Nascent Endothelium Initiates T_H2 Polarization of Asthma

Abstract

Asthma airway remodeling is linked to T helper-2 (T_H2) inflammation. Angiogenesis is a consistent feature of airway remodeling, but its contribution to pathophysiology remains unclear. We hypothesized that nascent endothelial cells in newly forming vessels are sufficient to initiate T_H2-inflammation. VE-cadherin is a constitutively expressed endothelial cell adhesion molecule, which is exposed in its monomer form on endothelial tip cells prior to adherens junction formation. Antibody targeted to VE-cadherin monomers inhibits angiogenesis by blocking this adherens junction formation. Here, VE-cadherin monomer antibody reduced angiogenesis in the lungs of the allergen-induced murine asthma model. Strikingly, T_H2 responses including, IgE production, eosinophil infiltration of the airway, subepithelial fibrosis, mucus metaplasia and airway-hyperreactivity were also attenuated by VE-cadherin blockade, via mechanisms that blunted endothelial IL-25 and proangiogenic progenitor cell TSLP production. The results identify angiogenic responses in the origins of atopic inflammation.

Introduction

The origins of atopic inflammation in asthma are linked to epithelial and/or stromal cell derived cytokine products such as Thymic Stromal Lymphopoietin (TSLP) and Interleukin-25 (IL-25), and downstream T-helper 2 (T_H2) type lymphocyte and B-cell IgE responses. T_H2-activated Signal Transducer and Activation Transcription factor-6 (STAT6) is the central intracellular pathway for mucus hypersecretion of the epithelium, smooth muscle hyperplasia and hyperreactivity during initial stages of asthma(1–8). In chronic stages, however, STAT6-independent pathways may also be involved (1–8). It is increasingly recognized that innate host responses by structural cells, such as the epithelium, are critically important in the origin of T_H2 cytokine-driven inflammation (9). Despite the fact that vascular endothelium is robustly able to activate T-lymphocytes (10), very little is known regarding the contribution of lung vascular endothelium to the genesis of T_H2 inflammation. In support of endothelial mechanisms in asthma, neovascularization of the airway is a consistent histopathologic feature of asthma related to disease severity (7, 11–13). In the acute OVA murine allergic airway inflammation model of asthma, lung angiogenesis occurs prior to onset of airway reactivity, and far in advance of the development of T_H2 inflammation in the lung (12). Bone marrow-derived SCA-1⁺C-kit⁺VEGFR2⁺ proangiogenic progenitor cells home to the lungs within hours after allergen challenge and initiate an angiogenic switch. The lung neovascularization is followed temporally by influx of eosinophils (14, 15), suggesting that angiogenesis may be a first step in the onset of allergic inflammation. While angiogenesis is well recognized as a participant in the host inflammatory response, whether angiogenesis or disruption of stable vessels triggers inflammation is unknown. The endothelial junction protein vascular endothelial (VE)-cadherin is essential to the assembly of endothelial cells into mature stabilized vascular tubes (16). Blocking of VE-cadherin during development does not inhibit the initial alignment of endothelial cells, but induces rapid disassembly of nascent blood vessel (17). Genetic deletion provides conclusive evidence for an indispensible role of VE-cadherin; VE-cadherin deficient mice die in utero due to severely impaired vasculature(16). A biologic approach using a highly sensitive and specific VE-cadherin blocking antibody, E4G10, has been informative in investigating pathologic neovascularization in the adult animal (18–20). In quiescent vessels, VE-cadherin dimers are responsible for the tight endothelial homotypic adhesion, and the monomer form is not present on endothelial surfaces. Consistent with this, E4G10 only binds to nascent vascular endothelium that form during angiogenesis, when VE-cadherin is not yet engaged in the dimers of adherens junctions (19). Likewise, E4G10 is unable to bind or disrupt quiescent vascular endothelium in which VE-cadherin monomers are unexposed (18, 19). Indeed, E4G10 administration in vivo specifically inhibits nascent endothelium and blocks new vessel formation (18–20). Here, we used E4G10 in order to test the hypothesis that neovascularization is an initiating mechanistic step in the genesis of the T_H2 allergic airway inflammation in a murine model of asthma.

Materials & Methods

Animals

Immunocompetent and non-irradiated female BALB/c mice 6–8 week old were purchased from the Jackson Laboratory (Bar Harbor, ME) and used for ova or saline (vehicle) sensitization and challenge as described previously(12, 15). All experiments were approved by the local Institutional Animal Care and Use Committee (IACUC).

OVA sensitization, E4G10 treatment and methacholine challenge

Standard induction of allergic airways disease, was performed as reported previously(12, 15, 21, 22). For this purpose, BALB/c mice were first immunized by intraperitoneal injection (IP) with OVA (Sigma Chemicals) (10 µg, adsorbed in Al(OH)₃). After two weeks, mice were challenged with aerosolized OVA in a chamber that was kept saturated with nebulized OVA solution (1% w/v in sterile PBS). Animals were allowed to inhale the allergen for 40 min each day during 6 successive days, unless indicated differently. Intraperitoneal injections with E4G10 antibodies (kindly donated by Dr. Seema Iyer (ImClone Systems, a wholly owned subsidiary of Eli Lilly and Company)) or control IgG were performed according to the schedule in fig 2. To analyze if E4G10 treatment directly affected eosinophil recruitment, naïve mice were treated with antibodies 24 hours prior to intranasal eotaxin-I instillation (600ng/50ul saline). Number of eosinophils in the lung tissue was analyzed by staining for eosinophilic major basic protein twenty-four hours after eotaxin instillation. Twenty-four hours after the last OVA exposure, animals were anesthetized by IP injection with pentobarbital and placed on a rodent ventilator inside a body plethysmography chamber. Measurement of lung mechanics was done using the FlexiVent ventilator (FlexiVent, Scireq, Montreal, Canada). AHR and lung mechanics were measured on mice in response to increasing doses of inhaled methacholine as described (23, 24).

Fig 2. Inhibition of angiogenesis and progenitor cell recruitment to the lungs by blockade of VE-cadherin.

[A] Model system for investigation of effect of VE-cadherin and angiogenesis in asthma. Female BALB/c mice (6–8 weeks old) were sensitized with OVA allergen or vehicle. On indicated days (days -1, 2 and 5), mice received an intraperitoneal injection of 1.5 mg E4G10 or control antibodies. 2 weeks after sensitization, mice were exposed to a series of aerosolized allergen or vehicle challenges for 1 week after which analysis were performed. Each group contained 6 mice. [B] Low and high power images of lung tissue sections stained for von Willebrand Factor to visualize blood vessels and quantification of vessels density per lung tissue area. Scale bar = 150 micron meters. [C] Flow cytometric enumeration of progenitor cells in the lungs. After tissue digestion, single cell suspensions were stained for progenitor cell antigens. Numbers of progenitors per lung were quantified. [D] Effect of VE-cadherin on angiogenic endothelial cell proliferation, 24 hours after first allergen challenge, was measured by flow cytometry. Endothelial cells were gated as pan-hematopoietic cell marker CD45⁻ and CD31⁺, followed by quantification of cell proliferation marker Ki-67 expression in the gated endothelial cells. Number of proliferating endothelial cells and total number of endothelial cells were calculated based on total single cell counts after whole lung digestion. Mean and standard errors of 6 mice per group are shown.

Bronchoalveolar lavage fluid collection and characterization

Bronchoalveolar lavage fluid (BALF) was collected after instilling 700 µL of sterile saline and then withdrawing the fluid with gentle aspiration via the syringe. Typical BALF 400–600 µL BALF fluid was collected per mouse. Number of leukocytes was counted using a hemocytometer; after which cytologic examination was performed on cytospin preparations fixed and stained using Diff Quick (American Scientific Products, McGaw Park, IL). Differential counts were performed based on counts of 200 cells using standard morphologic criteria to classify inflammatory cells as eosinophils, lymphocytes, neutrophils or alveolar macrophages. All counts were performed by a single observer blinded to study groups.

Quantification of microvessel density

Lung microvessel density was quantified as described previously after staining tissue sections with polyclonal rabbit anti-Von Willebrand Factor antibodies (Dako Cytomation, Glostrup, Denmark) (12, 15).

Goblet cell metaplasia

Standard PAS staining of paraffin embedded lung tissue sections were used to visualize goblet cells. Goblet cell metaplasia was measured by quantification of percentage of goblet cell per bronchiole (25–27).

Trichrome staining and quantification

Trichrome staining was used to visualize and measure airway fibrosis in mouse lung tissue sections(25–27). To measure total fibrosis in the samples, digital mosaic images were collected for each section on a Leica DM 5000B upright microscope (Leica Microsystems CMS GmbH, Wetzlar, Germany) with a Prior H101 motorized stage (Prior Scientific Inc., Rockland, MA, USA) and a QImaging Retiga-SRV CCD camera (QImaging, Surrey, BC, Canada) using a HC Plan APO 10X/0.40 objective. Image-Pro Plus v.6.1 (Media Cybernetics, Inc., Bethesda, MD, USA) with the Pro-Series Turboscan plug-in (Objective Imaging, Cambridge, United Kingdom) was used to control the Prior stage and perform tile scanning of the entire section for each sample to generate a mosaic image. The area of fibrosis stained blue by trichrome (collagen deposition) and total tissue area of the mosaic images were quantified using Image-Pro Plus v.6.1. Fibrosis was quantified across all sections by first establishing the thresholding criteria for the blue staining using a test set of images in Image-Pro Plus v.6.1. The same selection criteria for fibrosis were then applied to all of the mosaic images to quantify the area of fibrosis. Area of fibrosis per total tissue area was then calculated.

Isolation of proangiogenic hematopoietic progenitor cells

Proangiogenic hematopoietic progenitors were isolated as described (28–31). In short, bone marrow mononuclear cells were seeded on fibronectin (1mg/cm² fibronectin, Sigma-Aldrich, Milwaukee, WI) coated 12-well plates at a concentration of 8 × 10⁶ cells/1ml in 20% FBS, 1% penicillin/ streptomycin in EBM-2 medium supplemented with 20ng/ml VEGF (Invitrogen, Carlsbad, CA) added. Colonies of adherent cells were trypsinized at day 7 and characterized by flow cytometric staining using anti-mouse CD45-FITC (eBioscience), SCA-1-APC, C-kit-FITC (eBioscience) and VEGFR2-PE (BD Bioscience San Jose, CA). Isotype controls were used as negative staining. In some experiments proangiogenic progenitor cells were directly detected among bone marrow mononuclear cells as described previously(12, 15).

Isolation of lung endothelial cells

Lung endothelial cells were isolated as described in details previously (15). Lungs were perfused with warm PBS via the heart to remove all blood cells, minced, and dissociated in 0.1% Collagenase A (Roche Applied Science, Indianapolis, IN), 0.04% DNase I (Sigma-Aldrich), 0.5mM CaCl₂ (Sigma-Aldrich) in 1ml of Dispase II (Roche Applied Science). Dead cells were removed by using MACS Dead Cell Removal Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Viable cells were labeled with anti-mouse CD31 (eBioscience, San Diego, CA) and sorted by magnetic activated cells sorting. Obtained endothelial cells were seeded on rat-tail collagen I coated plates in MCDB-131 complete medium (VEC Technologies, Rensselaer, NY). Purity of CD31⁺ fraction was assessed by flow cytometry analysis and 1,1’ –dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine (Dil)-acetylated LDL (Dil-AcLDL) (Molecular probes, OR) uptake. Where indicated single cell suspensions after lung digestions were used for flow cytometric analysis of endothelial cells proliferation. Cells were stained for cell surface expression of anti-mouse CD45-APC and CD31-PE-Cy7 (eBiosciences), followed by intracellular staining for proliferation marker Ki-67-FITC (BD Biosciences). Dead cells were excluded by staining with UV LIVE/DEAD dye (invitrogen). All samples were analyzed on LSRII flow cytometer and data were analyzed using FlowJo 9.0.2 software (TreeStar).

Secretion of T_H2 cytokines and plasma IgE

Lung endothelial cells (0.5 × 10⁶) or bone marrow-derived proangiogenic progenitors (100 × 10³) were incubated in MCDB-131 medium. At day 5 supernatants were collected and stored at −80°C until analysis. Levels of IL-25, TSLP and IL-33 were measured using BioLegend (IL-25) or R&D quantikine ELISA kits (TSLP and IL-33). Values from different experiments were normalized against PBS/PBS group. Eotaxin-2 and IL-13 levels in BALF were also measured by R&D quantikine ELISA kits. Values were normalized against total protein concentration. OVA-specific plasma IgE levels were analyzed using OVA-IgE ELISA kit (MD Bioproducts, MN).

STAT6 Electrophoretic mobility shift assay (EMSA)

Whole cell extracts (WCE) from mice lungs were prepared as previously described (32). The duplex oligonucleotide high-affinity sis-inducible element (hSIE)(5’-AGCTTCATTTCCCGTAAATCCCTAAAGCTA-3’) and duplex oligonucleotide (5’-GATCGCTCTTCTTCCCAGGAACTCAATG-3’) specific for STAT6 binding (33) with the consensus sequences underlined sequences used in EMSA were end-labeled with [γ-³²P] ATP by T4 polynucleotide kinase (Invitrogen, Carlsbad, CA). To specifically identify DNA-binding-factor in binding complexes, rabbit polyclonal anti-STAT-6 Ab (Santa Cruz Biotechnology, Santa Cruz, CA) was added to the binding reaction mix.

Statistical analysis

Data were analyzed by using JMP 5.1 software program. ANOVA or Student’s T-test were used for comparisons of parametric data, and Wilcoxon or Median test was used for comparison of nonparametric data, as appropriate. p-values <0.05 were considered as significant. Mean ± SEM value for each group is shown.

Results

Bone marrow hematopoietic proangiogenic progenitor cell and lung endothelial cell proliferation are initial events after allergen exposure

Twenty-four hours after the first airway allergen exposure, numbers of proangiogenic progenitors in bone marrow and lungs, and numbers of endothelial cells in the lung were analyzed (Fig 1). Hematopoietic proangiogenic progenitors isolated from the bone marrow were CD45, SCA-1, c-Kit, and VEGFR2 positive, consistent with previous reports (12, 15) (Fig 1A). Endothelial cells isolated from the lungs were ascertained to be endothelial phenotype by positivity for CD31, a prototypical endothelial cell surface marker, and by the uptake of acetylated low density lipoprotein labeled with dioctadecyl-tetramethyl-indocarbocyanineperchlorate (Dil-AcLDL) (Fig 1B). The number of proangiogenic progenitor cells in the bone marrow was increased after allergen OVA, but not after sham phosphate buffered saline (PBS), sensitization [proangiogenic progenitor cells × 10⁶/bone marrow aspirate: PBS/PBS 1.3 ± 0.14; OVA/PBS 2.3 ± 0.15; OVA/OVA 1.5 ± 0.1, ANOVA p<0.0001] [Fig 1C]. The lesser numbers of proangiogenic cells in the bone marrow of OVA allergen challenged mice (OVA/OVA) as compared to the PBS challenged mice (OVA/PBS) was consistent with mobilization from the bone marrow and recruitment into the lungs, as suggested by greater numbers of proangiogenic progenitors cells in the lungs of only the OVA/OVA mice [proangiogenic progenitor cells × 10⁶/lung: PBS/PBS 0.1 ± 0.2; OVA/PBS 0.1 ± 0.09; OVA/OVA 0.37 ± 0.6, ANOVA p<0.0001]. In parallel to the influx of proangiogenic progenitor cells in the lung, the number of CD45⁻ CD31⁺ lung endothelial cells in the lung doubled (Fig 1C) [endothelial cells × 10⁶/lung: PBS/PBS 1.6 ± 0.02; OVA/PBS 1.6 ± 0.1; OVA/OVA 3.9 ± 1.6, ANOVA p = 0.04]. Overall, the data confirm that bone marrow-derived proangiogenic progenitor cell mobilization and recruitment to the lung and lung endothelial cell proliferation are early events in the development of allergic airway disease.

Fig 1. Bone marrow proangiogenic progenitor cell and lung endothelial cell proliferation within 24 hours of murine asthma model.

OVA allergen or sham sensitized mice were sensitized and then exposed to aerosolized OVA or sham inhalation. Twenty four hours after 1^st allergen inhalation, samples were collected. [A] Bone marrow mononuclear cells were separated to isolate proangiogenic hematopoietic progenitor cells. Antigen expression was analyzed by flow cytometry. Filled histograms show level of cell surface expression, open histograms are control stainings. [B] Lung endothelial cells were isolated after whole organ digestion and dead cell removal, followed by CD45-depletion and sorting of CD31⁺ cells. CD31 purity of obtained endothelial cells, phase contrast image and Dil-AcLDL uptake by endothelial cells are shown. Scale bar = 50 micron meters. [C] Quantification of number of isolated proangiogenic progenitor cells and lung endothelial cells. Mean and standard errors of 3 mice per group are shown.

Blockade of VE-cadherin inhibits angiogenesis in asthma lung

To investigate whether the neovascularization is actively involved in airway inflammation and remodeling, nascent endothelial cell formation of junctions was blocked by E4G10 rat anti-VE-cadherin. E4G10 (1.5 mg/mouse) or control antibody (cIgG) was administered intraperitoneally three times per week as previously described (18, 19) (Fig 2A). At indicated time points, samples were collected for analysis. In a pilot experiment OVA/OVA group without control antibody injection was included and no differences were observed in airway inflammation and airway remodeling between cIgG treated mice and no control antibody treated animals at day 8 of analysis. Animals sensitized and challenged with OVA (OVA/OVA) had greater micro-vessel density in the lungs compared to PBS/OVA mice, demonstrating increased angiogenesis during allergic airway inflammation (Fig 2). Treatment of OVA/OVA mice with E4G10 inhibited angiogenesis [lung micro-vessel density: PBS/OVA 25.9 ± 1.4; OVA/OVA with cIgG 61.1 ± 5.8; OVA/OVA with E4G10 37.1 ± 3.6, ANOVA p<0.0001] (Fig 2B). SCA-1⁺C-kit⁺VEGFR2⁺) proangiogenic progenitor cell recruitment into the lungs was also significantly reduced in E4G10 treated mice, indicating that the blockade of VE-cadherin also blocks bone marrow proangiogenic progenitor recruitment (Fig 2C) [SCA-1⁺C-kit⁺VEGFR2⁺ × 10³/lung: PBS/OVA 5.5 ± 0.7; OVA/OVA with cIgG 20.8 ± 6.4; OVA/OVA with E4G10 13.0 ± 2.0, ANOVA p<0.0001]. In vivo blockade of VE-cadherin also decreased lung endothelial cell counts and endothelial cell proliferation, as measured by flow cytometric analyses for Ki-67 (Fig 2D) [% Ki-67⁺ lung endothelial cells: OVA/OVA with cIgG 2.3 ± 0.2; OVA/OVA with E4G10 1.5 ± 0.2, p = 0.049; Number of endothelial cells/lung (×10⁶): OVA/OVA with cIgG 3.4 ± 0.2; OVA/OVA with E4G10 1.7 ± 0.2, p = 0.001]. Thus, blockade of VE-cadherin reduced angiogenesis and inhibited mobilization or recruitment of proangiogenic progenitors into the lung from the circulation.

Decreased angiogenesis reduces airway inflammation, T_H2 cytokines and OVA-specific IgE

H&E of lung tissue sections revealed substantial reduction of airway inflammation in OVA/OVA mice treated with E4G10 (Fig 3A). The numbers of eosinophils in lung tissue sections (Fig 3B–C) and in the bronchoalveolar lavage fluid (BAL) (Fig 3D) were reduced by VE-cadherin blockade [eosinophils 10³/ ml BAL: PBS/OVA 0.0 ± 0.0; OVA/OVA with cIgG 57 ± 7; OVA/OVA with E4G10 20 ± 6, ANOVA p<0.0001], [10³/mm² lung tissue: PBS/OVA 0.11 ± 0.04; OVA/OVA with cIgG 1.5 ± 0.48; OVA/OVA with E4G10 0.61 ± 0.15, ANOVA p<0.003]. Intranasal eotaxin-1 instillation induced eosinophilia in naive mice [number of eosinophils/0.1mm² lung tissue: naïve mice 21 ± 6; eotaxin-1 instilled naïve mice 44 ± 4, p=0.03 (n=4mice/group)]. Eotaxin-1 induced eosinophil influx was similar among E4G10 or control IgG treated mice [number of eosinophils/0.1mm² lung tissue: cIgG treated and eotaxin-1 instilled naïve mice 40 ± 3; E4G10 treated and eotaxin-1 instilled naïve mice 44 ± 4, p=0.41 (n=4mice/group)]. On the other hand, E4G10 treatment in the OVA model abrogated levels of T_H2 cytokines eotaxin (Fig 3E) and IL-13 (Fig 3F) in BAL [eotaxin-2 in BAL (pg/mg protein): OVA/OVA cIgG 2434 ± 108; OVA/OVA E4G10 566 ± 286, p = 0.049. IL-13 in BAL: OVA/OVA cIgG 58 ± 18; OVA/OVA E4G10 14 ± 4 pg/mg protein, p = 0.02]. Statistical significance was not reached for levels of eotaxin-1 [eotaxin-1 in BAL (pg/mg protein): OVA/OVA cIgG 353 ± 197; OVA/OVA E4G10 89 ± 21, p = 0.19]. Immunohistochemistry staining showed expression of VEGF by airway epithelial cells, but no obvious differences were observed between cIgG and E4G10 groups (not shown). TSLP plays a critical role in IgE production via activation of dendritic cells, which primes T_H2 T-cells to induce isotype switching in B-cells (9). OVA specific IgE levels were strikingly reduced in E4G10 treated group (Fig 3G): [OVA specific IgE (ng/ml plasma): PBS/OVA not detectable levels; OVA/OVA cIgG 250.4 ± 48.8; OVA/OVA E4G10 37.9 ± 3.4, ANOVA p = 0.0002]. Overall, blockade of new vessel formation by VE-cadherin antibody reduced T_H2 cytokines and IgE production, indicating that blockade of angiogenesis inhibits the polarization of T-cells and differentiation of B-cells to IgE production, all of which resulted in the lesser airway inflammation.

Fig 3. VE-cadherin blockade reduces inflammation in the mouse asthma model.

[A] H&E staining of lung tissue sections at low power magnification (5×) (upper panel) and high power magnification (20×) (lower panel) show inhibition of inflammation by VE-cadherin blockade. a = airway and black arrows point to inflammatory infiltrates. Scale bar = 150 micron meters. [B–C] Eosinophil specific major basic protein staining shows less number of eosinophils per lung tissue area. In red eosinophil specific major basic protein staining and nuclei in blue (DAPI staining). [D] Eosinophils in BAL were less in mice receiving anti-VE-cadherin than control antibodies and similar to sham control mice. [E–F] T_H2 cytokines eotaxin and IL-13 concentrations were lower is BAL fluid of mice treated with anti-VE-cadherin. [G] Effect of VE-cadherin blockade on OVA-specific IgE levels. N.D. = Not Detectable. Mean and standard errors of 6 mice per group are shown.

Blockade of VE-cadherin inhibits airway remodeling

Airway fibrosis and goblet cell metaplasia are important remodeling outcomes of T_H2-reactions in asthma. Airway fibrosis analyzed by trichrome staining for collagen deposition was lower in animals treated with E4G10 compared to cIgG [trichome area mm²/100mm² lung: PBS/OVA 1.7 ± 0.2; OVA/OVA with cIgG; 2.7 ± 0.3 ; OVA/OVA with E4G10 1.3 ± 0.5, ANOVA p = 0.037] (Fig 4A). Mucus producing goblet cells in the airways detected by Periodic acid-Schiff (PAS) histochemistry were ~50% lower in mice treated with E4G10 as compared to control IgG (Fig 4B) [% PAS⁺ cell/bronchiole: PBS/OVA 4 ± 2; OVA/OVA with cIgG 84 ± 3; OVA/OVA with E4G10 36 ± 3, ANOVA p<0.0001]. Altogether the results show that asthmatic airway remodeling, including airway fibrosis and goblet cell hypertrophy, requires nascent endothelium with available VE-cadherin.

Fig 4. Airway fibrosis and goblet cell numbers reduced by VE-cadherin blockade.

[A] Trichrome masson staining of lung tissue sections shows collagen deposition (blue) in low and high power field pictures. Collagen deposition was less in mice treated with anti-VE-cadherin antibodies. Scale bars = 150 micron meters. For quantification µm² collagen tissue/100 µm² lung tissue is shown. [B] Mucus producing goblet cells stained with Periodic acid-Schiff stain (PAS) to detect carbohydrate macromolecules, such as mucus, were stained purple-magenta (black arrows). Asthmatic mice showed increased numbers of goblet cells that was inhibited by VE-cadherin blockade. Representative bronchiole from each group is shown. Scale bar = 100 micron meters. Mean and standard errors of 6 mice per group are shown.

Angiogenesis affects airway mechanics

Airway reactivity to methacholine was measured to analyze the effect of angiogenesis inhibition on airway hyperreactivity. As anticipated, airway resistance was greater in the cIgG-treated OVA/OVA group as compared to OVA/PBS group (Fig 5A). E4G10-treated OVA/OVA mice had methacholine reactivity significantly lower than cIgG-treated OVA/OVA mice. Airway elastance was significantly higher in the cIgG OVA/OVA mice compared to E4G10 OVA/OVA mice (Fig 5B). The data are consistent with the decreased collagen deposition and mucus secretion in E4G10 OVA/OVA mice and support the notion that angiogenesis participates in the cellular mechanisms that lead to airway inflammation and hyperresponsiveness in asthma.

Fig 5. VE-cadherin blockade abrogates airway reactivity.

[A] Dose response curves of airway hyperreactivity shows significantly lower airway resistance in mice treated with anti-VE-cadherin antibodies. [B] Dose response curves of lung elastance, which is inversely related to lung elasticity, was also reduced by VE-cadherin blockade. * Significant different between groups (all p <0.05). Mean and standard errors of 6 mice per group are shown.

Mechanisms by which VE-cadherin blockade inhibits asthma phenotype development: IL-25 and TSLP expression and STAT6 activation

To investigate mechanisms underlying the inhibitory effect of VE-cadherin blockade on the genesis of atopic inflammation, IL-25 and TSLP production by endothelial cells and proangiogenic progenitor cells isolated from mice after challenge and/or antibody exposure were analyzed. Lung endothelial cells from OVA/OVA mice secreted substantial levels of IL-25, but proangiogenic progenitor cells did not (Fig 6A–B). IL-25 expression by lung endothelial cells was reduced by VE-cadherin blockade in the OVA/OVA group. TSLP production by isolated endothelial cells was low (Fig 6C). Conversely, there was significant increase of TSLP production by proangiogenic progenitor cells isolated from OVA/OVA mice, which was prevented by VE-cadherin blockade (Fig 6D). Expression of IL-33, another main T_H2 cytokine was not observed. Recruitment of CD3⁺ T-cells was not affected by E4G10 [Number of CD3⁺ cells/0.1mm² lung tissue: OVA/OVA with cIgG 125 ± 14; OVA/OVA with E4G10 116 ± 11, p=0.51], which taken together with the low levels of T_H2-cytokines, indicated that the T-cells present in the lung were not polarized to T_H2-type.

Fig 6. Inhibition of T_H2 cytokine production and signal transduction by VE-cadherin blockade.

[A–D] Animals were treated with VE-cadherin antibodies, sensitized and challenged as shown in fig 2a. In these experiments analysis were performed 24 hours after the first allergen exposure. Lung endothelial cells and bone marrow progenitor cells were incubated ex-vivo and levels of IL-25 and TSLP in overlaying media were analyzed by ELISA, showing reduced secretion by in-vivo VE-cadherin inhibition. [E–G] In these experiment analysis were performed 30 minutes to 1 hours after the first allergen exposure. Animals were treated with VE-cadherin antibodies, sensitized and challenged as depicted in fig 2a, whole lung tissue was used for analysis. [E] Electrophoretic mobility shift assay (EMSA) for STAT6 DNA binding were performed. E4G10 treatment inhibits OVA allergen induced STAT6 activation in the lung (lane 8), while control antibody did not (lane 7). [F] Densitometric quantification of STAT6 activation. [G] Phospho-STAT6 (P-STAT6) in lung tissue was analyzed by immunohistochemistry. STAT-6 activation in airway epithelial cells and endothelial cells, indicated by arrows in OVA/OVA groups, were attenuated by E4G10 treatment. Scale bar = 200 micron meters. Mean and standard errors of 3 mice per group are shown.

Electrophoretic mobility shift assay (EMSA) for STAT6 DNA binding activity revealed STAT6 activation in OVA/OVA lungs, which was reduced by the VE-cadherin blockade (Fig 6E–F). Immunostaining for phosphoSTAT6 confirmed greater STAT6 activation in airway epithelial cells and endothelial cells of OVA/OVA mice (Fig 6G). The VE-cadherin blockade resulted in less STAT6 activation in endothelium and epithelium of OVA/OVA mice. Interestingly, cIgG treatment appeared to potentiate STAT6 activation. We cannot exclude that cIgG potentially provided an adjuvant-like effect, but the fact that E4G10, which is of the same isotype as the control IgG, inhibited STAT6 activation further supports the concept that VE-cadherin blockade regulates the origins of atopic asthma.

Discussion

More than a century has passed since the initial observation by Ellis that blood vessel density is increased in the airways of asthma patients (11). Since then the dogma had been that angiogenesis in asthma is a side-effect of inflammation, supported by the notion that inflammation induces angiogenesis(34) and by lack of evidence that pathological angiogenesis itself can induce inflammation. In parallel to tumors, an angiogenic switch occurs early in the genesis of asthma(12), supporting the possibility that angiogenesis may regulate airway inflammation and remodeling. Here we provide evidence by blocking nascent endothelial cells that angiogenesis is a formative first step to development of T_H2 inflammation via IL-25 and TSLP production by the resident endothelium and recruited bone marrow-derived proangiogenic progenitors. To the best of our knowledge this is the first report that nascent endothelium is an essential early step in the origins of the T_H2 response.

The process of angiogenesis requires bone marrow-derived proangiogenic hematopoietic progenitors and proliferation of endothelial cells. Proangiogenic progenitor cell mobilization and/or recruitment and angiogenesis during asthmatic airway inflammation are reduced moderately by blockade of adherens junction formation of nascent endothelial cells. Yet, inhibition of angiogenesis resulted in marked attenuation of airway inflammation and remodeling. Our findings are in line with reports that overexpression of vascular endothelial cell growth factor (VEGF), a key angiogenic factor (35, 36), in airway epithelial cells, results in angiogenesis and airway inflammation in mouse models via development of a spontaneous T_H2 response(37). Interestingly, VE-cadherin association with VEGFR2 is critical for VEGF signaling during angiogenesis(16), indicating a crucial cross-talk between these two pathways in new vessel formation and inflammation. The primary role of VE-cadherin in vascular stability (35, 38, 39) makes it challenging to use molecular approaches targeting the whole protein, which results in lethal vascular leakage(18–20, 40). However, fine distinctions in the different epitopes allowed development of a new class of inhibitors, such as humanized E4G10 antibodies, which only affect angiogenic endothelial cells, and not quiescent endothelial cells which form the majority of the endothelium (18–20).

While proangiogenic progenitors are known to be essential to the process of new vessel formation in the murine OVA-model of lung inflammation, here we show for the first time that TSLP is provided delivered to the vascular bed in trans by the circulating bone marrow-derived progenitors, complementing the IL-25 produced by the resident endothelium. Prior studies indicate that TSLP is not expressed by most hematopoietic cells, with the exception of myeloid cells including neutrophils, macrophages and mast cells (41–43). In this context, proangiogenic and mast cell progenitors share the same identifying surface markers(12, 15), suggesting that the hematopoietic myeloid progenitors that promote angiogenesis required in the very early stages of atopic inflammation perhaps, in part, by delivery of TSLP to the tissue bed. In support of this concept, TSLP is central to dendritic cell driven T and B cell expansion and differentiation of B-cells that results in allergen-specific IgE production (2). Here, inhibition of new vessel formation partially reduced recruitment of proangiogenic progenitors, but more importantly the recruited proangiogenic progenitors did not produce TSLP and this was associated with the expected loss of IgE production. Altogether, the findings reveal a vital link between circulating proangiogenic progenitors required for new vessel formation and initiation of T_H2-inflammation and IgE production in atopic lung inflammation.

Epithelial-derived IL-25, TSLP and IL-33 are currently defined as the three most proximal activators of the T_H2 cascade (9). Epithelium lining body cavities is the first cellular contact with the external environment. The epithelial IL-33 and IL-25 induced T_H2 response is the host response to pathogens, including allergens, fungi and helminth infections (9). Here, we show that endothelium lining vessels is also capable of activating T_H2 responses in synergy with bone-marrow derived proangiogenic progenitor cells via IL-25 and TSLP. Although they do not differentiate into endothelial cells, proangiogenic progenitors are potent paracrine-acting angiogenic cells (44), and now show to be important paracrine-acting cells for development of atopy. A recent paper reported that in organ transplantation, human endothelial cells promoted proliferation of proinflammatory T_H17 and regulatory CD4⁺ T lymphocytes (45), more evidence that the endothelium directly modulates the immune system. It’s also possible that unengaged VE-cadherin monomers on nascent endothelial cells is essential in auto-activation of vascular immune response(46). Here, results of eotaxin administration to mice show that E4G10 blockade does not act by affecting recruitment of eosinophils or T-cells. Similarly, expression of VE-cadherin by murine proangiogenic progenitors (20) was not observed in our hands (data not shown), in line with recent findings by others (44), reducing the possibility of off-target effects of the antibody.

Traditionally, asthma and atopy have been viewed as mucosal diseases. The data here suggest that the origin of mucosal atopic inflammation requires new vessel formation, a process in which endothelial cells and bone marrow-derived progenitors provide the combined signals to initiate T_H2 inflammation. In this context, subsets of bone marrow-derived progenitor cells, such as IL-25 responsive “type-2 multipotent progenitors” (47"nuocytes” (48) and “natural helper cells” (49) have been reported to play a critical role in the initiation of gut T_H2-response and “type 2 myeloid cells” in asthma(50). Bone marrow-derived proangiogenic progenitors and these other innate type-2 cells most probably originate from a common ancestor as indicated by their co-expression of SCA-1 and C-kit(47, 49). Overall, our findings expand the emerging concept of vascular immunology by adding T_H2-response to the list of immune responses initiated by the endothelium.