Transforming Growth Factor_β–Mediated Corneal Myofibroblast Differentiation Requires Actin and Fibronectin Assembly

Abstract

purpose. Recent studies indicate that transforming growth factor (TGF)_β is a potent inducer of corneal myofibroblast differentiation and expression of smooth muscle–specific, α-actin (α-SMA). Although TGF_β is known to enhance synthesis of extracellular matrix proteins and receptors, little is known about how it modulates the expression of smooth muscle proteins in nonmuscle cells. The purpose of this study was to identify the role of Arg-Gly-Asp (RGD)-dependent tyrosine phosphorylation in regulatingα -SMA gene expression and ultimately myofibroblast development.

methods. Because cell culture in serum-containing media mimics myofibroblast transformation, all experiments were performed on freshly isolated rabbit keratocytes plated in defined, serum-free media. Cells were exposed to TGF_β (1 ng/ml), Gly-Arg-Gly-Asp-D-Ser-Pro (GRGDdSP, 50 μM), Gly-Arg-AL-Asp-Ser-Pro (GRADSP; 100 μM), or herbimycin A (0.1–10 nM) at 24 hours (sparse) or 7 days (confluent). Cells were evaluated by immunocytochemistry and proteins and RNA collected for western and northern blot analyses using antibodies specific for α-SMA, fibronectin, focal adhesion proteins, and phosphotyrosine (clones 4G10 and PY20); and probes directed against rabbit α-SMA. All experiments were repeated at least three times.

results. Keratocytes exposed to TGF_β showed expression of α-SMA that coincided with the intracellular reorganization of the actin cytoskeleton and the extracellular assembly of fibronectin fibrils. Addition of RGD containing but not control peptides blocked the organization of intracellular actin, extracellular fibronectin, andα -SMA protein and mRNA. Immunoprecipitation of cell proteins with 4G10 or PY20 identified the TGF_β-associated tyrosine phosphorylation of paxillin, pp125^fak, p130, PLCγ, and tensin, which was blocked by addition of GRGDdSP. Addition of herbimycin A to keratocytes exposed to TGF_β showed a dose-dependent loss of α-SMA protein and mRNA which correlated with loss of tyrosine phosphorylation, absence of actin reorganization, and fibronectin assembly.

conclusions. The data suggest that TGF_β-mediated α-SMA gene expression leading to myofibroblast transformation may involve an RGD-dependent phosphotyrosine signal transduction pathway.

During the process of wound healing, invading corneal fibroblasts exhibit unique ultrastructural and physiological characteristics similar to smooth muscle cells, including prominent intracellular microfilament bundles and in vitro contractile responses to smooth muscle agonists. ^{1 2} These characteristics have been used to define these cells phenotypically as myofibroblasts. It has been further recognized that corneal myofibroblast differentiation involves the expression of the α-isoform of actin, specific for vascular smooth muscle cells. ³  

Many aspects of myofibroblast function have yet to be understood fully; however, of particular interest is the regulation and function of smooth muscle–specific actins in nonmuscle fibroblastic cells. There are two classes of differentially expressed actin isoforms present in mammalian cells, comprising the nonmuscle actins β and γ (class 1) and the three isoforms of α actins present in cardiac, skeletal, and vascular muscle (class 2). ⁴ Isoactins show considerable sequence homology, differing predominantly at the NH₂-terminal sequence for which α-SMA contains an Ac-EEED sequence important in α-SMA polymerization. ⁵ In general, α-actins appear to localize preferentially to microfilament bundles—that is, stress fibers, rather than the actin-rich cortex or lamellipodial ruffles that contain nonmuscleβ and γ actin. ⁶ α-SMA can be selectively removed from stress fibers by microinjection of either monoclonal antibodies specific for the Ac-EEED epitope or by the injection of small peptides containing the Ac-EEED sequence. ⁵ Displacement of α-SMA from stress fibers leads to physiologically enhanced cell motility and loss of focal contacts, suggesting that α-SMA plays a role in regulating cell adhesivity. ⁷  

Expression of α-SMA is both developmentally and environmentally regulated. It appears in high concentration early in development of cardiac and skeletal muscle, disappears later, ^{8 9} and is expressed at the leading edge of migrating neural crest cells invading the primary corneal stroma. ¹⁰ During wound healing, expression is exclusively localized to cells within the corneal wound and excluded from undamaged regions, even in migrating cells containing stress fibers. ³ In studies in vitro,α -SMA expression in nonmuscle cells has been shown to be cell-density dependent, ^{11 12} influenced by cell substrate composition, ¹³ and sensitive to regulation by various cytokines. ^{14 15 16}  

Recently, interest has focused on transforming growth factor (TGF)_β as a potent inducer of α-SMA expression and myofibroblast differentiation of dermal, ¹⁶ breast, ¹⁷ palatal, ¹⁸ lung, ¹¹ and corneal stromal fibroblasts ¹⁹ and of other nonfibroblastic cells, including lens epithelial and liver Ito cells. ^{20 21} TGF_β plays an important role in the wound healing response and has been shown to: enhance the synthesis and deposition of extracellular matrix proteins, including type I collagen and fibronectin ^{22 23 24} ; decrease the degradation of extracellular matrix by decreasing the synthesis of matrix metalloproteinases (e.g., collagenase and stromelysin) and enhance the synthesis of protease inhibitors ^{25 26} ; and promote cell–matrix interaction by upregulating the synthesis of membrane surface receptors including α5β1 integrin. ^{27 28} Moreover, inactivation of TGF_β by the application of neutralizing antibodies to dermal or corneal wounds in vivo blocks the development of fibrosis or scarring, inhibits the deposition of collagen and fibronectin, and overall, decreases the number of wound-healing fibroblasts or myofibroblasts. ^{29 30 31 32}  

More recent studies indicate that the ED-A domain of fibronectin is crucial for induction of α-SMA expression by TGF_β, suggesting an outside–in signaling mechanism involving extracellular fibronectin. ³³ In the present study we used a serum-free culture system to evaluate the effect of fibronectin assembly and Arg-Gly-Asp (RGD)-dependent signaling on the TGF_β-mediated expression ofα -SMA and myofibroblast differentiation of rabbit corneal keratocytes. Overall, the data indicate that the upregulation ofα -SMA expression by TGF_β involves RGD-dependent, tyrosine phosphorylation consistent with an outside–in signal transduction pathway and environmental control of myofibroblast differentiation. 

Materials and Methods

Cell Culture

Whole rabbit eyes were obtained from Pel Freez (Rogers, AR). The surface epithelium was initially scraped from the cornea with a no. 10 Bard Parker blade (Lance, Sheffield, UK), the corneas excised, and the endothelium removed with a sterile cotton-tipped applicator soaked in ethanol. The corneas were then digested in sterile 2.0 mg/ml collagenase (Gibco, Gaithersburg, MD) and 0.5 mg/ml hyaluronidase (Worthington, Freehold, NJ) in minimum essential medium (MEM; Gibco) overnight at 37^oC. Corneal keratocytes were then plated in MEM in serum-free media supplemented with RPMI vitamin mix and glutathione, nonessential amino acids, pyruvic acid, 1% glutamine and penicillin-streptomycin (Gibco) at 5.0 × 10⁴ cells/cm². Keratocytes were plated onto 100- mm dishes (Falcon Primaria; Becton Dickinson, Lincoln Park, NJ) for biochemical analyses or 12-mm diameter glass coverslips coated with collagen (Vitrogen; Collagen, Freemont, CA) for immunocytochemical localization. Primary serum-free cultures were used exclusively in all experiments. 

To evaluate the temporal effects of TGF_β on myofibroblast differentiation, primary keratocytes were treated with 1 ng/ml TGF_β1 (Gibco) after overnight plating or after 7 days in culture to allow cells to establish cell–cell contacts before exposure to TGF_β. Cells were then evaluated between 8 and 72 hours after exposure. To evaluate the effects of RGD-dependent cell–matrix interactions, cells were plated overnight and then treated with TGF_β1 (1 ng/ml) in combination with the following peptides from Gibco: Gly-Arg-Gly-Asp-Asn-Pro (GRGDNP, 10–500 μM), Gly-Arg-Gly-Asp-D-Ser-Pro (GRGDdSP, 50 μM), Gly-Arg-AL-Asp-Ser-Pro (GRADSP, 50–100 μM), and Gly-Pen-Gly-Arg-Gly-Asp-Ser-Pro-Cys-AL (GpenGRGDSPCA, 1–1000 μM). Cells were then cultured for 72 hours and evaluated. To study the effects of herbimycin A (Gibco), a tyrosine phosphokinase inhibitor, cells were plated overnight and then exposed to TGF_β.(1 ng/ml) in combination with herbimycin A at various concentrations from 0.1 nM to 600 nM. All conditions were evaluated in triplicate, and experiments were repeated at least two times. 

Immunocytochemistry

Keratocytes grown on glass coverslips were rinsed once in phosphate-buffered saline (PBS; pH 7.4) fixed in 1% paraformaldehyde in PBS, rinsed in PBS, and extracted in cold acetone (−20^oC). Cells were rehydrated in PBS and nonantigenic sites blocked by incubating with 1% ovalbumin (Sigma, St. Louis, MO) or 10 μg/ml goat serum (Cappel, Durham, NC). Cells were then reacted with primary antibodies including, mouse monoclonal anti-α-SMA (1:100), clone 1A4 (Sigma); mouse monoclonal anti-human vinculin (1:100), clone V284 (Serotec, Washington, DC); and fluorescein isothiocyanate or rhodamine-conjugated goat anti-human fibronectin (1:20; Binding Site, San Diego, CA) for 60 minutes. Cells were then washed in PBS and reacted with appropriate affinity-purified fluorescein-conjugated goat anti-mouse IgG (1:20; Cappel). Cells were also doubled labeled with rhodamine-conjugated phalloidin (Molecular Probes, Eugene, OR) to identify f-actin filaments. Cells were then washed in PBS, mounted onto glass slides using a 1:1 solution of glycerol-PBS containing 1 mg/ml phenylenediamine (Sigma), and observed with an epifluorescence microscope (Diaplan; Leica, Deerfield, IL). 

Western Blot Analysis

Proteins for western blot analysis were solubilized in buffer containing 25 mM Tris-HCl (pH 7.4) containing 1 mM EDTA, 1 mM EGTA, 10 mM dithiothreitol, 1% sodium dodecyl sulfate (SDS), 5 μg/ml antipain, 5 μg/ml pepstatin A, and 1 mM phenyl methyl sulfonyl fluoride (PMSF). Cells were kept on ice and scraped using a rubber policeman. After protein determination, samples were boiled for 5 minutes and run on a 10% acrylamide gel to identify α-SMA and on a 7.5% acrylamide gel to identify tyrosine-phosphorylated proteins. Proteins were then transferred to nitrocellulose paper, blocked with 5% dry milk in Tris-saline, and incubated for 2 to 3 hours with monoclonal antibody to α-SMA (Sigma) or antibodies to phosphotyrosine (clone 4G10; UBI, Lake Placid, NY). The nitrocellulose paper was then washed in Tris-saline and incubated in horseradish peroxidase–conjugated goat anti-mouse IgG (1:1000) for 1 hour (Cappel). Proteins were then visualized by enhanced chemiluminescence reagent (ECL; Amersham, Arlington Heights, IL). 

Immunoprecipitation of Tyrosine-Phosphorylated Proteins

Immunoprecipitation of tyrosine-phosphorylated proteins was performed on cell lysates diluted to 1 μg/μl, total cell protein, with immunoprecipitation buffer (1% Triton X-100, 150 mM NaCl, 10 mM Tris [pH 7.4], 1 mM EDTA, 1 mM EGTA, 0.2 mM sodium vanadate, 0.2 mM PMSF, and 0.5% NP-40). Tyrosine-phosphorylated proteins were precipitated with 1 to 5 μg 4G10 or PY20 (ICN Biomedicals, Costa Mesa, CA) monoclonal antibody. Immunoprecipitated proteins were run on 7.5% polyacrylamide one-dimensional sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel, transferred to nitrocellulose paper, and probed with anti-chicken tensin (p210; UBI), anti-p130 (Transduction Laboratories, Lexington, KY), anti-FAK (Transduction Laboratories), anti-paxillin (ICN Biomedicals), and anti-PLCγ (Transduction Laboratories). 

Generation of a cDNA Probe for Rabbit α-SMA mRNA

Rabbit corneal keratocyte RNA was used as a template for RT-PCR reactions to produce a cDNA product directed toward α-SMA mRNA. Primers used to generate RT-PCR products were designed against regions of the rabbit α-SMA mRNA sequence extending upstream from base pair 607 to 630 (5′-GTGACTACTGCTGAACGTGAGATT-3′) and downstream from base pair 1157 to 1180 (5′-TAGCCCACAACTGTGAATGTGTTG-3′). ³⁴ The PCR reaction product was then cloned into a vector (pCR II-TOPO; Invitrogen, Carlsbad, CA), sequenced to confirm the identity, and ³²P labeled using random-primer labeling (Stratagene, La Jolla, CA). 

Northern Blot Hybridization

RNA was extracted from cultured keratocytes by using the RNazol method (TRI reagent; Molecular Research, Cincinnati, OH). RNA was then electrophoresed (10 μg/lane) through an agarose–formaldehyde gel, transferred overnight onto a nylon membrane (GeneScreen; New England Nuclear, Boston, MA) in 20× SSC, rinsed in 2× SSC, UV cross-linked, and dried in an 80^oC oven for 2 hours. Membranes were prehybridized in 50% formamide, 5× SSC, 0.1% SDS, and 5× Denhardt’s reagent at 42^oC for 8 hours and hybridized to ³²P-labeled probes overnight at 42^oC. Blots were washed three times in 0.1× SSC and 0.1% SDS at 42^oC and autoradiographed. 

Results

Temporal Response of Corneal Keratocytes to TGF_β

Primary corneal keratocytes cultured under defined, serum-free conditions maintained a dendritic morphology (Fig. 1 A) characteristic of in situ keratocytes. ³⁵ The cytoskeletal organization of actin within these cells showed a predominantly a cortical localization (Fig. 1B) with few, if any, stress fibers and no focal adhesions (Fig. 1C) . Interestingly, keratocytes grown under these defined serum-free conditions failed to express α-SMA (Fig. 1D) , whereas approximately 10% of keratocytes grown in the presence of serum show α-SMA expression, ¹⁹ a prevalence similar to that observed in cultured dermal fibroblasts. ¹⁴ When exposed to 1 ng/ml TGF_β, serum-free cultured keratocytes appeared to retract their dendritic processes and extend fewer, but broader, cellular processes, assuming a more fibroblastic cell shape (Fig. 1E) that was apparent as early as 24 hours after exposure to TGF_β in confluent cultures. Concomitant with the marked change in cell shape there was a striking reorganization of the f-actin into prominent stress fibers (Fig. 1F) that terminated at focal adhesion complexes, visualized by anti-vinculin antibody staining (Fig. 1G , arrows). Reorganized actin filaments also showed staining with antibodies to α-SMA (Fig. 1H) , suggesting the development of a myofibroblast phenotype. 

Temporally, the transition of keratocytes to a myofibroblastic phenotype morphology and the appearance of stress fibers within TGF_β-modified cells preceded by 24 to 48 hours the appearance of α-SMA, which was not immunocytochemically detectable until 72 hours after treatment (Fig. 2 A, 2B, 2C). Interestingly, it is during this early phase, before the appearance of α-SMA, that fibronectin, synthesized by keratocytes in response to TGF_β, ¹⁹ is deposited and organized into fibronectin fibrils within the extracellular matrix (Fig. 2E 2F) . As shown previously, ¹⁹ keratocytes cultured under defined serum-free conditions synthesize and deposit little if any fibronectin which after 4 hours of exposure to TGF_β is limited to sparse, punctate deposits and rare, poorly formed fibronectin fibrils (Fig. 2D , arrows). By 24 hours there was increased fibronectin deposited within the extracellular matrix that was organized into fibrils running parallel to cellular processes (Fig. 2E , arrows). With continued culture there was increasingly more fibrillar fibronectin deposited around the cells (Fig. 2F , arrows). 

Effects of Fibronectin Fibril Assembly and RGD Binding on α-SMA Expression

The finding that fibronectin was synthesized, deposited, and organized into fibrils either preceding or coincident with the appearance of anti-α-SMA staining suggests that fibronectin may play a role in modulating the expression of α-SMA. This possibility is also supported by the in vivo finding that α-SMA expression in wound healing keratocytes is exclusively localized to the wound, ³ suggesting that environmental, extracellular matrix factors unique to the wound, and not solely soluble cytokines, play an important role in regulating α-SMA expression. Cell binding to fibronectin is mediated by cell surface integrin receptor recognition of the RGD sequence of fibronectin in the cell-binding domain. ³⁶ Addition of soluble RGD-containing peptides is known to block this cell–fibronectin interaction and thereby prevent cell attachment and fibronectin fibril assembly. ^{36 37} When peptides containing the RGD sequences were added with TGF_β to primary serum-free cultured keratocytes, there was complete inhibition of TGF_β-mediated change in cell shape, actin reorganization, and fibronectin fibril assembly (Fig. 3 , Table 1 ). TGF_β treated cells cultured in the presence of either GRGDNP, which prevents cell attachment to fibronectin and vitronectin, or GRGDdSP, which prevents cell attachment to fibronectin, maintained a dendritic, keratocyte morphology (Fig. 3A) and neither α-SMA expression (Fig. 3B) or stress fibers and fibronectin fibrils developed (data not shown). TGF_β upregulation of α-SMA was not effected by the addition of either control, non-RGD containing peptides, GRADSP (Fig. 3C) , or the RGD-containing peptide, GPenGRGDSPCA, that blocks cell binding to vitronectin but not fibronectin (Fig. 3D) . Overall, the effect of RGD peptides was dose dependent with maximal inhibition of α-SMA expression observed at concentrations that were an order of magnitude below the required dose for inhibition of cell attachment (Table 1) . ³⁶  

The effect of RGD-containing peptides was further verified by immunoblot analysis for α-SMA. In control, untreated keratocyte cultures there was no detectable α-SMA protein (Fig. 4 A, lane 1), whereas cultures treated with 1 ng/ml TGF_β for 3 days showed staining of a prominent protein band that migrated at approximately 45 kDa (Fig. 4A , lane 2), consistent with smooth muscle actin. Cultures treated with GRGDNP (50 μM) in combination with TGF_β (1 ng/ml) showed no detectable levels of α-SMA protein (Fig. 4A , lane 3), similar to that observed in the untreated control cultures. Cultures treated with the control peptide, GRADSP at 100 μM (Fig. 4A , lane 4), showed a level of expression of α-SMA similar to that observed in cultures treated with TGF_β alone. Northern blot analysis of RNA isolated from parallel cultures showed low-level binding of the α-SMA cDNA probe to a 1.7-kb RNA species of a size consistent with that expected for rabbit α-SMA (Fig. 4B , lane 1). When keratocytes were treated with TGF_β alone (1 ng/ml) for 3 days, there was a marked increase in the level of binding of the α-SMA cDNA (Fig. 4B , lane 2), comparable to the level of binding detected in RNA from serum-cultured keratocytes (Fig. 4B , lane 4). When keratocytes were cotreated with GRGDdSP (50 μM) and TGF_β (1 ng/ml), binding was substantially reduced to levels observed in control, untreated cells. These findings suggest that the upregulation of α-SMA mRNA induced by TGF_β is blocked when the binding of fibronectin to surface membrane receptors is inhibited and that the decrease inα -SMA protein observed after GRGDdSP treatment may be caused by a block in the upregulation of α-SMA message by TGF_β. However, because the isoforms of actin show considerable sequence homology, further work is necessary to demonstrate gene regulation at the transcriptional level. 

Tyrosine Phosphorylation and Myofibroblast Differentiation

Corneal keratocytes maintained under serum-free conditions for 7 days showed no detectable tyrosine-phosphorylated proteins (Fig. 5 A, lane 1), whereas keratocytes grown for 7 days and then treated with TGF_β (1 ng/ml) for 3 days showed tyrosine phosphorylation of at least five proteins (Fig. 5A , lane 2). Immunoprecipitation of the tyrosine-phosphorylated proteins using the antibody 4G10 (Fig. 5B , lane 1) followed by immunostaining for specific proteins, identified the five proteins to be: tensin (lane 2), PLCγ (lane 3), p130 (lane 4), pp125^fak (lane 5), and paxillin (lane 6). The two bands identified for p130 are consistent with the presence of both phosphorylated (130-kDa) and unphosphorylated (125-kDa) forms, ³⁸ whereas the lower molecular weight bands stained by anti-tensin antibodies most likely represent degradation products. Blots stained with antibodies to irrelevant proteins (actin or α-SMA, lane 7) showed no staining, indicating that the immunoprecipitates contained predominantly tyrosine-phosphorylated protein complexes. When keratocytes were treated with TGF_β and GRGDdSP (50 μM), the RGD-containing peptides blocked the tyrosine phosphorylation (Fig. 5A , lane 3). Control peptides, GRADSP (Fig. 5A , lane 4), had no effect on TGF_β-related tyrosine phosphorylation of proteins. 

Inhibition of Tyrosine Phosphorylation by Herbimycin A

In situ fibronectin fibril assembly is principally a cell-mediated event involving interactions between extracellular matrix fibronectin and integrin cell surface receptors. ³⁹ Integrins are a large family of heterodimeric transmembrane receptors comprised of α and β subunits that mediate cell adhesion to extracellular matrix. ⁴⁰ Signal transduction mediated by integrin receptors involves initial clustering of receptors and recruitment of tensin and pp125^fak followed by tyrosine phosphorylation and focal accumulation of actin, actin-binding proteins, and at least 19 signal transduction molecules in the formation of focal adhesions. ^{41 42} Inhibition of tyrosine phosphorylation by herbimycin A, an inhibitor of the Src family of tyrosine kinases, not only blocks aggregation of signaling molecules but also inhibits the formation of focal adhesions and stress fibers. ⁴³  

We therefore evaluated the role of focal adhesion assembly in mediating myofibroblast differentiation by exposing keratocytes to herbimycin A in combination with TGF_β. Treatment of serum-free cultured keratocytes with herbimycin A alone at a range of concentrations appeared to have no effect on keratocyte morphology or actin organization. However, when herbimycin A at 600 nM, a dose previously shown to block tyrosine phosphorylation of paxillin and pp125^fak in fibroblasts, ⁴³ was added simultaneously with TGF_β (1 ng/ml) to 7-day-old cultures, herbimycin A completely inhibited the formation of stress fibers and focal adhesions (data not shown). When herbimycin was removed from the culture media, continued treatment with TGF_β led to stress fiber and focal adhesion formation, indicating that the effect of herbimycin A was reversible and not toxic. The lowest dose of herbimycin A that inhibited stress fiber formation (Fig. 6f -Actin) and focal adhesion assembly (Fig. 6 , Vinculin) after TGF_β treatment (1 ng/ml) was 10 nM. At lower doses there was partial (1.0 nM) to no (0.1 nM) inhibition of the stress fiber and focal adhesion formation induced by TGF_β, indicating a dose-response effect. 

As expected, the loss of stress fibers and focal adhesions was also associated with the concomitant loss of α-SMA expression in a dose-dependent fashion (Fig. 6 , α-SM). Western blot analysis of proteins (Fig. 7 A) obtained from TGF_β alone (lane 1) and TGF_β in combination with herbimycin A at 0.1 nM (lane 2), 1.0 nM (lane 3), and 10 nM (lane 4) confirmed that the loss of α-SMA protein expression was dose dependent and correlated with the observed block of actin reorganization and focal adhesion assembly. Furthermore, northern blot analysis of RNA from similarly treated cultures (Fig. 7B) also showed a dose-dependent loss in binding of theα -SMA cDNA, consistent with the inhibition of α-SMA mRNA expression by herbimycin A in TGF_β-treated cultures. Overall, the data support the observations made when using RGD peptides and indicate that actin reorganization and focal adhesion assembly play an important role in downstream modulation of α-SMA expression. 

Discussion

This study suggests that the induction of corneal myofibroblast differentiation and expression of α-SMA is environmentally modulated; explaining the unique localization of this cell type to regions associated with corneal matrix organization and wound contraction. Specifically, TGF_β appears to enhance the biosynthetic activity of cells, leading to synthesis of extracellular matrix proteins including fibronectin. ¹⁹ Later, RGD-dependent interactions between the extracellular matrix and the appropriate cell surface receptors lead to actin reorganization, focal adhesion formation, and extracellular matrix assembly, which appear to precede or occur coincident with expression of α-SMA. Blockage of either matrix assembly using RGD peptides or focal adhesion and stress fiber formation using herbimycin A inhibits α-SMA protein expression and reduces mRNA levels that bind cDNA probes for α-SMA. Overall, these findings support the view that TGF_β regulates myofibroblast differentiation by modifying important cell–matrix signaling pathways controlling extracellular matrix assembly, cell shape, and actin organization that lead to expression ofα -SMA consistent with the downstream regulation of gene transcription. 

The finding that extracellular matrix interactions are important in regulating the expression of α-SMA and myofibroblast differentiation in corneal keratocytes is consistent with previous work by Newcomb and Herman, ¹³ showing that matrix synthesized by the vascular endothelium and defined extracellular matrix constituents modulatedα -SMA protein and mRNA expression in cultured pericytes. They are also consistent with the more recent observations of Serini et al. ³³ who showed that the ED-A domain of fibronectin was crucial for the TGF_β-mediated induction ofα -SMA expression in dermal fibroblasts. Cell density has also been shown to modulate α-SMA expression, with both high cell density ¹¹ and very low cell density appearing to increase expression. ¹² However, using serum-free culture conditions in the present experiments, neither density nor original substrate conditions influenced expression of α-SMA. Maintaining cells in culture for 7 days to establish cell–cell contacts, plating on fibronectin and collagen, or wounding cultures failed to induce α-SMA expression in serum-free cultured primary keratocytes (unpublished results). Only treatment of serum-free cells with TGF_β, regardless of cell density, age, or substrate, induced a change in expression. 

The absence of response of serum-free cultured keratocytes to environmental conditions known to modulate α-SMA expression is most likely related, in part, to the types of extracellular matrix receptors expressed by these cells. Although both in situ corneal keratocytes and serum-free cultured keratocytes show no detectable expression of the fibronectin receptor (i.e., α5β1 or αvβ3 integrin, ^{3 35 44} ) there is a marked upregulation of both integrins on exposure to serum or TGF_β in the absence of serum. ^{19 44} Both α5β1 and αvβ3 localize to focal adhesions and are involved in the assembly of fibronectin fibrils. ^{45 46} Interestingly, in wound tissue myofibroblasts, α5β1 is also an important constituent of the fibronexus junctions that link intracellular actin to extracellular matrix, ^{47 48} forming a putative contractile apparatus that exerts mechanical force and plays a role in wound matrix organization and wound contraction. ⁴⁹ Induction of α5β1–αvβ3 integrin expression by TGF_β may therefore play a physiologically critical role in the downstream signaling of α-SMA expression and myofibroblast differentiation by establishing a mechanochemical signaling pathway. ⁵⁰ Such a pathway may involve tyrosine phosphorylation of pp125^fak and p130 and downstream activation of the ras/MAP kinase signal transduction pathways. ⁴¹ Clearly, additional work is necessary to understand more fully the role of integrins in regulating α-SMA expression. 

Although TGF_β’s effects on integrin and extracellular matrix expression may be important, other factor(s) undoubtedly play a role in regulating α-SMA expression and myofibroblast differentiation. As shown in various fibroblastic cells, serum starvation results in loss of focal adhesion and stress fibers that rapidly reappear on addition of serum or other factors, including LPA and bombesin, a response shown to be dependent on Rho activation. ⁵¹ More important, the activation of Rho has been shown to be necessary in focal adhesion assembly, downstream gene transcription in the presence of integrin-extracellular matrix interaction, ⁵² and cell contractility. ⁵³ Therefore, expression of extracellular matrix and appropriate matrix receptors by TGF_β may not be sufficient by themselves to initiate actin–focal adhesion assembly and downstream signaling. Overall, the activation and signaling by Rho is complex and involves, in part, the activation of G-protein–coupled receptors, leading to activation of phospholipase C, which in turn activates a tyrosine kinase that signals downstream conversion of Rho-guanosine diphosphate to Rho-guanosine triphosphate, a process also controlled by guanine exchange factors. ⁵⁴ Activated Rho may then induce the tyrosine phosphorylation of pp125^fak, p130, and paxillin ⁵⁵ and phosphorylation of myosin light-chain kinase, ⁵³ thus potentially regulating both focal adhesion formation and contractility, respectively. 

Finally, studies of myofibroblast differentiation using low-density cell culture suggest that cell–cell contacts may play a role inα -SMA expression. ^{12 56} Fibroblasts expressing α-SMA also appear to show upregulation of cadherins, suggesting important changes in cell–cell adhesion complexes in myofibroblastic cells. Interestingly, myofibroblasts in corneal wound tissue show both the upregulation of connexin 43, shown by immunostaining ⁴⁹ and the presence of functional gap junctions, shown by dye injection. ⁵⁷ Myofibroblasts within wounds appear to maintain a remarkable degree of interconnected structure, with actin filaments appearing at times to extend between cells, supporting the contention that cadherin junctions may play an important role in myofibroblast function in wound contraction. Whether communication mediated through gap junctions or other intercellular adhesions plays a role in regulating myofibroblast differentiation has yet to be determined. 

 Reprint requests: James V. Jester, University of Texas, Southwestern Medical Center at Dallas, 5323 Harry Hines Boulevard, Dallas, TX 75235.

 Supported by Grants EY07348 and EY10556 and Senior Scientist Awards (JVJ, HDC) from the National Institutes of Health; a Manpower Award (WMP) and by an unrestricted grant from Research to Prevent Blindness.

 Submitted for publication November 2, 1998; revised February 23, 1999; accepted April 1, 1999.

 Proprietary interest category: N.

 

Figure 1.

 

Serum-free cultured primary corneal keratocytes cultured in absence (A through D) and presence (E through H) of TGF_β (1 ng/ml) for 3 days. Cells are shown using phase microscopy (A, E) or after staining with phalloidin (B, F), anti-vinculin (C, G), or anti-α-SMA (D, H). Stress fibers stained by phalloidin, focal adhesions (arrows) stained by anti-vinculin antibodies, and anti-α-SMA staining of stress fibers were only detected in the TGF_β-treated cells. Bars, (A, E) 100 μM; (B –through D, F through H) 25 μM.

Figure 2.

 

Immunofluorescent localization of anti-α-SMA staining (A, B, C) and anti-fibronectin staining (D through F) in confluent keratocytes (7-day cultures) treated with TGF_β (1 ng/ml) for 4 hours (A, D), 24 hours (B, E), and 72 hours (C, F). Note that fibronectin fibril assembly (arrows, D–F) can be detected before staining of stress fibers with anti-α-SMA antibodies (C, F). Bar, 25 μM.

Figure 3.

 

Phase contrast (A) and anti-α-SMA immunofluorescent (B through D) micrographs of keratocytes treated with TGF_β (1 ng/ml) in combination with 50 μM GRGDdSP (A, B), 100 μM GRADSP (C), and 1000μ M GPenGRGDSPCA (D). Note that the addition of GRGDdSP blocked the effect of TGF_β on cell shape and density andα -SMA staining, whereas the control peptide GRADSP and the peptide GPenGRGDSPCA, which blocks adhesion to vitronectin, had no effect. Bars, (A) 100 μM; (B through D) 25μ M.

Table 1.

 

Effect of Fibronectin Receptor Blockers on α-SM Actin Expression

Inhibitor	Conc	Inhibition
				Cell Attachment (Fn/Vn)*	α-SM Actin Expression
Peptide Sequences
GRGDNP		(++++/+), †
	10 μM	−	++
	50 μM	−	++++
	100 μM	−	++++
	500 μM	++++	CD
GRGDdSP		(+++/−)
	50 μM	−	++++
GRADSP		(−/−)
	50 μM	−	−
	100 μM	−	−
GRGE		(−/−)
	100 μM	−	−
GPenGRGDSPCA		(−/++++)
	1 μM	−	−
	10 μM	−	−
	100 μM	−	−
	1000 μM	−	−

 Fn, fibronectin; Vn, vitronectin; CD, Cell detachment

*  Results in parentheses taken from Pierschbacher and Ruoslahti. ³⁶

†  ++++ strong inhibition; ++ moderate inhibition; + Weak Inhibition;− , no inhibition.

Figure 4.

 

(A) Western blot (lanes 1 through 4) and corresponding Coomassie blue–stained proteins (lanes 5 through 8) showing expression ofα -SMA in untreated keratocytes (lanes 1 and 5), compared with keratocytes treated with TGF_β (1 ng/ml) alone (lanes 2 and 6) or in combination with GRGDdSP (50 μM; lanes 3 and 7) and GRADSP (100 μM; lanes 4 and 8). Note that GRGDdSP completely blocked the expression of α-SMA. (B) Northern blot analysis of keratocyte RNA obtained from cultures treated with serum-free medium alone (lane 1), TGF_β at 1 ng/ml (lane 2), TGF_β at 1 ng/ml, and GRGDdSP at 50 μM (lane 3), and 10% fetal bovine serum (lane 4). Blots were probed with a 525 bp cDNA for rabbit α-SMA message. In the lower panel, ethidium bromide staining of gels shows equal loading of RNA.

Figure 5.

 

(A) Western blot of proteins from samples obtained in experiment shown in Figure 4A but stained with monoclonal anti-phosphotyrosine antibody, clone 4G10. Keratocytes were either untreated (lane 1) or treated with TGF_β (1 ng/ml) for 3 days, alone (lane 2) or in combination with 50 μM GRGDdSP (lane 3) or 100 μM GRADSP (lane 4). (B) Identification of tyrosine-phosphorylated proteins in keratocytes after 3 days of treatment with TGF_β (1 ng/ml). Proteins were initially extracted and then immunoprecipitated using antibodies (clone 4G10) specific for phosphorylated tyrosine residues. At least five tyrosine-phosphorylated proteins were immunoprecipitated by anti-phosphotyrosine antibodies (lane 1) that had apparent molecular weights of 200 kDa, 150 kDa, 130 kDa, 125 kDa, and 65 kDa. Reaction of immunoprecipitated proteins with various antibodies to focal adhesion–associated proteins identified positive staining for antibodies to tensin (lane 2), PLCγ (lane 3), p130 (lane 4), pp125^fak (lane 5), and paxillin (lane 6). Staining of immunoprecipitated proteins with antibodies to α-SMA (present in original cell lysates) was negative (lane 7).

Figure 6.

 

Effect of herbimycin A (0.1 nM, 1.0 nM, and 10 nM) on keratocytes treated simultaneously with TGF_β (1 ng/ml) for 3 days. Cells were evaluated for α-SMA immunolocalization (α-SMA), stress fiber formation (f-Actin), and focal adhesion assembly (Vinculin). Herbimycin at low concentration (0.1 nM) when added with TGF_β to serum-free keratocytes had no effect on anti-α-SMA staining, stress fiber formation, or focal adhesion assembly. At 1.0 nM there was partial loss of anti-α-SMA staining, and cells showed a reduction in the formation of stress fibers and focal adhesions. Higher doses (10.0 nM and up) completely blocked the staining of cells with anti-α-SMA antibodies and inhibited the formation of stress fibers and focal adhesions. Bar, 25 μM.

Figure 7.

 

(A) Western blot (upper) and Coomassie blue–stained gels (lower) of proteins extracted from confluent keratocytes treated with TGF_β alone (lane 1) or in combination with herbimycin A at 0.1 nM (lane 2), 1.0 nM (lane 3), or 10 nM (lane 4). Note the loss of α-SMA in keratocytes treated with 1.0 nM and 10 nM. Striping nitrocellulose paper and reprobing with anti-phosphotyrosine antibodies showed a similar loss in tyrosine-phosphorylated proteins (not shown). (B) Northern blot of RNA extracted from untreated keratocytes (lane 1) and keratocytes treated with TGF_β (1 ng/ml) alone (lane 2) or in combination with herbimycin A at 0.1 nM (lane 3), 1.0 nM (lane 4), and 10 nM (lane 5). Blots were probed with a 525 bp cDNA for rabbit α-SMA message. Lower panel shows ethidium bromide staining of gel before transfer.