Release of biologically active TGF-β from airway smooth muscle cells induces autocrine synthesis of collagen

Amanda Coutts, Gang Chen, Newman Stephens, Stuart Hirst, Deborah Douglas, Thomas Eichholtz, Nasreen Khalil

Abstract

In severe or chronic asthma, there is an increase in airway smooth muscle cell (ASMC) mass as well as an increase in connective tissue proteins in the smooth muscle layer of airways. Transforming growth factor-β (TGF-β) exists in three isoforms in mammals and is a potent regulator of connective tissue protein synthesis. Using immunohistochemistry, we had previously demonstrated that ASMCs contain large quantities of TGF-β1–3. In this study, we demonstrate that bovine ASMC-derived TGF-β associates with the TGF-β latency binding protein-1 (LTBP-1) expressed by the same cells. The TGF-β associated with LTBP-1 localizes TGF-β extracellularly. Furthermore, plasmin, a serine protease, regulates the secretion of a biologically active form of TGF-β by ASMCs as well as the release of extracellular TGF-β. The biologically active TGF-β released by plasmin induces ASMCs to synthesize collagen I in an autocrine manner. The autocrine induction of collagen expression by ASMCs may contribute to the irreversible fibrosis and remodeling seen in the airways of some asthmatics.

  • plasmin
  • bovine
  • transforming growth factor-β latency binding protein-1
  • fibrosis
  • remodeling

the functional abnormality in asthma is characterized by obstruction to flow during expiration (1). The obstruction to flow may occur from increased airway reactivity and contraction of the smooth muscle cells surrounding the airways (15). In addition, airway obstruction occurs as a consequence of mucus plugs consisting of an inflammatory exudate and desquamated surface airway epithelial cells in the lumen (1). The wall of the airways is thickened by inflammatory cells such as eosinophils, lymphocytes, and mast cells as well as by mucosal edema and vasodilation (1, 15). Although these changes in the airways may be reversible, there are also more permanent changes that occur, characterized by subepithelial fibrosis due to collagens I, III, and V and fibronectin (32). In addition, airway smooth muscle cell (ASMC) hyperplasia and hypertrophy occur, and there is an increase in connective tissue proteins such as collagen, elastin, laminin, hyaluronan, versican, tenascin, and fibronectin around the smooth muscle cells (5, 6, 32).

On the basis of animal models of injury, it has been observed that there is recruitment and activation of inflammatory cells before fibrotic changes (7, 16, 34). Activated inflammatory cells and structural cells are induced to release a number of proinflammatory and fibrotic cytokines, including transforming growth factor-β (TGF-β) (7, 16, 18, 34, 39). TGF-β, a multifunctional protein, is one of the most potent regulators of inflammation and connective tissue synthesis (2). TGF-β exists in three isoforms in mammals: TGF-β1, TGF-β2, and TGF-β3 (2, 18,27). TGF-β1 appears to be the most common isoform associated with disorders characterized by inflammation and fibrosis (2) due to the observation that at sites of injury, TGF-β1 is released in large quantities by platelets (10) and inflammatory cells such as macrophages (16). TGF-β1 is a chemoattractant for inflammatory cells and fibroblasts as well as a mitogen to immature fibroblasts (7, 16). In addition, TGF-β induces fibroblasts to synthesize a variety of extracellular matrix proteins such as collagens, elastin, proteoglycans, and fibronectin (25). The role of TGF-β in airway fibrosis and remodeling is currently unclear. Minshall et al. (26) demonstrated that TGF-β1 mRNA and immunoreactivity were increased in the submucosal eosinophils of asthmatics but not of normal controls. In addition, Redington et al. (29) demonstrated an increase in the quantity of TGF-β1 in the bronchoalveolar lavage fluid from asthmatic patients compared with normal controls and that the levels of TGF-β were further increased in asthmatics after exposure to allergen. Together, these observations suggest that, in asthmatic airways, there is an increase and release of TGF-β1 that may be important in the pathogenesis of airway fibrosis and remodeling.

We had previously demonstrated that TGF-β1–3 (18,19) and TGF-β receptors are ubiquitously present in ASMCs (20). The intracellular presence of TGF-β would only be relevant if the ASMCs were able to secrete TGF-β, which, once released, could function in an autocrine or paracrine fashion. TGF-β is synthesized as a 100-kDa pro-TGF-β (27), and before secretion, the pro region, called the latency-associated peptide (LAP), is cleaved but remains noncovalently associated with TGF-β (27). When TGF-β is secreted in association with LAP as latent TGF-β (L-TGF-β), it cannot interact with its receptor and is biologically inactive (27). Because TGF-β and its receptors are so ubiquitously expressed, the most critical regulation of TGF-β action is the generation of a biologically active form of TGF-β by removal of the LAP (27). Additionally, in some instances, L-TGF-β1 is associated with the high-molecular-weight latent TGF-β binding protein-1 (LTBP-1) (35, 38). LTBP-1 bound to L-TGF-β1 targets TGF-β1 to the extracellular matrix (ECM), a process that serves as a reservoir of TGF-β1 (38). Plasmin is a serine protease and has been shown to release L-TGF-β1 from its association with LTBP-1 (38) and has been demonstrated to activate L-TGF-β1 by removal of the LAP (38). In this paper, we demonstrate that subconfluent primary cultures of ASMCs spontaneously release large quantities of biologically active TGF-β, which is regulated by plasmin. In addition, the TGF-β secreted by ASMCs is complexed with LTBP-1 generated by the same cells, which is also released by plasmin. Furthermore, the plasmin-mediated release of biologically active TGF-β can induce ASMCs to synthesize procollagen I in an autocrine manner. These findings demonstrate that ASMCs can generate biologically active TGF-β, which may contribute to the pathogenesis of airway fibrosis and remodeling.

MATERIALS AND METHODS

Materials.

α-Modified Eagle's medium (αMEM), fetal calf serum (FCS), insulin-transferrin-selenium, and other cell culture ingredients were from GIBCO BRL (Burlington, Ontario, Canada). [3H]thymidine was purchased from ICN (Irvine, CA). Anti-collagen type I (rabbit) was from Cedarlane Laboratories (Hornby, Ontario, Canada). Anti-α-smooth muscle actin (mouse), anti-fibronectin (mouse), α2-antiplasmin (α2-AP), and the plasmin substrateN-p-tosyl-Gly-Pro-Lys p-nitroanilide were from Sigma (St. Louis, MO). Recombinant human TGF-β3, porcine TGF-β1 and TGF-β2, and anti-LTBP-1 (mouse) were from R&D Systems (London, Ontario, Canada). Anti-pan-TGF-β1–3 (rabbit) was from Santa Cruz Biotechnology (Santa Cruz, CA).

Cell culture.

Bovine trachealis muscle was obtained from a local abattoir, B. J. Packers (Beausejour, Manitoba, Canada) or J. & L. Beef (Surrey, British Columbia, Canada), and the ASMCs were isolated as previously described (9). Briefly, after removal of the trachealis muscle, the tissue was minced and the slurry was suspended in a digestion buffer consisting of Hanks' balanced salt solution (GIBCO BRL), 600 U/ml collagenase (GIBCO BRL), 10 U/ml elastase (Sigma), and 2 U/ml protease (Sigma). The isolated cells were suspended in 10% FCS (GIBCO BRL) to neutralize the proteases, filtered through a 70-μm nylon mesh (Biodesign, Carmel, NY), and washed in αMEM with 10% FCS and antibiotic-antimycotic solution (GIBCO BRL). Cell numbers were determined using a hemocytometer, and cells were cultured in 100-mm dishes at 250–500,000 cells/dish. Experiments were performed when the cells were either subconfluent (60–70%) or confluent. For confluent experiments, the cells were grown in serum-free αMEM containing antibiotics plus 2 ng/ml insulin, 1.34 ng/ml selenium, and 1.1 ng/ml transferrin. In some experiments, confluent monolayers were cultured in the presence of TGF-β1, -2, or -3, anti-pan-TGF-β antibody, or plasmin in the absence or presence of anti-pan-TGF-β 1–3 antibody, aprotinin, or α2-AP. In experiments requiring in vitro wounding, the confluent ASMC layers were scratched with a rubber policeman. In experiments with in vitro wounds, monolayers with 24 wounds were cultured in the absence and presence of α2-AP, aprotinin, or anti-pan-TGF-β antibodies. In some experiments, the subconfluent cells were cultured in the absence and presence of the plasmin inhibitor α2-AP or aprotinin. The smooth muscle identity of the cells was confirmed immunohistochemically by their expression of smooth muscle α-actin (data not shown and Ref.9).

Collection of conditioned medium.

At various times after incubation of cells, the overlying cell-conditioned medium (CM) was removed, stored in sterile siliconized Eppendorf tubes in the presence of the protease inhibitors (leupeptin 1 mg/ml, Amersham; aprotinin 1 mg/ml and pepstatin 1 mg/ml, both from Sigma), and frozen at −80°C until used. In those instances where CM was used to detect plasmin activity, the CM was collected without the addition of protease inhibitors (17). In some instances, the cells remaining after removal of CM were used for protein extraction and Western analysis.

CCL-64 mink lung epithelial growth inhibition assay for TGF-β.

The CCL-64 growth inhibitor assay that was used to quantitate TGF-β has been extensively described by us (16-19). Briefly, subconfluent cells maintained in TGF-β-free 0.2% bovine calf plasma, αMEM, 10 mM HEPES (pH 7.4), penicillin (25 mg/ml), and streptomycin (25 mg/ml) and cultured at 4.5 × 104cells/0.5 ml in 24-well Costar plates (Flow Laboratories, Mississauga, Ontario, Canada) in neutral CM or CM that was acidified and subsequently neutralized were added. After 22 h, the cells were pulsed with 0.2 μCi of [3H]thymidine for 3 h, at 37°C and lysed with 0.5 ml of 1 N NaOH for 30 min at room temperature, and [3H]thymidine incorporation was quantitated using liquid scintillation counting techniques. A standard curve using porcine TGF-β1 was included in each assay. In those instances where the CM demonstrated quantities of TGF-β that were in excess of the limits of detection by the CCL-64 assay, the sample was diluted 1:4, 1:10, etc., before the TGF-β in the sample was quantitated.

TGF-β immunoassays.

Quantikine TGF-β1 and TGF-β2 immunoassay kits (R&D Systems, Minneapolis, MN) were used to identify the presence of TGF-β1 or TGF-β2 in acidified CM from ASMCs according to the manufacturer's instructions.

Western blotting and immune detection.

Whole cell protein extracts were obtained by removing cells into 1.5-ml Eppendorf tubes and lysing cells on ice with triple-detergent lysis buffer [50 mM Tris · HCl, pH 8.0, 0.15 M NaCl, 1% (vol/vol) Triton X-100, 0.1% (wt/vol) SDS, 5 mg/ml sodium deoxycholate, and 1 mM phenylmethylsulfonyl fluoride]. Protein concentration was calculated using the Bradford assay (Bio-Rad, Hercules, CA). Protein extracts were run on SDS-PAGE according to the method of Laemmli (22) at 200 V for 45 min at room temperature. Molecular size was determined by running prestained molecular markers (Amersham, Buckinghamshire, UK). Gels were transferred to nitrocelullose membrane (Bio-Rad) using CAPS transfer buffer [25 mM 3-(cyclohexylamino)-1-propanesulfonic acid, pH 10, and 20% (vol/vol) methanol] for 1 h at 120 V at 4°C. Blots were blocked for 1 h at room temperature in Tris-buffered saline containing 0.2% Tween 20 (TBS-T) and 5% skim milk (wt/vol). Blots were incubated overnight at 4°C with primary antibody (1 μg/ml) in 1% skim milk (wt/vol) TBS-T and incubated with the appropriate secondary antibody for 1 h at room temperature [1:2,000 in 1% skim milk (wt/vol) TBS-T]. Blots were washed three times for 10 min in TBS-T, and antibody detection was carried out using the ECL system (Amersham) according to the manufacturer's instructions. Relative absorbance was determined using the Quantity I imaging system (Bio-Rad).

Immunoprecipitations.

Bovine tracheal smooth muscle pieces were microdissected from whole tracheae and stored at −80°C before protein extraction. Pieces (∼1 g) were pulverized while frozen and homogenized in triple-detergent lysis buffer on ice. Samples were further sonicated on ice, and 500–1,000 μg were used for immunoprecipitation. Samples were precleared by incubation with 1 μg of rabbit IgG and 10 μl of protein A/G agarose for 30 min at 4°C with rocking. After a brief centrifugation at 12,000 rpm, the supernatant was removed to a fresh tube and incubated with 1–2 μg of anti-TGF-β1–3 (Santa Cruz) for 1 h at 4°C with rocking. Protein A/G agarose (20 μl) was added and incubated for 1 h at 4°C with rocking. Immune complexes were pelleted by centrifugation for 10 min at 12,000 rpm, and the pellets were washed three times with RIPA buffer [50 mM Tris · HCl (pH 7.5), 150 mM NaCl, 1% (vol/vol) Nonidet P-40, 0.5% (wt/vol) sodium deoxycholate, and 0.1% (wt/vol) SDS]. Pellets were resuspended in Laemmli sample buffer (nonreducing) and boiled before being run on SDS-PAGE.

Plasmin assay.

Plasmin was quantitated by measuring the increase in absorbance at 405 nm due to cleavage of the plasmin-specific chromogenic substrateN-p-tosyl-Gly-Pro-Lys p-nitroanilide (17). The analysis of all experimental and standard curve samples was performed in 96-well flat-bottom microtiter plates (Flow Laboratories) using a Titertek Multiskan MCC/340 spectrophotometer (Flow Laboratories). The standard curve was made by incubating 2 mM ofN-p-tosyl-Gly-Pro-Lys p-nitroanilide in the presence of a range of concentrations of plasmin (Sigma; from 1 × 10−4 to 1 × 10−2 U/ml). To measure plasmin generated by ASMCs, 100 μl of either ASMC-derived CM or serum-free αMEM control medium were incubated with 100 μl of the plasmin substrate. The samples were incubated for 5 h at 37°C. Absorbance was measured, and the background absorbance from serum-free αMEM was subtracted from all experimental and standard curve samples. The quantity of plasmin present was then derived from the values obtained in the standard curve and presented as units of plasmin activity per milliliter of medium.

Statistical analysis.

Statistical analysis of TGF-β in CM of subconfluent and confluent ASMCs was performed using the Graphpad InStat software program (San Diego, CA) or ANOVA. The statistical analysis of the relative absorbance of the Western blots was done using the Mann-Whitney test for independent samples.

RESULTS

Release of TGF-β from cultures of ASMCs.

To determine whether primary cultures of ASMCs release TGF-β, the serum-free CM from subconfluent and confluent cultures was examined using the CCL-64 bioassay (16-19). Neutral CM from subconfluent ASMCs, representing TGF-β in an already active form, contained large quantities of TGF-β activity (Fig.1 A). In addition, total TGF-β, representing the combination of TGF-β in its biologically active and latent forms, was also present in large quantities (Fig. 1 A). Under these conditions, 61.5% of the TGF-β released by subconfluent ASMCs was in an active form. However, ASMC cultures that had just reached confluence for 1 day contained less active and total TGF-β compared with subconfluent cultures (Fig.1 A). After 15 days of confluence, markedly less active TGF-β was detectable, and the total TGF-β detected in the CM was also decreased (Fig. 1 A). After 15 days of confluence, the percent of active TGF-β was 2.2%. With the use of an immunoassay to identify the isoforms of TGF-β present, CM from both subconfluent and confluent ASMCs secreted TGF-β1 almost exclusively (Fig.1 B).

Fig. 1.

Release of transforming growth factor-β (TGF-β) by airway smooth muscle cells (ASMCs) in culture. A: TGF-β activity was measured in serum-free conditioned medium with (CM) from subconfluent and confluent ASMCs with the CCL-64 bioassay. Measurements were made using neutral CM, representing TGF-β in an active form, and acidified CM, representing total (both active and latent) TGF-β activity. Results are expressed as fmol TGF-β/106 cells and represent means ± SE from 2–5 independent experiments. *P < 0.01 for the quantity of total TGF-β present in confluent monolayers compared with subconfluent monolayers after 1 day of confluence. P < 0.001 for the quantity of active TGF-β present in the CM from ASMCs that were confluent for 15 days compared with subconfluent cells.# P < 0.003 for the quantity of total TGF-β present in the CM from ASMCs that were confluent for 15 days compared with subconfluent cells (paired Student's t-test).B: isoforms of TGF-β in CM from subconfluent (1 day no serum) and confluent (15 days no serum) ASMC cultures. Results are expressed as fmol TGF-β/106 cells and represent means ± SE of 4 experiments. *P < 0.001 for the quantity of TGF-β1 present in CM from confluent compared with subconfluent cultures of ASMCs (Student's t-test).

We and others have demonstrated that plasmin, a serine protease, is important for the activation of L-TGF-β1 (14, 17, 27). CM from subconfluent ASMCs contained increased quantities of plasmin activity compared with CM from confluent cells where no plasmin could be detected (Fig. 2 A). In addition, when ASMCs were cultured in the presence of aprotinin, an inhibitor of plasmin activity (14), there was a marked reduction in plasmin activity detected in the CM (Fig. 2 A). In the presence of the same concentration of aprotinin, the generation of active TGF-β by subconfluent ASMCs was completely abrogated, whereas the quantity of total TGF-β secreted was not affected (Fig.2 B). Although aprotinin may inhibit other serine proteases (41), we had previously confirmed that aprotinin inhibits the conversion of L-TGF-β to active TGF-β, equivalent to α2-AP, which is a specific inhibitor of plasmin activity (17, 41). Furthermore, when ASMCs were cultured in the presence of α2-AP, the results were comparable to when aprotinin was used (see Fig. 4 A). The presence of aprotinin or α2-AP totally abrogated the activation of L-TGF-β (Fig. 2, A and B) (17, 41). These findings suggest that it is unlikely that other proteases such as calpain have a significant role in the conversion of L-TGF-β to active TGF-β. The observations confirm that ASMC-derived plasmin activity is important in the generation of active TGF-β by ASMCs. When confluent ASMCs were cultured in the presence of plasmin, increased quantities of active TGF-β were released into the CM (Fig.3 A). Furthermore, the release of TGF-β by plasmin could be totally abrogated by the presence of aprotinin (Fig. 3 A). Although ASMCs spontaneously secrete only the TGF-β1 isoform, the addition of plasmin to confluent ASMCs released both TGF-β1 and TGF-β2 from ASMCs (Fig. 3 B).

Fig. 2.

Plasmin activity in cultures of ASMCs. A: plasmin activity in CM from subconfluent and confluent ASMCs grown under serum-free conditions. Subconfluent ASMCs were cultured under serum-free conditions and incubated with and without aprotinin (100 μg/ml) for 24 h. Results are expressed as units of plasmin (× 10−3)/ml CM and represent means ± SE of 3 independent experiments. Plasmin activity in CM of ASMCs confluent for 1 day compared with subconfluent cells was reduced, P ≤ 0.001. Plasmin activity in CM confluent for 15 days compared with subconfluent conditions was reduced, P ≤ 0.0001.B: aprotinin reduced TGF-β activity in subconfluent ASMCs. TGF-β activity (active and total) was measured in CM from subconfluent ASMCs under serum-free conditions. Subconfluent ASMCs were treated with and without 100 μg/ml aprotinin for 24 h before collection of CM. Active TGF-β in CM after the addition of aprotinin reduced TGF-β, P ≤ 0.0001. Total TGF-β in CM was not affected. Results are expressed as fmol TGF-β/106 cells and represent means ± SE of 3 independent experiments.

Fig. 3.

Effects of plasmin on TGF-β activity in CM from confluent ASMCs. A: active TGF-β was measured in neutral CM from confluent ASMCs grown for 24 h in the absence of serum for 12 days with or without plasmin (0.002 U/ml) or plasmin 0.002 U/ml plus aprotinin 100 μg/ml. Results are expressed as fmol TGF-β/106 cells and represent means ± SE of 3 independent experiments. TGF-β increased in CM when plasmin was added compared with when no plasmin was present, P < 0.05. TGF-β was decreased when aprotinin and plasmin were present compared with the absence or presence of plasmin with a P < 0.0001. B: immunoassays were performed on acidified CM samples from confluent ASMC cultures grown in the absence of serum for 12 days and incubated in the absence and presence of plasmin (0.002 U/5 ml CM) for 24 h. Results are expressed as fmol TGF-β1 or TGF-β2/106 cells and represent means ± SE of 3 independent experiments.

A significant difference between confluent and subconfluent cultures is that, at confluence, the ASMCs are closely associated with each other. Because ASMCs that were confluent generated little plasmin and no active TGF-β compared with subconfluent monolayers (Figs. 1 and 2), it suggests that the loss of a close association of these cells with one another may result in release of active TGF-β. This possibility was confirmed when CM from confluent monolayers of ASMCs that were mechanically wounded with 18–24 scratches contained active TGF-β (Fig. 4 A). However, CM from confluent monolayers with no in vitro scratches or <18 scratches contained less active TGF-β (Fig. 4 A). Plasmin assays performed on CM after wounding were unable to detect increased plasmin activity, which may have been due to the limited sensitivity of the method. However, since the presence of aprotinin or α2-AP in cultures of wounded monolayers inhibited TGF-β activity, it suggests that the activation of L-TGF-β in this context is mediated by plasmin (Fig. 4 A). The TGF-β present in CM when there were 24 scratches and no additional reagents was 42.3 ± 11.7 fmol/106 cells, but the addition of anti-TGF-β1–3 antibodies decreased TGF-β to 8.3 ± 3.3 fmol/106cells (P < 0.008). There was no statistical difference between the TGF-β present in CM in the absence of scratches compared with the presence of 24 scratches and anti-TGF-β1–3 antibodies. The generation of total TGF-β remained unchanged. Immunoassay demonstrated both TGF-β1 and TGF-β2 to be present in CM after in vitro wounding of the monolayers (Fig. 4 B).

Fig. 4.

Wounding releases active TGF-β into CM from confluent ASMCs. A: TGF-β activity (active and total) was measured in CM from confluent ASMCs. Cells were grown in the absence of serum for 12 days before mechanical wounding of the cell monolayers with a sterile rubber policeman in the absence and presence of 100 μg/ ml aprotinin or 100 nM α2-antiplasmin (α2-AP) 24 h before the CM was obtained. Compared with untreated controls, the presence of 18 or 24 scratches increased TGF-β in CM, *P ≤ 0.01. In the presence of 24 scratches the addition of aprotinin or α2-AP decreased TGF-β in CM with a P ≤ 0.005. There was no significant difference in the quantity of active TGF-β in CM in the presence of 24 scratches and aprotinin or 24 scratches and plasmin. B: immunoassay was performed on acidified CM samples from confluent ASMC cultures. Cells were grown in the absence of serum for 12–15 days before mechanical wounding with a sterile rubber policeman and are expressed as fmol TGF-β1 or TGF-β2/106 cells and represent means ± SE of 6–8 experiments.

Induction of collagen synthesis by ASMCs is regulated by TGF-β.

Although many connective tissue proteins may be regulated by TGF-β (2, 4) the experiments in this study were designed to detect procollagen I as an index of connective tissue synthesis. Subconfluent ASMCs constitutively express large quantities of procollagen α1(I) and α2(I), which, in the presence of a pan-TGF-β1–3 neutralizing antibody, was markedly decreased (Fig. 5 A). These findings demonstrate that procollagen I synthesis by ASMCs is regulated by TGF-β in an autocrine fashion. Furthermore, confluent compared with subconfluent (Fig. 5 A) monolayers of ASMCs constitutively express small quantities of procollagen I. However, confluent ASMCs can be induced to synthesize increased quantities of procollagen I in a dose-dependent manner in the presence of purified TGF-β1, TGF-β2, and TGF-β3 (Fig. 5 B). It is of note that the representative blot selected for Fig. 5 Bdemonstrates that the expression of collagen I in the presence of 0.5 ng/ml of TGF-β2 is decreased. However, when the relative absorbance using densitometry was done on all the samples, there was a significant induction of collagen I expression in the presence of 0.5 ng/ml of TGF-β2.

Fig. 5.

Procollagen I expression by ASMCs. ASMCs were incubated with and without 10 μg/ml anti-pan-TGF-β1–3 (a-pan-TGF-β) neutralizing antibody or TGF-β isoforms for 24 h before harvest. Whole cell protein extracts were obtained and run on 10% SDS-PAGE under reducing conditions before transfer to nitrocellulose for immune detection. Arrows on the right denote positions of procollagen α1(I) and α2(I) bands detected using anticollagen I antibody. A: procollagen I expression by subconfluent ASMCs. The blot is representative of 4 independent experiments. Control represents conditions when no anti-pan-TGF-β1–3 antibodies (a-pan-TGF-β) are present. Compared with untreated controls, the relative absorbance decreased in the presence of a-pan-TGF-β antibody, P < 0.02.B: regulation of procollagen I expression in confluent ASMCs by TGF-β isoforms. Confluent ASMCs were grown under serum-free conditions for 12 days before treatment for 24 h with porcine TGF-β1, porcine TGF-β2, or recombinant human TGF-β3 at several concentrations. The blot is representative of 6 independent experiments. C, an untreated control. Compared with untreated controls, the relative absorbance increased in the presence of TGF-β isoforms, *P values between 0.01 and 0.05. The numbers on they-axis in the histograms denote relative absorbance by densitometry.

Confluent ASMC cultures incubated with quantities of plasmin sufficient to increase active TGF-β (Fig. 3 A) also resulted in an increase in procollagen I expression (Fig.6 A). Because the induction of procollagen I by plasmin could be inhibited by an anti-pan-TGF-β1–3 antibody (Fig. 6 A), it suggests that the induction of collagen synthesis when plasmin is added to ASMCs is due to release of active TGF-β. The findings also demonstrate that the release of TGF-β by plasmin induces collagen synthesis in an autocrine manner. It is of interest that although 0.002 units of plasmin resulted in induction of collagen I expression, there was a reduction of collagen I expression to baseline values in the presence of α2-AP (P ≤ 0.05; data not shown). These findings confirm that the release of TGF-β by plasmin induces collagen I synthesis. In addition, when confluent monolayers of ASMCs were mechanically wounded, conditions previously demonstrated to release active TGF-β, there was an increase in procollagen I expression (Fig. 6 B).

Fig. 6.

Procollagen I expression by ASMCs induced by plasmin and in vitro wounding. A: plasmin induces procollagen I expression. Confluent ASMCs were grown under serum-free conditions for 12 days before treatment for 24 h with plasmin at a number of concentrations (0.00–0.002 U/ml) in the absence and presence of a-pan-TGF-β. Whole cell protein extracts were obtained and run on 10% SDS-PAGE under reducing conditions followed by transfer to nitrocellulose for immune detection. Arrows on the rightdenote positions of procollagen α1(I) and α2(I) bands detected with anticollagen I antibody. The blot is representative of 4 independent experiments. Compared with untreated controls, the relative absorbance was increased, *P ≤ 0.001–0.01. In experimental conditions when 0.002 U/ml of only plasmin was present compared with 0.002 U/ml of plasmin plus a-pan-TGF-β antibody, the relative absorbance is decreased, P ≤ 0.002. B: wounding of confluent ASMCs increases procollagen I expression. Confluent ASMCs were grown under serum-free conditions for 12 days before mechanical wounding with a sterile rubber policeman 24 h before harvest. Whole cell protein extracts were treated as described above. Arrows on the right denote positions of procollagen α1(I) and α2(I) bands detected with anti-collagen I antibody. The blot is representative of 4 independent experiments. Compared with untreated controls, the relative absorbance was increased. *P ≤ 0.01 when 18 scratches were present. The numbers on the y-axis of the histograms denote relative absorbance by densitometry.

Association of TGF-β with LTBP-1.

Despite the release of large quantities of TGF-β by subconfluent ASMCs, the CM from the same cells at confluence contained no TGF-β (Fig. 1 A). TGF-β in association with its LAP can be complexed with LTBP-1 generated by fibroblasts (38). If this were to occur in cultures of ASMCs, then the TGF-β in CM would be diminished but localized on ASMCs by its association to LTBP-1. LTBP-1 was expressed equally by subconfluent and confluent ASMCs (data not shown). However, the CM of subconfluent cells compared with confluent cells contained large quantities of LTBP-1 (Fig.7 A). It is then possible that the TGF-β released by subconfluent ASMCs associates with the LTBP-1 present in the CM and that the LTBP-1/L-TGF-β complex then interacts with the ASMCs or the ECM generated by the ASMCs. It has previously been demonstrated that plasmin can release LTBP-1 from its association with ECM (27, 38). Because confluent ASMCs cultured in the presence of plasmin resulted in a dramatic increase in LTBP-1 immunoreactivity in the CM, these findings confirmed that LTBP-1 that is extracellular to ASMCs was present and could be released by the actions of plasmin (Fig. 7 B). Further confirmation that plasmin releases LTBP-1 from ASMCs was obtained when a marked reduction of LTBP-1 in CM was observed in the presence of aprotinin (Fig.7 B). Because plasmin also resulted in increased TGF-β activity in the CM (Fig. 3 A), these findings suggest that the plasmin-mediated release of TGF-β may be due to its actions on LTBP-1 and L-TGF-β associated with LTBP-1. To determine whether TGF-β1 from ASMCs is associated with LTBP-1, protein extracts from ASMCs (data not shown) or the trachealis muscle were immunoprecipitated with anti-TGF-β1–3 antibodies, and the presence of LTBP-1 was detected by using Western analysis (Fig. 7 C). Lane 1, which contains A/G-associated proteins, did not demonstrate any immunoreactivity with LTBP-1 antibody (Fig. 7 C). However, the proteins obtained by immunoprecipitation with TGF-β1–3 antibodies contained detectable LTBP-1 (lane 2, Fig. 4). These findings then demonstrate that TGF-β is associated with LTBP-1 expressed by ASMCs obtained directly from the airways.

Fig. 7.

Expression of TGF-β latency binding protein-1 (LTBP-1) by ASMCs and the association of latent TGF-β with LTBP-1. A: CM samples (1 ml) from subconfluent and confluent ASMC cultures were lyophilized to concentrate the media to 200 μl before being run on 10% SDS-PAGE under nonreducing conditions before transfer to nitrocellulose for immune detection using anti-LTBP-1 antibody. In CM from confluent monolayers compared with subconfluent conditions, the relative absorbance was decreased with a value of P < 0.05. B: plasmin increases LTBP-1 in the CM of confluent ASMCs. ASMCs were grown under serum-free conditions for 12 days before incubation with plasmin (0.002 U/5 ml) or plasmin (0.002 U/ ml) plus aprotinin (100 μg/ ml) for 24 h before collection of serum-free CM and immune detection of LTBP-1. Compared with untreated control, the relative absorbance was increased with a value of P ≤ 0.001 when plasmin was present, but P < 0.01 when plasmin and aprotinin were present. When plasmin and aprotinin were present, the relative absorbance decreased with a value ofP < 0.001 compared with when plasmin alone was present. C: TGF-β and LTBP-1 interact with bovine tracheal smooth muscle. Immunoprecipitations performed using protein obtained from bovine tracheal smooth muscle were run on 10% SDS-PAGE under nonreducing conditions, and immune detection was done using anti-LTBP-1 antibody. In lane 1, immunoprecipitation was done using protein A/G agarose alone, while for lane 2immunoprecipitation was done using anti-TGF-β1–3 before immunodetection using anti-LTBP-1. A–C: positions of prestained molecular mass markers are shown on the left.

DISCUSSION

Our findings demonstrate for the first time that the serine protease plasmin is generated by ASMCs, where it functions to release a biologically active form of not only TGF-β1 but also TGF-β2 from ASMCs. Furthermore, once released, the presence of TGF-β1 and -β2 induces the same cells to synthesize procollagen I in an autocrine fashion. Last, it is of interest that LTBP-1 has consistently been demonstrated to be associated with the extracellular matrix (37,38). However, this is the first observation demonstrating that not only can LTBP-1 be present in solution but there may also be a potential role for LTBP-1 in the smooth muscle layer of airways. This is also the first observation to demonstrate that wounding of confluent ASMCs results in plasmin-mediated release of active TGF-β1, which then leads to an autocrine induction of collagen I synthesis.

With rare exception, the TGF-β isoforms are secreted by cells in a biologically latent form (16, 17, 27). The most important mechanism in the regulation of TGF-β activity is dependent on the conversion of L-TGF-β to its active form (27). Several mechanisms of activation of L-TGF-β have been described (27), but ASMC-derived L-TGF-β is activated by plasmin. Plasmin is cleaved from the proenzyme plasminogen by the actions of two plasminogen activators (PA), tissue-type PA (tPA) and urokinase-type PA (uPA), which, in turn, are controlled by plasminogen activator inhibitors (PAIs) (24). Currently, there are no reports demonstrating the regulation of the plasmin/plasminogen system in ASMCs, but there are a few reports describing the role of this system in vascular smooth muscle cells (VSMCs). Although the mechanism is not known, in models of VSMC injury, there is an increase in the expression of tPA, uPA, and PAI-1, leading to pericellular activation of plasmin (30). In this study, the expression of all the members of the plasminogen/plasmin system in ASMCs has not been described, but it has been demonstrated that plasmin activity is present in CM from subconfluent cultures of ASMCs. The presence of plasmin in these conditions is associated with large quantities of active TGF-β1. Because aprotinin, an inhibitor of plasmin activity (24), totally abrogates the generation of active TGF-β without affecting the total TGF-β secreted by the same cells, these findings confirm the importance of plasmin-mediated posttranslational activation of L-TGF-β from ASMCs. The presence of plasmin when ASMCs are subconfluent results in the release of primarily the TGF-β1 isoform. However, when ASMCs are confluent, the addition of plasmin results in the release of both TGF-β1 and TGF-β2. Previously, the actions of plasmin have been described as being restricted to the release of TGF-β1 (37, 38). The current findings suggest that in addition to the release of TGF-β1, plasmin may be important in the release of biologically active TGF-β2 that may be associated with the ASMCs extracellularly. Demonstrating that plasmin is critical to the activation of L-TGF-β1 from ASMC is complementary to the observation that plasmin activates L-TGF-β1 from alveolar macrophages (17,21), endothelial cells (27), and VSMCs (8,23, 30). Collectively, these observations support plasmin as an important physiological substance in the activation of L-TGF-β1 by several different phenotypes of cells. Because the in vitro model is composed of a single cell monolayer of ASMCs, the source of plasmin must be the cells themselves. However, in vivo, additional sources of plasmin could be inflammatory cells such as macrophages, mast cells, eosinophils, neutrophils, and lymphocytes (5, 15, 37) that are present in the walls of asthmatic airways. These inflammatory cells generate proteases such as plasmin when they are activated in a nonspecific manner or are stimulated to facilitate migration into tissue (11, 36). It is then conceivable that at times of inflammation in the airways, an increase in plasmin activity occurs and may lead to the release of TGF-β.

Confluent monolayers of ASMCs in vitro may resemble some characteristics of ASMCs in situ (12, 13). When ASMCs are confluent, they do not release TGF-β. In vitro wounding of confluent monolayers or conditions of subconfluence could be considered analogous to in vivo injury to the smooth muscle layer of the airways (12,13). When the ASMC monolayers are disrupted, there is a release of TGF-β, plasmin, and procollagen I synthesis. In this context the presence of plasmin releases both TGF-β1 and -β2, which, in turn, induce collagen I synthesis. In vivo, an injury to the ASMCs could occur during airway inflammation that may be due to a myriad of inflammatory mediators such as proteases, oxygen radicals, and cytokines (34). Injury in a localized region could resemble the culture conditions of subconfluent cells or wounded monolayers. If this were to occur in vivo, the release of biologically active TGF-β in these localized areas could result in ASMC connective tissue synthesis. Such a cycle of injury and inflammation may be repeated many times in patients with asthma (15, 34), which could contribute to the remodeling of airways described earlier (6, 13, 15, 32).

At subconfluence, there are large amounts of active and latent TGF-β present in the CM. Yet the CM from the same cells when allowed to become confluent contains markedly decreased quantities of active and latent TGF-β. TGF-β is easily degraded and can become adherent to plastic (31), which may account for loss of some activity in the CM. However, it is also possible that the TGF-β1 released by subconfluent ASMCs may be preserved. It is of interest that plasmin-mediated release of active TGF-β results in the removal of LAP from L-TGF-β (3), but the LAP can reassociate with the active TGF-β, resulting in L-TGF-β (3). L-TGF-β1 can complex with the LTBP-1 (37, 38), which is present in abundant quantities in the CM of subconfluent ASMCs. The association of L-TGF-β with LTBP-1 could then interact with the ECM generated by the ASMCs (35, 37). Confluent ASMCs do not secrete TGF-β, but the addition of plasmin results in increased TGF-β activity and LTBP-1 in the CM. TGF-β complexed with LTBP-1 is susceptible to release by plasmin (35, 37), suggesting that the addition of plasmin to confluent ASMCs releases the LTBP-1 from its interactions with the ECM and TGF-β from its associations with LTBP-1. These findings are highly significant because they suggest that, if TGF-β is released by ASMCs at some point, the same TGF-β complexed with LTBP-1 can later be present on ASMCs or the ECM generated by ASMCs.

TGF-β is one of the most potent inducers of connective tissue synthesis (2, 40). Because connective tissue proteins, including collagens and fibronectin, have been shown to be increased in the airways of asthmatics (6, 32), it was of interest to observe that subconfluent monolayers and monolayers with wounds expressed increased quantities of collagen regulated by TGF-β in an autocrine manner. Furthermore, confluent ASMCs that synthesize small quantities of collagen can respond to TGF-β1–3 by induction of collagen I synthesis. This would suggest that, in the event of increase in TGF-β1, -β2, or -β3 levels in the airways, induction of connective tissue synthesis is likely to occur. Physiologically, a localized increase of TGF-β in the airways could occur by the actions of plasmin or ASMC injury or at times of inflammation in the airways of asthmatics when the sources of TGF-β1 could be eosinophils, macrophages, and lymphocytes (15, 26, 29, 34). Although TGF-β3 was detected in ASMCs by immunohistochemistry (18,19), the release of TGF-β3 could not be adequately identified by an assay used previously to detect and characterize TGF-β isoforms (18, 21). It is possible that TGF-β3 is not released by ASMCs under the conditions described, or, alternatively, TGF-β3 is released but in quantities that cannot be detected by the current assay.

In conclusion, our findings demonstrate that for in vitro ASMCs that resemble conditions of injury, there is plasmin-mediated release of TGF-β, which in an autocrine fashion results in collagen synthesis. These findings suggest that the TGF-β expressed by ASMCs may be important in the pathogenesis of airway connective tissue synthesis and remodeling.

Acknowledgments

We thank Carolin Hoette for helping prepare the manuscript and Dr. Kevin Craib for the statistical analysis.

Footnotes

  • * Amanda Coutts and Gang Chen contributed equally to this work.  Address for reprint requests and other correspondence: N. Khalil, Division of Respiratory Medicine, Univ. of British Columbia, 655 West 12th Ave., Vancouver, BC, Canada V5Z 4R4 (E-mail:nasreen.khalil{at}bccdc.hnet.bc.ca).

  • Funding for this work was provided by Glaxo Wellcome. A. Coutts was supported in part by a Manitoba Lung Association Fellowship. S. Hirst is the recipient of Wellcome Trust Research Career Development Fellowship 051435 and was a visiting assistant professor in the Department of Physiology, University of Manitoba, Canada.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 29.
  29. 30.
  30. 31.
  31. 32.
  32. 34.
  33. 35.
  34. 36.
  35. 37.
  36. 38.
  37. 39.
  38. 40.
  39. 41.
View Abstract