WIPF1 antagonizes the tumor suppressive effect of miR-141/200c and is associated with poor survival in patients with PDAC

This study was approved by the Committee for the Ethical Review of Research, Fujian Medical University Union Hospital (No. 2016-ZQN-34). Human pancreatic cancer samples were collected from patients undergoing surgery at Fujian Medical University Union Hospital, Fuzhou, China, from March 2016 to July 2017. Informed consent was obtained before sample collection. All patients received curative surgery and had histologically confirmed PDAC. None of the patients received neoadjuvant radiation or chemotherapy. The stage of each patient was assessed based on the American Joint Committee on Cancer version 7 (AJCC 7). The tissue samples were placed in liquid nitrogen or RNAlater immediately after dissection and stored at −80C° until DNA and RNA extractions were performed.

The human pancreatic duct epithelial cells (HPDE) and pancreatic cancer cell lines (PANC-1, BxPC-3, HPAF-II and SW1990) were obtained from the Cell Bank, Chinese Academy of Sciences (Shanghai, China) and propagated in our laboratory by culturing in complete growth medium as recommended by the supplier. All cell lines were genotyped for identity by Cell Bank, Chinese Academy of Sciences and tested for ruling out mycoplasma contamination. Cells were incubated at 37 °C in atmospheric conditions of 20% O₂ and 5% CO₂, treated with 4 μM 5-aza-2′-deoxycitidine (5-Aza-dC) (Sigma, St. Louis, MO, USA) for 3 days, before harvesting for assays for methylation and mRNA expression.

All animal experimental protocols were approved by the Ethics Committee for Animal Research of Fuzhou General Hospital of Chinese PLA Nanjing Military Command. (No. FZZY-2016-26). The male athymic nude (BALB/c-nu) mice of 4 to 5 weeks and male NOD/SCID mice of 5 weeks were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and housed in a pathogen-free facility of Comparative Medicine Center of Fuzhou General Hospital and maintained on a 12-h light-dark cycle. Food and tap water were provided ad libitum.

Genomic DNA was extracted from tissue samples and cell lines with the QIAamp DNA Mini Kit (QIAGEN, Dusseldorf, Germany) according to the manufacturer’s instructions. DNA concentration and purity were determined based on the absorbance at 260 and 280 nm. A total of 1.5 μg of genomic DNA from each sample was converted with sodium bisulfite using an EZ DNA Methylation-Gold Kit (Zymo research, Orange County, CA, USA) according to the manufacturer’s instructions. PCR primers were designed using Epidesigner (http://​www.​epidesigner.​com) and are shown in the Additional file 1: Table S1. Polymerase chain reaction (PCR) was performed with the PCR Accessory Set (Sequenom, San Diego, CA, USA) and the following parameters: hot start at 94 °C for 10 min, followed by denaturing at 94 °C for 45 s, annealing at 62 °C for 48 s, extension at 72 °C for 1 min for 10 cycles, then denaturing at 94 °C for 45 s, annealing at 57 °C for 48 s, extension at 72 °C for 1 min for 35 cycles and final incubation at 72 °C for 3 min. Unincorporated dNTPs were dephosphorylated using the MassCLEAVE Kit (Sequenom). The reaction mixture was incubated at 37 °C for 20 min and shrimp alkaline phosphate (SAP) was then heat inactivated for 5 min at 85 °C. After SAP treatment, 2 μl of the PCR products were used for in vitro transcription and RNase cleavage in accordance with the manufacturer’s instructions (Sequenom). The samples were conditioned and spotted on a 384-pad SpectroCHIP (Sequenom) using the Sequenom MassARRAY platform (CapitalBio, Beijing, China), based on the matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry. Mass spectra were acquired via MassARRAY Compact MALDI-TOF (Sequenom) and the methylation ratios were generated by the EpiTyper software (Sequenom).

Total RNA was extracted from cells or fresh-frozen tissues with TRIzol reagent (Invitrogen, San Diego, CA, USA) by following the manufacturer’s protocol. The miR-200c and miR-141 levels were quantified by qRT–PCR using TaqMan assay kits (Applied Biosystems, Foster City, CA, USA) with U6 small nuclear RNA as an internal normalization reference. WIPF1, ZEB1, E-Cadherin and Vimentin mRNA was measured by qRT–PCR using a SYBR Premix Ex Taq (Takara, Dalian, China) and normalized for GAPDH expression. The qRT-PCR reactions were performed using an ABI Stepone plus Real-Time PCR System (Applied Biosystems). The primer sequences are listed in the Additional file 1: Table S2.

All RNA oligoribonucleotides, including miR-200c mimic, miR-141 mimic, negative control mimic (NC), miR-141/200c inhibitors (anti-miR-141 and anti-miR-200c) and their NC were obtained from Genepharma (Shanghai, China) and their sequences are shown in the Additional file 1: Table S3. The short hairpin RNA (shRNA) sequence for human WIPF1–1 was 5’-GGCCAACAGGGATAATGATTCTTCAAGAGAGAATCATTATCCCTGTTGGCCTT-3′ and WIPF1–2 was GGGAAAGCAGATTCTACTTCCTTCAAGAGAGGAAGTAGAATCTGCTTTCCCTT. Negative Control shRNA vector (Genepharma) was used as a control for RNA interference. The oligonucleotide transfection was performed using the Lipofectamine 2000 reagent (Invitrogen) and transfection was performed according to the manufacturers’ recommendations.

Human miR-141 mimics, miR-200c mimics and WIPF1 shRNA oligonucleotides were designed and cloned into the LV3-pGLVH1/GFP + Puro vector (GenePharma). The coding sequence of WIPF1 without its 3’-UTR was amplified and cloned into another lentiviral expression vector, LV5-pGCMV/MCS/EF1a/GFP + Puro vector (GenePharma), to produce WIPF1. BxPC-3 and PANC-1 cells were infected with lentiviruses following the instructions of the GenePharma Recombinant Lentivirus Operation Manual. At 72 h after lentivirus infection, the medium was replaced with fresh medium containing puromycin to select for stably infected cells.

The region of the 3′-untranslated region (UTR) of human WIPF1 containing three putative miR-200c binding sites and one putative miR-141 binding site were selected to generate four mutant variants (TargetScan, Additional file 1: Figure S1A and B). Mutant 3’-UTR of human WIPF1 mRNA was generated by site-specific mutagenesis of the wild-type 3’-UTR segment of human WIPF1 mRNA using altered sequence in the complementary site (Additional file 1: Table S1B). The wild-type (WT) and four mutant variants (MUT) derived from the 3’-UTR segment of human WIPF1 mRNA were amplified and subcloned into restriction sites downstream of the luciferase reporter gene in the pmirGlo-vector (GenePharma). Luciferase activities were assayed using a Dual-Luciferase Reporter Assay system (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Transfections were performed in duplicate and repeated at least three times in separate experiments.

Cell proliferation was evaluated using the Cell Counting Kit-8 (CCK-8, Dojindo, Tokyo, Japan). BxPC-3 and PANC-1 cells were seeded in 96-well culture plates at 5 × 10³ cells/well, transfected with the indicated miRNA and incubated for 1, 2, 3, 4 and 5 days. After washed in PBS, 100 μl medium containing 10% CCK-8 solution was added to each well and incubated for 2 h at 37 °C. Samples are read directly in the wells using an absorbance of the 450 nm wavelength by an enzyme linked immunosorbent assay (ELISA) plate reader.

Cell migration and invasion were measured by a Transwell migration plates (24-well, 8 μm pore size, Corning Costar, NY, USA) and chamber invasion assay (Matrigel-coated membrane, Corning Costar). The upper chamber contained pancreatic cancer cells in serum-free medium, and the lower chamber contained culture medium with 10% fetal calf serum. After incubation for 8 h (for migration assay without Matrigel-coated membrane) or 24 h (for invasion assay with Matrigel-coated membrane) at 37 °C, non-invading cells were removed with cotton swabs, and cells that had invaded to the underside of the membrane were stained with 0.1% crystal violet. Then, the invaded cells were counted under an inverted microscope.

Pancreatic cancer cells (1 × 10⁷ cells in 100 μl PBS) were injected subcutaneously into the right axilla of each male athymic nude mice. The length (L), width (W) and height (H) of the tumors were measured weekly. The tumor volume (V) was calculated using the following formula: V = (L × W × H) × 0.5. Mice were sacrificed 28 days after the injection. Tumors were removed and weighed. Each group contained at least five animals.

An in vivo metastasis model was established as previously described [31]. Pancreatic cancer cells (1 × 10⁶ cells in 50 μl PBS) were injected into the spleens of NOD/SCID mice. Ten weeks after the injections, the animals were sacrificed under deep anesthesia (pentobarbital sodium [30 mg·kg^− 1]) and the liver and lungs were harvested. Tumor metastases were quantified by counting the number of metastatic colonies on one histological section at the middle portion of each liver or lung sample. Each group contained at least five animals.

Western blotting of WIPF1, YAP and TAZ in pancreatic cancer cells was with the methods described previously [32].

RNAseq and clinical data of 177 patients defined by the TCGA pathologist as PDAC were obtained from the TCGA Data Portal (https://​tcga-data.​nci.​nih.​gov). All data were downloaded from the November 22, 2017 standard dataset. For RNAseq data, expression levels were TPM-normalized and ENSG-ID transformed.

We quantified the promoter DNA methylation levels of the miR-200 family members miR-200a/200b/429 and miR-200c/141 using the MassARRAY compact system with MALDI-TOF mass spectrometry which permits high-throughput identification of methylation sites and semi-quantitative measurement at single or multiple CpG sites. The CpG sites of the promoter region used for the methylation assays in each miR-200 family member are indicated (Fig. 1a). As shown by the heatmap (Fig. 1b), the levels of CpG methylation of miR-200a, miR-200b and miR-429 were significantly lower (hypomethylation) in the 10 paired human pancreatic cancer tissues (A01-A10), compared to the surrounding non-cancerous tissues (B01-B10, hypermethylation). This result is similar to that was previously reported [14]. In contrast, the levels of the CpG methylation of miR-141 and miR-200c were significantly higher in human PDAC. To confirm this result, we derived the mean values from all the CpG sites and found that the levels of CpG methylation of miR-141/200c were significantly higher in the PDAC tissues compared to the surrounding non-cancerous tissues (Fig. 1c). When the mean methylation values were calculated from each individual CpG of all the 10 paired samples, the result was similar (Fig. 1d). The opposite result for the miR-200a/200b/429 promoter was obtained as expected (Additional file 1: Figure S2A and B). Collectively, these results indicated that the miR-141/200c promoter is hypermethylated while which ofmiR-200a/200b/429 is hypomethylated in PDAC.

To verify whether the CpG hypermethylation mediates the miR-141/200c silencing, we evaluated miR-141 and miR-200c expression in the 10 paired human PDAC tissues by qRT-PCR. As expected, the relative expression levels of miR-141 and miR-200c in PDAC were significantly lower than that of the non-cancerous tissues (Fig. 2a). To further confirm this result, we investigated the correlation between the level of CpG methylation of miR-141 promoter and the level of miR-141 expression in 37 patients with PDAC and found an inverse correlation (r = − 0.627, P < 0.001, Fig. 2b and Additional file 1: Table S4). Similar result was obtained for miR-200c (r = − 0.782, P < 0.001, Fig. 2c). We then analyzed the CpG methylation of miR-141 and miR-200c DNA promoter in all the cell lines tested. All four human PDAC cell lines (PANC-1, BxPC-3, HPAF-II and SW1990) showed significantly higher methylation levels, compared to the benign pancreatic tissue human pancreatic duct epithelial cells (HPDE, Fig. 2d). We treated the pancreatic cell lines with the DNA-demethylating agent 5-Aza-dC and found that 5-Aza-dC decreased the level of promoter methylation of miR-141 and miR-200c (Fig. 2e) and restored the expression of miR-141 and miR-200c (Fig. 2f). These results indicated that CpG hypermethylation silenced the expression of miR-141 and miR-200c in PDAC.

To determine whether epigenetically regulated miR-141/200c confers tumor-suppressive function in PDAC cells, we stably infected BxPC-3 and PANC-1 cells with a retroviral construct carrying miR-141, miR-200c or with an empty vector as control. We verified the overexpression of these constructs (Additional file 1: Figure S3A and B) and showed that forced expression of anti-miR-141 and anti-miR-200c suppressed the expression of miR-141 and miR-200c (Additional file 1: Figure S3C and D). Using the CCK8 assay, we found that overexpression of miR-141 significantly inhibited cell proliferation while miR-141 inhibitor (anti-miR-141) stimulated cell proliferation (Fig. 3a and b). Interestingly, overexpression of miR-200c or miR-200c inhibitor (anti-miR-200c) showed no significant effect on the cell proliferation (Fig. 3a and b). Using xenograft by injecting the PDAC cells subcutaneously into the athymic nude mice, we found that overexpression of miR-141 but not miR-200c significantly suppressed the tumor growth (Fig. 3c and d).

We next investigated the role of miR-141 and miR-200c on the cell migration and invasion of PDAC using transwell chambers with or without Matrigel-coated membranes. Both miR-141 and miR-200c suppressed the migration and invasion of PANC-1 and BxPC-3 cells (Fig. 4a-d). Furthermore, miR-141 inhibitor and miR-200c inhibitor reverse the inhibition (Additional file 1: Figure S4A and B). We also determined that miR-141 and miR-200c suppressed the expression of epithelial-to-mesenchymal transition (EMT) markers ZEB1 and Vimentin while stimulated the expression of E-cadherin, a suppressor of EMT (Fig. 4e-g and Additional file 1: Figure S4C). These results indicate that both miR-141 and miR-200c suppress the stemness of PDAC cells.

To investigate the role of miR-141 and miR-200c on PDAC metastasis, PDAC cells were infected with a lentiviral vector expressing miR-141 or miR-200c, or an empty vector as control and the resulting cells were then injected into the spleens of NOD/SCID mice. Ten weeks after the injection, the animals were euthanized, and the organs were harvested. The overexpressing miR-141 significantly suppressed the tumor growth in the spleen where the cells were injected, while miR-200c had little effect (Additional file 1: Figure S4D). However the mice injected with the cells overexpressing miR-141 or miR-200c had significantly lower number of metastatic nodules in the liver and lungs compared with the mice injected with the cells infected with an empty vector (Fig. 4h and i). These findings demonstrate that miR-141 but not miR-200c retained the ability to inhibit the cell proliferation in vitro and the growth of primary tumor in xenograft, while both were capable of suppressing the migration, invasion and metastasis of PDAC cells.

To investigate the mechanism underlying the tumor suppressive function of miR-141 and miR-200c, we set out to identify the putative binding sites of their potential target genes. It is well established that microRNAs regulate mRNA expression by targeting the 3′-UTR of mRNAs with a complementary seed sequence [33]. With the use of TargetScan software program, we found that 3′- UTR of WIPF1 contained putative miR-141/200c binding sites (Additional file 1: Figure S1A). To verify their interaction, a dual-luciferase reporter system assay was performed. The wild-type (WT) or the mutant (MUT) miR-141 or miR-200c binding sequences in the 3′-UTR of human WIPF1 (Additional file 1: Figure S1B) was cloned to the downstream of firefly luciferase reporter gene, then co-transfected with miR-141 or miR-200c mimic or an unrelated sequence as control into 293 T cells. As is shown in Fig. 5a, the relative luciferase activity with WT but not the mutant 3′-UTR of WIPF1 was significantly suppressed by the miR-141 or miR-200c mimic. These results indicate that both miR-141 and miR-200c suppress WIPF1 expression through binding to their binding sites on the 3′-UTR of WIPF1. Consistent with these results, using HPDE as positive control we found that both the mRNA and protein levels of WIPF1 in both PANC-1 and BxPC-3 cells were downregulated by the miR-141 and miR-200c mimics, while upregulated by anti-miR-141 and anti-miR-200c (Fig. 5b, and Additional file 1: Figure S5).

To further confirm that WIPF1 is the target of miR-141/200c, we investigated if there is a correlation between their expression. We examined 37 individual human pancreatic cancer tissues and found that the expression of miR-141 inversely correlated with the expression of WIPF1 mRNA (r = − 0.734, p < 0.001, Fig. 5c, left panel). Similar result was obtained for the expression of miR-200c and WIPF1 (r = − 0.733, p < 0.001, Fig. 5c, right panel). We also examined 10 paired pancreatic cancer tissues and the surrounding non-cancerous tissues and found that mRNA expression of WIPF1 was highly elevated in the PDAC tissues compared to the surrounding non-cancerous tissues (P = 0.005, Fig. 5d). These results indicate that the silencing of miR-141 and miR-200c by CpG methylation in human PDAC leads to increased expression of WIPF1.

To investigate the functional roles of WIPF1 in PDAC, we constructed a cell line with stable knockdown of WIPF1 using a lentivirus vector expressing a short hairpin WIPF1 (shWIPF1) and a control cell line with vector only (shCtrl). Knockdown of WIPF1, as confirmed by the diminished protein and mRNA expression (Additional file 1: Figure S6A-D), reduced cell proliferation (Fig. 6a and b) and tumor growth of primary site (Fig. 6c). This result is nearly identical to that obtained with the overexpression of miR-141 (Fig. 3a-d). Knockdown of WIPF1 also inhibited migration and invasion of both PANC-1 and BxPC-3 cells (Fig. 6d-g), also nearly identical to that obtained with the overexpression of either miR-141 or miR-200c (Fig. 4a-d, h and i). Moreover, knockdown of WIPF1 decreased the number of metastatic nodules in the liver and lungs (Fig. 6h-j). This result is also nearly identical to that obtained with the overexpression of either miR-141 or miR-200c (Fig. 4h and i). These data demonstrate that in opposite to miR-141/200c, WIPF1 promotes cell growth, migration, invasion and metastasis of PDAC cells.

To determine whether miR-141/200c targets WIPF1 to suppress the invasion and metastasis of PDAC cells, PANC-1 and BxPC-3 cells were transfected with a plasmid carrying the cDNA that encodes the entire coding sequence of WIPF1 but with the deletion of the 3’-UTR (WIPF1/− 3’-UTR). We found that WIPF1/− 3’-UTR was resistant to the suppressive effect of miR-141 and miR-200c on its expression (Additional file 1: Figure S7A). Moreover, WIPF1/− 3’-UTR overcame the suppressive effect of miR-141/200c on the migration, invasion and metastasis of PDAC cells (Fig. 7a-d and Additional file 1: Figure S7B-D). These results demonstrate that WIPF1 is indeed a functional target of miR-141/200c.

To determine the impact of WIPF1 expression on the survival of patients with pancreatic cancer, we identified 177 PDAC cases from the TCGA database and examined the correlation between the level of expression of WIPF1 and overall survival (OS). These patients had a mean age of 64.7 and majority (85%) had stage II and III disease (Additional file 1: Table S5). We found 38 cases with low expression and 139 with high expression of WIPF1 in the tumor tissues (cut-off value = 4.81; the partition was constructed by the length of OS). Using Kaplan-Meier estimate, we found that the patients with low WIPF1 expression was associated with significantly longer OS compared to the patients with high WIPF1 expression (P = 0.03). The median OS was 38 months in the low WIPF1 expression group versus 19 months in the high WIPF1 expression group (Fig. 7e). This result indicates that WIPF1 expression is associated with a more aggressive disease course and possibly resistance to treatment in human PDAC.

WIPF1 was found to drive tumor progression by stabilizing the YAP/TAZ complex [32]. Previously it was also shown that miR-141 inhibited the expression of YAP1 in pancreatic cancer [17]. We investigated if miR-141/200c suppress YAP/TAZ expression by inhibiting WIPF1. We found that expression of both YAP and TAZ was indeed suppressed by the overexpression of miR-141 or miR-200c in both PANC-1 and BxPC3 cells (Fig. 8a). While miR-141 or miR-200c inhibitor (anti-miR-141 or anti-miR-200c) increased the expression of YAP and TAZ (Fig. 8b). Knockdown of WIPF1 decreased the expression of YAP and TAZ (Fig. 8c). In addition, forced expression of the mutant WIPF1 with the deletion of its 3′-UTR (WIPF1/− 3’-UTR) stimulated the expression of YAP/TAZ and blocked the suppressive effect of miR141/200c on YAP/TAZ expression (Fig. 8d). Hence, WIPF1 mediates the suppressive effect of miR-141/200c on YAP/TAZ. These results demonstrate that silencing of miR-141/200c stimulates PDAC growth and metastasis by activating the WIPF1-YAP/TAZ pathway, leading to poor patient survival (Fig. 8e).