Myofibroblast Differentiation by Transforming Growth Factor-β1 Is Dependent on Cell Adhesion and Integrin Signaling via Focal Adhesion Kinase*

Myofibroblast differentiation and activation by transforming growth factor-β1 (TGF-β1) is a critical event in the pathogenesis of human fibrotic diseases, but regulatory mechanisms for this effect are unclear. In this report, we demonstrate that stable expression of the myofibroblast phenotype requires both TGF-β1and adhesion-dependent signals. TGF-β1-induced myofibroblast differentiation of lung fibroblasts is blocked in non-adherent cells despite the preservation of TGF-β receptor(s)-mediated signaling of Smad2 phosphorylation. TGF-β1 induces tyrosine phosphorylation of focal adhesion kinase (FAK) including that of its autophosphorylation site, Tyr-397, an effect that is dependent on cell adhesion and is delayed relative to early Smad signaling. Pharmacologic inhibition of FAK or expression of kinase-deficient FAK, mutated by substituting Tyr-397 with Phe, inhibit TGF-β1-induced α-smooth muscle actin expression, stress fiber formation, and cellular hypertrophy. Basal expression of α-smooth muscle actin is elevated in cells grown on fibronectin-coated dishes but is decreased on laminin and poly-d-lysine, a non-integrin binding polypeptide. TGF-β1 up-regulates expression of integrins and fibronectin, an effect that is associated with autophosphorylation/activation of FAK. Thus, a safer and more effective therapeutic strategy for fibrotic diseases characterized by persistent myofibroblast activation may be to target this integrin/FAK pathway while not interfering with tumor-suppressive functions of TGF-β1/Smad signaling.

Cell differentiation and phenotypic plasticity of cells, including that of adult stem cells, is critically dependent on the nature of the cellular microenvironment. Rapid and dynamic alterations of this microenvironment occur in the setting of tissue injury, inflammation, and repair. Fibroblasts are active participants in such processes and are characterized by their exceptional ability to undergo various interconversions between related but distinctly different cell types. This phenotypic plasticity of fibroblasts allows them to play an active role in the tissue repair and remodeling process.
Fibroblast phenotype heterogeneity has been described in normal physiologic and diverse pathologic conditions (1,2). Gabbiani et al. (3) first described the transient appearance and disappearance of so-called myofibroblasts in the granulation tissue of healing cutaneous wounds. Myofibroblasts possess ultrastructural features intermediate between fibroblasts and smooth muscle cells and have been defined by their ability to express contractile proteins, particularly ␣-smooth muscle actin (␣-SMA) 1 protein (3,4). This "contractile phenotype" is functionally important for the closure of cutaneous wounds (4). In addition, these cells represent an "activated" fibroblast phenotype with high synthetic capacity for extracellular matrix (ECM) proteins (5,6), growth factors/cytokines (7), growth factor receptors (8), integrins (9), and oxidants (10). The presence/ activation of myofibroblasts appears to be a consistent finding in the pathology of human fibrotic diseases involving diverse organ systems such as the lung, liver, and kidney (11). Thus, persistent myofibroblast proliferation and/or survival represent a pathologic repair process that can result in aberrant architectural remodeling of tissues associated with end-stage fibrosis and organ failure.
Transforming growth factor-␤1 (TGF-␤1) has been linked to most of the fibrotic diseases in humans (11). TGF-␤1 has been shown to induce myofibroblast differentiation both in vitro (12) and in vivo (13), but regulatory mechanisms for this effect are unclear. TGF-␤ receptor(s) activation results in the rapid recruitment and phosphorylation of the Smad proteins (14). We have shown previously (10,15) that TGF-␤1 induces delayed protein tyrosine phosphorylation in human lung fibroblasts, an effect that is associated with extracellular H 2 O 2 production by a cell surface oxidase. ECM signals, either chemical or physical, can modulate TGF-␤1-induced myofibroblast differentiation (16,17), suggesting a role for cell adhesion receptors in mediating/promoting myofibroblast differentiation.
Integrins are the major cell adhesion receptors for ECM ligands that mediate adhesion-dependent signaling in diverse biological processes including cell proliferation, migration, and apoptosis (18). Adhesion-dependent integrin activation recruits focal adhesion kinase (FAK) and activates it at focal adhesions (19,20). FAK activation is mediated primarily by autophosphorylation of tyrosine 397 that creates a binding site for the Src homology 2 domain of Src (21) and phosphatidylinositol 3-kinase (22). Recent studies (23,24) suggest reciprocal catalytic activation of FAK and Src kinases at focal adhesions, specifically in adhesion-dependent integrin aggregation.
Cellular proliferation and differentiation, in general, activate distinct and often opposing signaling pathways (25,26). Yet both processes are critically dependent on cellular adhesion to the ECM (27), suggesting that at least some components of adhesion-dependent signaling might be common to both cell proliferation and differentiation. This study was undertaken to examine the role of adhesion-generated signal(s) in myofibroblast differentiation of normal, untransformed human lung fibroblasts by TGF-␤1.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-All experiments were performed on early passage normal human fetal lung fibroblasts (IMR-90; Institute for Medical Research, Camden, NJ). The cells were maintained in medium consisting of Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum (Sigma), 100 units/ml penicillin/ streptomycin (Sigma), and fungizone (Invitrogen); medium was changed every 3 days. Cells were plated on 35-or 100-mm dishes at a density of 10 6 cells/dish and incubated in 5% CO 2 , 95% air. For immunofluorescence studies, IMR-90 cells were plated on glass coverslips placed in 35-mm Petri dishes. In some experiments, cells were plated on dishes pre-coated with various ECM proteins (BD Biocoat cellware, BD Biosciences). Cells were growth-arrested by reducing the concentration of fetal calf serum in the medium to 0.01% for 48 h prior to stimulation with TGF-␤1. Porcine platelet-derived TGF-␤1 was obtained from R & D Systems, Minneapolis, MN. PP2 (4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo [3,4-d]pyrimidine) and its inactive analog, PP3, were from Calbiochem. LipofectAMINE transfection reagent was from Invitrogen. All other reagents were from Sigma.
Immunoprecipitation and Western Blotting-Cells grown on tissue culture plates were gently washed with 5 ml of phosphate-buffered saline and lysed in 0.5 ml of cold RIPA lysis buffer (1% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS, 0.15 M NaCl, 0.01 M NaH 2 PO 4 , 2 mM EDTA, 0.5 mM NaF) containing 2 mM sodium orthovanadate and 1:100 dilution of protease inhibitor mixture III (Calbiochem). Protein concentrations were determined using the BCA protein assay (Pierce). For each immunoprecipitation, equal amounts of protein were incubated for 4 h in 1:100 dilution of antibody and protein A-agarose beads (15 l/500-l sample) (Santa Cruz Biotechnology, Santa Cruz, CA).
Immunoprecipitated protein samples or whole cell lysates were mixed with a 1:5 v/v ratio of 6ϫ electrophoresis sample buffer (0.2 M EDTA, 40 mM dithiothreitol, 6% SDS, 0.06 mg/ml pyronin, pH 6.8) and boiled at 95°C for 5 min to denature protein. Sample mixtures were then loaded and subjected to electrophoresis in a 4 -20% polyacrylamide gradient gel. Proteins were electrophoretically transferred to polyvinylidene difluoride membrane (Immobilon-P, Millipore Inc., Bedford, MA) and incubated in blocking buffer containing 75 mM sodium phosphate, 70 mM sodium chloride, and 0.1% Tween 20 (pH 7.4) with 5% bovine serum albumin for 1 h at room temperature. The blot was treated with a 1:1000 dilution of primary antibody in blocking buffer overnight at 4°C. Three washes with a buffer containing 10 mM Tris, 100 mM NaCl, and 0.1% Tween 20 were then performed prior to incubation with a secondary antibody conjugated to horseradish peroxidase. The washes were repeated, and membrane was incubated with SuperSignal Substrate Western blotting reagent (Pierce) for 10 min. The blot was then exposed to chemiluminescent-sensitive Kodak X-Omat AR film (Eastman Kodak). Image analysis was performed using the public domain NIH Image program available on the internet at rsb.info.nih.gov/nih-image.
Expression Plasmids and Transient Transfections-The mammalian expression plasmid pKH3 encoding mutant FAK (substitution of Phe for Tyr-397; Y397F-FAK) was a gift from Dr. Hong-Chen Chen (Institute for Biochemistry, National Chung Hsing University, Taiwan) and has been described previously (22). Plasmid encoding green fluorescent protein (GFP) was from Clontech, Palo Alto, CA. Co-expression of FAK plasmid constructs with GFP was by transient transfections of IMR-90 cells using the cationic lipid reagent, LipofectAMINE (Invitrogen) according to manufacturer's instructions. Optimal ratio of DNA (g) to LipofectAMINE (l) was determined to be ϳ1:5 for IMR-90 cells. Cells were incubated with DNA-lipid complexes in serum-free RPMI medium for 4 -5 h prior to introducing serum (10%) for 16 -20 h. The next day, transfection medium was removed, and cells were again placed in serum-free medium for 4 h prior to treatment with TGF-␤1 (2 ng/ml) for 24 h. Cells were then analyzed by immunofluorescence staining.
Immunofluorescence Staining-Cells grown on glass coverslips were initially rinsed with Dulbecco's modified Eagle's medium (Invitrogen) for 30 s at ambient temperature and then fixed in 4% formaldehyde for 5 min. Cells were then washed three times in phosphate-buffered saline prior to permeabilization and after each subsequent step. Permeabilization was performed in buffer consisting of 0.1% Triton in 50 mM PIPES (pH 7.0), 90 mM HEPES (pH 7.0), 0.5 mM MgCl 2 , 0.5 mM EGTA, and 75 mM KCl for 30 s at room temperature. Coverslips were sequentially incubated with mouse monoclonal anti-␣-SM antibody (Dako Corp., Carpenteria, CA) and rhodamine-labeled anti-mouse antibody (Jackson ImmunoResearch, West Grove, PA), each for 60 min at room temperature. Cells were then visualized and photographed using a Zeiss fluorescence microscope.
Assay for Plasminogen Activator Inhibitor-1 Promoter Activity-Mink lung epithelial cells (MLECs-clone 32) stably transfected with an 800-bp fragment (Ϫ700 to ϩ71) of the 5Ј end of the human plasminogen activator inhibitor-1 (PAI-1) gene fused to the firefly luciferase reporter gene in a p19LUC-based vector containing the neomycin resistance gene from pMAMneo was a gift from Dr. Dan Rifkin (Department of Cell Biology, New York University Medical Center, New York). Assays were performed as described previously (28) with minor modifications.

TGF-␤1-induced Myofibroblast Differentiation Is Dependent on Cell
Adhesion-In cultured human lung fibroblasts (IMR-90), TGF-␤1 (2 ng/ml) induces a time-dependent steady increase in ␣-SMA protein expression (Fig. 1A). The up-regulation of ␣-SMA was dose-dependent up to 10 ng/ml and remained elevated for up to 5 days without repeat dosing (data not shown), consistent with a stably differentiated state. This effect of TGF-␤1 was associated with morphological changes of cellular hypertrophy and well formed actin stress fibers ( Fig.  2A, middle panels), characteristic of myofibroblasts. To examine the role of cell adhesive events to modulate TGF-␤1-induced myofibroblast differentiation, the effect of TGF-␤1 on cells stimulated under non-adherent conditions was examined. Cells were detached by EDTA treatment, dispersed by pipetting, and washed with Ca 2ϩ -and Mg 2ϩ -containing Hanks' balanced salt solution. Cells were then stimulated with TGF-␤1 (2 ng/ml) in suspension, and the effect on ␣-SMA protein expression was assessed at 24 h. TGF-␤1 was unable to induce ␣-SMA expression in suspended cells (Fig. 1B). The inability to up-regulate ␣-SMA expression in non-adherent cells following TGF-␤1 stimulation is not explained by loss of cell viability because these cells are capable of re-attaching and growing when replated on culture dishes; moreover, up-regulation of ␣-SMA expression was partially restored when cells treated with TGF-␤1 were returned to an adherent state (data not shown). This suggests that adhesion-dependent signal(s) mediate and/or maintain the myofibroblast differentiated state induced by TGF-␤1.

Pharmacologic Inhibition of FAK/Src Kinase(s) Activity Blocks TGF-␤1-induced Myofibroblast Differentiation-
Adhesion-dependent signal(s) may be transmitted by integrins, and these adhesion receptors activate FAK and Src kinases at focal adhesion complexes (23,24). PP2 is an inhibitor of both Src kinases and FAK (23). To determine whether FAK/Src kinase(s) may be required for TGF-␤1-induced myofibroblast differentiation, the effect of PP2 on the formation of actin stress fibers was examined. Fig. 2A demonstrates that PP2 (10 M) is able to inhibit completely the formation of stress fibers as well as the associated cellular hypertrophy induced by TGF-␤1. The effect of varying doses of PP2 and its inactive analog, PP3, on the expression of ␣-SMA was also examined. Fig. 2B shows a dose-dependent inhibition of TGF-␤1-induced ␣-SMA expression by PP2, with mild inhibition at 1 M, greater than 50% inhibition at 5 M and complete inhibition to basal levels at 10 M. PP3 (10 M) and vehicle control (Me 2 SO) had no effect on TGF-␤1-induced ␣-SMA expression. These results suggest that FAK/Src kinase(s) activity may be required for differentiation of adherent lung fibroblasts to myofibroblasts in response to TGF-␤1.
TGF-␤1 Induces Delayed, Adhesion-dependent FAK Autophosphorylation-Based on the finding that FAK/Src kinase(s) may be involved in TGF-␤1-induced myofibroblast differentiation, the ability of TGF-␤1 to activate this pathway in lung fibroblasts was examined. Activation of Src kinases, measured by in vitro kinase assays of Src, Yes, and Fyn autophosphorylation, in cells stimulated with TGF-␤1 was not observed at early time points (15 and 30 min), and delayed autophosphorylation of Src-Y416 did not reveal significant and consistent responses. However, TGF-␤1 consistently induced phosphorylation of FAK on Tyr-397, its major autophosphorylation site, in a time-dependent manner (Fig. 3A). These results were further confirmed by immunoprecipitation of cell lysates obtained 16 h following TGF-␤1 (2 ng/ml) and immunoblotting with antiphosphotyrosine antibody (PY20) and vice versa. In addition to demonstrating tyrosine phosphorylation of FAK, these studies indicated that this effect of TGF-␤1 is inhibitable by PP2 but not by PP3 (Fig. 3B). The delayed induction of FAK autophosphorylation by TGF-␤1 suggested that new protein synthesis might be required to mediate this effect.  TGF-␤1. FAK autophosphorylation was increased almost 2-fold by 6 h while significant ␣-SMA induction required about 12 h and steady increases thereafter (Fig. 3, C and D). These results demonstrate that TGF-␤1 induces FAK autophosphorylation by a mechanism that requires new protein synthesis, and this effect precedes the induction of ␣-SMA by TGF-␤1.
Early TGF-␤ Receptor Signaling Does Not Require Cell Adhesion-TGF-␤ receptor(s) signaling is primarily mediated by the rapid phosphorylation and activation of the Smad proteins (14). To address the question directly as to whether early TGF-␤1 signaling events require cellular adhesion, TGF-␤1induced Smad2 phosphorylation was examined in adherent and suspended cells. Rapid phosphorylation of Smad2 was observed within 15 min of TGF-␤1 stimulation of cells under both conditions (Fig. 4A). Interestingly, Smad2 phosphorylation in response to TGF-␤1 was higher in suspended cells (Fig.  4, A and B). In contrast, delayed autophosphorylation of FAK was almost completely abrogated in suspended cells (Fig. 4, A and C), consistent with the requirement for cell adhesion to mediate FAK autophosphorylation/activation. To determine whether Smad-dependent transcriptional events were altered under non-adherent conditions, the effect of TGF-␤1 to induce activation of the plasminogen activator inhbibitor-1 (PAI-1) gene was assessed utilizing the stably transfected PAI-1 gene promoter-reporter MLECs. Similar to the effects on Smad2 phosphorylation, PAI-1 promoter activity in response to TGF-␤1 was higher in suspended versus adherent cells, 11.9versus 7.5-fold, respectively, relative to adherent control cells (Fig. 4D). This suggests that loss of cell adhesion does not interfere with the ability of TGF-␤1 to bind/activate its receptor and mediate early Smad signaling; furthermore, the transcriptional activation of a well characterized TGF-␤1-responsive gene (PAI-1) is not altered under non-adherent conditions. FAK Autophosphorylation Is Required for TGF-␤1-induced Myofibroblast Differentiation-Because protein kinase inhibitors have the potential to target multiple protein kinases, the specific role of FAK autophosphorylation in TGF-␤1-induced myofibroblast differentiation was examined by expressing kinase-deficient FAK in these cells. Kinase-deficient FAK (Y397F-FAK, substitution of Phe for Tyr-397) was co-expressed with green fluorescent protein (GFP; transfected in 1:10 ratio by weight) by transient transfections. Cells expressing mutant FAK (GFP-encoding "green" cells) were incapable of expressing ␣-SMA (red fluorescence), whereas untransfected cells (GFPnegative) in the same field expressed high levels of ␣-SMA in response to 2 ng/ml TGF-␤1 (Fig. 5A). To confirm these findings at the level of protein expression, cells were transfected with Y397F-FAK or "mock"-transfected, and the effect of TGF-␤1 on ␣-SMA expression was determined by Western blotting. In cells transfected with mutant FAK (Y397F-FAK), protein expression of ␣-SMA in response to TGF-␤1 was markedly diminished (Fig. 5B). Together, these results indicate that autophosphorylation of FAK on Tyr-397 is required for induction of myofibroblast differentiation by TGF-␤1.
TGF-␤1 Induces Expression of Fibronectin and the Integrin Receptor Subunits ␣ 4 , ␣ 5 , and ␤ 1 in Association with Enhanced FAK Autophosphorylation-The observation that adhesion-dependent signals involving FAK autophosphorylation are required for TGF-␤1-induced ␣-SMA expression and the relatively delayed nature of the response suggested a secondary effect of this cytokine to modulate adhesion signaling. Therefore, we examined the ability of TGF-␤1 to alter the protein To localize individual cells that had been successfully transfected, a plasmid encoding green fluorescent protein (GFP) was cotransfected in a 10:1 (w/w) ratio. Cells were stimulated with TGF-␤1 (2 ng/ml) in serum-free medium for 24 h, and immunofluorescent cell staining was performed with a monoclonal antibody against ␣-SMA. The same fields were photographed under green and red fluorescent light for GFP and ␣-SMA, respectively (inner panels); the merged image is also shown (right panels). B, expression of kinasedeficient FAK inhibits the TGF-␤1-induced increase in ␣-SMA. Cells were transiently transfected with a kinase-deficient FAK mutant (Y397F-FAK) plasmid or without plasmid (transfection reagents alone; control transfection) prior to incubation in the absence (lane C) and presence (TGF) of TGF-␤1 (2 ng/ml) for 24 h. Cell lysates were obtained and subjected to SDS-PAGE followed by immunoblotting with antibodies against ␣-SMA and ␤-actin. expression of fibronectin and the integrin receptor subunits, ␣ 4 , ␣ 5 , and ␤ 1 in these cells. All three integrin subunits as well as fibronectin were up-regulated by TGF-␤1 in a time-dependent manner (Fig. 6). These effects of TGF-␤1 are associated with adhesion-dependent activation of FAK (Figs. 3A and 4C). This provides further support for the concept that alterations in cell adhesive capacity, by up-regulated expression/activation of integrin receptors and fibronectin by TGF-␤1, may be important in FAK activation and myofibroblast differentiation.
Regulation of Myofibroblast Differentiation on Different ECM-coated Surfaces-Integrins bind to ECM ligands with varying and, sometimes, overlapping specificities (29). To determine whether ECM proteins differentially regulate myofibroblast differentiation, cells were plated on tissue culturetreated dishes known to promote cell adhesion by integrin binding, dishes coated with ECMs (collagen I, fibronectin, laminin and collagen IV), and dishes coated with a non-integrinbinding polypeptide, poly-D-lysine. Fig. 7A demonstrates that the basal expression of ␣-SMA varies significantly depending on the ECM protein substrate where cells are grown. Cells grown on fibronectin expressed the highest level of ␣-SMA, whereas expression on laminin and poly-D-lysine was decreased (Fig. 7A). ␣-SMA expression also correlated with levels of both FAK autophosphorylation and total FAK, which appeared to be enhanced on fibronectin (results not shown). These results further support a role for integrin signaling in myofibroblast differentiation and suggest that specific integrin-ECM interactions may differentially regulate this process. When adherent cells on different ECMs are stimulated with TGF-␤1, the ability to differentiate into myofibroblasts is restored under all conditions (Fig. 7B), supporting the concept that the ability of TGF-␤1 to induce expression of integrin/ECM proteins and activate FAK may be critical for this differentiation response. DISCUSSION Myofibroblasts are active participants in normal wound repair, but persistence of these cells in injured tissues prevents normal healing and promotes a dysregulated repair process characterized by progressive connective tissue remodeling and fibrosis (30). Transformation of fibroblasts to the myofibroblast phenotype is primarily mediated by the pro-fibrotic cytokine, TGF-␤1 (12); however, molecular mechanisms for this effect are unclear. An understanding of these mechanisms is critical for the development of new therapeutic approaches for these otherwise fatal and largely treatment-unresponsive disorders. In this study, we demonstrate that cell adhesion/integrin-dependent autophosphorylation and activation of FAK is required for the induction of myofibroblast differentiation by TGF-␤1.
Combinatorial signalings involving integrins and mitogenic growth factor receptors have been well characterized, and synergistic activations of downstream signals such as mitogenactivated protein kinase activation are essential for anchoragedependent cell cycle progression (31). For example, growth factor-induced activation of MEK, an activator of mitogenactivated protein kinase/extracellular signal-regulated kinase kinase, is dependent on cell adhesion (32). The mechanisms by which cell adhesion/integrin signals regulate cellular differentiation are less clear, although a critical role for the ECM is FIG. 8. Schematic of the proposed adhesion-dependent and -independent regulatory pathways involved in myofibroblast differentiation by TGF-␤1 in human lung fibroblasts. TGF-␤ receptor(s) signaling is primarily mediated by rapid phosphorylation/ activation of Smad proteins, which occurs by an adhesion-independent mechanism. Other unidentified non-Smad pathways may also be activated early post-TGF-␤ receptor(s) activation. FAK autophosphorylation/activation is delayed relative to Smad signaling and is associated with TGF-␤1-induced expression of both integrin subunits and fibronectin/collagens. Integrin signaling via FAK autophosphorylation/activation is essential for induction/maintenance of the stably differentiated myofibroblast phenotype.
FIG. 6. TGF-␤1 induces the expression of integrin subunits (␣ 4 , ␣ 5 , and ␤ 1 ) and fibronectin in adherent human lung fibroblasts. Quiescent, adherent human lung fibroblasts (IMR-90) were treated with TGF-␤1 (2 ng/ml) for varying times as indicated. Cell lysates were then obtained and subjected to SDS-PAGE followed by immunoblotting with specific antibodies against the proteins indicated. FIG. 7. A, expression of ␣-SMA in human lung fibroblasts grown on different ECM substrates. IMR-90 cells were plated at the same density on cell culture-treated dishes (tissue culture plastic), dishes coated with various ECM proteins (collagen I, fibronectin, laminin, and collagen IV) and a non-integrin-binding adhesion polypeptide (poly-D-lysine). At ϳ90% confluency, cells were serum-deprived for 48 h and cell lysates obtained. Cell lysates were then subjected to SDS-PAGE followed by immunoblotting with antibodies against ␣-SMA and ␤-actin. B, effect of TGF-␤1 on induction of ␣-SMA on ECM substrates. Quiescent cells were stimulated in the presence/absence of TGF-␤1 (2 ng/ml) for 24 h prior to cell lysis and Western blotting as described above. well recognized (27,33). Our data strongly support a role for adhesion/integrin-dependent FAK activation in the differentiation response to TGF-␤1. However, in contrast to the model for adhesion-dependent proliferative responses, growth factor receptor(s) activation does not appear to require simultaneous integrin activation. Several lines of evidence support this notion. First, we found efficient and rapid activation of TGF-␤ receptor(s) signaling of Smad phosphorylation even in the absence of cell adhesion. In fact, the phosphorylation of Smad2 by TGF-␤1 was enhanced in non-adherent cells; moreover, these changes were associated with preserved (and even enhanced) PAI-1 promoter activity in non-adherent cells. Second, adhesion signals provided several hours after early TGF-␤ receptor(s)-mediated signaling had been transduced was, at least partially, sufficient to induce differentiation. Third, FAK activation in response to TGF-␤1 is adhesion-dependent but occurs in a delayed manner relative to early post-receptor signaling. FAK autophosphorylation, although appreciable at 3 h poststimulation with TGF-␤1, is maximal at 16 h. We have observed previously (15) a similar response to this cytokine in the induction of protein tyrosine phosphorylation in lung fibroblasts but had not identified FAK as a potential candidate molecule for this regulation. Finally, the phosphorylation of FAK is temporally associated with the up-regulation of integrin receptors and fibronectin, suggesting that these effects of TGF-␤1 may be related. Cumulatively, our data suggest a "serial" regulatory pathway by which TGF-␤1 modulates adhesion signaling that is critical for the induction and, perhaps more importantly, for the maintenance of a stably differentiated myofibroblast phenotype. It is important to recognize that adhesion signal(s), although necessary, are unlikely to be sufficient for full expression of this phenotype. Plating cells on fibronectin (in the absence of TGF-␤1) only partially promoted ␣-SMA expression. Components of early TGF-␤ receptor(s) signaling may function in a "parallel" (or synergistic) manner to transmit other signals for myofibroblast differentiation (see Fig. 8).
The findings of this study also support the concept that specific integrin-ECM interactions may differentially regulate this cell adhesion-dependent pathway. The basal expression of ␣-SMA was lower in cells grown on laminin, relative to other integrin-binding ECM surfaces, and was highest on fibronectin. The observation that fibronectin promotes this differentiation response is consistent with reports that the ED-A spliced variant of fibronectin is essential for TGF-␤1-induced myofibroblast differentiation (16). Interestingly, this variant of fibronectin is preferentially produced by fibroblasts grown on rigid planar surfaces and in response to TGF-␤1 (34,35). This may explain the relatively strong induction of ␣-SMA on "rigid" tissue culture dishes even in the absence of ECM coating of dishes. Furthermore, the ability of TGF-␤1 to up-regulate fibronectin and integrin receptors may serve to augment and sustain this fibroblast phenotype by promoting cell adhesive interactions and altering the biochemical/biomechanical microenvironment of the cell. We have shown recently (36) that biochemical, and potentially biophysical, alterations of the ECM can be induced in an inflammatory milieu by activated myofibroblasts via the generation of extracellular hydrogen peroxide and oxidative cross-linking of ECM proteins. More recent data from our laboratory also suggest that such oxidative modifications occur preferentially on fibronectin. 2 Thus, complex and dynamic changes in the cellular microenvironment influence the ultimate phenotypic fate of cells and, eventually, the outcome of the tissue injury and repair process.
The involvement of cell adhesion/integrin-FAK signaling in promoting the myofibroblast phenotype has important implications for potential new approaches to the therapy of fibrotic diseases. Although blocking TGF-␤ signaling is likely to be of benefit in abrogating the fibrotic response (37), it could also prove detrimental because this pathway functions in the control of epithelial cell proliferation and tumor suppression (38). This is further complicated by the reported increased incidence of epithelial cell malignancies in fibrotic diseases (39). Thus, the identification of a non-tumor-suppressive pathway critical for myofibroblast differentiation provides an ideal target for the development of new therapies for fibrotic diseases characterized by TGF-␤1 and myofibroblast activation.