Transforming Growth Factor-β–Stimulated Connective Tissue Growth Factor Expression during Corneal Myofibroblast Differentiation

Abstract

purpose. Transforming growth factor β1 (TGF-β) stimulates the differentiation of myofibroblasts as indicated by the nascent expression of α-smooth muscle (α-SM) actin protein and its organization into stress fibers. Downstream messengers of TGF-β in the conversion from the fibroblast to the myofibroblast phenotype were investigated. Whether TGF-β increases the transcription of a second growth factor, connective tissue growth factor 1 (CTGF), which could mediate myofibroblast differentiation, was evaluated. CTGF, a newly identified growth factor, is highly expressed in dermal granulation tissue.

methods. In this study, primary cultures of rabbit corneal fibroblasts were exposed to growth factors to investigate CTGF mRNA and protein expression during myofibroblast differentiation. Statistical analysis was used to evaluate the impact of growth factor treatment on myofibroblast differentiation.

results. TGF-β treatment induced both CTGF mRNA and protein in rabbit corneal fibroblasts; in contrast, fibroblast growth factor-2 (FGF) and heparin led to a decrease in CTGF mRNA. Addition of recombinant CTGF to rabbit corneal fibroblast cultures did not significantly increase α-SM actin mRNA or protein nor did it appear to affect assembly of α-SM actin stress fibers.

conclusions. This is the first study to present evidence for the induction of CTGF by TGF-β treatment of corneal fibroblasts. It is doubtful that CTGF is the TGF-β mediator of the corneal fibroblast to myofibroblast transition because CTGF does not induce α-SM actin in subconfluent fibroblast cultures. CTGF may play a supporting role in myofibroblast differentiation.

In corneal healing the normally quiescent keratocyte network is converted into motile secretory fibroblasts and later into contraction-ready myofibroblasts. ^{1 2 3} Myofibroblasts are characterized by the expression of α-smooth muscle (α-SM) actin, which is organized into stress fibers. ³ Myofibroblasts appear in situ adjacent to a wound site within a week of wounding. ^{4 5} Although the myofibroblasts promote wound closure, they also produce contracture and scarring; therefore, the respective contributions of fibroblasts and myofibroblasts is crucial for healthy repair. ⁶  

In situ and in vitro studies indicate that transforming growth factorβ 1 (TGF-β) stimulates myofibroblast differentiation. ^{7 8} However, if the cells are confluent, addition of TGF-β will not induce myofibroblast differentiation. ⁹ This density-dependent differentiation correlates with the finding that high- and low-density cells have a different capacity to respond to TGF-β, because high-density cells express fewer receptors (TGF-β receptor [TβR]II and TβRI) than low-density cells. ^{10 11}  

Wounds that have completely healed contain few if any myofibroblasts, presumably because they revert to the fibroblast phenotype or undergo apoptosis during wound healing. ² Recent work in our laboratory indicates that the treatment of myofibroblast cultures with fibroblast growth factor-2 (FGF) and heparin will induce the conversion of myofibroblasts to the fibroblast phenotype. ¹² The observation that FGF and TGF-β have opposing effects on phenotype is consistent with the findings that these two growth factors participate in different signaling pathways and that these pathways converge on the regulation of Smad proteins downstream of TβRI and -II. ^{12 13 14} Translocation of Smad2 to the nucleus in corneal fibroblasts is associated with myofibroblast differentiation. ¹¹  

Connective tissue growth factor 1 (CTGF) has been implicated as downstream of TGF-β in wound healing. ¹⁵ CTGF is a cysteine-rich, heparin-binding peptide originally identified in media conditioned by human umbilical vein endothelial cells. ¹⁶ Subsequently, it was demonstrated that CTGF was also expressed in fibroblasts after treatment with TGF-β. ^{17 18} CTGF is M _r 38,000 and is a member of the highly conserved CCN family of peptides that include the immediate early genes cef10, cyr61, nov, and fisp12, a putative avian proto-oncogene. ^{19 20 21 22 23} This protein family is characterized by an absolute conservation of 38 cysteine residues that constitute >10% of the total amino acid content. ^{16 24} Twisted gastrulation (twg), a Drosophila gene controlling medial mesoderm induction during dorsal-ventral axis pattern formation is also related to the CCN gene family. ²⁵ Additional members of the family are aberrantly expressed in human colon tumors. ²⁶ These include Elm-1/Wisp-1, Cop-1/Wisp-2/CTGF-3, and Wisp-3.  

TGF-β and CTGF genes are coordinately regulated during normal tissue regeneration in situ. ¹⁸ This type of regulation supports a cascade model where initiators of wound healing activate expression of secondary factors that control and sustain specific cellular processes in the regenerating tissue. ²⁷ Regulatory cascades involving TGF-β and CTGF appear to be critical in early development as well as in wound repair. ²⁷ In the early Drosophila embryo a cascade of gene products controls pattern formation during dorsal-ventral axis organization. ^{28 29} Several of the genes that have been identified in this cascade encode proteins are members of the TGF-β superfamily. ^{30 31 32} Another gene in this cascade is the CCN family member, twisted gastrulation (twg). ²⁵ These findings suggest that the genetic pathways regulating tissue formation are highly conserved and that the wound repair cascade may be derived from or possibly identical with the pathways that specify mesodermal tissue formation and organization during embryogenesis in the invertebrate. ²⁷ Indeed, recent data indicate a specific temporo-spatial pattern during embryogenesis that supports a role for CTGF in cellular differentiation and development during prenatal life. ³³  

CTGF gene expression is induced by TGF-β in fibroblasts but not in other cell types. ^{17 18} A brief exposure of fibroblasts to TGF-β (1 hour) is sufficient to induce a prolonged high-level expression of the CTGF transcript (24–36 hours). ²⁴ The regulation of CTGF appears to be controlled primarily at the level of transcription and is dependent on the action of a specific TGF-β response element (TβRE) demonstrated in human and murine CTGF promoters. ²⁴ The TβRE sequence within the human CTGF promoter had not been described previously as a control element. ²⁴ Collectively, these observations regarding CTGF gene expression strongly suggest that CTGF functions as a downstream mediator of TGF-β action on fibroblasts. ²⁴  

We sought to evaluate CTGF as a mediator of TGF-β–induced myofibroblast differentiation. Controlling the balance of fibroblasts and myofibroblasts is thought to be critical in corneal wound repair, ⁶ and the identification of the essential growth factors in this process may provide therapeutic solutions to disturbances in the healing process. Because recent work has shown that CTGF is present in healing stromal tissue of the cornea but not in normal corneal tissue, ³⁴ a role for CTGF in corneal wound healing is suggested, but it remains to be determined what physiological roles CTGF plays in this process. 

We have investigated whether TGF-β induces CTGF expression in corneal fibroblasts and whether CTGF may contribute to the transition of corneal fibroblasts to myofibroblasts. We have used a standard cell culture model for these studies, in which TGF-β’s ability to induce myofibroblast differentiation has been well documented. ⁸  

Materials and Methods

Cell Culture

After removing the epithelium and endothelium from extirpated rabbit corneas, the stroma was digested with collagenase (Worthington, Lakewood, NJ) to release the keratocytes as described previously. ³⁵ Cells were cultured in Dulbecco’s modified Eagle’s medium supplied with Ham’s nutrient mixture F-12 (DMEM-F12; Sigma, St. Louis, MO) with 10% fetal bovine serum (FBS; Gibco BRL, Grand Island, NY) and antibiotic–antimycotic mix (10,000 units penicillin, 10 mg streptomycin, 25 mg amphotericin B per milliliter in 0.9% sodium chloride) and gentamicin solution (10 mg/ml; Sigma) and used in passages 3 through 8. All cells in these studies were passaged at defined cell densities because we have previously demonstrated that the fibroblast or myofibroblast phenotypes are directly related to cell density. ⁹ For routine passage, cells are plated at high density (500 cells/mm²) or approx. 1 × 10⁶ cells in a 100-mm dish. These cells contact one another within 24 hours, but the culture does not become quiescent before 72 hours, at which time cell proliferation has ceased, all the cells have adhered to the dish, and cell–cell contacts have been established. The majority of these cells are fibroblasts. 

For Western and Northern blot analysis experiments, confluent corneal fibroblast cultures were split 1:4. As a result, corneal fibroblasts were plated at 50 cells/mm² or 1–4 × 10⁵ cells in a 100-mm dish (intermediate density). ³⁶ After 18 hours the medium was changed to DMEM-F12 medium with one of the following additions: (1) 10% FBS, (2) 10% FBS, 20 ng/ml recombinant FGF (Gibco BRL) + 5 μg/ml heparin (Sigma), (3) 1% FBS, or (4) 1% FBS with 1 to 10 ng/ml recombinant TGF-β1 (rTGF-β; Boehringer-Mannheim, Indianapolis, IN) or 1 ng/ml human platelet TGF-β1 (hTGF-β; Becton Dickinson, Bedford, MA), or (5) 1% FBS with 30 to 100 ng/ml recombinant CTGF ¹⁵ for 24 hours. Cells cultured for Western blot analysis were treated with 1 ng/ml rTGF-β for 24 and 48 hours. 

Immunocytochemical Identification of Myofibroblasts

Myofibroblasts were identified by immunodetection of α-SM actin. ³⁷ Cells were grown for 3 to 5 days on coverslips and were fixed with 3% p-formaldehyde (Fisher Scientific, Fair Lawn, NJ) in phosphate-buffered saline (PBS; pH 7.4), 15 minutes at room temperature (RT), permeabilized in 0.01% Triton X-100 in PBS for 1 minute at RT, followed by quenching aldehyde–induced fluorescence with NH₄Cl (50 mM). After blocking nonspecific binding with 3% normal serum, cells were incubated for 60 minutes with anti–α-SM actin conjugated to a fluorophore, Cy3 (1:400; Sigma). After rinsing with PBS and a 1-minute exposure to Hoechst 33258 (0.06 μg/ml; Sigma), coverslips were rinsed and mounted with antifade agent. ³⁸ Cells were viewed with a Zeiss Axiophot microscope (Thornwood, NY) equipped for epifluorescence and photographed on TMAX 3200 film (Eastman Kodak, Rochester, NY) exposed at 6300 ASA. 

To address whether CTGF induced myofibroblast differentiation as indicated by α-SM actin, corneal fibroblasts were plated at intermediate density, grown, and treated with 30 ng/ml CTGF ¹⁵ in 1% FBS and were compared with 0.25 ng/ml hTGF-β in 1% FBS or 20 ng/ml FGF and 5 μg/ml heparin in 10% FBS for 24 hours. Furthermore, CTGF + hTGF-β were added together to one set of coverslips in each experiment. Morphologic studies in our laboratory have routinely used hTGF-β, ³⁶ and comparable results are seen with rTGF-β (data not shown). α-SM actin was localized immunocytochemically, and nuclei were detected by Hoechst dye 33258 (Sigma). All cells in these experiments were fixed, stained, and viewed as above. The proportion of myofibroblasts and fibroblasts in each experimental condition was determined by counting the number ofα -SM actin–stained cells and the total number of cells visualized by Hoechst stained nuclei in three randomly chosen microscopic fields at× 400 final magnification (>100 cells total per coverslip). 

In experiments in which protein and RNA were isolated from the cells, a coverslip with cells from each dish was fixed and stained for α-SM actin, as described above. This method was used to verify the fibroblastic or myofibroblastic identity of the culture. 

Immunocytochemical Detection of CTGF

To investigate the ability of different growth factors to induce CTGF, corneal fibroblasts were grown for 18 hours after plating and then treated with 0.25 ng/ml hTGF-β in DMEM-F12 + 1% FBS or the FGF-heparin in 10% FBS for 24 hours. CTGF was localized immunocytochemically by goat anti-CTGF IgG ³⁹ followed by anti-goat IgG–FITC (Jackson Laboratories, West Grove, PA). 

To confirm the initial Golgi localization of the CTGF after hTGF-β treatment, cells were treated with 10 μg/ml Brefeldin A (Epicentre Technologies, Madison, WI), which disassembles the Golgi apparatus. Brefeldin A, by disrupting the Golgi apparatus, traps recently synthesized proteins in the endoplasmic reticulum (ER) and enhances their immunodetection. ⁴⁰ We found that commercially available anti-Golgi antibody gave nonspecific staining in rabbit cells. For these studies, corneal fibroblasts were treated on coverslips with 0.25 ng/ml hTGF-β for 6 hours, with Brefeldin A present for the last 2 hours. CTGF protein was localized immunocytochemically as above. 

SDS-PAGE and Western Blot Analysis

Adherent cells were treated with growth factors for 24 hours and lysed in cold Tris-acetate–nonidet P-40 (NP-40) lysis buffer (10 mM Tris-acetate, pH 8, 0.5% NP-40) in the presence of protease inhibitor cocktail tablets (Boehringer Mannheim) and 2 mM phenylmethylsulfonyl fluoride on ice. ^{41 42} Protein concentrations were determined using the Bio-Rad protein assay (Bio-Rad, Hercules, CA), and 20 μg of protein/lane was separated under reducing conditions in a 7.5% SDS-PAGE gel. ⁴³ Proteins were transferred to nitrocellulose membranes (Protran; Schleicher & Schuell, Keene, NH) and air-dried. Nonspecific sites were blocked in 5% BSA in Tris-buffered saline with 0.05% Tween-20 (TBS-T). The membranes were incubated with mouse monoclonal anti– α-SM actin (1:4000; Sigma, clone no. 1A4) in TBS-T with 5% BSA, overnight at 4°C. This was followed by incubation with HRP-conjugated goat anti-mouse IgG (1:5000; Boehringer Mannheim) for α-SM actin, 30 minutes at RT, and detected by enhanced chemiluminescence (ECL; Pierce, Rockford, IL) according to the manufacturer’s instructions. Quantitation of the gel bands was determined by scanning the film and analyzing the signal using an image analysis program (ImageQuant; Molecular Dynamics, Sunnyvale, CA). 

Northern Blot Analysis

Total RNA was extracted from cultured cells after 24 hours of growth factor treatment using a total RNA isolation system from Qiagen (Santa Clarita, CA). Total RNA samples (10 μg/lane) were separated on 1.2% agarose-formaldehyde gels, blotted onto positively charged membranes (Bright Star-Plus; Ambion, Austin, TX) with 0.5× TBE, and dried in an 80°C oven. The membranes were hybridized with a ³²P-labeled α-SM actin cDNA probe at 37°C for 16 to 18 hours. This cDNA probe (nucleotides 1114–1335) ⁴⁴ includes a unique sequence within the 3′ untranslated region of rabbitα -SM actin. The membranes were washed two times in 2× SSC, 0.1% SDS for 5 minutes at 37°C, washed two more times in 0.2× SSC, 0.1% SDS for 30 minutes at 37°C, and finally were washed once in 0.2× SSC for 5 minutes at 37°C and exposed to film. To detect CTGF mRNA, the membranes were hybridized with a ³²P-labeled cDNA probe corresponding to a 1.0-kb cDNA insert containing the human CTGF open reading frame ¹⁶ at 46°C for 16 to 18 hours. The membranes were washed as described previously ¹⁶ and exposed to film. 

To normalize the RNA loading in each lane, filters were hybridized with a ³²P-labeled plasmid cDNA insert of the 18S (housekeeping) gene (DECA template-18S-mouse; Ambion) in NorthernMax hybridization buffer (Ambion) and were washed using solutions from Ambion (NorthernMax Wash Buffer System). For each detection, the blot was exposed to Biomax MS film (Eastman Kodak). The autoradiograms were scanned with a densitometer and quantified using Image Quant software (Molecular Dynamics). The intensity of each mRNA band was normalized to the 18S signal. 

Statistical Analysis

Multiple regression analysis was applied to compare the effect of 1% FBS, hTGF-β, CTGF, or hTGF-β and CTGF combined on the expression of α-SM actin protein in corneal fibroblast cells (Table 1) . Within each experiment three replicates were included. Statistical analysis was conducted using the JMP program (SAS Institute, Cary, NC). 

Results

TGF-β–Induced CTGF Protein Expression in Corneal Fibroblasts

We began the present study by asking if TGF-β–treated rabbit corneal fibroblasts express CTGF at the protein level in a cell culture system used to model wound healing. Subconfluent cultures of rabbit corneal fibroblasts were treated with 0.25 ng/ml hTGF-β for 6 hours. CTGF protein was detectable in the ER surrounding the nucleus and prominent in the Golgi apparatus (Fig. 1B) . In contrast, the control cells grown in 1% FBS have little detectable CTGF, and it is limited to the nuclear envelope (Fig. 1A) . Brefeldin A treatment enhanced the detectable CTGF in the ER almost filling the cytoplasm, an index of TGF-β–induced new synthesis of CTGF (Fig. 1C) . Thus, TGF-β induces CTGF protein synthesis in corneal fibroblasts. 

Because our laboratory has recently found that FGF plus heparin reverses myofibroblast differentiation by promoting the fibroblast phenotype ¹² and because corneal fibroblasts treated with FGF-heparin maintain the fibroblast phenotype, we asked whether FGF might inhibit CTGF protein expression. In FGF-heparin–treated cultures, a minority of cells (Fig. 1E) had CTGF staining that was limited to the nuclear envelope. This suggests that FGF downregulates CTGF expression. 

Enhanced Expression of CTGF mRNA Induced by TGF-β in Corneal Fibroblasts

We next asked, on the basis of the ability of TGF-β to induce and FGF to inhibit CTGF expression, how these growth factors affect CTGF mRNA expression. We compared the ability of growth factor treatments to induce changes in CTGF mRNA expression in rabbit corneal fibroblasts. rTGF-β and hTGF-β were used to evaluate their respective effects on mRNA induction. Figure 2 illustrates that rTGF-β in 1% FBS elevates CTGF mRNA compared with 1% FBS alone (Fig. 2C , cf. lanes 1–3 and 1–4). The CTGF signal was normalized to the expression of the 18S ribosomal subunit housekeeping gene. rTGF-β treatment resulted in a 1.75- to 2-fold increase in CTGF expression compared with the 1% FBS control (Fig. 2C , cf. lanes 1–3 and 1–4). This result may be important for corneal healing because TGF-β and CTGF genes are coordinately regulated during normal tissue regeneration in situ. ¹⁸ Of particular interest is the observation that 1% FBS induces a twofold increase in the level of CTGF mRNA expression compared with 10% FBS (Fig. 2C , cf. lanes 1 and 6). This suggests the presence of an inhibitory factor within the serum that has a negative effect on CTGF expression. 

Downregulation of CTGF mRNA by FGF-Heparin in Corneal Fibroblasts

Given the opposing effects of TGF-β and FGF in corneal fibroblast differentiation and the capacity of TGF-β to induce CTGF mRNA and protein expression in cells that acquired the myofibroblast phenotype, we determined the effect of FGF on CTGF mRNA. Figure 2 indicates that FGF and heparin in 10% FBS downregulates CTGF mRNA in corneal fibroblasts (Fig. 2C , cf. lanes 5 and 6). Densitometric analysis showed that FGF-treated cells decreased CTGF mRNA threefold compared with the 10% FBS control (Fig. 2C , cf. lanes 5 and 6). This result suggests that the presence of FGF within 10% FBS may contribute to our finding of less CTGF expression than in cells grown in 1% FBS. 

Expression of α-SM Actin mRNA Induced by TGF-β, but Not by CTGF Treatment

Expression of α-SM actin mRNA as an index of myofibroblast differentiation was analyzed in corneal fibroblast cells in response to growth factor treatment. The autoradiogram shown in Figure 3 shows an intense 1.3-kb α-SM actin signal for rTGF-β–treated cells (Fig. 3A , lanes 3 and 4), corresponding to the size of the α-SM actin mRNA listed in GenBank (accession no. NM007392). A weak 1.7-kb signal was seen for all treatment groups with a slightly stronger 1.7-kb signal for rTGF-β–treated cells. The 1.3- and 1.7-kb bands are also present in a recent paper describing myofibroblast differentiation. ⁴⁵  

Densitometric analysis of the autoradiogram indicated that although rTGF-β treatment resulted in an eight- to ninefold increase in 1.3-kbα -SM actin mRNA expression over 1% FBS control (Fig. 3C , cf. lanes 1–3 and lanes 1–4), CTGF did not lead to an increase in 1.3-kb α-SM actin mRNA over 1% FBS control (Fig. 3C , cf. lanes 1–5 and lanes 1–6). FGF-heparin-treated cells had a negligible level of 1.3-kbα -SM actin mRNA expression consistent with FGF not promoting the myofibroblast phenotype. Densitometric evaluation gave the same reading for FGF and the 10% FBS control (Fig. 3C , cf. lanes 7 and 8). 

Lack of Induction of α-SM Actin Protein Expression in Corneal Fibroblasts After CTGF Treatment

Next, we analyzed α-SM actin expression in Western blot analysis of cultured corneal fibroblasts after growth factor treatment (Fig. 4) . Cells treated for 48 hours with rTGF-β led to a 3.5-fold increase in α-SM actin signal intensity compared with cells treated with 1% FBS (Fig. 4C , cf. lanes 3 and 7), whereas 24-hour treatment with rTGF-β led to a 2.8-fold increase (Fig. 4C , cf. lanes 6 and 7). FGF-heparin–10% FBS treatment did not stimulate α-SM actin protein expression over the 10% FBS control (Fig. 4C , cf. lanes 4 and 1), whereas a twofold elevation in α-SM actin protein is seen in comparison to 1% FBS–treated cells (Fig. 4C , cf. lanes 4 and 7). Cells treated with CTGF did not increase α-SM actin signal intensity compared with the 1% FBS control (Fig. 4C , cf. lanes 2–7 and lanes 5–7), although CTGF did have a small dose effect (Fig. 4C , cf. lanes 2 and 5). 

The α-SM actin signal seen in the CTGF and 1% FBS lanes suggests that there is a low level of α-SM actin in all the cells or that there is a small proportion of myofibroblasts in these cultures. ⁹ Most cells maintain a high cytosolic concentration of unpolymerized actin monomers in the form of an ADP–actin monomer complex, stably bound to an actin monomer binding protein. ^{46 47} This would explain the presence of α-SM actin protein at 24 hours, despite the absence of 1.3-kb α-SM actin mRNA for cells that were not treated with TGF-β. 

Furthermore, the presence of 10% serum primes fibroblasts for stress fiber elongation ⁴⁸ fueled by a large diffusible pool of profilin–actin monomer complexes. ⁴⁹ A large, stable population of profilin–actin monomer complex may explain the α-SM actin signal intensity from cells exposed 10% FBS (high serum), which is twofold higher than that of 1% FBS (low serum; Fig. 4C , cf. lanes 1 and 7). Addition of FGF-heparin to 10% FBS does not significantly alter α-SM actin signal intensity compared with 10% FBS alone (Fig. 4C , cf. lanes 4 and 1). 

Although Western blot analysis of CTGF-treated cells (24 hours) did not have increased α-SM actin protein, we wondered whether CTGF might influence assembly of α-SM actin containing stress fibers and thus promote myofibroblast differentiation in conjunction with TGF-β. After an 18-hour growth period, corneal fibroblasts were treated with 30 ng/ml CTGF ¹⁵ or 0.25 ng/ml hTGF-β (Becton Dickinson) or a combination of both growth factors at the designated concentrations for up to 5 days. Cells grown in 1% serum with or without CTGF (Figs. 5A 5B) resulted in <50% of the cells with the myofibroblast phenotype. hTGF-β treatment leads to the majority of cells expressing the myofibroblast phenotype (Fig. 5C) . After treatment with this combination of growth factors (Fig. 5D) , the number of cells with fully assembled α-SM actin stress fibers was approximately that of cells treated with hTGF-β alone, indicating that CTGF did not antagonizeα -SM actin stress fiber assembly. 

Table 1 presents the results of the statistical analysis of this data. Multiple regression analysis indicated that TGF-β had a highly significant impact on myofibroblast differentiation, whereas CTGF had a borderline effect. The statistical analysis does not show an interaction between TGF-β and CTGF; therefore, the effect of TGF-β on differentiation does not depend on CTGF under the conditions of these experiments. Although CTGF does not induce an increase in α-SM actin stress fiber assembly, it may have another effect on differentiation. CTGF may be chemotactic ¹⁶ and promote corneal fibroblast migration before differentiation. ⁵⁰ Because cell motility and adhesion are driven by the Rac/Rho pathway, ⁴⁷ CTGF may intersect with this pathway to promote wound healing. 

Discussion

CTGF has been implicated as an effector of TGF-β in several wound-healing responses. ^{15 18} We report that in corneal fibroblasts TGF-β induces expression of CTGF mRNA and protein. However, although TGF-β upregulated 1.35-kb α-SM actin mRNA and protein in corneal fibroblasts, CTGF did not. We also found that FGF-heparin inhibits CTGF expression, α-SM actin expression, and the development of the myofibroblast phenotype. We conclude that although CTGF did not appear to induce the corneal myofibroblast phenotype directly, it may influence other biological events leading to differentiation. 

Previous reports of a large upregulation of CTGF by TGF-β in confluent, quiescent fibroblasts ³⁹ were confirmed in corneal quiescent, confluent, serum-free cultures (data not shown). Because there is evidence that CTGF plays a role in proliferation and migration, ¹⁶ CTGF may promote the early events of myofibroblast precursor proliferation and migration. ⁵⁰  

CTGF is not only mitogenic and chemotactic for fibroblasts, ¹⁶ it also stimulates the synthesis of at least two extracellular matrix components: type I collagen and fibronectin. ¹⁵ Recent evidence also indicates an indirect role for CTGF in matrix degradation. Fibroblast adhesion to CTGF has been found to induce a prolonged activation of the metalloproteinases, MMP-1 and MMP-3. ⁵¹ Because wound healing requires the degradation of a provisional matrix as well as the synthesis of new matrix, ^{52 53} regulation of gene expression by CTGF may be central to the matrix remodeling of granulation tissue. These aspects of CTGF activity are currently being studied in corneal cell culture. 

The finding that TGF-β is involved in regulating the expression of CTGF mRNA and protein in corneal fibroblast cell culture supports previous reports that CTGF is selectively induced by TGF-β in fibroblasts. ^{17 18} A novel response element (TβRE) interacting with unknown transcription factors implies that the regulation of CTGF gene expression by TGF-β functions by a different mechanism than other TGF-β–regulated genes. ²⁷ The response element (TβRE) identified in the CTGF promoter shares partial sequence homology with the consensus sequence for the cAMP response element, and its activation by TGF-β is inhibited by cAMP analogs and agents elevating intracellular cAMP. ^{24 54} Because cAMP regulates cell proliferation ²⁷ and collagen synthesis ⁵⁵ in fibroblasts stimulated with TGF-β, it is possible that signal transduction initiated by TGF-β converges with cAMP-inducible pathways. ⁵⁵ Therefore, one may postulate that CTGF-dependent actions are responsive to intracellular cAMP levels and that a change in cAMP modulates corneal fibroblast CTGF mRNA and protein expression. 

The effects of FGF and heparin on CTGF mRNA and protein observed in these studies are interesting findings with respect to the fibroblast–myofibroblast balance. When corneal fibroblasts are treated with FGF-heparin, they remain fibroblasts in culture and do not progress to the myofibroblast phenotype (unpublished observation). Furthermore, when CTGF mRNA is assessed in these cells, there is a dramatic downregulation in response to FGF treatment. This downregulation of CTGF may contribute to the maintenance of the fibroblast phenotype by FGF. Our results extend those of Igarashi et al., ¹⁸ who found that FGF does not induce CTGF in confluent fibroblasts under serum-free conditions. In our study FGF downregulated CTGF mRNA in subconfluent cells grown in serum. The different growth status in which the cells were confluent in the earlier study and subconfluent in the present study may contribute to this difference in regulation. 

Recent findings by Jester et al. ⁴⁵ indicate that α-SM actin stress-fiber assembly is required for the myofibroblast phenotype. The serum used to culture our corneal fibroblast cells contains lysophosphatidic acid (LPA), which activates the small GTP-binding protein Rho, a central regulator of stress-fiber assembly. ⁴⁸ Because LPA may also be produced and released by growth factor-stimulated fibroblasts, it is an appropriate candidate mediator of the activated fibroblast phenotype in corneal wounds. ⁵⁶  

We did not find that CTGF had a significant impact on the assembly ofα -SM actin stress fibers in corneal fibroblasts. Although CTGF does not induce α-SM actin mRNA or protein, it may function in a supportive manner to enhance differentiation. Recent studies have shown that secreted CTGF is an adhesive substrate that has specific signaling capabilities in fibroblasts. Plating fibroblasts on CTGF promotes extensive and prolonged formation of filopodia and lamellipodia,β -actin reorganization, and activation of intracellular signaling molecules including focal adhesion kinase (FAK), paxillin, and Rac. ⁵¹ Because CTGF stimulates chemotaxis in fibroblasts, ¹⁶ it may play a supportive role by activating the reorganization of β-actin required for corneal fibroblast movement ⁴⁷ before differentiation. 

Jester et al. ⁴⁵ investigated the role of TGF-β in corneal myofibroblast differentiation in serum-free, confluent cultures of corneal keratocytes and found that TGF-β induces an upregulation of 1.7- and 1.3-kb mRNAs after a 72-hour treatment period. In the present study, we show that in serum-containing, subconfluent cultures of corneal fibroblasts, a 24-hour TGF-β treatment induces 1.7- and 1.35-kb mRNA expression. Northern blot analysis detection using theα -SM actin mRNA probe typically detects two bands: 1.3 and 1.7–1.8 kb. It is likely that the 1.3-kb mRNA detected in Northern blot analysis represents α-SM actin, based on the size of α-SM actin sited in GenBank (accession no. NM007392). However, actin mRNA is highly conserved among the α-, β-, and γ-forms, and cDNA probes can cross-react on Northern blot analysis (possibly at 1.7-kb). Detection of β-actin (1.8-kb mRNA; accession no. NM001101) by theα -SM actin probe has been suggested by Mitchell. ⁵⁷  

Here we present evidence for the induction of CTGF by TGF-β treatment of corneal fibroblasts. We find that it is unlikely that CTGF is the TGF-β mediator in the corneal fibroblast to myofibroblast transition. CTGF does not induce α-SM actin mRNA in subconfluent cultures. However, CTGF may play a supporting role in myofibroblast differentiation. In addition to being actively synthesized and released by granulation tissue fibroblasts as a TGF-β–induced immediate early gene, ^{17 18} CTGF is mitogenic and chemotactic for fibroblasts. ¹⁶ TGF-β–induced CTGF secretion by corneal fibroblasts may be important for migration into the wound, a phenomenon that does not require α-SM actin. ^{58 59 60} The current report demonstrates that CTGF, a protein involved in connective tissue formation during wound repair, is induced during the transition of the corneal fibroblast to the myofibroblast phenotype, suggesting a role in the critical differentiation events during corneal healing but not in the induction of the myofibroblast phenotype itself. 

 Supported by National Institutes of Health (NIH) Grants F32 EY06907 (PAF) and R01 EY09414 (SKM), NIH Core Center Grant EY01867, an unrestricted grant from Research to Prevent Blindness, and the Finley-Turobiner Eye Fund (SKM).

 Submitted for publication June 14, 2000; revised May 1 and June 6, 2001; accepted June 19, 2001.

 Commercial relationships policy: N.

 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.

 

TGF-β induces CTGF synthesis in corneal fibroblasts. CTGF was immunodetected in fixed rabbit corneal fibroblast cultures. (A through C) CTGF is increased in organelles of the secretory pathway in cells treated with hTGF-β (0.25 ng/ml). (A) A minority of fibroblasts grown in media with 1% FBS (control for hTGF-β) have CTGF localized in the nuclear envelope (arrow). (B) In fibroblasts treated with hTGF-β (6 hours, in 1% FBS), CTGF was localized to the Golgi apparatus (arrows). (C) In fibroblasts treated with hTGF-β (6 hours, in 1% FBS) and Brefeldin A for the 2 hours before fixation, the newly synthesized CTGF was highly expressed in the ER, which fills the cytoplasm of cells (arrows). (D) A minority of cells grown in media with 10% FBS (control for FGF-heparin) have immunodetectable CTGF (arrow). (E) In contrast to hTGF-β–treated cells, the majority of cells treated with FGF and heparin in 10% FBS for 6 hours have little demonstrable CTGF, and it is restricted to the nuclear envelope (arrow). Bar, 50 μm.

Figure 2.

 

TGF-β induces, whereas FGF inhibits CTGF mRNA expression. (A) Northern blot analysis of total RNA extracted from rabbit corneal fibroblasts treated for 24 hours with 1% FBS (lane 1), 1 ng/ml hTGF-β (lane 2), 1 ng/ml rTGF-β (lane 3), 10 ng/ml rTGF-β (lane 4), 20 ng/ml FGF-heparin in 10% FBS (lane 5), and 10% FBS (lane 6). Blots were probed with a 1000-bp cDNA for CTGF mRNA. A 2.37-kb mRNA is detected, which is the size of the CTGF mRNA. (B) Northern blot analysis shown in (A) reprobed for the 18S RNA (2.0 kb) to standardize mRNA levels. (C) The graph represents a densitometric analysis of the autoradiogram showing the expression of CTGF, normalized to the 18S signal, in each sample.

Figure 3.

 

α-SM actin mRNA is induced by TGF-β. (A) Northern blot analysis of total RNA extracted from corneal fibroblasts treated for 24 hours with 1% FBS (lane 1), 1 ng/ml hTGF-β (lane 2), 1 ng/ml rTGF-β (lane 3), 10 ng/ml rTGF-β (lane 4), 100 ng/ml CTGF (lane 5), 30 ng/ml CTGF (lane 6), 20 ng/ml FGF-heparin in 10% FBS (lane 7), and 10% FBS (lane 8). Blots were probed with a 221-bp cDNA for α-SM actin mRNA that detected the 1.3-kb mRNA as well as the 1.7-kb mRNA. (B) Northern blot analysis shown in (A) was reprobed for 18S RNA (2.0 kb). (C) Densitometric analysis of the autoradiogram showing the expression of 1.3-kb α-SM actin, normalized to the 18S signal, in each sample. (D) Densitometric analysis of the autoradiogram showing the expression of the 1.3- and 1.7-kb bands combined, normalized to the 18S signal, in each sample.

Figure 4.

 

TGF-β treatment increases α-SM actin in cultured fibroblasts, and CTGF does not. (A) Western blot of lysates of corneal fibroblast control cultures grown in 10% FBS for 24 hours (lane 1), 100 ng/ml CTGF treatment in 1% FBS for 24 hours (lane 2), 1 ng/ml rTGF-β in 1% FBS for 48 hours (lane 3), 20 ng/ml FGF-heparin in 10% FBS for 24 hours (lane 4), 30 ng/ml CTGF in 1% FBS for 24 hours (lane 5), 1 ng/ml rTGF-β in 1% FBS for 24 hours (lane 6), and 1% FBS for 24 hours (lane 7). (B) Protein detected by the Ponceau staining of the SM α-actin blot in (A) that corresponds to the region detected by anti–SM actin. (C) Histogram of α-SM actin signal (Western), normalized to the protein per lane (Ponceau), for each sample. TGF-β treatment induced a large increase in α-SM actin protein after 48 hours in culture (lane 3).

Figure 5.

 

Myofibroblast differentiation is promoted by TGF-β or TGF-β plus CTGF. Immunofluorescent micrographs of subconfluent corneal fibroblasts treated for 5 days with (A) 1% FBS (control), (B) CTGF (30 ng/ml), (C) hTGF-β (0.25 ng/ml), and (D) CTGF (30 ng/ml) and hTGF-β (0.25 ng/ml). Cells were double stained with Hoechst dye and cy3-anti–α-SM actin to identify myofibroblast cells.

Table 1.

 

α-SM Actin Induction in Myofibroblasts

Treatment	α-SM Actin
1% FBS	14.45 ± 1.87
CTGF	35.99 ± 18.98
hTGF-β	83.36 ± 15.31
hTGF-β+ CTGF	99.0 ± 1.73

 Values are the percentage of the number of cells expressingα -SM actin stress fibers (myofibroblasts) compared with the total number of cells in the field of view. All values represent the mean ± SD (n = 3). Cells were stained after 5 days in culture as seen in Figure 5 . Multiple regression analysis was used to compare the effect of 1% FBS, hTGF-β, CTGF, or hTGF-β and CTGF combined on the expression of α-SM actin protein. Values of P < 0.001 were considered to be statistically significant. The analysis indicated that CTGF did not significantly alter the expression of α-SM actin protein.

The authors thank Carol Bodian and Sylvan Wallenstein of the Department of Biomathematical Sciences at Mount Sinai School of Medicine for statistical analysis, Sevastiani Petridou and Olga Maltseva for excellent technical advice, and members of the Masur laboratory for stimulating discussions.