Transforming Growth Factor-β1 Specifically Induce Proteins Involved in the Myofibroblast Contractile Apparatus*

Transforming growth factor-β1 (TGF-β1) induces α-smooth muscle actin (α-SMA) and collagen synthesis in fibroblast both in vivo and in vitro and plays a significant role in tissue repair and the development of fibrosis. During these processes the fibroblasts differentiate into activated fibroblasts (so called myofibroblasts), characterized by increased α-SMA expression. Because TGF-β1 is considered the main inducer of the myofibroblast phenotype and cytoskeletal changes accompany this differentiation, the main objective of this investigation was to study how TGF-β1 alters protein expression of cytoskeletal-associated proteins. Metabolic labeling of cell cultures by [35S]methionine, followed by protein separation on two-dimensional gel electrophoresis, displayed ∼2500 proteins in the pIinterval of 3–10. Treatment of TGF-β1 led to specific spot pattern changes that were identified by mass spectrometry and represent specific induction of several members of the contractile apparatus such as calgizzarin, cofilin, and profilin. These proteins have not previously been shown to be regulated by TGF-β1, and the functional role of these proteins is to participate in the depolymerization and stabilization of the microfilaments. These results show that TGF-β1 induces not only α-SMA but a whole set of actin-associated proteins that may contribute to the increased contractile properties of the myofibroblast. These proteins accompany the induced expression of α-SMA and may participate in the formation of stress fibers, cell contractility, and cell spreading characterizing the myofibroblasts phenotype.

A characteristic feature of asthma is the remodeling of the subepithelial compartment of the bronchial airway wall, sometimes referred to as subepithelial fibrosis. The remodeling involves increased production of collagen I, III, and V, fibronectin, and proteoglycans (1). One of the mechanisms for the excess production of extracellular matrix molecules has been proposed to be due to an increase in the number of differentiated bronchial myofibroblasts (2). The myofibroblasts play a central role in wound healing (3)(4)(5)(6)(7) and appear to be involved in the formation and repair of the extracellular matrix (8 -10; for review see 6,7,11). The myofibroblasts are characterized by antibody reactions against components in the intracellular microfilaments such as ␣-smooth muscle actin (␣-SMA) 1 and members of the intermediary filament like vimentin, desmin, and laminin (12,13). In an activated state, the myofibroblasts also stain positive for muscle heavy chain myosin (also known as tropomyosin) (14,15). Another distinguishing feature of the myofibroblasts is their ability to secrete large amounts of matrix molecules, such as collagens and proteoglycans, compared with the fibroblasts (16,17), which are proteins found in increased levels in fibrotic tissue (18 -23). The transformation of fibroblasts to myofibroblasts can be caused by different exogenous stimuli, where the main factor is considered to be transforming growth factor ␤ 1 (TGF-␤ 1 ) (24 -27). Both in vivo and in vitro, TGF-␤ 1 alters the fibroblast phenotype to one that resembles the features of myofibroblasts, including increased expression of ␣-SMA (21, 26 -33), collagen, fibronectin, and proteoglycan synthesis (24, 27, 28, 31, 34). TGF-␤ 1 is also one of the main growth factors in tissue repair and in the development of fibrosis (19,27,34,35). TGF-␤ 1 signals via two serine/threonine kinase receptors TRI and TRII, and the SMAD-signaling pathway constitutes the main signal transduction route downstream of the these receptors. In fibroblasts, TGF-␤ 1 induces rearrangements of the actin filament system and rapid formation of lamellipodia (36). This response was shown to be independent of the SMAD-signaling pathway; instead, it required the activity of the Rho GTPases Cdc42 and RhoA (36). Long-term stimulation with TGF-␤ 1 resulted in an assembly of stress fibers; this response however required both signaling via Cdc42, RhoA, and SMAD proteins (36). Activation of RhoA by TGF-␤ 1 is also important for formation of stress fibers in the epithelial to mesenchymal transdifferentiation (37). Fibroblasts positive for ␣-SMA have a different phenotype than fibroblasts negative for ␣-SMA (38). The increased level of ␣-SMA found in myofibroblast is potentially important for the contractile features of these cells under repair and fibrotic conditions. Therefore, an increased knowledge of the molecular mechanism behind the induction of ␣-SMA would be of interest. Previous findings from our group show that by stimulating HFL-1 cells with TGF-␤ 1 several tropomyosin isoforms were up-regulated (39). The main aim of this study was to provide a more detailed quantitative study of changes in actin-associated proteins induced by TGF-␤ 1 using [ 35 S]methionine metabolic-labeling human fibroblasts and two-dimensional gel electrophoresis (2-DE).

Materials
The chemicals used were purchased from Merck (Stockholm, Sweden). The recombinant human cytokines tumor necrosis factor ␣ (TNF-␣) (R&D Systems, Minneapolis, MN) and TGF-␤ (R&D Systems) were diluted from stock solutions according to the manufacturer's recommendations.

Cell Culture
The human primary fetal lung cell line (HFL-1) was purchased from American Tissue Culture Collection (Manassas, VA). The HFL-1 cells were grown in T-75 flasks (Costar; Myriadindustries, San Diego, CA) with minimum Eagle's medium (MEM) (M-4655; Sigma, Stockholm, Sweden) supplemented with 10% fetal calf serum (FCS) (Life Technologies, Paisley, United Kingdom) and 2 mM L-glutamine (Life Technologies) at 37°C in 5% CO 2 until confluence. All experiments were performed below passage 18. Low serum medium (LSM) containing 0.4% FCS was used for labeling and stimulation experiments. For the immunohistological staining, the HFL-1 cells were grown in 12-well plates and treated with TGF-␤ 1 . After 48 h, the cells were detached with trypsin, washed with phosphate-buffered saline, adhered to adhesion slides (Bio-Rad, Hercules, CA), fixed in ethanol, and air dried and stored in -80°C until staining. Staining was performed with anti-human ␣-SMA mouse monoclonal antibody (Dako, Copenhagen, Denmark) or isotype for control. Visualization was made with a Vectastain ABC kit. The ␣-SMA positive cells were quantified by counting 400 cells per sample, and the experiments were repeated four times.

Metabolic Labeling of Cells with [ 35 S]Methionine
For Analytical Gels-The protocol for metabolic labeling has been described previously (40). Briefly, the cells were plated in a 1:2 dilution series in 96-well plates (1 well ϭ 0.32 cm 2 ) starting with 2.5 ϫ 10 6 cells/ml and 100 l/well. After 24 h, wells with confluent cells were selected and the medium in these wells was changed to LSM. For comparison, cells were plated in a 96-well plate at a concentration of 12,500 cells/ml and were allowed to grown to confluence for a week, followed by switching to LSM. After an additional 24 h, the cells were washed once with Hanks' buffered saline and once with labeling medium. Labeling medium consisted of MEM without methionine (M-3911; Sigma) supplemented with 0.4% FCS, 2 mM L-glutamine, and 0.5 g/ml cold methionine (M-7520; Sigma). The [ 35 S]methionine (SJ235; Amersham Pharmacia Biotech, Uppsala, Sweden) was lyophilized in a speedvac overnight before use to remove the stabilizing agent ␤-mercaptoethanol and redissolved in labeling medium to a final concentration of 1000 Ci/ml. After washing, 100 l of medium containing [ 35 S]methionine was added to the wells and incubated for various lengths of time so that the cells were exposed to radiolabel for an equivalent period of 20 h in all experiments. Thus, the control cells (t ϭ 0) were radiolabeled for 20 h in 110 l of medium. For 30 min of TGF-␤ stimulation, the cells were prelabeled with medium containing [ 35 S]methionine for 19.5 h and then 10 ng/ml TGF-␤ 1 was added (t ϭ 30 min). For 6 h of TGF-␤ stimulation, the cells were prelabeled for 14 h and then 10 ng/ml TGF-␤ 1 was added (t ϭ 6 h). At harvest, the For Preparative Gels-the cells were plated at 2.5 ϫ 10 5 cells (in 2 ml) in six-well plates and grown to confluence. The medium was changed to LSM and the procedure described for the analytical gels was followed, except that only 100 Ci [ 35 S]methionine per well was used. The cells were lysed in 500 l 8 M urea and 2% CHAPS. The sample was precipitated with ice-cold acetone to avoid excess of salt and lipids present in the cell preparations. The sample was mixed with acetone to a final concentration of 80% acetone and incubated on ice for 40 min and then centrifuged 20,000 rpm in a Sorvall centrifuge.

Two-dimensional Gel Electrophoresis
Immobiline Dry strips (180 mm, pH 3-10 NL) were rehydrated in 350 l of the solubilized cell lysate supplemented with 10 mM dithiothreitol (DTT) and 0.5% immobilized pH gradient 3-10 buffer (Am-ersham Pharmacia Biotech). The isoelectrofocusing step was performed at 20°C in an IPGphor™ (Amersham Pharmacia Biotech) and run according to the following schedule: (1) 30 V for 10 h, (2) 500 V for 1 h, (3) 1000 V for 1 h, and (4) 8000 V until ϳ65,000 Vh was reached. The strips were equilibrated for 10 min in a solution containing 65 mM DTT, 6 M urea, 30% (w/v) glycerol, 2% (w/v) SDS, and 50 mM Tris-HCl, pH 8.8. A second equilibration step was also carried out for 10 min in the same solution except for DTT, which was replaced by 259 mM iodoacetamide. The strips were soaked in electrophoresis buffer (24 mM Tris base, 0.2 M glycine, and 0.1% SDS) just before the seconddimensional gel electrophoresis. The strips were applied on 14% homogeneous Duracryl slabgel. The strips were overlaid with a solution of 1% agarose in electrophoresis buffer (kept at 60°C). Electrophoresis was carried out in a Hoefer™ DALT gel apparatus (Amersham Pharmacia Biotech) at 20°C and constant 100 V for 18 h.
After electrophoresis, the gels were dried using a gel dryer (model 583; Bio-Rad). Small pieces of filter paper were dipped in diluted isotope solution and dried in air and used as markers. The markers were glued on the dried gel before placing it in the imager cassette. Gel Staining-In some cases, the gels were silver stained as described previously (41) or stained with SyproRuby (Molecular Probes, Eugene, OR) according to the manufacturer's recommendations.

Computer-assisted Analysis of Protein Expression in 2-D Gels
Spot analysis was performed using the PDQuest (version 6.1.0) two-dimensional gel analysis system (Bio-Rad). Every spot on the gel is given an integrated optical density (IOD) value that was compared with the total from all valid spots. Thus, each spot has a relative intensity expressed as parts per million (PPM) of the total IOD of all valid spots. Each spot on the gel from stimulated samples was compared with the corresponding spot on the gel with the controltreated sample.

Identification by Mass Spectrometry
Tryptic Digestion-After image analysis, the gel spots were excised from the gel and transferred to separate Eppendorf vials. The gel pieces were washed with ammonium bicarbonate buffer (50 mM) followed by acetonitrile extraction three times. Reduction and alkylation of the excised spots was followed by tryptic digestion overnight using 10 l cold trypsin solution 10 ng/l (Boehringer, Mannheim, Germany).
Mass Spectrometry-The matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) instrument was a DE-PRO Voyager mass spectrometer (Applied Biosystems, Framingham, MA). The instrument, equipped with a delayed extraction ion source, utilized a nitrogen laser at 337 nm and was operated in reflector mode at an accelerating voltage of 20 kV. The sample probes were made of polished stainless steel. Alpha-cyano-4-hydroxycinnamic acid (␣-CHCA) was used as matrix employing the dried droplet technique or elution of peptides purified on reversed-phase microcolumns (42). Tandem mass spectrometry was performed using a nano-electrospray ionization (ESI) spray on a quadrupole TOF mass spectrometer (Micromass, Manchester, United Kingdom). The sample was loaded in precoated borosilicate nano-electrospray needles (MDS Protana, Odense, Denmark; now Protana Engineering A/S). The source temperature was set to 40°C. A potential of 1 kV was applied to the nano-electrospray needles, and 50 V was applied as the cone voltage. Argon was used as a collision gas at a pressure of 6 -7 ϫ 10 Ϫ5 mbar, and the collision energy was 20 -30 V. Product ions were analyzed using an orthogonal TOF analyzer. Data processing was performed with Mass Lynx Window NT PC system.

Statistics
Mean Ϯ SE were calculated. Student's t test was used to evaluate the differences of the means between groups.

[ 35 S]Methionine Labeling of HFL-1 Cells Prior to TGF-␤ 1 Stimulation
The number of regulated spots observed after silver staining was limited. We therefore modified a methodology for [ 35 S]methionine metabolic labeling, which is a more sensitive compared with both silver and fluorescent staining techniques (44,45). Silver staining reveals proteins present in the sample, whereas metabolic labeling focuses on newly synthesized proteins and can thus more effectively reveal cellular responses to a stimulus. The developed protocol increased the number of high-quality protein spots observed in the master gels from 800 to 2400 (Fig. 4). The reproducibility of this protocol was better than that of silver staining when using 100 g total amount of protein. The correlation coefficient of 0.9 within replicate sets was considered satisfactory (n ϭ 24 gels) (46) (see Fig. 4). An additional gain by using [ 35 S]methionine labeling as compared with silver staining is the much wider linear dynamic range of about five orders of magnitude compared with the 40-fold linear dynamic range before saturation for silver staining. The differences in the appearance between the gels stained with silver and the [ 35 S]methionine metabolic labeling protocol largely reflects the latter method focused on newly synthesized proteins. Additional differences may account for the fact that the two staining methods rely on different amino acids.

Induced Expression of Actin-associated Proteins upon
Exposure to TGF-␤ 1 The increased number of visualized spots was used to monitor the effect of TGF-␤ 1 on HFL-1 cells over a 20-h time period. TGF-␤ stimulation was performed by incubating the HFL-1 cells with [ 35 S]methionine for a total of 20 h, including TGF-␤ 1 stimulation for 30 min, 6 h, and 20 h. From the triplicate 2-D gels from every time point it can be seen that the total number of expressed proteins over time was rather constant while the number of proteins regulated by TGF-␤ 1 increased with time from ϳ50 to ϳ100. The procedure, including cell culture, stimulation, harvest, 2-D PAGE, and image analysis, was repeated five times. The combined output from three different strategies was used to select statistically regulated spots: 1) lower level match set where all gels at a certain time point (e.g. 24 h of stimulation) were included into a single match set and the statistical variation was calculated, 2) higher level match set, and 3) export of the raw data from PDQuest into a relational database in which regulated spots could be selected. Eighteen of the regulated spots were identified from the preparative gels and are shown in Fig. 5 plus 11 identified landmark proteins. The protein names are shown in Table I, and the expression pattern of the regulated proteins are shown in Fig. 6. The proteins were not induced at 30 min nor 6 h, but after 20 h of TGF-␤ 1 stimulation. Most were known to be associated with the cytoskeleton, either by binding to actin or by involvement in contractile features of the cells (Fig. 6). Proteins that have previously been shown to be induced by TGF-␤ are marked with a star in Table I. The proteins found to be differentially expressed in this study were SPARC precursor (osteonectin), 2.3-fold (p Ͻ 0.03); TGF-␤-  Table I and Fig. 6. induced protein, 2.8-fold (p Ͻ 0.045); four isoforms of tropomyosin, 1.5-to 11.1-fold (p Ͻ 0.01-0.05); CYR61 protein (caldesmon), 1.7-fold (p Ͻ 0.03); two isoforms of hsp27, 3and 5-fold (p Ͻ 0.05-0.001); two isoforms of smooth muscle protein 22-␣ (transgelin 1 and 2), 1.8-and 3-fold (p Ͻ 0.05-0.01); profilin, 2.6-fold (p Ͻ 0.001); cofilin, 3-fold (p Ͻ 0.01); and calgrizzin, 1.6-fold (p Ͻ 0.05) (number of experiments n ϭ 3). Identities are included in the reference image showing a protein number corresponding to the number in Table I (see Fig. 5).
In addition, it is also shown in the magnified spot trails in Fig. 7 how the expression of different isoforms of calponin is affected by TGF-␤ 1 stimulation (spots 12-15 in Fig. 5). Three of the calponin isoforms were significantly up-regulated by a factor of 4.6 (p Ͻ 0.001), 2.6 (p Ͻ 0.01), and 2.1 (0.05), whereas one was induced (1.7-fold) but not significantly. The sequence coverage of the three isoforms were 23, 42, 38, and 34%, respectively and are displayed in Fig. 7B. The type of post-translational modification cannot be determined from MS spectra alone and needs to be further investigated. Nonetheless, the spot pattern informs us that there is a relatively large shift in isoelectric point between the isoforms, which is the pattern normally associated with changes in phosphorylation. DISCUSSION TGF-␤ is considered one of the main inducers of the differentiation of fibroblasts into myofibroblasts (25). Because ␣-SMA is important for the phenotypic features and contractile properties of the myofibroblasts, we wished to extend our knowledge of cytoskeletal proteins accompanying this upregulation. To provide a more comprehensive study of the protein changes upon TGF-␤ 1 stimulation of human lung fibroblasts, a 2-D gel proteomic approach was used comparing two methods of protein visualization and quantification. Significant up-regulation of a number of actin-associated and calcium-binding proteins was observed.
When using silver staining, a typical protein expression pattern consisted of 1000 -1500 spots in each gel. Silver staining of two-dimensional gels in a pI interval of 3-10 does to some extent enable study of the regulation of proteins belonging to the contractile apparatus of the fibroblast. How- a The numbers correlate to the marked numbers on the master gel in Fig. 5. b Proteins previously shown to be induced by TGF-␤ (see text for details). c A.K.A., Also known as. d Swissprot accession number. e M r , molecular mass. ever, by developing a protocol for metabolic labeling with [ 35 S]methionine there was a marked 3-fold improvement in the overall number of high-quality proteins spots detected on the master gels (from 800 to 2400). The major protein spots in a silver-stained gel were also found in the [ 35 S]methionine gels, and the overall pattern was similar. Nonetheless, the metabolic [ 35 S]methionine approach has the advantage of only displaying newly synthesized proteins. In addition, the linear dynamic range before saturation for silver-stained gels is about 40-fold, whereas use of radioisotope provides five orders of magnitude. TGF-␤ 1 induced a strong expression of ␣-SMA in the majority of the HFL-1 cells. The functional role of the ␣-SMA regulation is to increase the contractility of the myofibroblast (47). The protein expression pattern induced by TGF-␤ 1 stimulation was compared with the pattern seen after TNF-␣ treatment to determine the specificity of the response. Even though some of the regulated proteins responded to both TGF-␤ 1 and TNF-␣, like hsp27, most regulated proteins were cytokine specific. It is thus possible to obtain specific cytokine-driven protein regulation patterns in HFL-1 cells. In addition, when comparing the TGF-␤ 1 effects on Mv1Lu lung epithelial cells determined by 2-DE, the induced proteins were vastly different, indicating a cell-specific response to TGF-␤ 1 (48). In this study, the main effect of TGF-␤ 1 stimulation of human lung fibroblast was the induction of several proteins  Table I. PPM-IOD is the optical density of the spots correlated to the total optical density for all spots present in the gel. responsible for an increase of stress fibers. Several of these proteins has previously been shown to be induced by TGF-␤ 1 : SPARC precursor (osteonectin) (49), TGF-␤-induced protein (50), tropomyosin (39), hsp27 (51), smooth muscle protein 22-␣ (transgelin 1 and 2) (52), and calponin (53). However, in addition we find that CYR61 protein (caldesmon), Rho-GDP dissociation inhibitor, profilin, cofilin, and calgizzarin are induced after 20 h of TGF-␤ 1 stimulation, thus showing that in parallel to the induction of ␣-SMA other actin-associated proteins are induced. Cofilin and profilin activates actin filament polymerization (54). Profilin catalyzes the exchange of ADP for ATP, returning subunits to the pool of ATP-actin bound to profilin, ready to elongate barbed ends of the actin filaments as they become available (55). Cofilin is required for the formation of barbed ends at the leading edge of the actin filaments (56,57) and severs actin, thereby increasing the number of free barbed ends leading to enhanced Arp2/3 complex activation (58). In this study, we found that tropomyosin was also induced, which in contrast to cofilin and profilin, stabilizes the actin filaments and inhibits nucleation and branch formation of the Arp2/3 complex (59) and the cofilin-F-actin interaction in vitro (60 -62). The mutual up-regulation of tropomyosin, profilin, and cofilin seems contradictory, but may be explained by a model proposed by DesMarais et al. (63). They found that, upon epidermal growth factor stimulation, all isoforms of tropomyosin were absent from the ex-FIG. 7. Zoom segment illustrating calponin isoforms regulated upon TGF-␤ 1 stimulation. HFL-1 cells were stimulated with TGF-␤ for 20 h and the cell lysate was separated by 3-10NL strips and 14% homogenous gels. The proteins were visualized with [ 35 S]methionine. A, TGF-␤ 1 induces the expression of four calponin isoforms with similar molecular mass but with different pI, indicative of a post-translational modification. B, The four spots were identified as calponin with a sequence coverage of 23, 42, 38, and 34%, respectively. Peptides displayed matched to calponin in the database with a mass error of less than 50 ppm. The peptides sequences marked with a star contain previously described phosphorylation sites, all of which are found as unmodified in the four MS spectra. treme dynamic leading edge compartment of lamellipodia, but rather co-localized deeper in the lamellipodia and in stress fibers, whereas cofilin was localized in the extreme leading edge of the lamellipodia. Thereby, the tropomyosin-free-actin filaments in the extreme leading edge just under the membrane can participate in actin branching and polymerization, whereas tropomyosin inhibits the branching and stabilizes the actin-filaments elsewhere in the cell (63).
Calponin, calgizzarin, smooth muscle protein 22-␣ (transgelin), and tropomyosin have previously been characterized as smooth muscle markers and are induced in smooth muscle differentiation (64). Additionally in some tissues, myofibroblasts stain positive for muscle heavy chain myosin or tropomyosin (old terminology), caldesmon, and desmin (6,14,15), perhaps explaining the contractile properties of the myofibroblast. Calponin is a thin filament-associated protein that has also been implicated in the modulation of the contractile apparatus in smooth muscle cells via its interaction with actin (65). The suggested role of calponin is mediated by inhibition of the actin-activated myosin Mg-ATPase (65). When calponin is phosphorylated either by PKC or Rho at Thr 170 , Ser 175 , Thr 180 , and Thr 259 , the binding of calponin to F-actin is inhibited (66). An up-regulation of the phosphorylated isoforms suggest increased contractile capability of the fibroblast. However, in this study, all of the potential phosphorylation sites were, within the matched peptides, indicating that these peptides are unmodified or that given sensitivity limits the spectra do not resolve this modification. An explanation can be that the peptides in fact are phosphorylated but not detected because they do not bind well to the micropurification columns used or the documented fact of ion suppression when detecting the phosphopeptides, which necessitates different MS strategies for their detection (67). Therefore additional experiments are required to determine the type of posttranslational modification seen in this case. Calgizzarin is a calcium-binding protein (68) that has been found in developing skeletal muscles (69). Expression and subcellular localization of calgizzarin has been shown to be affected in aging (70) and immortalization of fibroblasts (71,72). Thus, we find in this study that TGF-␤ 1 induced several proteins that are either actin-or calcium-binding proteins and whose function has been implicated in the formation of contractile stress fibers. Several of the proteins are also known smooth muscle cell markers, suggesting that TGF-␤ 1 induces a more contractile hybrid fibroblast-smooth muscle cell-like phenotype.
In summary, in this 2-D gel proteomic approach, we compared two methods of protein visualization and quantification. The modified protocol for metabolic labeling with [ 35 S]methionine showed a marked 3-fold improvement in the overall number of high-quality proteins spots detected and provided a more comprehensive study of the protein changes accompanying TGF-␤ 1 stimulation of human lung fibroblast. The results show that TGF-␤ 1 induces a whole set of actin-associated proteins involved in both actin polymerization and stabilization. These proteins accompany the induced expression of ␣-SMA and participate in the formation of stress fiber, cell contractility, and cell spreading characterizing the myofibroblasts phenotype. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Current address: Active Biotech, AB Box 724, 220 07 Lund, Sweden.