ENDOTHELIUM-DERIVED MICROPARTICLES INDUCE ENDOTHELIAL DYSFUNCTION AND ACUTE LUNG INJURY

Abstract

Acute lung injury (ALI) carries a high mortality in critically ill patients. Recent reports correlate elevated concentrations of endothelium-derived microparticles (EMPs) with diseases of endothelial dysfunction. Many of these diseases have ALI sequelae. We hypothesize that EMPs contribute to endothelial cell (EC) dysfunction and development of ALI. To test this hypothesis, we treated isolated vessels with EMPs and examined changes in vasodilation. Endothelial cell cultures were incubated with EMPs and examined for changes in stimulated nitric oxide (•NO) production and nitric oxide synthase (eNOS) activation. Finally, EMPs were injected into rats and mice and lungs examined for ALI. In both mouse and human ex vivo vessel preparations, we found a marked attenuation of endothelium-mediated vasodilation after EMP treatment (4 × 10⁶/mL). This dysfunction was not corrected by pretreatment of EMPs with free radical scavengers. Coincubation of EMPs with EC cultures yielded a three-fold reduction in A23187-stimulated •NO release. Western analysis of these cells showed a corresponding decrease in eNOS phosphorylation at Ser1179 and a decrease in hsp90 association. Measurements of lung permeability, myeloperoxidase activity, and histology of EMPs-treated Brown Norway rats demonstrated pulmonary edema, neutrophil recruitment, and compromise of the endothelial-alveolar barrier as a second hit phenomenon. In C57BL/6 mice, exogenous EMPs caused a significant rise in pulmonary capillary permeability both as a primary and secondary injury. These findings demonstrate EMPs are capable of inducing significant lung injury at pathophysiologically relevant concentrations. Endothelium-derived microparticles inhibit endothelium-mediated vasodilation and •NO generation from eNOS. Once elucidated, EMP mechanisms of inducing ALI and endothelial dysfunction may present new therapeutic targets.

INTRODUCTION

Endothelium-derived microparticles (EMPs) occur at relatively low levels under normal physiologic conditions and are found in increased number in human disease states injurious to the vascular system. Current case series and reports describe significant elevations of plasma EMPs in patients with sickle cell disease, systemic lupus erythematosus, meningococcal sepsis, thrombotic thrombocytopenic purpura, systemic inflammatory response syndrome, antiphospholipid syndrome, multiple sclerosis, acute coronary syndrome, pediatric vasculitides, preeclampsia, and miscarriage (^1-11). Endothelial injury is a common feature of these conditions, which are often associated with sequelae of pulmonary capillary leak, microvascular thrombosis, and physiologic shunt.

Endothelium-derived microparticles (0.1- to 4-μm diameter) are accompanied by platelet- and monocyte-derived microparticles in normal circulation (¹¹). They are comprised of a lipid bilayer and a number of normal endothelial cell (EC) surface proteins (e.g., CD31, tissue factor, VE-cadherin) (^11-13). Shet et al. found a strong correlation between the severity of sickle cell disease and circulating plasma EMPs concentrations (⁷). Such findings suggest that EMPs represent not only a measure of EC injury but also a potential cellular source of inciting agents. For example, patients in sickle cell crisis are more likely to develop acute chest syndrome, an example of acute lung injury (ALI) after systemic endothelial activation (¹⁴). In vitro studies have recently shown LPS in combination with cytokines increases EC production of microparticles (¹⁵) that, in a recent review, was interpreted as playing a potential role in the mechanisms by which complement increases ALI (¹⁶). Exactly how EMPs participate in the mechanisms of ALI remains unknown; however, recent reports demonstrated that platelet microparticles are able to bind to leukocytes which can complicate flow analysis (¹⁷). Accordingly, it becomes important to develop in vitro systems for generating well-characterized EMPs for use in delineating mechanisms relevant to activated endothelium. As stated above, increased concentrations of EMPs occur in pathologic states that appear quantitatively related to severity of microvascular dysfunction. Indeed, many EMP-associated diseases have ALI sequelae. These observations led us to hypothesize that EMPs contribute to downstream EC dysfunction. Here, we present new data demonstrating that EMPs induce endothelial dysfunction, attenuate EC •NO release, alter endothelial nitric oxide synthase (eNOS) activation, and promote ALI.

MATERIALS AND METHODS

All protocols utilizing human tissue and laboratory animals were approved by the Medical College of Wisconsin Institutional Review Board and Animal Research Committee. The experiments were performed in adherence to the National Institutes of Health Guidelines on the Use of Laboratory Animals.

Generation and characterization of EMPs

Passage four human umbilical vein endothelial cells (HUVEC) were grown to confluency in 1% gelatin-coated T75 flasks using EGM-2 medium (Clonetics) containing 20% fetal bovine serum (FBS). Cultured cells were maintained at 37°C in 5% humidified CO₂. At the time of EMP generation, cells were washed three times with HBSS and serum starved at 37°C in nonsupplemented EBM-2 (Clonetics) for 2 h. After serum starvation, cells were washed again and stimulated with 5 mL EBM-2 supplemented with 10 ng/mL human plasminogen activated inhibitor-1 (PAI-1, American Diagnostica 1095). Three hours later, the EMP-rich supernatant was collected and centrifuged (300 × g, 10 min) to remove cell debris. The supernatant was ultracentrifuged (10⁵ × g, 60 min) and EMPs were resuspended in phosphate-buffered saline (PBS) at room temperature for experimentation.

Flow cytometric analysis of EMP was as follows. For characterization, EMPs from five confluent T75 flasks were pooled and concentrated into seven 1.5-mL samples. Samples of EMP were incubated with the following fluorophore-conjugated antihuman antibodies: tissue factor (1.5 μg, Innovative Research IAHTF-9264), E-selectin (3 μg, Pharmingen 551145), VE-cadherin (3 μg, Bender MedSystems BMS158FI), CD31 (1.5 μg, Pharmingen 555445), Annexin V (37 μg, Pharmingen 556420), CD31 (PECAM, 1.5 μg, Pharmingen 555445), and isotype control (3 μg, Pharmingen 555748). One sample was left unstained to control for EMP autofluorescence. All samples were rotated at room temperature for 1 h.

Flow cytometry was performed on a BD LSR II (20 mW, 488 nm) using fluorescein isothiocyanate (FITC) and phycoerythrin (PE) channels for fluorescence detection. Voltage settings for detection were optimized using 3-μm latex beads (Sigma LB-30), CD-31 FITC staining of EMPs, and E-selectin PE staining of EMPs. A gate was set to include particles with forward scatter less than the latex bead standard (3-μm diameter) and greater than noise occurring in sterile-filtered HBSS. Counts were completed in this gate at a low flow rate (12 μL/min), with significant staining defined as greater than first-decade FITC intensity. Data were analyzed using FACSDiva software (BD Biosciences). For experiments, an aliquot of the EMPs was stained with anti-CD31, mixed with a known volume of 3-μm latex beads, and analyzed by flow cytometry to determine the concentration and relative size.

Microvessel preparation and vasodilation measurements

To determine EMP effects on endothelium-mediated vasodilation, mouse facialis artery and human colon submucosal arterioles were used (^18-20).

C57BL/6 facialis arteries-

Arteries (~200-μm inner diameter) were atraumatically dissected from adult male C57BL/6 mice and maintained in cooled MOPS solution (in mM: NaCl 145, KCl 4.7, CaCl₂ 2.0, NaH₂PO₄ 1.2, MgSO₄ 1.2, pyruvate 2.0, Na₂EDTA 0.02, MOPS 3.0, and glucose 5.0, at pH 7.4). Vessels were then submerged in a MOPS bath, mounted on MOPS-containing glass capillaries, warmed to 37°C, and internally pressurized to 60 mmHg. Internal vascular diameters were measured with a calibrated video-microscope (Olympus CK2 microscope, Panasonic WV-BL200 CCD camera, and Boeckeler Instruments VIA-100K video micrometer). After 30 min of warming, measurement of the vessel diameter was recorded (D_max). U46619 (10⁻⁷M), a thromboxane analog, was used to preconstrict the vessel to 30% to 50% of D_max. A control dose-response curve using acetylcholine (ACh, 10⁻⁷-10⁻⁴M) was determined and vessel diameter recorded 2 min after each cumulative addition of ACh. Acetylcholine is an endothelium-dependent dilator that activates eNOS. Vasodilation was calculated as a percent of maximal vessel diameter change ([D_ACh−D_U46619]/[D_max−D_U44619]) to maintain pH and oxygenation. Buffers were changed every 15 min throughout the studies to maintain vessel viability.

Vessels were subsequently treated with EMPs (4 × 10⁶/mL), EMPs pretreated with butylhydroxytoluene (BHT, 1 mM), or EMPs pretreated with superoxide dismutase (SOD) (200 U/mL). After 10 min, an identical ACh dose-response curve was performed and diameters recorded. To ensure vascular smooth muscle viability after treatments, vasodilation to papaverine (10⁻⁴M) was determined.

Human colon microvessels-

Submucosal arterioles were dissected from full-thickness surgical specimens. Patients undergoing colon resection for cancer, diverticulitis, or focal inflammatory bowel disease were included; diffuse inflammatory bowel disease, shock, peripheral vascular disease, coronary artery disease, history of smoking, and diabetes were excluded. Specimens were immediately placed in 4°C Krebs buffer solution (in mM: NaCl 118, KCl 4.7, CaCl₂ 2.5, KH₂PO₄ 1.2, MgSO₄ 1.2, NaHCO₃ 20, Na₂EDTA 0.026, and dextrose 11, at pH 7.4). Microvessels (arterioles with maximal diameter 126.5 ± 12 μm) were carefully dissected from the uninvolved sections (i.e., relatively healthy sections) of submucosal surface of the bowel from five different individuals and transferred to a 6-mL tissue chamber containing Krebs solution. Venules were readily identified and excluded from these experiments. Arteriole ends were mounted on glass micropipettes filled with Krebs buffer and connected to a hydrostatic reservoir and pressure adjusted to 60 mmHg. Vessel diameters were measured by videomicroscopy as described in previous experiments using facialis arteries (^18-20). The chamber solution was continuously aerated with 75% N₂, 20% O₂, and 5% CO₂ and maintained at 37°C. All pharmacological agents were added to the bathing solution.

After a 1-h stabilization period, vessels were constricted to 30% to 50% of D_max as described earlier in "Materials and Methods." As with facialis artery experiments, an ACh dose-response curve was determined by the cumulative addition of ACh (10⁻⁹-10⁻⁴M) to the external bath. Vessel diameters were measured 2 min after the addition of ACh. Next, buffers were changed, the vessel allowed to re-equilibrate. Endothelium-derived microparticles (4 × 10⁶/mL) were added to the circulating bath for 10 min. Next, the vessels were once again preconstricted, and a second ACh dose-response curve was determined. At the end of each experiment, vessels were treated with papaverine (10⁻⁴M), a smooth muscle relaxant. Marked increases in vasodilation in response to papaverine were interpreted to mean that the impaired vasodilation after EMP treatments were due to EC dysfunction, not smooth muscle cell dysfunction. All chemicals were obtained from Sigma Chemical Co. (St. Louis, MO).

•NO measurements

Bovine aortic ECs contain much more eNOS protein than HUVEC and therefore generate robust •NO and eNOS phosphorylation-dependent signals upon stimulation, greatly increasing the possibility of delineating altered mechanisms of eNOS activation (²¹). Production was measured as an increase in •NO release after stimulation with 5 μM A23187 (Sigma C-5149), relative to basal •NO levels. Passage four cells were grown to 80% confluence in four P-100 dishes in RPMI 1640 (Invitrogen) + 10% FBS (Clonetics) at 37°C and 5% CO₂. A23187 is a receptor-independent agonist that mobilizes calcium to active eNOS to increase •NO generation. A23187 is used for these studies because it bypasses receptor-dependent mechanisms of eNOS activation, making it possible to examine the effects of experimental treatments on eNOS activation and protein-protein interactions independent of cell signaling via receptors. The EC cultures were serum-depleted for 4 h in RPMI containing 0.5% FBS. Each plate was then washed with HBSS containing 25 μM l-arginine three times. Two plates were incubated at 37°C for 30 min in 6 mL HBSS containing 25 μM l-arginine and EMPs (8 × 10⁶/mL). The remaining two plates were incubated in HBSS containing 25 μM l-arginine without EMPs. Next, all plates were washed in HBSS containing 25 μM l-arginine. One dish from the control group and one from the EMP group were incubated in 6 mL HBSS containing 25 μM l-arginine and 5 μM A23187. The remaining dishes were incubated in 6 mL HBSS containing 25 μM l-arginine. After 10 min, 500-μL aliquots of supernatant were collected from each dish and •NO (determined as NO₂⁻ + NO₃⁻) concentrations determined by ozone chemiluminescence using a Sievers NO Analyzer, model 280i. A23187-stimulated •NO production was calculated from the difference in •NO concentration in each group without and with ionophore. Results were normalized to total EC protein in each dish.

Western blot analysis

Endothelial cell from •NO measurements were lysed in 500 μL of 4°C modified RIPA buffer (20 mM MOPS, pH 7.0, 2 mM EGTA, 5 mM EDTA, 30 mM sodium fluoride, 40 mM β-glycerophosphate, pH 7.2, 10 mM sodium pyrophosphate) containing 5 μL protease inhibitor (Sigma P-8340) and 5 μL phosphatase inhibitor cocktail I (Sigma P-2850). Phospho-eNOS (Ser-1179) and eNOS levels were determined by Western blot analysis. Briefly, EC lysates were sonicated twice for 30 s on ice at a power setting of 1.25 to 1.5 using a dismembranator (Fisher Scientific Model 550). After sonication, the lysates were centrifuged (20,900 × g, 4°C, 10 min) to remove cell debris. Supernatants were collected and protein concentration was measured by bicinchoninic acid assay (Pierce 23225) (²²). Aliquots of equal protein (50 μg/50 μL) were denatured in Lammeli buffer (95°C, 5 min) and loaded (20 μg/20 μL/lane) on a 7.5% SDS-PAGE gel. To determine how EMPs alter heat shock protein 90 (hsp90) interactions with eNOS, H32 antibody against eNOS was incubated with aliquots of precleared cell lysates protein (1 μg/100 μg) and associated proteins isolated as previously described (²¹). The proteins were transferred to nitrocellulose membranes and then the membranes were blocked in 5% nonfat dry milk in TBS-T (0.15 M NaCl, 0.01 M Tris-HCl, and 0.1% Tween-20) for 1 h at room temperature. The blocked membranes were then incubated with anti-phospho-eNOS (Cell Signaling 9571) or anti-hsp90 (Transduction Laboratories, H38220) overnight at 4°C in TBS-T as per manufacturer's instructions. Bands were visualized using the appropriate horseradish peroxidase-linked secondary antibody and the ECL Plus detection kit (Amersham Biosciences RPN2132) (²³). After probing for phospho-eNOS (S1179), blots were stripped and reprobed for eNOS (Santa Cruz 654). Bands in the autoradiograms were digitally captured using the Adobe Photoshop and Magic Scan software and band densities quantified using NIH Image v1.62. Relative levels of phospho-eNOS to eNOS were calculated from the densitometric ratios.

Western blot analysis of lysed and denatured HUVECs, PAI-1-stimulated HUVECs, and EMPs was performed using similar protocols. After protein quantification, 10 μg of protein from each sample was loaded per lane on a 10% Tris-HCl gel, run for 1 h, and transferred overnight (4°C). After a 1-h blocking period at room temperature (5% Blotto in TBS-T), nitrocellulose membranes were blotted for 2 h at room temperature with the following monoclonal antibodies: CD31(PECAM, Bender BMS137, 1:1000), Annexin V (Bender BMS147, 1:800), Caveolin-1 (Pharmingen 610406, 1:1200), Xanthine Oxidase (NeoMarkers MS-474-P, 1:1000), eNOS (Santa Cruz 654, 1:10,000), PAI-1 (Oxford Biomedical PI 31, 1:1500), C3 (Serotec AHC007, 1:1000), and C5 (Quidel A217, 1:1000). Isotype controls were mouse (Santa Cruz 2025, 1:500) and rabbit IgG (Santa Cruz 2027, 1:500). Bands in the autoradiograms were digitally captured as described above.

Lung injury assays

Studies of EMP-induced ALI were conducted in both 8-week-old male Brown Norway rats (∼250 g, Harlan) and 8-week-old male C57BL/6 mice (∼25 g, Harlan). Two animal models were used to control for species-specific pathogens and the concept synergy between EMPs and a second pathogenic agent or mechanism increases ALI.

Brown Norway lung injury-

Rats were divided into three treatment groups (n = 8). Lungs from four rats in each group were assayed for myeloperoxidase activity and histology. The remaining rats in each group were injected with ¹²⁵I-BSA to assess changes in lung permeability. Animals were killed 6 h after intraperitoneal injection of PBS (250 μL), intraperitoneal injection of S. Typhii LPS (4 mg/kg body weight in 250 μL PBS, Sigma L6511), or intravenous EMPs (4 × 10⁶ per 0.07 g body weight in 250 μL PBS via dorsal penile vein). All treatments were carried out on anesthetized animals [ketamine/xylazine (16.5 mg/1.7 mg)].

Morphometric analysis of hematoxylin and eosin (H&E) was performed using the following parameters: polymorphonuclear cell (PMN) count per 100 periarteriolar cells, edema (0 = none, 1 = moderate, 2 = severe) and RBC (0 = none, 1 = compromise of endothelium-alveolus barrier, 2 = diffuse hemorrhage).

Rats assayed for lung permeability were anesthetized 30 min before they were killed. ¹²⁵I-BSA of 1.6 μCi in quantity (2 mCi/mL, MP Biomedicals 68031.2) was diluted in 200 μL PBS and injected via dorsal penile vein. Before the animals were killed, they were reanesthetized and 3 mL of whole blood was collected from the inferior vena cava for gamma counting. The left atrium was incised, the right ventricle was cannulated, and lungs were slowly flushed with 20 mL of PBS. Lungs were then harvested for gamma counting. Samples were counted for 60 s (Wallac 1470 Wizard). Permeability index was calculated by dividing gamma counts of lung parenchyma by gamma counts in whole blood.

To assay myeloperoxidase activity, an indicator of PMN cell recruitment, the left lung of non-¹²⁵I-BSA-treated rats was removed and homogenized and the rate of peroxide production quantified based on the oxidation of o-dianisidine dihydrochloride (SigmaD-3252, absorbance 460 nm) using the technique of Goldblum et al. (²⁴).

Right lungs from non-¹²⁵I-BSA-treated rats were stored in xylene. Lungs were embedded in Sakura Tissue-Tek O.C.T. compound (IMEB Inc. 4583) and sectioned at 4 μm on a cryostat. Slides were fixed and stained with H&E. Morphometric analysis was performed by a pathologist who was blind to the identity of the slide's treatment group. Three separate and independent areas from each lung were analyzed and values averaged to obtain a global score for each specimen.

C57BL/6 lung injury-

Mice were killed after 6 h of the following treatments: intravenous EMP (4 × 10⁶/mL plasma) + intratracheal LPS (25 μg in 40 μL PBS, E. coli O111:B4, Sigma L2360), intravenous EMPs alone, intravenous PBS + intratracheal LPS, and intravenous PBS alone. All intravenous injections were 200 μL in volume and contained 0.8 μCi of ¹²⁵I-BSA. Intraperitoneal ketamine/xylazine (3.75:0.375 mg in 1 mL [100 μL/25 g mouse]) was administered before all injections and before the animals were killed. Upon killing the animals, 100 μL of whole blood was collected from the inferior vena cava. Lungs were flushed with 3 mL of PBS. Gamma counts (60 s) were performed on a Packard Cobra II and permeability index was calculated as described above for studies using Brown Norway rats.

Statistical analysis

Data are presented as mean ± standard error of the mean. Comparisons between means were either by ANOVA with Bonferroni as a post hoc test or by the Student's paired t test. Statistical significance between means was set at P < 0.05.

RESULTS

Characterization of EMPs

Flow cytometry was used to characterize EMPs and confirm previously reported studies (^{11, 13, 25}). Forward scatter analysis of EMPs in the presence of 3-μm latex beads shows particles predominantly 3 μm in diameter and smaller (Fig. 1). Per previous reports, ultracentrifugation leads to some microparticle conglomeration (^{11, 13, 25}). Relative to isotype control-stained EMPs, histograms demonstrate significant shifts in CD31, tissue factor, Annexin V, and E-selectin. Similar results have been reported with TNF-stimulated EMP release (¹¹). Unstained EMPs demonstrated minimal autofluorescence.

Fig. 1: Characterization of EMPs by flow analysis. Scatterplot shows gate used to detect particles smaller than 3 μm by forward scatter (FSC, SSC axis depicts side scatter). Three-micrometer latex beads are pictured to the right of gate. Adjacent histogram of gated population (P3) shows CD31 staining at excitation wavelength 488 nm in the FITC channel. The P4 subgroup, CD31 positive EMPs, are shown in yellow on the scatterplot. A similar profile is seen in the phycoerythrin channel (PE, data not shown). Endothelium-derived microparticles were not found to have significant autofluorescence (data not shown). In separate experiments, probes for surface CD31, Annexin V, E-selectin, VE-cadherin, and tissue factor in the gated population are shown below. Depicted thresholds in these plots represent fluorescence beyond the first decade, with significant fluorescent labeling occurring above this level on each scatterplot. Latex beads gated separately are shown last with no significant autofluorescence.

Western blot analysis confirms CD31, Annexin V, Caveolin-1, and PAI-1 are present to vary degrees in EMP lysates (Fig. 2). Xanthine oxidase and complement components C3 and C5 were not detected. A faint, but notable band for eNOS was detected in this EMP preparation. PAI-1-stimulated HUVECs differed from nonstimulated HUVECs by the presence of Annexin V (Fig. 2). All HUVECs contained eNOS, but the relative amount of eNOS in EMP preparations was consistently much lower than that observed in the unfractionated HUVEC cultures.

Fig. 2: Characterization of EMPs by Western Analysis. Immunoblots of EMPs lysates isolated from HUVECs stimulated with PAI-1 versus immunoblots of unstimulated HUVEC lysates. Endothelium-derived microparticles and HUVEC lysate proteins were loaded in equal concentrations (10 μg/lane).

Effects of EMPs on vasodilation

Endothelium-derived microparticles markedly inhibited ACh-induced vasodilation of facialis arteries isolated from C57BL/6 mice (Fig. 3). Pretreatment of EMPs with BHT, a lipophilic free radical scavenger, did not significantly improve vasodilation. Similarly, pretreatments of EMP with SOD, an aqueous phase free radical scavenger, did not prevent EMPs from impairing vasodilation. Papaverine dilated all vessels to 95% or greater maximal diameter, demonstrating that the defect in vasodilation was endothelium-dependent. These data suggest that reactive oxygen species are not a primary mechanism by which EMPs impair endothelium vasodilation.

Fig. 3: Endothelium-derived microparticles inhibit endothelium-mediated vasodilation. Acetylcholine (ACh)-mediated vasodilation in facialis arteries pretreated with EMPs (4 × 10⁶/mL) was markedly impaired compared with untreated control vessels (P < 0.01). The inhibitory effects of EMPs were not significantly reversed by pretreating the EMPs with butylhydroxytoluene (BHT; 1 mM) or superoxide dismutase (SOD; 200 U/mL).

Endothelium-derived microparticles significantly attenuated the vasodilation of human microvessels (Fig. 4). Papaverine increased vasodilation in these same vessels to greater than 95% of the original maximal diameter for all vessels suggesting, once again, that this impaired vasodilation in EMP-treated vessels resulted from EC dysfunction, not smooth muscle cell dysfunction. Although inflammation-induced extravascular leak occurs at the postcapillary venule, these studies show that EMPs can acutely induce EC dysfunction as assessed by a nearly complete loss of vasodilatory responses. These findings provide a proof of concept that EMPs are fully capable of impairing EC function in human tissues.

Fig. 4: Endothelium-derived microparticles inhibit endothelium-mediated vasodilation in human microvessels. Acetylcholine (ACh)-mediated vasodilation in human colon arterioles was significantly impaired in microvessels pretreated with EMPs (4 × 10⁶/mL) (P < 0.01). Please note that error bars for EMP-treated vessels are smaller than the open circle symbol.

As a separate control, human colonic arterioles were also treated with PAI-1 (0.083 μg/mL). This dose is equal to the amount of PAI-1 that could be present if 100% of PAI-1 used to generate EMPs were bound to the EMPs used to treat the microvessels. Vessels treated with PAI-1 dilated normally after EMP treatments (91 ± 6% control, 90 ± 8% treated), indicating that attenuation of vasodilation was not a result of PAI-1 contamination.

•Effects of EMPs on EC •NO generation

Coincubation of EMPs with EC cultures decreased A23 187-stimulated •NO production by three-fold (Fig. 5A). To determine the cellular mechanisms mediating this decreased •NO production, we next performed Western blot analysis of phosphorylated Ser1179 on eNOS and total eNOS in EL lysates as well as association of hsp90 with eNOS from the test groups. Phosphorylation of eNOS at serine 1179 increases electron flow through the reductase domain and is often used as an index of eNOS activation (^{26, 27}). Although other phosphorylation sites were probed (serine 116 and threonine 495), no significant differences were found (data not shown). Band densities for Ser1179 on eNOS were normalized to the band densities of the corresponding eNOS for each lane. As can be seen, EMP treatments decreased the phosphorylation (activation) of eNOS at Ser1179 by 50% (Fig. 5B) and association of hsp90 with eNOS (Fig. 5C) in A23187-stimulated EC cultures. Such decreases in eNOS activation and hsp90 association are consistent with the decrease in •NO production in EMP-treated EC cultures and with the impaired vasodilation in EMP-treated facialis arteries and human intestinal microvessels.

Fig. 5: Endothelium-derived microparticles inhibit stimulated EC •NO generation by inhibiting eNOS activation. A, A23187-stimulated (5 μM, 10 min) •NO release from EC cultures in the absence and presence of EMPs (8 × 10⁶/mL). These data show that •NO production in EC cultures incubated with EMPs is significantly attenuated compared with •NO production in untreated EC cultures (P < 0.05). B, EMPs inhibit eNOS activation. Immunoblots show that EMPs decrease eNOS phosphorylation at S1179 in EC cultures stimulated with A23187 (5 μM, 10 min), an index of electron flow through the enzyme. C, EMPs inhibit hsp90 association with eNOS. Immunoblots show that EMPs decrease hsp90 association with eNOS in EC cultures stimulated with A23187.

EMP-induced lung injury

Analysis of ALI in Brown Norway rats treated with PBS, LPS, and EMPs reveal that EMPs induced marked increases in lung injury (Fig. 6). H&E staining of representative periarteriolar sections from each group demonstrates perivascular neutrophil aggregation in the LPS and EMPs groups, with an obvious compromise of the endothelial/alveolar barrier in the EMPs group.

Fig. 6: Effects of EMPs on acute lung injury in Brown Norway rats. A, Representative H&E sections of perivascular lung tissue from each Brown Norway rat treatment group. Note the progression of increasing numbers of perivascular PMNs and edema. Frank hemorrhage is present in the EMP sections (H&E, 20×). Morphometric analysis (see the table illustrate at the bottom) reveals that EMPs induce significant injury. B, Pulmonary endothelial permeability index in rats treated with LPS or EMPs. C, Effects of EMPs on myeloperoxidase activity. Endothelium-derived microparticles significantly increased lung MPO activity compared with the positive control, LPS.

Pulmonary edema and similar interstitial neutrophil counts are seen in both the LPS and EMPs groups. Histology data indicate that EMPs induced marked increases in pulmonary inflammation and capillary leak.

Measured permeability indexes of PBS-, LPS-, and EMP-treated groups were 0.0693 ± 0.080, 0.166 ± 0.037 (P < 0.01), and 0.336 ± 0.123 (P < 0.03), respectively. Similarly, mean tissue myeloperoxidase activities for the groups were 0.071 ± 0.017, 0.109 ± 0.020 (P < 0.01), and 0.137 ± 0.039 (P < 0.02) respectively.

After completing these experiments, necropsy revealed that the Brown Norway rats also carried the following pathogens: Helicobacter hepaticus, Helicobacter rodentium, and Tritrichomonas foetus. The individual contributions of these pathogens to ALI are unknown. However, this discovery and subsequent variability in findings with noncolonized Brown Norway rats suggested that EMPs may be more significant in vivo in a "second hit" type of injury. In other words, colonized or infected animals may be more likely to respond pathologically when exposed to a second stimulus or injury (i.e., an increase in EMPs). To test this hypothesis, EMPs were injected into certified pathogen-free C57BL/6 mice using in a single- and two-hit model. Figure 7 shows that EMPs alone induced significant increase in the permeability than PBS (0.49 ± 0.09, n = 9 vs. 0.27 ± 0.03, n = 10 and P < 0.05), a single-hit model. As a second hit, EMPs induced an even greater increase in permeability than LPS alone (0.86 ± 0.18, n = 8 vs. 0.43 ± 0.04, n = 10 and P < 0.01) or EMPs alone (0.86 ± 0.18 [n = 8] vs. 0.49 ± 0.09 [n = 10] and P < 0.05).

Fig. 7: Effects of EMPs on a one- and two-hit model of ALI in C57BL/6 mice. Endothelium-derived microparticles independently significantly increased lung permeability in a one-hit model compared with PBS controls (0.49 ± 0.09 [n = 9] vs. 0.27 ± 0.03 [n = 10] and P < 0.05). In conjunction with intratracheal LPS, EMPs induced greater increases in lung permeability in a two-hit model: EMPs + LPS versus LPS (0.86 ± 0.18 [n = 8] vs. 0.43 ± 0.04, [n = 10] P < 0.01) and EMPs + LPS versus EMPs (0.86 ± 0.18 [n = 8] vs. 0.49 ± 0.09 [n = 9], P < 0.05).

DISCUSSION

In aggregate, our findings demonstrate that EMPs propagate endothelial dysfunction in vitro and in vivo. We have shown that EMPs impair endothelium-mediated vasodilation, increase pulmonary endothelial permeability, and enhance a pulmonary inflammatory response. Endothelium-derived microparticles likely impair EC function via multiple mechanisms to induce altered systemic vascular responses. Butylhydroxytoluene, SOD, and Western blot data argue against EMPs inducing EC dysfunction by generation of reactive oxygen species or by exposure to complement. Endothelium-derived microparticles may induce EC dysfunction by a release of some unknown signal or by direct EMP-EC interactions. Regardless of the initial step, our eNOS phosphorylation data demonstrate that EMPs, in pathologically relevant concentrations, decreased endothelial intracellular signaling with respect to eNOS phosphorylation and enzyme activation. Such findings provide strong support for the notion that EMPs impair vascular dysfunction by altering EC signaling to decrease generation of •NO, a major mechanism of protection.

Investigating mechanisms by which EMPs impair vascular function requires careful preparation of EMPs and characterization of the EMP preparation. Flow cytometry reveals that our EMP preparations are similar to EMP preparations in studies reported earlier (^{7, 11, 13}). Western blot analysis reveals that our EMPs contained higher concentrations of caveolin-1 than unfractionated stimulated and unstimulated HUVECs. This new information suggests that the release of EMPs may involve direct packaging by perturbed ECs upon stimulation. Our flow data, showing differential inclusion of EC surface proteins on EMPs, support the concept that EMPs are formed by orchestrated release of "ballast" rather than random cavitation of the luminal face of the endothelium. Thus, EMP generation may actually be an early adaptive response of activated or injured ECs aimed at protecting the endothelium. This "ballast" hypothesis prompts a question of vascular ecology: If injured or stimulated ECs release EMPs as an act of self-preservation, could the specific release of EMPs also adversely affect EC function downstream? Our data suggest that this is a very real possibility in that EMPs, in concentrations typically observed in nonacute sickle cell disease, significantly impaired EC function at multiple levels. Further support comes from a recent report from this laboratory demonstrating that EMPs inhibit VEGF-stimulated EC proliferation of human cardiac valve ECs (²⁸). In vitro studies reveal that EMPs impair vasodilation by what appears to be a pro-oxidant mechanism (²⁹), whereas in vivo studies find that EMP counts directly correlated with EC dysfunction in coronary artery disease (³⁰).

At first glance, findings here may appear to be similar to previous studies showing that microparticles inhibited vasodilation (³¹). However, the previous studies used T-cell microparticles, which are not the same as EMPs, vascular tissues only from mice, not humans, and finally, incubations for 24 h rather than 10 min to impair vasodilation. It is important to note that although all cells are capable of generating microparticles, not all microparticles are the same. Recent proteomic studies reveal that even the methods for isolating microparticles can influence protein composition (³²). Intuitively, because membrane protein and lipid composition differ between platelets, T cells and ECs, the effects of microparticles from these different cells may also differ. Evidence supporting this concept comes from the vascular responses to T-cell microparticles and EMPs. After 24-h exposure of aortic rings to T-cell microparticles, ACh dose-response curves were modestly impaired (from ~80% down to ~50%) compared with the nearly ablated vasodilation in murine facialis and human submucosal arteries treated with EMPs for only 10 min. Possible prolonged incubations blunt cell signaling in vascular tissues in response to acute exposure. As no studies were conducted with vessels from humans, it remains unclear how T-cell microparticles will alter human vessel physiology. Another difference between our work and the studies by Martin et al. (³¹) lies in the experimental approach for determining the effects of microparticles on eNOS function. We showed that acute exposure of EC cultures to EMPs decreased stimulated eNOS phosphorylation and association of hsp90, which correlated with a decrease in •NO generation by these same cultures. Although 24-h incubation of EC cultures with T-cell microparticles decreases eNOS protein and increases caveolin-1, no studies were performed to quantify •NO generation in these same cultures. Finally, we showed for the first time that EMPs are not only sufficient to induce ALI but that, in combination with LPS, actually enhance ALI.

The in vivo effects of EMPs observed here suggest that an EMP concentration threshold may exist beyond which endothelial barrier function and vasodilation become compromised. More importantly, such a threshold might be lowered by coinfection or by a second hit as we demonstrated, i.e., intratracheal LPS. Accordingly, these observations lend relevance to clinical pathologic states where EMPs are notably increased (e.g., systemic lupus erythematosus, meningococcal sepsis, thrombotic thrombocytopenic purpura, systemic inflammatory response syndrome, antiphospholipid syndrome, multiple sclerosis, acute coronary syndrome, pediatric vasculitides, preeclampsia, and miscarriage). It should be noted that all of these disease states are capable of causing multiple organ system dysfunction beyond the inciting insult. Indeed, experimental ALI research is embracing multiple hit models in an attempt to approximate the clinical conditions in which lung injury arises (³³). Our findings demonstrate that EMPs alone in pathophysiologically relevant concentrations not only are capable of inducing pulmonary capillary leak but also, in conjunction with a second hit, induce marked increases in ALI.

Exactly how EMPs alter vascular function during sepsis is unclear. Previous studies by Connelly et al. (³⁴) show that LPS induces a transient increase in eNOS phosphorylation (S1177) which peaks within 15 min. Such activation of eNOS during sepsis may actually offset the decrease in eNOS phosphorylation when ECs are exposed to EMPs and then subsequently activated.

Another aspect of this emerging field that probably should be investigated is whether the cellular source of microparticles influences blood and vascular cell function. For example, platelet microparticles are believed to play a role in increasing procoagulant activity (¹⁶). As microparticles containing high levels of tissue factor in sickle cell disease were also found to positively correlate with crisis (⁷), it is interesting to speculate that it may be useful to determine if EMP composition also plays a role in altering downstream vascular EC function. Future studies using detailed proteomic analysis, however, will be required to address this possibility. Finally, therapies aimed at optimizing or maintaining vascular ecology may be an important first step in preventing deleterious effects of the release of EMPs on EC function downstream in a variety of human disease states.

ACKNOWLEDGMENTS

Authors would like to acknowledge Thomas A. Neff for assistance with the mouse lung permeability model. Much thanks to Tara L. Sander and Denise B. Klinkner for intellectual support. Thanks to Hope Albertz and Corbett J. Reinbold at the Blood Research Institute, Milwaukee, WI, for assistance with flow cytometry. Finally, thanks to Emily M. Densmore and William J. Densmore for inspiration, patience, and unwavering support.