p38 Inhibitors Prevent TGF-β–Induced Myofibroblast Transdifferentiation in Human Tenon Fibroblasts

Abstract

purpose. The role of mitogen-activated protein kinase (MAPK) pathways in TGF-β–induced myofibroblast transdifferentiation of human tenon fibroblasts (HTFs) was investigated to identify potential pharmacologic targets for the inhibition of scarring after glaucoma surgery.

methods. TGF-β–dependent activation of Smad2, p38, and Erk-1/2 was examined by Western blot analysis. TGF-β–induced mRNA expression of collagen Iα1, fibronectin, and the myofibroblast transdifferentiation marker alpha smooth muscle actin (α-SMA) was analyzed by real-time RT-PCR. α-SMA protein expression and subcellular distribution were determined by Western blot analysis and immunofluorescence cytochemistry. Fibroblast contractility was assessed in three-dimensional collagen gel contraction assays, stress fiber assembly with rhodamine-phalloidin stains, and confocal microscopy. Cell proliferation was measured with an MTT assay. Specific pharmacologic kinase inhibitors were used to characterize the involvement of MAPK-dependent pathways.

results. TGF-β stimulation of HTF induced a rapid and transient activation of Smad2 and Erk, whereas p38 activation was biphasic and sustained. After 24 hours of TGF-β stimulation, increased levels of collagen Iα1, fibronectin, and α-SMA transcripts were detected. After 3 days of stimulation, HTF displayed increased α-SMA protein levels, enhanced contractility, and assembly of actin stress fibers. TGF-β also induced HTF proliferation. Specific p38 inhibitors prevented all these aspects of TGF-β–induced myofibroblastic transdifferentiation.

conclusions. Pharmacologic inhibition of p38 abrogates TGF-β–induced myofibroblast transdifferentiation, reduces extracellular matrix protein expression and HTF proliferation, and may therefore serve to inhibit scarring after glaucoma surgery.

Amajor problem of surgical glaucoma therapy lies in the postoperative scarring process. The transdifferentiation of fibroblasts to myofibroblasts is a crucial step in wound healing and scarring and is required for subsequent tissue remodeling. ¹  

Myofibroblasts share ultrastructural features of fibroblasts and smooth muscle cells. They exert increased contractile activity, which is associated with the de novo expression of alpha smooth muscle actin (α-SMA), ² the actin isoform typically expressed in smooth muscle cells. The increased amounts of α-SMA are incorporated into actin stress fibers as part of the contractile apparatus ^{2 3} to ensure sufficient wound closure. ⁴ Moreover, myofibroblasts represent an “activated” fibroblast phenotype with increased synthesis of ECM proteins, ^{5 6} growth factors, ⁷ growth factor receptors, ⁸ and integrins. ⁹  

Although myofibroblasts are only transiently present during normal wound healing, long-term persistence of myofibroblasts is associated with excessive scarring. ^{10 11 12 13} Myofibroblast transdifferentiation is regulated by growth factors, particularly those of the transforming growth factor-β (TGF-β) family, which have a key role in wound healing throughout the body. ^{14 15 16} Experiments with neutralizing antibodies against TGF-β1 and TGF-β2 showed reduced scar formation in rat dermal excisional wounds. ¹⁷ All three TGF-β isoforms have been identified in the eye, ^{18 19} with TGF-β2 the predominant isoform associated with ocular scarring diseases such as proliferative vitreoretinopathy and posterior lens capsule opacification. ^{20 21}  

TGF-β signaling is mediated by binding of TGF-β to its heteromeric receptor complex, consisting of two serine-threonine kinase receptors designated TGF-β type I and type II receptor, leading to phosphorylation of the signal transducer proteins Smad2 and Smad3. On phosphorylation, these proteins form complexes with the signaling molecule Smad4 and translocate to the nucleus to regulate gene transcription. ^{22 23}  

Recent reports ^{24 25} show that not only the Smad pathway is activated upon TGF-β stimulation; members of the mitogen-activated protein kinase (MAPK) signaling cascades are also stimulated by TGF-β. Activation of these signaling pathways may occur as a direct or an indirect result of TGF-β receptor I and II oligomerization. Direct activation of p38 relies on TGF-β–mediated activation of TAK and MKK3/6 without the involvement of Smad proteins. ^{25 26} An indirect p38-activation mechanism through expression of the stress-inducible protein GADDF45β has also been reported. ²⁷  

In addition to TGF-β, mechanical tension and integrin signaling are prerequisites for the acquisition and maintenance of the myofibroblastic phenotype. ^{1 28 29} Therefore, fibroblast-populated three-dimensional gel matrices represent a model that approximates particularly well the in vivo situation during wound healing. ^{30 31 32}  

Here, we report that selective inhibition of the MAPK p38 prevents the TGF-β–induced transdifferentiation of human tenon fibroblasts to myofibroblasts. These results suggest that p38 inhibitors may serve as noncytostatic drugs to prevent postoperative filtration bleb failure in glaucoma surgery. 

Materials and Methods

Reagents

A total RNA purification kit (RNeasy) and RNase-free DNAse were purchased from Qiagen (Hilden, Germany). Taq Polymerase and M-MLV Reverse Transcriptase Impro II were purchased from Promega (Mannheim, Germany). Antibodies raised against the following proteins were used: α-SMA, β-Actin, and tubulin (Sigma, St. Louis, MO), active Erk1/2 and active p38 (Promega), total Erk (Santa Cruz Biotechnology, Santa Cruz, CA), total p38 (Biosource, Nivelles, Belgium), active Smad2/3 (Zymed/Zytomed, Berlin, Germany), total Smad2/3 (Chemicon, Temecula, CA), Alexa-488–conjugated goat anti–mouse antibody (Molecular Probes, Eugene, OR), and horseradish peroxidase (HRP)–conjugated secondary antibodies (Jackson/Dianova, Hamburg, Germany). 

Recombinant TGF-β1 was obtained from R&D (Wiesbaden-Nordenstadt, Germany), and MEK1/2 inhibitor U0126 was from Promega. p38 Inhibitors SB203580, SB239068, and SB220025 were purchased from Merck Biosciences (Bad Soden, Germany). 

Cell Culture

Small tenon biopsy samples were obtained during standard intraocular surgery after comprehensive information and written consent were received from selected patients. The tenets of the Declaration of Helsinki were followed, and institutional ethics committee approval was granted. Primary human tenon fibroblasts were gained as an expansion culture of the human tenon explants and were propagated in Dulbecco’s modified Eagle medium (DMEM; PAA Laboratories GmbH, Pasching, Austria) supplemented with 10% heat-inactivated fetal calf serum (Gibco Life Technologies, Karlsruhe, Germany), 100 U/mL penicillin, and 100 μg/mL streptomycin (Biochrom, Berlin, Germany). Cells were maintained in the logarithmic growth phase. For all experiments, cells from passages 2 to 6 were used. Cell proliferation assays were performed four times; all other experiments were performed three times and yielded similar results. 

MAPK Inhibitors

Stock solutions (10 mM) of the MAPK inhibitors were prepared in Me₂SO (DMSO). Inhibitors were diluted in unsupplemented DMEM, added to the cell culture 30 minutes before stimulation, and present with the stimuli as indicated. 

RNA isolation and Real-Time Reverse Transcription–Polymerase Chain Reaction

Total RNA was harvested using RNeasy spin columns (Qiagen) according to the manufacturers’ recommendations. To eliminate contamination with genomic DNA, DNAse digestion was performed for 15 minutes. First-strand cDNA was synthesized (M-MLV Reverse Transcriptase; ImproII) at 42°C using 500 ng total RNA extract. 

For experiments investigating TGF-β effects at the transcriptional level, human tenon fibroblasts (HTFs) were seeded in 24-well plates and grown to subconfluence. After serum deprivation for 12 hours, cells were stimulated with TGF-β for 24 hours. RNA extraction and translation to cDNA were performed as described. 

Real-time reverse transcription–polymerase chain reaction (RT-PCR) was conducted in a 96-well microtiter plate with a total reaction volume of 25 μL containing 2 μL of a 1:4 dilution of first-strand reaction product, 0.4 μM specific upstream and downstream primers, 240 μM dNTP mix, 1× reaction buffer, Taq polymerase, and SYBR-green (Sigma-Aldrich, St. Louis, MO) as a fluorescent marker. Amplification and analysis of cDNA fragments were carried out (iCycler; Bio-Rad, Munich, Germany). Cycling conditions were initial denaturation at 95°C for 3 minutes, followed by 40 cycles consisting of a 15-second denaturation interval and a 1-minute interval for annealing and primer extension at 60°C. Amplification of the housekeeping gene hypoxanthine-guanine-phosphoribosyltransferase 1 (HPRT1) mRNA, which served as a normalization standard, was carried out with HPRT1 forward (GACCAGTCAACAGGGGACAT) and HPRT1 reverse (ACACTTCGTGGGGTCCTTTT) primers. Side-strand specific primers for α-SMA, collagen Iα1, and FN were as follows: α-SMA forward (CTGTTCCAGCCATCCTTCAT), α-SMA reverse (CCGTGATCTCCTTCTGCATT), collagen Iα1 forward (GAGAGCATGAC-CGATGGATT), collagen Iα1 reverse (CCTTCTTGAGGTTGCCAGTC), FN forward (AATATCTCGGTGCCATTTGC), and FN reverse (AAAGGCATGAAGCACTCAAT). α-SMA, collagen Iα1, and FN mRNA levels were measured as CT threshold levels and normalized with the individual HPRT1 control CT values. Induction on TGF-β stimulation is indicated as fold increase compared with unstimulated cells. 

Western Blot Analysis

Cells were rinsed with ice-cold PBS, and total cell protein extracts were prepared using a TNE lysis buffer (20 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100) containing phosphatase inhibitors (1 mM sodium vanadate, 50 mM NaF) and protease inhibitors (0.1% phenyl methyl sulfonyl fluoride [PMSF]; Complete Protease Inhibitor; Roche, Mannheim, Germany). Protein concentrations were measured by a BCA assay (KMF, Lohmar, Germany). Ten micrograms protein extracts were boiled in Laemmli sample buffer and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Amersham, Braunschweig, Germany) using a Bio-Rad gel-blotting apparatus. Membranes were stained with Poinceau red to control for equal loading, followed by blocking of the membranes in 3% BSA in TBST (10 mM Tris HCl [pH 7.5], 150 mM NaCl, 0.1% Tween 20) for 1 hour. Membranes were incubated with primary antibody overnight at 4°C and with a peroxidase-conjugated secondary antibody for 45 minutes at room temperature. After each incubation step, membranes were washed in TBST for 20 minutes. Peroxidase was visualized by enhanced chemiluminescence (ECL; Amersham, Braunschweig, Germany) and exposure to ECL film (Hyperfilm; Amersham) for appropriate times. 

Immunofluorescence Confocal Microscopy

HTF cells were seeded on fibronectin (FN; Biochrom)–coated glass coverslips in FCS-supplemented DMEM and were incubated for 24 hours to allow efficient attachment. Cells were serum-deprived for 12 hours, preincubated with SB203580 or vehicle for 30 minutes, and stimulated with TGF-β (2 ng/mL) as indicated. Subsequently, cells were fixed in 2% paraformaldehyde and permeabilized in 0.2% Triton X-100, blocked in 2% normal goat serum (Jackson-Immuno, Hamburg, Germany), and labeled with primary antibody against α-SMA (monoclonal mouse antibody) at a dilution of 1:500 in blocking buffer. Alexa-488–conjugated secondary antibody against mouse IgG was used at a 1:500 dilution in blocking buffer. Rhodamine-phalloidin was used to counterstain the F-Actin cytoskeleton. Cells were viewed with a laser scanning confocal microscope (Leica Microsystems, Bensheim, Germany). 

Fixed Collagen Gel Cultures

Experiments were conducted essentially as described by Grinnel et al. ³³ Briefly, 24-well plates were precoated with 0.2% BSA (KMF, Lohmar, Germany) for 1 hour. HTFs resuspended in DMEM supplemented with 0.2% FCS were added to 16 parts neutralized collagen solution (Vitrogen-100; Cohesion, Invitrogen, Palo Alto, CA), 2 parts 10× DMEM (Sigma), and 1 part NaHCO₃ (Sigma) in a 1:1 ratio, yielding a final concentration of 150,000 cells/mL and 1.2 mg/mL collagen. 

Collagen cell suspension was added to each well and incubated at 37°C for 1 hour for polymerization. Gels were preincubated with synthetic p38 inhibitor SB203580 or vehicle for 1 hour. After stimulation with TGF-β1 (2 ng/mL) for 48 hours in the presence or absence of SB203580, gels were detached and contraction was digitally photodocumented at various time points. Contraction quantification was performed using NIH image software (rsb.info.nih.gov/nih-image/Default.html). 

F-Actin Staining in Collagen Gel Cultures

Collagen lattices were rinsed in PBS, fixed in 2% paraformaldehyde for 20 minutes, and permeabilized with 1% Triton X-100 for 10 minutes. Lattices were then blocked in a mixture of 2% normal goat serum (DAKO, Hamburg, Germany) and 2% fetal calf serum (Biochrom) at 4°C overnight. F-Actin stress fibers were stained with rhodamine-phalloidin and were viewed with a laser scanning confocal microscope (Leica Microsystems). 

Cell Proliferation Assay

To assess cell proliferation, we used a proliferation assay (CellTiter 96; Promega) according to the manufacturer’s recommendations. Briefly, 4000 HTF cells/well were seeded in a 96-well microtiter plate, starved in serum-deprived medium for 16 hours, preincubated with inhibitor SB 203580 as indicated, and stimulated with 2 ng/mL TGF-β, 10% FCS, or vehicle control for 48 hours. Thereafter, 20 μL tetrazolium-containing solution (AQ_ueous Solution; Promega) was added, and absorption at λ = 490 nm was measured. 

Results

TGF-β Upregulates α-SMA and Activates Smad and MAPK Signaling Pathways in HTFs

A crucial step in wound healing is the transdifferentiation of fibroblasts to myofibroblasts, characterized by de novo synthesis of α-SMA. ¹ To address this process in vitro, we stimulated HTF with TGF-β1 (2 ng/mL) under serum-free conditions (Fig. 1) . The expression of α-SMA mRNA as measured by quantitative real-time RT-PCR increased above background noise after 12 hours (twofold) and was pronounced after 24 hours (15-fold; Fig. 1A ). The expression of α-SMA protein increased after 24 hours and was pronounced after 48 hours (Fig. 1B) . To control for unspecific alterations in the expression levels of cytoskeletal proteins, we also examined tubulin protein expression and found equal amounts of tubulin in unstimulated and stimulated cells (Fig. 1B) . Therefore, the increase of α-SMA was a qualitative shift in the cytoskeleton composition representing the myofibroblast transdifferentiation process rather than an unspecific increase in cytoskeletal proteins. 

Thereafter, we were interested in characterizing intracellular signaling events that might have contributed to the increased α-SMA expression. We observed biphasic activation of p38 with early-phase ceasing after 6 hours and a second, sustained phase starting after 12 hours (Fig. 1B) . In contrast, Erk1/2 was temporarily activated for 2 hours, and Smad2 activation lasted for up to 24 hours. Delayed p38 activation coincided with the onset of α-SMA RNA transcription and preceded the expression of α-SMA protein by 6 to 12 hours, suggesting potential involvement of p38 in α-SMA expression. 

TGF-β–Induced α-SMA Protein Expression Is p38 Dependent

To address a possible involvement of p38 in TGF-β–induced α-SMA expression, we studied the effects of specific p38 inhibitors. HTFs were stimulated with TGF-β1 for 3 days under serum-free conditions in the presence of different concentrations of the p38 inhibitors SB203580, SB239068, and SB220025 or vehicle control. All three inhibitors diminished α-SMA expression in a dose-dependent manner (Fig. 2) . These findings further support the critical involvement of p38 in TGF-β–induced α-SMA expression. 

TGF-β–Induced Development of α-SMA–Positive Stress Fibers Is p38-Dependent

Myofibroblasts incorporate α-SMA into the actin cytoskeleton, enabling them to exert enhanced contractile activity. ^{1 30 34} To assess the localization of α-SMA and its recruitment to actin stress fibers in our system, immunofluorescent stainings were performed. HTFs plated on fibronectin-coated glass coverslips were left untreated or were stimulated with TGF-β in the presence of SB203580 [10 μM] or vehicle control. Although untreated cells showed only weak cytosolic α-SMA staining (Fig. 3A) , TGF-β treatment induced an assembly of α-SMA–positive stress fibers in approximately 50% of the cells (Figs. 3D 3F) . Addition of the p38 inhibitor SB203580 prevented the TGF-β–induced expression of α-SMA and its incorporation into actin stress fibers (Figs. 3G 3I) . 

TGF-β–Mediated Contraction of Fibroblast-Populated Collagen Lattices Is p38-Dependent

At the functional level, the transition of fibroblasts to myofibroblasts is characterized by increased contractile activity. To address this issue, the assessment of three-dimensional collagen lattice contraction has become an accepted model system. ^{30 31 32 35} Although free-floating gels provide a model system for relaxed tissues, as in resting dermis or in a very early stage of wound healing, ⁴ attached matrices resemble the granulation phase of wound healing with increased mechanical tension. The latter model allows the study of morphologic changes typically observed during wound healing in vivo, such as the development of stress fibers and the maturation of focal adhesions. ^{36 37}  

We therefore applied the tethered gel model. After seeding HTFs in neutralized collagen solution, cells were incubated in the presence or absence of TGF-β (2 ng/mL) for 48 hours. TGF-β treatment caused a highly significant increase of contractility (gel size reduction to 36% of initial gel size in TGF-β–treated cells compared with 72% in untreated cells) (Figs. 4A 4B) . Addition of the p38 inhibitor SB203580 a half hour before TGF-β stimulation blocked this effect in a dose-dependent manner (48% and 63% gel contraction at 1 μM and 10 μM SB203580, respectively; Figs. 4A 4B ). 

At the end of the gel contraction experiments, we fixed the collagen lattices with formaldehyde, labeled the actin fibers of the embedded fibroblasts with fluorescent phalloidin, and visualized them with a confocal laser scanning microscope. Although unstimulated cells were characterized by cortical F-Actin and a dendritic phenotype (Fig. 4C) , cells treated with TGF-β showed intracellular stress fibers and were of retracted bipolar to stellate shape (Fig. 4D) . Treatment of cells with the p38 inhibitor SB203580 blocked these TGF-β–induced morphologic changes (Fig. 4E) . 

Endogenous Contraction of Fibroblast-Populated Collagen Lattices Is p38-Dependent

In the tethered collagen gel contraction assay, even unstimulated gels contract spontaneously to a certain extent. It has been speculated that this “endogenous” contraction is mediated by endogenously expressed TGF-β and by additional factors such as ECM prestress. To assess the general importance of p38 in tissue contraction, we assessed the effect of p38 inhibitors in endogenous contraction without TGF-β stimulation. Because the amount of contraction was smaller in this system than when TGF-β was applied, we followed the spontaneous contraction process over a longer time period (3 days compared with 3 hours with TGF-β stimulation). 

The Erk inhibitor U0126 or the p38 inhibitor SB 203580 was used. While U0126-treated cells showed contractility similar to that of control cells at low doses (2 μM; 32% and 33% of initial gel size 3 days after gel detachment; Figs. 5A 5B ) and even increased contractility at a higher dose (10 μM; 20% after 3 days), cells treated with the p38 inhibitor SB203580 exerted reduced gel contraction in a dose-dependent manner (42%, 47%, and 61% at 1 μM, 3 μM, and 10 μM, respectively; Figs. 5A 5B ). 

Confocal laser microscopy revealed intracellular actin condensation in untreated control gels, albeit organized stress fiber formation as seen in TGF-β–treated gels (see, for example, Fig. 4D ) was not observed (Figs. 5C 5D) . In contrast, cells treated with 10 μM SB203580 were larger and showed only a delicate subcortical actin cytoskeleton with almost no intracellular actin condensation (Fig. 5E) . Lower concentrations of SB203580 caused a gradual shift from the slightly retracted phenotype of untreated cells with intracellular actin condensations to the widespread phenotype of cells treated with 10 μM SB203580 (not shown). 

TGF-β–Induced Extracellular Matrix Protein Transcription Is Reduced by p38 Inhibitors

Because the deposition of extracellular matrix proteins is another important function of myofibroblasts, we studied the effect of p38 inhibitors on TGF-β–induced transcription of the ECM components collagen Iα1 and fibronectin. The effect on α-SMA transcription was studied as control. Serum-deprived HTFs were pretreated with the specific p38 inhibitor SB203580 (10 μM) or vehicle for 30 minutes; this was followed by 24-hour stimulation with 2 ng/mL TGF-β1 in the presence or absence of the inhibitor. To allow for comparisons, quantitative real-time RT-PCR was performed. Stimulation with TGF-β caused a 14.8-fold increase in collagen Iα1 mRNA and a fourfold increase in fibronectin mRNA abundance (Fig. 6) . Treatment with SB203580 strongly inhibited this effect: collagen- Iα1 mRNA levels increased only 2.5-fold, and fibronectin mRNA remained at baseline levels (Fig. 6) . TGF-β–induced transcription of α-SMA was inhibited by SB203580 in a similar fashion (Fig. 6) . Data are presented as levels relative to the housekeeping gene HPRT1 transcripts and normalized to the ratio in unstimulated cells. 

TGF-β–Induced Cell Proliferation Is Reduced by p38 Inhibitors

We further sought to evaluate the significance of p38 in TGF-β–induced fibroblast proliferation. HTFs treated with TGF-β showed even more pronounced proliferation than a positive control treated with 10% FCS (Fig. 7) . TGF-β–induced proliferation was significantly reduced by the p38 inhibitor SB203580 (10 μM; Fig. 7 ). Therefore, p38 signaling seems to have contributed to TGF-β–induced proliferation. 

Taken together, the inhibition of p38 signaling prevented TGF-β–induced myofibroblast transdifferentiation with respect to α-SMA expression, increased cell contractility, stress fiber development, and ECM protein transcription. TGF-β–stimulated cell proliferation was also reduced. 

Discussion

Postoperative scarring is the major long-term problem in fistulating glaucoma surgery. ^{38 39 40} In ocular wound healing and scarring processes, TGF-β has various roles. ⁴¹ The chemoattractant effects of TGF-β on inflammatory cells in the early phases of wound healing can effectively be controlled with anti-inflammatory drugs such as steroids. In contrast, the TGF-β effects on tenon fibroblasts in the later wound healing phases, comprising transdifferentiation of fibroblasts into myofibroblasts ⁴² and enhanced deposition of ECM, ⁵ constitute a more difficult therapeutic dilemma. Treatment with antimetabolites such as mitomycin C and 5-fluorouracil is often associated with serious adverse effects. Other, more specific compounds are therefore desirable to control myofibroblast transdifferentiation in the later stages of ocular wound healing. 

In the present study, we provide evidence for MAPK p38 dependence of TGF-β–induced myofibroblast transdifferentiation, proliferation, and ECM production. p38 may serve as a new therapeutic target in wound healing modulation. 

Myofibroblasts, which are characterized by de novo synthesis of α-SMA, exert critical functions in wound healing such as granulation tissue formation ^{43 44} and wound contraction. ¹⁰ To study this process in vitro, HTFs were plated on cell culture dishes and stimulated with TGF-β for up to 72 hours (Fig. 1) . While unstimulated cells did not show relevant α-SMA expression, incubation with TGF-β increased α-SMA mRNA expression after 12 hours and α-SMA protein expression after 24 to 48 hours, similarly to what has been observed in other cell types. ^{42 45 46}  

Two prerequisites for the transition of fibroblasts to myofibroblasts are well established in vitro and in vivo: mechanical stress, ^{29 30 31 36} which induces stress fiber formation leading to a protomyofibroblast phenotype, and the growth factor TGF-β, ^{42 47 48} which is critical for the final accomplishment of myofibroblastic conversion. 

The signaling pathways responsible for TGF-β–induced α-SMA expression during the myofibroblast transdifferentiation process are not completely understood. Blocking platelet-derived growth factor (PDGF) by neutralizing antibodies or blocking integrin signaling by arginine-glycine-aspartic acid (RGD) peptides significantly reduced TGF-β–induced α-SMA expression and actin reorganization. ^{28 29} A TGF-β response element has been identified in the α-SMA promoter. The first 125 bp of the promoter are sufficient for TGF-β responsiveness, but information on upstream signaling pathways is lacking. ⁴⁹  

Therefore, we examined TGF-β–induced signaling pathways potentially involved in α-SMA expression. Members of the MAPK signaling cascades were reported to be relevant for TGF-β signaling in addition to “classical” Smad signaling. ^{22 24 25 50}  

In HTFs, we found temporary activation of the Smad and Erk signaling pathways on TGF-β stimulation, whereas the p38 signaling cascade showed a biphasic activation pattern, with early activation ceasing after 6 hours and a second, sustained activation phase starting after 12 hours (Fig. 1) . This second activation phase coincided with α-SMA transcription and preceded the onset of α-SMA protein expression by 6 to 12 hours, compatible with a possible role of p38 in TGF-β–mediated α-SMA expression. 

It is unclear how the second, delayed p38 activation and the putatively consecutive α-SMA expression were initiated. The time course proposes an indirect mechanism. A potential mediator might be connective tissue growth factor (CTGF). CTGF drew considerable attention as a main cofactor in TGF-mediated fibrotic processes (for a review, see Leask and Abraham ⁵¹ ). TGF-β stimulation increases the expression of CTGF in NRK cells and in HTFs (Duncan et al. ⁵² and unpublished data). Antisense constructs against CTGF efficiently inhibited TGF-β–mediated myofibroblast transdifferentiation, ^{53 54} although CTGF alone was insufficient for conversion. ^{54 55}  

Another mode of TGF-β–mediated delayed p38 activation was reported by Takekawa et al. ²⁷ TGF-β induces GADDF45β, which in turn can activate MTK1, a MAPKKK upstream in the p38 pathway. Takekawa et al. ²⁷ demonstrate that this pathway is responsible for the delayed thrombospondin-1 expression on TGF-β stimulation. Further work is necessary to elucidate the mechanism of delayed p38 activation in our system. 

To test our hypothesis that p38 signaling is necessary for TGF-β–induced α-SMA expression, we used different pharmacologic inhibitors. We observed a dose-dependent reduction of TGF-β–mediated α-SMA expression on p38 inhibition at the mRNA and protein levels (Fig. 2) . These data were further supported by immunocytochemical studies addressing the subcellular distribution of α-SMA protein (Fig. 3) . In unstimulated cells, no α-SMA was found in actin stress fibers, and only faint cytosolic staining was detected. TGF-β induced a marked incorporation of α-SMA into actin stress fibers that was prevented by the p38 inhibitor SB203580. 

In addition to increased α-SMA protein levels, myofibroblasts also show increased contractile activity. Because α-SMA expression correlates with enhanced contractility, ^{2 36} we were interested in studying whether the attenuation of α-SMA expression by p38 inhibitors accompanies reduced contractility. Indeed, the p38 inhibitor SB203580 blocked the TGF-β–induced increase in HTF contractility in a dose-dependent manner (Fig. 4) . 

TGF-β1 is reported to stimulate cell-populated collagen gel contraction in a variety of cells, including primary skin fibroblasts and mouse embryonic fibroblasts, ⁴ corneal fibroblasts, ⁵⁶ and trabecular meshwork cells. ⁴⁵ PKC, myosin light-chain kinase (MLCK), and intracellular calcium are necessary for TGF-β1–induced contractility in trabecular meshwork cells. ⁴⁵ Moreover, in contrast to wild-type fibroblasts, embryonic fibroblasts originating from Smad3 knockout mice failed to contract type 1 collagen gels on TGF-β stimulation. ⁵⁷ Garret et al. ⁵⁴ showed that specific silencing of CTGF expression by antisense oligonucleotides in keratocytes attenuated the increase of contractility and α-SMA-expression after TGF-β stimulation. TGF-β–mediated contraction thus seems to rely on multiple factors and may vary among cell types. 

Increased contractility is associated with morphologic characteristics such as actin stress fibers and a retracted bipolar cell phenotype. ^{1 30 58} In our experiments, these TGF-β–induced morphologic changes could be averted by SB203580 (10 μM). In epithelial cells, TGF-β–mediated induction of actin stress fibers involves p38 and Smad signaling followed by increased expression of tropomyosins, ⁵⁹ thus supporting our observations. 

Fibroblast-populated collagen lattices contract even without exogenous TGF-β stimulation, as demonstrated in Figure 5and by Grinnell. ³² Endogenous TGF-β secretion by fibroblasts is a possible explanation for this phenomenon. TGF-β itself can activate autocrine secretion of growth factors. It has been proposed that PDGF mediates TGF-β–induced proliferation of mesenchymal cells. ⁶⁰ Moreover, TGF-β can positively regulate its own production, referred to as autoinduction, in primary subconjunctival fibroblasts ⁶¹ and other cell types. ⁶² Our observation that p38 inhibitors avert actin condensation and endogenous gel contraction is in accordance with studies in epithelial cells, in which p38 signaling was found to be mandatory for tropomyosin expression and subsequent stress fiber formation. ^{59 63} Moreover, activated Ras and Erk impair tropomyosin expression and stress fiber formation, and inhibition of Erk signaling restores stress fiber formation in Ras-transformed epithelial cells. ^{59 63} This is in accordance with our results of facilitated endogenous gel contraction in the presence of the Erk inhibitor U0126 (Fig. 5) . 

In addition, p38 signaling is involved in TGF-β–initiated epithelial-to-mesenchymal transition (EMT), an important step in tumorigenesis. ^{64 65} Edlund et al. ⁶⁶ showed participation of the p38 signaling pathway in TGF-β–induced cytoskeletal reorganization and formation of stress fibers. Furthermore, the p38 signaling pathway has been implicated in TGF-β–stimulated migration of smooth muscle cells. ⁶⁷ It is thus conceivable that this pathway is also essential for TGF-β–induced myofibroblast transdifferentiation. 

The p38 inhibitor SB203580 blocked TGF-β–induced transcription of collagen Iα1 and fibronectin (Fig. 6) . Studies in skin fibroblasts support these findings because they show an involvement of Smad3 in TGF-β–mediated collagen 1 induction. ^{68 69 70} More recently, Kimoto et al. ^{71 72} reported the inhibition of TGF-β–mediated collagen 1 expression by the p38 inhibitor SB203580 in the RPE cell line ARPE-19. One could speculate that Smad and p38 signaling cooperate in activating the collagen 1 promoter in a direct fashion. Another interpretation is that the inhibition of myofibroblast transdifferentiation reduces ECM production because α-SMA–positive myofibroblasts are known as the main source of ECM proteins. ^{5 6}  

A mitogenic effect of TGF-β on HTF has been reported. ⁷³ We tested a p38 inhibitor in this regard and found a reduction in TGF-β–induced HTF cell proliferation (Fig. 7)that may be beneficial in postoperative wound healing modulation. However, our data point toward an only partial p38 dependence of TGF-β–induced cell proliferation because it was not completely blocked by the p38 inhibitor. This might have been caused by the indirect nature of TGF-induced mitogenesis, which was shown to depend on TGF-β–mediated FGF-2 expression. ⁶¹ The role of p38 in this signaling pathway is unclear. 

In summary, we were able to block different aspects of myofibroblast transdifferentiation with p38 inhibitors: α-SMA-expression, increases in contractility, development of actin stress fibers, transcription of ECM proteins and cell proliferation. These results indicate a critical involvement of p38 signaling in TGF-β–mediated myofibroblast transdifferentiation. Given that the prolonged persistence of myofibroblasts is a cause for postoperative scarring in glaucoma surgery, it is intriguing to modulate the emergence of myofibroblasts by specific pharmacologic intervention. Modulating specific steps of the ocular wound healing process by kinase inhibitors might represent a new approach for avoiding hypertrophic scar formation after fistulating glaucoma surgery. 

⁴  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

 Supported by the Interdisziplinäre Zentrum für Klinische Forschung, Würzburg, Germany.

 Submitted for publication March 22, 2005; revised September 30, 2005; accepted February 14, 2006.

 Disclosure: T. Meyer-ter-Vehn, None; S. Gebhardt, None; W. Sebald, None; M. Buttmann, None; F. Grehn, None; G. Schlunck, None; P. Knaus, None

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

 

Figure 1.

 

Effect of TGF-β on α-SMA expression and intracellular signaling. (A) TGF-β1 upregulated α-SMA mRNA expression. Cells were treated with TGF-β1 (2 ng/mL) for the indicated time periods. α-SMA mRNA abundance was analyzed by real-time RT-PCR, and the housekeeping gene HPRT1 transcript was used for normalization. Data are shown as fold induction compared with unstimulated controls. This experiment was performed 3 times with similar results. (B) TGF-β1 upregulated α-SMA protein expression and activated several signaling pathways. Cells were treated with TGF-β1 (2 ng/mL), as indicated. Subsequently, cell extracts were prepared, and 10 μg total protein was analyzed by Western blot. Activation of p38, Erk, Smad2/3 was detected using phosphospecific antibodies. Antibodies against total proteins or tubulin served as controls. TGF-β1 induced α-SMA expression and caused p38 phosphorylation in a biphasic manner, whereas Erk and Smad signaling pathways were only transiently activated.

Figure 2.

 

TGF-β–mediated α-SMA protein expression was p38 dependent. HTFs were treated with different p38 inhibitors 30 minutes before and during stimulation with TGF-β1 (1 ng/mL), as indicated. After 72 hours, protein extracts were prepared and analyzed by Western blot for α-SMA expression. As a control for equal loading, the blots were reprobed with an antibody specific for β-Actin.

Figure 3.

 

TGF-β–mediated development of α-SMA–positive stress fibers is p38-dependent. HTFs were plated on fibronectin-coated glass coverslips, serum-starved for 16 hours, and stimulated with vehicle (A–C), TGF-β (D–F), or TGF-β and the p38 inhibitor SB203580 (G–I) for 2 days, as indicated. α-SMA expression and localization were analyzed by confocal immunofluorescence microscopy (A, D, G). F-Actin was stained with rhodamine-phalloidin (B, E, H). The composite image (C, F, I) shows α-SMA (green) and F-Actin (red) and allows assessment of α-SMA integration into the actin cytoskeleton.

Figure 4.

 

TGF-β–mediated contraction of fibroblast-populated collagen lattices is p38-dependent. HTFs were seeded in neutralized collagen solution and incubated with vehicle, TGF-β (2 ng/mL), or TGF-β (2 ng/mL) in the presence of the p38 inhibitor SB203580 (10 μM) for 48 hours, as indicated. Gels were detached, and subsequent gel contraction was digitally photodocumented (A) and measured as a reduction of gel surface 3 hours after detachment (B). (C–E) Cell morphology in contracted collagen gels. After 2 days of contraction, collagen lattices were fixed, and F-Actin was stained with rhodamine-phalloidin and viewed with a laser scanning microscope. Arrow indicates stress fibers. Scale bar, 50 μm.

Figure 5.

 

Endogenous contraction of fibroblast-populated collagen lattices is p38 dependent. HTFs were seeded in neutralized collagen solution and were incubated with the MEK1/2 inhibitor U0126 or the p38 inhibitor SB203580 for 48 hours, as indicated. Gels were detached, and subsequent gel contraction was digitally photodocumented (A) and measured as a reduction of gel surface 3 days after detachment (B). (C–E) Cell morphology in contracted collagen gels. After 3 days of contraction, collagen lattices were fixed, and F-Actin was stained with rhodamine-phalloidin and viewed with a laser scanning microscope. Arrows indicate stress fibers. Scale bar, 20 μm.

Figure 6.

 

p38 inhibitors prevented TGF-β–induced extracellular matrix (ECM) protein transcription. HTFs were serum starved for 16 hours, pretreated with vehicle or SB203580 (10 μM) for 30 minutes, and stimulated with vehicle, TGF-β1 (2 ng/mL), or TGF-β1 and the p38 inhibitor SB203580 (10 μM) for 24 hours. Collagen Iα1 and fibronectin transcription were assessed by real-time PCR, and the housekeeping gene HPRT1 transcript was used for normalization. α-SMA was analyzed for comparison. Data are shown as fold induction compared with unstimulated controls.

Figure 7.

 

p38 inhibitors reduced TGF-β–induced HTF proliferation. HTFs were serum-starved for 16 hours, pretreated with 10 μM SB203580 or vehicle, as indicated, and stimulated for 48 hours with vehicle, TGF-β1 (2 ng/mL), or 10% FCS in the presence or absence of the p38 inhibitor SB203580 (10 μM). Cell proliferation was assessed using an MTT-based proliferation assay (CellTiter). Each data point represents 5 independent measurements.

The authors thank Barbara Katzenberger, Katherina Knoke, and Nadine Kehl for excellent technical assistance. They also thank Michael Sendtner and coworkers at the Institute for Clinical Neurobiology for sharing equipment and reagents.