Myofibroblast Differentiation of Normal Human Keratocytes and hTERT, Extended-Life Human Corneal Fibroblasts

Abstract

purpose. The purpose of this study was to determine whether TGFβ induces myofibroblast differentiation in cultured human keratocytes and in telomerase (hTERT)-immortalized human corneal fibroblast cell lines.

methods. Normal human corneal keratocytes were isolated from donor corneas of various ages and grown under serum-free (cultured keratocytes) or serum-added (corneal fibroblasts) conditions. Corneal fibroblasts were infected with the MPSV-hTERT retroviral vector, and selected clones were isolated and characterized by chromosomal karyotyping. The responses of normal cultured keratocytes and serum-starved corneal fibroblasts to TGFβ in the presence or absence of Arg-Gly-Asp (RGD)-containing peptides and neutralizing antibodies to platelet-derived growth factor (PDGF) were characterized by immunocytochemistry, Western blot analysis, and real-time PCR, to identify assembly of actin filaments, formation of focal adhesions, and expression of α-smooth muscle actin (α-SMA).

results. Treatment of cultured keratocytes with TGFβ (1 ng/mL) induced cell spreading, assembly of actin filaments, formation of focal adhesions, and expression of α-SMA, which was blocked by the addition of RGD-containing peptides (100 μM). A similar response was identified in hTERT-expressing human corneal fibroblast cell lines, showing a 69-fold increase in α-SMA message. Furthermore, treatment of hTERT corneal fibroblasts with RGD or anti-PDGF inhibited myofibroblast differentiation. Karyotype analysis of hTERT corneal fibroblasts identified age-dependent chromosomal aberrations in cells of older donors but not in those of a 10-year-old donor.

conclusions. Induction of myofibroblast differentiation by TGFβ in cultured human keratocytes and hTERT corneal fibroblasts occurs through a similar signal transduction pathway to that previously identified in the rabbit, which involves an autocrine PDGF feedback loop.

Recent studies have shown that transforming growth factor (TGF)-β induces the differentiation of rabbit and bovine stromal keratocytes to a myofibroblast phenotype characterized by the development of prominent actin filament bundles and the expression of the smooth-muscle-specific α isoform of actin (α-SMA). ^{1 2 3} Myofibroblasts play a critical role in wound repair by organizing into a unique, interwoven network of interconnected cells that exert mechanical force through a smooth-muscle-like contractile mechanism, resulting in wound contraction and matrix organization. ^{4 5} Myofibroblasts also show unique differences in expression of cell-cell and cell-matrix adhesion proteins that further support a contractile role in the wound repair process. ^{6 7} More recently, the appearance of myofibroblasts has been associated with the development of corneal haze after refractive surgery in patients and animals, ⁸ and the appearance of haze in the rabbit has been effectively blocked by treatment with neutralizing antibodies to TGFβ. ^{9 10}  

Studies elucidating the role of TGFβ in corneal myofibroblast differentiation suggest that TGFβ is synthesized by the corneal epithelium ¹¹ or, alternatively, by corneal keratocytes in an autocrine fashion after injury or culture at low cell density, at least in the rabbit. ^{1 12} TGFβ binding to transmembrane serine-threonine kinase receptors then initiates a phosphorylation cascade, leading to the activation of signaling peptides belonging to the similar-to-mothers-against-decapentaplegic (Smad) family. ^{13 14} In corneal keratocytes, phosphorylation of Smad 2 leads to nuclear transport effecting downstream expression of proteins related to myofibroblast differentiation—for example, α-SMA, fibronectin, and α-actinin. ^{15 16} Final myofibroblast differentiation may also require synergistic interactions with extracellular matrix, matrix receptors, and other growth factors with fibronectin, α₅β₁ integrin, and platelet-derived growth factor (PDGF) putatively playing important roles in controlling the downstream differentiation process. ^{17 18}  

Although much of this work supports the hypothesis that TGFβ is the key regulator of myofibroblast differentiation, studies establishing the requirement of synergistic signal-transduction pathways that, when inactivated, lead to the inhibition of myofibroblast differentiation in the presence of TGFβ suggest multiple levels of control. Furthermore, studies in different animal systems suggests that these downstream modulators may be species specific with, for example, bovine corneal keratocytes requiring exogenous growth factors or serum components in contrast to the autocrine regulatory pathways identified in the rabbit (Funderburgh JL, Funderburgh ML, Mann MM, ARVO Abstract 4958, 2001). ³  

Whether human corneal keratocytes show a similar complexity in the TGFβ-induced myofibroblast differentiation process has yet to be established, in part because of the difficulty of obtaining sufficient tissue to conduct appropriately controlled studies. Recently, the generation of extended-life-span cells has been developed by using transfection with the human telomerase reverse transcriptase (hTERT) gene. Telomerase is a ribonucleoprotein that synthesizes telomeric DNA (tandem repeats of the sequence TTAGGG) onto the chromosomal ends. ¹⁹ Adult mammalian cells show no telomerase activity, and thus tissues show shortened chromosomal telomeres with increased age. It has been proposed that telomere shortening limits the number of divisions a cell can undergo before cell senescence in culture, and it is believed, but has not been formally shown, that this may have important consequences in in vivo cellular aging. ²⁰ Restoration of telomerase activity in cells without telomerase results in lengthening of the telomeres and extension or unlimited proliferation (immortalization), of the cell’s life span, ²¹ whereas excision of the inserted telomerase gene leads to a reestablishment of a defined life span. ²² The absence of telomerase in human corneal epithelial and endothelial cells has also been demonstrated, suggesting that telomere shortening may be responsible for senescence of human corneal cells in culture. ²³ There are several unique advantages to extension of cellular life span by the hTERT approach, in that cells maintain normal phenotypic or morphologic characteristics, exhibit normal contact inhibition and loss of growth under reduced serum concentrations, and maintain a normal karyotype, unlike many virally transformed cell lines. ²¹  

This article presents for the first time experimental evidence suggesting that TGFβ-induced myofibroblast differentiation of normal human keratocytes in culture requires a synergistic, RGD-dependent signal-transduction cascade similar to the rabbit keratocyte. In addition, we report that human corneal fibroblast cell lines derived from normal human keratocytes expressing the hTERT gene had a myofibroblast differentiation pathway similar to that of normal human and rabbit keratocytes, in that the induction of the myofibroblast phenotype required both integrin and multiple growth factor signaling. 

Materials and Methods

Cell Culture

Human corneas were obtained fresh from the Eye Bank at the University of Texas Southwestern Medical Center. Corneas were washed in sterile DMEM culture medium (Gibco-BRL, Life Technologies; Grand Island, NY), and an 8-mm diameter corneal button was obtained from the central cornea with a trephine. The corneal epithelium was scraped from the corneal button with a scalpel blade, and Descemet’s membrane was removed. The stromal buttons were digested overnight with 2.0 mg/mL collagenase and 0.5 mg/mL hyaluronidase in DMEM (all from Gibco-BRL) at 37°C. Isolated cells were washed in DMEM and resuspended in DMEM supplemented with 10% fetal calf serum (Gibco-BRL) 1% glutamate and 1% penicillin-streptomycin for hTERT transfection or cultured under serum-free conditions in DMEM supplemented with 1% RPMI vitamin mix, 100 μM nonessential amino acids, 1 mM pyruvate, 100 μg/mL ascorbic acid, and 1% glutamate with antibiotics (all from Gibco-BRL) for characterization of normal keratocyte response to TGFβ. 

For the purposes of this article, hTERT-expressing corneal keratocytes will be referred to as corneal fibroblasts to denote their exposure to serum factors that induce a fibroblastic phenotype differentiated from a normal corneal keratocyte phenotype, as recently demonstrated by Beales et al. ²⁴ Human corneal keratocytes cultured in serum-free conditions will be referred to as cultured keratocytes, in that studies in rabbits suggest that these cells maintain, at least partially, a normal quiescent keratocyte phenotype in cytoskeletal organization and expression of extracellular matrix proteins. ^{2 24}  

For studies characterizing the effects of TGFβ on cultured keratocytes, freshly isolated keratocytes were plated as primary cultures onto 1-cm diameter, type I collagen-coated (Vitrogen; Cohesion, Palo Alto, CA) glass coverslips at 5 × 10⁴ cells/cm². Cells were allowed to attach and grow for 4 to 5 days before treatment with TGFβ (1 ng/mL; Sigma, St. Louis, MO), in the presence or absence of different concentrations of the RGD-containing peptide GRGDdSP, specific for the fibronectin receptor, or RAD control peptide, both made at the Biopolymers Facility of the Howard Hughes Medical Institute, University of Texas Southwestern Medical Center. Cells were then cultured for 3 to 7 days before they were fixed with 1% paraformaldehyde in phosphate-buffered saline and processed for immunocytochemistry, as will be described later. 

For studies characterizing the effects of TGFβ on hTERT corneal fibroblasts, cells were serum starved by culturing in serum-free medium for 1 week followed by continued passage in serum-free medium for up to four to five passages. Cells were then plated onto 100-mm diameter, polystyrene surface modified tissue culture dishes (Primaria; Falcon Labware; BD Biosciences, Franklin Lakes, NJ) or type I collagen-coated (Vitrogen; Cohesion) glass coverslips at a density of 5 × 10⁴ cells/cm² for respective biochemical and immunocytochemical studies. Cells were allowed to attach for 48 to 72 hours before the addition of growth factors and/or inhibitors including TGFβ1; PDGF-BB (GibcoBRL); neutralizing antibodies to PDGF-AB (R&D, Minneapolis, MN); the peptide sequence GRGDdSP (GibcoBRL), known to block fibronectin receptor ligand binding ^{17 18} ; and RAD (GibcoBRL), the control peptide sequence to GRGDdSP. Cells were grown for various times and then fixed in 1% paraformaldehyde in phosphate-buffered saline or protein collected for Western blot analysis. All experiments were repeated at least three times. 

hTERT Transfection and Cloning

Primary corneal keratocytes were plated onto 100-mm tissue culture plates (Falcon Labware; BD Biosciences) and allowed to attach and grow for 48 to 72 hours in serum-containing medium. Subconfluent cultures during the active proliferative phase were then infected with the hTERT gene, using a modification of a previously described method. ²¹ In brief, the myeloproliferative sarcoma virus MPSV-hTERT vector was cloned into a retroviral vector (pBABE-puro). Retroviral vector particles were generated by transfecting 30 μg of vector DNA into PhoenixE cells with a transfection reagent (Fugene; Roche Molecular Biochemicals, Indianapolis, IN), according to the manufacturer’s instructions. Supernatants collected 48 hours after transfection were used to infect the amphotrophic packaging cell line PA317 (CRL-9078; American Type Culture Collection, Manassas, VA). After a 7-day selection with puromycin (3 μg/mL), supernatants containing the retroviral particles were harvested for infections and filtered through a 0.45-μm filter. 

Infecting medium containing 4 μg/mL polybrene was added and the cells cultured overnight. Additional infecting medium was then added and the cells cultured for an additional 12 hours. Infecting medium was then replaced with fresh serum-containing culture medium and the cells cultured for an additional 72 hours before passing at low density (10³/100-mm dish) and selection with puromycin (1000 ng/mL). Individual clones were then isolated and expanded for later characterization. 

Karyotype Analysis

Selected myoplasma-free cell lines were submitted to the University of Texas Southwestern Medical Center Cryogenic Laboratory for karyotype analysis. Cells were initially plated in serum-containing medium for 48 hours, treated with colcemid (0.01 mg/mL), lysed for 20 minutes in a hypotonic solution of medium containing three parts medium and 1 part 0.075 M KCl and Na citrate, and fixed in methanol-acetic acid. Chromosomes were stained with Wright solution, and 20 cells in metaphase from each cell line were analyzed. 

Immunocytochemistry and Western Blot Analysis

Cells were washed in fresh, serum-free medium and fixed in medium containing 1% paraformaldehyde for 3 minutes. Fixed cells were then permeabilized with acetone (−20°C) and stained with the following probes: FITC-conjugated phalloidin (Molecular Probes, Eugene, OR) to label F-actin, anti-α-SMA (Sigma) to identify myofibroblast differentiation, and anti-vinculin (Serotec, Raleigh, NC) to label focal adhesions; both monoclonal antibodies stained with rhodamine-conjugated goat anti-mouse IgG (ICN, Aurora, OH). Cell staining was evaluated by epifluorescence microscope (Diaplan; Leica, Deerfield, IL) adapted for digital image capture. To colocalize probes, images were overlaid on computer (Photoshop; Adobe, San Jose, CA). 

Proteins were run on 10% SDS polyacrylamide gels, transferred to nitrocellulose, and probed with antibodies to α-SMA (Sigma) or phosphotyrosine (Clone 4G10; Upstate Biotechnology, Lake Placid, NY). 

Measurement of α-SMA mRNA Expression

Cells (4 × 10⁶) were lysed with 1 mL reagent (TRI; MRC, Cincinnati, OH), directly in a culture dish, and total RNA was extracted. The quality of each RNA preparation was checked by measuring optical density (OD)₂₆₀/OD₂₈₀ and electrophoretically by detection of intact 18S and 28S ribosomal RNA bands. First-strand cDNA was synthesized by with reverse transcriptase (Omniscript; Qiagen, Valencia, CA) and oligo-dT primer with 2 μg total RNA. For PCR amplification, primers were based on the GenBank (http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) human sequences for α-SMA between base pairs 922 and 1277 (upper primer: 5′-AGGAAGGACCTCTATGCTAACAAT-3′ and lower primer: 5′-AACACATAGGTAACGAGTCAGAGC-3′) and for human β-actin between base pairs 1212 and 1681 (upper primer: 5′-ACTTAGTTGCGTTACACCCTTTCT-3′; lower primer: 5′-TTCATACATCTCAAGTTGGGGGAC-3′). Real-time PCR was performed in triplicate for each sample, with a commercial system (iCycler; Bio-Rad, Hercules, CA; and QuantiTect SYBR Green PCR; Qiagen). Reaction volumes of 50 μL were run in 96-well plates covered with optical sealing tape (Bio-Rad). Amplification was performed at 95°C for 15 minutes for initial activation and then 30 cycles of denaturation at 95°C for 1 minute, annealing at 55°C for 1 minute, and elongation at 72°C for 1 minute were performed. To confirm the specificity of the PCR reaction, products were separated by electrophoresis on a 1% agarose gel and visualized by ethidium bromide staining. 

Message levels for α-SMA and β-actin were quantified by comparing the critical threshold (C_T) obtained from unknown samples with the C_T obtained from purified PCR products from α-SMA and β-actin. Quantities are reported based on log quantity in nanograms of the known products. Individual samples were run in triplicate and the experiment repeated twice. Values are reported as mean and SD of results in a single experiment. 

Results

Normal Human Keratocytes

Human keratocytes when cultured under serum-free conditions maintained a dendritic or astrocyte-like morphology that has been previously observed in both cultured rabbit and bovine keratocytes (Fig. 1A) . ^{2 24 25} In addition, staining of cells with phalloidin showed a sparse actin network predominantly organized around the cell cortex with few or no intracellular actin filament bundles (Fig. 1B) . Consistent with these observations was the finding that cultured keratocytes showed no staining with antibodies for α-SMA specific for myofibroblasts and only minimal staining with antibodies to vinculin, which localizes to adhesion complexes (Fig. 1C 1D , respectively). Overall, these morphologic characteristics were consistent with those of normal and cultured keratocytes from other species. 

Addition of TGFβ (1 ng/mL) to serum-free medium and continued culture for 3 to 7 days resulted in the proliferation and spreading of cultured keratocytes consistent with myofibroblast differentiation (Fig. 2A) . Compared with cultured keratocytes, TGFβ-treated cells had greatly enhanced staining with phalloidin, showing the organization of actin filaments into prominent intracellular bundles consistent with the formation of stress fibers (Fig. 2B) . When cells were stained with antibodies to α-SMA, staining was localized to actin bundles, indicating that cultured keratocytes had differentiated to a myofibroblast phenotype (Fig. 2C) . Prominent anti-vinculin staining was also detected, indicating a marked upregulation of focal adhesion formation, again consistent with myofibroblast differentiation (Fig. 2D) . Overall, the temporal development of α-SMA expression was somewhat variable, with keratocyte cultures established from older individuals requiring a longer exposure time to TGFβ than younger keratocytes. Nevertheless, staining for α-SMA was always observed by 10 days in culture, with younger cells showing marked staining by day 3. 

Previous studies using cultured rabbit keratocytes have shown myofibroblast differentiation to be dependent on integrin signaling and that RGD-containing peptides known to block integrin binding to extracellular fibronectin inhibited both TGFβ-mediated integrin signaling and expression of α-SMA. ^{17 18} We therefore evaluated the role of integrin signaling in the differentiation of cultured human keratocytes into myofibroblasts by treating cells with TGFβ alone (1 ng/mL) or in combination with the RGD-containing peptide GRGDdSP (100 μM), specific for the fibronectin receptor, ¹⁸ or the control peptide RAD (100 μM). Cells were cultured under these conditions for periods varying from 3 to 10 days, depending on donor age, and then stained with phalloidin to evaluate actin filament organization and stained for anti-α-SMA to identify myofibroblast differentiation (Fig. 3) . Cultures treated with TGFβ alone showed cell spreading and formation of prominent actin filament bundles that stained for the presence of α-SMA (A, B, respectively). When GRGDdSP was added with TGFβ in parallel cultures from the same donor, prominent actin filament bundles did not develop in the cultured keratocytes nor did they stain for the presence of α-SMA (Figs. 3C 3D , respectively). An important finding was that cells treated with the control peptide RAD showed a normal response to TGFβ, having prominent actin filament bundles and expression of α-SMA (Figs. 3E 3F , respectively). Overall, the results of these experiments indicate that myofibroblast differentiation of normal human corneal keratocytes requires synergistic integrin and TGFβ signaling. Further biochemical and molecular characterization of the TGFβ signal-transduction cascade was not performed, because establishing a sufficient number of quiescent normal human keratocytes in culture would require large amounts of donor corneal tissue. 

hTERT Human Corneal Fibroblasts

Human Corneal Fibroblast Karyotype.

Corneal fibroblasts were isolated from donor corneas obtained from a 62-year-old and a 10-year-old donor. After infection with hTERT, 40 clones were selected and isolated from the 62-year-old donor, and 22 clones were selected and isolated from the 10-year-old donor. After clones were isolated, cells were expanded in culture and evaluated for chromosomal abnormalities (Fig. 4) . Corneal fibroblasts from the 62-year-old donor showed extensive chromosomal abnormalities that included loss of one sex chromosomes, polyploidy, and various chromosomal translocations and deletions (Fig. 4A) . Karyotype analysis of uninfected corneal fibroblasts from an older donor showed similar chromosomal abnormalities consistent with the findings of Pattenati et al., ²⁶ which indicates that chromosomal aberrations develop in human corneal keratocytes after the age of 18. Karyotype analysis of the 10-year-old donor showed a normal chromosomal pattern of 46, XY (Fig. 4B) . One clone was then selected from the 62-year-old donor and 15 clones selected from the 10-year-old donor for further characterization and continued growth. The clone from the 62-year-old donor (HTK cells) has undergone a total of 144 population doublings, and the oldest clone from the 10-year-old (HTKm cells) has undergone 89 population doublings. Characterization of clones from both the older and the younger donor showed similar responses, regardless of donor age. Results are therefore presented for only the HTK cells but also apply to the HTKm cells. 

Effects of Serum Starvation and TGFβ on Telomerized Human Corneal Fibroblasts.

HTK cells in serum culture showed a typical spindle-shaped fibroblast morphology with a small population (<10%) of cells showing expression of α-SMA, a pattern similar to that already shown in rabbit corneal fibroblasts grown in serum culture (data not shown). ² HTK cells cultured under serum-free conditions maintained their characteristic spindle-shaped, fibroblastic morphology (Fig. 5A) and contained phalloidin stained, F-actin filaments (Fig. 5B) terminating in anti-vinculin-stained focal adhesions (Fig. 5C) , albeit in a reduced amount compared with serum cultured cells. Serum-free culture of human hTERT corneal fibroblasts resulted in a substantial reduction (<1%) of cells staining with α-SMA (Fig. 5D) . Treatment of serum-starved hTERT corneal fibroblasts with TGFβ (1 ng/mL) resulted in marked spreading of cells by 24 to 48 hours after treatment (Fig. 5E) and greatly enhanced phalloidin (Fig. 5F) and anti-vinculin (Fig. 5G) staining, suggesting marked upregulation of the formation of actin stress fiber. In addition, TGFβ-treated cells showed marked staining for α-SMA in greater than 75% of cells, suggesting myofibroblast differentiation of telomerized human corneal fibroblasts (Fig. 5H) . 

Effects of RGD Peptides on α-SMA Expression.

Addition of the RGD peptide GRGDdSP at a concentration of 100 μM to TGFβ containing serum-free medium (Fig. 6A) effectively blocked the cell spreading of human corneal fibroblasts that was detected with TGFβ alone (not shown) or TGFβ in combination with the control peptide RAD, at a concentration of 150 μM (Fig. 6B) . The RGD peptide also inhibited, whereas the RAD peptide had no effect on, the TGFβ-mediated upregulation of actin filament assembly detected by phalloidin staining (Figs. 6C 6D , respectively) or formation of focal adhesion detected by anti-vinculin staining (Figs. 6E 6F , respectively). Finally, RGD peptides blocked the TGFβ-mediated induction of α-SMA expression, whereas the RAD peptide had no effect (Figs. 6G 6H , respectively). To confirm that RGD-containing peptides block TGFβ-mediated α-SMA expression, hTERT corneal fibroblasts were cultured in various concentrations of GRGDdSP in the presence of 1 ng/mL TGFβ for 3 days. Cells were then collected and evaluated for expression of α-SMA by Western blot analysis. As shown in Figure 7 , there was a dose-dependent decrease in the TGFβ-induced expression of α-SMA with increasing concentrations of GRGDdSP from 25 to 75 μM. Furthermore, the level of α-SMA expression in cultures treated with 75 and 100 μΜ GRGDdSP in the presence of TGFβ was not detectably different from that of serum-starved corneal fibroblasts not exposed to TGFβ. Expression of message for α-SMA was also evaluated by real-time PCR (Fig. 8) . Addition of TGFβ (1 ng/mL) to serum-starved cultures of corneal fibroblasts resulted in a 69-fold increase in mRNA specific for α-SMA, whereas GRGDdSP completely blocked this effect (ratio = 1). No effect on β-actin message was detected. 

Effects of Neutralizing Antibodies to PDGF on Human Corneal Myofibroblast Differentiation.

Earlier studies in the rabbit have shown that the induction of myofibroblast differentiation by TGFβ involves a synergistic interaction between integrin, TGFβ, and at least one additional growth factor: PDGF. ¹⁷ To determine whether a similar synergistic signaling cascade is present in human corneal fibroblasts, the effects of neutralizing antibodies to PDGF on TGFβ-mediated myofibroblast differentiation were evaluated (Fig. 9) . Neutralizing antibodies to PDGF added at concentrations of 25 μg/mL completely blocked the TGFβ-induced expression of α-SMA (Fig. 9A) as well as the reorganization of actin filaments (data not shown) and formation of focal adhesions, as indicated by anti-vinculin staining (Fig. 9B) . When excess PDGF (100 ng/mL) was added to cultures in combination with TGFβ and anti-PDGF, corneal fibroblasts showed both expression of α-SMA (Fig. 9C) and the formation of prominent actin stress fibers and focal adhesions (Fig. 9D) . PDGF alone added to serum-starved cultures of hTERT human corneal fibroblasts maintained a fibroblastic, spindle-shaped morphology with little or no expression of α-SMA (Fig. 9E) , scant actin filament bundles and focal adhesions localized predominantly to the cell periphery (Fig. 9F) . 

Discussion

Recent studies in rabbits indicate that differentiation of corneal keratocytes into a myofibroblast phenotype involves a synergistic interaction between RGD-dependent, integrin signaling, coupled with TGFβ and PDGF receptor signal transduction (Funderburgh JL, Funderburgh ML, Mann MM, ARVO Abstract 4958, 2001). ^{3 17} The findings in the current study suggest for the first time that human corneal keratocytes and telomerized human corneal fibroblast cell lines show a similar dependence on integrin, TGFβ, and PDGF for induction of myofibroblast differentiation. Furthermore, the action of TGFβ in the human appears to involve an endogenous, PDGF-dependent, autocrine loop that, when blocked by the addition of specific neutralizing antibodies, inhibits TGFβ-induced myofibroblast differentiation. Although this finding is similar to that previously shown in the rabbit, it differs from that of others working with bovine keratocytes, which show that TGFβ does not induce myofibroblast differentiation without the addition of exogenous growth factors or serum (Funderburgh JL, Funderburgh ML, Mann MM, ARVO Abstract 4958, 2001). ^{3 27} These human, rabbit, and bovine studies convincingly demonstrate that there are multiple interactions regulating myofibroblast differentiation in various species, any one of which, when blocked, inhibits downstream induction of myofibroblast formation. The presence of these multiple-interacting signal-transduction pathways suggests that the control of corneal fibrosis in the cornea may be accomplished through diverse approaches. 

Understanding the underlying mechanisms controlling human corneal myofibroblast differentiation has important implications for the control of wound healing in patients, particularly after keratoplasty, trauma, and refractive surgery. However, prior studies of corneal myofibroblast differentiation have relied on animal cell culture models because of the scarcity and cost of human donor tissue necessary to conduct in-depth biochemical and molecular investigations. Because recent reports have suggested unique species and environmental differences in the response of keratocytes to TGFβ (Funderburgh JL, Funderburgh ML, Mann MM, ARVO Abstract 4958, 2001), ^{1 18} studies of human corneal cells have become increasingly important, and the identification of an appropriate immortalized human cell line is clearly needed. Whereas many studies have reported on the successful immortalization of corneal epithelial and endothelial cells using various oncogene approaches including the large T antigen of simian virus (SV)40 and E6/E7 of the human papilloma virus, ^{28 29 30 31 32 33 34} only two studies have been reported on the immortalization of corneal fibroblasts, using SV40 infection of keratocytes from rabbit ³⁵ and human. ³⁶ As noted by Hoffschir et al., ³⁶ a problem with the use of the SV40 oncogene is the possibility of chromosomal instability after immortalization. Furthermore, both SV40 and E6/E7 block the retinoblastoma and p53 gene products, interfering with downstream cell differentiation ^{37 38} as well as resulting in loss of contact inhibition and growth control. ³⁴ Finally, 50% of the rabbit keratocyte clones immortalized with SV40 show constitutive expression of α-SMA with no upregulation on addition of TGFβ. ³⁵  

Recent studies indicate that control of prolonged cell growth in mammalian cells is regulated by telomerase, with human adult cells showing no telomerase activity and shortened telomeres. ³⁹ Restoration of hTERT activity in these cells leads to lengthening of telomeres, restoration of growth, and unlimited proliferation. ²¹ Cells infected with hTERT also show normal differentiation and growth control, such as contact inhibition not exhibited by cells immortalized with oncogenes. In the present study, infection of human keratocytes with hTERT generated extended-life cell clones that have continued in culture for more than 144 population doublings without any observable decrease in cell proliferation. Of the 62 clones isolated, it is notable that none have shown constitutive expression of α-SMA, indicating normal control of myofibroblast differentiation, unlike SV40-immortalized rabbit corneal keratocytes. Clones isolated from the older donor have shown chromosomal aberrations, but these aberrations are the result of chromosomal changes arising within the donor tissue. ²⁶  

Unfortunately, hTERT-immortalized cells appear to maintain a spindle-shaped, fibroblast phenotype, different from the in vivo keratocyte phenotype characterized by a dendritic morphology and cortical actin organization. ⁴⁰ Prolonged serum-starvation of hTERT corneal fibroblasts did not restore the keratocyte phenotype, nor did culture in various defined growth factors, including FGF, PDGF, and IGF (data not shown). However, persistence of the corneal fibroblast phenotype after culture in serum is also a characteristic of growth-limited cells, suggesting that the fibroblast phenotype maybe irreversible. Nevertheless, the generation of this immortalized human cell line provides for the first time the opportunity to study human corneal fibroblast responses to wound-healing cytokines. In addition, strategies to induce keratocyte differentiation from fibroblasts can now be systematically evaluated. 

In our studies of normal growth-limited human keratocytes, the induction of myofibroblast differentiation by TGFβ appeared to be age dependent, in that keratocytes isolated from older individuals required a longer exposure time. Although it was not possible to characterize this age dependence more fully, it should be noted that donor age was also associated with the appearance of chromosomal abnormalities, as previously described by Pattenati et al., ²⁶ in whose original work keratocytes from donors older than 18 years showed a much greater likelihood (17/24 donors) of having chromosomally abnormal cells, most commonly involving aneuploidy with gains or losses of sex chromosomes, along with various deletions and translocations in 10% to 91% of cells. Our finding that corneal fibroblasts from a 62-year-old donor showed an abnormal karyotype is consistent with their finding, although the appearance of polyploidy was not reported. Karyotype analysis of normal, untelomerized cultured corneal fibroblasts from an older donor identified a 46, X, +10 karyotype, more characteristic of the abnormalities identified by Pattenati et al. ²⁶ The karyotype obtained from 10-year-old donor cells that had been transfected with hTERT was shown to be normal (46, XY) indicating the hTERT infection was not directly associated with the development of polyploidy in the cell lines from the older donor. It is interesting to speculate on whether these chromosomal aberrations may influence the TGFβ response of human keratocytes and, in part, explain the apparent age-dependent differences in TGFβ noted in culture. In this regard, Pattenati et al. ²⁶ have noted that the most frequent chromosomes exhibiting aneuploidy, deletions, and translocation are those important in the regulation of collagen synthesis, adhesion, and turnover, suggesting that the adult cornea may be deficient in essential keratocyte responses during corneal repair. Furthermore, corneal wound repair in the adult human has been shown to be highly variable, with healing of partial incisional wounds ranging from a typical fibrotic response and complete healing to retention of an epithelial plug with no wound fibrosis. ⁴¹ This variation in the wound-healing responses explains, in part, the highly variable refractive results noted after radial keratotomy—an early form of refractive surgery for which older patients show a slightly more pronounced effect (0.7 D per decade of life) ⁴² that can be linked to the poorer wound-healing response. ⁴¹ Of interest, chromosomal aberrations in the keratocytes are also highly variable and weakly correlated with age. ²⁶  

Although the cause of chromosomal aberration in the human keratocyte is not known, a possible explanation includes environmental factors for which oxidative damage may be important. The keratocyte lacks or minimally expresses many of the protective mechanisms present in the corneal epithelium, including superoxide dismutase and ferritin. ^{43 44} It has further been proposed that the absence of antioxidants explains the heightened sensitivity of corneal keratocytes to oxidative insults, including hydrogen peroxide, sodium perborate, and sodium hypochlorite compared with the relatively insensitive corneal epithelium. ⁴⁵  

Whether the absence of response of bovine keratocytes to TGFβ involves an age- or environment-dependent change in keratocyte chromosomes is not known. However, it should be noted that bovine keratocytes are isolated from older animals that may be exposed to greater environmental insults than rabbit keratocytes isolated from younger rabbits reared indoors. Certainly, further work is necessary to establish more clearly the extent of these chromosomal aberrations, their significance, and their cause. Nevertheless, it is important when evaluating any tissue culture model of corneal fibroblasts to take into consideration that the cell population under study is heterogeneous and has underlying chromosomal defects that may influence any results. The establishment of hTERT-immortalized cell lines that show normal chromosomal patterns such as the cell line reported herein, may be of significant help in more precisely defining the cell biology of the corneal stroma. 

 Supported in part by grants from the Gillette Program Project, Centers for Alternatives to Animal Testing, Baltimore, Maryland; the Eye Bank Association of America; the Lions Eye Bank Foundation; National Eye Institute Grant EY07348; and an unrestricted grant from Research to Prevent Blindness.

 Submitted for publication September 20, 2002; revised November 4 and November 20, 2002; accepted November 23, 2002.

 Disclosure: J.V. Jester, None; J. Huang, None; S. Fisher, None; J. Spiekerman, None; J.H. Chang, None; W.E. Wright, None; J.W. Shay, None

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

 

Figure 1.

 

Normal cultured keratocytes showing dendritic morphology by (A) differential interference contrast microscopy with (B) cortical phalloidin staining, no (C) anti-α-SMA, and (D) minimal anti-vinculin staining. Bar, (A) 100 μm; (B-D) 50 μm.

Figure 2.

 

Cultured keratocytes treated with TGFβ (1 ng/mL) for 5 days showed cell proliferation and spreading (A, differential interference contrast), formation of prominent phalloidin-stained actin filament bundles (B), intense α-SMA staining localized to actin filament bundles (C), and marked anti-vinculin staining of focal adhesions (D). Bar, (A) 100 μm; (B-D) 50 μm.

Figure 3.

 

Cultured keratocytes from the same donor (left and right eyes combined) treated for 10 days with TGFβ (1 ng/mL) alone (A, B) or in combination with GRGDdSP (100 μM; C, D) or RAD (100 μM; E, F). Cells were stained with phalloidin (A, C, E) or anti-α-SMA (B, D, F). Bar, 50 μm.

Figure 4.

 

Karyotype analysis of the corneal fibroblasts from a 62-year-old male donor (A) and a 10-year-old male donor (B). The karyotype of the 62-year-old was 89(4n), XX, −4 whereas the karyotype of the 10-year-old was 46, XY.

Figure 5.

 

HTKs cells grown under serum-starved conditions alone (A-D) or treated with 1 ng/mL TGFβ (E-H) and observed by differential interference contrast microscopy (A, E) or stained with phalloidin (B, F), anti-vinculin antibodies (C, G), or anti-α-SMA antibodies (D, H). Bar: (A, E) 100 μm; (B-D, F-H) 50 μm.

Figure 6.

 

Effect of 100 μM GRGDdSP (A, C, E, G) and 150 μM RAD (B, D, F, H) on TGFβ-induced cell spreading (A, B), phalloidin staining (C, D), anti-vinculin staining (E, F), and anti-α-SMA staining (G, H) of HTKs cells. Bar: (A, B) 100 μm; 50 μm (C-H).

Figure 7.

 

Western blot (top) and densitometry graph (bottom) identifying expression of α-SMA extracted from telomerized corneal fibroblast treated with 1 ng/mL TGFβ (+) in combination with GRGDdSP (0–100 μM) or RAD (150 μM). Representative of two separate experiments.

Figure 8.

 

Real-time PCR of mRNA extracted from HTK cells treated with TGFβ (1 ng/mL) alone or in combination with GRGDdSP (100 μM) or RAD (150 μM) and probed with primers specific for α-SMA or β-actin. Data represent the C_T based on log quantity in nanograms compared with purified PCR products for α-SMA and β-actin.

Figure 9.

 

Effect of neutralizing antibody to PDGF on the TGFβ-mediated myofibroblast differentiation of HTK cells. Cultures were stained with anti-α-SMA (A, C, E) or anti-vinculin (B, D, F) after treatment with (A, B) 25 μg/mL anti-PDGF and 1 ng/mL TGFβ, (C, D) 100 ng/mL PDGF in combination with anti-PDGF and TGFβ, or (E, F) 100 ng/mL PDGF alone. Bar, 50 μm.