|Home | About | Journals | Submit | Contact Us | Français|
A critical component of corneal scarring is the TGFβ-induced differentiation of corneal keratocytes into myofibroblasts. Inhibitors of this differentiation are potentially therapeutic for corneal scarring. In this study, we tested the relative effectiveness and mechanisms of action of two electrophilic peroxisome proliferator-activated receptor gamma (PPARγ) ligands: cyano-3,12-dioxolean-1,9-dien-28-oic acid-metheyl ester (CDDO-Me) and 15-deoxy-Δ-12,14-prostaglandin J2 (15d-PGJ2) for inhibiting TGFβ-induced myofibroblast differentiation in vitro. TGFβ was used to induce myofibroblast differentiation in cultured, primary human corneal fibroblasts. CDDO-Me and 15d-PGJ2 were added to cultures to test their ability to inhibit this process. Myofibroblast differentiation was assessed by measuring the expression of myofibroblast-specific proteins (αSMA, collagen I, and fibronectin) and mRNA (αSMA and collagen III). The role of PPARγ in the inhibition of myofibroblast differentiation by these agents was tested in genetically and pharmacologically manipulated cells. Finally, we assayed the importance of electrophilicity in the actions of these agents on TGFβ-induced αSMA expression via Western blotting and immunofluorescence. Both electrophilic PPARγ ligands (CDDO-Me and 15d-PGJ2) potently inhibited TGFβ-induced myofibroblast differentiation, but PPARγ was only partially required for inhibition of myofibroblast differentiation by either agent. Electrophilic PPARγ ligands were able to inhibit myofibroblast differentiation more potently than non-electrophilic PPARγ ligands, suggesting an important role of electrophilicity in this process. CDDO-Me and 15d-PGJ2 are strong inhibitors of TGFβ-induced corneal fibroblast to myofibroblast differentiation in vitro, suggesting this class of agents as potential novel therapies for corneal scarring warranting further study in pre-clinical animal models.
Corneal scarring, the second most common cause of blindness in the world, can be due to ocular infections (Whitcher, et al., 2001), trauma (Farjo, et al., 2008), and surgeries (Marchini, et al., 2006; Saini, et al., 2003; Wilson, et al., 2001). Safe and efficacious medical therapies for the treatment of corneal scarring are limited.
Keratocytes are transparent cells of the corneal stroma that are involved in corneal wound healing and scarring (Fini and Stramer, 2005; Fini, 1999; Jalbert, et al., 2003; Muller, et al., 1995). Corneal wounds result in the release of several cytokines and chemokines that drive keratocyte activation, migration, and differentiation into fibroblasts and myofibroblasts (West-Mays and Dwivedi, 2006). Corneal myofibroblasts provide wound contraction and produce extracellular matrix molecules (ECMs), such as Type I and III collagen (Funderburgh, et al., 2001) and fibronectin (FN) (Welch, et al., 1990), which are important in regenerative wound repair (Jester, et al., 1995; West-Mays and Dwivedi, 2006). However, myofibroblasts also contribute to corneal haze and corneal shape change through decreased crystallin production (Jester, et al., 1999; 2008) and excessive and disordered production of collagen and other extracellular matrix molecules (Marchini, et al., 2006; Netto, et al., 2006; Pal-Ghosh, et al., 2004; Saini, et al., 2003; Sosne, et al., 2002; Wilson, et al., 2001). Both haze and changes in corneal shape decrease visual acuity (Marchini, et al., 2006; Netto, et al., 2006; Pal-Ghosh, et al., 2004; Saini, et al., 2003; Sosne, et al., 2002; West-Mays and Dwivedi, 2006; Wilson, et al., 2001).
Differentiation of keratocytes to myofibroblasts in vitro and in vivo is driven primarily by transforming growth factor beta (TGFβ) (Jester, et al., 1996), a cytokine released by corneal epithelial cells, corneal fibroblasts, the lacrimal gland, and conjunctival cells under inflammatory conditions (Buehren, et al., 2008; Wilson, et al., 1992). Netto and colleagues (2006) demonstrated that by damaging the corneal epithelial basement membrane, penetrating keratectomy in mice causes the release of TGFβ into the corneal stroma, stimulating keratocyte to myofibroblast differentiation. While factors other than TGFβ are involved in the corneal scarring phenomena, including platelet-derived growth factor (PDGF) (Kaur, et al., 2009a; 2009b) and integrin signaling (Jester, et al., 2002), pharmaceutical inhibitors of TGFβ have been shown to decrease myofibroblast differentiation and corneal opacification in several animal models of corneal scarring (Buehren, et al., 2008; Moller-Pedersen, et al., 1998; Jester, et al., 1997).
PPARγ ligands have anti-fibrotic properties and have been studied as agents capable of inhibiting TGFβ-induced myofibroblast differentiation in different tissues (Ferguson, et al., 2009; Galli, et al., 2002; Kawai, et al., 2009), including the cornea (Pan, et al., 2009; Pan, et al., 2010). PPARγ gene transfer decreased corneal opacification in an alkali burn mouse model of corneal scarring (Saika, et al., 2007). A non-electrophilic PPARγ ligand, pioglitazone, was found to inhibit TGFβ-induced keratocyte to myofibroblast differentiation in vitro (Pan, et al., 2010). However, a recent in vitro study found two electrophilic peroxisome proliferator-activated receptor gamma (PPARγ) ligands, cyano-3,12-dioxolean-1,9-dien-28-oic acid (CDDO) and 15-deoxy-Δ-12,14-prostaglandin J2 (15d-PGJ2), to inhibit TGFβ-induced lung myofibroblast differentiation more potently than non-electrophilic PPARγ ligands (Ferguson, et al., 2009). 15d-PGJ2 (Kliewer, et al., 1995), CDDO, and CDDO derivatives bind PPARγ with high affinity (Wang, et al., 2000). Once a PPARγ ligand binds to PPARγ, the latter forms a heterodimer with the retinoid X receptor (RXR) and its ligand (Ferguson, et al., 2009; Liby, et al., 2007; Rizzo and Fiorucci, 2006). The heterodimer then translocates to the nucleus and interacts with PPAR response elements (PPRE), leading to PPARγ-induced gene transcription (Ferguson, et al., 2009; Liby, et al., 2007; Rizzo and Fiorucci, 2006). Electrophilic PPARγ ligands have α/β-unsaturated ketone rings with electrophilic carbons (Brookes, et al., 2007; Ferguson, et al., 2009; Shi and Greaney, 2005). These electrophilic carbons are highly susceptible to Michael addition reactions (Shi and Greaney, 2005), which enables them to form covalent bonds with intracellular nucleophiles, such as thiol groups or cysteine residues (Brookes, et al., 2007 and Chintharlapalli, et al., 2005; Ferguson, et al., 2009; Ray, et al., 2006; Straus, et al., 2000; Suh, et al., 1999). For this reason, electrophilic PPARγ ligands, such as CDDO (Brookes, et al., 2007; Chintharlapalli, et al., 2005; Ferguson, et al., 2009; Suh, et al., 1999) and 15d-PGJ2 (Ray, et al., 2006; Straus, et al., 2000), can also exert PPARγ-independent effects.
The present experiments examine the ability of two electrophilic PPARγ ligands, a CDDO derivative, CDDO-methyl ester (−Me), and 15d-PGJ2 to inhibit differentiation of corneal fibroblasts to myofibroblasts in vitro. The anti-scarring properties of these agents have been studied in fibroblasts in the lung (Ferguson et al. 2009). However, due to the heterogeneity of fibroblasts between tissues (Smith, et al., 1995), this study is an essential first step towards assessing the potential effectiveness of these molecules for treating corneal scarring. CDDO in particular has already been used systemically in Phase I and II clinical trials as a chemotherapeutic agent (Dezube, et al., 2007). Since it was shown to have a favorable safety profile when administered systemically in humans (Dezube, et al., 2007), this significantly increases its translational potential for a topical application such as eye drops in the treatment of corneal scarring.
Human corneal fibroblast cell strains were derived from anatomically normal donor rim corneal tissue donated to the Rochester/Finger Lakes Eye and Tissue Bank. Human tissue was obtained and handled in accordance with the tenets of the Declaration of Helsinki. The cells derived from these explants were isolated and then cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco BRL, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Sigma Aldrich, St. Louis, MO), gentamicin (0.05 mg/ml; Invitrogen, Carlsbad, CA), amphotericin (.25 mg/ml; Invitrogen), and 25 mM HEPES at pH 7.4 at 37°C in 7% CO2. The cells were morphologically consistent with fibroblasts, expressing vimentin, but not CD45, factor VIII, or cytokeratin. Cells from three different patient strains were used at passages 4–10. The PPARγ ligands, 15d-PGJ2 (Biomol, Plymouth Meeting, PA), 9,10-dihydro-15-deoxy-D12,14-PGJ2 (CAY10410; Cayman Pharmaceuticals, Ann Arbor, MI), rosiglitazone (Cayman Pharmaceuticals), CDDO-Me (a kind gift from Dr. Michael Sporn, Dartmouth University), were prepared as 10 mM stocks in dimethyl sulfoxide (DMSO) and added to cell cultures to the final concentration indicated. GW-9662 (Cayman Pharmaceuticals), an irreversible, small-molecule PPARγ antagonist, was prepared in the same manner. This molecule has been used by us (Burgess, et al., 2005; Feldon, et al., 2006; Ferguson, et al., 2009) and others (Fu, et al., 2001; Mix, et al., 2009) and been shown to block PPARγ activity. Recombinant human TGFβ1 (#240B) was purchased from R&D systems (Minneapolis, MN).
The calcein AM/EthD-I Live/Dead Viability/Cytotoxicity Kit (Molecular Probes, Eugene, OR) was used to assess cell viability. Calcein AM specifically stains live cells via their intracellular esterase activity, and EthD-I specifically stains dead cells that have lost plasma membrane integrity. Corneal fibroblasts were plated in 25 ml flasks (Greiner Bio-One) in 10% FBS media and were either untreated, treated with TGFβ alone, or with PPARγ ligands with or without TGFβ for 72 hours. PPARγ ligands were added 30 minutes prior to TGFβ. Prior to harvesting the cells, one of corneal fibroblast samples were killed using 0.1% saponin treatment for 15 minutes. At harvest, cells were trypsinized and washed with PBS twice and then stained with 1 μM calcein AM and 8 μM EthD-1 for 10 minutes. Samples were analyzed on a FACSCanto II (BD Biosciences). The live cell gate was set based on cells staining with calcein AM in the untreated corneal fibroblast culture. The dead cell gate was set based on cells staining with EthD-1 in the 0.1% saponin treated culture. Treated samples were analyzed using these gates to assess percent of live and dead cells.
Primary human corneal fibroblasts were plated in 60 or 100 mm dishes (Falcon/Becton Dickson, Franklin Lakes, NJ) or 6 well plates (Falcon/Becton) in 10% FBS media and were either untreated, treated with TGFβ alone, or treated with PPARγ ligands with or without TGFβ for 72 hours. PPARγ ligands were added 30 minutes prior to TGFβ. For certain experiments, the fibroblasts were infected with an adenovirus vector expressing a dominant negative PPARγ or an empty vector for 24 hours before treatment with TGFβ and PPARγ ligands. In other experiments, the fibroblasts were treated with GW9662 for 4 hours before treatment with TGFβ and PPARγ ligands. Cell lysates were prepared using either a 1× SDS lysis buffer or with the commercial ActiveMotif Nuclear Extract kit (Carlsbad, CA). Proteins were separated by 10% SDS-PAGE under reducing conditions, transferred to a polyvinylidene fluoride (PVDF) membrane and examined for expression of αSMA (1:4000 dilution, Sigma). Precast 4–15% SDS-PAGE gradient gels (Bio-Rad, Hercules, CA) were used to examine expression of panfibronectin (1:1000 dilution, Sigma), and collagen I (1: 1000 dilution, Santa Cruz, Carlsbad, CA). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (1:3000 dilution, Abcam, Cambridge, MA) was used as a loading control. Western Lightning Chemiluminescence Reagent Plus (Perkin-Elmer, Wellesley, MA) was used to visualize protein bands. Densitometry of the resulting bands was performed using QuantityOne Software (Biorad) and normalized to GAPDH loading control.
Primary corneal fibroblasts were cultured at a density of 1,500 cells per well on eight-well chamber slides (Nalge Nunc International, Napersville, IL) in in 10% FBS media and either untreated, treated with TGFβ alone, or treated with PPARγ ligands with or without TGFβ for 72 hours. Slides were fixed with 2% paraformaldehyde for 15 minutes and sequentially treated with mouse monoclonal αSMA antibody (1:800 dilution, Sigma) and anti-mouse-Alexafluor 555 (Sigma). Mounting media containing 4,6-diamidino-2-phenylindole (DAPI; Sigma) was used to mount the slides and stain nuclei. Slides were viewed using a Zeiss Axioplan microscope (Zeiss, Thornwood, NY).
Primary human corneal fibroblasts were plated in 25ml flasks (Greiner Bio-One, Frickenhausen, Germany) in 10% FBS media and treated as described above. The cells were harvested, stained for αSMA (Sigma) and Draq5 (Axxora, San Diego, CA), and analyzed using Imagestream System 100 (Amnis Corporation, Seattle, WA) as described in Ferguson et al. (2009). Briefly, both the intensity and aspect ratio of the nuclear stain Draq5 were used in a gating strategy to isolate true single cells from clumps and debris. Subsequently, objects identified as single cells were analyzed for αSMA expression, which was defined by both the intensity of the αSMA signal and the fractional area of the cell covered by αSMA signal. A gate was created encompassing 1% of cells in the untreated sample with high αSMA signal intensity and area, and this was identified as the myofibroblast region in order to compare relative myofibroblast differentiation across samples.
Primary human corneal fibroblasts were plated in 100 ml dishes (Falcon/Becton Dickson, Franklin Lakes, NJ) in 10% FBS media and either untreated, treated with TGFβ alone, or treated with PPARγ ligands with or without TGFβ for 24 hours. RNA was isolated using RNeasy (Qiagen, Valencia, CA), following the manufacturer's protocol. RNA (1.0 μg) was incubated with PCR buffer, 0.5 μg of oligo (dT)12–18 primer (Invitrogen), and 10 mM deoxy-nucleotide-triphosphate (dNTP) for 10 minutes at 70°C and 5 minutes in ice water. Subsequently, 40 U of recombinant RNasin RNase inhibitor (Promega, Madison, WI), 0.1 mM DTT, and 200 U of Superscript III reverse transcriptase (RT; Invitrogen) were added to the mixture and incubated for 5 minutes at room temperature, 60 minutes at 50°C, and 15 minutes at 70°C. The reaction contents were diluted to 80-μl volume and stored at −20°C. Negative controls did not contain RT enzyme.
Quantitative real-time RT-PCR reactions were performed using a Bio-Rad iCycler with SYBR Green Supermix (Bio-Rad) according to the manufacturer's recommended protocol, which was modified such that the reactions contained 3 mM MgCl2 and 0.2 μM of each primer. Oligomer primers were ordered from Integrated DNA Technologies (Coralville, IA) (COL3A1, ACTA2, and GAPDH). The primer sequences were as follows: GAPDH sense, 5'- AAG GTC GGA GTC AAC GGA TTT GGT-3'; GAPDH anti-sense, 5'- AGC CTT GAC GGT GCC ATG GAA TTT-3'; ACTA2 sense, 5'- TCT GGA GAT GGT GTC ACC CAC AAT-3'; ACTA2 anti-sense, 5'-AAT AGC CAC GCT CAG TCA GG-3'; COL3A1 sense, 5'-AGC AA AGT TTC CTC CGA GGC CAG-3'; COL3A1 antisense, 5'-GGA GAA TGT TGT GCA GTT TGC CCA-3'.
Primary human corneal fibroblasts were plated in 6 well plates (Falcon/Becton) in 10% FBS media and infected with DN PPARγ (a kind gift from Drs. E. J. Lee and J. L. Jameson (Northwestern University)) or control adenovirus, AdDL70-3, at a multiplicity of infection (MOI) of 30 plaque forming units (pfu)/cell for 24 hours. The infected cells were then either untreated, treated with TGFβ alone, or treated with CDDO-Me with or without TGFβ. DN PPARγ adenovirus vector construction and production has been previously described by us and others (Burgess, et al., 2005; Park, et al., 2003). The DN PPARγ is able to bind the ligand; however, it does not activate transcription at the PPRE (Burgess, et al., 2005; Park, et al., 2003). Control adenovirus construction and production has been previously described (Bett, et al., 1994; Sime, et al., 1997). The control adenovirus has no inserted transgene.
Primary human corneal fibroblasts were plated in 75 cm2 flasks (Falcon/Becton) in 10% FBS media and infected with DN PPARγ or control adenovirus as described above. After a 24 hour infection, cells were harvested and PPRE × 3-firefly luciferase (Ray, et al., 2006) and SV40-renilla luciferase (Promega) plasmid DNAs were introduced using an Amaxa nucleofector device (Lonza AG) following the manufacturer's protocol. Nucleofected cells were plated into 6 well plates and allowed to grow for 6–8 hrs. Either DMSO (vehicle) or 5 β-M 15d-PGJ2 was then added to the cultures for an additional 14–18 hrs. Following incubation, cells were washed two times in 1XPBS and lysed directly in plates using the Dual-Glo luciferase assay buffer (Promega). Firefly and renilla luciferase readings were read on a Varioskan Flash luminometer (Thermo Fisher) following manufacturer's instructions. Luciferase readings were normalized to the control adenovirus, vehicle treated samples for statistical analysis.
Collagen gels were prepared from lyophilized rat tail collagen (Roche Diagnostics GmbH, Mannhein, Germany) as previously described (Mio, et al., 1996; Skold et al., 1999). Collagen was dissolved in 0.2% acetic acid (v/v) and mixed with sterile 2× MEM and 0.1M NaOH. 1× DMEM media with or without cells was added to this mixture, and 667 μl of the collagen mixture with or without 15,000 corneal fibroblasts per well was added to BSA coated 24 well plates (Falcon). The plates were incubated at 37°C for 2 hours to polymerize, forming a 3-dimensional collagen gel. The gels were separated from the wells using a 200 μl pipette tip. The gels were then either untreated, treated with TGFβ alone, or treated with PPARγ ligands with or without TGFβ in 10% FBS media. The gels were imaged and weighed after 48 hours of treatment. Since measuring the two-dimensional surface area of the gels fails to encompass the three-dimensional changes occurring in each gel, quantitative analysis of this assay was accomplished using gel weight. Based on the weight of the gels, the percent contraction of the gels with cells, compared to the gel without cells, was calculated as previously described by Lehmann and colleagues (2011).
One-way ANOVA with post-hoc Tukey HSD test was used for statistical analysis. Statistical analysis was performed using GraphPad Prism 5 (GraphPad Software, LaJolla, CA) software. A p value less than 0.05 was considered statistically significant.
To study the ability of different electrophilic PPARγ ligands to inhibit TGFβ-induced corneal myofibroblast differentiation, human corneal fibroblasts were treated with 5 ng/ml TGFβ after 30 mins pretreatment with varying doses of the electrophilic PPARγ ligands, CDDO-Me and 15d-PGJ2,. The maximum concentrations used (0.350 μM CDDO-Me and 10 μM 15d-PGJ2) were chosen for having the strongest inhibitory effect on TGFβ-stimulated αSMA up-regulation without effect on cell viability (data not shown). Although there were slight differences in sensitivity to TGFβ and effectiveness of the PPARγ ligands at reducing TGFβ-induced αSMA expression between the three human corneal fibroblast strains, pre-treatment with CDDO-Me and 15d-PGJ2 inhibited TGFβ-induced αSMA expression in all three strains (Figure 1A and B). Subsequent experiments were completed in the strain of corneal fibroblasts with the greatest sensitivity to TGFβ (strain 2). CDDO-Me and 15d-PGJ2 were also able to significantly inhibit TGFβ-induced αSMA mRNA expression (Figure 1C). CDDO-Me and 15d-PGJ2 treatment alone led to a non-significant decrease of αSMA mRNA compared to untreated cells (Figure 1C).
Further evidence about the ability of electrophilic PPARγ ligands to inhibit TGFβ-induced myofibroblast differentiation came from imaging flow cytometry, a technique that allows thousands of labeled cells to be imaged while measuring their fluorescent emissions (Figure 2). TGFβ alone increased the percentage of cells within the myofibroblast region to 29% compared to 1% in the untreated sample (Figure 2B). Pretreatment with CDDO-Me and 15d-PGJ2 reduced the percent of cells within the myofibroblast region from 29% to 9% and 19%, respectively.
In addition to reducing TGFβ-induced αSMA expression, CDDO-Me and 15d-PGJ2 also inhibited TGFβ-induced collagen I and fibronectin expression in cultured human corneal fibroblasts (Figure 3A). Furthermore, TGFβ-induced collagen III mRNA expression was inhibited by these electrophilic PPARγ ligands (Figure 3B). CDDO-Me and 15d-PGJ2 treatment alone led to a non-significant decrease of collagen III mRNA compared to untreated cells (Figure 3B).
A collagen gel contraction assay was used to assess whether electrophilic PPARγ ligands were effective at not only decreasing TGFβ-induced production of extracellular matrix molecules, but also at decreasing TGFβ-induced contraction of collagen gels by corneal myofibroblasts. Due to the increased contractile properties of myofibroblasts relative to fibroblasts, collagen gels with myofibroblasts are expected to be smaller in size and weigh less (Kurosaka, et al., 1998; Lehmann, et al., 2011; Skold, et al., 1999; Tingstrom, et al., 1992; Xi-Wen, et al., 2007). The percent contraction of the gels with cells was calculated based on the weight of the gels with cells compared to the gel without cells. As predicted, collagen gels containing corneal fibroblasts exhibited greatly reduced surface area (Figure 4A) and weight (Figure 4B) following treatment with TGFβ. CDDO-Me (p<0.001) and 15d-PGJ2 (p<0.05) treatment attenuated the collagen gel surface area contraction and weight reduction induced by TGFβ. As expected, collagen gels without cells did not change in size over time (Figure 4A).
We used both a pharmacological and a genetic approach to study the role of PPARγ in inhibition of TGFβ-induced myofibroblast differentiation by electrophilic PPARγ ligands. Pharmacologic inhibition of PPARγ was accomplished by using GW9662, a highly specific, irreversible PPARγ antagonist that renders PPARγ unable to bind with other ligands by forming a covalent bond with a cysteine residue within the PPARγ binding pocket (Feldon, et al., 2006, Guo, et al., 2011). Corneal fibroblasts were pre-treated with GW9662 prior to treatment with TGFβ and/or CDDO-Me or 15d-PGJ2. Western blotting was used to measure levels of αSMA after 72 hours. GW9662 treatment did not prevent CDDO-Me or 15d-PGJ2 from inhibiting TGFβ-induced αSMA expression (Figure 5A).
Genetic inhibition of PPARγ was accomplished by infection of corneal fibroblasts with replication-deficient adenovirus expressing a mutated dominant negative (DN) PPARγ gene. Corneal fibroblasts were treated with the DN PPARγ or empty adenovirus (AdDL70-3) for 24 hours and then treated with TGFβ or TGFβ and CDDO-Me. 72 hours after treatment, Western blotting for αSMA demonstrated that CDDO-Me inhibited TGFβ-induced αSMA expression in both the control virus infected fibroblasts and in the DN PPARγ infected fibroblasts (Figure 5B and C). The ability of the DN PPARγ adenovirus to block PPARγ-dependent transcriptional activity in corneal fibroblasts was demonstrated using a PPARγ reporter luciferace assay (Supplementary Figure 1). Corneal fibroblasts infected with AdDL70-3 and a PPARγ ligand, displayed a two-fold increase in PPARγ-dependent transcriptional activity as demonstrated by luciferase reporter assays. Corneal fibroblasts infected with the DN PPARγ adenovirus displayed almost no PPARγ dependent transcriptional activity in the presence or absence of a PPARγ ligand.
The ability of CDDO-Me to inhibit TGFβ-induced αSMA expression in the absence of PPARγ activity suggests that the inhibitory effects of this molecule on myofibroblast differentiation are largely PPARγ independent. However, an interesting observation was that corneal fibroblasts treated with GW9662 had a 1.43-fold increased TGFβ-induced αSMA compared to non-GW9662 treated cells. Similarly, corneal fibroblasts infected with the DN-PPARγ virus had a 1.36-fold increased TGFβ-induced αSMA compared to control virus infected cells. However, these differences were not statistically significant.
An important question in the context of the present studies is whether electrophilicity of CDDO-Me and 15d-PGJ2 was a critical factor in their ability to inhibit the differentiation of corneal fibroblasts into myofibroblasts. To answer this question, we contrasted the effects of these two electrophilic PPARγ ligands with two non-electrophilic PPARγ ligands (rosiglitazone and CAY10410), on TGFβ-induced αSMA expression. CAY10410 is a non-electrophilic structural analog of 15d-PGJ2. Western blotting and immunohistochemistry showed that all of the PPARγ ligands were capable of inhibiting TGFβ-induced αSMA expression (Figure 6A and B). However, the natural electrophilic PPARγ ligand 15d-PGJ2 inhibited αSMA expression much more potently than both non-electrophilic PPARγ ligands (Figure 6A). The synthetic electrophilic PPARγ ligand CDDO-Me inhibited αSMA expression even more potently than 15d-PGJ2, with 0.175 μM CDDO-Me and 2.5 μM 15d-PGJ2 reducing αSMA expression to about the same extent (Figure 6A). Thus, the electrophilic property of PPARγ ligands appears to be important for their inhibition of TGFβ-induced αSMA expression in human corneal fibroblasts.
PPARγ and its ligands have been previously studied in the context of corneal wound healing and have been shown to have anti-inflammatory and anti-scarring properties (Pan, et al., 2009; 2010; Saika, et al., 2007). Electrophilic PPARγ ligands, such as CDDO and 15d-PGJ2 have been shown to inhibit TGFβ-induced myofibroblast differentiation in lung fibroblasts more potently than non-electrophilic PPARγ ligands (Ferguson, et al., 2009). Here, we report for the first time, that the CDDO derivative, CDDO-Me, and 15d-PGJ2 can also function as potent anti-fibrotic agents for cultured human corneal fibroblasts stimulated to differentiate into myofibroblasts by TGFβ. In fact, CDDO-Me was able to inhibit TGFβ-induced myofibroblast differentiation to the same extent as 15d-PGJ2, but at a 14-fold lower concentration.
In order to elucidate the mechanism through which CDDO-Me and 15d-PGJ2 inhibit TGFβ-induced myofibroblast differentiation, we first investigated the PPARγ-dependence of their anti-fibrotic properties. Using pharmacologic and genetic methods to block PPARγ, we demonstrated that the anti-fibrotic properties of CDDO-Me and 15d-PGJ2 are largely PPARγ-independent. Ferguson and colleagues (2009) demonstrated similar findings with CDDO (rather than CDDO-Me) and 15d-PGJ2 in human lung fibroblasts and hypothesized that both PPARγ-dependent and PPARγ-independent pathways (Bishop-Bailey and Wray, 2003; Rizzo and Fiorucci, 2006) may generally modulate myofibroblast differentiation in that tissue. Both CDDOMe and CDDO bind to PPARγ with high affinity and activate PPRE (Chintharlapalli, et al., 2005).
CDDO-Me has two electrophilic α/β-unsaturated ketones, while 15d-PGJ2 has one. The electrophilic α/β-unsaturated ketone is susceptible to Michael addition reactions (Shi and Greaney, 2005) with intracellular nucleophiles, such as glutathione and cysteine residues (Brookes, et al., 2007). Non-electrophilic PPARγ ligands such as rosiglitazone and CAY10410 (the non-electrophilic structural analog of 15d-PGJ2), were much less potent inhibitors of TGFβ-induced myofibroblast differentiation than 15d-PGJ2 and CDDO-Me. This suggests that the ability of CDDO-Me and 15d-PGJ2 to inhibit TGFβ-induced myofibroblast differentiation is related to the electrophilic nature of these two compounds. Furthermore, CDDO-Me was a more potent inhibitor of TGFβ-induced proteins at a much lower concentration than 15d-PGJ2, which may be attributable to the presence of an additional α/β-unsaturated ketone. Other studies have also demonstrated that CDDO and its derivatives are effective inhibitors of TGFβ-induced proteins at much lower concentrations than 15d-PGJ2 (Ferguson, et al., 2009; Kulkarni, et al., 2011). Reactivity with intracellular nucleophiles may be a potential mechanism through which these compounds are able to inhibit myofibroblast differentiation. Indeed, data show other electrophilic agents, such as PGA2 and DSPS to also inhibit TGFβ-induced lung myofibroblast differentiation (Ferguson, et al., 2009).
It should be noted here that we did not examine the ability of 15d-PGJ2 and CDDO-Me to inhibit TGFβ-induced myofibroblast differentiation using keratocytes grown in serum-free media. Serum free conditions enable maintenance of the keratocyte phenotype in vitro, while culturing keratocytes in serum results in their activation, entry into the cell cycle, and phenotypic changes (spindle shape, actin stress fiber assembly, etc.) that are consistent with those seen in activated, repair fibroblasts in a wounded cornea.(Fini and Stramer, 2005; Jester, et al., 1996) Our decision to culture cells in serum was based partly on the fact that CDDO-Me requires a significant amount of protein in the media in order to remain soluble and partly because we wanted to test the agents of interest in a setting that mimicked the wounded cornea in vitro, rather than a non-inflammatory setting, generated by using serum free media.
Finally, a limitation of our study is related to the capability of bone-marrow derived cells to differentiate into αSMA expressing cells (Barbosa, et al., 2010). Since we used corneal fibroblast cultures, we cannot draw conclusions on the capability of CDDO-Me or 15d-PGJ2 to inhibit TGFβ-induced differentiation of bone-marrow derived cells to myofibroblasts. Our ongoing experiments examining the ability of electrophilic PPARγ ligands to inhibit corneal scarring in vivo are designed to directly answer this question.
In summary, we have shown that CDDO-Me and 15d-PGJ2 are potent inhibitors of TGFβ-induced corneal myofibroblast differentiation by examining their ability to decrease levels of multiple myofibroblast proteins and mRNAs. We show that the ability of these PPARγ ligands to inhibit myofibroblast differentiation is largely independent of PPARγ, but dependent on electrophilicity. Further studies are needed to determine the exact mechanisms of action of CDDO-Me and 15d-PGJ2 in corneal fibroblasts. For example, these electrophilic PPARγ ligands may alter the function of intracellular mediators of TGFβ signaling. Due to its established favorable safety profile in human trials, CDDO-Me appears promising as a proximate therapeutic agent for corneal scarring. However, preclinical animal models of corneal scarring should be utilized to assess its efficacy, as well as that of 15d-PGJ2 and other electrophilic agents.
DN PPARγ adenovirus infection of corneal fibroblasts inhibits PPARγ transcriptional activity. Primary human corneal fibroblasts were infected with DN PPARγ or control adenovirus (AdDL70-3) for 24 hours. The cells were then treated with DMSO or the PPARγ ligand, 15d-PGJ2, for 14–18 hours. PPARγ transcriptional activity was measured using a luciferase assay. Luciferase readings were normalized to the control adenovirus, vehicle treated samples (relative activity = 1).
The authors thank Dr. Nancy Chin (Department of Preventative and Community Medicine, University of Rochester School of Medicine and Dentistry) for her contributions.
Financial Support: TL1 RR024135, EY017123, EY015836, K23EY019353, Research to Prevent Blindness Unrestricted Grant, a research grant from the Rochester/Finger Lakes Eye & Tissue Bank, HL095402, and HL075432
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest: No author has a conflict of interest or proprietary interest in the contents of this manuscript.