Vesicle-associated membrane protein 7-mediated eosinophil degranulation promotes allergic airway inflammation in mice

Eosinophil degranulation is a determining factor in allergy-mediated airway pathology. Receptor-mediated degranulation in eosinophils requires vesicle-associated membrane protein 7 (VAMP-7), a principal component of the SNARE fusion machinery. The specific contribution of eosinophil degranulation to allergen-induced airway responses remains poorly understood. We generated mice with VAMP-7 gene deficiency exclusively in eosinophils (eoCRE/V7) from a cross using eosinophil-specific Cre recombinase-expressing mice crossed with VAMP-7f/f mice. Eosinophils from eoCRE/V7 mice showed deficient degranulation responses in vitro, and responses continued to be decreased following ex vivo intratracheal adoptive transfer of eoCRE/V7 eosinophils into IL-5/hE2/EPX−/− mice. Consistent with diminished degranulation responses, reduced airway hyperresponsiveness was observed in ovalbumin-sensitized and challenged eoCRE/V7 mice following methacholine inhalation. Therefore, VAMP-7 mediates eosinophil degranulation both in vitro and ex vivo, and this event augments airway hyperresponsiveness.

E osinophil activation is a major factor in exacerbation of asthma. Approximately 50% of all asthmatic subjects exhibit evidence of eosinophils in their airways 1 . Corticosteroids are the main treatment used in anti-inflammatory treatment of asthma and reduction in eosinophil levels, but these fail to ameliorate symptoms in all patients even at enhanced doses. Thus, a more tailored approach for intervention in asthma exacerbation is required to target underlying inflammatory events. This is evident from recent clinical trials showing that eosinophil depletion by monoclonal anti-IL-5 or IL-5 receptor antibody (e.g., mepolizumab, reslizumab, and benralizumab) leads to corticosteroid dose tapering and reduced exacerbations and diminished rates of hospitalization [2][3][4] .
In asthma, airway eosinophilia is a feature of atopic and severe non-atopic late-onset asthma. Eosinophils are considered predominantly pro-inflammatory, and contribute to tissue damage by the release of cytotoxic cationic granule proteins through degranulation 5 . Eosinophil degranulation is defined as regulated receptor-mediated release of granule products that serves to directly deposit secreted mediators toward a target 6 . Receptormediated degranulation in eosinophils in response to a range of inflammatory stimuli is dependent on the final fusion step occurring between secretory granules and the plasma membrane, known as regulated exocytosis 7 . This requires distal binding of membrane-bound fusion-competent soluble N-ethylmaleimide sensitive factor attachment protein (SNAP) receptors (SNAREs) on granule and cell membranes, which facilitate fusion of granules and the subsequent release of contents to cell exterior. SNAREs comprise core components of the fusion machinery necessary for exocytosis in all secretion-competent cells 8 , and are expressed in eosinophils [9][10][11] . Core components of SNARE complexes expressed by eosinophils include vesicle-associated membrane protein (VAMP) 2, 7, 8, syntaxin-4, and SNAP-23 11,12 . The R-SNARE VAMP-7 co-localizes with eosinophil peroxidase (EPX), an oxidoreductase that is uniquely expressed in eosinophils and stored within crystalloid granules 11,13 . VAMP-7 is required for agonist-activated release of EPX via degranulation from streptolysin-O-permeabilized human eosinophils 11 . However, no reports have demonstrated a causal link between physiologically induced eosinophil degranulation and the development of airway hyperresponsiveness (AHR) or airway remodeling. Here we sought to determine the role of VAMP-7mediated eosinophil degranulation in allergic airway inflammation as an important exacerbatory event in asthma.
To confirm co-localization with granules, eosinophils were fractionated and separated on a linear density gradient (0-45%). Eosinophils showed enriched EPX activity in crystalloid granulecontaining fractions that overlapped with peak VAMP-7 immunoreactivity (fractions 5-10; Fig. 1b and Supplementary Figure 1a) and less so with plasma membrane SNAP-23 (fractions [10][11][12][13][14], a cognate Q-SNARE binding partner for VAMP-7 9 . Eosinophil-specific VAMP-7 gene deletion in eoCRE/V7 mice. To investigate a possible role for VAMP-7 in eosinophil degranulation and allergic inflammation, we crossed mice expressing Cre recombinase from the eosinophil-specific EPX promotor (eoCRE mice) strain to mice with LoxP-flanked VAMP-7 alleles (VAMP-7 f/f mice) to generate VAMP-7 f/f eoCRE mice (called 'eoCRE/V7' here). High efficiency Cre-mediated VAMP-7 gene deletion in eosinophils was achieved using an established eoCRE-loxP strain associated with the eosinophil-specific EPX promoter 15 . The level of Cre-recombined allele was assessed using a four-primer strategy (see Methods section; Fig. 2a). Decreased reporter signals were observed in VAMP-7 null samples, in which the ΔΔCt ratio indicated fewer copies of the target region. Thus, the value of P-a-P-b approaches zero with Cre-mediated deletion of ex 3 and 4 of the VAMP-7 gene (see Methods section for details on primers used in this figure). DNA isolated from white blood cells (WBCs) of a ubiquitous VAMP-7 knockout mice (Zp3/V7) served as control; these cells had ΔΔCt values of nearly zero (Fig. 2b). Eosinophils derived from bone marrow progenitors from VAMP-7 f/f mice and heterozygous eoCRE +/− had higher baseline EPX expression levels than homozygous eoCRE −/− mice 15 , so heterozygous eoCRE +/− mice were used as controls in all experiments. Both VAMP-7 f/f and eoCRE +/− eosinophils had comparable excised:un-excised VAMP-7 DNA ΔΔCt ratios, suggestive of intact VAMP-7 genes. In contrast, eoCRE/V7 bone marrow eosinophils exhibited reduced ΔΔCt ratios, suggestive of a VAMP-7 null allele (Fig. 2b). Cytospin analysis revealed that the majority (>90%) of bone marrow cells were eosinophils, while a small percentage (~2-3%) were neutrophils, which accounts for slightly elevated ΔΔCt ratios observed in eoCRE/V7 bone marrow eosinophils.
To augment our molecular analyses, peripheral blood eosinophils from eoCRE/V7/I5 mice were tested for VAMP-7 protein expression by the western blot. VAMP-7 expression and molecular weight were similar in eoCRE/I5 eosinophils derived from bone marrow and peripheral blood, as well as human peripheral blood eosinophils ( Fig. 2e and Supplementary  Figure 1c), while VAMP-7 protein expression was ablated in peripheral blood eosinophils isolated from eoCRE/V7/I5 mice (96% purity) ( Fig. 2e and Supplementary Figure 1c).

Deficient ex vivo degranulation responses in
OVA treatment to reduced VAMP-7-mediated release of EPX.
To determine the role of VAMP-7 in eosinophil-induced lung pathologies, an acute OVA sensitization and challenge protocol was employed to induce pulmonary inflammation in eoCRE +/− and eoCRE/V7 mice. Specifically, eoCRE +/− controls and eoCRE/V7 mice were sensitized on days 0 and 14 with intraperitoneal (i.p.)-injected OVA plus alum or saline, then challenged intra-nasally with aerosolized 1% OVA or saline on days 24, 25, and 26 ( Fig. 5a). No significant difference in elevated BAL cellularity was observed in OVA-treated eoCRE +/− and eoCRE/V7 mice on day 28 (p > 0.05, one-way ANOVA; Fig. 5b). Similarly, eosinophil numbers were increased to equivalent levels in both strains upon OVA treatment (Fig. 5c). Levels of EPX and MBP in BAL samples from these mice were also assessed. While eoCRE +/− control and eoCRE/V7 mice injected with saline did not show EPX release into BAL samples, EPX levels in OVA-treated eoCRE/V7 mice were reduced bỹ 20% compared to OVA-treated eoCRE +/− mice (Fig. 5d), and MBP release was reduced by~23% in eoCRE/V7 mice (Fig. 5e). These findings show evidence that VAMP-7 gene deletion led to a reduction of EPX and MBP release in this acute model of allergic airway inflammation.
VAMP-7-mediated eosinophil degranulation contributes to AHR. Eosinophil recruitment and activation has been associated with AHR and airway tissue remodeling in numerous animal models of airway diseases 20 . Consequently, the influence of eosinophil-specific VAMP-7 gene deficiency in AHR was evaluated following aerosolized methacholine challenge. eoCRE +/− mice responded to OVA sensitization and challenge as indicated by increased airway resistance (Fig. 6a). However, airway resistance in OVA-treated eoCRE/V7 mice treated with OVA upon methacholine challenge was significantly lower compared with similarly treated eoCRE +/− mice (p < 0.05, ANOVA; Fig. 6a). These findings show that VAMP-7-mediated eosinophil degranulation contributes to AHR in this model. Lung sections from eoCRE +/− mice obtained 48 h after the final OVA challenge showed similar increases in inflammatory infiltrates compared to OVA-treated eoCRE/V7 mice (Fig. 6b). Increases in periodic acid-Schiff (PAS)-positive, mucin-containing epithelial cells could also be observed in lung sections from OVA-treated eoCRE +/− and eoCRE/V7 mice (Fig. 6c). No difference in eosinophil recruitment between eoCRE +/− and eoCRE/V7 mice treated with OVA was observed following anti-MBP staining (Fig. 6d), in agreement with our findings for BAL eosinophil numbers following OVA challenge (Fig. 5c). Higher magnification of anti-MBP staining showed a qualitative decrease in extracellular MBP in eoCRE/V7 mice treated with OVA compared to control, which correlated with our findings for MBP release in BAL samples (Fig. 6e).

Discussion
This study demonstrates a contributory role for the R-SNARE, VAMP-7, in mediating physiologically induced release of granule-derived proteins from eosinophils. Previously, VAMP-7dependent degranulation in eosinophils was only observed in permeabilized human eosinophils activated by intracellular applications of artificial agonists in vitro 11 . Because eosinophils are end-differentiated cells that are difficult to transfect, an understanding of the physiological role of VAMP-7 in receptormediated secretion from intact eosinophils could only be achieved using a targeted cell-specific gene knockout model such as the Cre recombinase system.
Degranulation is defined as the release of granule products by regulated exocytosis or necrosis (cytolysis) 5,6 . A heterogeneous group of granules are located in the cytoplasm of eosinophils, suggesting that eosinophils can undergo regulated exocytosis or piecemeal degranulation 24 . Eosinophil crystalloid granules are involved in classical regulated exocytosis 7 . In addition, a pool of small, rapidly mobilizable secretory vesicles forming part of a tubulovesicular complex, that shuttles proteins from the crystalloid granules to the cell membrane, is associated with piecemeal degranulation 7 . VAMP-7 is a critical component of the vesicular SNARE core complex, and its function is related to regulated exocytosis as well as piecemeal degranulation 11 . In human eosinophils, VAMP-7 was found expressed in secretory vesicles, as well as crystalloid granules, and inhibition of VAMP-7 specific antibody in permeabilized cells led to dose-dependent inhibition of EPX and eosinophil-derived neurotoxin (EDN) release, suggesting that VAMP-7 may be involved in all forms of regulated exocytosis in eosinophils, similar to neutrophils 11 . Exocytosis is a synergistic process that involves numerous intracellular proteins. VAMP-7 deletion in eosinophils diminished but did not completely abolish degranulation at all three levels of biological analysis (in vitro, ex vivo, and in vivo). Incomplete inhibition of granule protein secretion from stimulated eoCRE/V7-derived eosinophils suggests that a complex mechanism of selective sorting, transport, and/or fusion exists with a certain level of redundancy. Other R-SNARE proteins may contribute to granule fusion with the plasma membrane, including VAMP-2 and −8, which bind to cognate target Q-SNAREs syntaxin-4 and SNAP-23, expressed in human eosinophils 9,11,14 . Another possible reason for incomplete inhibition of degranulation by VAMP-7 gene deletion in eosinophils may be related to a separate mechanism of degranulation known as cytolysis in which eosinophils release intact granules following necrosis, a major mechanism occurring in allergic responses 7,25 . However, we found that eosinophil viability was high (≥90%) and eosinophils were intact following stimulation with PAF, ionomycin, and IL-33, suggesting that cytolysis was not a major mechanism of granule protein release in vitro.
Despite the finding that both VAMP-7 and -8 were localized to small secretory vesicles and crystalloid granules in human eosinophils 11 , only VAMP-7 was associated with both EPX and EDN release 11 . VAMP-2, on the other hand, was absent on crystalloid granules, but was found to co-localize with secretory vesicles containing the chemokine CCL5, and traffic to the plasma membrane in response to IFN-γ stimulation 14 . Inhibition of VAMP-2 using a specific antibody in permeabilized cells led to a partial reduction in EPX release but not EDN, suggesting that VAMP-2 is predominantly associated with piecemeal degranulation 14 . The reduction in EPX and MBP release, but not cytokines, from crystalloid granules in eoCRE/V7-derived eosinophils implies that a selective VAMP-7-dependent mechanism exists for sorting and mobilization of various granule components during receptor-mediated exocytosis.
The underlying mechanisms that link eosinophil degranulation with allergic AHR are not well understood. Several cationic products of eosinophil granules, including MBP and EPX, have been demonstrated to have cell damaging or activation effects on airway epithelial cells, airway smooth muscle cells, mast cells, and many other airway cells 5 . However, individual gene knockouts of MBP and EPX failed to alter AHR in this allergic model 26,27 . Our findings suggest that complex mechanisms may be involved in multiple granule protein-dependent airway effects that involve additive or synergistic effects of many products of eosinophil degranulation.
Physiological evidence collected in this study indicated that VAMP-7-mediated eosinophil degranulation influences the development of airway immune responses to allergens. AHR was attenuated in eoCRE/V7 mice compared with controls. However, a recent study has indicated that eosinophil degranulation is not an essential mechanism for the induction of lung pathologies in chronic inflammation 27 , which is in apparent contradiction to our findings. Several reasons could explain the discrepancy in these findings. One is that the cited study examined individual knockouts of MBP or EPX genes, which do not prevent eosinophil degranulation. Eosinophil crystalloid granules contain up to 17 different cytokines and chemokines 28 , as well as many other proteins and enzymes that can alter immune function. An advantage of our present study is that we were able to attenuate both MBP and EPX release without encountering off-target effects resulting from concurrent knockout of MBP and EPX; mainly, the complete ablation of eosinophils from animals 23 . Further studies are warranted to determine the specific contribution of individual eosinophil degranulation products in asthmatic inflammation.
Allergic asthma is intimately associated with the chronicity and increased severity of airway remodeling orchestrated by many cells including eosinophils 1 . Eosinophils manufacture and secrete cytokines and many granule proteins, and these factors are involved in mesenchymal transition and airway epithelial changes; both events are considered to be major driving forces for remodeling. From a clinical perspective, eosinophils are present in~50% of asthmatic patients 1 . Along with evidence acquired by animal models of allergic airway inflammation, it has been suggested that eosinophil effector functions play an active role in mediating airway inflammation 21,22,29 . Our allergen provocation experiment suggests that VAMP-7-mediated degranulation from eosinophils is important in exacerbation of AHR.  30 . All antibodies were used at 10 μg mL −1 or 1:100 dilution for all staining or immunoassays. All antibodies were also validated by the Mayo Clinic (Scottsdale, AZ, USA) for in-house use or by the respective commercial sources on mouse proteins using specific ELISAs.
Mice. All mice (Mus musculus) used in this study were bred against the C57BL/6 background for at least eight generations. Mouse strains included C57BL/6J (WT), C57BL/6-transgenic (Tg) IL-5NJ.1638/Lee Labs (I5) 17 , and double Tg IL-5/human eotaxin-2 (hE2) crossed with EPX-deficient mice (I5/hE2/EPX −/− ; Lee Labs 19 ). The eosinophil-specific B6.129P2-EPX tm1(Cre) /Lee Labs/Ozgene strain was also used (eoCRE, Lee Labs, Mayo Clinic and Ozgene, Bentley DC, WA, Australia 15 , as well as C57BL/6-Tg(Zp3-Cre)93Knw/J (Zp3/V7, Jackson Laboratory, MA, USA), B6; and 129-Vamp-7 tm1 /RIKEN BRC (VAMP-7 f/f , RIKEN BioResource Center, Ibaraki, Japan). Zp3/V7 refers to nearly complete gene deletion of VAMP-7 based on widespread gene expression of the zona pellucida 3 (Zp3) gene. eoCRE mice crossed with VAMP-7 f/f mice generated a new strain carrying eosinophils deficient in VAMP-7 (eoCRE/V7) 31,32 . Hemizygous experimental eoCRE mice (eoCRE +/− ) were used as previously described 15 . To elicit substantial numbers of eosinophils required for in vitro and ex vivo experimental analyses, I5 mice were crossed with eoCRE +/− or eoCRE/V7 strains to generate eoCRE/I5 and eoCRE/V7/I5 offspring, respectively, which contained abundant circulating peripheral blood eosinophils. In other experiments, VAMP-7 f/f mice were crossed with I5 mice to generate VAMP-7 f/f /I5 mice. Male and female mice were maintained and bred in ventilated microisolator cages housed in the Mayo Clinic's pathogen-free (SPF) animal facility. The following mouse strains are available for distribution from the Mayo Clinic: eoCRE, I5 transgenic, eoCRE/I5, eoCRE/V7/I5, and I5/hE2. VAMP-7 f/f are available from Riken, Japan, while the eoCRE/V7 strain has been discontinued at our site. Progeny of all crosses were viable and fertile, and mice at 8-12 weeks old were used for experiments. Littermates were equally divided for treatments and controls. Samples were allocated into control and experimental groups according to the transgenic background of mouse strains used. A formal randomization tool was not employed. All procedures were performed according to University of Alberta (Edmonton, AB, Canada) and Mayo Clinic's animal care ethics committee guidelines, and for human eosinophils isolated from voluntary donor blood samples, procedures were approved by the University of Alberta's human research ethics committee.
Cre-loxP recombination. To overcome problems associated with ubiquitous gene ablation and to address complex phenotypes associated with the loss of a single gene, tissue-specific genetic recombination was achieved by the eoCRE-loxP system 15,31 . A cell-specific, characterized promoter (i.e., EPX promoter in eosinophils) was used to drive Cre recombinase expression, thereby restricting enzyme expression in eosinophils which are the only cells expressing EPX in mice 15 . Using embryonic stem cell manipulation, the loxP sequences were inserted in flanking regions adjacent to the VAMP-7 gene locus (VAMP-7 f/f ) mice to generate cell lineage-specific KO mice 32 . The offspring generated from eoCRE bred with VAMP-7 f/f mice were designated 'eoCRE/V7' mice.
Eosinophil isolation. Eosinophils used in in vitro experiments were isolated at >98% from the peripheral blood of I5 and I5-crossed strains as previously described 17,33 . Briefly, peripheral blood was layered on top of Histopaque 1119 (Sigma-Aldrich) to yield an enriched eosinophil stratum. The eosinophil-containing layer was quickly treated with red blood cell-lysing ice-cold distilled water and washed with 1× PBS prior to isolation. Anti-mouse CD45R (B220)-and CD90 (Thy 1.2)-coated magnetic beads coupled with the MACS system was employed to isolate eosinophils by negative selection according to manufacturer's instructions (MACSxpress Eosinophil Isolation Kit, Miltenyi Biotec). Human eosinophils were isolated using MACS eosinophil isolation kits (Miltenyi Biotec). Eosinophil population purity was determined by Diff-Quik-stained cytospin samples.
DNA isolation for genotyping. Tail biopsies were cut and digested using a DNeasy Blood and Tissue kit (Qiagen) according to the manufacturer's instructions. Samples were stored at 4°C until PCR analysis.
Cre identification by polymerase chain reaction. DNA recovered from tail biopsies was used as a PCR template for genotypying animals that potentially carried the Cre-recombined VAMP-7 allele. Cre-recombined VAMP-7 null allelepositive animals were identified using a four-primer (P-1, P-2, P-a, and P-b) strategy (Fig. 2a). The amount of product amplified was indicated using the standardized ΔΔCt value generated by the fluorescence intensities of dsDNA resulting from PCR amplification. Primers P-a and P-b were designed based on the sequence between exons (ex) 2 and 3 of the VAMP-7 gene. The qPCR-amplified product generated using P-a and P-b primers resulted from excision of ex 3 and 4 of the VAMP-7 gene following Cre-mediated recombination while primers P-3 and P-4 were designed based on the sequence within the un-excised region (ex 1). The ΔCt values, defined in equations (1) and (2), of the amplicon were generated by dividing primers P-1 and P-2 or primers P-a and P-b by the constant (primers P-3 and P-4).
Dividing equation (1) by equation (2) provided a ΔΔCt value ratio for each experimental group, as shown in equation (3).
The control (eoCRE +/− ) allele (328 bp PCR amplicon) was derived from ex 2 and 3 using P-a and P-b while the Cre-recombined allele (326 bp amplicon) was identified using ex 5-derived primers P-a and P-b. Reaction conditions were as follows: 94°C for 2 min (annealing) followed by 35 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min (elongation), and a final extension at 72°C for 5 min. PCR amplicons were analyzed using a 2% agarose gel electrophoresis coupled with ethidium bromide DNA staining. Primer sequences used for PCR and qPCR in this study are shown in Supplementary Table 1. qPCR. Two primers (P-1 and P-2) were designed based on the sequence located immediately up-stream of the first loxP site in ex 3 (104 bps). Primers P-3 and P-4 were designed based on the sequence in ex 1 (i.e., the un-excised control region; 54 bps). VAMP-7 excision efficiency was evaluated using the standard ΔΔCt method and by comparing the qPCR (BioRad My iQ Single Color Real-Time PCR Detection System; BioRad iQ5 software) results from the excised region of VAMP-7 to that of ex 1; the lower the ratio, the more Cre-recombined allele present. Reaction conditions were as follows: 95°C for 10 s (40×), 55°C for 30 s, 95°C for 1 min, and 55°C for 1 min, followed by a melting cure up to 95°C at 10 s intervals. qPCR data were normalized to VAMP-7 f/f mice. Primer sequences used for qPCR in this study are in Supplementary Table 1.
Bone marrow isolation and mouse eosinophil derivation. Both femur and tibia were surgically removed from CO 2 -killed mice, and a scalpel was used to remove the epiphyses of long bones. Femurs and tibias were flushed, and bone marrow suspensions were transferred to in vitro derivation was carried out using the procedure described by Dyer et al. 35 . Bone marrow cell cultures were incubated with recombinant murine stem cell factor (100 ng mL −1 , PeproTech, Rocky Hill, NJ, USA) and Flt-3 ligand (100 ng mL −1 , PeproTech).
On day four of culture, part of the cell culture (15 mL) was resuspended in RPMI supplemented with recombinant mouse IL-5 (rmIL-5; 10 ng mL −1 ; R&D Systems), and incubated. Prior to incubation, cells were sampled for cytospin and DNA extraction, as well as for PCR identification of the Cre recombined allele. An additional sample was removed from cell culture for DNA extraction 2 days later. On day eight, cells were subsampled before the entire culture was resuspended in RPMI supplemented with 10 ng mL −1 rmIL-5 and returned to the incubator. After cell culture media were renewed on day 10, cells were resuspended for cytospin and DNA extraction, while supernatants were collected for ELISA analysis of EPX. Prior to cell culture termination on day 12, cells were sampled for cytospin, and visual inspection of Diff-Quik-stained cytospin preparations revealed that >90-100% of total cells were eosinophils.
WBC isolation by red blood cell lysis. Heat lamp-warmed mice were placed in a holder to collect peripheral (tail) blood from animals. Blood added to heparincontaining Eppendorf tubes was inverted, and then filled with lysis buffer (BD Pharm Lyse™, BD Biosciences). After lysis, cell suspensions were centrifuged, and cell pellets resuspended by tapping or by gentle vortexing. This lysis process was repeated for up to 10 min or until a clear WBC pellet was visible. The pellet was washed with PBS-containing 0.25% BSA and 2 mM EDTA.
Granule protein and cytokine/chemokine assays. EPX release was measured using an ELISA kit developed by the Mayo Clinic (Scottsdale, AZ, USA) 16 . Standards for EPX analysis were prepared by purifying eosinophils (14.6 × 10 6 cells per mL) isolated from the peripheral blood of I5 mice. After the cell suspensions were spun and their supernatants (200 μL) removed, 250 µL of 0.22% hexadecyltrimethylammonium bromide (Sigma) in 0.3 M sucrose solution was added to lyse cell pellets. Lysates were vortexed (1 min), flash-frozen in liquid nitrogen, and stored at −80°C. Lysates standards were thawed on ice and pulse-spun (16,000×g) immediately prior to use.
The release of MBP was assessed using an immunoblot assay as previously reported 18 . Briefly, 10 μL aliquots of samples were pipetted onto nitrocellulose membranes, which were then dried for 15 min and then incubated with biotinylated rat anti-mouse MBP for 1 h at room temperature, blocked with 1% "Blocker Casein" in PBS (Pierce) for 45 min, and thrice washed with PBScontaining 0.05% Tween 20 prior to incubation with streptavidin-alkaline phosphatase conjugate (Roche Applied Science) for 1 h at room temperature. Membranes subsequently washed with PBS-containing Tween 20 were incubated with Lumi-Phos WB Chemiluminescent Substrate (alkaline phosphatase; Pierce) and chemiluminescence intensity detected using CL-XPosure Film (Pierce). Samples were compared to a protein lysate standard curve derived from known eosinophil numbers.

Airway inflammation induction by acute OVA sensitization and challenge.
Experimental eoCRE +/− and eoCRE/V7 mice were sensitized and challenged with chicken OVA according to the procedure described in work published previously 36 . Briefly, male mice were sensitized on days 0 and 14 by i.p. injection of 20 µg OVA (Sigma-Aldrich) and 2.25 mg Imject Alum Adjuvant (Thermo Fisher Scientific) resuspended in 100 µL 0.9% sodium chloride (saline control; Hospira, Saint-Laurent, QC, Canada). On days 24, 25, and 26, sensitized mice were exposed for 25 min to aerosolized (1% w/v) OVA dissolved in saline within individual compartments of a mouse 'pie' chamber (Braintree Scientific, MA, USA) using a Pari IS2 nebulizer (Sun Medical Supply, Kansas City, KS, USA) connected to air compressor (30 mg pressure; PulmoAID, DeVilbiss, Somerset, PA, USA) while control mice received an aerosol challenge with saline for the same amount of time. Mice were rested on day 27 and the next day assessed for pulmonary infiltrate, histopathology, and lung function. Several mice were then killed, and BAL collected as described below.
Preparation and quantification of BAL-derived cells. BAL fluid was collected according to a procedure described by Lee et al. 17 . Briefly, tracheotomy was performed to expose the trachea of sodium pentobarbital-killed mice. After an 18gauge catheter (1.3 × 30 mm; BD Angiocath, BD Biosciences) was inserted into the trachea, the lungs were lavaged with 1 mL aliquots of ice-cold PBS-containing 0.2% BSA (Thermo Fisher Scientific); 0.7-0.9 mL instilled lavage fluid was typically recovered. Collected BAL samples were centrifuged (500×g; 5 min; 4°C) to pellet cells later used for cytospin and differential cell counts. Lysis buffer (1×; 70 µL; BD Pharm Lyse) and 5% BSA in PBS (500 µL) were added to the cell pellet if undesired red blood cells were present. Total BAL cell counts were carried out using a hemocytometer. All BAL supernatants were centrifuged further (16,000×g; 10 min; 4°C) to remove lung debris and stored at −80°C until further analysis.
Cytospin staining for differential cell counts. Samples (~100,000 cells per mL) were loaded onto pre-coated (10% FBS) ColorFrost Plus microscope slides (Thermo Fisher Scientific) and spun (500 rpm; 5 min; RT) in a cytocentrifuge (Thermo Scientific Cytospin 4, A78300003). Air-dried slides were differentially stained using a modified Wright Stain technique (Siemens Diff Quik Stain Kit, Thermo Fisher Scienfic). Then, slides were immersed in xylene and cover-slipped using Shandon Consul-Mount (Thermo Fisher Scientific). Cells (300 cells per sample) were counted to obtain differential cell counts of BAL samples.
Histopathology and IHC. Lung tissues from eoCRE +/− and eoCRE/V7 mice were inflated in situ with 10% formalin, fixed by 4% (w/v) paraformaldehyde perfusion, embedded in paraffin, and placed onto slides. Histopathological changes of the airways were assessed using conventional staining procedures described elsewhere 17,21 . Briefly, lung sections (4 µm) prepared from formalin-fixed, paraffin-embedded tissue blocks were stained with H&E, Masson's Trichome or PAS. H&E staining allowed for a general assessment of histopathology including inflammatory cell infiltrates, epithelial and airway smooth muscle hypertrophy, and/or hyperplasia. Masson's Trichrome facilitated the evaluation of collagen deposition and fibrosis while goblet cell metaplasia and airway mucin accumulation were assessed by H&E staining. Eosinophils were detected by rabbit polyclonal anti-mouse eosinophil MBP antiserum 40 .
IHC of rat anti-mouse MBP mAb stained lung sections (4 μm) have been described elsewhere 41 . Rat anti-mouse MBP mAB (200 µL; 2 µg mL −1 ) diluted in 1.5% normal rabbit serum or negative control (Rat IgG, Vector Labs) was added to each slide; unbound antibodies were removed and bound rat anti-mouse MBP mAb was detected by anti-rat IgG biotinylated mouse adsorbed reagent (secondary antibody; 200 µL; 0.4 µg mL −1 ; Vector Labs). After appropriate counterstaining, rinsed slides were air-dried and cover-slipped with Shandon Consul-Mount. Our clinical lung pathologist was blinded to all histology during quantification. Slides were provided with a simple numbering identification. The pathologist provided a written report of the findings without knowledge of the groups.
Assessment of methacholine challenge-induced AHR. Lung function was evaluated 48 h after final OVA challenge using previously described methods 21 . Briefly, tracheotomy was performed on pentobarbital sodium (diluted 1:5 in saline; Abbott Laboratories)-anesthetized mice (i.p. injection; 90 µg per g body weight). A fire-polished, glass endotracheal cannula (18G), inserted into the trachea and secured by sutures, was attached to both a pneumotachograph (model 8410, Hans Rudolph) and an ultrasonic nebulizer (Porta-Sonic model 8500 C, DeVillbiss Health Care) for airflow obstruction with aerosolized methacholine (Sigma-Aldrich). Pancuronium bromide (Sigma-Aldrich)-paralyzed mice (0.5 µg per g body weight) were placed on a 37°C heating station to maintain body temperature and a calibrated computer-controlled ventilator (Flexivent, SCIREQ, Montreal, QC, Canada; tidal volume: 8 mL per kg; frequency: 2.5 Hz) provided ventilation; the ventilator expiratory line was submerged in water to apply a positive endexpiratory pressure (2-3 cm H 2 O). Increasing concentrations (0, 3, 6, 12, 25, 50, and 100 mg per mL) of aerosolized methacholine in sterile saline was delivered to the animal's trachea (20 breaths per min; 30 s; tidal volume of ventilator piston: 0.8 mL) to assess airway responses. To re-establish a standard volume, lungs were inflated to total lung capacity (30 cm H 2 O) prior to each aerosol challenge. Seven breaths were collected during regular ventilation or after each aerosol exposure to measure the animal's baseline airway function and resistance, respectively.
Statistical analyses. Experiments were repeated independently at least three times. All attempts at replication were successful, and no experiment was found to be irreproducible. GraphPad Prism 7 (GraphPad Software Inc., La Jolla, CA, USA) and Adobe Illustrator were used for statistical analysis and to create graphs, respectively. No data were excluded from analysis. Data are presented as mean ± standard error of the mean (SEM). Data were analyzed for significance using either a one-way analysis of variance (ANOVA) with a Tukey's multiple comparison test or two-tailed Student's t-test. Significance levels were set at either p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), or p < 0.0001 (****). No relevant data existed to perform an accurate power calculation; therefore, sample sizes were not chosen a priori but rather to ensure reproducibility.
Data availability. The data that support the findings of this study are available from the corresponding author upon reasonable request.