From 2012 to 2014, 26 PC patients (15 males and 11 females), ranging from 34 to 72 years of age (mean age, 53.6 years), who underwent radical resections were recruited in the current investigation with informed consent. Research consent was approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. All patients were aware of the potential risks and complications of the proposed treatment scheme. The corresponding non-tumor tissues were collected at least 3 cm away from the margin of the tumor. All the specimens including tumor and paired non-tumor tissues were cut into small pieces and placed in liquid nitrogen immediately. All of the samples were submitted for routine pathologic evaluation and diagnostic confirmation. The human PC cell lines Capan-1, CFPAC-1, SW-1990, BxPC-3 and PANC-1 were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA) and PATU-8988 was purchased from Nanjing KeyGen Biotech Co., Ltd. (Nanjing, China). All tumor cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) or RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) at 37°C in 5% CO₂. Total RNA and genomic DNA were isolated according to previously reported methods (15). Additionally, PC tissue array of 29 patients was purchased from Shanghai Outdo Biotech Co., Ltd (Shanghai, China).

The full-length cDNA of LIFR was obtained using a reverse transcription kit (Takara) and total RNA was extracted by RT-PCR from human PC tissues using TRIzol reagent kit (Invitrogen, Carlsbad, CA, USA). The primers for the coding sequence (CDS) of double-strand DNA fragments of LIFR were: 5′-GGATCCAT GATGGATATTTACGTATGT-3′ (forward) and 5′-ACGCGT TTAATCGTTTGGTTTGTTCTG-3′ (reverse), and were subcloned into the pWPI-GFP vector to generate pWPI-GFP/LIFR. Transfection of the constructed plasmid and empty vector into PATU-8988 cells was performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's procedure. Cells that had been transfected with the constructed plasmid were then selected by antibiotic resistance in cell culture medium containing 1,500 µg/ml G418 to obtain cell strains with stable expression of LIFR. After 6 weeks of culture in the presence of G418, the remaining cells were isolated. Positive stable clones with pWPI-GFP/LIFR were maintained with pWPI-GFP empty vector transfection (PATU-8988/vector) as control.

Based on human LIFR gene data, synthesized DNA nucleotide fragments, encoding shRNA for knockdown of LIFR, were inserted into pL/shRNA/F lentiviral vector to obtain pL/shRNA/shR-LIFR. The constructions were further confirmed by DNA sequencing. The packaging of the lentivirus and establishment of stable knockdown cell clones were performed as mentioned above. The sequences of these synthesized oligonucleotides for RNAi LIFR were as follows: 5′-GATCCCCGCTGATTTCTCAACCTCTACATTCAAGAGATGTAGAGGTTGAGAAATCAGCTTTTTGGAA-3′ (forward) and 5′-AGCTTTTCCAAAAAGCTGATTTCTCAACCTCTACATCTCTTGAATGTAGAGGTTGAGAA ATCAGCGGG-3′ (reverse). Then, the pL/shRNA/shR-LIFR lentiviral vector was transfected into Capan-1 cells. Stable clones (Capan-1/sh) were established by selection with 5 µg/ml blasticidin. The irrelevant nucleotides in sh-NC did not target any annotated human genes and served as a negative control.

Paraffin-embedded tissue samples from PC specimens underwent a heat pre-treatment of 60°C for 1 h, then dewaxed in xylene, rehydrated in a series of ethanol and treated with 0.01 mol/l citrate buffer (pH 6.0) for antigen retrieval. After inhibition of endogenous peroxidase activity for 30 min with methanol containing 0.3% H₂O₂, the sections were stained with rabbit anti-LIFR antibody (C-19) (1:300; Santa Cruz Biotechnology, Santa Cruz, CA, USA) or the anti-Ki67 antibody (ab15580) (1:100; Abcam, Cambridge, MA, USA) at 4°C overnight. For tissue arrays, after being dewaxed, hydrated and blocked of non-specific binding sites, the microarray was incubated with rabbit anti-LIFR antibody (C-19) (1:300; Santa Cruz Biotechnology) at 4°C overnight. The following experimental procedure was performed according to the manufacturer's instructions for the LSAB+ kit (Dako, Carpinteria, CA, USA). Three pathologists who were blinded to any patient data independently examined the cellular location of LIFR and compared the staining between the tumor and normal tissues. Immunohistochemistry staining score = positive cell score + staining intensity score. The percentage of positive cells was classified according to five grades (percentage scores): <10% (grade 0), 10–25% (grade 1), >25–50% (grade 2), >50–75% (grade 3) and >75% (grade 4). Immunohistochemical staining intensity was graded as follows: 0 (no staining), 1 (bright yellow), 2 (orange) and 3 (brown). The total scores of ≤3, >3–5, and ≥6 were defined as negative, weak and strong positive, respectively.

Tumor and cell samples were collected and lysed using RIPA buffer (Solarbio, Beijing, China) in the presence of protease inhibitor cocktail and protein concentration was measured by BCA protein assay kit (both from Pierce, Rockford, IL, USA). An equal amount of total cellular protein was electrophoresed by 10% SDS-PAGE, then transferred to polyvinylidene difluoride (PVDF) membranes. The membranes were blocked with 5% skim milk for 2 h, and then incubated with primary antibodies overnight at 4°C. Primary antibodies were as follows: LIFR (C-19) (1:1,000; Santa Cruz Biotechnology), vimentin (D21H3) (1:1,000), N-cadherin (D4R1H) (1:1,000), β-catenin (D10A8) (1:1,000), slug (C19G7) (1:1,000) (all from Cell Signaling Technology, Danvers, MA, USA) and GAPDH (Abcam). After the membranes were incubated with the secondary antibody for 2 h at room temperature, the proteins were visualized using an enhanced chemiluminescence detection system (Amersham Biosciences, Piscataway, NJ, USA) according to the manufacturer's protocol.

For the colony formation assay, 1,000 cells were layered onto 6-well plates and cultured at 37°C for ~14 days in order to let the colonies develop. After visible colonies of cells were observed, the cells were washed twice with phosphate-buffered saline (PBS), and fixed with 4% paraformaldehyde and stained with crystal violet for 30 min. Colonies containing 50 cells or more were counted.

For the cell migration assay, a total number of 1×10⁵ cells were re-suspended in serum-free culture solution and added to the Transwell upper chambers (8 µm; 24-well format; Corning, Lowell, MA, USA). To measure the invasion, the upper chamber's membrane was pre-coated with Matrigel (BD Biosciences, San Diego, CA, USA) according to the manufacturer's protocols before cell solution was added to the upper chamber. The lower chambers included 0.6 ml of medium containing 10% FBS as a chemoattractant. After 24 h of culturing, the chambers were fixed with 10% methanol and stained with 0.5% crystal violet solution for 15 min, and then washed by PBS. Cells in the lower chamber were observed and counted in six random fields under an inverted microscope.

Cells were layered onto 6-well plates and cultured to confluency. The 200-µl pipette tips were used to scratch three separate wounds on the monolayer of cells. Plates were washed with fresh medium after the cells had been cultured for 0, 12 or 24 h, and then photographed. The distances between wound edges were measured.

PC xenografts were established in nude mice. Four-week-old male BALB/c nude mice were purchased from the Institute of Zoology, Chinese Academy of Sciences of Shanghai. The animal research was approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. All experiments were performed in accordance with the official recommendations of the China Zoological Society and animals received humane care according to the criteria outlined in the ‛Guide for the Care and Use of Laboratory Animals'. Briefly, Capan-1/sh, Capan-1/nc, PATU-8988/vector and PATU-8988/LIFR cells were re-suspended in PBS (pH 7.4). The suspension, containing 1×10⁶ cells, was subcutaneously injected into the right flank of nude mouse. The length (L) and width (W) of each tumor were measured every 7 days with a digital caliper, and the tumor volume was calculated using the formula: Tumor volume = (Width² × Length)/2. Mice were sacrificed by cervical decapitation 5 weeks after injection. Tumors were weighed and fixed for hematoxylin and eosin (H&E) and immunohistochemical staining. The tumorigenic experiments in vivo were performed with 5 mice in each group.

The tail vein injection assay was employed to evaluate the role of LIFR in tumor metastases in vivo. The stable clones of Capan-1/sh, Capan-1/nc, PATU-8988/vector and PATU-8988/LIFR were injected into athymic nude mice via tail veins. After 6 weeks, the lung metastasis lesions were detected by micro-PET/CT. PET/CT imaging was performed by the Department of Nuclear Medicine, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine according to Meng's method (16). Then, the mice were sacrificed and all the suspicious lung metastasis sites were evaluated by histological examination.

Statistical analyses were performed using SPSS 13.0 software (SPSS, Inc., Chicago, IL, USA). For comparison among groups, an analysis of variance (ANOVA) and Student's t-test were used. Differences between tumor volumes were assessed by Mann-Whitney U test. The Chi-square test was used for the comparison of categorical data. A P-value <0.05 was considered to indicate a statistically significant result.

To illuminate the expression of LIFR in multiple cancer tissues, Oncomine data-mining analysis analyses were first performed. Pei Pancreas Statistics confirmed that with all four probes testing it showed significantly decreased levels of LIFR mRNA in PC compared with that in normal pancreas tissues (Fig. 1A and B, available at Oncomine website).

To confirm our above findings, we then examined the expression level of LIFR in 26 PC clinical cases and 29 PC tissue microarrays, including 53 pancreatic duct adenocarcinomas and 2 pancreatic adenosquamous carcinomas. Immunohistochemical staining of 55 paired tissues showed that LIFR expression was significantly lower in the tumor tissues than that in the non-tumor tissues (negative, 32:4, positive, 21:16; strong positive, 2:35) (Table I) (Fig. 1C and D). Furthermore, LIFR was downregulated in 70.9% (39/55) of the PC cases. These results were consistent with those from the Oncomine analyses. PC tissue microarray was further employed to examine the correlation between LIFR expression and clinicopathological features. The results showed that downregulated LIFR was associated with lymph node metastasis (P=0.014), local invasion (P=0.047) and TNM stage (P=0.002), but not with other clinicopathological factors including gender, age and tumor location (Table II). Further analysis showed that only TNM stage of the patients was correlated to LIFR expression (P=0.002), not gender, age, tumor location or tumor size (Table III). All these findings suggested that downregulation of LIFR plays a critical role in PC development and LIFR may be an independent prognostic factor of PC.

Based on the above results, to establish a paired differential expression model for further study, we first quantitatively analyzed the expression of LIFR in a series of PC cell lines (CFPAC-1, PATU-8988, Capan-1, SW-1990, BxPC-3 and PANC-1). Among these PC cells, the protein level of LIFR in Capan-1 cells was higher than that in the other PC cells and LIFR in highly aggressive PATU-8988 cells was the lowest (Fig. 2A). Therefore, Capan-1 cells were selected for the knockdown assay to reveal the role of LIFR in the following study. The lentiviral-mediated shRNA was employed to knockdown LIFR in the Capan-1 cells. Meanwhile, we used the pWPI-GFP/LIFR vector to generate a PATU-8988/LIFR cell line ectopically overexpressing LIFR. After the establishment of stable clones of shRNA-mediated knockdown of LIFR in the Capan-1/sh cells and effective overexpression of LIFR in the PATU-8988/LIFR cells, western blotting was employed to confirm the expression level of LIFR in the Capan-1/sh and PATU-8988/LIFR cells (Fig. 2B and C). Colony formation assays indicated that silencing of endogenous LIFR in the Capan-1/sh cells significantly increased the ability of colony formation compared with the mock cells (328.5±27.5 vs. 106.5±8.5; P=0.016) (Fig. 2D and E). On the contrary, overexpression of LIFR in the PATU-8988/LIFR cells markedly impaired the ability of colony formation (177.5±9.5 vs. 438.0±15.0; P=0.005) (Fig. 2F and G). Collectively, these results indicated that LIFR may play a functional role in pancreatic carcinogenesis.

A close correlation between clinical invasive characteristics in PC patients and expression level of LIFR was noted. This suggested that LIFR may play a negative regulatory role in PC tumor metastasis. To further confirm the effect of LIFR on the metastasis of PC cells, Transwell migration and invasion assays were performed. As shown in Fig. 3A and B, migration and invasion in the Capan-1/sh cells were significantly enhanced after silencing of endogenous LIFR (migration, 750.0±18.0 vs. 382.5±15.5, P=0.004; invasion, 281.0±16.0 vs. 97.0±10.0, P=0.010). In contrast, by overexpressing LIFR in PC cells, the migratory and invasive abilities of the PATU-8988/LIFR cells were apparently reduced (migration, 215.0±9.0 vs. 490.0±11.0, P=0.003; invasion, 130.5±7.5 vs. 330.5±18.5, P=0.010) (Fig. 3C and D).

Furthermore, we quantitatively investigated the effect of LIFR on migratory ability by wound healing assay. The results in Fig. 3E and G showed that silencing of LIFR significant shortened the distance between the wound edge in the Capan-1/sh cells compared with the Capan-1/nc cells (P=0.034). Consistently, we observed a longer distance in wound healing after overexpressing LIFR in the PATU-8988/LIFR cells (P=0.013) (Fig. 3F and G). Taken together, these results indicated that the aggressive and highly metastatic phenotype of PC cells could be regulated by LIFR in vitro.

For in vivo confirmation, a recombinant lentivirus harboring LIFR was transfected into the PATU-8988 cells. The stable clone expressing ectopic PATU-8988/LIFR was subcutaneously injected into the flank of each athymic nude mouse, and an equal volume of cells transfected with the empty vector was injected into the opposite flank of the same mouse as the negative control. As shown in Fig. 4A and B, the PATU-8988/LIFR cells caused smaller tumor masses than the mock vector control after 5 weeks of observation (PATU-8988/LIFR, 47.9±12.5 mm³; PATU-8988/vector, 407.2±143.2 mm³; P<0.05). Meanwhile, tumors derived from the offspring subclones with LIFR-overexpressing cells were significantly lighter than those in the control group (PATU-8988/LIFR, 0.106±0.032 g; PATU-8988/vector, 0.828±0.273 g; P<0.05) (Fig. 4C). Furthermore, tumor sections from the nude mouse model were immunohistochemically stained for Ki67. We observed that Ki67 expression was decreased in the PATU-8988/LIFR group compared with the PATU-8988/vector group (Fig. 4D). As expected, opposing results were observed in the LIFR-silenced group compared with the control group. Tumors derived from the offspring subclones with inhibited LIFR were significantly larger and heavier than those in the control after 5 weeks of observation (tumor volume, 355.5±26.6 vs. 163.1±25.3 mm³, P<0.01; tumor weight, 0.712±0.055 vs. 0.326±0.052 g, P<0.01) (Fig. 4B and C). Thus, the data from the in vitro and in vivo experiments revealed that LIFR negatively regulates the tumorigenesis of PC cells.

Based on the results above, an in vivo model was employed to further confirm the effect of LIFR on tumor metastasis. Stably transfected Capan-1/sh cells (1×10⁶ cells) were injected into the nude mouse via tail vein to observe the distant metastasis of PC cells in vivo. After 6 weeks, we observed that the weight of mice in the Capan-1/sh group was significantly lower than that in the Capan-1/nc group (Fig. 4E; P<0.01). Lung metastasis was then examined by micro/PET-CT and confirmed by pathological study. Intriguingly, 5/5 mice injected with Capan-1/sh cells developed larger and greater numbers of nodules of metastatic lung tumors whereas 3 metastatic tumors out of 5 were observed in the Capan-1/nc mice (Fig. 4F; P<0.01). The presence of lung metastases in these mice was also confirmed by histological analysis (Fig. 4G). Opposing results were observed in the LIFR-overexpressing groups compared with the control groups (Fig. 4E and F). In addition, we identified spinal metastases in one mouse from the Capan-1/sh group, and none in the Capan-1/nc group (Fig. 4H). Based on the in vitro and in vivo data, the results supported that LIFR negatively regulated the metastasis of PC cells both in vitro and in vivo.

Research has demonstrated that epithelial-mesenchymal transition (EMT) is associated with cancer metastasis. Based on the results above, we further studied the possible molecular mechanisms by which LIFR contributes to PC cell proliferation and migration. As downstream regulation pathways of EMT, the activation of vimentin, β-catenin, slug and E-cadherin was detected to explore the underlying regulatory network. As shown in Fig. 5A and B, overexpression of LIFR in the PATU-8988/LIFR cells significantly inhibited the expression of vimentin, β-catenin and slug and induced the expression of E-cadherin. Opposing results were observed in the LIFR-silenced groups compared with the control groups (Fig. 5A and B). Collectively, these findings suggest that LIFR negatively regulates the metastasis of PC cells via EMT.