Biochemical Journal

Research article

Transforming growth factor-β2 promotes Snail-mediated endothelial–mesenchymal transition through convergence of Smad-dependent and Smad-independent signalling

Damian Medici , Scott Potenta , Raghu Kalluri

Abstract

EndMT (endothelial–mesenchymal transition) is a critical process of cardiac development and disease progression. However, little is know about the signalling mechanisms that cause endothelial cells to transform into mesenchymal cells. In the present paper we show that TGF-β2 (transforming growth factor-β2) stimulates EndMT through the Smad, MEK [MAPK (mitogen-activated protein kinase)/ERK (extracellular-signal-regulated kinase) kinase], PI3K (phosphinositide 3-kinase) and p38 MAPK signalling pathways. Inhibitors of these pathways prevent TGF-β2-induced EndMT. Furthermore, we show that all of these pathways are essential for increasing expression of the cell-adhesion-suppressing transcription factor Snail. Inhibition of Snail with siRNA (small interfering RNA) prevents TGF-β2-induced EndMT. However, overexpression of Snail is not sufficient to cause EndMT. Chemical inhibition of GSK-3β (glycogen synthase kinase-3β) allows EndMT to be induced by Snail overexpression. Expression of a mutant Snail protein that is resistant to GSK-3β-dependent inactivation also promotes EndMT. These results provide the foundation for understanding the roles of specific signalling pathways in mediating EndMT.

  • endothelial–mesenchymal transition (EMT)
  • GSK-3β (glycogen synthase kinase-3β)
  • Smad
  • Snail
  • transforming growth factor-β (TGF-β)

INTRODUCTION

EndMT (endothelial–mesenchymal transition) is an essential mechanism of endocardial cushion formation during cardiac development [13]. EndMT also has an essential role in cancer progression by causing formation of cancer-associated fibroblasts in the tumour microenvironment [4]. Many of the fibroblasts formed during cardiac and renal fibrosis have been shown to be of endothelial origin [5,6]. EndMT has also been implicated in atherosclerosis [7], pulmonary hypertension [8], diabetic nephropathy [9] and wound healing [10].

EndMT is characterized by loss of cell–cell adhesion and changes in cell polarity inducing a spindle-shaped morphology. These changes are accompanied by reduced expression of the endothelial markers, such as VE-cadherin (vascular endothelial cadherin) and CD31, and increased expression of the mesenchymal markers including FSP-1 (fibroblast-specific protein-1), α-SMA (α-smooth muscle actin), N-cadherin (neural cadherin) and fibronectin [11]. Loss of cell–cell adhesion is mediated by transcription factors such as Snail, Slug, ZEB-1 (zinc finger E-box-binding homoeobox 1), SIP-1, Twist and LEF-1 (lymphoid enhancer-binding factor-1) that suppress transcription of genes encoding proteins involved in formation of adherens junctions and tight junctions [1218].

TGF-β (transforming growth factor-β) signalling ligands are potent inducers of converting epithelial cells into mesenchymal cells [19,20]; however, EndMT appears to be stimulated primarily by the TGF-β2 isoform [12,17,2123]. Ablation of TGF-β2 in mice prevents EndMT-mediated cardiac development. TGF-β1- or TGF-β3-knockout mice show no significant effects on EndMT and heart development [3]. TGF-β2 has been described to promote EndMT by signalling through the TGF-β type 1 receptors ALK (activin receptor-like kinase) 2 and ALK5 [24,25], yet little is know about the downstream signalling events that occur to stimulate this process. The goal of the present study was to identify essential signalling pathways that mediate TGF-β2-dependent EndMT and expression of the EndMT-inducing transcription factor Snail.

EXPERIMENTAL

Cell culture

HCMECs (human cutaneous microvascular endothelial cells) were provided by Dr Bjorn R. Olsen (Department of Cell Biology, Harvard Medical School, Boston, MA, U.S.A.) and were isolated as described previously [26]. The cells were tested previously for purity and found to express no markers of lymphatic endothelial cells or stromal cells (smooth muscle cells, fibroblasts or pericytes) [27]. The cells were grown in culture using EGM (endothelial growth medium)-2 (Cambrex), containing 10% FBS (fetal bovine serum) and 1% penicillin/streptomycin, followed by human endothelial serum-free medium (Gibco) 24 h prior to all experimental conditions. Recombinant TGF-β2 (R&D Systems) was added to the serum-free culture medium for all relevant experiments at a concentration of 10 ng/ml. The cells were treated for 15 min to assess ERK (extracellular-signal-regulated kinase) 1/2, AKT, p38 MAPK (mitogen-activated protein kinase) or GSK-3β (glycogen synthase kinase-3β) phosphorylation, 1 h to measure Smad activity or 48 h to induce EndMT. A DN (dominant-negative) Smad4 (DN-Smad4) adenoviral construct (provided by Dr Diane Simeone, Department of Surgery, University of Michigan, Ann Arbor, MI, U.S.A.) was produced as described previously [28] and was used at a 1:100 dilution in serum-free medium, then added to cells for 24 h prior to treatment with TGF-β2. Small-molecule inhibitors were added to cultures 1 h prior to treatment with TGF-β2. The p38 MAPK inhibitor SB202190 (Tocris Bioscience) was used at a concentration of 25 μM, the PI3K (phosphinositide 3-kinase) inhibitor LY294002 (Cell Signaling Technology) was used at a concentration of 50 μM and the MEK (MAPK/ERK kinase)1/2 inhibitor U0126 (Cell Signaling Technology) was used at a concentration of 10 μM. The cells were transfected with 1 μg of pcDNA3-Snail (provided by Dr M. Angela Nieto, Instituto de Neurociencias de Alicante, CSIC-UMH, Alicante, Spain), pcDNA3 empty vector (Invitrogen), Snail-WT (wild-type) or Snail-6SA (provided by Dr Mien-Chie Hung, Department of Molecular and Cellular Oncology, The University of Texas, M.D. Anderson Cancer Center, Houston, TX, U.S.A.) using Lipofectin™ and Plus reagents (Invitrogen). LiCl was added at a concentration of 20 mM to cultures 24 h after transfection with expression plasmids. All experiments for the present study were performed at minimum in triplicate.

RNA interference

siRNA (small interfering RNA) gene expression knockdown studies were performed using the TriFECTa RNAi kit (Integrated DNA Technologies) and the corresponding protocol. Each 27-mer siRNA duplex was transfected into cells using X-tremeGene siRNA transfection reagent (Roche) following the manufacturer's guidelines. siRNA was synthesized (Integrated DNA Technologies) with the following sequences: Snail, 5′-CCACAGAAAUGGCCAUGGGAAGGCCAC-3′; negative control, 5′- UCACAAGGGAGAGAAAGAGAGGAAGGA-3′.

Luciferase reporter gene assays

Luciferase reporter gene assays were conducted using the Luciferase Assay System (Promega) and its corresponding protocol. All plasmids (500 ng) were transfected into cells using Lipofectin® and Plus reagents (Invitrogen) according to the manufacturer's guidelines. Light units were measured with a Luminometer TD-20/20 (Turner Designs). The assays were normalized for transfection efficiency by co-transfecting cells with a β-galactosidase control plasmid and were detected with the Luminescent β-galactosidase control assay kit (Clontech). Experimental (luciferase) results were divided by the β- galactosidase results to provide normalized data. The p3TP-Lux reporter plasmid was provided by Dr Joan Massague (Cancer Biology and Genetics Program, Memorial Sloan-Kettering Cancer Center, New York, NY, U.S.A.).

Immunoblotting

Western blot analysis was performed with the following antibodies using dilutions and protocols recommended by the respective manufacturers: anti-phospho-ERK1/2, anti-phospho-p38 MAPK (Millipore), anti-phospho-GSK-3β, anti-VE-cadherin, anti-Snail (Santa Cruz Biotechnology), anti-phospho-AKT, anti-AKT, anti-ERK1/2, anti-p38, anti-GSK-3β (Cell Signaling Technology), anti-CD31 (Dako), anti-α-SMA, anti-β-actin (Sigma–Aldrich) and anti-FSP-1 (Abnova). The samples were run with Criterion precast SDS/PAGE (10%) gels (Bio-Rad Laboratories). HRP (horseradish peroxidase)-conjugated IgG TrueBlot reagents (eBioscience) were used at a dilution of 1:1000.

Real-time quantitative PCR

RNA extractions were performed using the RNeasy Mini kit (Qiagen) and protocol. Real-time PCR experiments were conducted using the SYBR Green PCR system (Applied Biosystems) on an ABI 7500 cycler, with 40 cycles per sample. Cycling temperatures were: denaturing, 95 °C; annealing and extension, 60 °C. The following primers were used: Snail, forward 5′-ACCACTATGCCGCGCTCTT-3′ and reverse 5′-GGTCGTAGGGCTGCTGGAA-3′; GAPDH (glyceraldehyde-3phosphate dehydrogenase), forward 5′-ACCACAGTCCATGCCATCAC-3′ and reverse 5′-TCCACCCTGTTGCTGTA-3′.

Statistical analyses

One-way ANOVA was performed and confirmed with a two-tailed paired Student's t test using GraphPad Prism 4 software. P values less than 0.05 were considered significant.

RESULTS

We transfected HCMECs with the p3TP-Lux reporter plasmid in order to determine Smad transcription factor activity. We then treated the cells with recombinant TGF-β2 for 1 h and found that it greatly increased reporter activity. However, in cells that were also infected with an adenoviral DN-Smad4 expression construct, luciferase activity was dramatically reduced (Figure 1A). To assess Smad-independent signalling, immunoblotting was performed to detect phosphorylation of ERK1/2, AKT and p38 MAPK. Upon treatment of HCMECs with TGF-β2 for 15 min, it was found that phosphorylation levels of these kinases were all increased. Chemical inhibitors of MEK1/2 (U0126), PI3K (LY294002) and p38 MAPK (SB202190) prevented the TGF-β2-induced phosphorylation of ERK1/2, AKT and p38 MAPK respectively (Figures 1B–1D).

Figure 1 TGF-β2 activates Smad, MEK, PI3K and p38 MAPK signalling pathways

(A) p3TP-Lux reporter gene assay showing increased Smad activity upon treatment of HCMECs with TGF-β2. Expression of DN-Smad4 inhibited this increase in activity. The results represent means±S.D., n=3; *P<0.01 for TGF-β2 compared with vehicle; **P<0.05 for TGF-β2 and DN-Smad4 compared with TGF-β2 alone. (BD) Immunoblotting for phosphorylation levels of ERK1/2 (B), AKT (C) and p38 MAPK (D) showing that TGF-β2 increases the phosphorylation of these kinases. Chemical inhibitors against MEK1/2 (U0126; 10 μM), PI3K (LY294002; 50 μM), and p38 MAPK (SB202190; 25 μM) inhibit the increases in ERK1/2, AKT and p38 MAPK phosphorylation respectively. P-, phospho-.

Treatment of HCMECs with TGF-β2 for 48 h caused a dramatic change in cell morphology from the endothelial cobblestone-like form to an elongated spindle-shaped form that is characteristic of EndMT. However, pre-treatment of the cells with DN-Smad4, the MEK1/2 inhibitor, PI3K inhibitor or p38 MAPK inhibitor all prevented this change in cell morphology (Figure 2A). Real-time quantitative PCR was performed to assess gene expression of the EndMT-inducing transcription factor Snail. TGF-β2 dramatically increased Snail mRNA levels, which was inhibited in the presence of inhibitors to Smad4, MEK1/2, PI3K or p38 MAPK (Figure 2B). We performed immunoblotting to assess protein expression of Snail, the endothelial markers VE-cadherin and CD31, and the mesenchymal markers FSP-1 and α-SMA. TGF-β2 treatment caused a decrease in the expression of VE-cadherin and CD31, and an increase in the levels of FSP-1, α-SMA and Snail. Exposure of the cells to DN-Smad4, the MEK1/2 inhibitor, PI3K inhibitor or p38 MAPK inhibitor prevented these expression changes (Figure 2C).

Figure 2 TGF-β2 promotes EndMT through Smad-dependent and Smad-independent signalling

(A) DIC (differential interference contrast) imaging showing a change in cell morphology consistent with EndMT in HCMEC cultures treated with TGF-β2. Inhibitors against Smad4 (DN-Smad4), MEK1/2 (U0126; 10 μM), PI3K (LY294002; 50 μM), or p38 MAPK (SB202190; 25 μM) prevented the TGF-β2-induced change in morphology. Scale bar, 20 μm. (B) Real-time quantitative PCR analysis showing that TGF-β2 increases Snail gene expression, which is prevented by inhibitors of Smad4, MEK1/2, PI3K or p38 MAPK. Results are means±S.D., n=3; *P<0.01 for TGF-β2 compared with vehicle; **P<0.01 for all TGF-β2 and inhibitors compared with TGF-β2 alone. (C) Immunoblotting showing that TGF-β2 decreases the expression of VE-cadherin and CD31, and increases the expression of FSP-1, α-SMA and Snail. Inhibitors of Smad4, MEK1/2, PI3K or p38 MAPK prevent these expression changes.

Since all four signalling pathways are necessary for increasing expression of Snail, the importance of Snail was assessed in EndMT by inhibiting its expression with siRNA. HCMECs were transfected with either negative control siRNA or Snail siRNA and then treated with TGF-β2 for 48 h. TGF-β2 induced the mesenchymal morphology in cells transfected with control siRNA, but not in those tranfected with Snail siRNA (Figure 3A). Immunoblotting showed that inhibition of Snail expression was sufficient to prevent TGF-β2-induced decreases in VE-cadherin and CD31 and increases in FSP-1 and α-SMA, suggesting that Snail expression is necessary for EndMT (Figure 3B).

Figure 3 Snail activity is essential for TGF-β2-induced EndMT

(A) DIC imaging showing a change in cell morphology in cultures transfected with control siRNA treated with TGF-β2. No EndMT was observed in cultures transfected with Snail siRNA. Scale bar, 20 μm. (B) Immunoblotting showing that Snail siRNA inhibits TGF-β2-induced expression changes in VE-cadherin, CD31, FSP-1 and α-SMA.

Next we sought to determine whether Snail expression was sufficient to induce EndMT. HCMECs were transfected with a Snail expression plasmid for 48 h; however, we observed no change in cell morphology (Figure 4A). Immunoblotting confirmed a dramatic increase in Snail expression in cells transfected with the Snail expression plasmid, but no significant changes were observed in expression levels of VE-cadherin, CD31, FSP-1 and α-SMA (Figure 4B), suggesting that Snail expression alone is not sufficient to induce EndMT.

Figure 4 Snail expression is not sufficient to induce EndMT

(A) DIC imaging showing no effect of snail overexpression on cell morphology. Scale bar, 20μm. (B) Immunoblotting confirming a dramatic increase in Snail gene expression in cells transfected with the Snail expression construct. No significant changes in expression of the endothelial markers VE-cadherin and CD31 or the mesenchymal markers FSP-1 and α-SMA were observed.

The signalling kinase GSK-3β has been shown to control Snail activity by impairing its function, and inhibitory phosphorylation of GSK-3β promotes Snail transcriptional activity [29]. GSK-3β can be phosphorylated by other kinases such as AKT [30]. We showed in Figure 1 that AKT is activated by TGF-β2 through the PI3K pathway. Immunoblotting using lysates from our cultures showed that TGF-β2-treated cells have a large increase in GSK-3β phosphorylation. This increase in phosphorylation was blocked in cells exposed to the PI3K inhibitor (Figure 5A). The cells transfected with the Snail expression plasmid showed no phosphorylation of GSK-3β. LiCl, a chemical inhibitor of GSK-3β, increased GSK-3β phosphorylation in the cultures (Figure 5B). Snail protein levels were increased in cells transfected with the Snail expression plasmid and treated with LiCl, compared with cells transfected with the Snail expression plasmid in the absence of LiCl (Figure 5B).

Figure 5 Inhibition of GSK-3β allows Snail-induced EndMT

(A) Immunoblotting showing increased phosphorylation of GSK-3β in endothelial cells treated with TGF-β2. Inhibition of PI3K with LY294002 (50 μM) was sufficient to block GSK-3β phosphorylation induced by TGF-β2. (B) Immunoblotting demonstrating no phosphorylation of GSK-3β when overexpressing Snail. LiCl was sufficient to induce the phosphorylation of GSK-3β in cells transfected with pcDNA3 or pcDNA3-Snail plasmids. Snail expression is increased in cells transfected with pcDNA3-Snail and treated with LiCl. (C) DIC imaging showing that the GSK-3β inhibitor LiCl was sufficient to transform endothelial cells transfected with pcDNA3-Snail to mesenchyme. Scale bar, 20 μm. (D) Immunoblotting confirming expression of Snail, a decrease in the expression of endothelial markers VE-cadherin and CD31, and an increase in the expression of mesenchymal markers FSP-1 and α-SMA in cells containing pcDNA3-Snail and treated with LiCl. P-, phospho-.

We next attempted to determine whether inhibition of GSK-3β with LiCl could allow EndMT to be induced by Snail overexpression. Cells transfected with the pcDNA3-Snail expression plasmid then treated with LiCl did take on the mesenchymal morphology (Figure 5C). Snail overexpression was confirmed by immunoblotting (Figure 5D). LiCl caused pcDNA3-Snail-transfected cells to have a reduced expression of VE-cadherin and CD31, and an increased expression of FSP-1 and α-SMA (Figures 5D). These results suggest that inhibition of GSK-3β allows Snail to induce EndMT.

For further confirmation of GSK-3β-dependent regulation of Snail activity we transfected HCMECs with expression plasmids encoding either Snail-WT or a mutant Snail (Snail-6SA) that is resistant to GSK-3β-dependent inhibition. The mutant Snail construct caused acquisition of the mesenchymal morphology (Figure 6A), as well as a decrease in the expression of the endothelial markers VE-cadherin and CD31 and an increase in the expression of the mesenchymal markers FSP-1 and α-SMA (Figure 6B), further suggesting a critical role for GSK-3β in regulating Snail-induced EndMT.

Figure 6 Induction of EndMT by a GSK-3β-resistant mutant form of Snail

(A) DIC imaging demonstrating EndMT of cells transfected with a mutant GSK-3β-resistant Snail (Snail-6SA) construct. Scale bar, 20 μm. (B) Immunoblotting showing a decrease in the expression of endothelial markers (VE-cadherin and CD31) and an increase in the expression of mesenchymal markers (FSP-1 and α-SMA) in cells expressing mutant Snail, but not Snail-WT.

DISCUSSION

The results of the present study provide novel insight into the signalling mechanisms that mediate EndMT. It was found that TGF-β2 signals through the Smad, MEK, PI3K and p38 MAPK pathways, and all of these pathways are essential for inducing EndMT. Furthermore, all of these pathways are necessary for promoting increased expression of the EndMT-inducing transcription factor Snail, which suppresses cell adhesion and promotes EndMT [13]. Smad proteins have been shown to bind directly to the promoter of the Snail gene to regulate its transcription [3134]. Transcription factors induced by Smad-independent signalling that regulate Snail transcription remain unclear, but we show that these pathways have a critical role in controlling Snail gene expression.

In epithelial and cancer systems, studies have shown that transfection of cells with a Snail expression plasmid is sufficient to induce EMT [35,36]. Surprisingly, we found that expression of Snail alone is insufficient to induce EMT. These results suggest that other mechanisms are necessary to mediate this change in endothelial cell morphology. One such mechanism is inhibition of GSK-3β.

Snail protein stability and nuclear transport capability are inhibited through phosphorylation by GSK-3β [29]. Therefore inhibitory phosphorylation of GSK-3β by kinases such as AKT, a downstream signalling molecule from PI3K [30], is sufficient to prevent GSK-3β-dependent inhibition of Snail. The results of the present study show that inhibition of GSK-3β by TGF-β2-induced PI3K signalling or by direct inhibition with LiCl allows Snail to function to induce EndMT. PI3K is also necessary for controlling Snail gene expression, demonstrating a dual role for this pathway in mediating EndMT.

Although other transcription factors and signal transduction pathways probably play a critical role in EndMT, the present study has established a foundation for understanding this cellular transformation by identifying four major signalling pathways that co-operate to control a known regulator of cell–cell adhesion and cellular plasticity. Identifying signalling pathways that control EndMT is necessary for translational applications in clinical medicine. Such knowledge may prove beneficial for designing therapeutic strategies for EndMT-associated diseases, such as fibrosis, cancer, diabetes and atherosclerosis [11]. Targeting the Smad, MEK, PI3K or p38 MAPK pathways, or their downstream target Snail, with small-molecule inhibitors should perturb this detrimental mechanism of disease progression.

AUTHOR CONTRIBUTION

Damian Medici designed the experiments, performed the research, analysed the data and wrote the paper. Scott Potenta performed the research and analysed the data. Raghu Kalluri designed the experiments, analysed the data and wrote the paper.

FUNDING

This work was supported by the National Institutes of Health [grant numbers CA125550, CA155370, CA151925, DK55001, DK81576 (to R.K.) and F30HL095319 (to S.P.)], and the Champalimaud Foundation (to R.K.).

Acknowledgments

We thank Dr Bjorn Olsen (Harvard Medical School) for providing cells and antibodies necessary for the present study. We also thank Dr Mien-Chie Hung (M.D. Anderson Cancer Center) for providing the Snail plasmids.

Abbreviations: ALK, activin receptor-like kinase; α-SMA, α-smooth muscle actin; DIC, differential interference contrast; DN, dominant-negative; EndMT, endothelial–mesenchymal transition; ERK, extracellular-signal-regulated kinase; FSP-1, fibroblast-specific protein-1; GSK-3β, glycogen synthase kinase-3β; HCMEC, human cutaneous microvascular endothelial cell; MAPK, mitogen-activated protein kinase; MEK, MAPK/ERK kinase; PI3K, phosphinositide 3-kinase; siRNA, small interfering RNA; TGF-β, transforming growth factor-β; VE-cadherin, vascular endothelial cadherin; WT, wild-type

References

View Abstract