Four human pancreatic cancer cell lines, PANC-1 (cat. no. CRL-1469), AsPC-1 (cat. no. CRL-1682), BxPC-3 (cat. no. CRL-1687) and MIA PaCa-2 (cat. no. CRL-1420) were purchased from American Type Culture Collection (ATCC). The cell lines (PANC-1, AsPC-1, BxPC-3 and MIA PaCa-2) had been identified by specialized STR profiling and tested for mycoplasma contamination. PANC-1 and MIA PaCa-2 were maintained in Dulbecco's Modified Eagle's Medium (DMEM), while AsPC-1 and BxPC-2 were maintained in RPMI-1640 medium (both from Corning Incorporated). The medium was supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.), 1% penicillin/streptomycin and 1 mM glutamine (Invitrogen; Thermo Fisher Scientific, Inc.). The cells were grown in 5% CO₂ at 37°C in a humidified atmosphere.

EMT Antibody Sampler Kit (cat. no. 9782), NF-κB pathway Sampler Kit (cat. no. 9936), Phospho-EGF Receptor Antibody Sampler Kit (cat. no. 9922) and anti-phospho-STAT3 (cat. no. 9145) antibody were purchased from Cell Signaling Technology, Inc., except for anti-ERK (cat. no. sc-154) and anti-phospho-ERK (cat. no. sc-7976-R), anti-AKT (cat. no. sc-81434) and anti-phospho-AKT (cat. no. sc-81433) and anti-STAT3 (cat. no. sc-482) antibodies (Santa Cruz Biotechnology, Inc.). The recombinant human AREG protein (cat. no. 262-AR-100) and normal goat IgG control (cat. no. AB-108-C) were purchased from R&D Systems. PD153035 (EGFR tyrosine kinase inhibitor), U0126 (MEK1/2 inhibitor) and QNZ (NF-κB inhibitor) were purchased from Selleckchem. PD153035, U0126 and QNZ were dissolved in dimethyl sulphoxide (DMSO) and maintained at −20°C until use. AREG and IgG were dissolved in sterile PBS containing 0.1% FBS just before use.

The cells were grown to 80–90% confluence in 25-cm² flasks and then rinsed three times with serum-free medium. Then, the cells were incubated with serum-free medium for 48 h. The conditioned medium was harvested and centrifuged (335 × g for 10 min, 4°C). After centrifugation, the supernatant was collected and stored at −80°C for the subsequent enzyme-linked immunosorbent assay (ELISA).

Detailed procedures for total RNA extraction and RT-qPCR have been previously described (17). The primer sequences were as follows: AREG, 5′-TGAGATGTCTTCAGGGAGTG-3′ (sense) and 5′-AGCCAGGTATTTGTGGTTCG-3′ (antisense); EGFR, 5′-GGATGCCGACGAGTACCTC-3′ (sense) and 5′-GCTTTGCAGCCCATTTCTAT-3′ (antisense); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-GGCATCCTGGGCTACACTG-3′ (sense) and 5′-GTGGTCGTTGAGGGCAAT-3′ (antisense). GAPDH was used as a control. The mRNA levels were analysed using the comparative quantification cycle (Cq) method (2^−ΔΔCq) (18).

Western blot analysis was performed as previously described (17). Dilution rates for all primary and secondary antibodies were 1:1,000 and 1:10,000, respectively. The blots were probed with ECL Plus Western Blotting Detection kit (Pierce Biotechnology; Thermo Fisher Scientific, Inc.). The densities of the bands were scanned and analysed using ImageJ software (version 1.48v; National Institutes of Health).

The secretion level of AREG was assessed using an AREG ELISA kit (cat. no. DAR00; R&D systems) according to the manufacturer's instructions.

The siRNAs targeting AREG were as follows: AREG-Homo-390 (siAR-390), 5′-GGAUUUGAGGUUACCUCAATT-3′ (sense) and 5′-UUGAGGUAACCUCAAAUCCTT-3′ (antisense); AREG-Homo-817 (siAR-817), 5′-GCAUGAUUGACAGUAGUUUTT-3′ (sense) and 5′-AAACUACUGUCAAUCAUGCTT-3′ (antisense); and AREG-Homo-348 (siAR-348), 5′-UCUGGGAAGCGUGAACCAUTT-3′ (sense) and 5′-AUGGUUCACGCUUCCCAGATT-3′ (antisense). The siRNA control sequences were 5′-UUCUCCGAACGUGUCACGUTT-3′ (sense) and 5′-ACGUGACACGUUCGGAGAATT-3′ (antisense). All siRNAs were purchased from Shanghai GenePharma Co., Ltd. AREG was transfected into AsPC-1 cells in 6-well plates using Lipofectamine RNAiMAX Reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Total protein and RNA were extracted, and the knockdown efficiency of AREG was verified by western blotting and RT-qPCR analyses, respectively.

AsPC-1 cells were transfected with AREG siRNA (siAR-390) or control siRNA (NC) for 48 h. PANC-1 cells were treated with AREG (100 ng/ml, AREG), or normal IgG (100 ng/ml, NC) for 48 h. A total of 2×10⁵ cells were seeded on coverslips. After 24 h, serum-free culture medium was removed from each well and replaced with 4% paraformaldehyde, and the plates were incubated for 15 min at room temperature for fixation. Then, the cells were washed with PBS three times. After blockade of non-specific epitopes by 10% normal goat serum for 1 h at room temperature, primary antibodies against E-cadherin (1:100 dilution; cat. no. 3195; Cell Signaling Technology, Inc.) or vimentin (1:200 dilution; cat. no. 5741; Cell Signaling Technology, Inc.) were added. After overnight incubation at 4°C, the primary antibody was washed by rinsing three times with PBS, followed by incubation of secondary antibodies (Alexa Fluor^®488 Conjugate, cat. no. 4412 or Alexa Fluor^®594 Conjugate, cat. no. 8889, Cell Signaling Technology, Inc., 1:1,000 dilution) for 1 h at room temperature. The cells were stained with DAPI and observed under a fluorescence microscope (Olympus Corp.).

Transwell chambers (8-µm-pore; Corning Inc.) were used to evaluate cell migration and invasion. Matrigel was applied in the invasion assay. Briefly, after transfection or different treatment conditions for 48 h, 2×10⁴ cells were resuspended in 200 µl of serum-free medium and added to the upper chamber. Then, medium with 10% FBS was added to the opposite (lower) chamber. After 24 h, the cells remaining in the upper chamber were wiped away. The cells on the Transwell membrane were fixed and stained (by 4% paraformaldehyde and 0.1% crystal violet, respectively, for 15 min each at room temperature). Finally, the number of cells was counted in five random fields (magnification, ×100; BX53 upright microscope; Olympus Corp.).

PANC-1 cells were treated with AREG (100 ng/ml) alone (AREG), AREG combined with QNZ (20 µM, AREG+QNZ), or normal IgG (100 ng/ml, NC) for 48 h. A total of 5×10⁵ cells were seeded in a 6-well plate and grown to 90% confluence. A sterile 200-µl pipette tip was used to form a constant width gap. After removal of cell debris, 1 ml of serum-free medium was added to each well. The scratched areas were imaged at 0, 24, 48, and 72 h using a DP-73 scientific camera on a phase-contrast microscope (Olympus Corp.). The cell-free area was calculated using ImageJ software (version 1.48v; National Institutes of Health), and the percentage of the original cell-free zone at each time-point was used to evaluate the speed of wound healing. The values were the means of three independent experiments.

Short hairpin RNA (shRNA) targeting AREG was designed and synthesized by Shanghai GeneChem Co., Ltd., and used to construct the experimental plasmids (pGV248/AREG-shRNA); the shRNA was inserted into an AgeI- and EcoRI-linearized pGV248. The negative control vector (pGV248/control-shRNA) was constructed similarly by inserting an unrelated shRNA sequence. All inserted sequences were verified by sequencing. To generate stable cell lines, recombinant lentiviruses (namely, LV-AREG and LV-con) were generated and subsequently used to infect AsPC-1 cells.

Twelve six- to seven-week-old female athymic nude mice (BALB/c nu/nu) were randomly assigned to 2 groups. Six mice with LV-AREG cells were distributed to the experimental group, while the other 6 mice with LV-con cells were in the control group. In detail, the mice were anaesthetized and placed in a supine position; the abdomen was disinfected with 70% alcohol, and then, an incision of <1 cm was performed to expose the pancreas. AsPC-1 cells (2×10⁶) were injected into the tail of the pancreas, and the wound was sutured. The mice were fed in SPF conditions and weighed twice weekly. The mice were sacrificed by CO₂ asphyxiation on the 39th day after the injection. The pancreas was removed, and the tumours were weighed. The stomach, spleen, liver, intestine and kidney were examined for the presence of metastases. This study was conducted with the approval of the Institutional Animal Care and Use Committee of Peking Union Medical College Hospital and was in accordance with the National Policy on Use of Laboratory Animals.

All data were sorted and analysed using SPSS software for Windows, version 17.0 (SPSS, Inc.). The results are expressed as the mean ± SD. Significance was evaluated by performing Student's t-tests or one-way analysis of variance (ANOVA) with Tukey's post hoc test. A P-value <0.05 was considered to indicate a statistically significant difference.

To investigate EGFR and AREG expression in pancreatic cancer cell lines, RT-qPCR, western blot analysis and ELISA were performed. These analyses revealed that EGFR expression was significantly high in the AsPC-1 cells and BxPC-3 cells compared with that in MIA PaCa-2 cells (Fig. 1A and B). AREG expression was significantly high in the AsPC-1 cells, BxPC-3 cells and MIA PaCa-2 cells, compared with that in PANC-1 cells (Fig. 1C and D). AsPC-1 (high EGFR expression and high AREG expression) and PANC-1 cells (high EGFR expression and low AREG expression) were selected for the subsequent loss-of-function and gain-of-function experiments. To characterize the biological effects of AREG, AREG siRNAs were used to suppress the endogenous expression of AREG in AsPC-1 cells. Three different AREG siRNAs and 1 control siRNA were used. The results revealed that siAR-390 was most effective in inhibiting AREG mRNA (65% reduction) and protein (64% reduction) expression in AsPC-1 cells (Fig. S1). Therefore, siAR-390 was used in the subsequent experiments.

Transwell assays were performed to monitor whether AREG could affect pancreatic cancer cell migration and invasion. The results revealed a nearly 50% reduction in cell migration and 47% reduction in cell invasion in AREG siRNA-transfected AsPC-1 cells (P<0.05 and P<0.01, Fig. 2A). Knockdown of AREG significantly decreased the migration ability of BxPC-3 cells (P<0.01, Fig. S2). To confirm the involvement of AREG in pancreatic cancer cell migration and invasion, recombinant human AREG treatment was used. Exposure to exogenous AREG (100 ng/ml) significantly promoted the migration and invasion of the PANC-1 cells (P<0.05 and P<0.01, Fig. 2B). It was revealed that AREG was expressed in activated pancreatic stellate cells (PSCs), and co-culture with PSCs significantly increased the migration and invasion of PANC-1 cells (Fig. S3). The results revealed that AREG facilitated the migration and invasion of pancreatic cancer cells.

E-cadherin and ZO-1 expression was upregulated, however, vimentin, β-catenin, Snail, Slug, and ZEB-1 levels were downregulated in the AREG siRNA-transfected AsPC-1 cells (Fig. 3A). Knockdown of AREG upregulated the expression of E-cadherin and downregulated the expression of β-catenin and vimentin in BxPC-3 cells (Fig. S2). As anticipated, these proteins displayed the opposite expression profiles in the AREG-treated (100 ng/ml) PANC-1 cells, whereas ZEB-1 expression exhibited no evident change (Fig. 3B). The changes in the expression of E-cadherin and vimentin induced by AREG were also examined by immunofluorescence. The results revealed the upregulated expression of E-cadherin and the downregulated expression of vimentin in AREG siRNA-transfected AsPC-1 cells. In contrast, the downregulated expression of E-cadherin and the upregulated expression of vimentin were observed in AREG-treated PANC-1 cells (Fig. 3C).

To further explore the upstream signalling that may be responsible for AREG-induced EMT, the phosphorylation of EGFR, ERK, AKT and STAT3 was evaluated by performing western blot analyses. The phosphorylation levels of EGFR, ERK, AKT and STAT3 levels were downregulated in the AREG siRNA-transfected AsPC-1 cells (Fig. 4A). As anticipated, the exogenous treatment of AREG upregulated the phosphorylation levels of EGFR, ERK, AKT and STAT3 to varying degrees in PANC-1 cells. It was revealed that the small molecule inhibitors of EGFR (PD153035) suppressed the AREG-induced ERK, AKT and STAT3 phosphorylation (Fig. 4B). The expression levels of total EGFR, ERK, AKT, and STAT3 were not significantly altered in pancreatic cancer cells (Fig. 4A and B). The results revealed a significant decrease in phospho-ERK in siRNA-transfected AsPC-1 cells (P<0.01) and a significant increase in phospho-ERK in exogenous AREG-treated PANC-1 cells (P<0.05). To further characterize the effect of AREG on ERK activation, ERK phosphorylation was examined at various time-points after AREG stimulation; the peak phosphorylated ERK level occurred 30 min after AREG stimulation (Fig. 4C). The small molecule inhibitor of MEK1/2 (U0126) significantly reduced the phosphorylation level of ERK in exogenous AREG-treated PANC-1 cells.

Previous studies have suggested that NF-κB plays an essential role in the induction and maintenance of EMT (19,20). We proposed that the role of AREG-induced pancreatic cancer cell migration, invasion and EMT may be through regulation of the NF-κB signalling pathway. To evaluate this possibility, the nuclear accumulation level of NF-κB/p65 and the phosphorylation level of IκBα were analysed by western blot analysis. It was revealed that the nuclear accumulation level of NF-κB/p65 and the phosphorylation level of IκBα were significantly decreased in AREG siRNA-transfected AsPC-1 cells compared with the siRNA control group (P<0.05, Fig. 5A). In contrast, the nuclear accumulation level of NF-κB/p65 and the phosphorylation level of IκBα were significantly increased in AREG-treated PANC-1 cells. U0126 attenuated the AREG-promoted nuclear accumulation of NF-κB/p65 in PANC-1 cells (P<0.05, Fig. 5B). Furthermore, the role of the NF-κB signalling pathway in AREG-induced cell migration, invasion and EMT in pancreatic cancer cells was investigated using the selective NF-κB inhibitor QNZ. PANC-1 cells were pre-treated with 20 µM QNZ for 1 h, followed by incubation with 100 ng/ml AREG for 48 h. Based on the wound-healing assay, QNZ significantly inhibited the migratory ability of PANC-1 cells induced by AREG at 72 h (P<0.01, Fig. 5C). Transwell assay results revealed that QNZ abolished the invasive properties of PANC-1 cells induced by AREG (P<0.01, Fig. 5D). According to the western blot analysis, pre-treatment of PANC-1 cells with QNZ resulted in the upregulation of E-cadherin and ZO-1 and the downregulation of vimentin, Slug, and Snail (Fig. 3B). All of the aforementioned results indicated that inhibition of the NF-κB signalling pathway suppressed AREG-induced cell migration, invasion and EMT in pancreatic cancer cells.

The survival rate in our model was 91.7%; one mouse in the LV-AREG group died the day after the injection. All mice in the LV-AREG and LV-con groups exhibited a tumour at the site of the injection of cells into the pancreas tail. The incidence of regional and distant metastasis was lower in the LV-AREG group compared with the LV-con group (Table I; Fig. 6A). Macroscopically, the tumours appeared as firm, white, irregular masses. There had been a downward trend in tumour growth with AREG silencing in the orthotopic model of pancreatic cancer (0.74±0.31 g vs. 1.04±0.17 g); however, no significant differences were revealed between the LV-AREG group and LV-con groups (P>0.05, Fig. 6B). H&E staining confirmed pancreatic tumour development in mice (Fig. 6C).