Integrin αvβ5-Mediated TGF-β Activation by Airway Smooth Muscle Cells in Asthma

Airway hyperresponsiveness, a key feature of asthma, is the enhanced contraction of the airway smooth muscle (ASM) layer in response to inhaled stimuli, which leads to variable airway obstruction. This phenomenon is responsible for the acute exacerbations associated with asthma. Recurrent exacerbations are a feature of severe asthma. Similarly, the structural changes of airway remodeling are also commonly associated with cases of severe asthma (1) and deteriorating lung function over time (2, 3). It has recently been shown that bronchoconstriction can induce features of airway remodeling in patients with mild asthma, including collagen deposition and goblet cell hyperplasia (4); however, the responsible mechanism is unknown.

TGF-β has been implicated in the pathogenesis of airway remodeling in asthma (5). TGF-β promotes airway remodeling, in part because of its effects on ASM cell proliferation, epithelial cell apoptosis, and potent profibrotic actions, including increasing synthesis of collagen and fibronectin (6–8). TGF-β promotes extracellular matrix deposition, ASM proliferation, and mucous production in an animal model of allergic asthma without affecting existing airway inflammation (9). Overexpression of Smad2, a TGF-β signaling protein, causes thickening of the ASM layer and deposition of collagen following allergen challenge (10). Moreover, the importance of TGF-β signaling in asthma pathogenesis is supported by a genome-wide association study demonstrating a link between a single nucleotide polymorphism in the SMAD3 gene and asthma (11).

TGF-β is secreted from cells in noncovalent association with its propeptide, the latency-associated peptide, which renders it inactive. Activation of latent TGF-β is the rate-limiting step in its bioavailability (12), and mechanisms of TGF-β activation are fundamental to disease. Several mechanisms of activation have been described in vitro, including proteolytic activation by plasmin, matrix metalloproteases and tryptase, physical activation by extremes of heat and oxidation, and activation by thrombospondin-1 (13–17). Several studies have described increased TGF-β activity in asthma (18–20). Activation of TGF-β in asthma can occur by several mechanisms. Epithelial cells can activate TGF-β in response to damage to the epithelial layer. Mast cells, which are present in large numbers in the asthmatic bronchial mucosa, can activate TGF-β proteolytically through the release of proteases from their granules (15, 21–23). In vivo integrins appear to have a major role in TGF-β activation, at least in development (24, 25), and recently fibroblast-specific deletion of the αvβ8 integrin has been shown to reduce airway remodeling by reducing TGF-β–induced CCL2 and CCL20 dependent dendritic cell migration (26). However, it is known whether smooth muscle cell TGF-β activation can directly contribute to airway remodeling.

Integrins are heterodimeric cell surface molecules involved in cell–cell interactions and cell–matrix interactions. Six of the 24 currently described integrins recognize and bind arginine-glycine-aspartate motifs in the latency-associated peptide of both TGF-β1 and TGF-β3. Four of these have been reported to activate TGF-β in vitro, including αvβ6 (27), αvβ8 (28), αvβ3 (29), and αvβ5 (30). Integrin-mediated TGF-β activation has been best characterized for the αvβ6 and the αvβ8 integrins. Activation of TGF-β by the αvβ8 integrin involves MMP14 and proteolytic cleavage of the latent TGF-β molecule, whereas αvβ3, αvβ5, and αvβ6 integrins activate TGF-β by a mechanism requiring an intact cytoskeleton and cell contraction (27, 30–32). Activation of TGF-β by αvβ6 integrins is spatially restricted to epithelial cells, whereas αvβ5, and to a lesser extent αvβ3, are present on mesenchymal cells and are able to activate mesenchymal TGF-β (30, 33, 34). This raises the possibility that cellular contraction during bronchoconstriction promotes TGF-β activation via cell surface integrins.

The aims of this study were to investigate whether contraction agonists could promote TGF-β activation in human ASM (HASM) cells, and determine whether this process was dysregulated in asthma. We found that lysophosphatidic acid (LPA) induced TGF-β activation by HASM cells via an integrin αvβ5-mediated mechanism that involved reorganization of the cytoskeleton. Furthermore, methacholine also induced TGF-β activation by HASM cells. HASM cells isolated from asthmatic patients activated more TGF-β in response to both contraction agonists than cells from nonasthmatic individuals. Furthermore, a polymorphism in the cytoplasmic domain of the integrin β5 (itgb5) gene resulted in a β5 subunit that was unable to activate TGF-β. Using the OVA model of airway remodeling we demonstrate coassociation of αvβ5 integrins and phospho-Smad2 staining in the ASM layer of remodeled airways, as well as global increases in TGF-β activation. Finally, using two distinct murine models of asthma, we show that both inhibition and genetic loss of the αvβ5 integrin results in significantly less ASM surrounding the airways despite enhanced inflammation. This finding suggests that the αvβ5 integrin has an important role in mediating airway remodeling, as well as supporting the notion that inflammation and remodeling can be dissociated in asthma. These data suggest a potential novel mechanism through which contraction of the ASM layer during asthma attacks could promote airway remodeling in patients with poorly controlled disease.

Tracheal and bronchial HASM cells were used. Tracheal HASM cells were obtained from postmortem tracheal specimens from donors with no history of respiratory disease or evidence of airway abnormalities as previously described (35). Bronchial HASM cells were derived from bronchial biopsy specimens and were supplied by Chris Brightling (University of Leicester, Leicester, U.K.). Asthmatic subjects recruited at Glenfield Hospital (University of Leicester) were carefully characterized and presented with an appropriate history, objective evidence of variable airflow obstruction, and airway hyperresponsiveness as described previously (36). The subjects were free from exacerbations requiring systemic corticosteroids or antibiotics for 6 wk prior to the bronchoscopy. Cells were cultured in DMEM containing 10% FCS, penicillin G (100 U/ml), streptomycin (100 μg/ml) and l-glutamine (4 mM). All cells were used at passage 6 and were growth arrested in serum-free medium for 24 h prior to experiments. CS-1 cells, a nonadherent hamster melanoma cell line, were a gift from Dean Sheppard (University of California, San Francisco). They were cultured in RPMI 1640 plus 10% FCS, l-glutamine (4 mM), penicillin (100 U/ml), streptomycin (100 μg/ml), and amphotericin B (2.5 μg/ml). Following transfection with integrin β5 constructs, the cells became adherent to plastic; they were then cultured as before with the addition of the antibiotic G418 at 500 μg/ml, because this was the concentration that killed 100% of nontransfected CS-1 cells after 1 wk of culture (data not shown).

Transformed mink lung epithelial cells (TMLCs) that stably express a portion of the plasminogen activator inhibitor-1 (PAI1) promoter driving a luciferase gene were used as reporter cells to detect TGF-β activity (32). HASM cells were plated at 2.5 × 10⁵ cells/ml and growth arrested for 24 h. TMLCs were plated directly on top of HASM cells at 5 × 10⁵ cells/ml. The cells were then stimulated as required for each experiment. After 16 h, the cells were lysed using a luciferase reporter assay kit (Promega), and the luminescence was measured as relative luciferase units (RLU).

Following stimulation total cell RNA was extracted using NucleoSpin RNA II kit (Macharey Nagel) and reverse transcribed into complimentary DNA (cDNA) using Moloney murine leukemia virus reverse transcriptase. cDNA was subjected to quantitative RT-PCR analysis to assess expression of the TGF-β–inducible gene PAI1 using β2-microglobulin (β2-M) as a housekeeping gene. Distinct primer sets were used to amplify murine DNA and hypoxanthine-guanine phosphoribosyltransferase (HPRT) was used as a housekeeping gene in these cases. Primer sequences were as follows:

• PAI1 sense: 5′-TCTGCAGACCTGGTTCCCAC-3′,

• PAI1 antisense: 5′-AGCCCCGTAGTTCCATCCTG-3′,

• β2-M sense: 5′-AATCCAAATGCGGCATCT-3′,

• β2-M antisense: 5′-GAGTATGCCTGCCGTGTG-3′,

• Fibronectin sense: 5′-CAGGATCAACTTACGGAG-3′,

• Fibronectin antisense: 5′-GCCAGTGACAGCATACA-3′,

• Collagen sense: 5′-CCAAATCTGTCTCCCCAGAA-3′,

• Collagen antisense: 5′-TCAAAAACGAAGGGGAGATG-3′,

• murine PAI1 sense: 5′-GGCCAAAGGAAAAAGCACTGT-3′,

• murine PAI1 antisense: 5′-GGGCCACCATTTGATCTGTCT-3′,

• murine HPRT sense: 5′-TGAAAGACTTGCTCGAGATGTCA-3′, and

• murine HPRT antisense: 5′-CCAGCAGGTCAGCAAAGAACT-3′.

Amplification was performed using an MXPro3000 (Stratagene) with SYBR Premier Ex Taq (Takara Bio) on the following program: initial denaturation at 95°C for 30 s followed by 40 cycles of 95°C for 5 s, 60°C for 30 s, and 72°C for 15 s. Amplification of a single DNA product was confirmed by melting curve analysis. Data were expressed as relative expression using the ΔΔCt equation as described previously (37).

Nuclear and cytoplasmic protein fractions were isolated from LPA-stimulated HASM cells as described previously (38, 39) using the NXtract CelLytic nuclear extraction kit (Sigma) according to the manufacturer’s protocol.

Levels of Smad2 and Smad3 in cytoplasmic and nuclear fractions were determined by Western blotting. Protein samples (30 μg per lane) were subjected to electrophoresis on a 10% SDS-polyacrylamide gel and electroblotted to a polyvinylidene fluoride membrane. After blocking for 1 h (5% nonfat milk in TBS plus 0.1% Tween 20), the membrane was incubated overnight at 4°C with monoclonal anti-Smad2/3 Ab (1:500 in blocking buffer; Cell Signaling Technologies). After washing, the membrane (PBS pH 7.4 plus 0.3% Tween-20) was incubated with HRP-conjugated secondary Ab (1:2000 in blocking buffer) for 1 h at room temperature. The membrane was incubated with ECL Western blotting detection reagent and visualized by exposure to Hyperfilm-ECL.

The Cell Titer-Glo Luminescent Cell Viability Assay (Promega) was used to determine cell number in coculture experiments comparing nonasthmatic and asthmatic HASM cells. Additional wells of cells were prepared and treated in the same manner as the experimental wells. After the incubation period of 16 h, 100 μl Cell Titer-Glo reagent was added to each well, and the plate was placed on a rocker for 10 min at room temperature. The luminescence was measured and data were converted into actual cell numbers using a standard curve of known cell numbers.

DNA plasmid constructs corresponding to the common full-length integrin β5 subunit and a polymorphic integrin β5 subunit in the vector pc.DNA3.1 were provided by Prof. E. Kawahara (Kanazawa University, Kanazawa, Japan). The transfection reagent Transfast (Promega) was used to stably transfect CS-1 hamster melanoma cells according to the manufacturer’s protocol.

Cell surface expression of the integrin αvβ5 was assessed by flow cytometry as described previously (32). Cells were counted, and 0.5 × 10⁶ cells were first blocked in goat serum for 20 min and then stained with an anti-αvβ5 Ab (Ab clone name ALULA raised in mouse; a gift from Prof. Dean Sheppard) at 25 μg/ml for 20 min. After washing the cells in PBS, the cells were stained with a PE-labeled anti-mouse secondary Ab for 20 min. Negative control cells were blocked with goat serum but only stained with the secondary PE-labeled Ab. Cells were analyzed using a BD FACSCanto flow cytometer.

The interaction of integrin β5 subunit with the cytoskeletal protein talin was investigated using a Universal Magnetic coimmunoprecipitation kit (Active Motif) according to the manufacturer’s instructions. A rabbit polyclonal Ab directed against integrin β5 and a mouse monoclonal directed against talin 1 and 2 (clone 8D4) were used (both from Abcam). Protein (500 μg) was immunoprecipitated with 5 μg Ab and protein G magnetic beads. After separation of proteins by SDS-PAGE, the membrane was probed with second Ab.

All animal care and procedures were approved by the University of Nottingham Ethical Review Committee and were performed under Home Office Project and Personal License authority within the Animal (Scientific Procedures) Act 1986. Female BALB/c 6-wk-old mice were purchased from Charles River (Margate, Kent, U.K.) and housed in the Biomedical Services Unit, University of Nottingham for at least 7 d after delivery. Animals had free access to food (Tekland Global 18% protein rodent diet; Bicester, Oxen, U.K.) and water. On day 0 of the study, mice were sensitized by i.p. injection of 10 μg OVA (Sigma) diluted 1:1 with the adjuvant Alum (Sigma), followed by a subsequent sensitization on day 12. At day 19, the mice were challenged daily by oropharyngeal administration of either 400 μg/ml OVA in 50μl saline or 50 μl saline alone for 6 d, followed by additional challenges on days 26, 28, 30, and 33. The mice were sacrificed on day 34. During the study of the inhibition of αvβ5, mice were sensitized as described and treated every 2 d from day 18 with 4 mg/kg anti-αvβ5 (clone ALULA) by i.p. injection until the end of the study. Bronchoalveolar lavage was performed using 1 ml PBS. Both frozen and formalin-fixed tissues were collected for immunohistochemistry. For formalin-fixed tissue, the left lobe was inflated with formalin and fixed in formalin overnight and then embedded in paraffin wax. For frozen tissue, the left lobe was inflated with 1 ml OCT and then frozen in precooled isopentane and stored at −80°C. The remaining four lobes were separated and snap frozen in liquid nitrogen until needed.

The University of California–San Francisco Committee on Animal Research (IACUC) approved the use of all mice for all reported experiments. The animals were allowed free access to food (Tekland Rodent Laboratory Chow) and water. β5 knockout mice (itgb5^−/−) were generated as described previously (40) and bred onto a 129 SvJae background. The murine model of Aspergillus fumigatus allergic lung disease was used with methods that have been described (41). Isoflurane-anesthetized mice were given 10 μg (in 40 μl saline) of A. fumogatus (Hollister-Stier Laboratories) or 40 μl saline to the nostrils using a micropipette with the mouse held in the supine position. After three treatments per week for a total of nine doses, mice were euthanized 48 h after the last intranasal challenge. Bronchoalveolar lavage was performed using three separate washes with 1 ml PBS. The separate washes were combined and centrifuged at 1500 rpm for 5 min, and the resulting cell pellet was resuspended in 1 ml PBS. Total inflammatory cells were counted using a hemocytometer. Formalin-fixed tissue was collected for immunohistochemistry. The left lobe was inflated with formalin, fixed in formalin overnight, and embedded in paraffin wax.

Sections of frozen murine lung tissue (3 μm thick) were used. After fixation in Zamboni’s fixative for 20 min, the sections were boiled in 2 mM citrate buffer for 3 min and then blocked with avidin and biotin blocking solutions (Vector Laboratories) each for 15 min in PBS. The tissue was then blocked in 5% goat serum plus 0.1% BSA in PBS for 30 min and incubated with an αvβ5 Ab (Abcam 15459, 1:200) overnight. After washing in PBS, the slides were incubated with a Dylight649 conjugated secondary Ab. The cells were then stained with a 1:100 dilution of an α-smooth muscle actin (SMA) Ab (clone 1A4) using a mouse-on-mouse Ab detection kit (Vector Laboratories) according to the manufacturer’s instructions. Staining was visualized using a laser-scanning confocal microscope and Nikon NIS Elements image analysis software. Images were acquired at a single plane through the tissue.

Sections of paraffin wax-embedded tissue (5 μm thick) were used. After dewaxing and rehydration, Ag retrieval was performed by boiling the sections for 20 min in citrate buffer. The sections were blocked in 5% normal goat serum plus 0.1% BSA. The sections were incubated with anti–Phospho-Smad2 overnight (1:200) and then a DyLight594 conjugated secondary Ab. The sections were then counterstained with an anti-mouse anti-αSMA Ab using a mouse-on-mouse fluorescein kit (Vector Laboratories) according to the manufacturer’s instructions. Staining was visualized using a laser-scanning confocal microscope and Nikon NIS Elements image analysis software. Images were acquired at a single plane through the tissue.

αSMA immunohistochemistry was performed on 5-μm thick sections of paraffin wax-embedded tissue. These sections were dewaxed in xylene, rehydrated in increasing concentrations of ethanol, and then washed in PBS. Ag retrieval was performed by boiling the sections for 20 min in citrate buffer. The sections were blocked to inhibit nonspecific binding of the primary Ab in 5% normal goat serum plus 0.1% BSA in PBS for 30 min. The sections were incubated with primary Ab overnight (1:1000). Sections were then incubated with a DyLight488-conjugated secondary Ab for 30, and the sections were counterstained with a DAPI nuclear stain. Negative control slides that were not stained with the primary Ab prior to incubation with the secondary Ab were performed for each experiment. αSMA quantification around smaller airways was performed using NIS Elements software (Nikon) measuring area in micrometers of αSMA staining. The definition of a small airway in mice is currently ambiguous; however, a previous study has defined a small airway as one with a diameter of 100–200 μm (42). This study used this definition. Only cross-sectional airways with a radius of less than 100 μm were quantified.

Snap-frozen lung lobes were defrosted in ice-cold lysis buffer (20 mM Tris-HCl, 1% Triton-X 100, 137 mM NaCl, 2 mM EDTA, 25 mM β-glycerophosphate, 10% glycerol, 1 mM Na3VO4, 1 mM phenylmethanesulfonyl fluoride, 10 μg/ml leupeptin, 1 mM DTT, and 0.1 U/ml protease inhibitor mixture). The lungs were homogenized using a hand-held homogenizer. Samples were centrifuged at 13,000 rpm for 10 min, and the supernatant was collected. The concentration of murine PAI1 in lung homogenates from OVA-sensitized, saline, and OVA-challenged mice was determined using a murine PAI1 total Ag sandwich ELISA (Patricell) according to the manufacturer’s instructions.

Statistical analysis was performed using GraphPad Prism 4 software. When data measured differences between groups, the mean values for pooled data were compared. When mechanistic pathways in normal cells were being assessed, the experiments were replicated at least three times, representative examples were shown, and experimental replicates were assessed statistically. Comparison of two data sets was performed by two-tailed unpaired t test. Comparison of more than two data sets was performed by two-way ANOVA; p < 0.05 was accepted as significant.

To determine whether contraction agonists activated TGF-β in HASM cells, these cells were stimulated with LPA and methacholine, and TGF-β activity was measured using three independent assays. Both LPA and methacholine induced concentration-dependent increases in active TGF-β as measured by coculture assay (Fig. 1A, 1B). There was no effect of either agonist on the reporter cells alone (data not shown). LPA also induced a time-dependent increase in PAI1 mRNA expression, which was maximal at 8 h (Fig. 1C). This increase was abrogated by a pan–TGF-β neutralizing Ab confirming that LPA induced PAI1 gene expression was mediated by TGF-β. Methacholine was also able to induce PAI1 mRNA in a TGF-β dependent manner (Fig. 1D). Finally, both LPA and methacholine stimulation induced the translocation of Smad2 and smad3 to the nucleus (Fig. 1E, 1F). These data confirm that the contraction agonists LPA and methacholine activated TGF-β in HASM cells.

To identify the mechanism of TGF-β activation in HASM cells, an αvβ5 blocking Ab was used. LPA- and methacholine-induced TGF-β activation was assessed by measuring PAI1 mRNA levels, and both were abrogated by the αvβ5 blocking Ab (Fig. 2A, 2B). Because cytoskeletal proteins are central to contraction-induced TGF-β activation (27, 30), the pharmacologic inhibitor of actin reorganization (cytochalasin D) was used. LPA-induced TGF-β activation was inhibited by cytochalasin D (Fig. 2C). To determine whether bronchodilation inhibited LPA-induced TGF-β activation, the β2-adrenoreceptor (β2) agonist formoterol was used. The addition of formoterol completely inhibited LPA-induced increases in PAI1 gene expression (Fig. 2D).

To establish whether cytoskeletal interactions with the β5 cytoplasmic domain were involved in mediating TGF-β activation, the effect of a previously described β5 polymorphism on LPA-induced TGF-β activation was investigated (43). This polymorphism involves a 9-bp deletion within the integrin β5 cytoplasmic domain, resulting in a β5 subunit that lacks amino acids FNK^767–769. Full-length DNA constructs corresponding to both the common allele of β5 (FNKFNK) and the polymorphic β5 (FNK; both were gifts from Prof. E. Kawahara) were stably transfected into CS-1 hamster melanoma cells, which do not express any endogenous ITGB3 or ITGB5 genes. After transfection, the cell surface expression of αvβ5 was assessed by flow cytometry. Both cell lines transfected with the common and polymorphic allele expressed αvβ5 integrin on the cell surface (Fig. 3A, 3B), although levels of expression were higher after transfection of the polymorphic allele. TGF-β activation was assessed in these cell lines by coculture assay. Coculture of cells expressing the common allele with TMLC reporter cells demonstrated that the common allele was able to activate TGF-β basally (Fig. 3C), as shown by a decrease in luciferase activity in the presence of a TGF-β neutralizing Ab. However, cells transfected with the polymorphic allele were unable to activate TGF-β basally (Fig. 3D).

The deletion of FNK^767–769 in the polymorphic β5 subunit occurs within an NxxY talin binding domain. To assess whether talin interactions with the β5 cytoplasmic domain were altered in cells expressing the polymorphic β5 subunit, coimmunoprecipitation studies were performed. Initially the β5 integrin subunit was immunoprecipitated from CS-1 cell lysates expressing either full-length or polymorphic β5 and immunoblotted for talin (Fig. 4A). To further investigate these interactions, talin was immunoprecipitated and immunoblots for β5 integrin subunit were performed (Fig. 4B). Full-length talin (220 kDa) bound to both common and polymorphic β5 subunits, whereas the talin rod domain (190 kDa) consistently bound only to the polymorphic allele of the β5 subunit.

Having demonstrated that the contraction agonists LPA and methacholine induced αvβ5 integrin-mediated TGF-β activation in HASM cells via the cytoskeleton, we wanted to determine whether there was any difference between HASM cells from patients with asthma compared with controls. When asthmatic HASM cells were stimulated with increasing concentrations of LPA, they activated more TGF-β compared with nonasthmatic cells (Fig. 5A; p < 0.005). LPA also induced TGF-β activation in a time-dependent manner as measured by PAI1 mRNA levels (p < 0.05), and asthmatic HASM cells elicited an exaggerated response to LPA stimulation at each time point tested (p < 0.05), which was a maximum of 6 h after stimulation (Fig. 5B). Furthermore, asthmatic HASM cells activated more TGF-β than did nonasthmatic cells in response to methacholine stimulation, as measured by both coculture assay (Fig. 5C) and by PAI1 quantitative RT-PCR (Fig. 5D). To determine whether this increased TGF-β activation resulted in increased expression of genes associated with airway remodeling we assessed expression of both fibronectin and collagen type 1A (COL1A) by quantitative RT-PCR in response to LPA stimulation. Asthmatic HASM cells expressed greater levels of fibronectin mRNA in response to LPA treatment than did nonasthmatic cells (Fig. 5E). Although there was a trend toward increased COL1A expression, no statistically significant differences between diseased and nondiseased HASM cells were detected (Fig. 5F). Finally, total TGF-β levels were measured and found to be 2-fold higher in unstimulated asthmatic HASM cells compared with nonasthmatic HASM cells (p < 0.005; Fig. 5G).

To determine whether the enhanced TGF-β activity observed in asthmatic HASM cells was mediated by the αvβ5 integrin, we stimulated HASM cells from asthmatic patients with LPA in the presence of a αvβ5 blocking Ab (Fig. 6A, 6B). The αvβ5 integrin blocking Ab completely abrogated LPA-induced increases in TGF-β activity, as assessed by coculture assay in asthmatic and nonasthmatic HASM cells (Fig. 6A, 6B). This finding indicated that the increased LPA-induced TGF-β activation observed in asthmatic HASM cells was mediated via the αvβ5 integrin. To determine whether the increased αvβ5 integrin-mediated TGF-β activation was due to higher levels of the αvβ5 integrin on the cell surface of asthmatic HASM cells, β5 integrin subunit cell surface expression was measured by flow cytometry and compared with nonasthmatic cells (Fig. 6C, 6D). There was no difference in the mean fluorescent intensity of cell surface β5 integrin between asthmatic HASM cells or nonasthmatic HASM cells.

To investigate the role of the αvβ5 integrin in vivo, we used the OVA model of airway remodeling. Histologic sections from the saline-challenged mice showed normal lung architecture, few inflammatory cells, and an intact epithelial layer (Fig. 7A). However, histologic sections from mice sensitized and challenged with OVA showed widespread airway inflammation and evidence of airway remodeling, such as a thickened ASM layer (Fig. 7B). The ASM layer stained positive for both αSMA and αvβ5 in saline- and OVA-treated animals, with αvβ5 integrins localized at the ends of actin fibers at cell-cell contacts and on the cell surface (Fig. 7C, 7D). When smooth muscle bundles from OVA-treated animals were visualized in cross section, it was apparent that ASM cells express αvβ5 integrin on the external surface of the actin fibers at the cell surface (Fig. 7E). To confirm that ASM cells can activate TGF-β, histologic sections were dual labeled with Abs against αSMA and P-Smad2. ASM cells, as demonstrated by αSMA expression, showed nuclear expression of P-Smad2 (Fig. 7F, arrow). Global levels of TGF-β activation in the lung were elevated in OVA-treated mice, as estimated by measuring levels of the TGF-β inducible protein PAI1 in murine lung homogenates by sandwich ELISA (Fig. 7G).

To assess the importance of αvβ5 integrin-mediated TGF-β activation on airway remodeling, an αvβ5 blocking Ab (ALULA) was administered to mice undergoing OVA-induced airway remodeling and the ASM layer around smaller airways was assessed. There was minimal staining for αSMA around the smaller airways of saline treated mice (Fig. 8A), and the αvβ5 blocking Ab had no effect in saline-treated animals (Fig. 8B). In sections from mice treated with OVA and an isotype control Ab, there was significant thickening of the ASM layer (Fig. 8C), and this was reduced in animals receiving an αvβ5 blocking Ab (Fig. 8D). The thickness of the ASM layer was quantified, confirming that mice treated with OVA plus an anti-αvβ5 Ab had significantly less αSMA staining around the airways compared with animals treated with OVA and an isotype control Ab (Fig. 8E). However, there appeared to be enhanced inflammation on histologic sections from mice with OVA-induced remodeling treated with the αvβ5 integrin blocking Ab, compared with mice treated with isotype control (Fig. 8G, 8H). Therefore, the inflammatory cell infiltrate was quantified on BAL fluid (BALF). OVA treatment increased the total number of inflammatory cells present in BALF (Fig. 8F), which appeared to be enhanced in animals treated with an anti-αvβ5 blocking Ab; however, these differences were not statistically significant. There was no difference in the differential cell counts between isotype control and αvβ5 Ab treated animals (data not shown).

To further elucidate the role of αvβ5 in the development of airway remodeling, we used an A. fumigatus model of allergic airway remodeling in asthma in itgb5^−/− mice. Both itgb5^−/− and control animals, following saline challenge, had minimal αSMA-positive cells surrounding their airways (Fig. 9A, 9B). Treatment of control mice with A. fumigatus caused thickening of the ASM layer (Fig. 9C), which was reduced in A. fumigatus exposed itgb5^−/− animals (Fig. 9D). This reduction in ASM thickness was statistically significant when αSMA staining was quantified (Fig. 9E). Similar to the observation in OVA exposed mice, A. fumigatus caused an increase in the total number of inflammatory cells present in control mice (p < 0.005), which was further enhanced in itgb5^−/− mice exposed to A. fumigatus (p < 0.05).

The aims of this study were to investigate whether contraction agonists could induce TGF-β activation in HASM cells and to explore the mechanism of activation and its relevance to asthmatic airway remodeling. We have shown that LPA and methacholine activate TGF-β in HASMs, and we have identified a novel mechanism of TGF-β activation in these cells involving the cytoskeleton, talin, and the αvβ5 integrin. We have shown that HASM cells isolated from asthmatic patients activated more TGF-β via this pathway than did cells isolated from nonasthmatic individuals. To our knowledge, this is the first time that this mechanism of increased TGF-β activation has been described in cells from patients with asthma. Furthermore, we show that murine ASM cells express the αvβ5 integrin in vivo and can activate TGF-β in an OVA model of asthmatic airway remodeling. Finally, we show that impaired αvβ5 integrin function in vivo, through both pharmacologic inhibition and gene deletion, results in reduced allergen-induced ASM thickness but increased airway inflammation. These data identify a key mechanism that dissociates inflammation from airway remodeling in asthma and supports the novel hypothesis that recurrent bronchoconstriction might lead to contraction-induced TGF-β activation by ASM cells promoting the development of asthmatic airway remodeling.

To our knowledge, this study is the first to describe αvβ5 integrin-mediated TGF-β activation in ASM cells. Activation of TGF-β by the αvβ5 integrin occurs in scleroderma fibroblasts promoting the transition of fibroblasts into fibrogenic myofibroblasts (44). Furthermore, myofibroblasts activate TGF-β from extracellular stores by transmitting contractile force via the αvβ5 integrin to latent TGF-β, which is inhibited by αvβ5 function-blocking Abs (30). Smooth muscle cells are contractile cells, and both LPA and methacholine augment ASM contraction via effects on rhoA and rho kinase (45–47). Our data demonstrate that contraction-induced TGF-β activation was inhibited by blocking the αvβ5 integrin, and LPA-induced TGF-β activation was inhibited by the inhibitor of actin assembly (i.e., cytochalasin D) and the β2 adrenergic receptor agonist formoterol. These data suggest that contraction agonists induced αvβ5 integrin-mediated TGF-β activation via smooth muscle cell contraction. It is possible that formoterol inhibited TGF-β activation via effects on the cAMP pathway independent of its bronchodilator effects, although the effects of cytochalasin D and observations in myofibroblasts (29) would suggest that cytoskeletal changes are central to αvβ5 integrin-mediated TGF-β activation.

Further evidence to support the role of the cytoskeleton in αvβ5 integrin mediated TGF-β activation comes from data obtained from experiments using constructs encoding wild type and a naturally occurring polymorphism in the β5 integrin subunit (43). This polymorphism leads to a 9-bp deletion within the NxxY talin binding region of the β5 subunit’s cytoplasmic domain and results in the loss of 3 aa (FNK^767–769) from the cytoplasmic tail. This polymorphism has not previously been shown to have any functional effect (43), but our data demonstrated that the polymorphism resulted in a β5 subunit that was unable to activate TGF-β. Talin is a cytoplasmic protein capable of binding to the cytoplasmic domains of β integrin subunits and is fundamental for inside-out activation of integrins (48). A role for talin and inside-out activation of integrins in integrin-mediated TGF-β activation has been suggested (49). The data presented in this study show that talin was still able to interact with the polymorphic β5 subunit; however, the rod domain interactions were aberrant. The functional role of the talin rod domain is largely unclear; however, this would suggest that the rod domain of talin could be central for mediating cytoskeletal forces that promote integrin mediated TGF-β activation.

To our knowledge, these data show for the first time that HASM cells from asthmatic patients activated more TGF-β than did cells from smooth muscle from nonasthmatic individuals in response to both LPA and methacholine stimulation. It is possible that the enhanced TGF-β activity is merely due to the enhanced total TGF-β secreted by HASM cells from asthmatic patients; however, we do not favor this possibility. The level of total TGF-β in HASM cells from nonasthmatics and asthmatics was in considerable excess of levels of active TGF-β following stimulation with LPA, as found in other cell systems (15, 32). The enhanced TGF-β activity detected in asthmatic cells was completely inhibited by blocking the αvβ5 integrin, suggesting that the augmented TGF-β activity observed in asthmatic cells was completely dependent on the αvβ5 integrin. However, no increase in cell surface expression of αvβ5 integrin was found in asthmatic HASM cells compared with nonasthmatic cells. Several studies have suggested that asthmatic HASM cells has intrinsic hypercontractility compared with nonasthmatic HASM cells (50–52). Future work will investigate whether hypercontractility of asthmatic HASM cells is responsible for the enhanced TGF-β activation in response to contraction agonists.

It has previously been shown in in vivo models that TGF-β activity is increased in asthma (20, 53), but the cell type responsible for the increased activation and the mechanism of activation have not been investigated. By assessing lung sections from a murine OVA model of asthma, we have shown that ASM cells express αvβ5 and can activate TGF-β in vivo, suggesting that ASM cells are an important source of TGF-β in asthma. A role for TGF-β activating integrins in asthmatic airway remodeling has been suggested because blocking all TGF-β binding integrin function using an RGDS (Arg-Gly-Asp-Ser) peptide can prevent airway remodeling in an OVA asthma model (54), and fibroblast-specific deletion of the itgb8 integrin also reduces OVA-induced airway remodeling (26). Whereas deletion of fibroblast itgb8 reduces TGF-β–induced dendritic cell migration and inflammation leading to reduced airway remodeling (26), our data provide evidence that blockade of smooth muscle cell αvβ5 integrins significantly reduces allergen-induced increases in ASM regardless of effects on inflammation. The finding that ASM thickness was not completely reduced to levels comparable with saline-treated animals implies that αvβ5-mediated TGF-β activation is not the sole mechanism of TGF-β activation promoting increased ASM cell mass. Although αvβ8 integrin-mediated TGF-β activation contributes to the development of asthmatic airway remodeling via effects on inflammation, its effects on ASM were not investigated (26). Furthermore, the mechanism of TGF-β activation by the αvβ8 integrin does not require cytoskeletal integrity, and thus will not be directly affected by bronchoconstriction.

Although it is clear that controlling inflammatory responses are important in asthmatic airway remodeling (26, 55, 56), there is also a positive correlation between airway responsiveness to methacholine and subepithelial fibrosis (57). More recently it has been shown that recurrent bronchoconstriction promotes airway remodeling independent of airway inflammation (4). Furthermore, the muscarinic receptor antagonist tiotropium, which inhibits methacholine-induced contraction, reduces allergen-induced increases in ASM mass (58). The mechanism responsible for bronchoconstriction-induced airway remodeling has not been determined, but our data suggest that bronchoconstriction results in αvβ5 integrin-induced TGF-β activation by HASM cells, which then drives structural changes associated with airway remodeling. In addition to reducing allergen-induced increases in ASM mass, blockade or loss of αvβ5 appeared to increase airway inflammation, at least in the Aspergillus model. The anti-inflammatory effects of TGF-β are well documented (27, 59, 60); therefore, it is possible that inhibiting TGF-β activity on immune and inflammatory cells may exacerbate the inflammation associated with allergen challenge as observed in our studies. Additional studies are required to investigate a possible role for αvβ5 in regulating airway inflammation.

These data demonstrate that contraction agonists can lead to TGF-β activation in smooth muscle cells via an αvβ5 integrin-mediated pathway in vitro, and that this mechanism is accentuated in asthmatic HASM cells. Furthermore, inhibiting the function of the αvβ5 integrin reduces ASM mass in vivo, without reducing inflammation. It is conceivable that this mechanism is responsible for the bronchoconstriction-induced airway remodeling observed by other groups (4, 57). Thus, while global inhibition of αvβ5 integrin-mediated TGF-β activation is unlikely to be a useful therapeutic strategy in asthma because of the effects on the inflammatory response, defining cell-specific pathways of αvβ5 integrin-mediated TGF-β activation, through careful in vitro cell biology, may aid the development of novel therapies for airway remodeling in patients with poorly controlled disease.

We thank Prof. E. Kawahara for providing the β5 constructs and Prof. Daniel Rifkin for supplying the TMLC reporter cells, Katherine Huang for assistance with the A. fumigatus in vivo model, Amha Atakilit for producing and supplying the αvβ5 blocking Ab for the in vivo studies, the nurses at Glenfield Hospital, Leicester and City Hospital, Nottingham for subject recruitment, and Ruth Saunders for isolation and characterization of primary HASM cells.

G.J. has a research agreement with GlaxoSmithKline to study pulmonary fibrosis and D.S. has a sponsored research agreement with Stromedix. The other authors have no financial conflicts of interest.