Injury-Induced Shedding of Extracellular Vesicles Depletes Endothelial Cells of Cav-1 (Caveolin-1) and Enables TGF-β (Transforming Growth Factor-β)–Dependent Pulmonary Arterial Hypertension

Abstract

Objective—

To determine whether pulmonary arterial hypertension is associated with endothelial cell (EC)–Cav-1 (caveolin-1) depletion, EC-derived extracellular vesicle cross talk with macrophages, and proliferation of Cav-1 depleted ECs via TGF-β (transforming growth factor-β) signaling.

Approach and Results—

Pulmonary vascular disease was induced in Sprague-Dawley rats by exposure to a single injection of VEGFRII (vascular endothelial growth factor receptor II) antagonist SU5416 (Su) followed by hypoxia (Hx) plus normoxia (4 weeks each–HxSu model) and in WT (wild type; Tie2.Cre⁻;Cav1^lox/lox) and EC-Cav1^−/− (Tie2.Cre⁺;Cav1^fl/fl) mice (Hx: 4 weeks). We observed reduced lung Cav-1 expression in the HxSu rat model in association with increased Cav-1+ extracellular vesicle shedding into the circulation. Whereas WT mice exposed to hypoxia exhibited increased right ventricular systolic pressure and pulmonary microvascular thickening compared with the group maintained in normoxia, the remodeling was further increased in EC-Cav1^−/− mice indicating EC Cav-1 expression protects against hypoxia-induced pulmonary hypertension. Depletion of EC Cav-1 was associated with reduced BMPRII (bone morphogenetic protein receptor II) expression, increased macrophage-dependent TGF-β production, and activation of pSMAD2/3 signaling in the lung. In vitro, in the absence of Cav-1, eNOS (endothelial NO synthase) dysfunction was implicated in the mechanism of EC phenotype switching. Finally, reduced expression of EC Cav-1 in lung histological sections from human pulmonary arterial hypertension donors was associated with increased plasma concentration of Cav-1, extracellular vesicles, and TGF-β, indicating Cav-1 may be a plasma biomarker of vascular injury and key determinant of TGF-β–induced pulmonary vascular remodeling.

Conclusions—

EC Cav-1 depletion occurs, in part, via Cav-1+ extracellular vesicle shedding into the circulation, which contributes to increased TGF-β signaling, EC proliferation, vascular remodeling, and pulmonary arterial hypertension.

Visual Overview—

An online visual overview is available for this article.

Endothelial cell (EC) injury is a primary event in several vascular diseases, including pulmonary arterial hypertension (PAH).¹ PAH is a multifactorial disease caused by heritable genetic mutations, persistent inflammation due to parasite (Schistosoma mansoni) or virus (HIV) infection, collagen vascular disease, or of unknown pathogenesis (ie, idiopathic).^2,3 As demonstrated previously by our group, EC Cav-1 (caveolin-1) and BMPRII (bone morphogenetic protein receptor II) depletion are hallmarks of acute lung vascular inflammation,⁴ but whether and how that leads to chronic vasculopathy is unknown. During a persistent or unresolved inflammatory response, increased proliferation and migration of immune, endothelial, and mural cells leads to increased vasoconstriction and microvessel thickening culminating in the formation of vascular lesions (plexiform lesions)—the hallmark of PAH.^5,6 In this sense, although vascular remodeling is a primary component of PAH, the mechanisms associated with pathogenesis and progression from healthy to severely remodeled vasculature are not fully established.

Vessel outgrowth to alleviate hypoxia or to promote vascular repair after injury requires a dynamic switch in resting ECs to a proliferative and migratory phenotype.⁷ In this process, the orchestrated cross talk between evolutionary conserved signaling pathways such as Notch, Wnt/β-catenin, and TGF-β (transforming growth factor-β)/BMP (bone morphogenetic proteins) promotes vascular homeostasis, although persistent activation may also lead to vascular pathology.⁸ Cav-1 and BMPRII are highly expressed in lung ECs and thought to play critical roles in modulating signaling pathways required for physiological EC functions, such as the activation of eNOS (endothelial NO synthase) and NO production. In this sense, autocrine growth factor signaling such as that promoted by BMPs and VEGF (vascular endothelial growth factor) maintains EC morphofunctional heterogeneity by modulating vessel homeostasis and also angiogenesis through an interplay between stalk and tip cell phenotypes.⁹ However, the mechanisms that regulate this cellular switch and the annealing of properly aligned ECs into tubes are not fully understood and may contribute to the disorganization of normal blood vessels resulting in remodeled vasculature as observed in PAH.

Murine lung ECs (MLECs) from global Cav1^−/− mice stimulated with TGF-β undergo endothelial-to-mesenchymal transition^4,10 indicating that Cav-1 expression may be required to maintain the differentiated EC phenotype. Both global Cav1^−/− and conditional deletion of Bmpr2 predisposes mice to develop vascular pathology,^11,12 although the role of altered EC signaling in chronic inflammatory response has not been fully elucidated. Thus, based on accumulating evidence in support of a key role for Cav-1 in maintaining the differentiated EC phenotype and vessel homeostasis, which has been ascribed to some extent to TGFβ/BMP and VEGF signaling,¹³ we assessed in vitro and in vivo whether these signaling pathways are also associated with onset and progression of PAH. Our data support the hypothesis that Cav-1 depletion is at least, in part, dependent on shedding of Cav-1+ extracellular vesicles (EVs) and that in the absence of Cav-1, eNOS-derived oxidants may promote disassembly of endothelial junctions, reduce BMPRII expression, and favor hyperactivation of TGF-β/pSmad2/3 signaling. Cav-1 is thus an important regulator EC homeostasis, and its depletion promotes proliferation of an EC population unable to orchestrate a physiological post-natal angiogenic response, maintain blood vessel homeostasis, or promote vascular repair after endothelial injury, which leads to the development of PAH.

Methods

All supporting data are available within the article (and its online-only Data Supplement).

Human Donors Lung Tissue and Plasma

Deidentified human lung sections (formalin-fixed paraffin-embedded; tissue deemed nonsuitable for transplant) from 1 female and 2 male control donors (age, 27–62 years) and 3 female PAH donors (age, 35–38 years) were acquired from the Center for Heart Lung Innovation lung registry (Protocol Ethics No. H00–50110; University of British Columbia, Vancouver, BC, Canada). Plasma samples were obtained from 14 controls (9 females and 5 males; age, from 22 to ≤53) and 13 idiopathic PAH (9 females and 4 males; age, from 18 to ≤68) donors from the Lerner Research Institute, Cleveland Clinics Foundation (Cleveland, OH; CC IRB No. 10–1117).

Animal Models

Sprague-Dawley rats (250–300 g) were subcutaneously injected with SU5416 (20 mg/kg) and subsequently kept under hypoxia for 4 weeks (10% O₂) plus normoxic conditions (21% O₂) for 4 weeks. SU5416-induced EC injury was evaluated in rats after 1, 3, and 7 days. In addition, C57BL6 (8–12 weeks; Jackson Laboratory, Bar Harbor, ME), EC Cav-1 null mice (EC-Cav1^−/−) generated by crossing Tie2.cre+ mice with Cav1^lox/lox mice,⁴ strain-matched Tie2.Cre⁻, and EC-specific preproendothelin-driven Cav-1 reconstituted (EC Cav-1 RC) mice were also kept under normoxia or exposed to hypoxia for 1 day or 1 month. Pulmonary hypertension (PH) was used in reference to these animal models. Cav1^−/−;eNOS^−/− mice were generated by crossing Cav1^−/− and eNOS^−/− parental strains and maintained for 6 generations by inbreeding. Flk1^GFP−/+ knock-in mice were backcrossed with WT (wild type) and Cav1^−/− B6:129SF2/J for at least 7 generations to achieve 99% chance of congenic background genome in Cav1^+/+;Flk1^+/−GFP and Cav1^−/−;Flk1^+/−GFP hybrid strains. Females differ significantly from males in their response to oxidative stress, which has been associated with the loss of pulmonary EC Cav-1 expression. Therefore, because the present study focused on oxidative stress–linked depletion of Cav-1 in the onset of PAH, only male mice and rats were used. In all cases, strain- and age-matched mice or rats were used as approved by the University of Illinois at Chicago Institutional Animal Care and Use Committee.

Hemodynamics, Right Ventricular Hypertrophy, Lung and Blood Collection

Hemodynamic measurements and assessment of right ventricular (RV) hypertrophy were conducted in anesthetized animals (Ketamine/Xylazine at 100 and 10 mg/kg, respectively) as described previously.¹⁴ Briefly, a Millar Mikro-Tip catheter transducer (model PVR-1030) was inserted into the RV via the right jugular vein. RV systolic pressure (RVSP) was calculated using a MPVS-300 system connected via a Powerlab A/D converter (AD Instruments, Colorado Springs, CO). After recordings, the animals were ventilated, and blood was collected with heparin-treated or 3.8% sodium citrate–treated syringes via cava vein. After full lung perfusion with cold PBS, the lung lobes were either removed and snap-frozen in liquid nitrogen for Western blotting or carefully inflated with 4% paraformaldehyde solution for histological analysis. Finally, the heart was dissected for RV hypertrophy evaluation (RV/left ventricle+septum weight ratio).

Immunocytochemistry and Immunohistochemistry

Cultured ECs were fixed with 4% paraformaldehyde solution for 5 to 15 minutes at room temperature. After permeabilization, cells were blocked with 10% donkey or goat serum diluted in PBS for 1 hour (room temperature) followed by overnight incubation with primary antibodies at 4°C (in humidified chamber). After washing, slides were incubated with secondary antibodies, washed again, and mounted using Prolong antifade mounting media plus DAPI (4’,6-diamidino-2-phenylindole). After antigen retrieval, deparaffinized lung sections were blocked and incubated as described above. Masson trichrome images were collected using Zeiss Apotome brightfield microscope, and lung microvessel area and thickness were quantified using ImageJ software (https://imagej.nih.gov/ij/). Fluorescent images were collected using confocal microscope (Carl Zeiss).

Plasma Extracellular Vesicle Isolation and Characterization

Differential centrifugation and filtration methods were adapted for isolation of plasma microvesicles, exosomes, and small apoptotic bodies (sABs).^15,16 Briefly, blood collected using 3.8% sodium citrate solution was slowly centrifuged for 10 minutes (4°C; 300g). Then, 250 μL of plasma from mice and humans and 700 μL from rats, respectively, was centrifuged twice for 20 minutes (1500g; 4°C). Platelet-free plasma was either filtered (0.8 μm mesh) and centrifuged at 12 200g (40 minutes; 4°C) for isolation of microvesicles or 16 100g for 20 minutes for collection of microvesicles+sABs. In both cases, the pellet was either resuspended in PBS or RIPA (radioimmunoprecipitation assay) buffer and supernatant was filtered (0.2 μm mesh) and ultracentrifuged at 100 000g (60 minutes; 4°C) for exosome isolation. EVs were diluted in double-filtered PBS for determination of size and concentration by Nanoparticle Tracking Analysis (NanoSight; Malvern Panalytical) or resuspended in RIPA buffer plus protease and phosphatase inhibitor cocktail for Western blot. EV morphology was assessed by transmission electronic microscopy.

Isolation and Culture of Bone Marrow–Derived Macrophages

Rat and murine bone marrow–derived macrophages (BMDMs) were isolated and differentiated as described previously.¹⁷ Briefly, bone marrow cells were washed from the femur and tibia and differentiated for 7 days in RPMI (Roswell Park Memorial Institute media) containing 10% heat-inactivated fetal bovine serum (Gemini Bio-Products, West Sacramento, CA), 1% (vol/vol) antibiotic-antimycotic, 1 mmol/L sodium pyruvate (both from Thermo Fisher Scientific, Waltham, MA) and 10% to 15% L929 conditioned cell culture medium. Differentiated BMDMs were either treated with 1 ug/mL lipopolysaccharide, 10 ng/mL IL (interleukin)-13, or EVs derived from control mice and rats, D7 SU5416-treated rats, and lipopolysaccharide-treated mice (24 hours after exposure to 10 mg/mL nebulized lipopolysaccharide for 1 hour).

ELISA and Blood Cell Counting

Heparinized blood was used to determine blood cell percentage by automatic cell counter. TGF-β1 levels were measured in plasma from WT and EC-Cav1^−/− mice, human subjects, and BMDM supernatants according to manufacturer instructions (R&D ELISA kits). Human or rat plasma Cav-1 levels were also measured according to manufacturer instructions (Biomatik ELISA kits).

Human Pulmonary Artery ECs

Human pulmonary artery ECs (HPAECs; from Lonza; passage 3–7) were exposed to 5 μmol/L SU5416, 10 ng/mL VEGF, 10 ng/mL TGF-β, and 0.5 or 1 mmol/L 3-morpholinosydnonimine (SIN-1) daily for ≤72 hours. HPAECs were observed at 24, 48, and 72 hours by phase-contrast microscopy. Cell lysates were prepared using RIPA buffer containing protease and phosphatase inhibitor cocktail.

Western Blotting

Lung tissue and HPAECs lysates (10 or 30 μg, respectively) were loaded and run on gradient SDS-PAGE gels (8%–12%) or 12% gel and transferred to nitrocellulose membranes as described.¹⁸ Total protein loading in the membrane was assessed by Pounceau Rouge staining, and membranes were either scanned or washed with TBS-Tween 1×. Membranes were then blocked using 5% milk or BSA for 1 hour followed by primary antibody incubation (overnight at 4°C or 2–3 hours at 37°C). After washing, membranes were incubated for 1 hour with specific HRP-conjugated antibody, and proteins were detected by using an ECL kit (Amersham, Piscataway, NJ) and scanned with a Li-Cor Odyssey CLx (Lincoln, NE). Similarly, dot blot was performed to evaluate TGF-β expression in BMDM supernatant. Data were normalized to β-actin or GAPDH loading controls and analyzed using ImageJ Software (https://imagej.nih.gov/ij/).

Quantitative Real-Time Polymerase Chain Reaction

After treatments, total RNA from BMDMs was isolated using TRIzol reagent (Life Technologies) according to the manufacturer’s instructions. RNA was quantified using a spectrophotometer (NanoDrop), and cDNA was synthesized from 1 μg of total RNA using the high-capacity cDNA reverse transcription kit with RNAse inhibitor (Invitrogen). SYBR Select Master Mix (Applied Biosystems) was used for quantitative real-time polymerase chain reaction. Reactions were performed using 2 μL cDNA and 100 nmol/L of each of the reverse and forward primers. The following primers were used for quantitative real-time polymerase chain reaction: NOS2 forward: 5′-CAC AGT GTC GCT GGT TTG AA-3′; NOS2 reverse: 5′-TCT CCG TGG GGC TTG TAG TT-3′; Arg1 forward: 5′-TGG ACC CTG GGG AAC ACT AT-3′; Arg1 reverse: 5′-GTA GCC GGG GTG AAT ACT GG-3′; Tgfb1 forward: 5′-CTT TGT ACA ACA GCA CCC GC-3′; Tgfb1 reverse: 5′-TAG ATT GCG TTG TTG CGG TC-3′; Gapdh forward: 5′-AGT GCC AGC CTC GTC TCA TA-3′; Gapdh reverse: 5′-GAC TGT GCC GTT GAA CTT GC-3′. Reactions were performed in a ViiA7 Real-Time System (ThermoFisher Scientific). Gapdh was used as an internal control to calculate relative mRNA levels.

BrDU Incorporation Assay

Control or Cav-1 shRNA-transfected HPAECs were plated in a 96-well plate (black plates; clear bottom) and exposed to 10 ng/mL TGF-β daily for ≤48 hours. BrDU (bromodeoxyuridine/5-bromo-2’-deoxyuridine) incorporation indicative of cell proliferation was measured by fluorimetry according to the instructions of the manufacturer (Novus Biologicals).

EC Isolation and Culture

MLECs were isolated as described previously.¹⁹ Briefly, perfused lungs from WT;Flk1^+/−GFP, Cav1^−/−;Flk1^+/−GFP, and Cav1^−/−;eNOS^−/− mice were minced into 2- to 5-mm fragments, rinsed with PBS, and incubated in 0.1% to 0.2% collagenase I solution in DMEM (30–60 minutes at 37°C under agitation [≈80 rpm]). After fragments were dissociated and filtered (100 and 40 μm mesh), cell suspension was centrifugated (300g; 8 minutes), resuspended in growth medium, and assayed for viability using Trypan blue reagent (2–4×10⁶ cells per adult lung). Isolated cells were incubated for 6 to 8 h (37°C; 5% CO₂), and outgrowths with characteristic features of ECs were selected and subcultured using cloning disks. For sorting, cells were resuspended in 0.5% BSA/2 mmol/L EDTA solution and sorted with a MoFlow high-speed sorter (Dako-Cytomation). Purity of ECs was verified by CD31/PECAM-1 (cluster of differentiation 31/platelet and endothelial cell adhesion molecule 1) staining, and in case of contamination by CD31⁻ cells, cultures were either discarded or recloned and reevaluated using the same criteria.

Growth Media Formulation

Basal medium was a 1:1 (vol:vol) mixture of high glucose DMEM and EBM-2 (endothelial basal medium-2) containing 15% HyClone characterized fetal bovine serum, 0.5 μg/mL of heparin, 10 ng/mL VEGF-165, 5 μg/mL endothelial cell growth supplement, glutamine, 50 μg/mL penicillin, and 10 μg/mL streptomycin.

Statistical Analysis

Data were analyzed using GraphPad Prism v7. Normally distributed data are presented as the arithmetic means and SD. Non-normally distributed data are presented as median with interquartile range. Shapiro-Wilk test was used to determine normality of data and Brown-Forsythe or F test to determine equality of variances, and then a parametric or nonparametric test was performed accordingly. Parametric statistical analysis was performed using unpaired Student t test (2 tailed) between 2 groups and 1-way ANOVA followed by post hoc analysis (Bonferroni, Dunnett, or Newman-Keuls Multiple Comparison test) for analysis of differences between >2 groups. Nonparametric statistical analysis was performed using Mann-Whitney U test. Tests used are specified in the legend of each figure, and P <0.05 was considered statistically significant.

Reagents and Antibodies

Antibodies

Rabbit polyclonal anti-eNOS, anti-TGFRI (transforming growth factor-β receptor I), and anti-GAPDH were acquired from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal anti-CD31 (PECAM-1), rabbit polyclonal anti-pSMAD2/3, and anti-BMPRII were purchased from Cell Signaling Technology (Danvers, MA). Rabbit polyclonal anti-TGF-β1 was obtained from Bioss Antibodies (Woburn, MA). Mouse monoclonal anti-eNOS, rabbit polyclonal anti–Cav-1, and rat anti-mouse CD31 (clone MEC 13.3) were purchased from BD PharMingen (San Jose, CA). Alexa-Fluor 488-, 555-, and 546-conjugated goat anti-mouse and anti-rabbit IgG were purchased from Life Technologies (Grand Island, NY). Anti-mouse and anti-rabbit HRP-conjugated IgG were purchased from Cell Signaling Technology (Danvers, MA) or Kirkegaard & Perry Laboratories (Gaithersburg, MD; Table I in the online-only Data Supplement).

Reagents

SIN-1 hydrochloride, SU5416, Escherichia coli lipopolysaccharide, RIPA buffer, protease inhibitor cocktail, collagenase type I, heparin, and sucrose were purchased from SIGMA Chemical, Co (St. Louis, MO); recombinant TGF-β and IL-13 were obtained from Peprotech (Rocky Hill, NJ). Mounting media was obtained from Vector (Burlingame, CA). DMEM, penicillin, streptomycin, glutamine, trypsin-EDTA, and prolong antifade mounting media were from Invitrogen (Carlsbad, CA); EBM-2/EGM-2 (endothelial cell growth medium-2) SingleQuots and fetal bovine serum were purchased from Lonza (Walkersville, MD) and Gemini Bio-Products (West Sacramento, CA); endothelial cell growth supplement and bovine plasma fibronectin were from Biomedical Technologies (Stoughton, MA); HyClone characterized fetal bovine serum and poly-L-lysine–coated microscope slides were from Fisher Scientific (Pittsburgh, PA); hrVEGF-165 (human recombinant vascular endothelial growth factor-165) and hrbFGF (human recombinant basic fibroblast growth factor) were from Shenandoah Biotechnology (Warwick, PA); donkey or goat serum from Jackson ImmunoResearch (West Grove, PA); stock solutions were prepared in 100% dimethylsulphoxide, buffered physiological solution, or sterile PBS and diluted daily in buffered physiological solution or sterile PBS. The highest final concentration of the solvent was 0.1% (v/v) and had no effect on the experiments.

Results

Reduced EC Cav-1 Expression Is Associated With Vascular Remodeling

Immunohistochemical analysis of lung sections from patients with PAH showed reduced Cav-1 expression in microvascular CD31+ cells when compared with healthy control donors (Figure 1A), indicating Cav-1 depletion is associated with vascular remodeling. Quantitative analysis of rat lung Cav-1 expression after HxSu exposure—a widely-accepted animal model of PAH (Figure IA and IB in the online-only Data Supplement)—revealed a similar reduction in Cav-1 expression (Figure 1B). To further investigate the role of EC Cav-1 in the onset of PAH, WT (Tie2.Cre⁻;Cav1^lox/lox), EC-Cav1^−/− (Tie2.Cre⁺;Cav1^lox/lox), and EC Cav-1 RC mice were exposed for 1 month to hypoxia or kept under normoxia. In response to hypoxia alone, RVSP increased in WT mice and was even higher in EC-Cav1^−/− mice. However, RVSP in EC Cav-1 RC mice exposed to hypoxia was similar to the WT group (Figure 1C). Increased RVSP was associated with mild cardiac hypertrophy in all groups after hypoxia exposure (Figure 1D). There was a significant increase in vessel area and thickness observed in EC-Cav1^−/− mice when compared with other groups, suggesting the absence of EC Cav-1 promotes murine pulmonary microvascular remodeling (Figure 1E through 1G). No difference in total Cav-1 expression was observed in the lung lysates of WT mice exposed to hypoxia when compared with normoxia (Figure IC online-only Data Supplement) indicating 1 month of hypoxia alone does not reduce Cav-1 expression in the murine pulmonary vasculature. Taken together, these data implicate EC injury induced by VEGFRII (vascular endothelial growth factor receptor II) inhibition in the mechanism of loss of EC Cav-1 expression.

Figure 1.

Endothelial Cav-1 (caveolin-1) depletion induces pulmonary vascular remodeling and pulmonary arterial hypertension. A, Lung sections from normal control and pulmonary arterial hypertension (PAH) donor patients showing Cav-1 (red) and CD31 (cluster of differentiation 31; green) expression in pulmonary microvessels (scale bar=100 μm). Nuclei were stained using DAPI (4’,6-diamidino-2-phenylindole). Images are representative of 3 different patients from each group. Inset: negative control (NC). MERGE indicates overlay of red and green colors. B, Western blot showing Cav-1 expression in total lung lysates from Sprague-Dawley rats under normoxia (Nor) or after hypoxia plus a single injection of SU5416 (VEGFRII [vascular endothelial growth factor receptor II] antagonist; 20 mg/mL; HxSu (hypoxia+SU5416); n=4 animals per group). WT (wild type; Tie2.Cre⁻;Cav1^lox/lox), endothelial cell (EC)-Cav1^−/− (Tie2. Cre⁺;Cav1^fl/fl), and EC Cav-1 reconstituted (EC Cav-1 RC) mice kept under Nor or hypoxia for 1 mo were used to evaluate (C) right ventricular (RV) systolic pressure (RVSP) and (D) heart hypertrophy (ratio of the RV by the left ventricle [LV] plus the septum [S]: RV/[LV+S]). Mouse lung sections from WT and EC-Cav1^−/− were stained using Masson trichrome (E) and the microvasculature area (F; μm²) and thickness (G; μm²) were quantified from micrographs (n=7–8 animals per group; scale bar=20 μm). Student t test or 1-way ANOVA followed by Newman-Keuls was used for statistical comparisons. CTR indicates control. *P<0.05, **P<0.01, ***P<0.001.

EC Cav-1 Depletion Is Partially Dependent on EV Shedding

The mechanism by which EC Cav-1 is depleted in the lungs is not fully understood. Because Cav-1 is a membrane protein, we hypothesized that Cav-1 depletion may be due in part to membrane blebbing and shedding into the circulation in the form of EVs (ie, exosomes, microvesicles, and apoptotic bodies [ABs]). Plasma EVs isolated from WT and global Cav1^−/− mice revealed Cav-1 was highly expressed in EVs >100 nm (ie, microvesicles/sABs) when compared with smaller EVs (Figure 2A and 2B). Expression of CD31 was detected in all fractions (data not shown). The HxSu rat model or 1-day exposure of mice to Hx alone did not significantly alter plasma Cav-1 levels compared with controls, but Cav-1 expression in EVs was significantly increased in response to lipopolysaccharide-induced acute EC injury (Figure IIA in the online-only Data Supplement). Similarly, VEGFRII inhibition can promote EC injury/death,^20,21 which we confirmed using 3 days of SU5416 treatment in vitro (Figure IIB in the online-only Data Supplement). Thus, to test whether SU5416-induced EC injury alone promotes Cav-1+ EV shedding, rats were treated with SU5416 for 1, 3, or 7 days. No significant difference in plasma microvesicles/sABs and Cav-1 level was observed between rats treated with vehicle or SU5416 for 1 or 3 days. However, plasma Cav-1 concentration increased 7 days after administration of SU5416 to rats (Figure 2C and 2D). Analysis of the histogram of the population of EVs indicates there is a time-dependent increase in plasma microvesicles/sABs after SU5416 exposure and that an increase in plasma Cav-1 level is primarily because of shedding of microvesicle/sABs (Figure 2E and 2F). In addition, at the same time point that plasma Cav-1+ microvesicles/small ABs peak, circulating monocytes were significantly decreased (Figure 2G), which may indicate there is an interplay between SU5416-induced Cav-1 depletion and monocyte recruitment and extravasation. No differences in the percentage of lymphocytes, neutrophils, basophils, or eosinophils were observed. Thus, while vascular injury may prime ECs by inducing EV release and Cav-1 depletion, chronic hypoxia may provide the second hit that promotes vascular remodeling.

Figure 2.

Endothelial cell Cav-1 (caveolin-1) depletion is partially dependent on shedding of Cav-1+ extracellular vesicles. A, Western blot showing Cav-1 expression in different sizes of plasma extracellular vesicles (EVs; 1: <1 μm; 2: 1–4 μm; 3: <0.2 μm) isolated from WT (wild type) or Cav1^−/− mice. LAMP-1 (lysosome-associated membrane protein 1) was primarily expressed in small EVs and Cav-1 in larger EVs. Inset: transmission electron micrographs showing a microvesicle (MV; top) and exosome (exo; bottom; scale bar=200 nm). B, Quantitative analysis of Cav-1 expression in the different WT EVs (MVs; small apoptotic bodies [ABs], and exo) normalized by Ponceau staining. C, Rat plasma Cav-1 level from vehicle control (Ctr) or rats treated with 20 mg/kg SU5416 and collected after 1, 3, and 7 d (Su D1, D3, and D7). Inset: representative image of Cav-1 level measured by ELISA. D and E, Nanosight analysis of plasma EVs isolated from Ctr rats or rats treated with Su (D1, D3, or D7). F, Cav-1 concentration (ELISA) associated with isolated microvesicles (MVs)/small ABs vs MV/AB-free plasma from vehicle Ctr or after 7-d SU5416 treatment. G, Percentage of blood lymphocytes, neutrophils, monocytes, basophils, and eosinophils in Ctr or Su D1, D3, or D7 (*P<0.05; n=3–4 animals per treatment). One-way ANOVA followed by post hoc test Newman-Keuls was used for statistical comparisons.

EC Cav-1 Depletion Is Associated With Altered TGF-β/ALK5 Signaling

The increase in plasma EV density on EC injury raised the possibility that the Cav-1+ EVs may act as EC damage–associated signals or as decoys for the immune system. To test whether Cav-1–enriched EVs interact with macrophages, EVs isolated from D7-treated rats were coincubated with BMDMs, and expression of inflammatory mediators was measured. Both NOS2 and Arg1 expressions were upregulated by SU5416-derived EVs when compared with control EVs. However, Tgfβ1 expression was only upregulated on exposure to the higher concentration of EVs from SU5416-treated rats (Figure 3A). Similarly, increased active TGF-β1 was detected in the supernatant of BMDMs after treatment with lipopolysaccharide-derived EVs (Figure 3B). Thus, to determine the role of Cav1^−/− microvesicles/ABs in BMDM-mediated TGF-β production, BMDMs were treated with WT or Cav-1^−/−–derived microvesicles/ABs and exposed to normoxia or hypoxia in vitro for 48 hours. No differences in particle number (Figure IIIA in the online-only Data Supplement) or TGF-β production induced by microvesicles/ABs from WT versus Cav1^−/− mice were observed at baseline. However, under hypoxia, a significant increase in TGF-β production was observed in response to WT mouse microvesicles/ABs compared with Cav1^−/− microvesicles/ABs, indicating injury-induced shedding of Cav-1+ microvesicles/ABs promotes TGF-β production by macrophages (Figure 3C). IL-13 was used as positive control (mean value, 489.6 pg/mL). In vivo, hypoxia-induced PH in EC-Cav1^−/− mice was associated with increased plasma levels of TGF-β when compared with control mice (Figure 3D). Moreover, plasma TGF-β level was increased in 1-year-old global Cav1^−/− mice at the same time RVSP was elevated as compared with WT control 1-year-old mice (Figure 3E and 3F). Interestingly, plasma TGF-β level was not different in 1-year-old EC-Cav1^−/− mice. Then, to further investigate whether increased TGF-β level correlates with activation of TGF-β signaling, we assessed expression levels of TGFRI(ALK5 [activin receptor-like kinase 5]), BMPRII, and pSMAD2/3 in the lung of Cav1^−/− mice aged 3, 8, and 12 months. Western blot analysis revealed no difference in total Cav-1 or TGFRI expression (Figure IIIB through IIID in the online-only Data Supplement; Figure 3F). However, expression of BMPRII (short form+long form) was reduced in 1-year-old Cav1^−/− mice when compared with mice aged 3 and 8 months. No significant differences were observed in WT mice (Figure 3G and 3H). Consistent with the decrease in BMPRII, we observed an increase in pSMAD2/3 expression (Figure 3G and 3I). Taken together, these data indicate prolonged absence of Cav-1 expression or exposure to hypoxia contribute to altered TGFRI(ALK5)/BMPRII signaling and development of PH.

Figure 3.

Pulmonary hypertension mediated by loss of Cav-1 (caveolin-1) is associated with activation of TGF-β (transforming growth factor-β) signaling. A, NOS2, Arg1, and Tgfβ1 expression in rat bone marrow–derived macrophages (BMDMs) treated with control (Ctr) or SU5416-derived extracellular vesicles (EVs; 5×10⁹ or 1×10¹⁰ EV/mL) for 24 h. Lipopolissacarideo (LPS; 1 μg/mL) and IL-13 (interleukin-13; 10 ng/mL) were used as positive Ctr. B, TGF-β1 level in the supernatant of murine BMDMs treated with Ctr or LPS-derived EVs (10⁷–10¹⁰ EVs/mL) for 24 h. LPS (1 μg/mL) was used as Ctr. Inset: dot blot showing total TGF-β expression in the supernatant of BMDMs after treatments. C, TGF-β1 level in supernatant of BMDMs treated with 10¹⁰ Ctr mouse–derived microvesicle/ABs and exposed to normoxia or hypoxia in vitro for 48 h. D, WT (wild type; Tie2.Cre⁻;Cav-1^lox/lox) and endothelial cell (EC)–specific-Cav1^−/− (Tie2. Cre⁺;Cav-1^fl/fl) mice exposed to normoxia or hypoxia for 1 mo were used to evaluate plasma level of TGF-β. WT and global Cav-1 knockout mice (Cav1^−/− KO) aged 3, 8, or 12 mo were used to evaluate right ventricular systolic pressure (RVSP; E) and plasma level of TGF-β (F); lung expression level of Cav-1, TGFβRI/ALK-5 (transforming growth factor-β receptor I/activin receptor-like kinase 5), (F); BMPRII (bone morphogenetic protein receptor II) short form (SF) and long form (LF; G and H), and pSMAD2/3 (G and I; *P<0.05 and ##P<0.01; n=4–6 animals per group). Student t test or 1-way ANOVA followed by post hoc test (Bonferroni or Newman-Keuls) was used for statistical comparisons.

Reduced Cav-1 Expression May Promote Remodeling by Altering TGF-β Signaling

We next evaluated the role of TGF-β signaling and VEGFRII inhibition in vitro. HPAECs were treated with SU5416, SIN-1, VEGF, and TGF-β and observed for ≤72 hours. Western blot analysis of adherent cell lysates revealed a reduction in the expression of Cav-1 as reported previously (Figure 4A and 4B).^20,21 In this pool of cells, we also observed a partial reduction in BMPRII expression. There was no effect of VEGF and TGF-β on Cav-1 and BMPRII expression at this time point (Figure 4A through 4C). Reduced EC Cav-1 expression was previously shown to be associated with increased peroxynitrite production via dysfunctional eNOS.²² While in vitro treatment with peroxynitrite donor (SIN-1) did not significantly alter Cav-1 expression, a significant reduction in BMPRII expression was observed (Figure 4A through 4C). In addition, treatment with SIN-1 altered EC morphology from quiescent cobblestone to spindle-shape mesenchymal-like morphology based on the junctional pattern of VE-cadherin staining (Figure 4D). This mesenchymal-like cell profile was also observed in MLECs isolated from Cav1^−/− mice as compared with WT MLECs and prevented by the additional deletion of eNOS^−/− in Cav1^−/− mice (Figure 4E). Moreover, in the absence of Cav-1, discontinuous junctional PECAM-1 staining was noted (Figure 4F), which was also normalized in Cav1^−/−;eNOS^−/− MLECs. Finally, Cav1 shRNA-transfected HPAECs treated with TGF-β showed greater BrDU incorporation when compared with Cav1 shRNA alone or control shRNA-transfected cells (Figure 4G). At 48 hours, no significant difference was observed between Cav1 shRNA or TGF-β treatment alone. These data suggest that in the absence of EC Cav-1 expression, dysfunctional eNOS may contribute to BMPRII depletion and loss of contact inhibition critical for the quiescent EC phenotype.

Figure 4.

Depletion of Cav-1 (caveolin-1) promotes expansion of a morphofunctionally altered endothelial phenotype. A–C, Cell lysates from adherent human pulmonary artery ECs (HPAECs) treated with SU5416, VEGF (vascular endothelial growth factor), TGF-β (transforming growth factor-β), or 3-morpholinosydnonimine (SIN-1) for 72 h were used to evaluate Cav-1 and BMPRII (bone morphogenetic protein receptor II) expression. D, Immunocytochemistry analysis of junctional VE-cadherin (green) and Cav-1 (red) expression after 1 mmol/L SIN-1 treatment of HPAECs (scale bar=20 μm). Top, Negative control (NC). E, Phase-contrast micrograph showing morphology of WT (wild type), Cav1^−/− and Cav1^−/−;eNOS^−/− murine lung endothelial cells (MLECs). F, Immunocytochemistry showing Cav-1 (pink), eNOS (endothelial NO synthase; green), and CD31 (cluster of differentiation 31; red) expression in WT, Cav1^−/−, and Cav1^−/−;e NOS^−/− MLECs, respectively. Nuclei were stained with DAPI (4’,6-diamidino-2-phenylindole). G, BrDU (bromodeoxyuridine/5-bromo-2’-deoxyuridine) incorporation in control (Ctr) or Cav-1 shRNA-transfected HPAECs after stimulation with vehicle (Veh) or 10 ng/mL TGF-β (*P<0.05; n=3 independent experiments). One-way ANOVA followed by post hoc test (Dunnet or Newman-Keuls) was used for statistical comparisons. LF indicates long form. *P<0.05.

Increased Circulating Cav-1, TGF-β, and EVs in Idiopathic PAH Patients

To investigate whether the results observed in animal models correlate with human pathology, plasma from patients with idiopathic PAH and healthy donors was used to evaluate the levels of microvesicles/sABs, Cav-1, and TGF-β. Consistent with rat and mouse PAH studies, a greater number of circulating microvesicles/sABs was observed in PAH donor patients when compared with healthy donors (Figure 5A). Plasma Cav-1 and TGF-β concentration were also increased (Figure 5B and 5C), suggesting shedding of Cav-1 into the circulation correlates with the development of PAH in 46% of the patient cohort evaluated. There was no evidence of correlation of plasma Cav-1 level or particle number with sex or age (data not shown). Thus, these data suggest EC Cav-1 depletion in lung microvessels (Figure 1A) and its appearance in the circulation in microvesicles/ABs may serve as a biomarker of EC injury and disorganized vascular repair associated with microvascular remodeling and PAH.

Figure 5.

Human plasma from patients with pulmonary arterial hypertension (PAH) showed a significant increase in particle number, Cav-1 (caveolin-1), and TGF-β (transforming growth factor-β) levels. A, Nanosight analysis showing average number of particles (microvesicles+small apoptotic bodies [ABs]) isolated from 250 μL of plasma from healthy control or idiopathic PAH donors (n=8 donors per condition). B, Cav-1 concentration (pg/mL) in the plasma from control or PAH donor patients measured by ELISA. C, TGF-β1 concentration (pg/mL) in the plasma from control and PAH donor patients measured by ELISA (Mann-Whitney U test; *P<0.05; n=13–14 donor patients per condition).

Discussion

A heterogeneous EC population in the lung vascular network is critical for tissue homeostasis.²³ In this sense, Cav-1 expressing ECs seem to be required for normal pulmonary vascular function, including the regulation of eNOS activity.²⁴ Previously, we and others reported that EC Cav-1 depletion is a key component of PAH,^25–27 and recent reports indicate pulmonary ECs lacking Cav-1 expression exhibit a non-EC phenotype. It has been suggested that the altered phenotype of Cav-1–depleted ECs may be because of TGF-β–induced endothelial-to-mesenchymal transition or simply reflect the intrinsic characteristics of an EC subpopulation with a pathogenic phenotype.^1,10,28 In this scenario, compelling evidence obtained by our group indicates that pulmonary EC Cav-1 depletion is a hallmark of inflammation-induced vascular pathology as observed in acute lung injury/acute respiratory distress syndrome⁴ and as shown here in HxSu-induced PH and human PAH. More importantly, the present study raised the possibility that EC Cav-1 depletion via release of EVs further functions as DAVES (damage-associated vesicular endothelial signals) that stimulate TGF-β–dependent reparative responses by activating and recruiting circulating immune cells.

Reduced Cav-1 expression has been reported in different pulmonary inflammatory diseases.²⁸ Previous work from our group showed Cav-1 nitrosation, phosphorylation, and ubiquitinylation in ECs from patients with idiopathic PAH that resulted in Cav-1 degradation despite increased Cav-1 mRNA expression.²⁵ Here, we have shown that injury-induced EC Cav-1 depletion occurs, in part, via shedding of Cav-1–containing EVs, indicating that EV shedding and proteasomal degradation can reduce the total level of Cav-1 protein in ECs during PAH. We also observed that Cav-1+ EVs interact with macrophages and stimulate expression of both proinflammatory and anti-inflammatory genes and secretion of TGF-β. In fact, increased TGF-β production was observed only in response to exposure to the highest concentration of EVs tested and in particular, in the presence of hypoxia. Thus, enhanced EV shedding in response to vascular injury, as opposed to basal shedding of EVs, plays an important role in promoting TGF-β–mediated signaling associated with PAH. Consistent with this conclusion, our data further showed an increase in the density of EVs and of Cav-1 immunoreactivity in plasma of patients with idiopathic PAH. Interestingly, an increase in total plasma EVs was also recently observed in patients with PH associated with congenital heart disease.²⁹ Cav-1 expressing EVs have also been associated with malignancy in breast cancer and recently implicated in EC-adipocyte communication.^30,31 Whether and how Cav-1–containing EVs interact only with macrophages or whether additional interactions contribute to the onset or severity of PAH remains to be determined.

During inflammatory vascular conditions, an imbalance in growth factor signaling promotes inhibition of signaling pathways necessary for EC quiescence and vascular repair, and this may progressively lead to microvascular injury and remodeling as observed in PAH.^4–6 Withdrawal of survival factors such as VEGF promotes EC apoptosis and subsequent selection of apoptosis-resistant EC subpopulations.^20,32 In this sense, we have demonstrated that SU5416-induced EC apoptosis is associated with reduction of EC Cav-1 expression via shedding of Cav-1+ EVs. In vitro, SU5416 induced both Cav-1 and BMPRII depletion, whereas the peroxynitrite donor SIN-1 only induced BMPRII depletion in HPAECs, suggesting reactive nitrogen species accumulating in the absence of Cav-1 may promote the depletion of BMPRII. In fact, TGFRI(ALK5) and reduced BMPRII signaling have been implicated in the development of PAH, although the role of TGF-β–mediated signaling in the regulation of homeostatic versus pathological EC phenotype is still controversial.^33,34

TGF-β is a potent stimulus of EC transdifferentiation³⁴ and promotes its effects by binding to the type II TGFβ receptor (TGFβRII), which then in turn phosphorylates and activates type I receptors. Two type I receptors, ALK1 and ALK5, can be activated by TGF-β.^35,36 Whereas ALK1 is known to be an EC-specific receptor that can be activated by BMPRII leading to Smad1/5/8 phosphorylation in healthy ECs, ALK5 promotes Smad2/3 phosphorylation and inhibition of healthy EC functions. In this regard, TGF-β signaling may contribute to vascular repair but can also promote formation of dysfunctional vasculature. Our data indicate that increased plasma levels of TGF-β and decreased BMPRII expression in global Cav1^−/− mice occurred at the same time the mice developed PH. Moreover, increased TGF-β–driven PH was also associated with EC Cav-1 depletion via shedding of Cav-1+ EVs indicating at least 2 mechanisms are involved in Cav-1 depletion–dependent PH, and both require TGF-β–dependent ALK5-activated Smad2/3 signaling in the pulmonary vasculature.

Previously, it was demonstrated that Cav-1 reconstitution driven by an EC-specific preproendothelin promoter rescued the aberrant lung vascular phenotype and PH observed in global Cav1^−/− mice.^37,38 It is also well established that Cav-1 negatively regulates eNOS activity.³⁹ We observed that reconstitution of EC Cav-1 expression using the Cav-1-RC mouse model also protects against hypoxia-induced PH. Thus, in the context of reduced EC Cav-1 expression, hyperactivated eNOS may contribute to the alterations in lung vasculature seen in Cav1^−/− mice, as suggested previously.^40–42 Dysfunctional eNOS (ie, uncoupled) is associated with reduced NO production in parallel with an increase in peroxynitrite accumulation mediated by the reaction of superoxide anion and NO. Moreover, eNOS dysfunction can be induced by PAH-associated cytokines IL-6 and TNF-α (tumor necrosis factor-α)^4,22 and chronic hypoxia.⁴¹ Peroxynitrite in turn may promote nitration of proteins, reduce BMPRII expression,⁴ disrupt junctional integrity, and lead to the activation of TGF-β/ALK5–mediated signaling. Although other reactive oxygen or nitrogen species may play a significant role in this process, data thus far indicate oxidative stress due to lack of dynamic negative feedback regulation of eNOS by Cav-1³⁹ plays a critical role in development of PH in the absence of cav1. Future experiments to determine how dysfunctional eNOS and oxidative stress reduce EC-BMPRII expression may unveil an important mechanism for rescuing the quiescent EC phenotype.

Based on the pulmonary vascular remodeling reported in global Cav1^−/− mice²⁷ and observations that Cav-1 degradation and Cav-1 mutation can promote PH/PAH,^25,26,43 HxSu rats, EC-Cav1^−/−, global Cav1^−/− mice, and ECs derived from these mice were explored as surrogate animal and cell models of vascular injury and remodeling. We showed that hypoxia-induced PH in EC-Cav1^−/− mice or spontaneous PH in Cav1^−/− mice are associated with increased TGF-β level, reduced BMPRII expression, and Smad2/3 phosphorylation, suggesting the absence of Cav-1 promotes TGF-β/ALK5 signaling to induce pathological vascular remodeling (Figure 6). Moreover, Cav-1 expression appears to play a crucial role as a regulator of junctional integrity required for contact-inhibited quiescence and maintenance of the differentiated state of ECs. In its absence, increased oxidant production, loss of BMPRII expression, and subsequent hyperactivation of TGF-β–driven ALK5 signaling contributes to the expansion of EC subpopulations associated with pulmonary vascular remodeling and progression of PAH.

Figure 6.

Schematic showing hypothesized sequel of pulmonary vascular remodeling induced by vascular injury followed by hypoxia. Normal lung vasculature endothelial cells (ECs) express Cav-1 (caveolin-1) and BMPRII (bone morphogenetic protein receptor II). Vascular injury promotes Cav-1+ microvesicles (MVs) and small apoptotic body (sAB) shedding from ECs (damage-associated vesicular endothelial signals), which we propose contribute to the recruitment and activation of TGF-β (transforming growth factor-β)–producing macrophages. In the absence of Cav-1 expression, accumulation of peroxynitrite reduces EC-BMPRII expression. Reduced BMPRII expression and increased levels of TGF-β facilitate transdifferentiation and proliferation of dysfunctional ECs. Finally, reprogramed pulmonary ECs unable to promote physiological repair of the damaged vasculature contribute to microvascular remodeling and pulmonary arterial hypertension. SMC indicates smooth muscle cells.

Highlights.

• Pulmonary Cav-1 (caveolin-1) depletion associated with pulmonary arterial hypertension is, in part, dependent on extracellular vesicle shedding into the circulation.

• Increase in circulating Cav-1+ extracellular vesicles is associated with TGF-β (transforming growth factor-β)–induced microvascular remodeling and pulmonary arterial hypertension.

• Cav-1–depleted endothelial cells contribute to vascular remodeling via TGF-β/pSmad2/3 signaling.

Nonstandard Abbreviations and Acronyms

AB

apoptotic body

BMDM

bone marrow–derived macrophage

BMPRII

bone morphogenetic protein receptor II

Cav-1

caveolin-1

EC

endothelial cell

eNOS

endothelial NO synthase

EV

extracellular vesicle

HPAEC

human pulmonary artery endothelial cell

IL

interleukin

MLEC

mouse lung endothelial cell

PAH

pulmonary arterial hypertension

RV

right ventricle

RVSP

right ventricular systolic pressure

sAB

small apoptotic body

SIN-1

3-morpholinosydnonimine

TGF-β

transforming growth factor-β

TNF-α

tumor necrosis factor-α

VEGF

vascular endothelial growth factor

VEGFRII

vascular endothelial growth factor receptor II

WT

wild type