Neutrophil-secreted extracellular vesicles induce blood-brain barrier leakage and tight junction disruption

Abstract

Extracellular vesicles (EVs) are potent mediators in cell-cell communication that regulate diverse cell functions through the delivery of their bioactive cargo molecules to recipient cells. Previous work from our group demonstrated elevated plasma EV levels in patients and animals following septic or inflammatory insults, with a substantial proportion originating from neutrophils. As frontline defenders against bacterial infection, activated neutrophils release germicidal factors, some of which circulate systemically and inflict collateral tissue damage at a distance, including the brain. The role of neutrophil-derived EVs in regulating blood-brain barrier (BBB) structure and permeability after septic injury remains poorly defined. In this study, we first characterized EV production by mouse neutrophils stimulated with bacterial lipopolysaccharide (LPS) and subsequently investigated their functional and mechanistic effects on BBB integrity under in vivo and in vitro settings. Nanoparticle tracking analysis (NTA), immunoblotting, and transmission electron microscopy (TEM) revealed that LPS stimulation of neutrophils promoted EV secretion, indicated by increased particle number and protein content. Systemic administration of these EVs in mice induced cerebral microvascular leakage of plasma tracers (sodium fluorescein, at 376-Da; and dextran at 3-kDa) as quantified by near-infrared (NIR) nano-imaging and fluorometric assays. In cultured brain microvascular endothelial monolayers, EVs from naïve unstimulated neutrophils exerted minimal effects, whereas EVs from LPS-stimulated neutrophils caused a concentration-dependent reduction in transendothelial electrical resistance (TER) and a significant increase in solute permeability, indicative of paracellular hyperpermeability. Confocal microscopy revealed that tight junction proteins claudin-5 and zonula occludens-1 (ZO-1), which normally form continuous belt-like structures at endothelial cell–cell contacts, appeared discontinuous or fragmented upon EV internalization. Consistently, endothelial cells exposed to activated neutrophil-derived EVs exhibited reduced expression of tight junction proteins. Furthermore, TEM of brain capillaries from EV-injected mice provided ultrastructural evidence of tight junction disruption. Collectively, these findings suggest that neutrophil activation in response to infection promotes BBB leakage through the release of EVs capable of compromising endothelial tight junction integrity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12987-025-00744-8.

Introduction

The brain microvascular endothelium forms a highly specialized interface between the systemic circulation and neuronal parenchyma. Endothelial cells juxtaposed via intercellular junctions constitute the BBB that imposes stringent restrictions on the transvascular passage of fluids, solutes, and cells [1, 2]. Compromise of BBB integrity represents a pivotal pathophysiological event in the initiation and progression of various disease or injury states with multisystem consequences, including central nervous system (CNS) dysfunction [1, 3–5]. In the context of bacterial infection, invading microorganisms and their associated virulence factors, typically bacterial cell wall lipopolysaccharide (LPS) [6], elicit a robust host immune response marked by neutrophil recruitment and release of germicidal agents [7, 8]. Some of these agents can travel through the bloodstream and extend damaging effects to host tissues at distant sites. Damage to the cerebral microvascular endothelium results in BBB leakage and brain dysfunction, manifesting as emotional distress, cognitive impairment, confusion or delirium, often seen in clinical conditions associated with severe infection or sepsis [9, 10]. Although canonical neutrophil-derived factors such as pro-inflammatory cytokines and granular proteases have been extensively investigated as principal effectors of barrier disruption [11], emerging evidence implicates EVs as an additional, mechanistically distinct class of mediators contributing to BBB dysfunction [12, 13].

EVs are membranous nanoparticles secreted by their parent cells into biological fluids such as blood [14]. They carry unique cargo molecules that not only indicate the identity and activity of their donor cells but also actively interact with recipient cells and alter cell function via cargo transfer [15, 16]. Circulating EVs are more stable and potent mediators than individual soluble plasma factors because they contain a collection of cargo molecules that are encapsulated by lipid bilayers and thus protected from enzymatic degradation or non-enzymatic destruction. As an example, cytokines contained by EV cargo are more stable than those existing in free form in the blood [17]. Furthermore, unlike most soluble proteins that are large and hydrophilic, often requiring receptors or transporters to enter cells, EVs are tiny, lipophilic particles that can easily enter cells via membrane fusion, endocytosis, and other transcellular pathways with or without receptors, thereby rapidly and efficiently delivering cargo contents to recipient cells [17–19]. It has been shown that the signaling strength of EV-presented molecules can be 1,000-fold higher than their soluble counterparts [20]. These unique features of EVs make them powerful vehicles, driving long-distance communication among multiple organs and tissues.

We have recently reported that EV production is increased in the blood of patients with sepsis [21], as well as in animals subjected to septic insults, including bacterial wall LPS [22]. A significant portion of the plasma EVs derived from neutrophils, and their cargo was enriched with molecules known as hyperpermeability or barrier-disrupting factors. While the involvement of EVs in inflammatory injury has been well recognized, their cell- and tissue-specific effects and underlying molecular mechanisms remain poorly understood [12, 23, 24]. Even less is known about neutrophil EV-induced BBB changes in the context of infection or septic injury. Therefore, the goal of this study was to examine the direct impact of neutrophil-derived EVs on cerebral microvascular endothelial junction integrity and BBB permeability under in vivo and in vitro settings.

Materials and methods

Animals

C57Bl/6J mice were obtained from The Jackson Laboratory and bred in-house at the University of South Florida animal care facility. Mice were housed under a 12-hour light/dark cycle with ad libitum access to food and water. Male and female mice, 2−6 months old, were used for this study.

Neutrophil isolation and preparation

Polymorphonuclear neutrophils (PMNs) were isolated from mouse bone marrow as previously described [25], with modifications. Femurs and tibias were flushed with RPMI 1640 medium containing 10% fetal bovine serum (FBS) and 2 mM EDTA. Red blood cells were lysed using RBC lysis buffer (BioLegend; #420301). Neutrophils were separated by density gradient centrifugation using Histopaque 1119 (Sigma; #11191) and 1077 (Sigma; #10771). Neutrophil purity and viability were confirmed by flow cytometry (Cytek^® Nothern Lights™) using the following antibodies: CD45-FITC (1:1200; Biolegend; #103108), CD3-APC/Cy7 (1:800; BioLegend; #103108), CD4-APC/Cy7 (1:800; BioLegend; #100221), CD11b-APC (1:800; BioLegend; #101212) and Ly6G-PE (1:800; BioLegend; #127608). Non-viable cells were identified by exclusion using Zombie NIR fixable viability dye (1:2000; BioLegend; #423105). Neutrophils (14−32 × 10⁶ cells) were stimulated with either LPS (E. coli serotype O55:B55; 1 µg/mL; Sigma; #L2637) or vehicle (PBS) in fresh RPMI 1640 media supplemented with 5% EV-depleted fetal bovine serum (FBS) and cultured for 4 h at 37 °C. Subsequently, the cell culture supernatant was collected and used for EV isolation as described below, yielding CTR-PMN and LPS-PMN EVs, respectively.

PMN-derived EV isolation

EVs were isolated from 8 to 10 mL of PMN supernatant as previously described [26]. The supernatant was first centrifuged at 400 x g for 10 min to remove floating cells, followed by 2,000 x g for 10 min to remove dead cells and cellular debris. The resulting supernatant was centrifuged at 18,000 x g (Optima XPN-90, Beckman Coulter) for 40 min to remove large particles. The supernatant was then filtered through a 0.22 μm pore filter and subjected to ultracentrifugation at 150,000 x g for 2 h. Lastly, the pellet containing crude EVs was washed once with ice-cold double-filtered PBS at 150,000 x g (Optima MAX-TL, Beckman Coulter) for 2 h. The washed pellet, enriched in small EVs, was resuspended in double-filtered PBS, aliquoted, and stored at -80 °C. In separate groups, EV-free supernatant (LPS-PMN EV-DEP) was prepared by collecting LPS-stimulated neutrophil supernatant and ultracentrifuged at 100,000 x g for 20 h, followed by filtration through a 0.22 μm pore filter. In some experiments, LPS-PMN EVs were sonicated three times for 40 s to disrupt EV membrane.

Characterization of PMN-derived EVs by NTA

EV preparation and characterization followed the MISEV 2023 guidelines [15]. NTA of EV particle size and concentration was conducted using our previously described methods with minor modifications [21, 22, 27, 28]. Briefly, EV samples were diluted (1:1000) in double-filtered PBS and analyzed using a NanoSight NS300 instrument (Malvern Panalytical) equipped with 488- and 532-nm lasers and a high-sensitivity CMOS camera. Ten independent replicates were performed for each EV preparation derived from vehicle-treated naïve PMNs and LPS-treated PMNs. Instrument settings, including temperature below 25 °C, sensitivity of 30–85 frames per second (fps), shutter speed of 55, camera gain 5, a camera level of 11, and a laser pulse duration equal to the shutter duration, were kept constant across all measurements. Data were processed using NTA software version 3.4. All relevant data from our experiments have been submitted to the EV-TRACK knowledgebase (EV-TRACK ID: EV2500700) [29].

Characterization of PMN-derived EVs by TEM

EV samples were prepared for TEM using a modified protocol based on previously published methods [21, 22, 30]. Briefly, EVs were fixed in 2.5% glutaraldehyde/0.1 M cacodylate buffer (Electron Microscopy Sciences; #16537-05) and deposited onto 200-mesh copper grids with a carbon-coated formvar film (Electron Microscopy Sciences; #FCF200-Cu-50) for 30 min. Excess liquid was removed, and grids were contrasted with 2% uranyl acetate (Thermo Fisher; #18-607-644) for 40 s and imaged (JEOL1400) at 80 kV.

Endothelial cell culture

Primary mouse brain microvascular endothelial cells (BMECs) from C57Bl/6J mice were purchased from Cell Biologics (#C57-6023) and cultured in the recommended endothelial cell medium (Cell Biologics; #M1168) supplemented with 10% FBS, 100 U/mL penicillin, and 100 µg/mL streptomycin (Thermo Fisher; #15140122). All culture surfaces used for BMEC growth were coated in 0.1% gelatin. For functional assays, cells were grown for 5–7 days post-confluence to allow endothelial maturation and tight junction formation. Experiments were performed using different batches of cells to ensure biological replicates.

Cell viability

Cell viability was assessed using the Cell Counting Kit-8 (GLPBIO; #GK10001) as previously described [27]. Briefly, 1 × 10⁵ BMECs were seeded into 96-well plates and treated with vehicle (CTR-PMN EVs) or LPS (LPS-PMN EVs) (5 × 10⁸ particles) for 6 and 24 h. CCK-8 reagent (10 µL) was added to 100 µL of culture medium and incubated at 37 °C for 1 h. Absorbance at 450 nm was measured using a microplate reader (SpectraMax M3; Molecular Devices).

TER measurements

The barrier function was assessed in cultured BMEC monolayers as previously described [31] by measuring cell-cell adhesive resistance to electric current using an electric cell-substrate impedance sensing (ECIS) system (Model Zθ; Applied Biophysics). Briefly, 2 × 10⁵ BMECs were seeded onto 8-well ECIS culture arrays (8W10E + PET; Applied Biophysics) pre-coated with gelatin. Arrays were connected to the ECIS system after 5−7 days post-confluence. A 1-V, 4-kHz alternating current was applied through a 1-MΩ resistor at 7-second intervals. Treatments, including vehicle (PBS), CTR-PMN EVs, LPS-PMN EVs, or LPS-PMN EV-DEP (5 × 10⁸ particles), were added at time zero. TER readings were continuously recorded and analyzed using the ECIS data analysis software. TER tracings are presented and normalized resistance to baseline values prior to treatments (t = 0).

Transwell solute flux assay

Solute permeability was determined by measuring solute flux across cultured BMEC monolayers as previously described [31]. BMECs were seeded into gelatin-coated transwell inserts (0.4 μm pore size; Corning Inc., #3401) and cultured to confluence. Cells were then treated with vehicle (PBS), CTR-PMN EVs (5 × 10⁸ particles), or LPS-PMN EVs (5 × 10⁸ particles) for 6 and 24 h. FITC-dextran 4-kDa (1 mg/mL) was added to the luminal chamber for 1 h, and fluorescence intensity in the abluminal chamber was measured at 6 and 24 h after treatment using a microplate reader (excitation/emission: 485/535 nm). The concentration of FITC-dextran 4-kDa was determined using a standard curve, and the solute permeability coefficient was calculated as [C]/t x 1/A x V/[L], where [C] is the concentration in the bottom chamber (in mg/mL), t represents time (in seconds), A denotes the cross-sectional area of the insert (0.33 cm²), V indicates the volume of the bottom chamber, and [L] is the initial concentration of the tracer in the top chamber. Solute flux was reported in units of cm s^{− 1}.

Immunofluorescence

BMECs (2 × 10⁵) were seeded onto gelatin-coated chamber slides (Lab-Tek; #1777399PK). EVs derived from vehicle- or LPS-treated PMNs were labeled with PKH67 (2 µM) for 2 min as previously described [26]. PKH67-labeled EVs were reconstituted in RPMI 1640 medium supplemented with 5% EV-depleted serum and isolated using size exclusion chromatography (qEV original, 70 nm; Izon) per manufacturer’s instructions. EV concentration and protein content in each fraction were quantified by NTA and Qubit protein assay, respectively. EV-rich fractions (fractions 5−12) were pooled and concentrated using a 10-kDa Amicon centrifugal filter (Merck Millipore; #UFC801024) to obtain purified EVs. PKH67-labeled EVs or PBS were added to the BMEC monolayers for 6 h. Following treatment, cells were washed and fixed with either 4% paraformaldehyde in PBS or ice-cold methanol for 20 min at room temperature. For claudin-5 and ZO-1 immunolabeling, cells were washed, blocked, and permeabilized with 5% normal donkey serum in 3% BSA/PBS containing 0.2% Triton X-100 for 1 h at room temperature. Cells were incubated overnight at 4 °C with primary antibodies: claudin-5 (Mouse; 1:100; Thermo Fisher; #35-2500) and ZO-1 (Rabbit; 1:100; Thermo Fisher; #40-22200) in 0.3% BSA/PBS with 0.2% Triton X-100. Slides were washed and incubated with Alexa Fluor^® 488-conjugated donkey anti-mouse (or anti-rabbit) IgG (1:500; Invitrogen; #A21202 or #A21210) or Alexa Fluor^® 594-conjugated donkey anti-mouse (or anti-rabbit) IgG (1:500; Invitrogen; #A21203 or #A21207) for 1 h at room temperature, followed by DAPI staining (Thermo Fisher; #62248) and mounted using ProLong™ Diamond Antifade Mountant (Invitrogen; #P36961). Cells were imaged under a Leica DMi8 STED confocal microscope HC PLAPO 63x/1.4 NA objective and analyzed as z-stacks with Leica Application Suite X (LAS X) software. The distribution and morphology of tight junction molecules were observed under the above-described confocal microscope and analyzed with ImageJ. For 3D reconstruction, the serial images captured along the z-axis were opened in Imaris 9.0, and the surface rendering of relevant stained areas was generated by intensity thresholding to identify them for visualization. Non-specific staining or background signals were eliminated by size filtering and manual deletion. Junction molecule (claudin-5, ZO-1) distribution was assessed by measuring their fluorescence intensity at cell-cell contact regions, with background fluorescence in adjacent non-junction regions subtracted for correction. Under each experimental condition, immunofluorescence measurements were taken from 3 independent experiments using different dishes of cells. In each experiment, 3 different microscopic fields (each containing 6−8 cell-cell contact regions) were selected, and junction fluorescence intensities were collected and averaged across the 3 fields. In addition, junction integrity was assessed by analyzing the continuity of junction structures lining the neighboring cell membrane, where “discontinuity” was defined by the number of interruptions or gaps present at junction regions per microscopic field.

Western blotting

For quantification, EVs, MBECs, or brain tissue homogenates were resuspended in lysis buffer consisting of 1x Laemmli sample buffer (Bio-Rad; #1610737), β-mercaptoethanol (Gibco; #21985023), and a protease/phosphatase inhibitor cocktail (Sigma; #78442). Samples were incubated on ice and sonicated. Lysates were centrifuged at 10,000 x g for 10 min at 4 °C and boiled at 75 °C for 10 min. Total protein concentration in cell lysates, EVs, and tissue homogenates was measured using Qubit protein assay (Thermo Fisher; #Q33212) and a Qubit 3.0 fluorometer according to the manufacturer’s instructions. Samples were loaded onto 4−20% Tris-glycine gels (Bio-Rad; #456–1094) and transferred to nitrocellulose membranes using the Trans-Blot Turbo system (Bio-Rad). Membranes were blocked for 1 h at room temperature in 5% non-fat dry milk (Cell Signaling; #9999) prepared in Tris-buffered saline (TBS-T) containing 0.01% tween-20 (Cell Signaling; #9997). Membranes were incubated overnight at 4 °C with primary antibodies: CD63 (Rabbit; 1:500; Abcam; #ab217345), CD81 (Rabbit; 1:500; Cell Signaling; #10037s), CD9 (Rabbit; 1:500; Cell Signaling; #98327), Cyto C (Rabbit; 1:500; Cell Signaling; #11940), claudin-5 (Mouse; 1:250; Invitrogen; #35-2500), ZO-1 (Rabbit; 1:500; Invitrogen; #40-2200), β-actin (Mouse or Rabbit; 1:1000; Li-Cor; #926-42212 or #926-42210) or GAPDH (Rabbit; 1:1000; Cell Signaling; #5174) diluted in blocking buffer. After washing with TBS-T, membranes were incubated with IRDye-conjugated secondary antibodies (Mouse or Rabbit; 1:10000; Li-Cor; #926-32212 or #926-68073) or HRP-conjugated secondary antibodies (Mouse or Rabbit; 1:2000; Cell Signaling; #7076S or #7074S) for 1 h at room temperature, followed by additional washes in TBS-T. Protein bands were visualized using the Odyssey CLx Imaging System (Li-Cor) for IRDye-conjugated secondary antibodies or SignalFire Elite ECL reagent for HRP-conjugated secondary antibodies (Cell Signaling; #12757S) and imaged using ChemiDoc MP Imaging system (Bio-Rad). Band densitometry was performed using ImageJ software. Data from 3−4 independent experiments are represented as mean ± standard error mean (SEM), normalized to GAPDH (for EV preparations) or β-actin (for cell and tissue homogenates).

In vivo measurement of BBB permeability

Transport of plasma tracer molecules of different sizes (376-Da and 3-kDa) across the BBB was assessed as we previously described [32]. Mice under 2% isoflurane anesthesia were injected via the retro-orbital venous sinus with 100−150 µL of PBS, CTR-PMN EVs (1 × 10¹¹ particles), LPS-PMN EVs (1 × 10¹¹ particles), or LPS-PMN EV-DEP (1 × 10¹¹ particles). At 24 h post-injection, mice received Alexa-Fluor 680-3-kDa dextran (5 µg/g of a 1 mg/mL solution; Invitrogen; #D34681) for 1 h. The dye was then cleared from the circulation after perfusion with PBS. Brains were harvested and imaged (as whole or coronal sections) using a NIR nanoimaging system (Odyssey CLx; Li-Cor) with 700- and 800-nm excitation wavelengths. Fluorescence intensity of 3-kDa dextran was quantified using Li-Cor system software and expressed as arbitrary units (A.U). In different groups, BBB permeability to a smaller tracer NaFl (376-Da) was measured. Mice were injected with NaFl (2.5 µL/g of a 200 mg/mL solution; Sigma; #6377) and allowed to circulate for 30 min. After clearance of the dye, brains were harvested, and tissues were weighed and homogenized in 20% trichloroacetic acid (Sigma; #T0699). Samples were diluted with an equal volume of 0.05 M borate buffer and centrifuged at 1,200 x g for 10 min at 4 °C. Proteins were precipitated with ethanol and centrifuged at 12,000 x g for 20 min at 4 °C. Fluorescence was measured from 100 µL aliquots of each sample, alongside blanks and standards, using a microplate reader. Samples within the linear range of the standard curve were used to determine NaFl amount in the brain. Fluorescence values were normalized to tissue weight, and NaFl amount was expressed as ng/mg tissue.

Ultrastructural analysis of brain endothelium by TEM

The ultrastructure of mouse brain capillary endothelium was examined as previously described [32]. Briefly, anesthetized mice were injected intravenously through the retro-orbital venous sinus with either PBS or LPS-PMN EVs (1 × 10¹¹ particles). After 24 h, mice were perfused with PBS followed by 2% glutaraldehyde/0.1 M cacodylate buffer (Electron Microscopy Sciences; #16536-15). Hippocampus samples were dissected from intact brains, cut into 1 mm cubes, and post-fixed in 2% glutaraldehyde/0.1 M cacodylate buffer overnight at 4 °C. Samples were rinsed in 0.1 M sodium cacodylate (Electron Microscopy Sciences; #11652) and then immersed in 2% aqueous osmium tetroxide with 1.5% potassium ferrocyanide buffer for 4 h at room temperature. After three rinses in 0.1 M cacodylate buffer, samples were dehydrated in an ethanol ascending series (50, 70, 95, and 100%) followed by a final change in acetone before resin/acetone gradient infiltration (2:1, 1:1, 1:2) (Embed812 eponate; Electron Microscopy Sciences; #14120). This was followed by an overnight immersion in pure fresh resin and embedding in silicon flat molds, which were polymerized in a 60 °C oven. Ultrathin (70 nm) sections were obtained using an ultramicrotome. To enhance contrast, grids were incubated with 100 µL of 2% uranyl acetate for 20 min. Images were acquired using a TEM (JEOL1400 or JEOL1400 flash) at 8 kV. ImageJ software was used to measure tight junction morphology (length and width). Measurements were taken from 10−14 capillaries from two mice per group.

Statistical analysis

Data were analyzed using one-way analysis of variance (ANOVA) to assess differences among multiple groups with normality distributed (parametric) data. Post-hoc comparisons were conducted using Tukey’s test. For comparisons between two groups, a two-tailed unpaired Student’s t-test was applied. Results are presented as mean ± SEM, and statistical significance was defined as p ≤ 0.05. Statistics and graphs were performed using GraphPad Prism version 10.

Results

LPS stimulation enhances EV release from neutrophils

We compared EVs isolated from control (naïve, unstimulated) neutrophils (CTR-PMN) and those stimulated with LPS (1 µg/mL; LPS-PMN). NTA analysis revealed a marked increase in EV concentration following LPS stimulation, with LPS-PMNs releasing significantly more EVs than CTR-PMNs (Fig. 1a, b). Protein quantification of EVs showed a significant elevation in protein content in LPS-PMN EVs compared to CTR-PMN EVs (Fig. 1c), suggesting enhanced vesicle biogenesis and cargo protein enrichment in response to LPS stimulation. Despite increase in EV production, EV size (both mean and modal size) remained constant across groups, centered around 100 nm (Fig. 1d). Morphological examination of EV samples by TEM confirmed the presence of intact, spherical vesicles approximately 100 nm in diameter under both conditions, indicating that LPS stimulation does not alter EV shape or morphology (Fig. 1e). As further confirmation of EV identity, Western blot analysis demonstrated robust expression of typical EV markers (tetraspanins) CD63, CD9, and CD81 in both EV populations. Additionally, the absence of cytochrome C (Cyto C) indicated that the EV samples were not contaminated with mitochondrial components from apoptotic cells (Fig. 1f; full blots shown in Supplementary Fig. 1).

Fig. 1.

LPS stimulation of neutrophils increases EV release. (a) NTA histograms showing size distribution of EVs isolated from mouse bone marrow PMNs incubated for 4 h with vehicle (PBS; CTR-PMN EVs) or LPS (1 µg/mL; LPS-PMN EVs). Both EV populations exhibited a size range (< 200 nm), with the LPS-PMN EVs group showing increased particle counts. (b) Quantification of EV concentration per 10⁶ neutrophils revealed a significant increase in particle release following LPS stimulation (***p < 0.001). (c) Total protein content was significantly elevated in LPS-PMN EVs compared to CTR-PMN EVs (**p ≤ 0.01), suggesting enhanced cargo protein loading. (d) Mean and mode EV sizes were comparable between groups, indicating no major change in vesicle size after LPS. (e) TEM images confirmed the presence of intact, spherical vesicles ranging from 50–100 nm in both groups. (f) Western blot analysis demonstrated enrichment of canonical EV markers CD63, CD81, and CD9 in both EV populations. Cytochrome C (Cyto C) was absent, indicating that EV samples were not contaminated with mitochondrial components from apoptotic cells. GAPDH served as a loading control. Data are represented as mean ± SEM (n = 10 each); statistical significance determined by unpaired Student’s t-test

Activated neutrophil-derived EVs disrupt endothelial barrier integrity

To assess the impact of neutrophil-derived EVs on endothelial barrier function, we measured TER and solute permeability on BMEC monolayers following treatment with EVs from unstimulated (CTR-PMN) and LPS-activated neutrophils (LPS-PMN). LPS-PMN EVs induced a significant, reduction in TER compared to vehicle and CTR-PMN EVs, with maximal barrier disruption observed at 24 h (Fig. 2a, b). This effect appeared to be concentration-related, with higher concentrations of LPS-PMN EVs (up to 5 × 10⁸ particles/mL) producing more pronounced decrease in TER (Fig. 2c, d). More specifically, the ECIS measurements showed basal TER values (ohms, Ω) of 2166 ± 63.7 and 2190 ± 25.3 in the PBS and LPS-PMN EV groups, respectively (at 0 h). Comparable values were seen with similar BMEC models in previous studies [33, 34]. By 6 h, TER was reduced in EV-treated cells vs. controls (1887 ± 54.5 vs. 2236 ± 52 Ω; *p ≤ 0.05, n = 10 per group). This difference was more significant at 24 h (1663 ± 52 vs. 2113 ± 55 Ω; ****p ≤ 0.0001, n = 10 per group). To verify that the TER reduction was primarily attributed to EVs, rather than LPS contaminated in the neutrophil supernatant, we performed additional control experiments with EV-depleted supernatant (via ultracentrifugation) and found that it (LPS-PMN EV-DEP) did not significantly alter TER (Fig. 2e, f). Likewise, EV preparations after sonication (break EVs by disrupting the membrane) failed to produce the TER response seen with intact EVs (Supplementary Fig. 2). Moreover, BMEC monolayers directly treated with LPS (1 µg/mL, same dose used to activate PMNs) did not exhibit obvious TER reduction as observed when they were treated with LPS-PMN EVs (Supplementary Fig. 2), further supporting the relative importance of EVs derived from activated PMNs in reducing endothelial barrier resistance. In addition to TER, we assessed solute permeability by measuring the transendothelial flux of small molecule (4-kDa dextran) across BMEC monolayers, which showed that treatment with LPS-PMN EVs resulted in significantly increased solute permeability (Fig. 2g). Cell viability assays showed no significant cytotoxicity at either 6- or 24-hours post-EV treatment, suggesting that barrier impairment was not due to cell death (Fig. 2h). Together, these results suggest that LPS-PMN EVs mediate hyperpermeability through a dynamic process that affects the barrier structure and function without causing cell death.

Fig. 2.

LPS-PMN EVs reduce barrier resistance and increase solute permeability in BMEC monolayers. (a-b) TER was measured as an indicator of cell-cell adhesive barrier function (paracellular permeability) over 24 h in BMEC monolayers treated with vehicle (PBS), EVs from unstimulated neutrophils (CTR-PMN EVs), or EVs from LPS-stimulated neutrophils (LPS-PMN EVs). LPS-PMN EVs induced a progressive and significant drop in TER over time compared to vehicle and CTR-PMN EVs, with maximal barrier disruption observed at 24 h (**p ≤ 0.01; n = 3 per group). (c-d) Concentration-related analysis of EVs showed that higher concentrations of LPS-PMN EVs (5 × 10⁸ particles/mL) produced more pronounced decreases in TER (***p ≤ 0.001 compared to vehicle; n = 3 per group). (e-f) TER measurements in BMEC monolayers treated with vehicle, LPS-PMN EVs, and EV-depleted supernatants from LPS-stimulated PMNs (LPS-PMN EV-DEP) (n = 3 per group). LPS-PMN EV-DEP slightly decreased TER values but did not reach statistical significance compared to vehicle. (g) Solute permeability assays showed increased flux of 4-kDa dextran in LPS-PMN EV-treated monolayers, while vehicle and CTR-PMN EVs had no significant effect (*p ≤ 0.05, **p ≤ 0.01; n = 8 per group). (h) Cell viability remained unchanged across all conditions at both 6 and 24 h after EV treatment, indicating that barrier disruption was not due to cytotoxicity (n = 4 per group). Data are represented as mean ± SEM; statistical significance determined by one-way ANOVA with Tukey’s post-test

Activated neutrophil-derived EVs increase BBB permeability in vivo

To evaluate the impact of neutrophil-derived EVs on BBB in vivo, we administered EVs from unstimulated (CTR-PMN EV), LPS-stimulated neutrophils (LPS-PMN EVs), and EV-depleted supernatants from LPS-stimulated PMNs (LPS-PMN EV-DEP) into mice and assessed BBB permeability using fluorescent tracers (Fig. 3). Brains showed enhanced 3-kDa dextran extravasation in mice injected with LPS-PMN EVs, indicating BBB leakage (Fig. 3a-d). In contrast, mice receiving CTR-PMN EVs or EV-depleted supernatants (LPS-PMN EV-DEP) showed minimal tracer extravasation, comparable to vehicle controls. Quantitative analysis confirmed a significant increase in 3-kDa dextran and NaFl uptake by brain tissues in LPS-PMN EV-injected mice, while CTR-PMN EV and LPS-PMN EV-DEP groups did not differ significantly from vehicle (Fig. 3d, e). Together, these results demonstrate that neutrophil EVs generated under inflammatory conditions can increase BBB permeability in vivo, supporting a mechanistic link between systemic inflammation and BBB dysfunction.

Fig. 3.

LPS-PMN EVs increase BBB permeability in vivo. (a-c) Representative NIR images of whole brains (top) and coronal sections (bottom) from mice injected with vehicle (PBS), EVs from unstimulated neutrophils (CTR-PMN EVs), EVs from LPS-stimulated neutrophils (LPS-PMN EVs), or EV-depleted supernatants from LPS-stimulated PMNs (LPS-PMN EV-DEP) at 24 h, showing distribution of 3-kDa dextran within the brain. LPS-PMN EV-treated mice showed increased tracer uptake, indicating enhanced vascular permeability. (d) Quantification of 3-kDa dextran fluorescence intensity (A.U., arbitrary units) showed significantly elevated tracer uptake in brains from LPS-PMN EV-injected mice compared to vehicle and CTR-PMN EVs (*p ≤ 0.05, **p ≤ 0.01; n = 5−16), while LPS-PMN EV-DEP did not change from controls. (e) Quantitative analysis of NaFl uptake (ng dye/mg tissue) confirmed increased leakage in LPS-PMN EV-injected mice, supporting an EV-dependent mechanism of barrier disruption (*p ≤ 0.05; n = 6 per group). Data are represented as mean ± SEM; statistical significance determined by one-way ANOVA with Tukey’s post-test

Activated neutrophil-derived EVs reduce claudin-5 and ZO-1 localization at cell-cell contacts

To examine the molecular events occurred in endothelial cells during EV challenge, we examined the expression of tight junction proteins in BMECs following treatment with vehicle (PBS), EVs from unstimulated neutrophils (CTR-PMN EVs), or LPS-stimulated neutrophils (LPS-PMN EVs) for 6 h. Representative images showed disrupted junctional localization of claudin-5 and ZO-1 in LPS-PMN EV-treated cells, consistent with compromised barrier integrity. In contrast, CTR-PMN EVs and vehicle controls showed preserved tight junction architecture (Fig. 4a). Quantification of fluorescence intensity confirmed a statistically significant decrease in claudin-5 and ZO-1 fluorescence intensity in BMECs treated with LPS-PMN EVs compared to vehicle or CTR-PMN EVs (Fig. 4b). These findings suggest that EVs released from activated neutrophils impair endothelial barrier function by altering tight junction protein organization at cell-cell contacts, which may contribute to the mechanism underlying cerebrovascular hyperpermeability during systemic inflammation.

Fig. 4.

LPS-PMN EVs reduce claudin-5 and ZO-1 junction location in BMECs. (a) Representative immunofluorescence images of BMEC monolayers treated for 6 h with vehicle (PBS), EVs from unstimulated neutrophils (CTR-PMN EVs), or EVs from LPS-stimulated neutrophils (LPS-PMN EVs), showing staining of claudin-5 and ZO-1 at cell-cell borders. (b) Quantitative analysis of fluorescence intensity showed a significant reduction in claudin-5 and ZO-1 signal in cells exposed to LPS-PMN EVs compared to vehicle and CTR-PMN EVs (*p ≤ 0.05). These results suggest that LPS-PMN EVs cause tight junction diffusion at endothelial cell borders as reflected by decreased fluorescence intensity. Data are represented as mean ± SEM (n = 3 per group); statistical significance determined by one-way ANOVA with Tukey’s post-test

Activated neutrophil-derived EVs reduce tight junction protein expression in vivo and in BMECs

To determine whether neutrophil-derived EVs alter tight junction protein expression in vivo, we analyzed brain tissue from mice injected with EVs from unstimulated (CTR-PMN) or LPS-stimulated neutrophils (LPS-PMN) at 24 h. Western blot analysis of claudin-5 and ZO-1 showed a significant reduction in protein levels in LPS-PMN EV-injected mice compared to vehicle controls (Fig. 5b). This effect was not observed in mice injected with CTR-PMN EVs (Fig. 5a), suggesting that inflammatory activation is required for EV-mediated tight junction disruption. Quantification of band intensity confirmed a significant decrease in claudin-5 and ZO-1 protein expression levels in LPS-PMN EV-injected mice (Fig. 5b), while CTR-PMN EVs had no significant effect (Fig. 5a; full blots shown in Supplementary Fig. 3). In BMECs, LPS-PMN EVs also suppressed tight junction protein expression levels, with claudin-5 and ZO-1 showing a sustained reduction at both time points, 6 and 24 h post-EV treatment (Fig. 5e, f) compared to BMECs treated with EVs from unstimulated neutrophils (CTR-PMN EV) (Fig. 5c, d; full blots shown in Supplementary Figs. 4 and 5). Together, these results demonstrate that activated neutrophil-derived EVs impair tight junction protein expression in both brain tissue and BMECs, supporting their role in BBB barrier disruption.

Fig. 5.

LPS-PMN EVs decrease expression levels of tight junction proteins in brain tissue and BMECs. (a-b) Western blot analysis of claudin-5 and ZO-1 in brain homogenates from mice injected with vehicle (PBS), EVs from unstimulated neutrophils (CTR-PMN EVs), or EVs from LPS-stimulated neutrophils (LPS-PMN EVs). LPS-PMN EVs significantly reduced claudin-5 and ZO-1 expression compared to vehicle and CTR-PMN EVs (*p ≤ 0.05, **p ≤ 0.01). (c-f) Representative blots and quantification of claudin-5 and ZO-1 expression in BMECs treated with CTR-PMN EVs or LPS-PMN EVs for 6 and 24 h. LPS-PMN EVs significantly reduced protein levels of claudin-5 and ZO-1 at both time points, with higher suppression observed at 24 h (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001). CTR-PMN EVs did not significantly alter expression compared to the vehicle. Quantification of band intensity was performed using ImageJ and normalized to β-actin. Data are represented as mean ± SEM (n = 4 per group); statistical significance determined by unpaired Student’s t-test

Activated neutrophil-derived EVs disrupt tight junction architecture in BMECs

To assess the structural impact of LPS-activated neutrophil-derived EVs on tight junction architecture, we performed immunofluorescence on claudin-5 and ZO-1 in BMEC monolayers treated with either LPS-stimulated neutrophil EVs (LPS-PMN EVs) or vehicle. Three-dimensional volume rendering of immunofluorescence images showed marked fragmentation and discontinuity (gaps) of claudin-5 and ZO-1 staining in LPS-PMN EV-treated BMECs compared to vehicle (Fig. 6a, b). Tight junction proteins appeared disorganized, with reduced linear continuity along endothelial borders. Quantitative image analysis confirmed a significant increase in the number of disconnected junctional components for both claudin-5 and ZO-1 in LPS-PMN EV-treated BMECs, indicating structural disruption of tight junctions (Fig. 6c). These changes were not observed in vehicle-treated controls, where tight junctions remained intact and continuous at cell-cell borders (Fig. 6a, b). Together, these findings demonstrate that neutrophil-derived EVs released under LPS stimulation compromise the spatial organization of tight junction proteins in brain endothelial monolayers, contributing to altered endothelial barrier function.

Fig. 6.

LPS-PMN EVs cause tight junction discontinuity in BMECs. (a-b) Representative 3D immunofluorescence images of claudin-5 and ZO-1 in BMEC monolayers exposed to vehicle (PBS), CTR-PMN EVs, or LPS-PMN EVs for 6 h. Inserts show higher magnification views highlighting the normal belt-like structure becoming fragmented upon EV internalization. (c) Quantification of disconnected junctional components showed a significant increase in claudin-5 and ZO-1 discontinuity in cells uptaking LPS-PMN EVs compared to vehicle controls (**p ≤ 0.01, ***p ≤ 0.001). Images were rendered using Imaris software; scale bars = 10 and 5 μm (a), 15 and 6 μm (b). Data are represented as mean ± SEM (n = 3 per group); statistical significance determined by one-way ANOVA with Tukey’s post-test

Changes in tight junction ultrastructure consistent with paracellular permeability

To investigate the ultrastructural changes induced by activated neutrophil-derived EVs on brain capillary tight junctions, we performed TEM on brains from mice injected with LPS-PMN EVs or vehicle (PBS). TEM images showed intact and continuous tight junctions in vehicle-treated mice (Fig. 7a, b) while LPS-PMN EV-treated mice exhibited disrupted junctional morphology, including widened paracellular gaps (Fig. 7c, d,f). Quantitative analysis confirmed an increase in tight junction width in LPS-PMN EV-treated mice compared to controls (Fig. 7f), with no changes observed in tight junction length in both groups (Fig. 7e). These findings suggest that tight junction opening and paracellular hyperpermeability may serve as an important mechanism responsible for activated neutrophil EV-induced BBB leakage.

Fig. 7.

LPS-PMN EVs alter brain capillary endothelium ultrastructure. (a-d) Representative TEM images of hippocampal capillaries of mice receiving intravenous injection of vehicle (PBS) or LPS-PMN EVs at 24 h, showing intact tight junctions in vehicle-injected mice and disrupted tight junctions in LPS-PMN EV-injected mice. (e-f) Quantitative analysis revealed a significant increase in tight junction width in LPS-PMN EV-injected mouse capillaries compared to controls (**p ≤ 0.01), with no changes observed in TJ length in both groups. Scale bar = 500 nm (a, c-d) and 200 nm (b). BM basement membrane, EC endothelial cell, TJ tight junction (white arrows). Data are represented as mean ± SEM; statistical significance determined by unpaired Student’s t-test

Discussion

This study provides novel evidence supporting the pathophysiological importance of neutrophil activation and secretion of EVs in mediating barrier dysfunction in the BBB endothelium. We incorporated state-of-the-art techniques into a comprehensive examination of cell-specific production of EVs and their function in animals and cultured brain microvascular endothelium. Major findings include: (1) LPS stimulation promoted neutrophil secretion of EVs enriched with proteins; (2) EVs produced by LPS-activated neutrophils were highly capable of causing barrier dysfunction in the brain microvascular endothelium, as indicated by a decreased TER and increased solute permeability in a concentration-related manner, whereas EVs produced by naïve unstimulated neutrophils exerted minimal effects; (3) in mice, intravenous injection of activated neutrophil EVs caused significant leakage of small plasma solutes (< 3-kDa) across cerebral microvessels, indicating tight junction hyperpermeability; (4) tight junction molecules, claudin-5 and ZO-1, exhibited reduced protein expression and junction localization in brain microvascular endothelium exposed to EVs from LPS-activated neutrophils; (5) confocal microscopic analysis of brain microvascular endothelium revealed continuous belt-like structures of claudin-5 junctions juxtaposed at cell-cell contacts under normal conditions; the belt became thinner, discontinuous or fragmented upon endothelial uptake of neutrophil-derived EVs; and (6) TEM confirmed tight junction disruption in intact brain capillaries of mice receiving intravenous injection of EVs derived from LPS-activated neutrophils. Taken together, these data suggest that activated neutrophils secrete EVs that can interact with brain capillary endothelium and disrupt the claudin-5-based tight junction, leading to BBB leakage.

The current study focuses on the pathological regulation of tight junction integrity and permeability in the BBB endothelium. Tight junctions are expressed at the apical side of endothelial cells lining along the luminal surface of the microvascular wall, formed by networks of transmembrane protein strands that pull neighboring cells together [35–37]. Claudin-5 and its associated proteins, including ZO-1, that provides a structural linkage to the cytoskeleton, are key components of tight junctions essential for restricting the paracellular passage of water and small solutes [35, 38–40]. Aside from their physiological importance in maintaining basal barrier properties, alterations in tight junction structure and function play a critical role in pathological conditions associated with infection, inflammation, or traumatic injury, where excessive blood fluid and proteins extravasate from microvessels into surrounding tissues, rendering tissue edema. Water and water-soluble molecules, such as small plasma proteins, traverse via diffusion and/or convection across the endothelium primarily through the paracellular route, rather than the transcellular route, which requires energy-supported transporters, receptors, or vesicular transcytosis. In the brain, tight junctions are considered a major element controlling water passage across the endothelium, whereas transcellular flux of solutes across the BBB is low due to limited presence of transporters, pinocytosis, and vesicular transport [38]. The structural features of the BBB endothelium support the rationale of our study focusing on tight junctions and paracellular permeability under the pathological conditions related to neutrophil activation.

As a hallmark of inflammatory response to septic insults, neutrophils in the peripheral circulation are activated and undergo a series of changes characterized by adhering to endothelium followed by chemotactic migration and releasing bactericidal agents; this process is a double-edged sword because while protective against infection, it could culminate with host tissue injury and blood fluid/protein leakage across microvessels [7, 41, 42]. Previous studies by our group and others have demonstrated that activated neutrophils alter microvascular barrier structure and function via both adhesion-dependent and adhesion-independent mechanisms [43–45]; the latter occurs through the secretion of barrier-disrupting agents, including cytokines, reactive oxygen species, and granular enzymes. While these canonical mediators are well recognized, attempts to treat inflammatory injury by antagonizing them or blocking their receptors have met with limited clinical success. Multiple factors have been ascribed to the current lack of therapies directly targeting the microvascular barrier, including our incomplete understanding of tissue/cell-specific mechanisms of barrier regulation, short lifespan and instability of soluble inflammatory mediators due to constant exposure to enzymatic and/or non-enzymatic destruction, and heterogeneous responses of various barrier tissues to different types of mediators. For example, it remains unclear whether and how certain types of neutrophil products can target barrier tissues that are distantly located or anatomically isolated, such as the BBB.

Our data show that bacterial endotoxin (LPS) stimulation of neutrophils led to an increased production of EVs that possess a potent capability to disrupt tight junctions in the brain microvascular endothelium, contributing to BBB leakage. As EVs have attracted rising interest in their diagnostic and therapeutic potential, their role in the pathogenesis of disease or injury has begun to be appreciated. Evidence has emerged supporting a causal link between EV production and the development of cardiovascular disease, cancer, and inflammatory disorders [46–48]. Since EVs produced by different parent cells under different physiological states carry unique cargo and exert pleiotropic effects on recipient cells, it is important to characterize their cell-specific production and function under defined clinical or experimental conditions [49, 50]. Within this context, our literature search indicates that limited information is available regarding the effects of neutrophil-derived EVs on the brain endothelial barrier in sepsis-relevant conditions.

In this study, we compared the particle concentration and molecular identity of EVs produced by naïve, unstimulated neutrophils with LPS-stimulated neutrophils. We then applied these EVs to cultured brain microvascular endothelial monolayers and measured their concentration response in reducing barrier resistance and increasing solute permeability. Importantly, when injected into the retro-orbital venous sinus of live animals, activated neutrophil EVs caused significant BBB dysfunction, evidenced by plasma tracer leakage and accumulation in brain tissues. Thus, our study provides unique evidence, both in vivo and in vitro, for the direct and causal role of neutrophil-secreted EVs in mediating BBB leakage during septic injury.

At the sub-cellular level, we revealed novel insights into the molecular events occurring at tight junctions of the brain microvascular endothelium, where EV internalization is coupled with claudin-5 and ZO-1 disorganization, seen as the normal belt-like structure at cell-cell contact becoming diffuse or fragmented. The morphological observation is supported by immunoblot assays showing reduced protein expression of claudin-5 and ZO-1 in endothelial cells treated with activated neutrophil EVs. Further support comes from the transmission electron microscopic study of brain capillary endothelium in mice receiving EV injection. Capillaries exposed to EVs from LPS-stimulated neutrophils often display ultrastructural changes characteristic of tight junction opening or disruption. Such a junction response could be attributed to the underlying mechanism of paracellular hyperpermeability. Indeed, both in vivo assessment of plasma transport and in vitro measurement of solute flux indicate that EVs induce hyperpermeability to small molecules of 376-Da to 3-kDa. This finding is in line with the notion that claudin-5-based tight junctions determine the paracellular permeability of small molecules below 800-Da [51]. Thus, our data further implicate the mechanistic importance of tight junctions in the BBB leak response. Consistent with these findings, a previous study demonstrated that EVs released upon fMLP stimulation of neutrophils decrease TER and tight junction protein abundance in BMECs, implicating neutrophil-derived EVs as mediators of tight junction-related BBB pathogenesis [52].

We acknowledge several limitations associated with this study. First, while the current focus is placed on the BBB endothelium as the EV target or recipient, how donor cells (neutrophils) package and secrete EVs in response to different stimuli is an interesting EV biology question remaining to be answered. Likewise, their cargo contents and specific cargo molecules responsible for the BBB-disrupting action remain to be identified. We have recently published studies on proteomics profiling of plasma EVs in human patients and animals with septic injury, demonstrating septic EV cargo enriched with permeability-increasing signaling molecules, including protein tyrosine kinases, proteases, and extracellular histones [22]. It is possible that these molecules are expressed in activated neutrophil EV cargo and contribute to EV-induced BBB hyperpermeability. As our proteomics profiling was done with whole blood or plasma, future studies can be carried out to identify neutrophil-specific EV cargo composition, as well as differential regulation of EV biogenesis in neutrophils under different stimulatory conditions. In this study, we used LPS to stimulate neutrophils because it is a typical septic agent with close relevance to the clinical conditions of bacterial infection, endotoxemia and sepsis, and it has been widely used in animal/cell models of inflammatory injury. Furthermore, we acknowledge that EV isolation via differential ultracentrifugation (dUC) has methodological limitations associated with suboptimal purity or potential contamination with other particular matter. However, compared to other methods, such as size exclusion chromatography (SEC), dUC has a better recovery rate for small-volume samples, as seen in the neutrophil supernatant used in this study. While combining dUC with SEC or other purification methods could enhance purity, these approaches often reduce yield through vesicle loss, adsorption, dilution, and fraction overlap [53, 54]. Finally, the potential beneficial effects of EVs and their role in maintaining cerebral fluid homeostasis should be recognized. Although we found that neutrophils after inflammatory stimulation produced EVs that, in general, exert a detrimental effect to tight junction integrity, we do not factor out the possibility that these EVs also carry barrier-protective cargo. This notion is supported by the data that EVs produced by naïve neutrophils did not damage the BBB endothelium or cause plasma leakage in cerebral microvessels. Further studies are necessary to evaluate their relative importance in physiological versus pathological regulation of BBB structure and function.

Conclusions

Our study demonstrates that bacterial LPS stimulation promotes neutrophil secretion of EVs with barrier-disrupting capability. When injected into the cerebral circulation or applied to cultured brain microvascular endothelial monolayers, they cause endothelial barrier dysfunction and BBB leakage. Further molecular and imaging analyses suggest that neutrophil-derived EVs interact with brain endothelial cells and induce claudin-5/ZO-1 tight junction disruption. We postulate that neutrophil activation in response to septic challenge leads to BBB leakage by releasing EVs into the circulation that can target the tight junction-based microvascular endothelial barrier structure. This process may play a crucial role in the development of CNS disorders and multiple organ dysfunction associated with infection or sepsis.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

BBB

Blood-brain barrier

Cyto C

Cytochrome C

dUC

Differential ultracentrifugation

ECIS

Electric cell-substrate impedance sensing

EVs

Extracellular vesicles

FBS

Fetal bovine serum

HRP

Horse-radish peroxidase

LPS

Lipopolysaccharide

NaFl

Sodium fluorescein

NIR

Near-infrared

SEC

Size exclusion chromatography

SEM

Standard error mean

TBS-T

Tris-buffered saline

TEM

Transmission electron microscopy

TER

Transendothelial electrical resistance

ZO-1

Zonula occludens-1

Author contributions

L.A.B. contributed to the study design and performed most of the EV experiments along with data collection and analysis. M.N.B. performed Western blot analyses and helped with EV experiments. N.V. performed TEM and helped with mouse experiments for in vivo measurements of permeability. L.S. performed EV isolation, as well as mouse surgery and experiments. B.J.C. provided technical assistance for the confocal microscopy and electron microscopy studies. C.K. helped with the Limulus Amebocyte Lysate (LAL) assay. Y.Y. participated in experimental design and data interpretation. M.H.W. and S.Y.Y. conceptualized, directed, and sponsored the work through all levels of development. All authors critically discussed the results and approved the manuscript.

Funding

This work was supported by funding from the National Institute of Health grants HL150732 (to S.Y.Y), GM142110 (to S.Y.Y. and M.H.W.), GM143138 (to M.H.W.), and Department of Veterans Affairs IK6BX004210 (to M.H.W.).

Data availability

Data from the EV experiments have been submitted to the EV-TRACK knowledgebase (EV-TRACK ID: EV250070). Published data and relevant information will be made available upon reasonable request.

Declarations

Ethics approval and consent to participate

All animal procedures were approved by the University of South Florida Institutional Animal Care and Use Committee (IACUC) and complied with the NIH Guide for the Care and Use of Laboratory Animals.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.