Hepatocyte Growth Factor Inhibits Epithelial to Myofibroblast Transition in Lung Cells via Smad7

Cell Culture Treatments

Rat lung epithelial (RLE)-6TN cells were obtained from American Type Culture Collection (Manassas, VA) and grown in 1:1 mixture of Dulbecco's modified Eagle's medium and nutrient mixture F-12 Ham (Gibco/Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Gemini Bioproducts, West Sacramento, CA), 40 mmol/L HEPES (Gibco/Invitrogen), 100 U/ml penicillin G, and 100 μg/ml streptomycin (Gibco/Invitrogen). For preparation of protein extracts for Western blotting, RLE-6TN cells were incubated in medium alone or in medium containing 2.5 ng/ml TGF-β1 (R&D Systems, Minneapolis, MN) ± 10, 50, or 100 ng/ml HGF (R&D Systems) for 24 hours for α-SMA protein and β-actin expression measurements by Western blotting. For phosphoprotein assessment, RLE cells were serum starved for 24 hours, followed by 100 ng/ml HGF treatment for 5 or 25 minutes. To inhibit specific cell signaling pathways, cells were pretreated with either a specific inhibitor of p38 (20 μM SB203580; Calbiochem, San Diego, CA), Akt inhibitor (10 μM; Calbiochem) or a specific inhibitor of MEK (50 μM PD98059; Calbiochem) for 30 minutes followed by treatment with 2.5 ng/ml TGF-β₁ alone or in combination with 100 ng/ml HGF.

Primary Cell Isolation

Isolation of alveolar type 2 epithelial cells was performed using a previously described protocol with some modifications (27). Briefly, lungs from 6- to 8-week-old C57BL6 male mice were perfused with PBS containing 20 U/ml dispase (Gibco/Invitrogen), 0.1 mg/ml collagenase/dispase (Roche, Branchburgh, NJ), and 0.1 mg/ml DNase II (Sigma Aldrich, St. Louis, MO). Lungs were then digested at 37°C for 30 minutes, followed by mincing and sequentially passing through 100-μm and 40-μm sieves (BD Biosciences, San Diego, CA). Cells were washed and red blood cells were lysed using PharmLyze (BD Biosciences). CD45+ and CD32+ cells were magnetically depleted using biotinylated antibodies specific for CD45 and CD16/32 (BD Biosciences), respectively, followed by incubation with anti-biotin magnetic beads (Miltenyi Biotech, Auburn, CA). The epithelial cell–enriched fraction was collected and resuspended in bronchial epithelial growth media serum-free media with growth supplements (Clontech, Mountain View, CA). Cells were plated on collagen-coated plastic chambered slides. A quantity of 100 μg/ml cis hydroxy proline (Sigma Aldrich) was added during plating to inhibit fibroblast proliferation.

Immunocytochemistry

RLE cells were plated on glass chamber slides at 20% confluence. After 24 hours, cells were fixed with 2% paraformaldehyde in PBS for 10 minutes at room temperature. Nonspecific binding was blocked with 10% goat serum in 5% BSA/PBS. The antibodies used to examine expression of specific proteins were: mouse monoclonal anti–α-SMA (1:100 dilution; Dako, Carpinteria, CA), rabbit polyclonal anti-zonula occludens-1 (ZO-1) (2 μg/ml; Zymed, San Francisco, CA), mouse monoclonal anti-fibronectin (5G7, Abcam), rabbit polyclonal anti-human Collagen I (AB675P; Chemicon, Billerica, MA) or rabbit polyclonal E-cadherin (24E10; Cell Signaling, Danvers, MA). Antibody incubation was performed for 1 hour at room temperature followed by incubation with secondary conjugated Alexa 555 and Alexa 488 antibodies (Molecular Probes/Invitrogen). Nuclei were stained with a 1:1,000 dilution of DRAQ5 (Alexis Biochemical, Lausen, Switzerland). For Smad7 subcellular localization experiments, RLE cells were plated at 20% confluence. After 24 hours, cells were treated with 50 mM MEK inhibitor for 30 minutes and HGF for 3 hours, then fixed in 2% paraformaldehyde as described above. Cells were sequentially immunostained with rabbit anti-Smad7 (1:100 dilution; Santa Cruz Biotechnology) overnight at 4°C and then with mouse anti–E-cadherin (1:50 dilution; BD Transduction Labs, Franklin Lakes, NJ) for 1 hour. Appropriate secondary conjugated Alexa 488 and Alexa 555 antibodies were added for 1 hour, followed by DRAQ5 nuclear stain. For Smurf1 immunostaining, cells were incubated with 1:75 dilution of anti-goat Smurf1 antibody (Novus Biologicals, Littleton, CO). Confocal images were captured using an Olympus Fluoview 1000 Confocal system with an Olympus IX81 inverted microscope (Olympus, Center Valley, PA). Acquired images were analyzed with MetaMorph 7.1 software (Molecular Devices, Sunnyvale, CA). To calculate intensity levels of nuclear Smad7 and Smurf1. Nuclear regions were defined by DRAQ5 nuclear stain.

Nuclear, Cytoplasmic, and Total Protein Extract Preparation and Western Blotting

RLE nuclear and cytoplasmic extracts were prepared as previously described (28). RLE total cell extracts were prepared in SDS lysis buffer (0.0625 M Tris pH 6.8, 1.5% SDS, 10% glycerol) warmed to 95°C. Thirty-microgram protein samples were separated using 4 to 15% gradient polyacrylamide gels, transferred to PVDF membranes, and probed with the following antibodies: mouse monoclonal anti–α-SMA (1:2,000), rabbit anti-Smad7 (1:750); and the following antibodies purchased from Cell Signaling Technology: rabbit anti-phospho Smad2 (pSmad2) (Ser465/467) (1:1,000), rabbit anti-phospho Smad3 (pSmad3) (1:1,000), rabbit anti-Smad2/3 (1:1,000), rabbit anti-pSmad3 (Ser433/435) (1:1,000), rabbit anti–phospho-p44/42 MAPK (Thr202/Tyr204) (1:1,000), rabbit anti–total p44/42 MAPK (1:1,000), rabbit anti–phospho-JNK (1:1,000), rabbit anti–total JNK (1:1000), rabbit anti-Akt (Thr 308) (1:1,000), rabbit anti-Akt (Ser 473) (1:1000), rabbit anti–total Akt (1:1000), rabbit anti–phospho-p38 MAP Kinase (Thr180/Tyr182) (1:1,000), rabbit anti-p38 MAP kinase (1:1,000), rabbit anti–β-actin (1:1,000). All blots were exposed for optimal lengths of time for visualization.

Smad7 Knockdown Using siRNA

RLE cells were left untransfected, or transfected with 10 or 50 nM On-target plus Smartpool Smad7 siRNA or 50 nM nonspecific siRNA (Dharmacon, Chicago, IL) with siPORT NeoFX (Ambion) according to manufacturer's recommendations. Twenty-four hours after transfection, fresh medium containing 100 ng/ml HGF was added to the cells, followed by 24 hours of incubation. Cells were scraped and Smad7, α-SMA, and β-actin proteins were assessed by Western blotting. As a control for transfection efficiency, fluorescently tagged siRNA (siGLO; Dharmacon) was introduced into cells using the above protocol.

Statistical Analysis

All experiments were done independently three times unless otherwise indicated. Data are presented as means ± SE. Data were analyzed using Student's unpaired t test or Mann Whitney t test. In each case, P ≤ 0.05 was considered significant.

RLE-6TN Cells Express Basal ZO-1, E-Cadherin, Aquaporin-5, and α-SMA

To characterize the cellular properties of the rat alveolar cell line, RLE-6TN, we performed immunostaining for the epithelial marker, ZO-1 (15, 29) and the myofibroblast marker, α-SMA. All cells expressed ZO-1 with a fraction of the cells also showing basal co-expression of α-SMA (Figures 1A–1C). In areas where cells exhibited cobblestone morphology, ZO-1 was plasma membrane–associated. A pattern of intracellular, fibril-associated α-SMA expression noted in some cells was consistent with several other reports (15, 30, 31). In addition, as shown by immunoblotting in Figure 1D, RLE cells express the epithelial cell markers, E-cadherin and aquaporin-5. Cells in lane 1 were plated at a low density and in lane 2 at a higher density, with no clear difference in protein expression suggesting that cell–cell contact did not influence myofibroblast or epithelial cell marker expression. Cumulatively, these results suggested that RLE-6TN cells possess characteristics of both epithelial cells and myofibroblasts. It should be noted that given the uniform expression of ZO-1 in all cells, it was clear that some cells spontaneously assumed a fibroblast phenotype. However, this property was not observed in other epithelial cell lines from human or rodent origin that are routinely maintained in our laboratory. This feature of RLE-6TN cells, however, prompted us to investigate whether the observed inhibitory effects of HGF on bleomycin-induced pulmonary fibrosis (20–24) could be studied at a mechanistic level using these cells as a model system.

HGF Prevents Expression of TGF-β–Induced Fibrogenic Markers in Type 2 Pneumocytes and RLE Cells

Studies in renal, hepatic, and lung fibrosis indicate that HGF may completely antagonize the effects of TGF-β and may contribute to the inhibition of epithelial to mesenchymal transition (7, 22–25, 32). We therefore sought to determine whether HGF may up-regulate the expression of epithelial proteins, and down-regulate fibrogenic proteins induced by TGF-β, in cultures of RLE cells. The expression of epithelial markers ZO-1 and E-cadherin was appreciably induced by HGF, but reduced by TGF-β. Simultaneous exposure to HGF and TGF-β did not inhibit the expression of epithelial markers (Figure 2A, panels i and iv). Cellular expression of pro-fibrotic markers such as α-SMA, collagen type 1, and fibronectin in response to HGF and TGF-β, alone or in combination, was also assessed in RLE cells. The results of our experiments indicate that the expression of these proteins was up-regulated by TGF-β but not by HGF, and HGF treatment along with TGF-β abrogated the effect of TGF-β (Figure 2A, panels i–iii).

We also investigated whether the effects of HGF could be observed in primary murine alveolar type II cells. Murine primary pneumocyte cultures were generated using a previously established protocol and used to study the ability of HGF to prevent EMT (Figure 2B, left panel). Robust expression of the alveolar type 2 epithelial marker pro–surfactant protein C (Figure 2B, middle panel), and minimal expression of the Clara cell–specific airway epithelial cell marker CC-10 (Figure 2B, right panel) indicated an enriched population of alveolar epithelial cells. Cells cultured in the presence of HGF demonstrated retention of epithelial morphology with intact cell–cell contacts and strong expression of ZO-1. Treatment of cells with TGF-β, however, caused a change in morphology to fibroblast-like spindle-shaped cells with lower expression of ZO-1 and up-regulation of α-SMA. Upon exposure of the cells to HGF and TGF-β in combination, we observed a marked induction of ZO-1 and reduction of α-SMA (Figure 2C). These results suggest that HGF inhibits TGF-β signaling and fibroblastic transition in lung epithelial cells.

Interference with TGF-β Type I Receptor Signaling Down-Regulates the Expression of α-SMA in RLE Cells

We next explored the basis for basal α-SMA protein expression in RLE cells. As noted in many studies, we chose α-SMA as a marker for the myofibroblast phenotype which has semblance to both fibroblasts and smooth muscle cells (33). We hypothesized that activation of the TGF-β receptor was responsible for the basal expression of α-SMA. Therefore, we treated cells with a TGF-β type I receptor kinase inhibitor, SB-431542, to block TGF-β signaling. This inhibitor has been shown to specifically block signaling downstream of the TGF-β type I receptor (activin receptor-like kinase 5 [ALK5]). It also inhibits the activin type I receptor ALK4 and the nodal type I receptor ALK7, which are very highly related to ALK5 in their kinase domains (34). Basal α-SMA protein decreased in response to increasing concentrations of the inhibitor (0.5–5μM) (Figure 3A). As expected, the inhibitor also blocked Smad2 phosphorylation (Figure 3B). This suggests that activation of the TGF-β type I receptor kinase or a related receptor kinase promotes α-SMA protein expression and a myofibroblast phenotype in RLE cells.

HGF Significantly Decreases α-SMA Protein Expression in RLE Cells

Through the activation of its tyrosine kinase receptor, HGF modulates cell motility, morphogenesis, survival, and proliferation. These effects may be important for the protective action of HGF in models of pulmonary fibrosis in vivo, as demonstrated in multiple studies (22–25). We examined whether HGF could alter levels of α-SMA protein in RLE cells. Cells were exposed to increasing concentrations (10–100 ng/ml) of HGF for 24 hours in the presence or absence of TGF-β, lysed, and assayed for α-SMA protein expression by immunoblotting. While lower concentrations of HGF did decrease basal and TGF-β–induced α-SMA (data not shown), a concentration of 100 ng/ml significantly inhibited expression of the protein (Figures 4A and 4B).

Inhibition of α-SMA Expression by HGF Is ERK1/2 Dependent

To elucidate the molecular mechanisms responsible for HGF-dependent inhibition of α-SMA, we first assayed the activation status of a variety of signaling molecules after HGF treatment. Exposure of RLE cells to HGF for 5 minutes significantly activated Akt (Ser 473 and Thr 308) and the MAPK family members ERK1/2, JNK, and p38 (Figure 5A). Akt activation continued through 25 minutes of HGF treatment, while ERK1/2, JNK, and p38 phosphorylation diminished slightly. Interestingly, we observed basal activation of p38 in the absence of HGF stimulation, suggesting that other endogenous factors, such as TGF-β, may be responsible for the basal levels of phospho-p38 (35, 36). We next examined the effect of blocking HGF-induced signaling molecules on α-SMA protein levels. RLE cells were exposed to optimal concentrations of pharmacologic inhibitors widely used to block specific signaling pathways. Either a MEK inhibitor (37, 38), an Akt inhibitor (39), or a p38 inhibitor (40, 41) was used for 30 minutes, followed by HGF treatment for 24 hours. The Akt inhibitor did not alter the HGF-mediated suppression of α-SMA protein. However, treatment of RLE cells with the MEK inhibitor, PD98059, significantly antagonized the HGF-mediated decrease in α-SMA (Figure 4B). Interestingly, exposure to the p38 inhibitor before addition of HGF further diminished the expression of α-SMA protein, tempting us to speculate that TGF-βR–dependent α-SMA protein production in these cells may potentially involve p38 activation (Figure 5B). Overall, these results showed that HGF reduces α-SMA protein in an ERK-dependent manner.

HGF Induces Smad7 Expression and Inhibits TGF-β Signaling

In two models of chronic renal disease, administration of HGF expression plasmids resulted in reduction in renal injury (42, 43). HGF has been reported to intercept Smad2/3 translocation to the nucleus in intestinal fibroblasts (44), up-regulate Smad transcriptional corepressor TGIF in mesangial cells (45), and increase expression of the transcriptional corepressor SnoN in tubular epithelial cells (46). To date, no report exists on effects of HGF on myofibroblast generation from lung epithelial cells.

To explore whether HGF intersects the TGF-β signaling pathway in rat lung epithelial cells, we assessed Smad7 protein expression in the cells. RLE cells were treated with HGF, total protein was extracted, and Western blots were probed for Smad7 protein. Smad7 levels began to increase at 3 hours after HGF stimulation, and remained elevated through 24 hours (Figure 6A).

To determine if HGF-dependent Smad7 induction involved MEK-ERK activation, RLE cells were pretreated with the MEK inhibitor for 30 minutes before HGF stimulation. The MEK inhibitor essentially blocked the HGF-dependent increase in Smad7 protein without any effect on basal Smad7 levels (Figure 6B). Given that the MEK inhibitor appreciably blocked the HGF-dependent induction of Smad7, we also wanted to investigate the effects of the inhibitor on pSmad2 and pSmad3 levels. In Figure 6B, we show that pSmad2 and pSmad3 profiles were different in response to HGF and the MEK inhibitor. There was a small increase in pSmad2 in response to HGF, peaking at 1 hour after stimulation, which was inhibited by the MEK inhibitor. pSmad3 levels peaked at 3 hours after treatment with HGF and were refractory to treatment with the MEK inhibitor. Thus, the decrease observed for pSmad2 was not due to unequal loading and the pSmad3 level was useful as a loading control. Since all three proteins—Smad2, Smad3, and Smad4—have been shown to be required for Smad7 transcription (47), inhibition of pSmad2 by the MEK inhibitor is in agreement with the reduced levels of Smad7 observed in Figure 6B.

In contrast to SnoN protein up-regulation by HGF in kidney epithelial cells (7), no effect on SnoN protein in rat alveolar epithelial cells was observed after HGF treatment (data not shown).

HGF Induces Nuclear Export of Smad7 and Smurf1

Smad7-dependent inhibition of TGF-β signaling occurs at the TGF-β receptor. Therefore, to assess cellular localization of Smad7 in the presence or absence of HGF, we performed immunocytochemistry in RLE cells with antibodies to Smad7 and a cell membrane–associated protein, E-cadherin. The majority of Smad7 was found to be localized to the nucleus in untreated RLE cells, as has been previously noted in many cell types (48). However, after 3 hours of HGF treatment, nuclear Smad7 decreased and cytoplasmic Smad7 increased (Figure 7). Quantitative analysis of 100 cells averaged across five fields per treatment revealed a significant shift of Smad7 out of the nuclear region after treatment with HGF. It should be noted that the average values shown in a bar graph format include cells that appeared to have lost all nuclear Smad7 and others that still retained an appreciable amount. These results show that HGF treatment promotes export of Smad7 from the nucleus, which is required for its inhibitory functions.

Smad ubiquitin regulatory factor 1 (Smurf1) exports Smad7 from the nucleus to the cytoplasmic region, where it can interact with the type I TGF-β receptor, leading to inhibition of TGF-β signaling (49, 50). Because we observed a change in the distribution of Smad7, we asked whether Smurf1 also displayed a change in localization after HGF exposure. RLE cells were treated with the MEK inhibitor for 30 minutes and HGF for 3 hours, fixed, and observed with confocal immunofluorescent microscopy. Intensity analysis of over 300 cells was conducted in MetaMorph using DRAQ5 labeling of the nuclei to define the nuclear region. In untreated cells, both Smad7 and Smurf1 were predominantly in the nucleus (Figure 8). After 3 hours of exposure to HGF, analysis of nuclear intensity of both Smad7 and Smurf1 was similarly decreased. However, pretreatment with the MEK inhibitor blocked the HGF-dependent nuclear export of Smad7 and Smurf. Exposure to HGF and the inhibitor resulted in maximum nuclear expression of both Smad7 and Smurf1. Linescan profiles (Figure 8) through the center of the nuclei underscore the results of the image analysis. Nuclear linescan regions are identified by the blue line, Smad7 by the green line, and Smurf1 by the red line. In the control cells, Smad7 and Smurf1 expression is predominantly in the nuclear region. After treatment with HGF, the nuclear expression of Smad7 and Smurf1 drops dramatically. After treatment with the MEK inhibitor and HGF, the majority of expression remains in the nuclear region. These results suggest that HGF regulates the Smad7-Smurf1 cellular localization in an ERK-dependent fashion.

Knockdown of Smad7 Relieves HGF-Mediated Inhibition of α-SMA Protein Levels

To establish that the HGF effect was dependent on Smad7 up-regulation, we examined the effect of Smad7-specific siRNA on α-SMA protein expression. Cells were either left untransfected or transfected with 10 or 50 nM Smad7 siRNA, treated with 100 ng/ml HGF for 24 hours, lysed, and protein collected for assessment of α-SMA, Smad7, and β-actin protein. As a control for the transfection, we used fluorescently tagged siRNA to label transfected cells in a separate experiment and observed a near 100% transfection efficiency of the protocol that was used in all other experiments (Figure 9A). Cells devoid of Smad7 siRNA and exposed to 100 ng/ml HGF for 24 hours demonstrated the characteristic decrease in α-SMA protein compared with untreated cells (Figure 9A). Increasing concentrations of Smad7 siRNA antagonized the HGF-dependent decrease in α-SMA protein (Figure 8A). Transfection of 50 nM Smad7 siRNA restored the level of α-SMA protein to 60% of the levels in untreated controls (Figure 9A). Transfection of nonspecific siRNA had no effect on Smad7 or α-SMA protein levels (Figure 9B). An overall 60% inhibition of Smad7 expression appreciably blocked the HGF-mediated inhibition of α-SMA protein levels. These results suggested that HGF-dependent down-modulation of α-SMA is regulated, in part, by induction of Smad7 and inhibition of Smad signaling.