Human pancreatic cancer PANC-1 and MIA PaCa-2 cells were cultured as described previously (25,26). These cells were incubated with recombinant human TGF-β (R&D Systems, Minneapolis, MN, USA) or A-83-01 (Takara, Shiga, Japan) at various concentrations and periods.

Short interference RNA (siRNA) against DEC1 was synthesized by Qiagen (Hilden, Germany). The sequences of the sense and anti-sense DEC1 siRNA and the negative control (scrambled) siRNA, and the siRNA transfection were performed as described previously (18).

DEC1 overexpression was induced using pcDNA vector as described previously (19). After transfection, the cells were incubated for 24 h and subjected to the invasion assay.

Cells were lysed using M-PER lysis buffer (Thermo Scientific, Rockford, IL, USA) and their protein concentration was determined using the bicinchoninic acid (BCA) assay. Their lysates were subjected to SDS-PAGE and detected the protein expression as described previously (18,19).

The membranes for western blotting were incubated with antibodies specific to DEC1 (1:10,000; Novus Biologicals Inc., Littleton, CO, USA), Smad3 (1:1,000; Epitomics Inc., Burlingame, CA, USA), pSmad3 (1:6,000; Epitomics), phospho extracellular signal-related kinases (p-ERK)½ (1:1,000; Epitomics), pSmad2 (1:5,000; Invitrogen, Carlsbad, CA, USA), slug (1:3,000; Cell Signaling Technology Inc.), snail (1:3,000; Cell Signaling Technology Inc., Danvers, MA, USA), ERK1/2 (1:30,000; Cell Signaling Technology Inc.), α-smooth muscle actin (α-SMA) (1:20,000; Sigma Chemical Co., St. Louis, MO, USA), vimentin (1:10,000; Epitomics), N-cadherin (1:10,000; ECM Biosciences, Versailles, KY, USA), E-cadherin (1:1000; Takara), claudin-1 (1:10,000; Invitrogen), claudin-4 (1:20,000; Invitrogen), claudin-7 (1:1,000; Invitrogen), and actin (1:30,000; Sigma), followed by horseradish peroxidase-conjugated secondary antibody (Immuno-Biological Laboratories, Fujioka, Japan). Can Get Signal immunoreaction enhancer solution (Toyobo Co. Ltd., Osaka, Japan) or Immunoshot immunoreaction enhancer solution (Cosmobio Co. Ltd., Tokyo, Japan) was used to dilute the primary antibody.

We prepared three independent RNA samples (n=3). Total-RNA was isolated and first-strand cDNA was synthesized as described previously (19). The real-time PCR was performed using SYBR-Green Master Mix (Life Technologies, Carlsbad, CA, USA). The sequences of the primers and the sizes of products are shown in Table I.

Immunofluorescent staining was performed as described previously (18). The permeabilized cells were incubated with anti-DEC1 (1:300), pSmad3 (1:300), snail (1:300), N-cadherin (1:300), E-cadherin (1:300), claudin-1 (1:300), claudin-4 (1:300), or claudin-7 (1:300) antibodies at 4°C overnight. The cells were then incubated for 1 h with goat anti-rabbit IgG antibody conjugated to Alexa 488 dye (Molecular Probes Inc., Tokyo, Japan), while nuclear staining was performed using 4′,6-diamidino-2-phenylindole (DAPI). These cells were visualized using confocal laser scanning microscopy LSM 710 (Zeiss, Wetzlar, Germany).

Cell stain was carried out using the CnT-ST-100 stain kit (CellnTEC Advanced Cell Systems AG, Bern, Switzerland), in accordance with the manufacturer’s instructions.

The invasion assay was performed using a BD BioCoat Matrigel invasion chamber kit (Becton Dickinson, Franklin Lakes, NJ, USA). PANC-1 cells were separated using cell dissociation solution (Sigma) and then (5×10⁴ cells/600 μl) were added to the top chamber of a cell culture insert in a 24-well companion plate. After overnight incubation, the cells that had invaded the lower surface of the membrane were fixed with methanol and subjected to Giemsa staining. The number of cells that had migrated was quantified by counting them in ten random distinct fields using a light microscope.

For the migration assay, PANC-1 cells were seeded in a 4-chamber slide glass, and an artificial ‘wound’ was carefully created at 0 h by scratching the confluent cell monolayer with the tip of a P-200 pipette. Microphotographs were taken at 0, 24 and 48 h.

We examined immunohistochemical analyses of surgically resected pancreatic cancers (n=17), which have been filed in Hirosaki University Hospital, Japan (Table II). All of the 17 cases examined were invasive ductal carcinoma of pancreas. Histological specimens were retrieved from the archives of our hospital according to the guidelines produced by the Japanese Society of Pathology.

Immunohistochemistry was performed as described previously (27). The sections were incubated overnight at 4°C with anti-DEC1 (1:100), pSmad3 (1:200), or claudin-4 (1:400) antibodies diluted in Can Get Signal Immunostain Solution (Toyobo Co.). Finally, the sections were counterstained with Mayer’s hematoxylin.

TGF-β treatment induced Smad3 phosphorylation, and upregulated the expression of DEC1, snail, α-SMA, vimentin and N-cadherin, whereas it downregulated E-cadherin, claudin-1 and claudin-4 in PANC-1 cells (Fig. 1A and B). The highest level of DEC1 expression was observed in the cells treated with 2 ng/ml of TGF-β for 24 h. TGF-β had little effect on the Smad2 phosphorylation and the expression of Smad3, slug and claudin-7 in PANC-1 cells. On the other hand, no apparent effects were observed on the expression of the aforementioned molecules after TGF-β treatment in MIA PaCa-2 cells. Next, we investigated the endogenous mRNA expression of these proteins in PANC-1 cells treated with TGF-β. The expression of DEC1, snail and N-cadherin was upregulated by TGF-β, whereas the expression of E-cadherin and claudin-4 was downregulated (Fig. 1C). A-83-01-an inhibitor of the TGF-β signaling pathway prevents Smad2/3 phosphorylation. We therefore investigated whether A-83-01 affects the Smad3 phosphorylation and DEC1 expression induced by TGF-β in PANC-1 cells. A-83-01 suppressed the Smad3 phosphorylation, and the expression of DEC1 and snail. On the other hand, claudin-4 expression was upregulated by A-83-01 (Fig. 1D).

The cells that were transfected with the control siRNA showed a spindle-shaped morphology after 24 h treatment with TGF-β, while the cells transfected with DEC1 siRNA displayed no morphological changes after the same treatment (Fig. 2A). As shown in Fig. 2B, in the absence of TGF-β, DEC1 siRNA upregulated the expression of claudin-4, claudin-7 and E-cadherin, and downregulated the expression of N-cadherin, whereas it had little effect on the Smad3 phosphorylation, and the expression of snail and claudin-1. In the presence of TGF-β, DEC1 siRNA decreased the Smad3 phosphorylation, and the expression of snail and N-cadherin, whereas it upregulated the expression of claudin-1, claudin-4, claudin-7, and E-cadherin. In the presence or absence of TGF-β, DEC1 siRNA had little effect on the ERK1/2 phosphorylation, and the expression of total-ERK1/2, α-SMA and vimentin. Another DEC1 siRNA oligonucleotide yielded similar results (data not shown). In order to examine whether DEC1 regulates EMT-related factors at the transcriptional level, we performed real-time PCR analysis. The altered mRNA expression patterns of snail, E-cadherin, claudin-4 and N-cadherin were compatible with the above protein results (Fig. 2C). In the presence of TGF-β, DEC1 siRNA also significantly decreased the expression of TGF-βRI, although DEC1 siRNA without TGF-β slightly decreased the expression of TGF-βRI. DEC1 siRNA regardless of TGF-β had little effect on the expression of TGF-βRII (data not shown).

As shown in Fig. 2D, DEC1 siRNA without TGF-β decreased the amount of N-cadherin in the cell membrane, while it increased the cell membrane levels of E-cadherin, claudin-4 and claudin-7. On the other hand, DEC1 siRNA without TGF-β had little effect on the amounts of nuclear/cytoplasmic pSmad3, snail and claudin-1. In the presence of TGF-β, control siRNA increased the levels of pSmad3 and snail in the nucleus compared with the absence of TGF-β, and slightly increased the level of N-cadherin in the cell membrane, whereas it decreased the amounts of E-cadherin and claudin-1 in the cell membrane, and decreased the amount of claudin-4 in the cytoplasm. However, it had little effect on the amount of claudin-7 in the cytoplasm or membrane. In the presence of TGF-β, DEC1 siRNA decreased the levels of pSmad3 and snail in the nucleus compared with control siRNA, and it also decreased the amount of N-cadherin in the cell membrane. On the other hand, a combination treatment of DEC1 siRNA with TGF-β significantly increased the levels of E-cadherin, claudin-1 and claudin-4 in the cell membrane compared with control siRNA. A combination treatment of DEC1 siRNA with TGF-β also slightly increased the amount of claudin-7 in the cell membrane. These findings demonstrated that DEC1 has inducible effects on EMT in PANC-1 cells during TGF-β treatment, which involved alterations in the cellular amounts of EMT-related factors.

We examined whether DEC1 expression was involved in the migration and invasion of PANC-1 cells. In the presence of TGF-β, DEC1 siRNA delayed cell migration for 24–48 h in comparison with those transfected with the control siRNA (Fig. 3). We performed an invasion assay in which we transiently transfected the cells with a DEC1 expressing plasmid. As a result, the number of invasive PANC-1 cells with a spindle-shaped morphology was increased in the presence of TGF-β (Fig. 4). The invasion assay also demonstrated that there were significantly more invasive spindle-shaped cells among the cells transfected with DEC1 than among those transfected with the control pcDNA vector.

We examined the immunohistochemical expression of DEC1, pSmad3 and claudin-4 in human pancreatic cancer tissues. Photographs of DEC1, pSmad3 and claudin-4 expression in representative cases are shown in Fig. 5. Significant DEC1 immunoreactivity was detected in the cancer tissues (100%, 17/17 cases) compared with the adjacent non-cancerous pancreatic tissues, and it was predominantly located within the cytoplasm of the cancer cells, although very weak DEC1 immunoreactivity was found in the non-cancerous pancreatic ductal cells of all cases. DEC1 immunoreactivity was detected in parts of the adjacent non-cancerous tissues (17%, 3/17 cases).

It is often difficult to distinguish between spindle-shaped cells of the cancer invasive front and fibroblasts in stroma. We distinguish them by dyskaryosis and size. Firstly, dyskaryosis was shown in spindle-shaped cells, whereas it was not shown in fibroblasts in stroma. Secondary, the sizes of spindle-shaped cells were larger than fibroblasts. The spindle-shaped cells of the cancer invasive front showed strong DEC1 immunoreactivity similar to that seen in the other cancer regions (case 9, right panel). Marked claudin-4 immunoreactivity was detected in the membrane and/or cytoplasm of non-cancerous pancreatic ductal cells, whereas that in the cancer cells was weak, except for 4 cases in which strong claudin-4 expression was detected in cancer cells. Significant pSmad3 expression was found in the nucleus of spindle-shaped cancer cells, whereas it was detected in non-cancerous ductal cells in 3 cases. The changes in the immunohistochemical expression of DEC1, pSmad3, and claudin-4 were found to be independent of the cancer grade and the patient’s age and gender.