Lipoxin A4 reverses mesenchymal phenotypes to attenuate invasion and metastasis via the inhibition of autocrine TGF-β1 signaling in pancreatic cancer

5(S), 6(R)-Lipoxin A4 (LXA4) and 5(S), 6(R)-7-trihydroxymethyl Heptanoate (BML-111) were purchased from Cayman Chemical (Ann Arbor, MI, USA). Antibodies against E-cadherin, N-cadherin, Vimentin, and Snail were obtained from Cell Signaling Technology (Boston, MA, USA). The antibodies against FPRL1 and 15-LOX were purchased from Abcam (Cambridge, UK). The anti-TGF-β1 neutralizing antibody was obtained from R&D Systems (Minneapolis, MN, USA), and recombinant human TGF-β1 (rhTGF-β1) was obtained from PeproTech (Rocky Hill, NJ, USA). Anti-β-actin was purchased from Sigma-Aldrich (St. Louis, MO, USA).

The human pancreatic cancer cell lines Panc-1 and MIA PaCa-2 were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Both of these cell lines were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS) (ExCell, South America), 100 U/mL penicillin and 100 μg/mL streptomycin (Gibco). The cells were cultured at 37 °C in an atmosphere of 5% CO2.

Matrigel invasion assays were performed to examine the invasive capacity of cancer cells. The upper surface of the membrane was coated with Matrigel (BD Biosciences, Franklin Lakes, USA). Cancer cells (5 × 10⁴) were suspended in FBS-free media then seeded in the upper chamber. In the lower chamber, DMEM supplemented with 10% FBS was added to generate a gradient to drive cell migration. Forty-eight hours later, the medium was aspirated and noninvasive cells in the upper chamber were removed by a cotton swab. The invading cells were then fixed in 4% paraformaldehyde and stained by crystal violet; the number of cells on the membrane in 10 random fields was counted under a microscope. The values reported here are the averages of triplicate experiments.

Wound-healing assays were performed to evaluate the migration ability of cancer cells. Pancreatic cancer cells were seeded in fibronectin-coated 6-well plates. After the cells reached the appropriate confluence, they were starved in media with 1% FBS overnight, and then, the monolayers were scratched with a 10-μL pipette tip. The cancer cells were washed 3 times with PBS and cultured in FBS-free medium. Photographs of the same locations were obtained at 0 and 24 h under a phase-contrast microscope at 200× magnification. The original borders were marked by red lines, and 24 h later, the number of cells between the borders was counted. This assay was repeated three times and the averages were calculated as the final results.

Panc-1 and MiaPaCa-2 cells were lysed in RIPA lysis buffer with proteinase inhibitor (Roche, Mannheim, Germany) as previously described [13]. Total protein was electrophoresed in a 10% SDS-PAGE gel and transferred to PVDF membranes (Roche). The membranes were blocked for 2 h in TBS containing 0.1% (vol/vol) Tween-20 and 10% (wt/vol) nonfat dry milk powder and then incubated with the primary antibodies overnight at 4 °C. Following the incubation with the secondary HRP-coupled antibodies for 2 h at room temperature, the membranes were washed with 0.1% TBS-Tween-20, and the immunocomplexes were detected using the enhanced chemiluminescence (ECL) kit and a Molecular Imager ChemiDoc XRS System (Bio-Rad Laboratories, Hercules, CA, USA).

SiRNA targeting FPRL1 (siFPRL1: 5′-CGGUUUGUCAUUGGCUUUATT-3′, 5′-UAAAGCCAAUGACAAACCGTT-3′) and a negative control (siNC: 5′-UUCUCCGAAGGUGUCACGUTT-3′, 5′-ACGUGACACGUUCGGAGAATT-3′) were synthesized by GenePharma (Shanghai, China). Cells (2 × 10⁵) were seeded into 6-well plates and treated with 100 nM siRNA using Lipofectamine™ 2000 (Invitrogen, CA, USA) according to the manufacturer’s instructions. The cells were used for further experiments after 48 h of transfection.

Cancer cells were washed with phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde. After permeabilization with 0.5% Triton X-100 and blocking in 5% bovine serum albumin (BSA), the cells were incubated with primary antibodies overnight at 4 °C then incubated with fluorescence-conjugated secondary antibodies at room temperature for 1 h. The nuclei were stained with 4′, 6-diamidino-2-phenylindole (DAPI) for 5 min. Images were pseudocolored using a Zeiss Instruments confocal microscope.

Cells were treated by corresponding agents for 24 h then cultured in FBS-free medium for 72 h. Then, the culture medium was collected and centrifuged at 1500 rpm for 5 min. The supernatant was frozen at −80 °C. TGF-β1 in the supernatant was detected by a commercial ELISA kit (R&D Systems) according to the manufacturer’s instructions.

Sixteen five-week-old female BALB/c nude mice were purchased and maintained in the animal center of the Medical College of Xi’an Jiaotong University, China. All experiments were approved by the Ethical Committee of the First Affiliated Hospital of Xi’an Jiaotong University.

A reliable liver metastasis model was established according to the methods of our previous study [14]. After anesthetization with 4% chloral hydrate (0.1 mL/10 g), a small incision, which was carefully anatomized to expose the spleen, was made in the left abdominal flank of the mice. Next, Panc-1 cells (5 × 10⁵), which were suspended in 20 μL of Ca²⁺- and Mg²⁺-free HBSS were injected slowly into the subsplenic membrane with a 30-gauge needle. The presence of a vesicle in the spleen was the criterion for successful inoculation; any mice in which leakage occurred were excluded from further analysis. The spleen was replaced and the peritoneum and skin were sutured. Finally, the liver metastasis model was successfully established in twelve mice.

The mice were randomly divided into two groups either treated with vehicle or BML-111. The administration schedule is shown in Fig. 5a. At the end of the eighth week, after the body weight was recorded, mice were sacrificed, and the livers and primary splenic tumors were gently removed. The weights of the livers and the visible metastatic lesions of each mouse were measured. Each primary tumor was randomly dissected into 6 pieces, which were observed under a microscope. The relative areas of the metastatic lesions were measured by Image Pro Plus 6.0 software, and the average number of relative metastatic areas per mouse was calculated.

The tumor samples were also fixed in 4% paraformaldehyde and embedded in paraffin. Serial sections of 4 mm were cut for hematoxylin and eosin (H&E) staining and immunohistochemical staining for TGF-β1 and Snail, as previously described [14].

Sixty-six paraffin-embedded pancreatic cancer samples from patients with complete medical records were collected from the First Affiliated Hospital of Medical College, Xi’an Jiaotong University. Sixty-two patients were diagnosed with pancreatic ductal adenocarcinoma, and all of the patients had undergone chemotherapy or radiation therapy prior to surgery. The histological grade and TNM status were confirmed by experienced pancreatic pathologists. All of the samples were subjected to immunohistochemical staining for 15-LOX, FPRL1 and TGF-β1. The staining was scored by two independent investigators according to the methods described previously [15, 16] as negative (0), weak (1), moderate (2) or strong (3).

The data are presented as the mean ± standard deviation (SD). Differences between two groups were determined by Student’s t test, while differences among more than two groups were analyzed by analysis of variance (ANOVA) with Dunnett’s test for post-hoc analysis. The correlation between the LES and the clinical-pathological features was determined by Pearson χ² test. P < 0.05 was considered significant. All statistical analyses were performed using SPSS 18.0 software.

To confirm the inhibitory functions of LXA4 on cell migration and invasion, Panc-1 and MIA PaCa-2 cell lines were treated with either vehicle or LXA4 (400 nM) for 24 h. Then, wound-healing assays were performed (Fig. 1a). We found that, after exposure to LXA4, the migration of both Panc-1 and MIA PaCa-2 cells was significantly suppressed compared to those cells treated with vehicle. To test cell invasion capability, Transwell assays were performed (Fig. 1b, c), and the results showed that cell invasion was significantly attenuated by LXA4 in Panc-1 (P = 0.0018) and MIA PaCa-2 (P = 0.0002) cells. These data revealed that LXA4 could suppress pancreatic cell migration and invasion.

Next, we set out to investigate the effect of LXA4 on the EMT of pancreatic cancer cells. Gradient concentrations of LXA4 were used to treat Panc-1 and MIA PaCa-2 cells, and then, the expression of E-cadherin, N-cadherin and Vimentin was determined by western blotting. We found that, in parallel with LXA4 concentration, the expression of E-cadherin was gradually up-regulated while the expression of N-cadherin and Vimentin was down-regulated. These data indicated that LXA4 could reverse mesenchymal phenotypes in a dose-dependent manner (Fig. 2a). Based on morphological observations, when these cells were treated with LXA4 (400 nM) for 24 h, both cell lines partially lost their mesenchymal morphology, such as spindle shapes and tight intercellular connections, and became cubic-shaped cells that were linked with one another (Fig. 2b).

As the anti-inflammatory function of LXA4 is mediated by its receptor, FPRL1, we wondered whether FPRL1 plays a role in the maintenance of mesenchymal phenotypes in pancreatic cancer. SiRNA targeting FPRL1 was applied to knock down the expression of FPRL1 in Panc-1 and MIA PaCa-2 cells; the expression of EMT-associated markers was then assessed. In the absence of LXA4, the knock down of FPRL1 expression demonstrated little effect on these markers (Fig. 2c). However, when FPRL1 was knocked down, LXA4 failed to reverse mesenchymal phenotypes in both cell lines (Fig. 2c), which suggested that the effects of LXA4 were mediated by FPRL1.

TGF-β1 is a canonical cytokine that promotes EMT and maintains the mesenchymal phenotype of cancer cells [17].Given the previous evidence that LXA4 can suppress many types of cytokines present during inflammatory responses [18], we inferred that LXA4 might down-regulate the level of TGF-β1. To verify our hypothesis, we determined the expression of TGF-β1 by western blotting. We found that TGF-β1 was down-regulated by LXA4, but when we knocked down FPRL1 by siRNA, LXA4 no longer exerted an effect on TGF-β1. Both Panc-1 and MIA PaCa-2 cells showed a similar result (Fig. 3a). Immunofluorescence was then performed to confirm the results. Treatment of both cell lines with LXA4 decreased the fluorescence intensity of TGF-β1, whereas when FPRL1 expression was knocked down prior to LXA4 administration, TGF-β1 exhibited a similar fluorescence intensity as the control and the siFPRL1 group (Fig. 3b). Finally, we tested the secretion of TGF-β1 by ELISA, which revealed that LXA4 could also suppress the secretion of TGF-β1. However, the knock down of FPRL1 eliminated the suppressive function of LXA4 (Fig. 3c).

TGF-β1 is a type of secreted protein that activates the TGF-β receptor, which influences cellular phenotypes [19]. To explore whether autocrine TGF-β1 signaling plays a role in the mechanism by which LXA4 suppresses the mesenchymal phenotypes in pancreatic cancer cells, anti-TGF-β1 neutralizing antibody and exogenous rhTGF-β1 were used to modulate the concentration of TGF-β1 in the cell culture media. When an anti-TGF-β1 neutralizing antibody (1 μg/mL) was added, E-cadherin expression was elevated while N-cadherin and Vimentin expression were decreased, which was similar to what was observed with LXA4 administration. Interestingly, when rhTGF-β1 (5 ng/mL) was administered after LXA4 treatment, the expression of EMT-associated markers was reversed (Fig. 4a). These data demonstrated that LXA4 attenuated mesenchymal phenotypes via the inhibition of autocrine TGF-β1 signaling. Then, Transwell invasion assay was performed to assess the invasive capacity of both Panc-1 and MIA PaCa-2 cells. Similarly, an anti-TGF-β1 antibody and LXA4 significantly decreased the number of passaged cells (P < 0.05), but rhTGF-β1 reversed the effect of LXA4 (P < 0.05) (Fig. 4b, c).

Our data revealed that LXA4 could attenuate the invasiveness of pancreatic cancer in vitro. Whether LXA4 could inhibit metastasis in vivo remained to be investigated. Thus, we established a liver metastasis model by injecting Panc-1 cells into the subsplenic membranes of mice. One week after the model was established, mice were treated with vehicle or BML-111, which is a stable analog of LXA4 and an FPRL1 agonist. The drug was administered for five weeks, and all of the mice were sacrificed at the end of the eighth week (Fig. 5a). We found that the mice that were treated with BML-111 exhibited fewer metastatic lesions and lower relative liver weights compared to mice in the control group (Fig. 5b, c). By microscopy, larger areas of metastatic lesions in the liver were found in mice treated with vehicle, whereas in mice treated with BML-111, the lesions were small and most of them were located near blood vessels (Fig. 5d, e). To confirm the mechanisms that were explored in vitro, we determined the expression of TGF-β1 and that of a key EMT-promoting transcription factor, Snail, by immunohistochemistry. We found that both TGF-β1 and Snail were down-regulated by BML-111 (Fig. 5f), which suggested that similar mechanisms occurred in vivo.

LXA4 is an endogenous lipid molecule that exerts its biological activity via FPRL1. We wondered whether the concentration of LXA4 in human pancreatic cancer tissues correlates with the metastatic potential. However, the concentration of LXA4 in tissues is extremely low (~0.1 ng/mL) [20] and is therefore difficult to extract. Above all, it is difficult to evaluate the biological activity of LXA4 solely by its concentration while the expression of its receptor FPRL1 is ignored. To solve this problem, we defined the “Lipoxin effect score” (LES) as the product of the immunohistochemistry scores of both 15-LOX and FPRL1. Because immunohistochemistry scores range from 0 to 3, the LES can be a number in the set (0, 1, 2, 3, 4, 6, 9). We also define a low LES (LES < 4) and a high LES (LES ≥ 4). Then, we analyzed the correlation between the LES and the clinical-pathological features of pancreatic cancer (Table 1). We found that the LES correlated with histological grade, primary tumor (T), lymph node metastasis (N), and distant metastasis (M). Specifically, a low LES tended to correlate with lymph node and distant metastasis, which confirmed our previous results in vitro and in vivo. In addition, immunohistochemical staining revealed that patients with a low LES tended to express more TGF-β1 and vice versa (Fig. 6). Statistical analysis revealed that the LES was also correlated with TGF-β1 expression (Table 2), which further confirmed that LXA4 attenuated cell invasion and metastasis via the inhibition of TGF-β1 expression.