.3. Neutrophils release EVs spontaneously and during challenge with bacteria

Given complications that arise from thrombosis, we questioned whether other types of bacteria would increase the production of procoagulant neutrophil EVs. To answer this, we selected bacteria that are prevalent in sepsis [SA, S. epidermidis (SE), E. coli (EC), and P. aeruginosa (PA)], challenged human neutrophils, and isolated EVs.¹ The success of EV recovery was confirmed following the MISEV guidelines.⁵⁰ Flotillin 1 and CD63 were used to confirm the presence of EVs, and the absence of GRP-94, an endoplasmic reticulum protein, was tested to determine the purity of each EV subtype. Relative to cell lysates and consistent with EV recovery, EVs were enriched in flotillin 1 and absent of GRP-94 (Fig. 3A). These EVs were also positive for CD63 and PS and were visualized by transmission electron microscopy (Fig. 3B–D), demonstrating that EV recovery occurred.

Fig. 3.

Characterization of neutrophil EVs following challenge with bacteria. EVs were isolated from neutrophils left alone buffer (spon) or challenged with SA, SE, EC, or PA. After isolation, proteins from cell lysates (CLs) and EVs were separated via sodium dodecyl–sulfate polyacrylamide gel electrophoresis and detected with immunoblotting for flotillin 1 (FLOT-1, 1 µg) and GRP94 (7 µg) (A, n ≥ 3). Shown are representative images from a minimum of 3 independent experiments. CD63 and PS were measured by staining with CD63-PE or Annexin-V–FITC as described above. EVs were identified by gating on CellVue⁺ beads, and the average percentage of CD63⁺ (B) or PS⁺ (C) events is shown (n ≥ 3 ± SD). EVs were processed and imaged by transmission electron microscopy on a Hitachi 7700 transmission electron microscope (D, n = 3). Representative images are shown. Scale bars equal 100 nm.

3.4. Bacteria increase the production of small EVs from neutrophils

After characterizing intact EVs, our focus shifted toward evaluating the potential of bacteria to upregulate EV biogenesis from neutrophils. Compared to EVs recovered from resting neutrophils (e.g. spontaneous EVs), EVs generated in response to bacteria had higher protein concentrations (Fig. 4A). So, we quantified actual EV production using nanoparticle tracking analysis. The size distribution remained comparable for each EV subtype. EVs had a mean size of 156 ± 7 nm and a median size of 139 ± 71 nm, categorizing them as small- to medium-sized EVs.⁵¹ Although the size of EVs was consistent between all groups, the particle count increased by 6.4- to 8.7-fold when neutrophils were challenged with bacteria, and no significant differences were observed among EV types (Fig. 4D and E). Therefore, increasing EV biogenesis seems to be a shared feature of these bacteria.

Fig. 4.

Neutrophils challenged with bacteria generate more EVs compared to resting cells. EV protein concentration was determined for each EV subtype by bicinchoninic acid assay (BCA) (A, n ≥ 19). Protein concentration was normalized to 1 × 10⁶ cell equivalents. Nanoparticle tracking analysis (NTA) was utilized to determine particle size (B and C) and particle concentration (D) with a sample size of 3 per group. The particle concentration was normalized to 1 × 10⁶ cell-equivalents. P values were determined by 1-way analysis of variance with Dunnett's posttest (*P < 0.05, **P < 0.01, ****P < 0.0001 vs spon-EVs). A representative histogram shows the size distribution of EVs by plotting the concentration of particles against particle size (nm) (E).

3.5. EVs produced in response to bacteria associate with more DNA

Again, we questioned whether DNA played a direct role in activating coagulation. First, we had to ascertain whether NETosis occurred and if DNA associated with SE-, EC-, and PA-EVs. We explored whether vital NETosis occurred in response to bacteria by monitoring the presence of extracellular DNA in the absence of lysis (Fig. 5A). Compared to control cells, challenge with SA, SE, and EC yielded an increase in extracellular DNA without membrane compromise, consistent with vital NETosis. However, challenge with PA resulted in a modest, but significant, degree of cell lysis, indicating vital NETosis was not the only mechanism contributing to increased extracellular DNA.

Fig. 5.

Bacteria cause neutrophil EVs to associate with human and bacterial DNA. Before isolating EVs, supernatants from neutrophil–bacteria coculture were collected and analyzed for extracellular DNA with Sytox Green staining and membrane damage by LDH release (A, n ≥ 3 ± SD). The levels of extracellular DNA are presented as fold-change relative to resting cells, while LDH release is calculated as a percentage of LDH found in the same number of lysed cells. P values were determined for each assay using a 1-way analysis of variance and Dunnett's posttest (*P < 0.05, **P < 0.01, ****P < 0.0001 vs buffer for Sytox staining or ^#P < 0.0001 vs buffer for LDH release). Human nucDNA and mtDNA from supernatants before and after pelleting EVs with ultracentrifugation (UC) were quantified with droplet digital polymerase chain reaction (ddPCR) (B, n ≥ 3 ± SD). Neutrophils were left in buffer or subjected to coculture with SA, SE, EC, or PA. Following this, EV-DNA was extracted, separated via electrophoresis on a 0.7% agarose gel, and visualized using Gel Red staining (n ≥ 3, C). ddPCR was utilized to quantify the total copy number of nucDNA, mtDNA, and bacDNA present in EV-DNA. The copy number of the resulting amplicons were normalized to 1 × 10⁶ cells (D, n ≥ 3 ± SD). P values were determined by 1-way analysis of variance with Dunnett's posttest (*P < 0.05 vs spon-EVs). Before DNA extraction, EVs underwent treatment with DNase I. Intraluminal nucDNA and mtDNA were subsequently quantified, and the percentage of DNase-susceptible and DNase-resistant DNA is reported for each EV subtype (E, n ≥ 3 ± SD). Cell lysates (CLs) or EVs (3 µg) were then separated by sodium dodecyl–sulfate polyacrylamide gel electrophoresis and subjected to immunoblotting for myeloperoxidase (MPO), neutrophil elastase (NE), or histone H3 (H3). Representative images of at least 3 experiments are shown (F).

Given that increased extracellular DNA was present following bacterial challenge, we characterized the interaction further by determining the copy number of nuclear (nuc) and mitochondrial DNA (mtDNA) before and after pelleting EVs (Fig. 5B). Using this approach, we detected both types of DNA and found over half of the extracellular DNA in supernatants pelleted with EVs. Further analysis of EV-associated DNA by gel electrophoresis revealed a 10-Kb band of DNA from SA-, SE-, EC-, and PA-EVs, whereas no DNA was detected from spontaneously released EVs (Fig. 5C).

Sequencing of DNA from SA-EVs showed complete coverage of the human nuclear, human mitochondrial, and bacterial genomes,²⁷ so we then quantified the total copy number of nucDNA, mtDNA, and bacterial DNA (bacDNA, Fig. 5D). Relative to spontaneous EVs, SA- and SE-EVs showed a notable 25-fold increase in nucDNA. EC- and PA-EVs also possessed 8.5-fold and 5.0-fold more nucDNA, respectively, but this increase was not significant. The other EV subsets possessed more mtDNA compared to spontaneous EVs and associated with bacDNA specific to the bacterial genus used to generate them. Furthermore, most human DNA associated with the surface of EVs and not inside a lipid membrane as 99% of nucDNA and 90% of mtDNA were susceptible to DNase I treatment (Fig. 5E). Finally, neutrophil elastase, myeloperoxidase, and histone H3 were elevated on EVs isolated following bacterial challenge (Fig. 5F), and these proteins may be attached to DNA (as NETs), EVs, or bound to both.

3.6. Neutrophil EVs promote factor XII–mediated contact-dependent thrombin generation

To determine whether SE-EVs, EC-EVs, and/or PA-EVs were procoagulant, we measured the ability of each EV subtype to amplify TG in PPP (Fig. 6A). EVs generated during bacterial challenge increased peak thrombin compared to the PBS control. Spontaneously released EVs also tended to increase TG, but the trend in this experiment was not statistically different from the control. To explore these results further, we elected to measure whether neutrophil EVs could initiate the coagulation cascade without prior activation. EV-dependent TG was measured in the absence of the TF–phospholipid trigger (Fig. 6B). Compared to the PBS control, spontaneously released EVs were not able to produce thrombin in the absence of the TF–phospholipid trigger, whereas EVs produced in response to bacteria could initiate coagulation on their own. Altogether, these data show that bacteria increase the production of EVs from neutrophils, and these bacterial-induced EVs have the capacity to amplify and initiate the coagulation cascade.

Fig. 6.

Coagulation factor XII regulates EV-amplified thrombin generation. Thrombin generation was evaluated in PPP after treatment with 15 µg/mL EVs to quantify the amplification of EV-induced coagulation (A, n ≥ 3 ± SD). P values were determined using a 1-way analysis of variance and Tukey's posttest (*P < 0.05, ***P < 0.001 vs spon-EVs). Furthermore, the initiation of the coagulation cascade by EVs was assessed in PPP without the presence of the phospholipid–TF trigger (B, n = 3 ± SD). P values were determined using a 1-way analysis of variance and Tukey's posttest (^aP < 0.05 vs PBS, ^bP < 0.05 vs spon-EVs). EV isolation was performed in the absence of neutrophils, and recovered material was measured for its ability to initiate TG (C, n = 3 ± SD). P values were determined using a 1-way analysis of variance and Tukey's posttest (^aP < 0.05 vs buffer, ^bP < 0.05 vs neutrophils + SA). EVs were isolated from dHL-60 challenged with S. aureus, and then CD66b⁺ EVs were depleted to evaluate the initiation of TG (D, n = 4 ± SD). P values were determined using a 1-way analysis of variance and Tukey's posttest (**P < 0.01). P values were determined using a 1-way analysis of variance and Tukey's posttest (^aP < 0.05 vs PBS, ^bP < 0.05 vs spon-EVs). Control, factor VII–deficient (FVII–), and factor XII–deficient (FXII–) PPP were treated with 15 µg/mL EVs. TG was measured over 60 min to determine which pathway EVs use to amplify coagulation (E, n = 3 ± SD). P values were determined using a 2-way analysis of variance and Tukey's posttest (*P < 0.05, **P < 0.01, ***P < 0.001).

To confirm that neutrophil EVs, and not bacterial EVs, mediated TG, we evaluated the procoagulant activity of bacteria and bacterial products. Viable bacteria are present in EV preparations, but opsonized bacteria did not cause prothrombin cleavage (Supplementary Fig. 2). It is possible that ultracentrifugation concentrates bacterial-derived EVs, and so we performed EV isolation in the absence of neutrophils to address this possibility (Fig. 6C). As expected, SA-EVs initiated TG compared to the PBS control, but material recovered from mock-treated bacteria did not cause TG. Next, we used CD66b as a marker for neutrophil EVs and depleted CD66b⁺ EVs following EV isolation from dHL-60 challenged with S. aureus (Fig. 6D). Fractions enriched in CD66b⁺ EVs initiated TG, but depletion of CD66b⁺ EVs significantly reduced TG. Finally, we found that EVs recovered following treatment with N-formylmethionyl-leucyl-phenylalanine (fMLF), a sterile agonist, could also initiate the coagulation pathway (Supplementary Fig. 3). These data exclude bacteria or bacterial products as a primary source of procoagulant activity.

Finally, we investigated the mechanism driving TG. We tested whether TG required coagulation factor VII, initiator of the TF pathway, or factor XII, initiator of the contact pathway. EVs were added to control, factor VII–deficient, or factor XII–deficient PPP, and TG was measured (Fig. 6C). All EV subtypes (including spontaneous EVs) were procoagulant in control PPP. Likewise, EVs increased thrombin production in factor VII–deficient plasma, indicating that TF plays no significant role in neutrophil EV-mediated coagulation. In contrast, EVs failed to increase TG in PPP deficient of factor XII, indicating that the contact-dependent pathway is activated by these EVs.

3.7. Neutrophil EVs promote coagulation through DNA-dependent and independent mechanisms

As DNA can activate the contact-dependent pathway and surface DNA enhances the procoagulant activity of SA-EVs, we sought to determine whether the surface DNA affected TG caused by EVs from SE, EC, and PA challenge. Since the amount of EV-DNA appeared to vary between EV subsets, we measured the threshold at which EV-DNA could enhance TG (Fig. 7A). We found that a concentration of 15 µg/mL EV-DNA was required to increase TG in PPP. Next, we quantified the concentration of surface DNA on each EV subtype using the Quant-iT PicoGreen dsDNA assay. Overall, EVs from bacterial challenge associated with over 8-fold more surface DNA compared to spontaneous EVs. However, SA- and SE-EVs associated with nearly double the amount of DNA compared to EC- and PA-EVs (Fig. 7B). Per 3 µg of EV protein (the amount of EVs added to PPP), SA- and SE-EVs associated with ∼15 µg/mL surface DNA, whereas EC- and PA-EVs associated with ∼7 µg/mL surface DNA.

Fig. 7.

The mechanism neutrophil EVs utilize to promote coagulation partially depends on the type of bacteria used as an agonist. PPP was treated with the indicated concentrations of purified EV-DNA, and TG was measured (A, n = 3, ± SD). P values were determined using a repeated-measures 1-way analysis of variance and Dunnett's posttest (*P < 0.05). Quantification of surface DNA associated with 3 µg EVs was conducted using the PicoGreen dsDNA assay (B, n ≥ 5 ± SD). Statistical significance was determined using a 2-way analysis of variance and Tukey's posttest (^aP < 0.05 vs spon-EVs, ^bP < 0.05 vs SA-EVs and SE-EVs). Then, 30 µg/mL EVs was treated with DNase I or vehicle, and amplification of TG was evaluated in PPP (C, n ≥ 3 ± SD). Prior to isolation, EV-containing supernatants were passed through a 0.2-µm filter, and unfiltered or filtered EVs were used to treat PPP and measure TG (D, n = 4 ± SD). The initiation of the coagulation cascade was measured by excluding the phospholipid–TF trigger, and 30 µg/mL DNase- or vehicle-treated EVs was added to PPP to evaluate TG (E, n ≥ 3 ± SD). P values were determined with paired t-tests (*P < 0.05, **P < 0.01).

To investigate whether surface DNA amplified EV-mediated coagulation, EVs were treated with or without DNase I and added to PPP (Fig. 7C). As expected, treatment of SA-EVs with DNase I resulted in a 45% reduction in TG (Fig. 7C). Likewise, SE-EVs showed a similar trend with a 71% reduction in procoagulant activity. Conversely, surface DNA on EC- and PA-EVs did not contribute to TG. Supporting these findings, as representatives, SA- and EC-EVs were filtered prior to isolation, which we have shown removes surface DNA.²⁷ Consequently, TG by SA-EVs was reduced by 57% postfiltration, whereas EC-EVs showed no change in procoagulant activity after filtration (Fig. 7D), corroborating the DNase results. Finally, as only EVs generated during bacterial challenge could initiate the coagulation cascade, we decided to determine if initiation was attributable to NET-DNA on the EV surface. EVs were treated with or without DNase I, and TG was measured in the absence of the TF–phospholipid trigger (Fig. 7E). Once again, only the surface DNA on SA- and SE-EVs contributed to TG. Taken together, these data suggest that while several types of bacteria cause EVs and NET-DNA to colocalize and assemble, only some bacteria cause the generation of EVs, which utilize NET-DNA as a mechanism to promote coagulation.

4. Discussion

To understand factors that promote coagulation, a component linked to unfavorable outcomes in sepsis,⁵² we investigated whether EVs and/or EV-associated DNA were produced by neutrophils following phagocytosis of sepsis-causing bacteria and tested if these neutrophil products were procoagulant.

We began by evaluating the procoagulant potential of SA-EVs and their associated DNA since we, and others, have shown that SA increases EV production and that SA-EVs associate with NETs.^27,39 SA-EVs were procoagulant, and this activity required intact, bioactive EVs (Figs. 1 and 2). EV-associated DNA contributed to the TG, and our data are consistent with a procoagulant role for intact NETs.^33,35 However, removal of DNA by blocking PS or by degrading DNA did not fully eliminate TG, so other factors likely also contribute to EV-mediated coagulation.

Sepsis has multiple etiologies, with bacterial infection being the most common.⁵³ Our approach to study neutrophil EV production in response to SA, SE, EC, or PA was guided by the clinical prevalence of these bacteria¹ and a recognition that using primary human cells and whole bacteria would provide deeply valuable insights into sepsis. We characterized EV subsets (Fig. 3 and 4) and found that EVs generated during bacterial challenge associate with more DNA than spontaneously released EVs (Fig. 5). Human DNA on the surface of SA-, SE-, and EC-EVs is likely a result of vital NETosis, but PA-induced DNA might originate from lytic NETosis or another cell death pathway, excluding apoptosis or necrosis, which fragments DNA. While a comprehensive understanding of the mechanisms involved in the release of all 3 types of DNA and the factors responsible for DNA attachment are not fully understood, EV-DNA could be explored as a diagnostic marker for subtyping in sepsis. Along these lines, high levels of EVs and mtDNA in septic patient plasma are predictive of increased mortality.^54,55

All EVs amplified TG in a factor XII–dependent manner, but EVs generated in response to bacteria (or fMLF) could initiate coagulation, as well (Figs. 6 and 7). Compared to PBS, spontaneous EVs amplified TG to some extent. This finding was significant when we used commercial plasma and an analysis of varaince as a statistical test (Fig. 6E) or if a Mann–Whitney test was used to compare 2 groups: PBS treatment and spontaneous EVs (P < 0.05). However, the procoagulant activity of spontaneous EVs may be disproportionately emphasized in this assay given that PPP was treated with an equivalent protein quantity rather than a particle number, and 3 µg of spontaneous EVs corresponded to a greater number of unstimulated neutrophils than did any of the other subsets.

Another surprising observation was that EV-associated DNA only contributed to TG when SA- and SE-EVs were used to initiate or amplify TG. Thus far, we have excluded histone or granule proteins as contributors since their abundance was similar in all subtypes. Other biomolecules that could potentially initiate thrombin generation include negatively charged proteins, proteoglycans, and inorganic polyphosphates. Neutrophils are rich in proteoglycans, which contain negatively charged glycosaminoglycans and can be found in EVs from various cell types.⁵⁶ These EV components may trigger thrombin generation by activating kallikrein via factor XIIa.^24,56–59 Polyphosphates, negatively charged inorganic phosphates, have been found on EVs from several cell types and can initiate coagulation.^60,61 Notably, polyphosphate⁺ EVs can cleave high molecular weight kininogen into cleaved kininogen, suggesting a role in coagulation.¹² Given that polyphosphates are highly conserved across all living cells, they might be present on the surface of neutrophil EVs, providing a possible mechanism for EVs to initiate coagulation.⁶² We are currently investigating how EVs, especially EC- and PA-EVs, activate coagulation. However, an -omics approach would be the best method to determine which molecules on the surface of EVs contribute to coagulation.

To date, several studies have examined the use of anticoagulants for treating blood clotting in sepsis, such as heparin and antithrombin, which target downstream coagulation factors but are associated with an increased risk of bleeding complications.⁶ Targeting the initiation of the coagulation cascade by preventing EV production might offer a more effective approach to treating patients. While the factors driving DNA-independent, EV-induced coagulation remain undefined, our data demonstrate that the mechanisms by which neutrophil EVs influence TG occur with and without the involvement of extracellular DNA. Given many researchers have suggested targeting and reducing NET production as a potential therapeutic strategy for thrombosis, a multifaceted approach that considers the influence of EVs and the type of infection could have clinical implications.^63–65 Still, our study has limitations because we did not assess the crosstalk between leukocytes, endothelial cells, platelets, and blood products, which is an important consideration when studying thrombosis. This is especially important since each bacterial-induced EV subtype associated with more neutrophil elastase, myeloperoxidase, and histone H3, which could also contribute to coagulation via endothelial and platelet activation.^34,66,67

In summary, we demonstrate that bacteria can drive thrombosis by upregulating the release of procoagulant EVs from neutrophils that elevate TG in a factor XII–dependent manner. Neutrophil EVs generated during bacterial challenge associated with NETs, but only the NET-DNA associated with SA- and SE-EVs influenced the procoagulant activity of EVs. This study provides the first evidence that the role of NET-DNA in coagulation is dichotomous, and NET-independent mechanisms should be explored. Overall, this study provides insights into the molecular mechanisms that could drive thrombosis and highlights the importance of experimentally addressing bacterial diversity in models of sepsis.

Author contributions

K.M.W-K. designed the study and wrote the manuscript, along with M.C.G-W., who also funded the work. A.R.M. provided methodology development. K.M.W-K., C.C.B., E.R.A., M.T.G., J.W.G., B.D.S., and M.C.G-W. contributed to the acquisition, analysis, and interpretation of data. All authors read, revised, and approved the manuscript.

Supplementary material

Supplementary materials are available at Journal of Leukocyte Biology online.