Mast Cell Fibroblastoid Differentiation Mediated by Airway Smooth Muscle in Asthma

Asthma is a major cause of morbidity and mortality (1). It is characterized by variable airflow obstruction and airway hyperresponsiveness (AHR) in association with Th2 eosinophilic airway inflammation and remodeling (2). Asthma is a heterogeneous condition, and the relationship between airway inflammation and dysfunction is complex (3). Indeed, anti–IL-5 therapy has resulted in successful reduction in eosinophilic inflammation without changes in AHR (4, 5), illustrating that eosinophilic inflammation and AHR are dissociated. Comparative studies between asthma and eosinophilic bronchitis, a common cause of cough (6), have been very informative about the cause of disordered airway physiology in asthma and suggest that the microlocalization of mast cells within the airway smooth muscle (ASM) bundle is critical for the development of AHR (7, 8).

Mast cell progenitor recruitment to the lung is in part mediated by leukotriene B4 (9) and chemokines (10) and is integrin dependent (11). In the airway, mast cell trafficking to different compartments is under the influence of several chemotactic factors (reviewed in Ref. 12). Recruitment to the ASM bundle in asthma is predominantly via activation of the CXCR3/CXCL10 axis (13). Within the ASM bundle, mast cells adhere to ASM cells, in part via heterophilic adhesion mediated by cell adhesion molecule-1 (CADM1) expressed by mast cells (14). However, the CADM1-independent adhesion mechanism remains unexplained. Mast cells within the ASM bundle are in an activated state with increased IL-4 and IL-13 expression (15, 16), are smaller than mast cells observed in the lamina propria, and undergo piecemeal degranulation as evidenced by electron microscopy (17). Whether these changes simply represent mast cell activation or a transition to an altered phenotype is unknown. ASM-bundle–resident mast cells are intimately associated with both the ASM cells and the local matrix environment. Therefore, the potential for mast cells to undergo phenotypic transition as a consequence of interactions with ASM and extracellular matrix (ECM) proteins needs to be explored further.

The ECM composition varies between different compartments in the airway wall, between asthma versus a healthy state, and across disease severity. This matrix is not a benign bystander that simply constitutes the scaffold to build airway structures and allow inflammatory cell trafficking, but can alter the biophysical properties of the airway wall and modulate cell function (18). In those with fatal asthma, compared with nonasthmatics and asthmatics who died of other causes, elastin fibers and fibronectin are increased in the ASM bundle (19). We therefore hypothesized that ASM-derived ECM proteins exert important effects on mast cells localized within the ASM bundle.

In this study, we have demonstrated that mast cells localized within the ASM bundle express fibroblast markers and their number in the ASM bundle is closely related to AHR. We have identified in vitro that this fibroblastoid mast cell transition is mediated by ASM-derived ECM proteins. These fibroblastoid mast cells are activated and release increased concentrations of histamine, and their number in the ASM bundle is closely related to AHR.

Asthmatic subjects and nonasthmatic controls were recruited from Leicester, U.K. Asthmatic subjects had a consistent history and objective evidence of asthma, as described previously (7, 13). Subjects underwent extensive clinical characterization including video-assisted fiberoptic bronchoscopic examination. The study was approved by the Leicestershire Ethics Committees. All patients gave their written informed consent.

Sequential 2-μm sections were cut from glycomethacrylate-embedded bronchial biopsies and stained using mAbs against α-smooth muscle actin (α-SMA; Dako, Ely, U.K.), mast cell tryptase (Dako), the fibroblast marker Thy-1 (Ab1) (Calbiochem, San Diego, CA) (20), and appropriate isotype controls (Dako). Thy-1⁺tryptase⁻ (fibroblast), Thy-1⁻tryptase⁺ (mast cell) and Thy-1⁺tryptase⁺ cells (fibroblastoid mast cell) were enumerated per square millimeter ASM by colocalization (7, 13, 15). The whole biopsy was assessed, and a minimum area of 0.1 mm² was considered assessable, as described previously (7). Double immunostaining for tryptase and Thy-1 fibroblast marker was performed using the EnVision Doublestain Kit (Dakocytomation, Carpinteria, CA) according to the manufacturer’s instructions. Thy-1 expression was developed with peroxidase and 3,39-diaminobenzidine tetrahydrochloride (brown reaction product) and for tryptase alkaline phosphatase and fast red (red reaction product). Sections were then counterstained with hematoxylin and mounted in an aqueous mounting medium (BDH Chemicals, Poole, U.K.). Appropriate isotype controls were performed in which the primary Abs were replaced by irrelevant mouse mAb of the same isotype and at the same concentration as the specific primary mAb.

Pure ASM bundles were isolated from bronchoscopic samples (n = 12; 10 asthmatic subjects, 2 nonasthmatic) and from lung resection (n = 10). Primary ASM was cultured and characterized as previously described (21). Cells used were passage 2–5. Primary fibroblasts were isolated from bronchial biopsies, characterized for the fibroblast markers Thy-1 and 1B10 (AbCam, Cambridge, U.K.), and used between passage 6 and 8 (n = 13). The human lung mast cells (HLMCs) were isolated and cultured from nonasthmatic lung (n = 19), as previously described (21). The human mastocytoma cell line (HMC-1) was a generous gift from Dr. J. Butterfield (Mayo Clinic, Rochester, MN), and was cultured as described previously (21).

HMC-1 cells (1 × 10⁵cells/well) were seeded onto six-well plates uncoated or coated with 5–40 μg/ml fibronectin (Sigma-Aldrich, Poole, Dorset, U.K.) or 60 μg/ml collagen (Inamed Biomaterials, Fremont, CA) in ITS Media (ASM media with 1% ITS+3 supplement [Sigma-Aldrich] in place of FBS). After 6 d, nonadherent cells were collected and counted. The percentage of elongated adherent HMC-1 cells was calculated from five random high-powered fields. Adherent cells were then harvested using Accutase (Insight Biotechnology, Wembley, London, U.K.) and the percentage of adherence determined as the proportion of adherent cells/total cells.

ASM cells (5 × 10⁴ cells/60 mm dish for HLMC coculture; 2.5 × 10⁵cells/T75 flask for HMC-1 cocultures) were incubated at 37°C for 48 h, prior to incubation with ITS media at 37°C for 24 h. HLMCs or HMC-1 cells were then added in ITS media at a 1 (mast cell):4 (ASM cell) ratio (based on the mean mast cell density in asthmatic ASM bundles in bronchial biopsies) and cultured over 15 d. Mast cell monoculture controls were established in parallel in ITS media plus SCF where appropriate.

To study the functional contribution of receptors for HMC-1/HLMC adhesion, cells were pretreated with anti-α5β1 (6.5 μg/ml, clone JBS5; Chemicon International, Hampshire, U.K.) and CADM1 [10 μg/ml, clone 9D2, a generous gift from Dr. A. Ito (22)] or appropriate isotype controls (IgG1 [Dako], chicken IgY [R&D Systems, Abingdon, Oxfordshire, U.K.]), either for the duration of 1 h prior to seeding onto ASM or fibronectin (40 μg/ml) or throughout the 7-d incubation for histamine release and PGD₂ release. After 6–7 d, the percentage of adherent cells was calculated.

HMC-1 cells were added in ITS media at a 1:4 ratio to Transwell membranes (0.4 μm; Corning, Corning, NY) inserted into wells containing ASM cells (1 × 10⁵ cells/well in six-well plates, incubated at 37°C for 24–48 h, prior to incubation with ITS media for 24 h) or supernatant from ASM incubated in ITS media for 24 h or ITS media alone. After 6 d, the percentage of adhesion was calculated.

Following ECM/coculture assays, cells were stained for mast cell/fibroblast markers, where appropriate, by flow cytometry. Elongation of mast cells in coculture with ASM was also visualized by immunofluorescence.

HMC-1 cells were stained, where appropriate, with Abs to Thy-1, 1B10, Collagen 1 (Chemicon, Temecula, CA), CD117 (BD Pharmingen, San Diego, CA), chymase (Chemicon, Hampshire, U.K.), and tryptase (Dako), or their appropriate isotype controls indirectly labeled with FITC (Dako) or allophycocyanin (R&D Systems) and analyzed by one-color or two-color flow cytometry (in coculture samples) on a FACSCanto (BD Biosciences, Oxford, U.K.). Mast cells were distinguished from ASM by prelabeling with CellTrace CFSE prior to coculture (according to manufacturer’s instructions, Invitrogen Molecular Probes, Paisley, Scotland, U.K.) or by staining with CD117-RPE (Dako).

Cells were stained, where appropriate, for surface Thy-1, 1B10, or appropriate isotype controls; indirectly labeled with FITC or allophycocyanin; and counterstained with DAPI (Sigma). Mast cells were distinguished from ASM cells by prelabeling with CellTrace CFSE or by staining with CD117-RPE and indirect labeling with NorthernLights-RPE (R&D Systems). Double immunofluorescence was performed on cocultured ASM and mast cells and mast cells adherent to fibronectin using biotinylated tryptase mAb (Promega, Madison, WI) and chymase mAb (Chemicon, Temecula, CA). The biotinylated Ab binding was detected using Streptavidin-Texas Red, and chymase Ab binding was detected using secondary FITC-labeled Ab. Sections were counterstained with DAPI. The slide was then mounted with fluorescent mounting medium and analyzed.

Following culture in fibronectin-coated six-well plates over 6 d, the extent of collagen gel contraction by HMC-1 cells was assessed. A total of 1.25 × 10⁵ HMC-1 cells were resuspended in 144 μl ITS media ± stimulus, with 299 μl collagen, 37 μl 10× DMEM (Invitrogen), and 20 μl sodium bicarbonate (Invitrogen), and added to 24-well plates. The mixture was left to polymerize into gels at 37°C for 90 min prior to detachment from the well; 500 μl ITS media ± stimulus was added and incubated over 5 d. Gel surface area was measured using ImageJ (http://rsb.info.nih.gov/ij) by a blinded observer. Collagenase (1.9 mg/ml; Sigma-Aldrich) for 20 min at 37°C was used to extract HMC-1 cells from the gels, to further characterize mast cell differentiation.

Histamine was measured by a sensitive radioenzymatic assay, as described previously, and corrected for cell number (23).

PGD₂ expression in cocultured cell supernatants was measured by ELISA (Cayman Chemical, Ann Arbor, MI).

Statistical analysis was performed using GraphPad Prism 4 (GraphPad, San Diego, CA). Parametric data are presented as mean ± SEM, log normally distributed data were log transformed and expressed as geometric mean (95% confidence interval), and nonparametric data are presented as median (interquartile range [IQR]). Comparisons across groups was analyzed by ANOVA and post hoc Tukey’s test or by Kruskal–Wallis testing with post hoc Dunn’s pairwise comparisons for nonparametric data. Between-group comparisons were made by paired and unpaired t tests, as appropriate. Differences were considered significant when p < 0.05.

To determine the localization of mast cells, fibroblasts, and cells coexpressing markers for both cell types in the airway wall, bronchial biopsies from 16 healthy controls and 26 subjects with asthma were studied. The clinical characteristics of these subjects are shown in Table I. Representative photomicrographs of mast cells (tryptase⁺ cells) and fibroblasts (Thy-1 [Ab1]⁺ cells) within the ASM bundle are shown in Fig. 1A–D. Mast cells were observed within the ASM bundles typically surrounded by matrix, as illustrated (Fig. 1C). Double immunostaining for Thy-1 and tryptase on the same section are shown in Fig. 1E, where the Thy-1 cells are stained individually and double stained with tryptase⁺ mast cells within the ASM bundles (n = 3 severe asthmatics). The median (IQR) mast cell number within the ASM bundle was increased in mild-moderate (6.8 [9.8] cells/mm² ASM) and severe asthma (6.2 [16.4] cells/mm² ASM) compared with healthy controls (1.7 [4.4] cells/mm² ASM; p = 0.001 Kruskal–Wallis, p < 0.01 asthma groups versus controls, Dunn’s pairwise comparison) (Fig. 1F).

The number of fibroblasts was also increased in the ASM bundle in mild-moderate (4.4 [11.5] cells/mm² ASM) and severe asthma (6.5 [10.2] cells/mm² ASM), compared with controls (1.5 [2.5] cells/mm² ASM; p = 0.003 Kruskal–Wallis, p < 0.05 asthma groups versus controls) (Fig. 1G). The proportion of mast cells that coexpressed fibroblast markers (tryptase⁺/Thy-1⁺ cells) was increased in the ASM bundle (18 ± 3%) compared with the lamina propria (2.9 ± 0.6%; p < 0.0001), and in the ASM bundle the proportion was increased in mild-moderate (22 ± 3%) and severe asthma (22 ± 4%), compared with healthy controls (4 ± 2%; p = 0.0002 ANOVA, p < 0.01 asthma groups versus controls, Tukey’s post hoc comparison) (Fig 1H).

In the 21 asthmatic subjects who underwent bronchial challenge, the number of mast cells within the ASM bundle was significantly inversely correlated with the degree of AHR (r = −0.45; p = 0.04; Fig. 1I). There was no significant correlation between the number of fibroblasts in the ASM bundle and AHR (data not shown). The number of mast cells that coexpressed fibroblast markers was strongly correlated with the degree of AHR (r = −0.61; p = 0.004; Fig. 1J). There was no significant correlation between FEV₁ percentage predicted and the number of mast cells, fibroblasts, or fibroblastoid mast cells in the ASM bundle (data not shown).

We therefore considered whether mast cells differentiate to a fibroblastoid phenotype in the presence of ECM proteins. Indeed, we found that in vitro the morphology of the HMC-1 changed over 6 d in the presence of plasma fibronectin, with cells becoming elongated (Fig. 2A). The percentage of elongated cells was related to the concentration of fibronectin in a dose-dependent manner (Fig. 2B). The number of elongated cells was greatest in the presence of fibronectin 40 μg/ml. Therefore this concentration of fibronectin was used throughout the study. At this concentration, HMC-1 cells adhered avidly to fibronectin after 6 d (81 ± 5%, n = 8; p < 0.001), and adhesion was markedly inhibited by anti-α5β1 blocking Ab (4.3 ± 0.4% versus isotype control, 78 ± 2%; p = 0.0005; n = 3; Fig. 2C).

Previous reports have shown HMC-1 cells to adhere to collagen at a maximum concentration of 60 μg/ml (6). At this concentration we found that the proportion of HMC-1 cells that adhered to collagen after 6 d was 59 ± 2% (n = 5; p < 0.001), and the proportion of elongated cells was increased (44 ± 2.4%; n = 4; p < 0.001).

The fibroblast markers Thy-1 and 1B10 were highly expressed on cultured fibroblast cells, as evidenced by immunofluorescence (Fig. 3A) and by flow cytometry (Fig. 3B). The proportion of fibroblast cells that expressed Thy-1 was 67 ± 7% (p = 0.001; n = 13), and 1B10 was 26 ± 5 (p = 0.003; n = 7) (Fig. 3C).

HMC-1 cells cultured for 6 d without fibronectin (control), HMC-1 cells cultured with fibronectin that did not adhere (nonadherent cells) and those cells that adhered (adherent cells) were analyzed by flow cytometry for surface Thy-1, IB10, CD117, and intracellular tryptase expression. 1B10 expression was significantly increased in the adherent HMC-1 cells (Δ geometric mean fluorescence intensity [GMFI] 137 ± 21; n = 6) compared with nonadherent cells (64 ± 14; p = 0.001) and controls (81 ± 23; p = 0.002). Example histograms are shown in Fig. 3D. No difference was seen in the expression of collagen I, CD117, or tryptase by the adherent or nonadherent HMC-1 cells (Fig. 3D, 3E). However, there was a significant increase in chymase expression in adherent HMC-1 cells (ΔGMFI 21 ± 2.3; n = 3) compared with control (6.7 ± 1.76; p = 0.002). HMC-1 cells cultured with or without fibronectin did not significantly express Thy1 (n = 5; data not shown).

In the presence of the anti-α5β1 blocking Ab, HMC-1 cells did not express 1B10 (Fig. 3F). We were unable to compare 1B10 expression in adherent versus nonadherent HMC-1 cells in the presence of the blocking Ab, as it markedly inhibited adhesion and therefore resulted in insufficient adherent cells for flow cytometry. Expression of 1B10 by adherent HMC-1 cells was confirmed by immunofluorescence (Fig. 3G).

In the presence of collagen (60 μg/ml) for 6 d, adherent HMC-1 cells demonstrated increased 1B10 expression (ΔGMFI 79 ± 5; n = 3) compared with nonadherent cells (33 ± 9; p = 0.025) and HMC-1 cells alone (28 ± 6; n = 3).

We considered whether during fibroblastoid transition in a three-dimensional matrix mast cells undergo shape change that induces contraction. Therefore, to test whether the HMC-1 cells that were nonadherent to fibronectin in two-dimensional culture after 6 d have the capacity to undergo fibroblastic transition in a three-dimensional matrix, we cultured adherent and nonadherent HMC-1 cells in collagen gels for 5 d.

We confirmed that after 6 d culture with fibronectin, the 1B10 expression was increased in the adherent versus nonadherent HMC-1 cells (ΔGMFI [95% confidence interval] 233 [140–325]; p = 0.003; n = 3) (Fig. 4A). Following culture in collagen gels for a further 5 d, there was no difference in 1B10 expression between the adherent and the nonadherent HMC-1 cells (Δ GMFI [95% confidence interval] 43 [−26–112]; p = 0.18), but the change from baseline in 1B10 expression in the nonadherent cells was significantly increased compared with that in the adherent cells (280; 95% CI [142–418]; p = 0.005) (Fig. 4A). This observation suggests that the nonadherent mast cells in the two-dimensional culture had indeed undergone fibroblastoid transition in collagen gels.

This HMC-1 fibroblastoid transition was associated with a significant reduction in the size of the collagen gel. Representative gels are as shown (Fig. 4B). After 5 d the mean ± SEM percentage size of gel compared with baseline was 66 ± 4% (n = 7) for nonadherent cells versus 82 ± 4% for adherent cells (p = 0.025) (Fig. 4C). This finding illustrates that the mast cells had undergone concomitant changes in cell shape and expression of fibroblast markers in the collagen gels.

To investigate the effects of mast cell-ASM interactions on mast cell phenotype, we extended our model to a coculture system. Primary HLMCs and HMC-1 cells cocultured with primary ASM cells over 7 d changed their cell morphology to become more elongated (Fig. 5A). This change in cell morphology was also confirmed by immunofluorescence (Fig. 5B). The percentage of tryptase/chymase-positive cells increased significantly after 10 d (27 ± 5.8%; p = 0.05; n = 13; Fig. 5C, 5D) compared with cells in coculture (9.85 ± 5.3%; n = 7) and cells adherent to fibronectin (10.1 ± 3.3%; n = 6; p = 0.02) over 1 d.

In contrast to HMC-1 cells, HLMCs expressed 1B10 in the absence of fibronectin and/or ASM (13 ± 5%; n = 3), whereas the percentage of Thy-1–positive HLMCs in the absence of fibronectin and ASM was low (4.7 ± 1.7%). Therefore, Thy-1 expression by mast cells was studied in coculture with ASM cells.

To track mast cells in coculture with ASM over 6–15 d, mast cells were labeled with the fluorescent marker CFSE, a stable dye that is not passed between cells upon adhesion. With use of flow cytometry, CFSE-labeled mast cells were gated and analyzed for Thy-1 expression. Example flow cytometry dot plots and histograms are as shown, illustrating Thy-1 expression by HLMCs, HMC-1 cells, and in cells cocultured with ASM (Fig. 5E). Following coculture with ASM, HLMC Thy-1 expression was increased compared with control after 7 d (22 ± 5%; p = 0.005; n = 6) and 15 d (45 ± 8%; p = 0.004; n = 6) (Fig. 5F). A small, albeit significant, increase in Thy-1 expression was observed for HMC-1 cells (9 ± 2%; p = 0.011; n = 7) cocultured with ASM for 6 d, compared with cells alone (Fig. 5G). Similarly, Thy-1 expression by HMC-1 cells cultured with ASM separated by a Transwell insert or with ASM-conditioned media was increased, compared with that in controls. This increased expression was reduced in comparison with that of HMC-1 cells in coculture with ASM (Fig. 5G; p < 0.05).

HMC-1 cell Thy-1 expression was not modulated by TGFβ-1 or 2 (0–10 ng/ml) (data not shown).

We considered whether this mast cell fibroblastoid transition was dependent upon mast cell-ASM adhesion or mast cell ECM-ASM interactions. The percentage adhesion of HMC-1 cells to ASM after 6 d coculture was 57 ± 7%, n = 4. In contrast to short-term culture (14), this HMC-1 adhesion to ASM after long-term culture was not inhibited by preincubation with the CADM1 blocking Ab (58 ± 9% versus isotype control 57 ± 9%; n = 4; p = 0.67) but was markedly inhibited by α5β1 blockade (26 ± 4% versus isotype control 68 ± 6%; p = 0.0008; n = 6) (Fig. 6A). The blockade of CADM1 and α5β1 in concert was not different to α5β1 alone (p = 0.49) (Fig. 6A).

In coculture with ASM and in the presence of the α5β1 blocking Ab, the proportion of Thy-1⁺ HMC-1 cells (6.4 ± 4.3%; n = 4) and HLMCs (4.9 ± 1.4%; n = 5) was significantly decreased compared with their isotype controls (27.4 ± 7.5% and 13.4 ± 3.1%, respectively; p < 0.05) (Fig. 6B). Blockade of CADM1 did not affect Thy-1 expression by HLMCs (15.8 ± 3.9% versus 17.7 ± 1.3%; p = 0.59).

The histamine concentration in the media corrected for cell number following coculture of HLMCs with ASM for 7 d was significantly reduced in the presence of the anti-α5β1 blocking Ab compared with its isotype control (2.4 ± 0.5 μg/10⁶ cells versus 3.3 ± 0.6 μg/10⁶ cells; p = 0.02; n = 6) (Fig. 6C). However, IgE–anti-IgE activation of HLMCs prior to incubating with ASM did not significantly reduce histamine release (Fig. 6C). PGD₂ release by HLMCs was not affected by anti-α5β1 (117.7 ± 47.2 ng/10⁶ cells; p = 0.9) compared with IgG1 control (109.9 ± 18.8 ng/10⁶ cells).

This study is the first to demonstrate that mast cells localized within the ASM bundle in asthma express fibroblast markers. In vitro we determined that this mast cell fibroblastoid differentiation is ASM-derived ECM protein dependent. These fibroblastoid mast cells have an increased expression of chymase and are in a heightened state of activation, with increased constitutive release of histamine. Taken together, our findings suggest that ASM-derived ECM-mediated mast cell fibroblastoid transition is important in the development of airway dysfunction in asthma.

Our study provides further evidence to support the view that, in asthma, mast cell localization to the ASM bundle is important in the pathogenesis of AHR. Several studies have consistently reported increased mast cell numbers in the ASM bundle in asthma (7, 8, 13, 16, 17, 24, 25). Using stepwise logistic regression to explore the association between AHR and immunopathological features of asthma, we observed that mast cell infiltration of the ASM bundle was the strongest independent predictor of AHR (8). In this paper, we extend these earlier studies to demonstrate that the association with AHR is stronger for the number of “fibroblastic” mast cells than for the total mast cells within the ASM bundle, suggesting that this fibroblastoid-mast cell transition may play a pivotal role in mast cell-ASM interactions and the development of disordered airway function.

Consistent with our in vivo findings, we confirmed in vitro that mast cells have the capacity to undergo fibroblastoid differentiation. Primary HLMCs and human mast cell lines changed their morphology to spindle-shaped cells in the presence of the ECM proteins fibronectin or collagen I. These mast cells expressed proteins usually associated with fibroblasts, confirmed with two different Abs using a combination of techniques. Importantly, although the mast cells acquired fibroblast-like features and increased chymase expression, there was no downregulation in their CD117 or β-tryptase expression. This fibroblastoid transition was replicated in coculture with ASM and was associated with a change in cell size and increased constitutive histamine release, but not PGD₂. This fibroblastoid transition was associated with an increase in gel contraction, which may be a consequence of cell size and function. This finding was consistent with earlier observations determined by electron microscopy whereby mast cells within the ASM bundle were found to be smaller with reduced granule content, as a consequence of piecemeal degranulation (17).

We have reported previously that HLMCs and HMC-1 cells in vitro adhere to ASM, in part via CADM1 (14), and in coculture with ASM mast cells survive and proliferate (23). In contrast, we report in this paper that the mast cell transition toward a fibroblastoid phenotype was CADM1 independent. Indeed, this phenotypic transition was dependent on mast cell-ECM adhesion. One possible explanation for this apparent paradox is that CADM1 is important for the initial adhesion of mast cells to ASM and may initiate important cell signaling events, but that sustained adhesion of mast cells within the ASM bundle is maintained by mast cell interactions within a local ECM protein environment. This observation provides an explanation for the lack of effect we reported for blocking β-integrins on short-term adhesion for 30 min between ASM and mast cells, as well as providing a rationale for our inability to inhibit mast cell adhesion to ASM by CADM1 blockade after coculture for several days. Mast cell adhesion to fibronectin has been consistently reported (26–28), and in some circumstances, stem cell factor and IgE cross-linking promote mast cell adhesion to fibronectin (29, 30). We report in this paper that the development of fibroblastoid mast cells cultured with ECM proteins increased in a dose-dependent manner. This also occurred in coculture with ASM and was present, albeit to a substantially reduced extent, in the absence of direct contact with ASM when using Transwells and ASM supernatants and may have been mediated by soluble fibronectin. All of these effects were inhibited by α5β1 blockade, implicating mast cell-fibronectin interactions. Importantly, fibronectin is increased in the ASM bundle in asthma deaths (19) suggesting that this ECM protein is important in asthma, particularly in severe disease, and therefore has the capacity to mediate important mast cell-ECM interactions. Consistent with this view, previous reports have shown that rat mast cells cocultured with fibroblasts (31) or fibronectin result in increased histamine release (26). Similarly, human mast cells that adhere to fibronectin become more oblong, with pronounced formation of filopodia (27). This adhesion is attenuated by TLR3 activation, which consequently abrogates mast cell attachment-dependent potentiation of IgE-mediated responses (28). Similarly, reduction in mast cell β1 integrin surface expression as a consequence of diminished surface receptor stability in rabaptin-5 knockdown also leads to inhibition of mast cell adhesion, migration, and activation (32). Perhaps of most importance, adhesion to fibronectin does not increase IgE-dependent histamine release from HLMCs (33), supported by data reported in this paper, in contrast to findings in rodent mast cells (34). This suggests that ASM-driven HLMC mediator release, which is in part dependent on this matrix protein, drives the chronic mast cell activation evident within the ASM bundle in asthma in vivo (17). Taken together, these data suggest that ASM-derived ECM-mast cell interactions are critical for the transition of mast cells to a fibroblastoid phenotype and are likely to be central for the functional effects of this altered mast cell phenotype.

Our findings lead us to propose a new paradigm for the development of disordered airway physiology in asthma whereby the location of the mast cell within the ASM bundle leads to important reciprocal phenotypic changes in both cell types, which are perpetuated via positive feedback. We have reported that in coculture IgE-independent mast cell release of β-tryptase results in an upregulation of α-SMA intensity mediated via autocrine TGF-β release (35). Consequently, the ASM undergoes a phenotypic switch to a more contractile phenotype. This is supported by a very strong correlation in vivo between the number of mast cells within the ASM bundle and the ASM α-SMA intensity. We found that TGF-β did not influence the mast cell phenotype. However, TGF-β is an important inducer of ECM protein production by ASM, in particular, fibronectin (36, 37). This increased release of fibronectin will in turn promote ASM synthetic function (38, 39) and mast cell activation (26, 40), as well as transition to fibroblastoid phenotype. This altered phenotype has the capacity to further drive the ASM to a more contractile phenotype, therefore propagating the cycle to promote AHR (35).

One potential criticism of our findings is the cross-sectional design of the bronchoscopy study, which does not allow us to determine whether the strong relationship between the number of activated fibroblastoid mast cells in the ASM bundle and AHR is causal or an epiphenomenon. To confirm that fibroblastoid mast cells and their products promote airway dysfunction, bronchoscopic evaluation is required as part of interventional studies of therapies that attenuate AHR. In addition, to date there is paucity of robust animal models of asthma in which mast cell infiltration of the ASM bundle is a feature. Thus there is a need for improved models and new therapies to further study the importance of this mast cell fibroblastoid transition in asthma. Whether this transition is reversible and the fibroblastoid mast cell phenotype is plastic is also important and warrants further study.

In conclusion, we report in this paper the first evidence that mast cells develop a fibroblastoid phenotype mediated via ASM-derived ECM proteins, dependent on the α5β1 integrin. This novel mast cell phenotype demonstrates increased mediator release of histamine, and its presence in the ASM bundle is closely related to AHR. This presents a new paradigm for asthma and underlines the importance of mast cell-ASM interactions as a therapeutic target.

We thank B. Hargadon and S. McKenna for assistance in clinical characterization of subjects who underwent bronchoscopy.

Disclosures The authors have no financial conflicts of interest.