Idiopathic pulmonary fibrosis (IPF) is a progressive and lethal lung disease characterized by the expansion of the fibroblast/myofibroblast population and aberrant remodeling. However, the origin of mesenchymal cells in this disorder is still under debate. Recent evidence indicates that epithelial-mesenchymal transition (EMT) induced primarily by TGF-β1 plays an important role; however, studies regarding the opposite process, mesenchymal-epithelial transition, are scanty. We have previously shown that fibroblast growth factor-1 (FGF-1) inhibits several profibrogenic effects of TGF-β1. In this study, we examined the effects of FGF-1 on TGF-β1-induced EMT. A549 and RLE-6TN (human and rat) alveolar epithelial-like cell lines were stimulated with TGF-β1 for 72 h, and then, in the presence of TGF-β1, were cultured with FGF-1 plus heparin for an additional 48 h. After TGF-β1 treatment, epithelial cells acquired a spindle-like mesenchymal phenotype with a substantial reduction of E-cadherin and cytokeratins and concurrent induction of α-smooth muscle actin measured by real-time PCR, Western blotting, and immunocytochemistry. FGF-1 plus heparin reversed these morphological changes and returned the epithelial and mesenchymal markers to control levels. Signaling pathways analyzed by selective pharmacological inhibitors showed that TGF-β1 induces EMT through Smad pathway, while reversion by FGF-1 occurs through MAPK/ERK kinase pathway, resulting in ERK-1 phosphorylation and Smad2 dephosphorylation. These findings indicate that TGF-β1-induced EMT is reversed by FGF-1 and suggest therapeutic approaches to target this process in IPF.
- pulmonary fibrosis
- mesenchymal-epithelial transition
idiopathic pulmonary fibrosis (IPF) is a progressive, irreversible, and lethal lung disease characterized by epithelial cell injury and activation, expansion of the fibroblast/myofibroblast population, and extracellular matrix remodeling, resulting in an irreversible distortion of the lung architecture (17, 31). Mesenchymal cells accumulate in discrete small collections of spindle-shaped fibroblasts and myofibroblasts within myxoid stroma forming the so-called fibroblastic foci (15). These foci are considered to represent areas of active disease. Myofibroblasts, the α-smooth muscle actin (α-SMA)-expressing fibroblasts, are shown to be the main source of type I collagen and fibrogenic cytokines in fibrotic lesions, and also contribute to the altered mechanical properties of affected lungs (27).
The origin of fibroblasts in these fibroblastic foci has not been definitely elucidated. Migration and proliferation of resident mesenchymal cells and recruitment of fibrocytes may account for a fraction of them; however, emerging evidence indicates that an important number of matrix-producing fibroblasts/myofibroblasts may arise through a mechanism of epithelial-mesenchymal transition (EMT) (2, 16, 20, 27, 31, 32, 41).
EMT and its opposite, mesenchymal-epithelial transition (MET), are crucial for germ layer formation and cell migration in the early vertebrate embryo (1). Thus, EMT processes are essential for the progress of embryonic development, and, although usually maintained in a silent state in the adult, it may transiently turn-on for wound healing and tissue repair (1, 13). However, the abnormal activation of EMT programs has been associated with tissue fibrosis, cancer invasion, and metastasis (1, 9, 13, 16, 20, 25, 34, 38, 41, 45).
EMT involves a functional transition of polarized epithelial cells into migratory mesenchymal cells. During this process, cells lose many of their epithelial hallmarks such as strong inter-cell adhesion and polarity, show decreased expression of epithelial markers such as E-cadherin and acquire a mesenchymal phenotype including spindle-shaped morphology switching expression from keratin- to vimentin-type intermediate filaments, and become isolated, motile, and resistant to anoikis (18).
TGF-β1, a potent profibrotic factor, plays a pivotal role promoting EMT in normal development and in pathological processes, including the alveolar epithelial cell transition to myofibroblasts in vitro and in vivo (1, 16, 20, 41, 42, 44). More recently, a number of mediators able to induce EMT, such as WNT1-inducible signaling protein-1 (WISP1) and endothelin-1, have also been described (10, 19). However, studies regarding the converse process MET are scanty. Recently, it was shown that hepatocyte growth factor induces the upregulation of Smad7 and prevents acquisition of a myofibroblast phenotype in lung epithelial cells stimulated with TGF-β, thereby potentially reversing EMT (35).
We have previously demonstrated that acidic fibroblast growth factor-1 (FGF-1) displays antifibrotic functions downregulating collagen expression and antagonizing some profibrotic effects of TGF-β (4, 29).
In this study, we present evidence that FGF-1 reverts TGF-β1-induced EMT in alveolar epithelial-like cells throughout MEK-ERK pathway inducing the dephosphorylation of Smad2.
MATERIALS AND METHODS
Epithelial to mesenchymal transition.
A549 human lung epithelial-like cells and RLE-6TN, rat alveolar epithelial-like cells, were obtained from American Type Culture Collection. Human cells were cultured in Ham's F-12 and rat cells in Ham's F-12 and DMEM (vol/vol) containing 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2.5 μg/ml amphotericin B at 37°C in a humidified atmosphere of 5%–95% CO2-ambient air. Experiments were performed when A549 and RLE-6TN epithelial cells reached 50% and 30% confluence, respectively. To induce EMT, cells were stimulated for 72 h in serum-free medium (SFM) with 0.1% BSA and TGF-β1 (5 ng/ml); then, to evaluate the effect of FGF-1, the culture medium was replaced with SFM supplemented with 0.1% BSA, TGF-β1 (5 ng/ml), and recombinant human FGF-1 (R&D, Minneapolis, MN; 20 ng/ml) plus heparin (100 μg/ml) or TGF-β1 alone, and cells were incubated for 48 additional hours. Control cells were maintained in SFM with 0.1% BSA for 120 h. Conditioned media (CM) were collected and frozen at −70°C until use. Cells were used for immunocytochemistry, RT-PCR, and Western blot.
Immunofluorescence and confocal microscopy.
A549 cells (1 × 104 cells/cm2) grown on cover slips and treated as specified above were fixed for 3 min with methanol at room temperature, followed by acetone at −20°C for 2 min; then they were washed and kept in PBS at 4°C. Nonspecific binding was blocked with 10% goat serum in PBS-Tween 20 0.5%. Cells were incubated with FITC-conjugated anti-α-SMA (1:100 dilution; Sigma, St. Louis, MO) and mouse MAb for human E-cadherin (1:25 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C and then washed with PBS-Tween 0.5% and incubated with Alexa 633-conjugated goat anti-mouse antibody (5 μg/ml; Molecular Probes, Invitrogen) for 1 h at room temperature. At the end, the preparations were treated with mounting medium containing DAPI to stain cell nuclei (Ultra Cruz Mounting Medium, Santa Cruz Biotechnology). For fluorescence analyses, a confocal laser scanning microscope (Leica TCS SP5; Wetzlar, Germany) was used. Sequential scanning for the different fluorophores was performed, and the images were then merged.
Signaling pathways analysis.
A549 cells were incubated for 72 h with TGF-β1, preincubated 1 h with specific inhibitors, and subsequently incubated for an additional 48 h with TGF-β1 plus FGF-1/H. Inhibitors included PD-098059 [MAPK ERK kinase (MEK) inhibitor, 40 μM; Calbiochem, San Diego, CA], LY-294002 (phosphatidylinositol 3-kinase, 50 μM; Cell Signaling Technology, Beverly, MA), SB-203580 (p38-MAPK pathway, 20 μM; Calbiochem), and SB-431542 (TGF-β1 type I receptor kinase, 10 μM; Calbiochem). Signaling pathways were examined by immunoblotting.
Cells were lysed in RIPA buffer M-PER (Pierce, Cheshire, UK) containing a protease inhibitor cocktail (P8340; Sigma). Protein concentrations were determined by Bradford method. Equal amounts of proteins (40 μg) were resolved in 10% SDS-PAGE and transferred onto a Hybond ECL membrane (Amersham, Buckinghamshire, UK). After blocking with non-fat dried milk, the membranes were incubated with primary mouse monoclonal antibodies for 1 h at room temperature followed by detection using horseradish peroxidase-labeled anti-mouse antibody. Primary antibodies included anti-human E-cadherin (Biogenex Lab, San Ramon, CA; 2:1,000 dilution), α-SMA (Sigma; 1:300), Smad2 (Santa Cruz Biotechnology; 1:1,000), phospho-Smad2 (Santa Cruz Biotechnology; 1:100), Erk1/2 (Santa Cruz Biotechnology; 2:1,000), phosphorylated Erk1/2 (5:1,000 dilution; Santa Cruz Biotechnology), and β-tubulin (1:200 dilution; Santa Cruz Biotechnology). Also, anti-rat E-cadherin (Biogenex Lab; 2:1,000 dilution) and α-SMA (Sigma-Aldrich, St. Louis, MO; 1:300 dilution) were used. The proteins were revealed by an enhanced chemiluminescence detection system (Amersham Biosciences). Band densities were digitalized and quantified using image analysis software (ID; Eastman Kodak, Rochester, NY). Results were expressed as a ratio of band density to total β-tubulin. Changes in levels of phosphorylated proteins were assessed with reference to the respective non-phosphorylated proteins.
Zymography was used to identify proteins with gelatinolytic activity in CM as described elsewhere (26). Briefly, samples were mixed with an equal volume of 2× sample buffer, resolved under nonreducing conditions on 7.5% SDS-PAGE containing 1 mg/ml gelatin as a substrate. CM from human lung fibroblasts and from U2-OS cells stimulated with PMA was used as MMP-2 and MMP-9 positive controls, respectively. Band densities were digitalized and quantified using image analysis software (ID).
Real-time quantitative RT-PCR.
Total RNA was extracted from epithelial cells using TRIzol reagent (Invitrogen Life Technologies, Grand Island, NY) and reversed transcribed into cDNA (Advantage RT-for-PCR Kit; Clontech, Palo Alto, CA) according to the manufacturer's instructions. RT-PCR amplification was performed with the i-Cycler iQ Detection System (BioRad, Hercules, CA) using TAQMAN probes (PE Applied Biosystems, Wellesley, CA) labeled with FAM for E-cadherin (Hs 01023895_M1), α-SMA (Hs 00909449_M1), and 18S rRNA (4352930E) as endogenous control. PCR was performed with the cDNA working mixture in a 25-μl reaction volume containing 3 μl of cDNA, 2 mM MgCl2, 0.2 mM dNTPs, 0.2 μM of each TAQMAN probe and 2.5 units of recombinant Taq DNA polymerase (Invitrogen). A dynamical range was built with each product of PCR on copy number serial dilutions from 1 × 108 to 1 × 101. Standard curves were calculated referring the threshold cycle (Ct) to the log of each cDNA. Results were expressed as the number of copies of the target gene normalized to 18S rRNA. The PCR conditions were 2 min at 94°C followed by 40 cycles of 15 s at 94°C and 1 min at 60°C.
Apoptosis of epithelial cells.
A549 cells were grown to 70% confluence in T-25 culture flasks. Apoptosis was determined by annexin-V staining assessed by flow cytometry in cells stimulated with recombinant human FGF-1 (20 ng/ml) and controls, as follows: harvested cells were incubated with Annexin V-PE and 7AAD (BD Bioscience, San Diego, CA) following the manufacturer's instructions. Apoptotic cell death (Annexin-V+, 7AAD-) was measured by flow cytometry using a FACSAria cytometer. The results were analyzed with Flow Jo software. Two independent experiments were performed by duplicate.
Tissue sections from IPF and controls were treated as previously described (33). Antigen retrieval was performed in citrate buffer (10 mM, pH 6.0) for 6 min in a microwave. Samples were incubated with anti-human anti-FGF-1 mouse monoclonal (Abcam, Cambridge, MA; 3 μg/ml) at 4°C overnight. A secondary biotinylated anti-immunoglobulin followed by horseradish peroxidase-conjugated streptavidin (BioGenex, San Ramon, CA) was used according to the manufacturer's instructions. 3-Amino-9-ethyl-carbazole (AEC; BioGenex) in acetate buffer containing 0.05% H2O2 was used as substrate. The sections were counterstained with hematoxylin. The primary antibody was replaced by nonimmune serum for negative control slides.
Results are presented as means ± SD of at least three independent experiments. Statistical analysis was performed with ANOVA followed by Dunnett's multiple comparison post hoc tests. P < 0.05 was considered statistically significant.
TGF-β1-induced EMT is reversed by FGF-1/H.
After TGF-β1 treatment, A549 human lung epithelial-like cells lost cell-cell contact and turned into a spindle-like mesenchymal phenotype (Fig. 1). A549 cell line treated with FGF-1/H in the presence of TGF-β1 recovered their polygonal/cobblestone phenotype. Similar results were obtained with rat alveolar epithelial-like cells RLE-6TN (Fig. 1). This morphological change was accompanied by a substantial reduction, demonstrated at the gene and protein level, of the characteristic epithelial phenotypic marker E-cadherin (Fig. 2, A and B) with concurrent and significant induction of the mesenchymal marker α-SMA (Fig. 2, C and D), and also of vimentin (not shown). After treatment with FGF1/H, the expression of both E-cadherin and α-SMA returned to control levels as demonstrated by Western blot and real-time RT-PCR (Fig. 2). Comparable results for E-cadherin expression was obtained when rat RLE-6TN cells were treated with FGF-1/H (Fig. 3A). Also, α-SMA that is usually expressed by these cells (22) was upregulated by TGF-β1 and returned to basal levels after treatment with FGF-1H (Fig. 3B). Treatment of epithelial cells with heparin alone showed no effect (not shown).
Confocal immunofluorescence microscopy using double-labeling for E-cadherin (red) and α-SMA (green) corroborated these results. As shown in Fig. 4, control A549 epithelial cells displayed intense E-cadherin labeling without α-SMA expression. Incubation of A549 cells with TGF-β1 strongly reduced E-cadherin immunostaining while an intense α-SMA expression was observed. When epithelial cells were incubated with TGF-β1 for 72 h followed by an additional 48 h with FGF-1/H, labeling for both E-cadherin and α-SMA regressed to control levels. Since we have previously demonstrated that FGF-1 induces cell death in mesenchymal cells (29), we evaluated this effect on A549 epithelial-like cells. Our results showed that FGF-1 did not induce apoptosis on this cell line (1.49 ± 0.26 in controls vs. 1.61 ± 0.33 in FGF-1-stimulated cells).
FGF1 reverts the increase of MMP-2 and MMP-9 induced by TGF-β1.
It is known that during TGF-β1-induced EMT, alveolar and other epithelial cells increase the expression of gelatinases, mainly MMP-2, that are necessary to acquire the cell migratory phenotype characteristic of mesenchymal cells (14, 21). As illustrated in the zymogram of Fig. 5, under basal conditions A549 cells show moderate activity bands of gelatinases MMP-2 and MMP-9. After TGF-β1 treatment, there is a strong increase of both MMP-2 and MMP-9 enzymes. When cells were incubated with TGF-β1 and then with FGF-1/H, MMP-2 and MMP-9 activities returned to control levels.
FGF-1 reverts TGF-β1-induced EMT through MEK-dependent signaling.
To analyze the signaling pathways involved in EMT reversion induced by FGF-1, we used several pharmacological inhibitors. A549 cells were incubated for 72 h with TGF-β1 and then preincubated 1 h with specific inhibitors and subsequently incubated for additional 48 h with TGF-β1 plus FGF-1/H and inhibitors. As previously demonstrated, downregulation of E-cadherin by TGF-β1 (Fig. 6) is reverted by FGF-1/H. When the effects of different pharmacological inhibitors were compared with the band intensity of FGF-1/H, the MEK inhibitor PD-098059 was the only one blocking significantly the reversion of E-cadherin suggesting that this pathway is implicated in the effect of FGF-1 on EMT. A modest effect was also observed when p38 and PI3K were antagonized, whereas inhibition of TGF-β1 type I receptor kinase showed no effect. We also used the TGF-β1 type I receptor kinase inhibitor in absence of FGF-1 observing that this inhibitor successfully abolished the effect of TGF-β1 on E-cadherin expression (Fig. 6).
The role of MEK kinase signaling pathway in the FGF-1-induced EMT reversion was further corroborated. As shown in Fig. 7, PD-098059 (MEK inhibitor) abolished the effect of FGF-1/H on E-cadherin. The inhibitor alone had no effect. Also, the inhibitor did not affect the reduction of E-cadherin expression induced by TGF-β1 indicating that MEK/ERK pathway is not involved in TGF-β1-induced EMT. Additionally, we demonstrated that FGF1/H phosphorylates Erk-1 (Fig. 8A).
FGF-1 inhibits Smad2 phosphorylation.
Although the molecular mechanisms governing TGF-β1-induced EMT remain unclear, most evidence indicates that it may be mainly regulated by Smads. Therefore, we explored the effect of FGF-1/H on TGF-β1-induced Smad phosphorylation. Epithelial cells were stimulated for 3 and 6 h with TGF-β1 alone or in the presence of FGF-1 plus heparin, and phosphorylation of Smad2 was assessed by Western blot. As shown in Fig. 8B, expression of phosphorylated Smad2 protein was increased by TGF-β1. This increase was markedly reduced by the treatment with FGF-1 at 3 and 6 h.
Immunolocalization of FGF-1 in IPF lungs.
The localization of FGF-1 was examined by immunohistochemistry in IPF and control lungs. Immunoreactive FGF-1 was found in discrete areas of the IPF lungs. As illustrated in Fig. 9, FGF-1 was localized primarily in the cytoplasm of endothelial cells (Fig. 9A) and in some alveolar epithelial-like cells (Fig. 9, B and C). In addition, FGF-1 was also observed bound to the extracellular matrix. Immunohistochemical staining for FGF-1 was negative in normal lungs as well as in lung tissue samples incubated without the primary antibody.
IPF is characterized by epithelial cell injury and activation leading to the formation of fibroblastic foci, the active sites of fibrogenesis (15, 31, 32). Although the origin of fibroblasts in IPF (and other fibrotic lung disorders) is unclear, a growing body of evidence indicates that EMT could be one of them (16, 20, 41). TGF-β1 appears to be the main responsible of this process and may induce the transition of alveolar epithelial-like cells to myofibroblasts, both in vitro and in vivo (16, 41, 44). Wnt/Wingless pathway, that crosstalk with a variety of growth factors including TGF-β, may also play a role in this process. The inhibition of GSK-3, a key component of the Wnt response, leads to the downregulation of E-cadherin and EMT in cultured epithelial cells (3), and recent work indicates that Wnt pathway is upregulated in IPF (19, 34).
Less is known regarding the opposite embryological process MET in adults, although a mesenchymal-to-epithelial reverting transition has been described during metastatic seeding (40). In kidney fibrosis, it has been shown that bone morphogenic protein-7 induces MET in adult renal fibroblasts and facilitates regeneration of injured kidney (47).
In lung fibrosis, it has been recently demonstrated that hepatocyte growth factor induces MET, probably through the upregulation of Smad7, an inhibitor of TGF-β signaling (35).
In the present study, we analyzed the potential role of FGF-1 to revert the TGF-β1-induced EMT. FGF-1 was chosen because previous work in our lab has revealed that it has strong antifibrotic properties and antagonizes TGF-β (4, 29). FGF-1 was used in combination with heparin since it is well known that for the full activation of the FGF receptor by its FGF ligand and subsequent signaling, the presence of heparan sulfate or heparin is required. The mechanism has not been fully elucidated. On one hand, several studies indicate a direct role for heparin in FGF1-FGFR interaction and receptor activation (30, 48). On the other hand, some recent work suggests that this molecule protects the naturally unstable FGF-1 against heat and/or proteolytic degradation but is not essential for a direct FGF1-FGFR interaction and receptor activation (46).
Our results showed that FGF-1 plus heparin blocked the TGF-β1-induced EMT, reverting the typical phenotype of spindle-like mesenchymal cells to a round/cobblestone epithelial appearance. Accompanying this phenotypic change, we showed that E-cadherin, a key component of adherens junctions and critical in the maintenance of epithelial integrity, was highly reexpressed after FGF-1 stimulation, while the expression of the mesenchymal markers α-SMA and vimentin was strongly downregulated. This effect was observed in two different epithelial cell lines obtained from human and rat lungs. Interestingly, FGF-1 was found in vivo in discrete areas of the IPF lungs and was expressed by elongated alveolar epithelial-like cells and endothelial cells.
The transition to mesenchymal cells induced by TGF-β1 was accompanied by a strong increase of MMP-2 and MMP-9 activities as demonstrated by gelatin zymography, and this effect was also abolished when cells regained the epithelial phenotype induced by FGF-1. It has been described that the transition from epithelial to a mesenchymal phenotype, both in fibrosis and cancer, results in an upregulation of MMP-2, a matrix-degrading enzyme (14, 28, 36). Increased activity of MMP-2 augments the migratory capacity of fibroblasts and enhances invasion and metastasis by cancer cells. Also, the developmentally regulated expression of MMP-2 correlates with the EMT that generates the neural crest, the sclerotome, and dermatome, suggesting that this enzyme is critically involved in the transformation of epithelia to mesenchyme, and also in the later dispersion of mesenchymal tissues (6). Upregulation of MMP-9 has also been described in some models of EMT (5, 8, 37). In a study that evaluated the potential for airway epithelium from lung transplant recipients to undergo EMT, a threefold increase of MMP-9 was demonstrated with a concomitant increase in the number of invasive cells (5).
The dissection of the signaling mechanisms that are activated in response to TGF-β and lead to EMT has demonstrated a pivotal role of the Smad signaling pathway (42). Increased expression of Smad2 or Smad3 with Smad4 induces EMT, whereas expression of dominant negative versions of Smad2 or Smad3, or the knockdown for Smad4 expression by RNA interference blocks TGF-β-induced EMT (12, 39).
In our study, alveolar epithelial-like cells stimulated with TGF-β1 showed strong phosphorylation of Smad2, which was clearly inhibited by FGF-1 in a time-dependent manner.
To identify the signaling pathway involved in the FGF-1-induced reversion of TGF-β1-elicited EMT, epithelial cells were preincubated with different specific pharmacological inhibitors. Activation of FGFR by FGF-1 allows the recruitment and activation of Src homology (SH2)- or phosphotyrosine (PTB)-containing proteins, leading to the activation of various cytoplasmic signal transduction pathways including PI3K/Akt, MEK1/2-ERK, and p38MAPK (22, 24). Occasionally, in some in vitro systems a dual signaling pathway has been identified. Thus, for example, FGF-1-upregulates MMP-9 expression in ENU1564 cell lines by increasing the activities of NF-κB and AP-1 involving the activation of both PI3K/Akt and MEK1/2-ERK (23).
The results of our study showed that the inhibition of MAPK ERK kinase pathway markedly attenuated the effect of FGF-1 on TGF-β-induced EMT. Thus, when alveolar epithelial-like cells were stimulated with TGF-β1 and then the MEK inhibitor was added the effect of FGF-1, restoring the E-cadherin expression to control levels was inhibited. FGF-1 with heparin alone (without a previous stimulus with TGF-β1) or the MEK inhibitor alone had no effect on E-cadherin expression.
Interestingly, several studies have indicated that ERK signals may repress considerable subsets of intermediate TGF-β target genes and that the MAPK/ERK pathway is also involved in Smad7 upregulation (7, 43).
In summary, this study identifies FGF-1 as an antagonist of TGF-β1-induced EMT inhibiting Smad2 phosphorylation through MEK signaling pathway. This finding suggests that FGF-1 might have a role in the therapy of fibrotic lung disorders where EMT contributes to the expansion of the fibroblast/myofibroblast population. However, additional confirmation with primary human alveolar epithelial cells is necessary.
M. Selman is a consultant of Boehringer Ingelheim.
- Copyright © 2010 the American Physiological Society