Smad4 regulates TGF-β1-mediated hedgehog activation to promote epithelial-to-mesenchymal transition in pancreatic cancer cells by suppressing Gli1 activity

Abstract

Pancreatic cancer (PC) is an aggressive and metastatic gastrointestinal tumor with a poor prognosis. Persistent activation of the TGF-β/Smad signaling induces PC cell (PCC) invasion and infiltration via epithelial-to-mesenchymal transition (EMT). Hedgehog signaling is a crucial pathway for the development of PC via the transcription factors Gli1/2/3. This study aimed to investigate the underlying molecular mechanisms of action of hedgehog activation in TGF-β1-triggered EMT in PCCs (PANC-1 and BxPc-3). In addition, overexpression and shRNA techniques were used to evaluate the role of Smad4 in TGF-β1-treated PCCs. Our data showed that TGF-β1 promoted PCC invasion and infiltration via Smad2/3-dependent EMT. Hedgehog-Gli signaling axis in PCCs was activated upon TGF-β1 stimulation. Inhibition of hedgehog with cyclopamine effectively antagonized TGF-β1-induced EMT, thereby suggesting that the hedgehog signaling may act as a downstream cascade signaling of TGF-β1. As a key protein that assists the nuclear translocation of Smad2/3, Smad4 was highly expressed in PANC-1 cells, but not in BxPc-3 cells. Conversely, Gli1 expression was low in PANC-1 cells, but high in BxPc-3 cells. Furthermore, knockdown of Smad4 in PANC-1 cells by shRNA inhibited TGF-β1-mediated EMT and collagen deposition. Overexpression of Smad4 did not affect TGF-β1-mediated EMT due to the lack of significant increase in nuclear expression of Smad4. Importantly, Gli1 activity was upregulated by Smad4 knockdown in PANC-1 cells and downregulated by Smad4 overexpression in BxPc-3 cells, indicating that Gli1 may be a negative target protein downstream of Smad4. Thus, Smad4 regulates TGF-β1-mediated hedgehog activation to promote EMT in PCCs by suppressing Gli1 activity.

Graphical Abstract

Highlights

• •TGF-β1 induces hedgehog-Gli signaling activation and promotes Smad2/3-dependent EMT in pancreatic cancer cells.
• •Hedgehog-Gli signaling plays a key role in TGF-β1-induced EMT in pancreatic cancer cells.
• •Smad4 and Gli1 have an obvious negative correlation in TGF-β1-induced EMT in pancreatic cancer cells.

1. Introduction

Pancreatic cancer (PC) is a gastrointestinal tumor with a high malignancy and poor prognosis. Epidemiological data have demonstrated that the 5-year survival rate for PC is less than 8% [1], [2]. Surgery is the mainstay of treatment for PC. However, most patients will miss the opportunity for surgery due to the invasion and metastasis of PC cells (PCCs). In addition, even if surgery is performed, the effect is not sufficient and prognosis is poor [3]. Therefore, exploring the underlying mechanisms of PCC invasion and metastasis may be a key factor in determining the prognosis of patients.

PC, mainly referred to as pancreatic ductal adenocarcinoma (PDAC), and its epithelial-mesenchymal transition (EMT) is the main reason for PCC invasion and metastasis [4]. During the EMT process, pancreatic ductal cells (PDCs) lose epithelial markers such as E-cadherin and ZO-1 under the stimulation of internal factors such as KRAS or TP53 mutations or external injury factors such as hypoxia. As a result, PDCs transdifferentiate into mesenchymal cells expressing alpha-smooth muscle actin (α-SMA). α-SMA-positive mesenchymal cells are more suitable for growth in damaged microenvironments and are more likely to migrate. At present, it is believed that the activation of the transforming growth factor beta-1 (TGF-β1)/Smad signaling pathway may be one of the main incentives to promote EMT [5]. TGF-β affects the phosphorylation of transforming growth factor beta receptor-2 (TGFBR2) by binding to its receptor TGFBR1. The latter can promote phosphorylation of the carboxyl terminus of the intracellular signal effector proteins Smad2 and Smad3 (classical pathway), and form a Smad multimer complex under the action of Smad4 in the cytoplasm that can enter the nucleus to bind specific regulatory elements of target genes. Consequently, activated TGF-β/Smad signaling induces the transcription of downstream target genes, and promotes a series of biological behaviors such as EMT via EMT-inducing transcription factors (EMT-TFs), including Snail1 and Snail2 (Slug). In addition, whole-genome sequence analysis has demonstrated that the Smad4 deletion was observed in 60% of PC patients [6]. Deletion of Smad4 results in a dysfunctional canonical TGF-β/Smad2/3 signaling that causes activation of the non-canonical TGF-β signaling pathway [7]. Non-canonical TGF-β signaling through RhoA, Ras, MAP3K7, and other pathways can also cause cellular EMT effects at the cellular level [8]. Therefore, the presence or absence of Smad4 can direct EMT in different ways, thereby leading to invasion and metastasis of tumor cells.

Previous studies have demonstrated that the hedgehog signaling is involved in the EMT process of tumor cells [9], [10], [11]. Hedgehog pathway is an evolutionarily conserved signaling pathway, which persistent activation has carcinogenic effects. The hedgehog signaling pathway is mainly composed of hedgehog ligands, two membrane receptors Ptch, Smoothened (Smo) and the downstream transcription factors Gli1/2/3. Ptch and Smo are transmembrane proteins that are located on the cell membrane, among which Ptch is a hedgehog receptor that can bind to hedgehog protein. In the absence of hedgehog protein, Ptch inhibits the activity of Smo, which in turn inhibits the transcriptional expression of downstream genes. When hedgehog protein binds to Ptch, the inhibitory effect on Smo is dismissed, and the downstream transcriptional regulator Gli1/2/3 is activated to induce the expression of target genes. These target genes include genes that are associated with cell proliferation, such as MYC, cyclin C, and cyclin D [12]. In addition, activated hedgehog signaling induces the expression of Snail1, which in turn promotes the EMT effect of tumor cells. Different Gli proteins mediate different biological activities. Studies using mutant mouse models have shown that Gli1 and Gli2 play a positive role in activating hedgehog signaling, and Gli3 has a reverse inhibitory effect [13]. However, in PDAC, different functional roles of Gli in the induction of EMT are not fully understood.

In tumor cells, there is a dialogue between TGF-β signaling and hedgehog signaling, which together affect biological functions, such as tumor cell invasion and infiltration [14], [15]. In cervical cancer cells, the TGF-β receptor TGFBR2 restrains the migration and proliferation abilities via mediating Smad4 to partially block the hedgehog signaling pathway [16]. In addition, inhibition of the hedgehog pathway by Gli1 siRNA or cyclopamine and GANT61 activates TGF-β-activated kinase in thyroid tumor cells [17]. Moreover, Smad4 physically interacts with Gli1 for concerted regulation of gene expression and cellular survival [18]. Thus, these findings support a possible cross-link between TGF-β signaling and hedgehog signaling via Smad4 and Gli1, respectively.

In this study, we aimed to explore the underlying molecular mechanism by which TGF-β1 triggers the activation of hedgehog signaling in PCCs (PANC-1 and BxPc-3). Firstly, the activity of hedgehog signaling was analyzed during TGF-β1-induced EMT in PCCs. In addition, TGF-β-induced effects on the induction of EMT were evaluated by using recombinant Shh protein to upregulate hedgehog activity or Smo-specific small-molecule inhibitor cyclopamine to downregulate hedgehog signaling. Furthermore, Smad4 knockdown in PANC-1 cells and Smad4 overexpression in BxPc-3 cells were selected to identify the role of Smad4 in the dialogue between TGF-β1 and hedgehog signaling. Our findings demonstrated that Smad4 regulated TGF-β1-mediated hedgehog activation to promote EMT in PCCs by suppressing Gli1 activity. Thus, these results provide new insights into the underlying mechanisms of PCC invasion and metastasis.

2. Materials and methods

2.1. Cells culture and reagents

Human PCC lines (BxPc-3, Patu8988, CFPAC-1, MiaPaCa-2, and PANC-1) and normal pancreatic ductal epithelial cell (hTERT-HPNE) were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). BxPc-3 cells were cultured in Roswell Park Memorial Institute-1640 medium (RPMI-1640, Invitrogen, Grand Island, NY, USA), and PANC-1 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Invitrogen). The medium contained 10% fetal bovine serum (FBS, Invitrogen), 100 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen), and cells were cultured in a 37 ℃-cell incubator with 5% CO₂. Cyclopamine was obtained from GLPBIO (Shanghai, China, #GC13441), recombinant human sonic hedgehog (Shh) protein was purchased from PeproTech (Suzhou, China, #100–45), and human recombinant TGF-β1 was obtained from MedChemExpress (MCE, Shanghai, China, #HY–P7118) [14], [19].

2.2. Construction and transient transfection of plasmids

Sequences used for knockdown and overexpression of the Smad4 gene were obtained from Sigma-Aldrich (Shanghai, China), and Sangon (Shanghai, China) was commissioned to design the relevant plasmids (Table S1). Plasmids specific for knockdown and overexpression of the Smad4 gene were added to the culture plate plated with PCCs at the concentration of 0.5 ng/ml. After 24 h of culturing, 2 μg/ml of puromycin (Sigma-Aldrich) and 400 μg/ml G418 (Invitrogen) were added to screen for successfully transfected cells [20].

2.3. Cell wound-healing assays

The migratory capacity of PCCs was tested using cell wound-healing assays [21]. First, PCCs at a density of 5 × 10⁵ cells/well were added to the six-well plate. When the cells were 100% confluent, a yellow pipette tip (200 μl) was used to evenly draw a straight line, and cells were scraped to form a scratch (cell exfoliation) and cells were washed with phosphate buffer saline (PBS). After making the scratch, cells were cultured in a medium containing 2% FBS, and corresponding drugs or recombinant proteins were added for 24 h. Using an inverted microscope (Nikon, Tokyo, Japan), images of the scratch were obtained. Experiments were independently performed at least in triplicate.

2.4. Cell invasion assays

The invasive ability of PCCs was determined using transwell assays, and pretreated transwell chambers (Costar, USA) were used [14]. Before the experiment, PCCs were resuspended in medium without FBS. In brief, 800 μl/well of complete medium containing 15% FBS was added into a 24-well plate, and 200 μl medium containing 2% FBS and 4 × 10⁴ PCCs were added into the transwell chamber. After the cells were thoroughly mixed, corresponding drugs or recombinant proteins were added. After PCCs were cultured for 24 h, the cells on the inner side of the membrane of the transwell chamber were removed. The chamber was gently washed three times with PBS, fixed with 4% Paraformaldehyde (Solarbio, China) for 30 min, and stained with crystal violet (Solarbio). Images were finally captured using a microscope (Nikon). Experiments were independently performed at least in triplicate.

2.5. Immunofluorescence staining

PCCs were uniformly dispersed in six-well plates containing glass slides at a density of 1.0 × 10⁵ cells/well. After 24 h with or without drug treatment, PCCs were washed with PBS and fixed with 4% histiocyte fixative for 30 min. Next, 0.1% Triton× 100 was added to permeate the membrane and to increase the efficiency of antibody entry into the cells. After blocking with goat serum for 30 min, PCCs were incubated with primary antibodies, including anti-E-cadherin, α-SMA, Collagen III, Snail1, Slug, Smad4, p-Smad3, Shh, Ptch1, Smo, Gli1, Gli2, and Gli3 (Table S2) at 4 ℃ for 24 h. Then, cells were washed three times with PBS, and conjugated with a fluorescently labeled horseradish peroxidase (HRP)-conjugated goat anti-rabbit secondary antibody (DyLight 488 or 594, ProteinTech, Wuhan, China). Finally, nuclei were stained blue with anti-fluorescence decay 4'6-diamidino-2-phenylindole (DAPI, Solarbio, China), and image acquisition was performed using a Leica DM4000B fluorescence microscope (Leica, Wetzlar, Germany) and the lens were used 10 times of the eyepiece and 40 times of the objective. [21]. Experiments were independently performed at least in triplicate.

2.6. Immunoblot analysis and Co-inmunoprecipitation experiments

Total protein was extracted from PCCs, and the protein concentration of each sample was determined using the enhanced bicinchoninic acid (BCA) protein concentration assay kit (Beyotime, Shanghai, China). Total proteins were electrophoresed using 10% SDS-PAGE gels. Then, the proteins on the gels were transferred to polyvinylidene fluoride immunoblotting membranes (PVDF, Sigma Aldrich) by a wet transfer method. Subsequently, membranes were blocked with blocking solution containing 5% nonfat milk at room temperature [21], [22]. After 2 h, membranes were washed three times, and incubated with primary antibodies, including anti-E-cadherin, N-cadherin, α-SMA, Vimentin, Collagen I, Collagen III, Snail1, Slug, VEGF, TGF-β1, Smad2/3, p-Smad2, p-Smad3, Smad4, Shh, Ptch1, Smo, Gli1, Gli2, and Gli3 (Table S3), at 4 ℃ for 24 h. The GADPH antibody was used as the internal reference. Finally, membranes were incubated with an HRP-conjugated secondary antibody for 1 h, and the protein bands were visualized using a Super Signal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific Inc., Waltham, MA, USA). The protein supernatant extracted by the above method was pre-incubated with protein A+G agarose beads at 4 ℃ for 30 min to weaken non-specific binding, and then uniformly added into different 1.5 ml EP tubes. Three of the samples were incubated overnight on the shaker at 4 °C with magnetic beads (pre bound to mouse IgG (Beyotime, China), anti Gli1, and anti Smad4), while the remaining samples were not processed. Finally, the protein was resuspended with the sample buffer and incubated at 95 ℃ for 5 min before western blot analysis.

2.7. Human PC tissues and immunohistochemical (IHC) staining

Human PC tissue samples (n = 6) and paracarcinoma tissue samples (n = 6) were obtained from the First Affiliated Hospital of Wenzhou Medical University and approved by the Ethics Committee in Clinical Research (ECCR, YJ2019–011-01). PC and paracarcinoma tissue samples were fixed and dehydrated, embedded in paraffin and sectioned into slices at a thickness of 4 µm. Before immunohistochemistry (IHC) staining, tissue sections were deparaffinized with xylene, and antigens were recovered by hydration with gradient 100%∼75% ethanol and sodium citrate buffer [23]. After treatment for 10 min with the endogenous peroxidase blocking agent (10%, BOSTER, Wuhan, China), 25 μl of goat serum (10%, BOSTER) was added dropwise, and sections were blocked for 1 h. Sections were incubated with primary antibodies, including anti-TGF-β1, Gli1, Gli2, and Gli3 (Table S4) at 4 °C for 24 h, then washed three times with PBS. Next, sections were incubated with HRP-conjugated secondary antibody for 1 h. Finally, sections were incubated with 3,3’-diaminobenzidine (DAB) chromogenic reagent (Beyotime). Then, the sections were dried in a 37 °C incubator and sealed with neutral resin. Images were taken using a microscope. Stained samples were semi-quantitatively or quantitatively assessed by two independent investigators who were blinded to the study groups.

2.8. Bioinformatic analyses

To investigate the mRNA expression of TGF-β1 in PC tissues and normal pancreatic tissues, the Gene Expression Profiling Interactive Analysis database (http:// http://gepia2.cancer-pku.cn/) was used [24]. In addtion, the correlation between TGF-β1 gene and Gli1, Gli2, or Gli3 was analyzed with Pearson correlation analysis. Next, the UALCAN database (http://ualcan.path.uab.edu/index.html), a comprehensive open access public platform, was utilized to analyze the relationship between TGF-β1 protein expression and the pancreatic tumor grade. Moreover, the TIMER2.0 database (http://timer.cistrome.org/) was used to evaluate the relationship between TGF-β1 expression and the infiltration of immune cells, including B cells, CD8⁺ T cells, CD4⁺ T cells, macrophages, neutrophils, and dendritic cells.

2.9. Statistical analysis

The statistical results of the above-mentioned experiments were obtained by double-blind statistics by more than two experimental investigators. Software used for statistical analysis and drawing of the results included Image J software (version 1.4.3, NIH, USA) and GraphPad Prism software (version 8.3.0, GraphPad Software, USA). Statistical plots across experiments are presented as the mean ± standard deviation (SD). To determine whether the differences between the experimental groups in each experiment were significant, the Student’s t-test and one-way analysis of variance (ANOVA) was used for parametric variables, and Wilcoxon rank sum test was used for non-parametric variables. P < 0.05 was set as the level of statistical significance.

3. Results

3.1. TGF-β1 promotes PCC invasion and infiltration via Smad2/3-dependent EMT

TGF-β1 is an important factor promoting PCC infiltration and metastasis. In this study, data from the database as well as the results of clinical samples and cell lines showed that both mRNA and protein expression of TGF-β1 were significantly increased in PC tissues (P < 0.05) (Fig. 1A, B, Fig. S1A) compared with normal pancreatic tissues, and positively correlated with clinical stages (P < 0.05) (Fig. S1B). Data collected from the databases also showed that elevated TGF-β1 expression was associated with the infiltration of various immune cells, including CD4⁺ T cells, macrophages, neutrophils, and dendritic cells (Fig. S1C), but was not significantly correlated with the infiltration of B cells and CD8⁺ T cells (Fig. S1C). Stimulation of PANC-1 or BxPc-3 cells with TGF-β1 markedly induced fibrosis-like changes in cell morphology (Fig. 1C, D). The wound-healing assay showed that TGF-β1 increased the mobility of PANC-1 (P < 0.01) or BxPc-3 (P < 0.05) cells (Fig. 1E, F). Moreover, transwell experiments showed that TGF-β1 enhanced the invasive ability of PANC-1 (P < 0.05) or BxPc-3 (P < 0.05) cells (Fig. 1G, H). Therefore, these results confirmed the promoting effect of TGF-β1 on PCC invasion and infiltration.

Fig. 1.

TGF-β1 promotes PCC invasion and infiltration. A. Immunohistochemical (IHC) staining of TGF-β1 in PC tissues and paracarcinoma tissues. Bar = 50 µm. B. Immunoblot analysis of TGF-β1 in pancreatic ductal epithelial cell (hTERT-HPNE) and PCCs (BxPc-3, Patu8988, CFPAC-1, MiaPaCa-2, and PANC-1). C, D. Cell morphology in TGF-β1-treated (10 ng/ml for 24 h) PANC-1 and BxPc-3 cells, Bar = 50 µm. E, F. Wound-healing assay of TGF-β1-treated PANC-1 and BxPc-3 cells s, Bar = 200 µm. G, H. Transwell analysis of TGF-β1-treated PANC-1 and BxPc-3 cells. Data were presented as the mean ± standard deviation, Bar = 50 µm. *P < 0.05, ^**P < 0.01, ^***P < 0.001.

Previous studies report that the invasion and infiltration of epithelial PCCs mainly depend on the EMT [21]. Therefore, we examined whether the EMT program was activated in the two cell lines treated with TGF-β1. After TGF-β1 stimulation, expression of the epithelial marker E-cadherin in PCCs was significantly downregulated, while the expression of mesenchymal markers N-cadherin, α-SMA, and Vimentin were significantly upregulated (Fig. 2A, B), thereby suggesting that TGF-β1 induced EMT in two types of PCCs, resulting in increased expression of extracellular matrix components, such as Collagen I and III (Fig. 2C, D). During EMT, the expression of E-cadherin and N-cadherin is regulated by EMT-Transcription Factors (EMT-TFs), including Snail1 and Snail2 (Slug). Our data showed that the expressions of Snail1 and Slug in both PCCs were significantly increased after TGF-β1 stimulation (Fig. 2E, F). In addition, the increased expression of VEGF caused by TGF-β1 suggested to some extent that tumor metastasis also involved angiogenesis (Fig. 2G). In vivo, TGF-β1 regulates the expression of EMT-TFs mainly through downstream Smad2/3, which are key nuclear transcription factors involved in PC development (Fig. S1D, E). In addition, TGF-β1 in tumor cells can activate EMT-TFs through non-Smad2/3 signaling [8]. To further clarify whether the pro-EMT effect of TGF-β1 in PC involves the activation of Smad2/3, we determined the expression and phosphorylation of Smad2/3 in PANC-1 or BxPc-3 cells stimulated by TGF-β1. Our data showed that TGF-β1 not only increased the phosphorylation levels of Smad2 and Smad3 (Fig. 2H), but also promoted the nuclear expression of p-Smad3 (Fig. 2I). Taken together, these results suggest that TGF-β1 may promote PCC invasion and infiltration via Smad2/3-dependent EMT.

Fig. 2.

TGF-β1 induces EMT in PCCs via Smad2/3 pathway. A. Immunoblot analysis of E-cadherin, N-cadherin, α-SMA, and Vimentin in TGF-β1-treated (10 ng/ml for 24 h) PANC-1 and BxPc-3 cells. B. Immunofluorescence staining of E-cadherin and α-SMA in TGF-β1-treated PANC-1 and BxPc-3 cells, Bar = 5 µm. C. Immunoblot analysis of Collagen I and III in TGF-β1-treated PANC-1 and BxPc-3 cells. D. Immunofluorescence staining of Collagen I and III in TGF-β1-treated PANC-1 and BxPc-3 cells, Bar = 5 µm. E. Immunoblot analysis of Snail1 and Slug in TGF-β1-treated PANC-1 and BxPc-3 cells. F. Immunofluorescence staining of Snail1 and Slug in TGF-β1-treated PANC-1 and BxPc-3 cells, Bar = 5 µm. G. Immunoblot analysis of VEGF in TGF-β1-treated PANC-1 and BxPc-3 cells. H. Immunoblot analysis showing the expression and phosphorylation of Smad2 and Smad3 in TGF-β1-treated PANC-1 and BxPc-3 cells. I. Immunofluorescence staining of p-Smad3 in TGF-β1-treated PANC-1 and BxPc-3 cells, Bar = 5 µm. ^***P < 0.001.

3.2. TGF-β1 promotes the activation of canonical hedgehog-Gli signaling axis in PCCs

Previous studies have revealed that canonical hedgehog signaling plays an important regulatory role in the occurrence and development of PDAC through different downstream Gli molecules [10], [25]. In this study, results from clinical samples showed that Gli1, Gli2 and Gli3 were highly expressed in PDAC tissues (Fig. 3A). Similarly, gene expression levels of Gli1 (P < 0.05), Gli2 (P < 0.05), and Gli3 (P < 0.05) were increased in PC from the Gepia2 database (Fig. 3B). In addition, protein expression levels of Gli1, Gli2, and Gli3 were higher in two types of PCCs (PANC-1 and BxPc-3) compared to a normal pancreatic ductal epithelial cell line (hTERT-HPNE) (Fig. 3C). Interestingly, unlike Gli2 and Gli3, the expression of Gli1 was significantly higher in BxPc-3 cells than that in PANC-1 cells. Considering that PANC-1 cells show normal expression of Smad4, while BxPc-3 cells have a homozygous deletion of Smad4, it is speculated that there may be a negative correlation between Smad4 and Gli1 expression.

Fig. 3.

TGF-β1 induces canonical hedgehog activation in PCCs by enhancing Gli1/2/3 activity. A. IHC staining of Gli1, Gli2, and Gli3 in PC tissues and paracarcinoma tissues. Bar = 50 µm. B. The Gepia2 database showing the mRNA expression of Gli1, Gli2, and Gli3 in PC tissues and normal pancreatic tissues. C. Immunoblot analysis of Smo, Gli1, Gli2, and Gli3 in pancreatic ductal epithelial cell (hTERT-HPNE) and PCCs (BxPc-3 and PANC-1). D. Immunoblot analysis of Shh, Ptch1, and Smo in TGF-β1-treated (10 ng/ml for 24 h) PANC-1 and BxPc-3 cells. E. Immunofluorescence staining of Shh, Ptch1, and Smo in TGF-β1-treated PANC-1 and BxPc-3 cells, Bar = 5 µm. F. Immunoblot analysis of Gli1, Gli2, and Gli3 in TGF-β1-treated PANC-1 and BxPc-3 cells. G. Immunofluorescence staining of Gli1, Gli2, and Gli3 in TGF-β1-treated PANC-1 and BxPc-3 cells, Bar = 5 µm. *P < 0.05, ^***P < 0.001.

Next, we further investigated whether hedgehog signaling in PCCs is activated under TGF-β1 stimulation and its role in EMT. In both types of PCCs, the expression of Shh and Smo significantly increased after stimulation with TGF-β1, while the expression of Ptch1 significantly decreased (Fig. 3D, E). As important downstream regulators of hedgehog signaling, expression of Gli1, Gli2, and Gli3 increased by TGF-β1 (Fig. 3F), and their nuclear expression was significantly enhanced (Fig. 3G). Collectively, these findings indicated that TGF-β1 promotes the activation of canonical hedgehog signaling in PCCs.

3.3. Hedgehog-Gli signaling plays a key role in TGF-β1-induced EMT in PCCs

In this study, recombinant human protein Shh was used to activate hedgehog signaling. We observed a marked increase in Shh expression in both PCCs (Fig. S2A), suggesting that recombinant protein Shh has drug activity. However, no significant changes were observed in the expression of Shh receptor Ptch1 and downstream Gli1, Gli2, and Gli3 (Fig. S2B, C). Similarly, Shh did not downregulate E-cadherin expression and did not increase the expression of α-SMA, Vimentin, Collagen I and III (Fig. S3A, B), thereby indicating that Shh did not induce the EMT effect. These findings are consistent with the fact that no significant changes were observed in the expression of Snail1 and Slug in the two types of PCCs (Fig. S3C). Further analysis revealed that Shh did not induce increased TGF-β1 expression, or downstream Smad2 and Smad3 phosphorylation and nuclear expression (Fig. S3D, E). Treatment with Shh did not affect downstream hedgehog signaling or alter EMT even in the context of a TGF-β1 effect (Fig. S4).

Secondly, the Smo-specific small molecule inhibitor cyclopamine was used in PANC-1 or BxPc-3 cells after TGF-β1stimulation. It was observed that the effect of cyclopamine significantly inhibited TGF-β1-mediated activation of hedgehog signaling, especially significantly decreased the expression of Gli1 in BxPc-3 cells (Fig. 4A-C). Furthermore, cyclopamine abolished TGF-β1-triggered EMT by increasing the expression of E-cadherin and reducing the expression of a-SMA, N-cadherin, and Collagen III (Fig. 4B, C)). Also, TGF-β1-induced an increase in Snail1 and Slug expression, which was antagonized by cyclopamine treatment (Fig. 4D). Thus, these results suggest that downregulated hedgehog activity caused by cyclopamine contributed to the abolishment of TGF-β1-mediated EMT. As shown by the results of the PANC-1 (P < 0.05) and BxPc-3 (P < 0.001) cells. TGF-β1-increased PCC migration was inhibited by cyclopamine (Fig. 4E). In addition, the TGF-β1-enhanced invasive ability of PANC-1 (P < 0.01) and BxPc-3 (P < 0.01) cells was antagonized by cyclopamine (Fig. 4F). Furthermore, cyclopamine treatment inhibited the phosphorylation and nuclear expression of Smad2 and Smad3 (Fig. 4G, H). Therefore, these results suggested that inhibition of hedgehog signaling by cyclopamine may prevent the activation of TGF-β1 via the Smad2/3 pathway.

Fig. 4.

Inhibition of hedgehog signaling prevents TGF-β1-mediated EMT. A. Immunoblot analysis of Smo, Gli1, Gli2, and Gli3 in TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine (10 μM for 24 h) treatment. B. Immunoblot analysis of E-cadherin, N-cadherin, α-SMA, and Collagen III in TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment. C. Immunofluorescence staining of Smo and α-SMA in TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment, Bar = 5 µm. D. Immunoblot analysis of Snail1 and Slug in TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment. E. Wound-healing assay of TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment, Bar = 200 µm. F. Transwell analysis of TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment, Bar = 50 µm. G. Immunoblot analysis showing the expression and phosphorylation of Smad2 and Smad3 in TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment. H. Immunofluorescence staining of p-Smad3 in TGF-β1-treated PANC-1 and BxPc-3 cells with or without cyclopamine treatment. Data were presented as the mean ± standard deviation. ns, not significant, Bar = 5 µm. *P < 0.05, ^**P < 0.01, ^***P < 0.001.

3.4. Smad4 knockdown enhances Gli1 expression in TGF-β1-treated PANC-1 cells

As previously mentioned, the SMAD4 gene is inactivated in approximately 60% of PDAC cases. In PCCs, PANC-1 cells normally express Smad4, whereas Smad4 is not expressed by homozygous deletion in BxPc-3 cells (Fig. 5A). Interestingly, like Gli1, Shh expression was significantly higher in Smad4-deficient BxPc-3 cells compared with PANC-1 cells, thereby suggesting that Smad4 may antagonize the hedgehog-Gli1 signaling axis. Smad4 was markedly enhanced in PANC-1 cells upon stimulation by TGF-β1 and was nuclearized (Fig. 5B, C). Therefore, shRNA was used to specifically knockdown the expression of Smad4 in PANC-1 cells (Fig. 5D, E). In addition, shSmad4 inhibited nuclear expression of phosphorylated Smad3 (Fig. 5E), which resulted in malfunctioning of the TGF-β1 signaling in the nucleus. Thus, the high expression of Snail1 and Slug induced by TGF-β1 was inhibited by shSmad4 (Fig. 5F, G), which ultimately reduced the accumulation of Collagen I and III, α-SMA, and Vimentin (Fig. 5G, H), and inhibited cell invasion and infiltration (Fig. S5A, B). This EMT antagonism by shSmad4 was not achieved by restoring E-cadherin expression (Fig. 5H). Further analysis showed that Smad4 knockdown did not affect the expression of Shh, Ptch1 and Smo, but suppressed the expression of Gli2 and Gli3 and increased the protein expression of Gli1 (Fig. 5H, I). Enhanced expression of Gli1 by shSmad4 is consistent with the higher expression of Gli1 in Smad4-deficient BxPc-3 cells. In addition, Smad4 knockdown did not induce the activation of non-Smad2/3 signaling, such as JAK2/STAT3 and AKT/mTOR pathways (Fig. S5C, D). Thus, these findings indicated that Smad4 and Gli1 have antagonistic effects in TGF-β1-induced EMT of PCCs.

Fig. 5.

Smad4 knockdown enhances Gli1 expression in TGF-β1-treated PANC-1 cells. A. Immunoblot analysis of Smad4 and Shh in BxPc-3 and PANC-1 cells. B. Immunoblot analysis of Smad4 in TGF-β1-treated PANC-1 cells. C. Immunocytochemical staining of Smad4 in TGF-β1-treated PANC-1 cells, Bar = 5 µm. D. Immunoblot analysis of Smad4 in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment. E. Immunocytochemical staining of Smad4 and p-Smad3 in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment, Bar = 5 µm. F. Immunoblot analysis of Snail1 and Slug in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment. G. Immunocytochemical staining of Snail1 and Collagen III in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment, Bar = 5 µm. H. Immunoblot analysis of E-cadherin, α-SMA, Vimentin, Collagen I and III in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment. I. Immunoblot analysis of Shh, Ptch1, Smo, Gli1, Gli2, and Gli3 in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment. J. Immunocytochemical staining of Smo in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment, Bar = 5 µm. K. Immunocytochemical staining of Gli1, Gli2, and Gli3 in TGF-β1-treated PANC-1 cells with or without shSmad4 treatment. ns, not significant, Bar = 5 µm. *P < 0.05, ^**P < 0.01, ^***P < 0.001.

3.5. Smad4 overexpression reduces Gli1 expression in TGF-β1-treated BxPc-3 cells

Next, we investigated the effect of Smad4 overexpression on the expression of Gli1 in TGF-β1-treated BxPc-3 cells. Smad4 was induced to be overexpressed in Smad4-deficient BxPc-3 cells, but did not affect its nuclear expression (Fig. 6A, B). Furthermore, overexpression of Smad4 did not increase the nuclear expression of phosphorylated Smad3 at the same concentration of TGF-β1 (Fig. 6B). Thus, overexpression of Smad4 did not exacerbate the TGF-β1-mediated increase in Snail1 and Slug (Fig. 6C, D), but it affects the expression of EMT-related proteins (Fig. 6D, E), as well as cell invasion and infiltration (Fig. S6A, B). In addition, Smad4 overexpression did not induce activation of non-Smad2/3 signaling, such as JAK2/STAT3 and AKT/mTOR pathways (Fig. S6C, D). The overexpression of Smad4 in BxPc-3 cells did not affect the expression of Shh, Ptch1, and Smo in hedgehog signaling (Fig. 6F, G). Interestingly, Smad4 overexpression in BxPc-3 cells inhibited Gli1 expression and nuclear translocation but did not affect Gli2 and Gli3 expression (Fig. 6H, I). Co-inmunoprecipitation experiments also identified the interaction between Smad4 and Gli1 protein (Fig. 6J). Thus, these results confirms again that Smad4 and Gli1 may have a significant negative correlation in TGF-β1-induced EMT in PCCs.

Fig. 6.

Smad4 overexpression reduces Gli1 expression in TGF-β1-treated BxPc-3 cells. A. Immunoblot analysis of Smad4 in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment. B. Immunocytochemical staining of Smad4 and p-Smad3 in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment, Bar = 5 µm. C. Immunoblot analysis of Snail1 and Slug in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment. D. Immunocytochemical staining of Snail1 and Collagen III in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment, Bar = 5 µm. E. Immunoblot analysis of E-cadherin, α-SMA, Vimentin, Collagen I and III in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment. F. Immunoblot analysis of Shh, Ptch1, and Smo in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment. G. Immunocytochemical staining of Smo in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment, Bar = 5 µm. H. Immunoblot analysis of Gli1, Gli2 and Gli3 in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment. I. Immunocytochemical staining of Gli1, Gli2 and Gli3 in TGF-β1-treated BxPc-3 cells with or without Smad4-overexpression treatment, Bar = 5 µm. J. Co-inmunoprecipitation experiments analyzed the interaction between Smad4 and Gli1 protein. OE, overexpression. ns, not significant. *P < 0.05, ^**P < 0.01, ^***P < 0.001.

4. Discussion

In the present study, we investigated the relationship between hedgehog signaling and TGF-β1 signaling during the EMT process in PCCs. As expected, TGF-β1 significantly induced invasion and infiltration of PCCs by activating associated EMT-TFs through Smad2/3. During TGF-β1-triggered EMT, hedgehog signaling was activated by increased expression of Gli1, as well as Gli2 and Gli3, thereby indicating that hedgehog signaling may be a cascade downstream of TGF-β1 signaling and is involved in the induction of EMT.

Multiple studies have demonstrated the cross-talk between hedgehog signaling and TGF-β1 signaling [26]. In fibrosis, epithelial TGF-β1 stimulates hedgehog signaling to promote EMT and collagen deposition after injury [27]. Pharmacological inhibition of hedgehog signaling can relieve TGF-β1-induced EMT effect. Similarly, TGF-β1-induced EMT can be exacerbated by recombinant sonic hedgehog protein. In hepatocellular carcinoma cells, TGF-β1 promotes cellular proliferation and invasion by activating Gli1 signaling [28]. In addition, Gli1 facilitates TGF-β1-induced EMT in gastric cancer cells [29]. Interestingly, in neuroblastoma cells, Gli1/2 was involved in TGF-β1-induced EMT that was independent of Smad signaling [30], suggesting that TGF-β1 activates hedgehog signaling in cancer cells to promote EMT possibly through a non-canonical Smad2/3 pathway. Our results showed that hedgehog signaling was activated after TGF-β1 stimulation both in Smad4-expressing PANC-1 cells and in Smad4-deficient BxPc-3 cells [31]. These findings suggest that blocking nuclear translocation of Smad2/3 does not prevent TGF-β1 from activating hedgehog signaling, and that TGF-β1-mediated hedgehog activation may occur in a non-canonical signal transduction manner.

Next, we investigated whether regulating hedgehog signaling affects the EMT-promoting effect of TGF-β1. Surprisingly, the application of recombinant protein increased Shh levels but did not activate hedgehog signaling. Consequently, Shh protein did not induce EMT in PCCs. In addition, TGF-β1-mediated EMT was not abolished by Shh protein. Previous studies have shown that PTCH1 and PTCH2, the ligands of Shh, are frequently mutated in cancer cells [32], [33], [34]. A mutated PTCH gene causes protein inactivation, which can relieve the inhibitory effect on the Smo protein, and activate downstream signal transduction, resulting in abnormal expression of target genes [35]. On the other hand, mutated PTCH can prevent binding to the Shh ligand, thereby rendering Shh ineffective, which makes it unable to perform its function of promoting hedgehog signaling activation [36].

Considering that a PTCH1 mutation may hinder recombinant protein Shh to fail to activate hedgehog signaling, we switched to cyclopamine, a specific inhibitor of Smo, to regulate hedgehog signaling via bypassing the Ptch1 protein. Cyclopamine treatment inhibited the expression of Gli1, Gli2, and Gli3 to inactivate hedgehog signaling after TGF-β1 stimulation. As a result, cyclopamine abolished TGF-β1-induced EMT through downregulation of the expression of EMT-TFs. Several studies have also shown that cyclopamine hinders EMT and inhibit the production of extracellular matrix production [37], [38]. In addition, cyclopamine inhibits tumor cell invasion and infiltration by antagonizing EMT [39], [40], which is consistent with our results. Interestingly, the EMT-inhibiting effect of cyclopamine seems to be related to the inactivation of Smad3, which suggested that there may be cross-talk with the downstream transcription factors of Smo, including Gli1, Gli2, and Gli3, with Smad3.

Smad4 is a suppressor of early PDAC tumors that blocks initial tumor progression, whereas in a subset of advanced tumors, intact Smad4 facilitates EMT and TGF-β-dependent growth. The nuclear expression of Smad3 to exert its pro-EMT effect requires the action of Smad4 [7], [41]. To further study the role of Smad4 in TGF-β1-induced hedgehog activation, shRNA technology was applied to knockdown the expression of Smad4 in TGF-β1-treated PANC-1 cells. As expected, Smad4 knockdown inhibited the nuclear expression of phosphorylated Smad3, and thereby antagonized the TGF-β1-induced EMT process by reducing partial EMT-TFs expression [42], [43]. Notably, Smad4 knockdown increased the protein expression of Gli1 in PANC-1 cells, which was consistent with the higher Gli1 expression in Smad4-deficient BxPc-3 cells. On the other hand, Smad4 overexpression in BxPc-3 cells was used to further verify the relationship between Smad4 and Gli1. Our study showed that Smad4 overexpression suppressed Gli1 expression and nuclear translocation in BxPc-3 cells, which was consistent with lower Gli1 activity in Smad4-expressing PANC-1 cells. These findings suggested a negative relationship between Smad4 and Gli1 in TGF-β1-induced EMT in PCCs. The most studies have shown that Smad4-mutant cancer cells cannot undergo EMT. However, EMT categorically precluded in Smad4mut tumors as human tumor transcriptomes show that Smad4 mutations are not underestimated in mesenchymal tumor samples, which is consistent with what we observed in the BxPc-3 cell line. The crystal diffraction structure of Smad4 shows that there is a triple helix bundle in the MH2 region of Smad4, a trimer formed by a tricyclic α-helix region, and an intermediate mouth sheet sandwich, which plays an important role in the recognition of intermolecular interactions [44]. We speculate that Smad4 may combine with Gli1 to form a complex, which prevents Smad4 from binding to phosphorylated Smad3 and subsequently translocate to the nucleus to regulate downstream target genes of Smad4. Similarly, the complex may prevent Gli1 from being cleaved for nuclear expression. Thus, Smad4 may play an important role in TGF-β1-mediated hedgehog activation in PCCs by targeted inhibition of Gli1 activity.

Our study has some limitations. First, a genetic approach to investigate the relationship between Smad4 and Gli1 needs to be presented. In addition, the direct or indirect mechanism of the action of the Smad4 and Gli1 proteins should be further evaluated by Glutathione-S-transferase-pull down (GST-pull down) assays on the basis of co-inmunoprecipitation (Co-IP) experiments. It is also worth mentioning that the target of the interaction between Smad4 and Gli1 protein needs to be further clarified through techniques such as molecular docking studies.

5. Conclusions

In conclusion, TGF-β1 induced hedgehog activation and EMT, and promoted the invasion and infiltration of PCCs. Inhibition of hedgehog signaling by cyclopamine effectively antagonized TGF-β1-mediated EMT. During EMT, Smad4 plays a vital role by assisting the nuclear expression of phosphorylated Smad3. Notably, a negative association was observed between Smad4 and Gli1 in TGF-β1-induced hedgehog activation. Gli1 was expressed at low levels in Smad4-expressing PANC-1 cells, but was highly expressed in Smad4-deficient BxPc-3 cells. Smad4 knockdown in PANC-1 cells enhanced Gli1 activity, but Smad4 overexpression in BxPc-3 cells suppressed Gli1 activity. Thus, these findings indicated that Smad4 regulates TGF-β1-mediated hedgehog activation to promote EMT in PCCs by suppressing Gli1 activity.

Funding

This study was sponsored by Key Laboratory of Diagnosis and Treatment of Severe Hepato-Pancreatic Diseases of Zhejiang Province (Grant No. 2018E10008).

Declaration of Competing Interest

The authors have declared that no competing interest exists.

Appendix A. Supplementary material

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.

Supplementary material

.