Fibroblast Growth Factor Reversal of the Corneal Myofibroblast Phenotype

Abstract

purpose. Keratocytes give rise to fibroblasts and myofibroblasts in wounded cornea. It is well established that treatment of fibroblasts with transforming growth factor (TGF) β will induce myofibroblast differentiation. We investigated whether this differentiation could be reversed by the administration of fibroblast growth factor (FGF).

methods. Cultured corneal myofibroblasts were plated at 200 cells/mm², and cells were grown in DMEM/F12 containing (1) 10% FBS or (2) 10% FBS with FGF and heparin or (3) 1% FBS or (4) 1% FBS with TGF-β. As distinguished from the fibroblast phenotype, the myofibroblast phenotype was identified by the assembly of α-smooth muscle (SM) actin protein into the stress fiber cytoskeleton. To further characterize growth factor regulation of the two phenotypes, the phenotypic expression of TGF-β receptor types I and II, cadherins, and connexin 43 by immunocytochemistry, Western blot analysis, and immunoprecipitation and of α-SM actin mRNA in Northern blot analysis were evaluated.

results. Corneal myofibroblasts replated and grown in the presence of FGF-1 or FGF-2 (20 ng/ml) plus heparin (5 μg/ml) in 10% FBS medium had decreased expression of α-SM actin protein, TGF-β receptors, and cadherins. Thus, FGF–heparin decreased the myofibroblast phenotype and promoted the fibroblast phenotype. Administration of either 20 ng/ml FGF or 5 μg/ml heparin alone was not effective. Addition of TGF-β further enhanced the expression of α-SM actin mRNA and protein and cell surface expression of TGF-β receptors in myofibroblast cultures.

conclusions. FGF-1 or -2 and heparin promoted the fibroblast phenotype and reversed the myofibroblast phenotype. This finding supports the idea that corneal myofibroblasts and fibroblasts are alternative phenotypes rather than terminally differentiated cell types.

The quiescent keratocytes of corneal stroma become activated by corneal wounding and are replaced by fibroblasts and myofibroblasts. Differentiation of myofibroblasts is induced by transforming growth factor (TGF) β. The myofibroblast phenotype is exemplified by the novel expression of α-smooth muscle (SM) actin and its assembly into stress fibers. Although both fibroblasts and myofibroblasts contribute to normal wound repair, in a fully healed wound few if any myofibroblasts are found. ¹ It is not known whether during healing all myofibroblasts undergo apoptosis or if some myofibroblasts revert to a fibroblast or a keratocyte phenotype. Pathologic conditions are associated with excess numbers of persistent fibroblasts or myofibroblasts. ^{2 3 4}  

To characterize each of these three phenotypes, recent studies have used cultures of freshly isolated keratocytes, passaged fibroblasts, and myofibroblasts. For example, freshly isolated keratocytes have the patterns of expression of integrins and metalloproteinases that correspond to that seen in situ. ^{5 6 7} Keratocytes are activated to fibroblasts by addition of serum to the culture medium. ^{8 9} Fibroblasts express novel integrins, cytokine receptors, cell junction molecules, matrix metalloproteases, and matrix components similar to expression patterns seen in situ after wounding. ^{10 11 12 13 14 15} Myofibroblasts are generated by adding TGF-β to keratocytes or to fibroblasts plated at an intermediate density or by plating fibroblasts at very low density. ^{8 16 17 18} Associated with the myofibroblast phenotype are differential expression patterns of integrins, metalloproteases, and cell junction molecules. ^{6 19 20}  

There have been no reports that cytokine treatment of corneal myofibroblasts induces them to “revert” to fibroblasts in vivo or in vitro. Therefore, we have evaluated whether treatment of cultures with FGF and heparin can “reverse” their myofibroblast phenotype. FGFs are members of a family whose core amino acid sequences homology allows for a common FGF tertiary structure. These common cores bind to heparin, resulting in dimerized FGF. Signaling through the FGF receptor is induced by dimerization of the FGF receptors and is promoted by dimerized FGF–heparin (reviewed in Ref. ²¹ ). FGF-1 and -2 have overlapping specificity for isoforms of the FGF receptors (FGFR-1, -2, -3, and -4; Table 1 in Ref. ²¹ ). The prototypic members of the FGF family, FGF-1 and -2, can regulate differentiation in various cell types. 

In addition by using the presence of α-SM actin as the myofibroblast signature, we have determined several membrane proteins whose expression is differentially regulated by FGF and TGF-β in proliferating corneal stromal cells in culture. These studies extend our knowledge of the role of these growth factors in regulating characteristics of the fibroblast and myofibroblast phenotypes. ^{17 22}  

Materials and Methods

Cell Culture

Corneas were dissected from New Zealand Albino rabbits immediately after they were killed. All procedures were performed according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Primary fibroblast cultures used in these studies were established from rabbit corneal keratocytes, released from the stroma using collagenase (500 units/ml; Worthington Biochemical Corporation, Lakewood, NJ), after removal of the epithelium and endothelium. Culture medium was a 1:1 mix of Dulbecco’s modified Eagle’s medium and Ham’s nutrient mixture F-12 (DMEM-F12; Gibco-BRL, Grand Island, NY), supplemented with 10% fetal bovine serum (FBS; Gibco), antibiotic-antimycotic mix (10,000 units penicillin, 10 mg streptomycin, and 25 mg amphotericin B/ml in 0.9% sodium chloride), and gentamicin (10 mg/ml; Sigma, St. Louis, MO). 

Corneal fibroblasts were maintained in culture up to eight passages. Within 16–18 hr of plating cells at intermediate density (10⁴ cells/ml or 200 cells/mm²), we induced the myofibroblast phenotype by adding 0.25–1.0 ng/ml TGF-β₁ in DMEM-F12 with 1% FBS (human platelet TGF-β₁; Collaborative Biomedical Products, Becton Dickinson Labware, Bedford, MA or recombinant human TGF-β₁; R&D Systems, Inc., Minneapolis, MN). Media plus specific growth factors were replaced every 2 days. At 3 days, the majority of cells expressed α-SM actin microfilaments, and at 4 to 5 days the cultures were confluent and were 100% myofibroblasts. The proportion of myofibroblasts was quantified by fluorescence microscopic colocalization of Hoechst-stained nuclei and immunodetected α-SM actin microfilaments (see Immunocytochemistry, below; Fig. 1 ). 

Confluent myofibroblast cultures were trypsinized (trypsin–EDTA–phosphate-buffered saline [PBS]; Invitrogen, Carlsbad, CA) and plated at intermediate density. Cultures were treated with experimental agents within 16 to 18 hours of plating, and media were replaced every other day. On the basis of our initial studies to promote fibroblast differentiation using either FGF-1 or -2 (see Results, Fig. 1 ), we standardly added fibroblast growth factor (FGF-1 or -2, 20 ng/ml; recombinant human; Gibco BRL) plus 5 μg/ml of heparin in DMEM/F12 medium, containing 10% FBS. These conditions are referred to as “FGF–heparin” and effectively decreased α-SM actin to a minimum (Fig. 2) . In contrast, we produced a totally myofibroblast culture when we added TGF-β, 0.25 ng/ml, in DMEM/F12 medium, 1% FBS. Therefore, two different FBS concentrations served as controls for specific growth factor addition: cultures were grown in DMEM/F12 medium, 10% FBS, to compare with FGF–heparin, or in DMEM/F12 medium, 1% FBS, to compare with TGF-β. Phenotype reversal was evaluated by Western blot analysis and immunocytochemistry, at intervals between 3 and 5 days after initial treatment with growth factors. The cultures were confluent at 5 days. 

Northern Blot Analysis of α-SM Actin mRNA

Total RNA was prepared from myofibroblast cultures replated at intermediate density. After 18 hours, the medium was changed to DMEM/F12 medium, as indicated with either 1% or 10% FBS and with FGF-2 (20 ng/ml), heparin (5 μg/ml), or TGF-β (1 ng/ml). We chose to lyse the cultures after 24 hours to capture changes in mRNA that precede the phenotypic expression of α-SM actin protein and its incorporation into microfilaments, which we detect after 3 to 5 days (see above). We lysed the cells in TriZol Reagent, following the manufacturer’s instructions (Molecular Research Center, Inc., Cincinnati, OH). Samples were separated on a 1% agarose-formaldehyde gel and transferred to a Nytran nylon membrane (Schleicher & Schuell, Keene, NH). The membrane was hybridized with a ³²P-labeled cDNA probe corresponding to a unique 111 nucleotide sequence of the 3′ untranslated region of rabbit α-SM actin (nucleotides 1114–1335) ²³ at 37°C for 16 to 18 hours. To normalize the RNA loading in each lane, the same filter was hybridized in NorthernMax Hybridization Buffer (Ambion, Austin, TX) with a ³²P-labeled plasmid cDNA insert of the internal standard, 18S ribosomal RNA (DECA templates; Ambion), and washed using solutions provided by Ambion (NorthernMax Wash Buffer System). ^{24 25} In each detection, the blot was exposed to Biomax MS film (Eastman Kodak, Rochester, NY) and developed in an X-Omat (Eastman Kodak). The autoradiograms were scanned with a BioRad 1650 densitometer (Hercules, CA) and quantified using Image Quant software (Molecular Dynamics, Sunnyvale, CA). 

Western Blot Analysis

At 4 to 5 days after passage, when the cultures were confluent, cells in 100-mm plates were scraped and suspended in PBS with protease inhibitors (Boehringer Mannheim, Indianapolis, IN), centrifuged at 1000 rpm, and then lysed in a small volume of Nonidet P-40 (NP-40) lysis buffer (0.5% NP-40, 150 mM NaCl, 10 mM tris-acetate buffer, pH 8.0) containing protease inhibitors. ¹⁴ Protein concentration of lysates was determined using Micro BCA kit (Pierce, Rockford, IL). Twenty-microgram samples were electrophoresed in 10% SDS-PAGE. Proteins were separated by SDS-PAGE on 8% gels and transferred from gels to nitrocellulose for Western blot analysis (Schleicher & Schuell). We verified that equal amounts of protein were loaded in each lane by staining the Western blot with Ponceau-S (Sigma). Immunoblotting was performed with monoclonal anti–α-SM actin (Sigma), monoclonal anti–pan-cadherin (Sigma), or affinity-purified polyclonal anti–cadherin-11 (Karen Knudsen, Lankenau Medical Center, Wynnewood, PA), anti–TGF-β receptor I (anti–TGF-βRII) and anti-TGF-βRII (V-22 and C-16, respectively; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and polyclonal anti–connexin 43 (Cx43; Elliot Hertzberg, Albert Einstein College of Medicine, Bronx, NY). Primary antibody was followed by the appropriate secondary antibodies conjugated to horseradish peroxidase (HRP; Jackson Laboratories, West Grove, PA) and detected by enhanced chemiluminescence (ECL; Pierce). 

Immunocytochemistry

We identified myofibroblasts by the presence of α-SM actin in stress fibers. Cells were fixed with 3% p-formaldehyde (Fisher Scientific, Fair Lawn, NJ) in PBS, pH 7.4, for 15 minutes at room temperature or in absolute methanol for 10 to 15 minutes at− 20°C. Nonspecific binding was blocked with 3% normal serum, and the cells were incubated with cy3 conjugated to mouse monoclonal antibodies against α-SM actin (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. ²⁶  

Proliferation

Two cytochemical methods were used to evaluate proliferation. Although we evaluated the myofibroblast phenotype at 3 and 5 days, cultures are often confluent at that time and are subject to contact inhibition. Therefore, we assessed proliferation after 2 days of cytokine treatment to evaluate the impact of the cytokines on proliferation before contact inhibition as the cells reach confluence. To detect DNA synthesis as an index of proliferation, bromodeoxyuridine (BrdU, final concentration 1 μM; Sigma) was added for the last 5 hours of incubation. Cells were fixed with 3% p-formaldehyde, DNA was denatured with 3 N HCl, and BrdU labeling was immunodetected with mouse monoclonal anti-BrdU antibody (Sigma), followed by anti-mouse IgG conjugated to FITC (Jackson Laboratories). Cells were counted and scored for presence and absence of BrdU in their nuclei in 5 or 6 randomly selected fields in each sample (minimum of 243 cells/coverslip were counted). Two independent proliferation studies detecting BrdU were performed, the results were averaged, and the mean was determined for each of the two experiments. 

As an independent index of proliferation, fixed cells were immunodetected with mouse monoclonal anti-Ki67 (Sigma) followed by anti-mouse IgM-FITC. Ki-67 is a nuclear proliferation factor expressed at all stages of the cell cycle except G₀. ²⁷ Nuclear staining by Hoechst dye 33258 detected essentially no apoptosis after 2 days in any of the growth conditions. ²⁸  

Cells were viewed with an epifluorescent Zeiss Axiophot (Thornwood, NY) and photographed on Kodak TMAX 400 film. 

Detection of Cell Surface Expression of TGF-β Receptors

As in the proliferation studies, we detected cell-surface expression of TGF-β receptors 2 days after plating. Cell surface proteins were detected by incubation of cultures with NHS-sulfo biotin, 0.5 mg/ml, for 30 minutes at 0 to 4°C (Pierce). NP-40 lysates with equalized protein were precleared with normal rabbit serum (Life Technologies). TGF-βRI and -RII were immunoprecipitated using polyclonal receptor specific antibodies (described above; Santa Cruz). Antibody–antigen complexes were captured with Protein A agarose beads (Life Technologies) and eluted by boiling in SDS. ¹⁴ Ten microliters of each immunoprecipitate was electrophoresed and transferred to nitrocellulose as described above. Biotinylated cell surface TGF-β receptors were detected with streptavidin-HRP (Jackson Laboratories) followed by ECL. ¹⁴  

Results

FGF and Heparin Administration Decreased α-SM Actin Protein Expression in Myofibroblast Cultures

We had previously noted that occasional cultures were myofibroblastic in their first passage and remained myofibroblastic in later passages without addition of TGF-β (unpublished observation). By treating fibroblast cultures with 0.25 to 1 ng/ml exogenous TGF-β, we produced myofibroblast cultures that were similarly “persistent” as defined by immunocytochemistry. These cultures were comprised of> 60% myofibroblasts at confluency in two successive high-density passages (1:2 or 1:3 split). In contrast, standard fibroblast cultures passaged and grown under the same conditions have <25% myofibroblasts. ^{7 14 17 29}  

Figure 1 illustrates the myofibroblastic nature of such a persistent myofibroblast culture (passage 5), fixed and immunodetected 3 days after passaging at intermediate density. Similar to its progenitor culture, the majority of cells are myofibroblasts (Fig. 1A) . 

We investigated the ability of FGF to induce fibroblast differentiation of these persistent myofibroblast cultures, “reversing” the myofibroblast phenotype. In Western blot analysis the myofibroblastic nature of the cultures was confirmed by α-SM actin expression in cells grown in 1% or 10% FBS (Fig. 2) . TGF-β treatment of the cells increased the immunodetectable α-SM actin protein (Fig. 2) . In contrast, decreased α-SM actin expression was seen in lysates of cells incubated with 5 μg/ml of heparin plus FGF-1 or -2 (Fig. 2) . By immunocytochemistry, we found that the majority of cells were fibroblasts after 3 days of FGF–heparin treatment (Fig. 1D) . Treatment with either 20 ng/ml FGF (Fig. 1B) or 5μ g/ml heparin alone (Fig. 1C) did not achieve this. On the basis of the protein and the immunocytochemical data, we used 20 ng/ml FGF-2 and 5 μg/ml heparin (FGF–heparin) in all subsequent experiments. FGF-1 with heparin was equally effective in inducing the fibroblast phenotype. 

Proliferation, as indicated by incorporation of BrdU, was stimulated by serum, and even more by FGF–heparin in 10% FBS (Table 1) . The lowest incorporation of BrdU was found in cells grown in 1% serum with or without 0.25 ng/ml TGF-β. Serum and growth factor effects were confirmed in separate coverslips stained for Ki-67 (data not shown). 

In some experiments in which the myofibroblast phenotype was induced by 1 ng/ml TGF-β treatment, we found that replated myofibroblast cultures lacked immunodetectable Ki67 and therefore were in G₀. FGF and heparin addition could not reverse the myofibroblast phenotype when the majority of myofibroblasts were in G₀, further suggesting that the ability to reverse phenotype is linked to the ability to proliferate. 

Serum Factors Influence Growth Factor Induction of Phenotype

As seen in Figure 2 , persistent myofibroblast cultures maintained their myofibroblast phenotype when grown in the presence of either 1% or 10% FBS without addition of specific growth factors. However, the effect of FGF–heparin was dependent on the concentration of serum in the culture medium and was most effective in promoting the fibroblast phenotype in 10% FBS. In fact the FGF–heparin–treated cultures in 10% FBS were fibroblastic even if TGF-β was added. In contrast, myofibroblast induction by TGF-β was more effective in low serum (1% FBS), and in contrast to cultures in 10% FBS, cultures in 1% FBS remained approximately 50% myofibroblasts even when FGF-2 and heparin were added. Thus, the myofibroblast phenotype was optimized by TGF-β in 1% FBS, and the fibroblast phenotype was optimized by FGF–heparin in 10% FBS. 

TGF-β Receptor Expression Is Influenced by Serum and Growth Factors

We confirmed our earlier finding that TGF-βRI and -RII were highly expressed in myofibroblasts ²² (Fig. 3B , lanes 1, 3, and 4). In contrast, FGF–heparin treatment decreased both total and cell surface expression of TGF–βRI and -RII as well as decreasing α-SM actin (FGF-hep in Figs. 3A and 3B ). 

Cadherin and Cx43 Expression after FGF–Heparin Treatment

Cadherins are calcium-dependent cell adhesion molecules that are expressed more at cell–cell junctions in myofibroblasts than fibroblasts. ^{20 22} The persistent myofibroblast cultures had high expression of cadherins detected with anti–pan-cadherin (Fig. 4) or with anti–cadherin-11 (not shown), paralleling the maintenance ofα -SM actin expression. In contrast, treatment of the replated cells for 3 days with FGF–heparin induced fibroblastic cultures that expressed less cadherin than the TGF-β–treated cultures (Fig. 4 , lane 2). 

Cx43 is a gap junction protein that mediates cell–cell communication, and we have reported that unlike cadherins, Cx43 is more highly expressed in fibroblasts than in myofibroblasts. ^{20 22} In the current studies, we compare the expression of Cx43 in fibroblasts and myofibroblasts derived from myofibroblasts and grown for 3 days. In these cultures, FGF–heparin treatment increased the Cx43 expression, paralleling the fibroblast phenotype induction (Fig. 4 , lane 2). Differences in cadherin and connexin expression between fibroblasts and myofibroblasts are greater with longer periods after replating (Taliana L, Masur SK, unpublished results, 2000). 

TGF-β Induced α-SM Actin mRNA Expression in Myofibroblasts

We evaluated the impact of replating and TGF-β or FGF–heparin on α-SM actin mRNA (Fig. 5) . As we had for protein expression, we also evaluated the impact of the concentration of FBS, 1% and 10%. Thus, we prepared RNA from myofibroblasts under various growth conditions: (1) 10% FBS alone or with (2) FGF-2 and heparin, (3) heparin alone or (4) FGF alone, and (5) 1% FBS alone or with (6) recombinant TGF-β or (7) human platelet TGF-β. 

We performed Northern blot analysis using cDNA for α-SM actin mRNA (Fig. 5A) . A 1.3-kb band corresponds to the predicted size for α-SM actin mRNA (arrowhead), based on the size of α-SM actin cited in GenBank (accession no. NM007392). We normalized for loading differences by detecting with a cDNA probe for 18S ribosomal RNA (Fig. 5B) . Myofibroblast cultures, trypsinized and replated, had minimal detectable α-SM actin mRNA except after TGF-β. Treatment with TGF-β increased the expression of the 1.3-kb band 2.3- and 2.9-fold (Fig. 1C) . This contrasts with the finding that cells grown with heparin or FGF or in 1% or 10% FBS continue to express α-SM actin protein (Figs. 1 and 2) . This suggests that α-SM actin protein is sufficiently stable to maintain the myofibroblast phenotype after replating in FBS without an increase in α-SM actin mRNA synthesis. The increase in α-SM actin mRNA induced by TGF-β, compared with cells grown in all other conditions, could occur either by new synthesis or by stabilization of α-SM actin mRNA. 

Discussion

Our cell culture studies support the hypothesis that FGF promotes fibroblast differentiation and TGF-β promotes myofibroblast differentiation, respectively. We have determined that treatment with FGF-1 or -2, heparin, and 10% FBS can induce myofibroblast cultures to convert to the fibroblast phenotype as characterized by reduced expression of α–SM actin, cadherins, and TGF-β receptors. The reverse, the conversion of fibroblasts to myofibroblasts induced by TGF-β₁, has been well-documented previously. The resultant myofibroblasts are characterized by increased expression of SM α-actin, TGF-β receptors and decreased Cx43 expression. ^{16 20 22}  

The current studies were initiated because we noted that on occasion myofibroblast cultures arose spontaneously. Typical fibroblast cultures have fewer than 25% myofibroblasts. In contrast, we defined“ persistent myofibroblast cultures” as cultures passaged and grown under standard culture conditions that were comprised of >60% myofibroblasts at confluence. It is of interest to speculate on the origin of the spontaneous myofibroblast cultures. In the present study we produced persistent myofibroblasts by adding 0.25 to 1 ng/ml TGF-β to primary cultures. Thus, we reason that the originating keratocytes in the spontaneously persistent myofibroblast cultures may have been exposed to large quantities of TGF-β before or early in their isolation. 

Our current finding that fibroblast phenotype induction by FGF requires heparin is consistent with previous demonstrations that heparin promotes FGF signaling. Most models of FGF–heparin interaction propose that heparin’s impact is on the binding of FGF to its cell surface receptor. ³⁰ Heparin causes oligomerization of FGF and binding of oligomerized FGF to its receptors results in receptor dimerization, thus activating receptor tyrosine kinase and initiating the biological responses to the growth factor. ³¹ It is possible that FGF-2, bound to heparan sulfate proteoglycan, is translocated to the nucleus where signaling may occur. ³² Although it has been demonstrated recently that the requirement for heparin may be eliminated by raising FGF concentration to 50 ng/ml, this was not true for the corneal myofibroblasts (data not shown). ³³  

We do not know whether the growth factor–induced changes are part of a cassette of genes that is regulated by each growth factor, if the changes are independent effects, or if they occur sequentially. It is of interest that FGF and TGF-β, which have opposite effects on the phenotype, signal via different pathways: FGF receptors are tyrosine kinases and TGF-β receptors are serine/threonine kinases. There is recent evidence that these pathways converge on regulation of the Smad proteins, which are downstream of the TGF-β receptors. ^{34 35} Smads are 45- to 70-kDa proteins with high sequence similarity to the Drosophila “Mad” proteins at their N- and C-terminal ends. Transcriptional activation by TGF-β requires translocation of Smad proteins into the nucleus. ^{22 36} Because FGF may induce phosphorylation of residues within the region linking inhibitory and stimulatory domains of Smads, signaling through the FGF receptor may prevent TGF-β–induced Smad translocation. ³⁴  

A different interplay between TGF-β and FGF in fibroblast proliferation has been reported, in which proliferation is promoted by TGF-β via stimulation of FGF secretion. ³⁷ We could not evaluate this in our current studies because the influence on proliferation of 1% versus 10% FBS was as great as the specific growth factor. However, in terms of phenotype regulation, our results indicate that TGF-β and FGF can induce opposite effects in promoting the myofibroblast or fibroblast phenotype. Future studies are needed to evaluate the intersection of serum effects on cell cycle and impact of these and other growth factors on phenotype regulation. ³⁸  

Our results reinforce the concept that the myofibroblast is not a terminally differentiated cell. We have previously demonstrated that the phenotype in myofibroblast cultures generated by passaging at very low cell density is reversible: if myofibroblasts generated from low-density passage are passaged at high density, the resultant culture is fibroblastic. ¹⁸ Our current findings demonstrate that FGF-heparin can induce the fibroblast phenotype in previously myofibroblast cultures. Thus, by controlling cell density, growth factor concentration, heparin and serum content, it is possible to push a population of cells to the fibroblastic or myofibroblastic phenotype and foster wound repair and closure while diminishing scar formation. We do not know whether the myofibroblast-to-fibroblast reversion will occur in situ, but these studies provide evidence that it is possible. 

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

 Submitted for publication January 3, 2001; revised May 23, 2001; accepted June 11, 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.

 

FGF and heparin in 10% FBS promote the fibroblast phenotype in a“ persistent myofibroblast” culture. Myofibroblasts are identified by anti–α-SM actin in (A) through (D). Nuclei are visualized by Hoechst stain in the identical microscopic fields in (E) through (H). All cells were passaged at 200 cells/mm²; growth factors were added 16 hours later in DMEM-F12, 10% FBS, and cells were fixed after 3 days. (A) The majority of cells expressed α-SM actin in this passage and the preceding passage. (B) Myofibroblasts remained the majority cell type when 20 ng/ml FGF-1 was administered, without heparin. (C) Similarly myofibroblasts are the majority cell type in cultures treated with 5 μg/ml heparin for 3 days, without FGF. (D) Fibroblasts are the predominant cell type in cultures treated with both 20 μg/ml FGF-1 and 5 μg/ml heparin. In (D), the few myofibroblasts (α-SM actin–positive cells) “sit” on top of fibroblasts, suggesting that cell–cell or cell–matrix contact also plays a role in myofibroblast differentiation. All micrographs are the same magnification. Bar in (D), 40 μm.

Figure 2.

 

FGF-1- or -2–heparin inhibits α-SM actin expression. In this representative set of cultures replated from a single myofibroblast culture, we added 1, 10, 20, 40, or 80 ng/ml FGF in the presence of 5μ g/ml heparin in 10% FBS for 3 days and then detected α-SM actin in the Western blot analysis of NP-40 cell lysates. FGF-1 (20 ng/ml) and FGF-2 (1 ng/ml) effectively inhibited α-SM actin expression. In contrast, control cultures grown in 10% FBS or 1% FBS for 3 days express high levels of α-SM actin and are classified as persistent myofibroblasts. Addition of 0.25 ng/ml TGF-β to 1% FBS medium increased the expression of α-SM actin.

Table 1.

 

Impact of Growth Factors and Serum on Proliferation (2 days)

Medium	BrdU Incorporation
10% FBS+ FGF	0.74, 0.61
10% FBS	0.50, 0.33
1% FBS+ TGF	0.21, 0.20
1% FBS	0.25, 0.18

 Cells were counted and scored for presence and absence of BrdU in their nuclei; coverslips were costained with Hoechst dye to obtain the total number of nuclei. Two independent proliferation experiments were performed. A minimum of 243 cells per coverslip were counted for each experiment using 5 or 6 randomly selected fields. The results are shown as the average incorporation for each of the two experiments, separated by a comma.

Figure 3.

 

FGF-2–heparin treatment reduces the expression of immunodetectable TGF-βRI and -RII. (A) Western blot analysis of total expression of TGF-β receptor types I (RI) and II (RII) in cell lysates. The receptors were immunodetected in equal amounts of protein in total cell lysates in separate, paired gels using polyclonal receptor specific antibodies and ECL. FGF-2–heparin treatment decreased expression of α-SM actin, and the total expression of TGF-βRI and -RII (lane 2). (B) Cell surface expression of TGF-β receptors was detected by labeling cells with biotin followed by immunoprecipitating the specific receptors from the cell lysate. Streptavidin-HRP was used to detect biotinylated cell surface receptors. FGF–heparin (20 ng/ml FGF-2) decreased the surface expression of TGF-βRI and -RII (lane 2), whereas, 1% FBS and TGF-β (1 ng/ml) treatment increased the cell surface expression of TGF-βRII (lanes 3 and 4).

Figure 4.

 

FGF–heparin (20 ng/ml FGF-2) treatment for 3 days decreased cadherin expression and increased Cx43 expression. The molecules were immunodetected in equal amounts of protein in total cell lysates in separate, paired gels using specific antibodies and ECL. Cadherins were detected with pan-cadherin antibody as well as anti–cadherin-11 (not shown) in all cultures grown in 10% or 1% FBS. Treatment with FGF–heparin decreased the amount of cadherin (lane 2). In contrast, Cx43 is more highly expressed after FGF-2–heparin as we have previously demonstrated in fibroblasts. ²⁰

Figure 5.

 

α-SM actin mRNA expression in myofibroblasts is increased by treatment with TGF-β. (A) Northern blots were analyzed forα -SM actin mRNA in total RNA as described in Material and Methods. The cultures were treated for 24 hours as indicated: (1) DME-F12 plus 10% FBS; (2) 20 ng/ml FGF-2, 5 μg/ml heparin in DME-F12, 10% FBS; (3) 5 μg/ml heparin in DME-F12, 10% FBS; (4) 20 ng/ml FGF-2 in DME-F12, 10% FBS; (5) DME-F12, 1% FBS; (6) 1 ng/ml recombinant TGF-β in DME-F12, 1% FBS; and (7) 10 ng/ml human platelet TGF-β in DME-F12, 1% FBS. (B) The Northern blot was reprobed for the expression of the 18S ribosomal RNA to normalize for loading differences. The expression of α-SM actin message at 1.3 kb was greatest in TGF-β–treated cells (lanes 6 and 7; A and C). TGF-β–treated cells had 2.3- to 2.9-fold α-SM actin mRNA compared with cells grown in other conditions. The signal at 4.4 kb is 28S ribosomal RNA and also indicates relative amounts of RNA loaded in each lane. ⁸ (C) Densitometric analysis of α-SM actin mRNA signal normalized to 18S RNA (y-axis) shows significant increases in SM α-actin mRNA after TGF-β treatment. x-axis: gel lane numbers correspond to descriptions for (A).

The authors thank Scott Henderson, Center for Microscopy, Mount Sinai School of Medicine, for advice on imaging; Mitchell Goldfarb, Brookdale Center for Molecular Biology, Mount Sinai School of Medicine for advice on FGF and heparin; and Ed Fisher for use of the BioRad 1650 densitometer and Image Quant software.