Original Paper | Published:

Collaborative activities of macrophage-stimulating protein and transforming growth factor-β1 in induction of epithelial to mesenchymal transition: roles of the RON receptor tyrosine kinase

Abstract

Epithelial to mesenchymal transition (EMT) is a process occurring during embryonic development and cancer progression. Using recepteur d'origine nantais (RON)-expressing epithelial cells as a model, we showed that RON activation causes spindle-shaped morphology with increased cell motilities. These activities resemble those observed in EMT induced by transforming growth factor (TGF)-β1 or by Ras–Raf signaling. By immunofluorescent and Western blot analyses, we found that constitutive RON expression results in diminished expression of E-cadherin, redistribution of β-catenin, reorganization of actin cytoskeleton, and increased expression of vimentin, a mesenchymal filament. RON expression is also essential for TGF-β1-induced expression of α-smooth muscle actin (α-SMA), a specialized mesenchymal marker. In the study of signaling pathways responsible for RON-mediated EMT, it was found that PD98059, a MAP kinase inhibitor, blocks the collaborative activities of RON and TGF-β1 in induction of α-SMA expression and restores epithelial cells to their original morphology. Moreover, we showed that RON expression increases Smad2 gene promoter activities and protein expression, which significantly lowers TGF-β1 threshold for EMT induction. These results suggest that persistent RON expression and activation cause the loss of epithelial phenotypes. These changes, collaborating with TGF-β1 signaling, could play a critical role in epithelial transdifferentiation towards invasiveness and metastasis of certain cancers.

Introduction

Epithelial to mesenchymal transdifferentiation (EMT) is a unique event occurring during embryonic development, wound healing, and tumor progression (Boyer et al., 2000; Savagner, 2001; Thiery, 2002). EMT is a complicated process characterized by the loss of epithelial characteristics and the gain of mesenchymal phenotypes (Boyer et al., 2000; Savagner, 2001). The typical EMT consists of spindle-shaped morphologies with reorganized cytoskeleton, reduced and/or delocalization of E-cadherin from cell junctions, and expression of specific mesenchymal cellular markers such as α-smooth muscle actin (α-SMA) (Boyer et al., 2000; Savagner, 2001). Most invasive and/or metastatic cancers are characterized by partial or complete EMT (Ruiz and Gunthert, 1996; Thiery, 2002). In these tumors, the epithelial phenotype, exhibiting strong cell–cell junction and polarity across the epithelial layer, is replaced by a mesenchymal phenotype with reduced cell–cell interactions, increased cell motilities, and expression of mesenchymal cellular protein markers (Ruiz and Gunthert, 1996; Thiery, 2002). These observations led to the conclusion that the process of EMT is a critical step required for the invasiveness and metastasis of certain epithelial cells in vivo (Ruiz and Gunthert, 1996; Savagner, 2001; Thiery, 2002).

Many growth factors and cytokines induce EMT (Boyer et al., 2000; Thiery, 2002). The typical examples are TGF-β1 (Piek et al., 1999a; Bhowmick et al., 2001) and hepatocyte growth factor/scatter factor (HGF/SF) (de Caestecker et al., 1998; Elliott et al., 2002; Lamorte et al., 2002). TGF-β1 binds to its transmembrane serine/threonine kinase type I and type II receptors (Piek et al., 1999b; Yue and Mulder, 2001). Activation of TGF-β1 signaling mediates partial or complete EMT in culture epithelial cells, such as NMuMG and Madin–Darby canine kidney (MDCK) cells (Piek et al., 1999a; Nicolas et al., 2003). Signaling pathways including Smad, Ras, and MAPK/ERK have been implicated in TGF-β1-induced EMT (Ellenrieder et al., 2001; Grande et al., 2002; Janda et al., 2002; Oft et al., 2002). Studies in vivo using transgenic mice have also demonstrated that blocking TGF-β1 signaling significantly reduces the metastatic potentials of epithelial tumors (Muraoka et al., 2002; Yang et al., 2002), suggesting that TGF-β1-induced EMT contributes to malignant behaviors of epithelial cancer in vivo. Similarly, HGF/SF activates the MET receptor tyrosine kinase that induces EMT in various types of epithelial cells (Thiery and Chopin, 1999). The crosstalk between HGF/SF-induced signaling cascades and TGF-β-Smad2 pathways has been documented (de Caestecker et al., 1998), indicating that the collaboration between MET and TGF-β1 might be important in induction of EMT.

The recepteur d'origine nantais (RON) receptor tyrosine kinase (Ronsin et al., 1993) is a member of the MET proto-oncogene family (Rubin et al., 1993). Mature RON is a 180-kDa heterodimer composed of a 40-kDa α-chain and a 150-kDa transmembrane β-chain with intrinsic tyrosine kinase activity (Ronsin et al., 1993). The ligand for RON has been identified as macrophage-stimulating protein (MSP) (Skeel et al., 1991; Gaudino et al., 1994; Wang et al., 1994a), also known as hepatocyte growth factor-like protein (Han et al., 1991). MSP belongs to the kringle protein family that includes plasminogen, HGF, and others (Gherardi et al., 1997). Activation of RON by MSP stimulates multiple signaling pathways including Ras (Li et al., 1995), PI-3 kinase (Wang et al., 1996a), MAP kinase (Santoro et al., 1998), and NFκB (Zhou et al., 2002). These pathways regulate a variety of cellular activities in different types of cells (Wang et al., 2003). One of the functions observed in epithelial cells upon RON activation is the induction of cell adhesion, dissociation (scattering), migration, and matrix invasion (Medico et al., 1996; Santoro et al., 1996; Wang et al., 1996b). These activities are collectively known as the motile-invasive phenotype (Comoglio et al., 1999), which, to a certain degree, resembles those found in EMT (Boyer et al., 2000; Savagner, 2001; Thiery, 2002). Thus, activated RON has the ability to transduce signals that initiate cell motile machinery leading to cell locomotion.

The present work was to study the role of RON in regulating transdifferentiation of epithelial cell towards mesenchymal phenotypes. MDCK cells and their RON transfectants were used as the model. Our results indicated that constitutive expression and activation of RON in MDCK cells results in cell scattering with spindle-shaped morphology. This change is accompanied by the conversion of epithelial properties toward mesenchymal phenotypes. The collaborative activities of MSP and TGF-β1 on EMT was further demonstrated by RON-mediated expression of Smad2 and α-SMA expression in RON expressing MDCK cells. Thus, by induction of EMT, MSP-induced RON activation, together with TGF-β1-induced signaling, could play a pivotal role in regulating malignant behaviors of certain epithelial cancer cells.

Results

Effect of RON expression and activation on morphological changes and cellular marker expression

MDCK cells were characterized by typical epithelial morphology with tight adherens junctions and expression of epithelial markers such as E-cadherin (Figure 1). Expression of human RON in MDCK cells results in a heterogeneous cell population called RE7 (Wang et al., 1994a), which express relatively high levels of RON (Figure 1B). The morphological differences between MDCK and RE7 cells were clearly visible when cells were cultured in DMEM with 10% FBS (Figure 1a). Upon MSP stimulation, RE7 cells underwent further shape changes with spindle-like morphologies (Figure 1A). The adherens junctions among RE7 cells were clearly reduced. Similarly, TGF-β1 induced spindle-like morphologies in both MDCK and RE7 cells (Figure 1A). The effect of RON-mediated cell motile activities was seen in cell migration assays. The percentages of migrated RE7 cells upon MSP stimulation increased significantly (34.7±7.4%) in comparison to those of unstimulated cells (13.2±3.2%). No changes were observed in control MDCK cells stimulated with (3.2±0.7%) or without MSP (3.3±1.1%). These results, together with those in Figure 1, suggested that persistent expression and activation of RON exert morphological and motile effects on MDCK cells, which resemble, to a certain degree, those induced by TGF-β1.

Figure 1
figure1

Effect of RON expression and activation on morphology and cellular protein expression. (A) MDCK and RE7 cells (4 × 104 cells per dish) were cultured overnight in DMEM with 10% FBS and then stimulated with or without MSP (2 nM) or TGF-β1 (2 ng/ml) for 48 h. Cell morphologies were examined under a phase-contrast microscope and photographed (magnification × 200). (B) Cellular proteins (50 μg per sample) from lysed RE7 or MDCK cells without any stimulation were subjected to 8% SDS–PAGE under reduced conditions. Western blotting was performed using antibodies specific to RON, vimentin, or E-cadherin. Results shown here are from one of three experiments with similar results. CTL, control

To determine if RON expression causes changes in expression of cellular marker proteins, Western blot analyses were performed using antibodies specific to RON, E-cadherin, or vimentin. The results are shown in Figure 1B. In MDCK cells, which did not express RON, high levels of E-cadherin were detected. However, in RE7 cells that express high levels of RON, E-cadherin expression was significantly diminished or lost. This effect was also seen in epithelial AA/C1 cells (Hague et al., 1993). E-cadherin expression was reduced up to 50% after cells were treated with MSP for 48 h (data not shown). We also tested the expression of vimentin, a mesenchymal filament, in MDCK and RE7 cells. Extremely low levels of vimentin were detected in MDCK cells. In contrast, the vimentin expression was significantly increased in RE7 cells. The switches between E-cadherin and vimentin expression suggest that RON expression affects not only morphology, but also gene transcription or protein expression. They also suggest that the persistent RON expression in MDCK cells results in the loss of epithelial characteristics and the gain of mesenchymal phenotypes, which are the features of EMT.

Effect of MSP or TGF-β1 on expression and redistribution of E-cadherin, β-catenin, and actin

The data obtained in Figure 1 prompted us to use immunofluorescent techniques to determine the expression and redistribution of E-cadherin, β-catenin, and actin in MDCK or RE7 cells stimulated with MSP. Results are shown in Figure 2. In MDCK cells, MSP stimulation did not result in any changes in expression or redistribution of E-cadherin, β-catenin, and actin, respectively, consistent with the inability of MSP to induce morphological changes shown in Figure 1A. In contrast, redistribution of β-catenin from membrane to cytosol in RE7 cells stimulated with and without MSP was observed (Figure 2j and 1). Reorganization of the actin cytoskeleton from cortical ring at cell junction into stress fibers was also observed in RE7 cells stimulated with or without MSP (Figure 2p and r). Consistent with those shown in Figure 1B, E-cadherin expression was not detected in RE7 cells (Figure 2d–f).

Figure 2
figure2

E-cadherin expression, β-catenin redistribution, and actin cytoskeleton reorganization in MDCK or RE7 cells stimulated with MSP. MDCK or RE7 cells (4 × 104 cells per dish) were stimulated with 2 nM MSP for 48 h. Cells treated with TGF-β1 were used as positive controls. After stimulation, cells were fixed with the methanol–acetone solution. Immunofluorescent staining was performed using FITC-conjugated antibodies specific to β-catenin, actin, or E-cadherin, and photographed under Nikon Eclipse E600 microscope (magnification × 400). CTL, control. One of three experiments with similar results

To further determine if the levels of β-catenin, actin, and E-cadherin are changed, Western blot analysis was performed. Results are shown in Figure 3. No significant changes were observed in the levels of β-catenin and actin in both MDCK and RE7 cells stimulated with or without MSP or TGF-β1. The only difference observed between MDCK and RE7 cells was the expression of E-cadherin. These results, together with those in Figure 2, suggested that RON-mediated RE7 cell morphological changes are associated with redistribution of β-catenin and reorganization of actin cytoskeleton.

Figure 3
figure3

Effect of RON expression on levels of E-cadherin, β-catenin, and actin expressed in RE7 cells. Cells were stimulated for 48 h as described in Figure 1. Cellular proteins (50 μg per sample) and separated in 10% SDS–PAGE followed by Western blot analysis using antibodies specific to E-cadherin (a), β-catenin (b) or actin (c). CTL, control. T+ M, TGF-β1+MSP. One of three experiments with similar results

Collaborative effect of MSP and TGF-β1 on induction of EMT in RE7 cells

One of the distinct features in complete EMT is the expression of a specific mesenchymal cellular marker α-SMA. To study if RON-mediated EMT is accompanied by α-SMA expression, MDCK and RE7 cells were stimulated with MSP alone or in combination with TGF-β1 for 7 days. Results are shown in Figure 4. MSP or TGF-β1 alone or in combination did not induce α-SMA expression in MDCK cells (data not shown). Also, MSP had no stimulatory effect on α-SMA expression in RE7 cells. Interestingly, in RE7 cells with spindle-like morphologies, α-SMA was readily induced by TGF-β1 (Figure 4b). Western blot analysis confirmed that α-SMA is induced by TGF-β1 but not by MSP in RE7 cells (Figure 5a). Interestingly, the TGF-β1-induced α-SMA expression was slightly reduced when MSP was included in RE7 cell cultures. The kinetic expression of α-SMA in RE7 cells was shown in Figure 5b. Upon TGF-β1 stimulation, the α-SMA protein was detected as early as day 4. The expression peaked from day 10 to 12 and then gradually reduced at day 14. In contrast, no transient or delayed α-SMA expression was detected in MDCK cells stimulated with TGF-β1 alone or in combination with MSP under similar conditions (data not shown). Moreover, it was found that after a 7-day stimulation, the minimal amount of TGF-β1 required for inducing α-SMA expression is 0.1 ng/ml. These results indicate that even though MSP is unable to induce α-SMA expression, RON expression is required for TGF-β1-induced α-SMA expression, which significantly lowers the TGF-β1 threshold for EMT induction.

Figure 4
figure4

Effect of MSP and TGF-β1 on morphological changes and α-SMA expression in RE7 cells. RE7 cells (1 × 104 cells per dish) were stimulated for 7 days with MSP (2 nM), TGF-β1 (2 ng/ml) or both. (a) Morphological changes were examined under the microscope and photographed (magnification × 200). (b) α-SMA expression in fixed cells was determined by immunofluorescent techniques using antibodies specific to α-SMA and photographed under the microscope (magnification × 400). One of three experiments with similar results

Figure 5
figure5

Expression of α-SMA in RE7 cells stimulated by TGF-β1 or in combination with MSP. MDCK or RE7 cells were stimulated with MSP (2 nM), TGF-β1 (2 ng/ml), or both at different time intervals. Expression of α-SMA was determined by Western blotting. (a) α-SMA expression in RE7 cells stimulated by TGF-β1 for 7 days. (b) Time courses of TGF-β1-induced α-SMA expression. Both membranes were reprobed for β-actin as the loading control. T+M, TGF-β1+MSP. One of three experiments with similar results

Since α-SMA expression is associated with long-lasting EMT (Oft et al., 2002), we wanted to determine if RON expression stabilizes TGF-β1-induced EMT. To this end, MDCK or RE7 cells were stimulated with TGF-β1 for 7 days and then cultured in TGF-β1-free medium for an additional 4 days. Morphological changes were monitored during the entire period. The results are shown in Figure 6. After TGF-β1 stimulation, MDCK cells gradually lost their epithelial properties and displayed spindle-shaped morphologies. However, these changes were fully reversed to their original cell morphologies after withdrawal of TGF-β1 (Figure 6a). Immunofluorescent analysis of E-cadherin expression confirmed the reassembly of adherens junctions in TGF-β1-removed MDCK cells (Figure 6b). In contrast, RE7 cells continued to display the spindle-shaped morphology, which persisted up to 11 days (Figure 6c). Studying the kinetic expression of α-SMA in RE7 cells after removal of TGF-β1 showed that α-SMA continues to be expressed after removal of TGF-β1, with the expression curve similar to those shown in Figure 5b. These results demonstrated that RON expression is required for TGF-β1-induced long-lasting EMT in MDCK cells.

Figure 6
figure6

RON-dependent long-lasting morphological changes in RE7 cells stimulated by TGF-β1. MDCK or RE7 cells were stimulated by TGF-β1 (2 ng/ml) for 7 days and washed with serum-free medium. Cells were then cultured in fresh TGF-β1-free medium for additional 4 days. Cell morphologies (a, c) and E-cadherin expression (b) were monitored at days 7, 9, and 11 under the microscope or by immunofluorescent staining, respectively. MDCK or RE7 cells cultured in medium without any stimulation for 11 days were used as controls (CTL). The data shown here are from one of three experiments with similar results

Effect of the MAP kinase inhibitor PD98059 on RON-mediated morphological changes

The collaborative activities observed in Figure 6 prompted us to determine potential signaling pathways involved in RON and TGF-β1-induced EMT. Various chemical inhibitors were used. The MAPK/ERK signaling pathway was the focus because of its critical role in EMT (Ellenrieder et al., 2001; Janda et al., 2002). After stimulation of RE7 cells with MSP or TGF-β1 at different time intervals, phosphorylation of ERK1/2 was determined by Western blot analysis using antibodies specific to phosphorylated ERK1/2. The results are shown in Figure 7. Both MSP and TGF-β1 induced ERK1/2 phosphorylation in a time-dependent manner with different kinetics (Figure 7a and b). In general, ERK1/2 phosphorylation induced by MSP was much earlier and higher than that by TGF-β1. The synergistic activities were also determined. As shown in Figure 7c, the levels of phosphor-p42 were increased after cells were stimulated for 60 min. The effect of MSP or TGF-β1 on ERK1/2 phosphorylation was also studied using PD98059, a specific MAP kinase inhibitor. As shown in Figure 7c, the ERK1/2 phosphorylation was completely inhibited by 50 μ M of PD98059.

Figure 7
figure7

MSP or TGF-β1-induced phosphorylation of ERK1/2 and its inhibition by PD98059. RE7 cells (2 × 106 cells per dish) were serum-starved overnight and then stimulated with MSP (a) or TGF-β1 (b) at different time intervals. Cells without stimulation were used as controls. RE7 cells were also stimulated for 60 min with MSP, TGF-β1, or both in the presence or absence of PD98059 (50 μ M) after initiation of culture (c). After treatment, cellular proteins (50 μg/sample) were separated in 10% SDS–PAGE and blotted with antibodies to phosphorylated p42 and p44. The same membranes were also reprobed with the antibodies specific to p42. It is noticed that the increase of phosphorylated p42 results in the decrease of nonphosphorylated p42 detected by the antibody to p42. β-Actin was used as the loading control. One of the three experiments with similar results

To determine if the inhibition of ERK1/2 phosphorylation by PD98059 correlates with MSP and TGF-β1-induced EMT, the effect of PD98059 on MSP or TGF-β1-induced morphological changes in RE7 cells was studied. As expected, MSP or TGF-β1 induced typical morphological changes within a week. The addition of PD98059 completely restored the polarized epithelial phenotypes (Figure 8e–g). The earliest time showing the inhibitory effect of PD98059 was seen 48 h after cells were treated. PD98059 also prevented collaborative effects of MSP and TGF-β1 in induction of spindle morphologies although its effect is not complete (Figure 8h). Consistent with these results, MSP-induced migration of RE7 cells (31.2±5.9%) was partially inhibited by PD98059 (19.8±3.6%) with control cells at 12.4±3.3%, suggesting that signal pathway(s) other than ERK/MAPK might also be involved in the dispersal of RE7 cells after stimulation with MSP and TGF-β1. However, in cells treated with different concentrations of wortmannin (an inhibitor of PI-3 kinase) or SB203580 (an inhibitor of P38/MAPK), no significant effect on morphological changes or E-cadherin expression were observed (data not shown).

Figure 8
figure8

Inhibitory effect of PD98059 on morphological changes of RE7 cells induced by MSP, TGF-β1, or their combination. RE7 cells (4 × 104 cells per dish) were stimulated for 7 days with MSP (2 nM), TGF-β1 (2 ng/ml), or both in the presence or absence of 50 μ M PD98059, respectively. Morphologic changes were monitored daily and photographed. The photos shown here are from cells incubated for 7 days (magnification × 200). CTL, control. One of three experiments with similar results

Immunofluorescent techniques were used to determine if PD98059 treatment results in the reverse of the β-catenin distribution and actin cytoskeleton organization induced by MSP or TGF-β1. Treatment of RE7 cells with PD98059 completely prevented the redistribution of β-catenin induced by MSP, TGF-β1, or their combination (Figure 9a–h). Similarly, PD98059 blocked the reorganization of the actin cytoskeleton stimulated by MSP, TGF-β1 or their combination (Figure 9i–p). The restored β-catenin distribution and actin organization are indistinguishable to those observed in nonstimulated MDCK cells (data not shown). Moreover, PD98059 treatment led to re-expression of E-cadherin in RE7 cells stimulated with TGF-β1 alone or with MSP (Figure 9q–x). The effect of wortmannin or SB203580 was also tested. No changes were observed (data not shown). These results, together with those in Figure 8, suggest that MAPK/ERK pathways are involved in MSP and TGF-β1-induced EMT.

Figure 9
figure9

Effect of PD98059 on redistribution of β-catenin, E-cadherin and reorganization of actin cytoskeleton in RE7 cells. RE7 cells were stimulated for 48 h with MSP (2 nM), TGF-β1 (2 ng/ml), or both in the presence or absence of PD98059 (50 μ M). After stimulation, the fixed cells were used to detect β-catenin, E-cadherin, or actin expression by immunofluorescent techniques and photographed (magnification × 400). One of the three experiments with similar results. CTL, control

Effect of PD98059 on expression of cellular marker proteins in RE7 cells

To determine if the effect of PD98059 on cell morphology correlates with expression of cellular marker proteins, Western blot analyses were performed to detect E-cadherin, vimentin, and α-SMA in RE7 cells. As shown in Figure 10a, E-cadherin was expressed in MDCK cells but not in RE7 cells. Stimulation of RE7 cells with MSP, TGF-β1, or both, did not induce E-cadherin expression. Similarly, PD98059 alone had no effect on E-cadherin expression. However, when RE7 cells were stimulated with TGF-β1 in the presence of PD98059, the expression of E-cadherin was detected. This effect was also seen when cells were stimulated with TGF-β1 and MSP, consistent with those shown in Figure 9v and x. The effect of SB203580 or wortmannin on E-cadherin expression was not observed (data not shown).

Figure 10
figure10

Effect of PD98059 on expression of E-cadherin, α-SMA, or vimentin in RE7 cells. MDCK or RE7 cells (4 × 104 cells per dish) were stimulated as described in Figure 9. PD98059 (50 μ M), SB203580 (5 μ M) or wortmannin (100 nM) was added simultaneously after initiation of cultures, respectively. Western blotting was performed using cellular proteins (50 μg per sample) after 10% SDS–PAGE. Expression of E-cadherin (a), vimentin (b), or α-SMA (c) was determined using individual antibodies, respectively. Membranes were also reprobed for β-actin as the loading control. T+M, TGF-β1+MSP. PD, PD98059; SB, SB203580; and WT, wortmannin. One of three experiments with similar results

Next, the effect of PD98059 on vimentin expression was studied (Figure 10b). In MDCK cells, the extremely low levels of vimentin were detected, consistent with those reported previously (Cano et al., 2000). In contrast, high levels of vimentin were detected in RE7 cells even without MSP stimulation. After TGF-β1 stimulation, the levels of vimentin did not change very much in both cells. PD98059 alone had no effect on vimentin expression in RE7 cells. However, in combination with TGF-β1, MSP or TGF-β1 plus MSP, PD98059 inhibits vimentin expression at various degrees. These results indicate that PD98059 has the ability to downregulate vimentin expression in RE7 cells in the presence of TGF-β1, MSP or both.

The effect of PD98059 on α-SMA expression was also studied (Figure 10c). In RE7 cells, expression of α-SMA was induced by TGF-β1 but not by MSP. Interestingly, TGF-β1-induced α-SMA expression was reduced in the presence of MSP. When PD98059 was added to RE7 cells stimulated with TGF-β1, the amount of α-SMA was significantly reduced with only about 10% of α-SMA remaining. However, the synergistic effect of MSP and PD98059 on TGF-β1-induced α-SMA was not observed. The effect of SB203580 or wortmannin on α-SMA expression was also studied. No significant inhibitory activities were observed. These results, together with those in Figure 10b, suggest that although PD98059 regulates vimentin and α-SMA expression in RE7 cells stimulated by TGF-β1, MSP or both, the underlying mechanisms are complicated.

Effect of RON expression or activation on Smad2 expression, transcription, phosphorylation, and nuclear translocation

Collaborative activities between MSP and TGF-β1 in EMT raised the possibility that the functions of RON might be channeled through TGF-β signaling. To this end, the effect of RON expression on Smad2/3 expression and activation was studied. Western blot analyses showed that the levels of Smad2/3 were significantly increased in RE7 cells in comparison to those in MDCK cells (Figure 11a). We also tested if RON activation increases the Smad2 promoter activities. The luciferase reporter assay was performed using RE7 cells transiently transfected with the pGL3-1a vector containing the human Smad2 promoter sequence (Takenoshita et al., 1998). As shown in Figure 11b, spontaneously elevated luciferase activities driven by the Smad2 promoter were observed in nonstimulated RE7 cells. Upon MSP stimulation, the luciferase activities were further increased. No luciferase activities were seen when pGL3-1a was transfected into MDCK cells. We then tested if the increased Smad2 expression increases TGF-β1-induced transcription activities of a TGF-β-responsive plasminogen activator inhibitor type (PAI)-1 luciferase gene reporter construct (Oft et al., 2002). As shown in Figure 11c, luciferase activities driven by the PAI-1 promoter in RE7 cells upon TGF-β1 stimulation were much higher than those in MDCK cells. To further confirm these results, a dominant-negative (Δ) Smad2 (Oft et al., 2002) was expressed in RE7 cells. The results showed that TGF-β1-induced PAI-1 promoter activities were significantly reduced in the presence of ΔSmad2, indicating that the increased Smad2 expression is responsible for the enhanced TGFβ1 signaling. We also studied if Smad2 overexpression enhances the sensitivity of MDCK cells to TGF-β1. It was found that the levels of TGF-β1-induced PAI-1 luciferase activities in MDCK cells transfected with human Smad2 cDNA is five times higher than those in cells without Smad2 transfection. These results demonstrate that the increased Smad2 expression is accompanied with the increased sensitivity of MDCK cells in response to TGF-β1 stimulation.

Figure 11
figure11

Effect of RON expression and activation on Smad2 expression, transcription, phosphorylation, and nuclear translocation. (a) Expression of Smad2 and Smad3 in MDCK and RE7 cells. Cellular proteins (50 μg/sample/lane) were used in Western blot analysis. (b) MSP-induced Smad2 promoter activities in RE7 cells. Transient transfection of RE7 or MDCK cells with the pGL3-1a vector containing the Smad2 promoter fragment was performed as detailed (Zhou et al., 2002). Luciferase activities were measured in cell lysates after cells were stimulated with or without 2 nM MSP for 24 h. s.d. is less than 15%. (c). TGF-β1-induced PAI-1 promoter activities in RE7 cells. Cells were transfected with the 3TPLux vector containing the PAI-1 promoter alone or with the expression vector containing ΔSmad2. After TGF-β1 stimulation, luciferase activities were measured. s.d. is less than 15%. (d) Effect of MSP on Smad2 phosphorylation. RE7 cells were serum-starved overnight and then stimulated with 2 nM MSP at different time intervals. Cells treated with TGF-β1 for 1 h were used as the positive control. Phosphorylated Smad2 was determined in Western blot analysis. (e) Effect of MSP on Smad2 nuclear translocation. Serum-starved RE7 cells (6 × 106 cells per dish) are stimulated with 2 nM MSP at different time intervals. Cells stimulated with TGF-β1 were used as the positive control. Smad2 and Smad3 in nuclei lysates were detected in Western blot analysis. One of three experiments with similar results

The effect of RON on Smad2 phosphorylation and nuclear translocation was also studied. Results in Figure 11d showed that TGF-β1 induces Smad2 phosphorylation. However, no Smad2 phosphorylation was observed when MSP was added to cell cultures. Also, MSP had no effect on Smad2 nuclear translocation (Figure 11e). These results suggested that RON activation has no ability to stimulate Smad2 activities.

Discussion

The purposes of this study were to determine the roles of RON in mediating trans-differentiation of epithelial cells toward mesenchymal phenotypes. Our results demonstrated that constitutive expression of RON in MDCK cells results in alterations of cell morphology with increased motile-invasive activities. These changes are characterized by the loss of epithelial markers such as adherent protein E-cadherin and the gain of mesenchymal phenotypes including the increased expression of vimentin. We found that RON expression is required for the induction of the specialized mesenchymal marker α-SMA expression during the process of TGF-β1-induced EMT. Inhibition of MAP kinase pathways by PD98059 blocks MSP and TGF-β1-induced morphological changes and suppresses α-SMA expression. Moreover, we showed that constitutive RON expression increases the Smad2 gene transcription and protein expression, which are essential for TGF-β1 signaling. These results, together with those published previously (Medico et al., 1996; Zhou et al., 2003), suggest that persistent expression and activation of RON cause epithelial cells to undergo a profound phenotypic change. This process resembles, to a certain degree, EMT, which is characterized by the loss of epithelial properties and the gain of mesenchymal phenotypes (Boyer et al., 2000; Savagner, 2001; Thiery, 2002). Thus, the collaborative activities of RON and TGF-β1 could be a driving force that regulates invasive and metastatic potentials of certain epithelial cancers in vivo.

The effects of RON on cell morphologies and motile activities have been reported previously (Medico et al., 1996; Santoro et al., 1996; Wang et al., 1996b). It has been shown that RON activation causes cell dissociation, migration, and matrix invasion in different types of epithelial cells (Medico et al., 1996; Santoro et al., 1996, 1998; Wang et al., 1996b; Maggiora et al., 1998; Zhou et al., 2003). The results from our current studies (Figure 1) demonstrated that RON expression has a profound morphological effect on MDCK cells. First, RE7 cells displayed elongated and spindle-shaped morphologies when cultured in DMEM with 10% FBS. Second, the adherens junctions among cells are significantly reduced, suggesting proteins involved in the junction formation are dysfunctional. Third, the RON-expressing cells displayed increased motile activities. These activities were further enhanced when cells were stimulated with MSP. In these cases, cells had undergone further elongation and scattering. The tightly formed cell–cell adjunctions were almost completely diminished. The redistribution of β-catenin and the reorganization of the actin cytoskeleton were clearly observed (Figure 2). In comparison with EMT induced by TGF-β1, it was noticed that RON-mediated morphological changes, together with cellular marker protein expression, are indistinguishable to those induced by TGF-β1. Thus, it is likely that RON activation directs a cellular program(s) that resembles the one induced by TGF-β1.

Analysis of cellular marker protein expression demonstrated that the effect of RON on MDCK cells is not limited to cell morphology. RON expression affects the transcriptions of certain genes that determine cellular phenotypes. As shown in Figure 1B, the persistent RON expression results in the loss of epithelial cellular marker E-cadherin and the gain of mesenchymal filament vimentin. It is noticed that vimentin is occasionally expressed in certain epithelial cells such as MDCK (Cano et al., 2000); however, increased expression of vimentin is associated with EMT (Gilles et al., 2003; Grille et al., 2003). Thus, the relevance of vimentin as a mesenchymal marker should be judged carefully.

The importance of E-cadherin in maintaining epithelial phenotype is well documented (Vleminckx and Kemler, 1999). The loss of E-cadherin is also a key event occurring during EMT (Thiery, 2002). Moreover, the diminished E-cadherin expression is a hallmark observed in many malignant cancers with invasive and metastatic potentials (Beavon, 2000; Cavallaro and Christofori, 2001). The finding that E-cadherin is expressed in MDCK but not in RE7 cells strongly suggests that the transition of epithelial towards mesenchymal phenotypes occurs in RON-expressing epithelial cells. Currently, it is unknown how RON expression results in the loss of E-cadherin expression. Different mechanisms might be involved. It has recently been reported that the transcription of the E-cadherin gene is inhibited by transcription repressors such as Snail 1 and SIP1 (Batlle et al., 2000; Cano et al., 2000; Comijn et al., 2001). We speculate that RON might transduce signals that activate Snail 1 or other transcription repressors leading to the inhibition of E-cadherin expression. Thus, determining the effect of RON on these transcription repressors with regard to E-cadherin expression will be worthy to try.

The collaborative activities between receptor tyrosine kinases and the members of TGF super-family in induction of EMT have been reported (Grande et al., 2002). For example, the MET and EGF receptors have been shown to cooperate with the TGF-β1 signaling to facilitate the process of EMT in several epithelial cells (de Caestecker et al., 1998; Grande et al., 2002). Since RON expression leads to EMT in MDCK cells, it is reasonable to think that RON collaborates with TGF-β1 in regulating EMT. Our results shown in Figures 4, 5 and 6 demonstrate that this is the case. Two important findings were obtained from these experiments. One is the expression of α-SMA, the specific mesenchymal cellular marker (Serini and Gabbiani, 1999). α-SMA expression is a hallmark that defines the changes of epithelial cells toward mesenchymal phenotypes (Masszi et al., 2003). α-SMA is expressed exclusively in mesenchymal cells (Serini and Gabbiani, 1999). It is known that TGF-β1 alone does not induce α-SMA expression in epithelial cells such as squamous carcinoma cells (Oft et al., 2002). Additional signals such as increased RAS activities are required for TGF-β1 to induce α-SMA expression (Fujimoto et al., 2001; Janda et al., 2002; Oft et al., 2002). The results presented in Figures 4 and 5 demonstrated that TGF-β1, MSP, or their combination fail to induce α-SMA expression in MDCK cells. Similarly, MSP alone in RE7 cells did not induce α-SMA expression. The induction of α-SMA expression was only seen in TGF-β1 stimulated RE7 cells. Thus, RON expression is required for α-SMA expression during TGF-β1-induced EMT. The interesting thing is that α-SMA expression by RE7 cells is moderately reduced in the presence of MSP. This could suggest that further activation of RON generates negative-feedback signals that modulate α-SMA expression. Thus, α-SMA expression during TGF-β1-induced EMT is a tightly regulated process requiring balanced and adequate RON expression and activation. Another important finding is the requirement of RON in sustaining EMT in TGF-β1-stimulated RE7 cells. As reported, spindle-like morphologies induced by most growth factors or cytokines are fully reversible within a short period after factor withdrawal. The stabilized EMT was only documented in cells with persistent oncogenic RAS signaling and autocrine TGF-β1 production (Oft et al., 1996; Lehmann et al., 2000). The data in Figure 6 showed that RON expression is critical in maintaining the long-lasting EMT upon TGF-β1 stimulation. In RE7 cells, persistent morphological changes and cellular marker expression exist even when TGF-β1 is removed from cell cultures. Thus, by stabilization of EMT, RON, together with TGF-β1, could facilitate differentiated epithelial cells toward migration and invasion.

At present, the precise signaling pathways that transduce RON signals leading to EMT are largely unknown. However, the results shown in Figures 7 and 10 provide important clues indicating that the activation of MAPK/ERK pathways might be critical. As shown in Figure 7, RON activation leads to increased phosphorylation of ERK1/2, the constitutive components of MAP kinase pathways. Blocking ERK1/2 activation by PD98059 prevents the MSP-induced EMT. The evidence supporting this notion include: (a) PD98059 blocks RON-mediated ERK1/2 phosphorylation in RE7 cells. This inhibitory effect was also seen in TGF-β1-induced ERK1/2 phosphorylation; (b) PD98059 restores altered morphologies to their original compact epithelial phenotypes even in the presence of MSP and TGF-β1; and (c) PD98059 prevents TGF-β1-induced α-SMA expression and restores E-cadherin expression in RE7 cells. In contrast, inhibition of RON-mediated activation of PI-3 kinase or p38 by wortmannin or SB203580, respectively, did not prevent the MSP or TGF-β1-induced morphological changes and cellular marker protein expression. Considering these data, it is concluded that the activation of the MAPK/ERK pathway is essential in MSP-induced EMT, although other signaling pathways may also be involved.

The findings that RON expression simultaneously increases Smad2/3 expression and Smad2 promoter activities are unexpected. Smad2 and Smad3 are critical components of TGF-β signaling pathways (Heldin et al., 1997; Yue and Mulder, 2001). Upon phosphorylation and nuclear translocation, Smad2 and Smad3 function as transcription factors that control a variety of gene expression (Yue and Mulder, 2001). HGF and EGF have been shown to activate Smad2 with phosphorylation and nuclear translocation (de Caestecker et al., 1998; Grande et al., 2002). Blocking Smad2 and Smad3 activities with the dominant-negative mutants severely impaired the TGF-β1-induced cellular functions including EMT (Oft et al., 2002). The data from our studies indicate that persistent RON expression result in increased Smad2/3 expression. These activities could represent a novel mechanism in which RON collaborates with TGF-β1 signaling. Thus, it is likely that by increasing Smad2 and Smad3 expression, RON-expressing epithelial cells could exert increased responsiveness upon TGF-β1 stimulation. The results shown in Figure 11c indicate that this might be the case. Thus, the collaboration between RON and TGFβ1 facilitates the complete process of EMT and promotes the invasive and metastatic potentials of malignant epithelial cancers.

Materials and methods

Cells and reagents

MDCK and RE7 cells were used as previously described (Wang et al., 1994a). MSP was provided by Dr EJ Leonard (National Cancer Institute, Frederick, MD, USA). Human TGF-β1 was from R&D Systems (Minneapolis, MN, USA). Rabbit IgG antibodies to human RON were used as described previously (Wang et al., 1994a). Goat IgG antibodies to actin or p-ERK were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Mouse mAb to β-catenin, E-cadherin, Smad2, or pan-ERK were from Transduction Laboratories (San Diego, CA, USA). Mouse mAb to vimentin and α-SMA were from Sigma Co (St Louis, MO, USA). Rabbit IgG antibodies to phosphorylated serine at residues 465 and 467 of human Smad2 was from Upstate Biotechnology (Lake Placid, NY, USA). PD98059, SB203580, and wortmannin were from CalBiochem (San Diego, CA, USA). Luciferase reporter vectors containing a human Smad2 gene promoter fragment (pGL3-1a) or a human PAI-1 gene promoter sequence (3TPLux) were from Dr K Hagiwara (Tohoku University, Sendai, Japan) or Dr Byron Hann (University of California at San Francisco [UCSF], CA, USA), respectively. The expression vector containing ΔSmad2 was from Dr Allan Balmain (UCSF, CA, USA). ECL reagents were from Amersham-Pharmacia (Arlington Heights, IL, USA).

Immunofluorescent staining

The assay was performed as described previously (Oft et al., 2002). Primary antibodies against E-cadherin, β-catenin, actin, or α-SMA, respectively, were added followed by secondary antibodies conjugated with FITC. Cells incubated with normal rabbit IgG were used as controls. All slides were examined and photographed under a Nikon Eclipse E600 upright microscope equipped with fluorescent devices.

Preparation of nuclear extracts

Nuclear extracts were prepared according to a method previously described (Oft et al., 2002). Briefly, cells (6 × 106 cells/dish) were treated with MSP, TGF-β1, or both, for various times and then suspended in 0.4 ml of buffer A (Oft et al., 2002). After 10 min at 4°C, NP-40 was added to reach a 0.5% concentration. Nuclei were collected by centrifugation, resuspended in 100 μl of buffer A supplemented with 20% glycerol and 0.4 M KCl and stored at −80°C or used immediately.

Western blot analysis

The methods were performed as described previously (Wang et al., 1994a). Individual proteins were detected with corresponding antibodies followed by second antibodies conjugated with HRP. The reaction was developed with ECL reagents, recorded on film, and quantitated using the Bio-Rad densitometry software.

Assays for DNA transfection and luciferase activity

These methods were performed as previously described (Zhou et al., 2003). Briefly, MDCK or RE7 cells (5 × 104 cells per dish) were transfected with the pGL3 reporter vector containing Smad2 or PAI-1 promoter fragments. After incubation for 24 h, transfected cells were treated with 2 nM MSP or 2 ng/ml of TGF-β1 for an additional 24 h. Luciferase activities were determined by a Turner Designs Luminometer TD20/20 (Promega). The results were expressed as the relative luciferase activities after normalization with control cells.

Cell migration assay

MSP-induced migration of MDCK or RE7 cells was determined using a multiwell migration chamber as previously described (Wang et al., 1994a). Migrated cells were counted in three randomly selected areas. Results are expressed as the percentage of input cells that migrated.

References

  1. Batlle E, Sancho E, Franci C, Dominguez D, Monfar M, Baulida J and Garcia De Herreros A . (2000). Nat. Cell Biol., 2, 84–89.

  2. Beavon IR . (2000). Eur. J Cancer, 36, 1607–1620.

  3. Bhowmick NA, Ghiassi M, Bakin A, Aakre M, Lundquist CA, Engel ME, Arteaga CL and Moses HL . (2001). Mol. Biol. Cell., 12, 27–36.

  4. Boyer B, Valles AM and Edme N . (2000). Biochem. Pharmacol., 60, 1091–1099.

  5. Cano A, Perez-Moreno MA, Rodrigo I, Locascio A, Blanco MJ, del Barrio MG, Portillo F and Nieto MA . (2000). Nat. Cell Biol., 2, 76–83.

  6. Cavallaro U and Christofori G . (2001). Biochim. Biophys. Acta, 1552, 39–45.

  7. Comijn J, Berx G, Vermassen P, Verschueren K, van Grunsven L, Bruyneel E, Mareel M, Huylebroeck D and van Roy F . (2001). Mol. Cell, 7, 1267–1278.

  8. Comoglio PM, Tamagnone L and Boccaccio C . (1999). Exp. Cell Res., 253, 88–99.

  9. de Caestecker MP, Parks WT, Frank CJ, Castagnino P, Bottaro DP, Roberts AB and Lechleider RJ . (1998). Genes Dev., 12, 1587–1592.

  10. Ellenrieder V, Hendler SF, Boeck W, Seufferlein T, Menke A, Ruhland C, Adler G and Gress TM . (2001). Cancer Res., 61, 4222–4228.

  11. Elliott BE, Hung WL, Boag AH and Tuck AB . (2002). Can. J. Physiol. Pharmacol., 80, 91–102.

  12. Fujimoto K, Sheng H, Shao J and Beauchamp RD . (2001). Exp. Cell Res., 266, 239–249.

  13. Gaudino G, Follenzi A, Naldini L, Collesi C, Santoro M, Gallo KA, Godowski PJ and Comogio PM . (1994). EMBO. J., 13, 3524–3532.

  14. Gherardi E, Gonzalez Manzano R, Cottage A, Hawker K and Aparicio S . (1997). Ciba Found. Symp., 212, 24–35.

  15. Gilles C, Polette M, Mestdagt M, Nawrocki-Raby B, Ruggeri P, Birembaut P and Foidart JM . (2003). Cancer Res., 63, 2658–2664.

  16. Grande M, Franzen A, Karlsson JO, Ericson LE, Heldin NE and Nilsson M . (2002). J. Cell. Sci., 115, 4227–4236.

  17. Grille SJ, Bellacosa A, Upson J, Klein-Szanto AJ, van Roy F, Lee-Kwon W, Donowitz M, Tsichlis PN and Larue L . (2003). Cancer Res., 63, 2172–2178.

  18. Hague A, Manning AM, Hanlon KA, Huschtscha LI, Hart D and Paraskeva C . (1993). Int. J. Cancer, 55, 498–505.

  19. Han S, Stuart LA and Degen SJF . (1991). Biochemistry, 30, 9768–9780.

  20. Heldin CH, Miyazono K and ten Dijke P . (1997). Nature, 390, 465–471.

  21. Janda E, Lehmann K, Killisch I, Jechlinger M, Herzig M, Downward J, Beug H and Grunert S . (2002). J. Cell Biol., 156, 299–313.

  22. Lamorte L, Royal I, Naujokas M and Park M . (2002). Mol. Biol. Cell, 13, 1449–1461.

  23. Lehmann K, Janda E, Pierreux CE, Rytomaa M, Schulze A, McMahon M, Hill CS, Beug H and Downward J . (2000). Genes Dev., 14, 2610–2622.

  24. Li BQ, Wang MH, Kung HF, Ronsin C, Breathnach R, Leonard EJ and Kamata T . (1995). Biochem. Biophys. Res. Commun., 216, 110–118.

  25. Maggiora P, Marchio S, Stella MC, Giai M, Belfiore A, De Bortoli M, Di Renzo MF, Costantino A, Sismondi P and Comoglio PM . (1998). Oncogene, 16, 2927–2933.

  26. Masszi A, Di Ciano C, Sirokmany G, Arthur WT, Rotstein OD, Wang J, McCulloch CA, Rosivall L, Mucsi I and Kapus A . (2003). Am. J. Physiol. Renal Physiol., 284, 911–924.

  27. Medico E, Mongiovi AM, Huff J, Jelinek MA, Follenzi A, Gaudino G, Parsons JT and Comoglio PM . (1996). Mol. Biol. Cell, 7, 495–504.

  28. Muraoka RS, Dumont N, Ritter CA, Dugger TC, Brantley DM, Chen J, Easterly E, Roebuck LR, Ryan S, Gotwals PJ, Koteliansky V and Arteaga CL . (2002). J. Clin. Invest., 109, 1551–1559.

  29. Nicolas FJ, Lehmann K, Warne PH, Hill CS and Downward J . (2003). J. Biol. Chem., 278, 3251–3256.

  30. Oft M, Akhurst RJ and Balmain A . (2002). Nat. Cell Biol., 4, 487–494.

  31. Oft M, Peli J, Rudaz C, Schwarz H, Beug H and Reichmann E . (1996). Genes Dev., 10, 2462–2477.

  32. Piek E, Heldin CH and Ten Dijke P . (1999b). FASEB J., 13, 2105–2124.

  33. Piek E, Moustakas A, Kurisaki A, Heldin CH and ten Dijke P . (1999a). J. Cell. Sci., 112, 4557–4568.

  34. Ronsin C, Muscatelli F, Mattei MG and Breathnach R . (1993). Oncogene, 8, 1195–1202.

  35. Rubin JS, Bottaro DP and Aaronson SA . (1993). Biochim. Biophys. Acta, 1155, 357–364.

  36. Ruiz P and Gunthert U . (1996). World J. Urol., 14, 141–150.

  37. Santoro MM, Collesi C, Grisendi S, Gaudino G and Comoglio PM . (1996). Mol. Cell Biol., 16, 7072–7083.

  38. Santoro MM, Penengo L, Minetto M, Orecchia S, Cilli M and Gaudino G . (1998). Oncogene, 17, 741–749.

  39. Savagner P . (2001). Bioessays, 23, 912–923.

  40. Serini G and Gabbiani G . (1999). Exp. Cell Res., 250, 273–283.

  41. Skeel A, Yoshimura T, Showalter SD, Tanaka S, Appella E and Leonard EJ . (1991). J. Exp. Med., 173, 1227–1234.

  42. Takenoshita S, Mogi A, Nagashima M, Yang K, Yagi K, Hanyu A, Nagamachi Y, Miyazono K and Hagiwara K . (1998). Genomics, 48, 1–11.

  43. Thiery JP . (2002). Nat. Rev. Cancer, 2, 442–454.

  44. Thiery JP and Chopin D . (1999). Cancer Metastasis Rev., 18, 31–42.

  45. Vleminckx K and Kemler R . (1999). Bioessays, 21, 211–220.

  46. Wang MH, Dlugosz AA, Sun Y, Suda T, Skeel A and Leonard EJ . (1996b). Exp. Cell Res., 226, 39–46.

  47. Wang MH, Montero-Julian FA, Dauny I and Leonard EJ . (1996a). Oncogene, 13, 2167–2175.

  48. Wang MH, Ronsin C, Gesnel MC, Coupeym L, Skeel A, Leonard EJ and Breathnach R . (1994a). Science, 266, 117–119.

  49. Wang M-H, Wang D and Chen Y-Q . (2003). Carcinogenesis, 23, 1291–1300.

  50. Yang YA, Dukhanina O, Tang B, Mamura M, Letterio JJ, MacGregor J, Patel SC, Khozin S, Liu ZY, Green J, Anver MR, Merlino G and Wakefield LM . (2002). J. Clin. Invest., 109, 1607–1615.

  51. Yue J and Mulder KM . (2001). Pharmacol. Therapeutics, 91, 1–34.

  52. Zhou YQ, Chen YQ, Fisher JH and Wang MH . (2002). J. Biol. Chem., 277, 38104–38110.

  53. Zhou YQ, He C, Chen YQ, Wang D and Wang MH . (2003). Oncogene, 22, 186–197.

Download references

Acknowledgements

We thank Drs EJ Leonard (NCI of NIH, Frederick, MD, USA) for human MSP; K Hagiwara (Tohoku University, Sendai, Japan) for the Smad2 promoter sequence. B Hann (UCSF) for the PAI-1 promoter sequence; and A Balmain (UCSF) for ΔSmad2. We are grateful to Ms Q Tanna (Denver Health Medical Center, Denver, CO, USA) for editing the manuscript. This study was supported in part by NIH Grant R01 CA91890 to MHW and The foundation of Chang-Jiang Scholar Endowment from the Chinese Ministry of Education.

Author information

Correspondence to Ming-Hai Wang.

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • receptor tyrosine kinases
  • transforming-growth factor
  • epithelial–mesenchymal transitions
  • signal transduction

Further reading