SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

In fibroblasts, transforming growth factor β (TGFβ) stimulates collagen synthesis and myofibroblast transdifferentiation through the Smad intracellular signal transduction pathway. TGFβ-mediated fibroblast activation is the hallmark of scleroderma and related fibrotic conditions, and disrupting the intracellular TGFβ/Smad signaling may provide a novel approach to controlling fibrosis. Because of its potential role in modulating inflammatory and fibrotic responses, we examined the expression of the nuclear hormone receptor peroxisome proliferator–activated receptor γ (PPARγ) in normal skin fibroblasts and its effect on TGFβ-induced cellular responses.

Methods

The expression and activity of PPARγ in normal dermal fibroblasts were examined by Northern and Western blot analyses, immunocytochemistry, flow cytometry, and transient transfections with reporter constructs. The same approaches were used to evaluate the effects of PPARγ activation by naturally occurring and synthetic ligands on collagen synthesis and α-smooth muscle actin (α-SMA) expression. Modulation of Smad-mediated transcriptional responses was examined by transient transfection assays using wild-type and dominant-negative PPARγ expression constructs.

Results

The PPARγ receptor was expressed and fully functional in quiescent normal skin fibroblasts. Whereas ligand activation of cellular PPARγ resulted in modest suppression of basal collagen gene expression, it abrogated TGFβ-induced stimulation in a concentration-dependent manner. This response was mimicked by overexpressing PPARγ in fibroblasts, and was blocked by a selective antagonist of PPARγ signaling or by transfection of fibroblasts with dominant-negative PPARγ constructs. Furthermore, PPARγ ligands abrogated TGFβ-induced expression of α-SMA, a marker of myofibroblasts. Stimulation of Smad-dependent transcriptional responses by TGFβ was suppressed by PPARγ despite the absence of a consensus PPARγ-response element in the targeted promoters. Ligand-induced activation of fibroblast PPARγ had no effect on protein expression of cellular Smad3 or Smad7.

Conclusion

By abrogating of TGFβ-induced stimulation of collagen gene expression, myofibroblast transdifferentiation, and Smad-dependent promoter activity in normal fibroblasts, PPARγ may play a physiologic role in the regulation of the profibrotic response. Furthermore, our results suggest that PPARγ activation by pharmacologic agonists may represent a novel approach to the control of fibrosis in scleroderma.

Abnormal synthesis and tissue accumulation of collagen are hallmarks of scleroderma and are responsible for the damage and failure of affected organs. Lesional scleroderma fibroblasts display an activated phenotype characterized by accelerated transcription of genes coding for collagen and other extracellular matrix proteins, increased expression of cell surface receptors for transforming growth factor β (TGFβ), and sustained production of TGFβ, connective tissue growth factor, interleukin-1α, and other profibrotic cytokines and growth factors (for review, see ref. 1). Furthermore, lesional fibroblasts show increased expression of the myofibroblast marker α-smooth muscle actin (α-SMA) and resistance to apoptosis (2). Although the nature of the stimulus that triggers fibroblast activation in scleroderma remains unknown, the close topographic association between activated tissue fibroblasts and infiltrating inflammatory cells suggests that mononuclear cell–derived signals may be responsible. Because TGFβ is consistently detected in lesional tissue and is known to be a potent inducer of extracellular matrix synthesis, growth factor production, and fibroblast proliferation, chemotaxis, and terminal differentiation, it is widely regarded as the pivotal mediator in pathologic fibrosis. It remains unclear, however, whether an excess of TGFβ or an exaggerated intensity or duration of the target cell response to TGFβ is primarily responsible for the activated scleroderma fibroblast phenotype.

Recent studies have elucidated the molecular mechanisms underlying the fibrotic responses elicited by TGFβ (3). Receptor-activated Smads (R-Smads) are directly activated by TGFβ/activin (Smad2 and Smad3) or by bone morphogenetic proteins (Smads 1, 5, and 8). Upon ligand binding to the surface TGFβ receptors, R-Smads are phosphorylated and then heterodimerize with Smad4 and translocate from the cytoplasm into nucleus. Once inside the nucleus, the Smad complex recognizes specific DNA sequences in TGFβ-regulated target genes, stimulating or repressing their transcription (3, 4). In contrast to R-Smads, Smad7 is an inhibitory member of the Smad family that abrogates TGFβ/Smad signaling.

We have previously demonstrated that in normal dermal fibroblasts, TGFβ-induced stimulation of type I collagen gene (COL1A2) transcription requires cellular Smad3 and is abrogated by Smad7 (5, 6). We also demonstrated that interaction of the DNA-bound Smad complex with transcription coactivators and histone acetyltransferases p300/CREB binding protein is required for maximal TGFβ/Smad3-induced type I collagen synthesis in normal dermal fibroblasts (7, 8). In scleroderma lesional fibroblasts, R-Smads display increased phosphorylation and nuclear accumulation in the absence of exogenous TGFβ, indicating intrinsic activation of the TGFβ/Smad signal transduction pathway (9). Substantial evidence indicates that altered regulation of Smad signaling is also implicated in the pathogenesis of lung, liver, and kidney fibrosis in humans and in experimentally induced fibrosis in animal models (for review, see ref. 10).

Peroxisome proliferator–activated receptors (PPARs) represent a family of nuclear hormone receptors that are expressed at high levels in adipose tissues and were originally identified as key regulators of adipocyte differentiation and insulin sensitivity (11). Three PPAR isoforms (α, β, γ) have been identified and have been shown to be encoded by separate genes (for review, see ref. 12). The PPARs function as ligand-dependent transcription factors that regulate the expression of target genes. Fatty acids, eicosanoids, and prostaglandins, such as 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2), have been proposed to function as naturally occurring ligands for PPARs (12, 13). The thiazolidinedione class of drugs used in diabetes and dyslipidemias have also been shown to activate PPARγ and, thus, are synthetic ligands.

Like other nuclear hormone receptors, PPARs are modular in structure, with N-terminal transcriptional activation domain, C-terminal ligand-binding domain, and the activation function 2 domain required for interaction with coactivators/corepressors (12, 14). The middle region contains the DNA binding domain, consisting of a zinc finger that specifically recognizes conserved DNA sequences called PPAR-response elements (PPREs) that are present in the promoters of PPAR-regulated genes linked to glucose homeostasis, apoptosis, and proliferation (15).

Recent discoveries suggest that signaling through PPARγ influences a wide range of cellular responses that are entirely unrelated to adipogenesis and insulin homeostasis. Macrophages, microglia, chondrocytes, T cells, and synovial fibroblast-like cells all express PPARγ. In these cells, activation of PPARγ is associated with potent antiinflammatory and immunomodulatory effects; these effects are due to suppression of the genes for tumor necrosis factor α, inducible nitric oxide synthase, cyclooxygenase 2, and interleukin-6 (16–20). The immunoregulatory activities of PPARγ involve mechanisms distinct from those that mediate insulin sensitization. Importantly, activation of PPARγ by naturally occurring ligands also appears to have antiinflammatory effects. In the joint, for example, monosodium urate monohydrate crystals have been shown to activate PPARγ on monocytes, presumably via endogenous 15d-PGJ2, and crystal-induced PPARγ activation was implicated as a potential mechanism to explain the spontaneous resolution of acute inflammation associated with gouty arthritis (21). In light of its potent antiinflammatory effects, PPARγ ligands hold substantial promise as novel immunomodulatory and antiinflammatory agents. These immunoregulatory effects appear to be cell-specific, however, since in monocytes, PPARγ activation resulted in induction, rather than suppression, of cyclooxygenase 2 (22).

The potential involvement of PPARγ in physiologic tissue remodeling, wound healing, and organ fibrosis has thus far received only scant attention. Investigation of PPARγ in these processes has focused primarily on the pancreas, liver, and kidney, the target organs in diabetes. It has been shown that through activation of cellular PPARγ, the thiazolidinedione antidiabetic drug troglitazone inhibited collagen synthesis in mesangial cells from diabetic rats (23) and in mesangial cells activated in vitro by glucose or TGFβ (24). Furthermore, naturally occurring or synthetic ligands of PPARγ have been shown to inhibit proliferation (25), myofibroblast transdifferentiation (26), and collagen synthesis (27, 28) in hepatic and pancreatic stellate cells. Long-term troglitazone administration prevented the development of glomerulosclerosis (29) and pancreatic fibrosis (30) in rodent models of diabetes. Together, these findings indicate that activation of PPARγ by naturally occurring ligands or synthetic agonists causes repression of profibrotic responses in vitro and is associated with reduction or prevention of organ fibrosis.

Virtually nothing is known about the expression, function, or mechanism of action of PPARγ in skin fibroblasts or the role of PPARγ in modulating fibrotic responses in the skin. We report here that PPARγ is constitutively expressed in normal skin fibroblasts, is up-regulated by TGFβ, and can be activated by naturally occurring ligands or by the thiazolidinedione class of antidiabetic drugs. Transient overexpression of PPARγ in fibroblasts dramatically enhanced their sensitivity to PPARγ ligands. Troglitazone and 15d-PGJ2 prevented TGFβ-induced stimulation of type I collagen synthesis at the level of transcription and abrogated α-SMA expression. The inhibitory effects of the PPARγ ligands on these profibrotic responses were specific and PPARγ-dependent. Furthermore, PPARγ activation in fibroblasts had no effect on the level of cellular Smad3 or Smad7 expression. These results indicate that PPARγ inhibits TGFβ-induced profibrotic responses in normal fibroblasts. Together, the findings suggest a novel potential role for PPARγ in the control of skin fibrosis.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Fibroblast culture.

Primary cultures of human dermal fibroblasts were established from newborn foreskins obtained from the delivery suite, as described previously (31). Fibroblasts were maintained at 37°C in an atmosphere of 5% CO2 in Eagle's minimum essential medium (EMEM) containing 5 mM glucose supplemented with 10% fetal bovine serum (FBS), 1% vitamins, 1% penicillin/streptomycin, and 2 mML-glutamine (all from BioWhittaker, Walkersville, MD). For all experiments, fibroblasts were studied between passages 3 and 7. Cell viability was determined by trypan blue dye exclusion.

Plasmids.

The 772COL1A2/CAT plasmid contains sequences from −772 to +58 of the human COL1A2 gene linked to the chloramphenicol acetyltransferase (CAT) reporter gene (32). The p(AOx)3-TK-Luc plasmid (obtained from Dr. Christopher Glass, University of California, San Diego) contains 3 copies of the PPRE from the acyl-coenzyme A (acyl-CoA) oxidase gene linked to thymidine kinase and luciferase genes. The expression vector pCMX-mPPARγ contains the full-length PPARγ complementary DNA under the cytomegalovirus (CMV) promoter in the pCMX vector. Dominant-negative PPARγ mutant L466A was generated by polymerase chain reaction–based site-directed mutagenesis (obtained from Dr. J. Larry Jameson, Northwestern University Medical School, Chicago, IL) (33). Dominant-negative PPARγ (L468A/E471A) was constructed by mutation of 2 highly conserved residues in the ligand-binding domain (obtained from Dr. Krishna K. Chatterjee, University of Cambridge, Cambridge, UK) (34). SBE4-TK-Luc contains 4 copies of the consensus Smad-binding element linked to thymidine kinase and luciferase genes (35). The p3637-TK-Luc construct contains 6 copies of the COL1A2 Smad-binding element (5′-ATGCAGACA-3′) sequences linked to thymidine kinase promoter in pGL3-basic vector (36). The pRL-TK renilla luciferase (pRL-TK-Luc) construct served as an internal control.

Transient transfection.

Fibroblasts were seeded in 6-well clusters at 105 cells/well. At confluence, fibroblasts were transiently transfected with reporter constructs along with PPARγ expression vectors or appropriate empty vector using Superfect reagent (Qiagen, Valencia, CA). Transfected fibroblasts were pretreated with troglitazone or 15d-PGJ2 (both from Biomol, Plymouth Meeting, PA) or vehicle (0.1% DMSO) for 1 hour, followed by TGFβ2 (Genzyme, Framingham, MA). To test the PPARγ dependence of ligand-mediated effects on fibroblasts, the PPARγ antagonist GW9662 (Cayman Chemical, Ann Arbor, MI) was added to cultures 30 minutes prior to the PPARγ ligand. Following a further 48-hour incubation, fibroblasts were harvested and CAT or luciferase activities were determined in triplicate samples containing equal amounts of proteins. For determination of CAT activities, a phase extraction procedure was used. Luciferase activities were determined by scintillation counting. Transfection efficiency was monitored by measuring renilla luciferase activity in each sample. Each experiment was repeated 2–3 times, and the results were consistent.

Western blot analysis.

At the end of the incubation period, confluent fibroblasts were harvested and whole cell lysates or cell fractions were prepared, as described previously (5). The amount of protein in the cell lysates or nuclear fractions was determined using the Bio-Rad protein assay kit (Bio-Rad, Hercules, CA). Samples containing equal amounts of proteins (10–15 μg) were subjected to electrophoresis in 4–20% Tris–glycine gradient gels. Proteins were then transferred to polyvinylidene difluoride membranes, blocked with 10% fat-free milk in TBST buffer (20 mM Tris HCl, 137 mM NaCl, and 0.05% Tween 20). The membranes were incubated with antibodies against PPARγ (1:200 dilution), Smads 1, 2, and 3 (1:200 dilution), actin (1:200 dilution), or histone H3 (1:3,000 dilution) (all from Santa Cruz Biotechnology, Santa Cruz, CA), Smad7 (1:500 dilution; Novus Biologicals, Littleton, CO), type I collagen (1:250 dilution; Southern Biotechnology, Birmingham, AL), or α-SMA (1:500 dilution; Sigma, St. Louis, MO) in TBST buffer overnight. Membranes were then washed 3 times and incubated with appropriate secondary antibodies for 45 minutes.

Antigen–antibody complexes were visualized using the enhanced chemiluminescence detection system (ECL; Amersham Biosciences, Piscataway, NJ) according to the manufacturer's instructions. Membranes were exposed to autoradiography, and signals were scanned for quantitation. The results were normalized against the intensity of the actin signal.

Northern blot analysis.

At the end of the incubation period, total RNA was isolated from confluent fibroblasts by a 1-step extraction procedure using TRIzol reagent (Life Technologies, Grand Island, NY), and analyzed by Northern blotting as described previously (37). Nitrocellulose filters were sequentially hybridized with 32P-dCTP–labeled probes for human COL1A2, PPARγ, or GAPDH. Membranes were exposed to autoradiography, and signals were scanned for messenger RNA (mRNA) quantitation. The results were normalized against the intensity of the 18S ribosomal RNA or GAPDH probe.

Immunocytochemistry.

Fibroblasts were cultured on 8-well chamber slides in EMEM with 0.1% FBS in the presence or absence of troglitazone or 15d-PGJ2 (10 μM). At the end of the incubation period, cells were fixed in methanol, washed in phosphate buffered saline (PBS), and incubated with primary antibodies against PPARγ (Santa Cruz Biotechnology) for 30–60 minutes, as described previously (9). Slides were then washed with PBS and treated with fluorescein-conjugated anti-mouse IgG (Santa Cruz Biotechnology) for 30 minutes. To stain the nuclei, chambers were mounted with Vectashield (Vector, Burlingame, CA). Nonimmunized IgG was used as a negative control.

Following stringent washing of the slides, the subcellular distribution of fluorescence was evaluated by immunofluorescence or confocal laser scanning microscopy using a Zeiss LSM 510 microscope (Zeiss, Wetzlar, Germany). Each experiment was repeated at least 3 times, and the results were consistent. Quantitative analysis was performed by scoring 100 individual fibroblasts from different microscopic fields as showing predominantly nuclear or predominantly cytoplasmic distribution of immunofluorescence. The observer was blinded to the identity of each section.

Flow cytometry.

Fibroblasts were incubated with 10 μM troglitazone or 15d-PGJ2 in the presence or absence of TGFβ. After 48 hours, fibroblasts were harvested by gentle trypsinization, washed in ice-cold buffer, and fixed with 1% paraformaldehyde, followed by permeabilization with 0.2% saponin (Sigma) for 10 minutes. Direct immunofluorescence staining was performed using fluorescein isothiocyanate–conjugated monoclonal antibody to α-SMA (1 μg) or an isotype control mouse IgG (BD Biosciences, San Diego, CA). Aliquots of equal numbers of cells (104) were analyzed using a FACSCalibur instrument (Becton Dickinson, Mountain View, CA) equipped with CellQuest software. The percentage of positive cells was measured from a cutoff set determined by using isotype-matched control, and the mean channel fluorescence was measured over the entire distribution. Data are expressed as the percentage of α-SMA–positive fibroblasts from 3 separate experiments.

Statistical analysis.

Values are expressed as the mean ± SD. Statistical differences between experimental and control groups were determined by analysis of variance. P values less than 0.05 by Student's t-test were considered significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Expression and activation of PPARγ in normal skin fibroblasts.

The nuclear hormone receptor PPARγ was originally detected in adipocytes. Subsequent studies demonstrated that PPARγ is also expressed in endothelial cells, monocytes, macrophages, chondrocytes, and synovial fibroblast-like cells (for review, see ref. 38). In order to examine PPARγ protein expression in skin fibroblasts, low-passage confluent dermal fibroblasts were incubated for 48 hours in EMEM, harvested, and whole-cell lysates were analyzed by Western immunoblotting. In unstimulated fibroblasts, a single ∼50-kd band corresponding to PPARγ was detected (Figure 1A). Incubation with TGFβ resulted in a time-dependent increase in cellular PPARγ levels, with a maximal 6-fold increase at 18 hours. Next, Northern blot analysis was used to examine fibroblast PPARγ mRNA expression. The results revealed the presence of a single 1.8-kb band corresponding to PPARγ mRNA (Figure 1B).

thumbnail image

Figure 1. Expression of peroxisome proliferator–activated receptor γ (PPARγ) in human skin fibroblasts. Confluent cultures of foreskin fibroblasts were incubated in media with 10% fetal bovine serum in the absence or presence of transforming growth factor β2 (TGFβ2; 12.5 ng/ml) for the indicated periods. A, Whole cell lysates were analyzed by Western blotting with antibodies against PPARγ or actin. A representative autoradiogram is shown. B, Total RNA was extracted from quiescent fibroblasts and analyzed by Northern blotting using radiolabeled PPARγ polymerase chain reaction products or GAPDH as probes.

Download figure to PowerPoint

Having demonstrated that normal dermal fibroblasts express PPARγ mRNA and protein, we used 2 complementary approaches to determine whether the PPARγ signaling pathway was functional in these cells in vitro. First, confluent fibroblasts were incubated with the naturally occurring PPARγ ligand 15d-PGJ2 or with the synthetic PPARγ ligand troglitazone (both at 10 μM concentration), stained with specific antibody against PPARγ, and examined by confocal immunofluorescence microscopy. The results showed the presence of low levels of PPARγ distributed throughout the cytosol in unstimulated fibroblasts (Figure 2A). Treatment with troglitazone or 15d-PGJ2 resulted in substantial nuclear accumulation of cellular PPARγ. Increased nuclear accumulation of PPARγ was detectable as early as 30 minutes (data not shown), and was maximal after 24 hours of exposure to the ligand. In the absence of primary antibody, no staining was detected (data not shown). There was no increase in cellular trypan blue staining in ligand-treated fibroblasts compared with fibroblasts treated with DMSO (data not shown).

thumbnail image

Figure 2. Peroxisome proliferator–activated receptor γ (PPARγ) is functional in human skin fibroblasts. A, Confluent fibroblasts were incubated with vehicle (DMSO) or the PPARγ ligands troglitazone (TGZ) or 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2) (both at 10 μM) for 24 hours. Fibroblasts were then fixed and processed for immunocytochemistry using specific antibodies. Left, Confocal microscopy images representative of 3 separate experiments. Green indicates PPARγ; blue indicates nucleus (original magnification × 250). DAPI = 4′,6-diamidino-2-phenylindole. Right, Quantitation of the data from the control and PPARγ ligand–treated fibroblasts. Values are the mean ± SEM of 3 separate experiments. B, Fibroblasts were incubated with DMSO or 15d-PGJ2 (10 μM) for 24 hours. Nuclear extracts were prepared, and equal amounts of proteins were subjected to Western blot analysis for PPARγ or for histone H3 to confirm that the proteins were nuclear. C, Fibroblasts transiently transfected with the PPRE-TK-Luc construct (PPARγ response element [PPRE]; 200 ng/well) were incubated in media containing 15d-PGJ2 (10 μM) and/or GW9662 (1 μM). Following 48 hours of incubation, cells were harvested and luciferase activity was measured. Values are the mean ± SD of triplicate determinations from a single experiment normalized against protein concentrations and are expressed in arbitrary units. Transfection efficiency was monitored using renilla luciferase assay of each sample. Results were consistent in 3 separate experiments. ∗ = P < 0.05 versus control.

Download figure to PowerPoint

To further document the stimulation of nuclear accumulation of cellular PPARγ, Western immunoblotting was performed. Exposure of fibroblasts to 15d-PGJ2 for 24 hours resulted in substantial increase in nuclear levels of PPARγ (Figure 2B). In contrast, PPARγ ligands failed to induce a change in the total levels of cellular PPARγ in these fibroblasts (Figure 3A).

thumbnail image

Figure 3. Abrogation of transforming growth factor β (TGFβ)–stimulated type I collagen synthesis in skin fibroblasts by peroxisome proliferator–activated receptor γ (PPARγ) activation. Confluent fibroblasts were pretreated with DMSO or troglitazone (10 μM) or with 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2; 10 μM), followed by incubation in media with or without TGFβ (12.5 ng/ml) for 48 hours. A, Whole cell lysates were prepared, and collagen and PPARγ levels were analyzed by Western immunoblotting. B, Total RNA was extracted and subjected to Northern blot analysis using a COL1A2 probe. 18S ribosomal RNA (18S rRNA) was used as loading control. A representative autoradiogram is shown. Bottom, The autoradiogram was scanned by laser densitometry to determine the intensity of bands. Results were normalized against 18S rRNA in each lane. Open bars show untreated fibroblasts; closed bars show TGFβ-treated fibroblasts.

Download figure to PowerPoint

To further confirm that the PPARγ axis was functional in dermal fibroblasts, cells were transiently transfected with reporter vector p(AOX)3-TK-Luc containing the minimal PPRE from the acyl-CoA oxidase gene and incubated with 15d-PGJ2 for 48 hours. The results showed that 15d-PGJ2 induced a >2-fold increase in PPRE-driven luciferase activity (Figure 2C). To confirm that the observed 15d-PGJ2 response in fibroblasts was due to activation of cellular PPARγ, a selective inhibitor was used. GW9662 is an irreversible PPARγ ligand that works as an effective and nontoxic PPARγ antagonist (39). Transfected fibroblasts were pretreated for 30 minutes with GW9662 (1 μM) and then stimulated with 15d-PGJ2 for 48 hours. Luciferase assays showed that whereas GW9662 by itself had no effect on PPRE-driven promoter activity, it completely abrogated ligand-induced stimulation, indicating that the response was mediated through activation of endogenous PPARγ (Figure 2C). Together, these results provide the first direct evidence of functional PPARγ expression in normal skin fibroblasts.

Abrogation of TGFβ-stimulated collagen synthesis by PPARγ ligands.

The presence of a functional PPARγ pathway in fibroblasts raised the possibility that it may be involved in regulation of collagen gene expression in these cells. We therefore sought to determine the effect of PPARγ activation on the rate of basal and TGFβ-stimulated synthesis of collagen. For this purpose, confluent skin fibroblasts were incubated with 15d-PGJ2 or troglitazone (both at 10 μM concentration) in the presence or absence of TGFβ2 (12.5 ng/ml). At the end of the 48-hour incubation period, whole cell lysates were prepared from fibroblasts and examined by Western immunoblot analysis using anti–type I collagen antibodies. The results revealed that while TGFβ caused a >2-fold increase in the levels of cellular type I collagen, pretreatment with 15d-PGJ2 or troglitazone reduced this stimulation by 55% and 73%, respectively (Figure 3A). No change in cellular PPARγ levels was induced by either PPARγ ligand.

Decreased levels of intracellular collagen may result from decreased synthesis, increased secretion, or turnover of newly synthesized collagen. In order to directly determine the effects of PPARγ ligands on collagen gene expression, total RNA from ligand-treated fibroblasts was subjected to Northern blot analysis. While, as expected, TGFβ induced a >2-fold increase in COL1A2 mRNA levels incubation of fibroblasts with 15d-PGJ2 or troglitazone caused a significant decrease in collagen mRNA expression (Figure 3B). Importantly, short preincubation with either ligand almost completely abrogated the stimulation of collagen mRNA expression induced by TGFβ. These results indicate that naturally occurring or synthetic ligands of PPARγ inhibited TGFβ-stimulated collagen synthesis in normal dermal fibroblasts. Significantly, PPARγ activation had only a relatively modest effect on basal levels of collagen in unstimulated fibroblasts, suggesting that the major function of PPARγ is repression of TGFβ-induced responses.

Prevention of the stimulation of α-SMA expression in fibroblasts by PPARγ.

In fibrosis, a subgroup of resident fibroblasts transdifferentiate into myofibroblasts that express high levels of α-SMA normally found only in smooth muscle cells. Because these specialized fibroblasts show accelerated synthesis of extracellular matrix proteins, are resistant to apoptosis, and have contractile properties, they have a significant functional role in pathologic fibrosis. A major fibrogenic effect of TGFβ is stimulation of fibroblast–myofibroblast transdifferentiation (for review, see ref. 40).

In order to investigate the effect of PPARγ activation on the myofibroblast response, quiescent fibroblasts were incubated with TGFβ2 (12.5 ng/ml) in the presence or absence of 10 μM troglitazone or 15d-PGJ2 for 48 hours. Western blot analysis showed that TGFβ induced a significant increase in cellular levels of α-SMA, detected as a single 42-kd band (Figure 4A). Preincubation of the fibroblasts with either PPARγ ligand had no effect by itself, but in both cases resulted in substantial suppression of TGFβ-stimulated α-SMA expression.

thumbnail image

Figure 4. Prevention of TGFβ-stimulated expression of α-smooth muscle actin (α-SMA) by PPARγ ligands. Confluent fibroblasts were incubated in media containing DMSO or troglitazone (10 μM) or containing 15d-PGJ2 (10 μM) in the presence or absence of TGFβ (12.5 ng/ml) for 48 hours. A, Whole cell lysates were analyzed by Western blotting using specific antibodies against α-SMA or actin. B, Fibroblasts were stained for α-SMA, and 104 cells were subjected to flow cytometric analysis, and the proportion of cells with high-level α-SMA expression was determined. Isotype staining has been subtracted from each set of data. Results from a representative experiment are shown. Open bars show untreated fibroblasts; closed bars show TGFβ-treated fibroblasts. See Figure 3 for other definitions.

Download figure to PowerPoint

To further characterize the effect of PPARγ on myofibroblast transdifferentiation and to determine the proportion of α-SMA-positive fibroblasts, flow cytometry was used. Approximately 20% of unstimulated fibroblasts were strongly positive for α-SMA. Stimulation with TGFβ resulted in a >2-fold increase in the proportion of fibroblasts with high levels of α-SMA expression (Figure 4B) and a similar increase in the mean intensity of α-SMA staining (data not shown). Consistent with the results of Western blot analysis, pretreatment with troglitazone caused a 30% reduction in the proportion of fibroblasts with high levels of α-SMA expression and decreased the mean intensity of staining as well. Thus, these findings demonstrate that in normal dermal fibroblasts, activation of PPARγ by either naturally occurring or synthetic ligands inhibits the induction of critical profibrotic responses (stimulation of collagen synthesis and of α-SMA expression) induced by TGFβ.

Prevention of COL1A2 stimulation by PPARγ activation.

The expression of collagen mRNA is regulated predominantly at the level of transcription. To determine whether PPARγ-induced inhibition of collagen gene expression involved transcriptional repression, fibroblasts were transiently transfected with a COL1A2 reporter construct. The 772COL1A2/CAT construct harbors a Smad-response element between −258 and −263 bp of the human COL1A2 5′-flanking region, but lacks a consensus PPRE sequence. Consistent with previous results (5–8), exposure to TGFβ caused a >2-fold increase in COL1A2 promoter activity in transiently transfected fibroblasts (Figure 5A). Treatment with either 15d-PGJ2 or troglitazone alone for 48 hours had little or no effect on basal levels of COL1A2 promoter activity, but caused a dose-dependent suppression of stimulation induced by TGFβ.

thumbnail image

Figure 5. Prevention of TGFβ stimulation of the COL1A2 promoter by PPARγ ligands. Confluent fibroblasts were transiently transfected with the 772COL1A2/CAT construct (2.5 μg/well). A, Fibroblasts were pretreated with the indicated concentrations of troglitazone or 15d-PGJ2 for 1 hour, followed by 12.5 ng/ml of TGFβ2. After a further 48-hour incubation, fibroblasts were harvested and chloramphenicol acetyltransferase (CAT) activities were determined. Values are the mean ± SD of triplicate determinations from an experiment representative of 3 separate experiments. Open bars show untreated fibroblasts; closed bars show TGFβ-treated fibroblasts. B, Fibroblasts were pretreated with GW9662 for 30 minutes followed by 15d-PGJ2 (5 μM) and TGFβ2 for 48 hours. ∗ = P < 0.005 versus control. C, Fibroblasts were transfected with wild-type or dominant-negative mutant PPARγ expression vectors (L468A/E471A or L466A) or empty vector, along with 772COL1A2/CAT reporter constructs. Cultures were incubated with DMSO or 15d-PGJ2 (10 μM) for 1 hour, followed by 12.5 ng/ml of TGFβ2. After a further 48-hour incubation, fibroblasts were harvested and CAT activities were determined. See Figure 3 for other definitions.

Download figure to PowerPoint

We next sought to determine whether suppression of TGFβ-induced stimulation was mediated via the PPARγ pathway by use of a selective PPARγ antagonist. For this purpose, fibroblasts were pretreated with GW9662, followed by treatment with 15d-PGJ2 and stimulation with TGFβ. Transient transfection analysis showed that suppression of TGFβ-induced COL1A2 activity by 15d-PGJ2 was abrogated by GW9662 in a dose-dependent manner, whereas GW9662 by itself had no effect on either basal promoter activity or its TGFβ-induced stimulation (Figure 5B and data not shown).

To further confirm that ligand-induced PPARγ signaling in fibroblasts was directly responsible for the suppressive effects of 15d-PGJ2, dominant-negative mutants of PPARγ were used. Both PPARγ mutant constructs L468A/E471A and L466A abrogated the inhibitory effects of 15d-PGJ2 on TGFβ-induced stimulation of COL1A2 promoter activity in transiently transfected fibroblasts (Figure 5C). Together, these findings suggest that the inhibitory effect of PPARγ on collagen transcription in transfected fibroblasts was mediated through cellular PPARγ.

Abrogation of COL1A2 stimulation by overexpression of PPARγ.

To further characterize the role of cellular PPARγ in the repression of TGFβ-induced collagen stimulation, fibroblasts transiently cotransfected with PPARγ expression vector along with 772COL1A2/CAT or PPRE-TK-Luc were incubated in the presence or absence of TGFβ for 48 hours. Transfection of PPARγ resulted in increased cellular PPARγ protein levels (Figure 6B, top). The stimulation of COL1A2 promoter activity induced by TGFβ was completely abrogated in the presence of overexpressed PPARγ (Figure 6A). As expected, the activity of the PPRE reporter construct was increased in PPARγ-transfected fibroblasts, even in the absence of ligand, and was dramatically further enhanced when the PPARγ pathway was activated by incubation of the fibroblasts with 15d-PGJ2 (Figure 6B). Together, these results suggest that overexpression of PPARγ results in complete suppression of collagen gene transcription stimulated by TGFβ, even in the absence of activating ligand. Overexpression of the PPARγ receptor by itself results in increased PPARγ-dependent transcription and causes dramatic sensitization of fibroblasts to the activating ligands. These observations suggest that the relatively low level of PPARγ expression in normal fibroblasts may be limiting for negative regulation of TGFβ responses.

thumbnail image

Figure 6. Prevention of TGFβ stimulation of the COL1A2 promoter by PPARγ. Fibroblasts were cotransfected with PPARγ expression vector or empty vector along with A, 772COL1A2/CAT (COL1A2) or B, PPRE-TK-Luc (PPARγ response element [PPRE]) reporter constructs. Following incubation in the presence (closed bars) or absence (open bars) of A, TGFβ or B, 15d-PGJ2 for 48 hours, fibroblasts were harvested and chloramphenicol acetyltransferase (CAT) or luciferase activities were determined. The blot at the top of B shows PPARγ protein expression in whole cell lysates prepared from empty vector–transfected and PPARγ expression vector–transfected fibroblasts. See Figure 3 for other definitions.

Download figure to PowerPoint

Disruption of intracellular TGFβ/Smad signal transduction by activation of PPARγ.

We showed that troglitazone or 15d-PGJ2 abrogated multiple TGFβ-induced responses in normal fibroblasts and that this inhibition involved activation of the endogenous PPARγ pathway. Transcriptional responses in fibroblasts elicited by TGFβ are known to be mediated in large part through the Smad intracellular signal transduction pathway (10). In particular, we have recently demonstrated that in dermal fibroblasts, Smad3 was both necessary and sufficient to mediate TGFβ-induced stimulation of collagen gene expression (5). Transactivation of COL1A2 by Smad3 is mediated through a Smad-binding regulatory element containing the canonical CAGACA sequence (6). Stimulation of the myofibroblast marker α-SMA induced by TGFβ was also shown to be dependent on Smad-mediated signal transduction (41, 42).

To determine if the repression of TGFβ-induced responses by PPARγ was due to disruption of intracellular Smad signaling, we used a minimal reporter construct containing 4 copies of the consensus binding sites for Smad3/Smad4. Fibroblasts transiently transfected with SBE4-TK-Luc plasmid were incubated with 15d-PGJ2 (10 μM) or vehicle in the presence and absence of TGFβ2 for 48 hours, followed by determination of reporter activity. Only relatively low levels of luciferase activity were detected in unstimulated fibroblasts, whereas TGFβ caused marked up-regulation (Figure 7A). However, the TGFβ response was almost completely abrogated when fibroblasts were pretreated with 15d-PGJ2. Essentially identical results were obtained when p3637-TK, which contains 6 copies of the COL1A2 promoter SBE sequence, was used as the reporter construct in transient transfections (data not shown). Together, these results strongly suggest that ligand activation of the PPARγ pathway disrupted the Smad-dependent intracellular TGFβ signaling.

thumbnail image

Figure 7. Prevention of Smad2/3-driven transcriptional responses by PPARγ activation. A, Fibroblasts were transiently transected with SBE4-TK-Luc (Smad-binding element [SBE]) reporter construct along with pRL-TK-Luc. Fibroblasts were pretreated with 15d-PGJ2 (10 μM) for 1 hour, followed by 12.5 ng/ml of TGFβ2. After a further 48-hour incubation, fibroblasts were harvested and luciferase activities were determined. Values are the mean ± SD of triplicate determinations. Open bars show untreated fibroblasts; closed bars show TGFβ-treated fibroblasts. ∗ = P < 0.05 versus control. B, Fibroblasts cotransfected with SBE4-TK-Luc reporter plasmid along with PPARγ expression vector or empty vector and pRL-TK-Luc reporter constructs were incubated with or without TGFβ2 for 48 hours. Cells were harvested, and cell lysates were subjected to luciferase assays. ∗ = P < 0.01 versus control. C, Skin fibroblasts were incubated with 15d-PGJ2 (10 μM) and or GW9662 in the presence and absence of TGFβ2 for 48 hours. Whole cell lysates were then prepared and analyzed by Western blotting using antibodies against Smad3, Smad7, or actin. See Figure 3 for other definitions.

Download figure to PowerPoint

To confirm the direct effect of PPARγ on Smad-dependent TGFβ-signaling, fibroblasts cotransfected with PPARγ expression vectors along with the SBE4-TK-Luc reporter construct were incubated for 48 hours in the presence or absence of TGFβ. The results showed that overexpression of PPARγ resulted in abrogation of TGFβ stimulation of the minimal Smad-responsive promoter (Figure 7B), indicating that PPARγ signaling can repress Smad-dependent transcriptional responses in the absence of ligand. Because there are no consensus PPRE sequences in either 772COL1A2 or SBE4-TK, these results further suggest that PPARγ-mediated suppression of TGFβ signaling in fibroblasts is independent of PPARγ interaction with its cognate DNA elements.

In normal fibroblasts, TGFβ causes rapid phosphorylation and nuclear translocalization of cellular Smad3 and increased levels of inhibitory Smad7 (37). The repression of Smad-dependent transcriptional responses observed in PPARγ-treated fibroblasts could be the result of decreased steady-state levels of cellular Smad3, interference with TGFβ-induced nuclear import of Smad3, or alternately, diminished induction of Smad7 in response to TGFβ. In order to determine if PPARγ activation in the fibroblasts modulates the expression level or activation state of cellular Smads, fibroblasts incubated for 48 hours with TGFβ in the presence or absence of 15d-PGJ2 were analyzed by Western immunoblotting.

The results showed that in whole cell lysates, the cellular levels of Smad3 and Smad7 proteins were comparable in vehicle-treated and 15d-PGJ2–treated fibroblasts (Figure 7C, compare the first and fifth lanes). Furthermore, at this late time point, TGFβ treatment resulted in a marked down-regulation of Smad3 both in the presence and in the absence of 15d-PGJ2 (compare the second and sixth lanes). Therefore, inhibition of Smad-mediated TGFβ responses by PPARγ was not associated with alterations in the expression levels of Smad3 or Smad7 in these fibroblasts. These results also indicate that PPARγ is not simply a nonspecific inhibitor of all TGFβ responses (such as repression of Smad3), which suggests that PPARγ antagonistic effects on TGFβ signaling are likely to involve interference with nuclear Smad activation or function.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In scleroderma, lesional fibroblasts are responsible for the development of tissue fibrosis. In contrast to the signals and pathways implicated in the activation of fibroblasts, the endogenous mechanisms that limit this response have thus far received scant attention. Yet, such mechanisms must clearly be important in order to regulate fibroblast function and to prevent unopposed activation. For example, during physiologic processes of tissue remodeling, such as organogenesis or wound healing, the synthesis and accumulation of collagen must be terminated precisely at the appropriate stage. Failure of the inhibitory mechanisms would result in an exaggerated magnitude, or prolonged duration, of profibrotic responses, culminating in aberrant repair and pathologic fibrosis. In the case of TGFβ, the most potent of the profibrotic mediators, several mechanisms that repress intracellular signaling have been identified. One such mechanism is the induction of Smad7, the inhibitory member of the Smad family that blocks Smad phosphorylation and functions as an endogenous negative feedback to terminate TGFβ responses. In addition, various ligands that repress TGFβ responses also can induce Smad7 expression (43–45). It is not surprising that defective expression and/or function of endogenous inhibitors of fibroblast activation is linked to deregulated tissue remodeling and pathologic fibrosis (46).

The findings of the present study indicate that quiescent normal dermal fibroblasts express PPARγ at the mRNA and protein levels. PPARγ was originally identified in adipocytes and was more recently shown to also be expressed in the liver, pancreas, kidney, and vascular tissues, as well as in a variety of inflammatory cells (11, 27, 28, 47, 48). In unstimulated fibroblasts, only relatively low levels of PPARγ expression was found, and the receptor was largely distributed in the cytoplasm. Interestingly, TGFβ induced a significant time-dependent increase in PPARγ protein levels in these fibroblasts. In contrast, a recent study in vascular smooth muscle cells found that TGFβ caused early stimulation and late inhibition of PPARγ expression (49), suggesting that PPARγ regulation by TGFβ may be cell type–specific.

Naturally occurring ligands of PPARγ, such as 15d-PGJ2, as well as the synthetic pharmacologic PPARγ agonist troglitazone, caused activation of the receptor, as indicated by its enhanced nuclear accumulation in the absence of significant change in cellular PPARγ levels and by stimulation of a PPARγ-responsive minimal promoter in transfected fibroblasts. Remarkably, transient overexpression of PPARγ in the fibroblasts dramatically enhanced their sensitivity to stimulation by PPARγ agonists, suggesting that the level of expression is limiting for cellular responsiveness to endogenous PPARγ ligands. It is noteworthy that in diabetic glomerulosclerosis and other forms of fibrosis, PPARγ expression is diminished in affected tissues, potentially rendering them resistant to the potentially protective effects of endogenous PPARγ ligands (24, 27). This loss of responsiveness to an antifibrotic mechanism may play a significant role in the pathogenesis of fibrosis in these conditions. It will be of great interest to determine whether scleroderma fibroblasts show altered PPARγ expression or functional activity.

Agonists of the PPARγ receptor in normal fibroblasts caused suppression of collagen gene expression. Both 15d-PGJ2, an endogenously produced prostanoid, and synthetic thiazolidinedione drugs were effective inhibitors of TGFβ-induced collagen synthesis and COL1A2 promoter activity. Because 15d-PGJ2 is known to have PPARγ-independent cellular effects (13), it was important to confirm that the repression of TGFβ-induced responses was mediated through PPARγ. By using the potent and selective PPARγ antagonist GW9662, which covalently modifies a cysteine residue in the ligand-binding site of PPARγ (50), we established that the effect of 15d-PGJ2 on repression of TGFβ-induced collagen stimulation could be prevented in a dose-dependent manner. In addition, dominant-negative PPARγ expression vectors blocked the inhibitory effects of 15d-PGJ2 on TGFβ-stimulated COL1A2 promoter activity.

These results further demonstrate that in addition to collagen, the PPARγ ligand also disrupted the induction of α-SMA expression. Because stimulation of α-SMA by TGFβ is one of the key steps in the transdifferentiation of normal fibroblasts into myofibroblasts (40), these results indicate that PPARγ can interfere with multiple cellular events that are important in the pathogenesis of fibrosis. The inhibitory effects of PPARγ on TGFβ-induced fibroblast activation were selective (see below), and were not attributed to cellular toxicity.

The results of the transient transfection studies indicated that inhibition of TGFβ-stimulated collagen synthesis was mediated at least in part through a transcriptional mechanism. The regulation of COL1A2 transcription in normal fibroblasts has been investigated extensively (for review, see ref. 51). Such studies indicate that TGFβ-induced activation of the Smad signal transduction pathway results in rapid nuclear accumulation of the Smad2/3/4 complex and its binding to a CAGACA sequence in the COL1A2 promoter region (5, 6). In addition, interaction of the Smad complex with multiple coactivators and cofactors is also required for optimal transcriptional stimulation of COL1A2 by TGFβ (7, 8, 51–56). Furthermore, activation of the p38 and ERK cascades have also been implicated in the stimulation of collagen transcription elicited by TGFβ in fibroblasts (57, 58).

PPARγ functions as a ligand-activated transcription factor that modulates target gene transcription through direct binding to its cognate PPRE element (12). Sequence analysis of the COL1A2 promoter reveals the presence of a consensus SBE, but there are no putative PPARγ-binding elements. This suggests that the inhibitory effect of PPARγ on COL1A2 transcription involves disruption of the cellular transcriptional machinery that mediates the stimulation elicited by TGFβ. Accordingly, we examined the effects of PPARγ on Smad-mediated signaling using reporter constructs containing either a consensus SBE or tandem repeats of the Smad-binding element of the COL1A2 promoter. The results indicated that PPARγ was able to prevent TGFβ stimulation of these minimal promoter constructs containing the binding sites for only Smad3 and Smad4. These results suggested that PPARγ was able to directly antagonize the activation and/or function of Smad3 in fibroblasts (Figure 8). Furthermore, PPARγ did not decrease the protein expression of stimulatory Smad3 or increase the expression of inhibitory Smad7. The results also demonstrated that the suppression of Smad3 induced by TGFβ was unaffected by PPARγ, indicating that PPARγ did not block all TGFβ responses nonselectively.

thumbnail image

Figure 8. Regulation of transforming growth factor β (TGFβ)–induced responses in fibroblasts by peroxisome proliferator–activated receptor γ (PPARγ). Through activation of the intracellular Smad signal transduction pathway, TGFβ stimulates collagen synthesis, α-smooth muscle actin (α-SMA) expression, and myofibroblast transdifferentiation of resident fibroblasts in the skin. These responses contribute to tissue fibrosis. Naturally occurring endogenous ligands or synthetic pharmacologic agonists activate cellular PPARγ, resulting in the disruption of TGFβ/Smad signal transduction and the blocking of profibrotic responses. The expression of PPARγ is enhanced on fibroblasts by TGFβ, thereby sensitizing them to the antifibrotic effects of ligands. Activation of PPARγ may represent an effective antifibrotic intervention strategy. 15d-PGJ2 = 15-deoxy-Δ12,14-prostaglandin J2; TGZ = troglitazone.

Download figure to PowerPoint

The transactivating function of R-Smads may be repressed by PPARγ through cytoplasmic retention of the activated Smad complex. Direct physical interaction of R-Smads has been demonstrated with PPARγ in vascular smooth muscle cells (48) and with the estrogen receptor, another nuclear hormone receptor, in kidney carcinoma cells (59). We are pursuing further studies in order to precisely delineate the level of antagonistic interaction between PPARγ and Smads that would account for the repression of the profibrotic TGFβ responses in fibroblasts.

In summary, the results of the present study indicate that normal quiescent skin fibroblasts display constitutive PPARγ expression and activation of PPARγ-dependent transcriptional responses elicited by endogenous and synthetic PPARγ ligands. Up-regulation of PPARγ receptor levels by transient transfection of a PPARγ expression plasmid in fibroblasts markedly enhanced sensitivity to activation by PPARγ agonists. While TGFβ enhanced the expression of endogenous PPARγ in fibroblasts, both endogenous and synthetic ligands of PPARγ caused selective abrogation of TGFβ-induced profibrotic responses. The inhibitory effects of PPARγ on TGFβ-dependent responses involved direct antagonism of Smad signal transduction (Figure 8).

Together, these results suggest that PPARγ represents an important physiologic mechanism in the control of TGFβ responses in normal fibroblasts. Diminished PPARγ expression in affected tissues may be associated with aberrant repair. Accordingly, enhancing fibroblast sensitivity to exogenous PPARγ may represent a novel strategy for the treatment of fibrosis. Furthermore, in light of their potent antiinflammatory properties, pharmacologic PPARγ agonists may be particularly effective in scleroderma and related fibrotic conditions in which both inflammation and fibrosis play prominent roles.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We are grateful to Drs. Christopher Glass (University of California, San Diego) for the gift of the p(AOx)3-TK-Luc and PPARγ expression vectors, Krishna K. Chatterjee (University of Cambridge, Cambridge, UK), Thomas P. Burris (Lilly Corporate Center, Indianapolis, IN), and J. Larry Jameson (Northwestern University, Chicago, IL) for the wild-type and mutant PPARγ expression vectors, Jean-Michel Gauthier (Glaxo Wellcome, Les Ulis, France) for the p3637-TK-Luc plasmid, and Leigh Zawel (Johns Hopkins University, Baltimore, MD) for the SBE4-TK-Luc plasmid.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES