Nobiletin, LY294002, TGF-β1 was from Merck Millipore (Darmstadt, Germany). AZD1080, SKL2001 and diamidino-phenyl-indole (DAPI) were purchased from Funakoshi (Tokyo, Japan). Slug plasmid (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was transfected into U343 cells according to the manufacturer's protocol. Antibodies against p38, p-p38, ERK, p-ERK, p-AKT, AKT, β-catenin, GSK-3β, p-GSK-3β, p-Smad3 were from Cell Signaling Technology (Danvers, MA, USA). Antibodies against E-cadherin, N-cadherin, occluding, fibronectin, Slug, Twist1, and Snail were obtained from Santa Cruz Biotechnology.

U87, U251 and U43 cells from the American Type Culture Collection (ATCC, Manassas, VA, USA) were maintained at 37°C, 5% CO₂, in DMEM supplemented with 10% fetal bovine serum, penicillin (500 U/ml) and streptomycin (100 µg/ml). Cell proliferation was determined using the CCK-8 (Beyotime Institute of Biotechnology, Shanghai, China) method. U87 and U251 cells (3×10³ cells/well) were plated in 96-well plates. At 24 h incubation with nobiletin, CCK-8 reagents were added and incubated for another two hours. The absorbance at 450 nm represents the number of viable cells with a Multiskan Spectrum microplate reader.

In the migration assays, scratches were made in the cells and presented by the average of distance differences between 24 and 0 h. To assess invasion potential, Matrigel-coated Transwell was used. Nobiletin-treated cells were seeded in the upper chamber of the Transwell insert. The migration capacity over the Matrigel and the membrane towards the bottom chamber containing 10% FBS was examined. Cells were fixed in formaldehyde solution, and stained with crystal violet and counted.

Cells were grown on glass coverslips and blocked with 5% normal fetal bovine serum containing 0.1% Triton X-100 for 2 h at room temperature, thereafter, coverslips were incubated with anti-E-cadherin antibodies overnight with DAPI (Molecular Probes). Fluorescence was visualized with a microscope and Image-Pro Plus 4.0 software.

Signaling experiments were performed as described previously with slight modifications (16). Nuclear and cytosolic fractions were carried out according to manufacturer's instructions (Pierce Biotechnology, Rockford, IL, USA). In western blot assays, protein samples were separated on SDS polyacrylamide gel and then blotted onto nitrocellulose membranes using a Semi-dry electrotransfer mini Trans-Blot Cell (Bio-Rad, Hercules, CA, USA) and transfer buffer. The membranes were incubated overnight at 4°C with primary antibodies and then incubated with anti-IgG Antibody. Protein G-agarose beads (Invitrogen) were used for immunoprecipitations. Pierce Clean Blot IP Detection reagent (Thermo Fisher Scientific, Rockford, IL, USA) were applied for immunoprecipitation experiments. ECL substrates were used for protein band visualization (Bio-Rad). The bands densitometry was analyzed with ImageJ software.

ChIP assay was performed as previously described with slight modification (17). β-catenin antibody was used, and qPCR was performed with specific primers to the slug promoter. PCR amplifications were conducted using Blend Taq DNA polymerase and the PCR products were analyzed by electrophoresis on agarose gel and then qPCR. Cross-link released chromatin was saved and reversed by proteinase K digestion and phenol/chloroform extraction. Immunoprecipitated samples were compared to the housekeeping gene RPL30.

Athymic nude mice (4 weeks old) were maintained under pathogen-free conditions. U87-Luc cells were harvested and resuspended. Suspended cells (1×10⁷) were subcutaneously injected into the flank of each nude mouse. U87-Luc-bearing nude mice were randomly divided into four groups as follows: Vehicle group (saline daily); Cyclophosphamide positive group (20 mg/kg, i.p. every 3 days); nobiletin group (15 and 30 mg/kg, oral administration daily). The tumor growth was examined with a caliper every 3 days and calculated by the following equation: V = length×width²/2. Mice were observed by the IVIS in vivo imaging system after 24 day-treatment.

The primary tumors were immunostained for E-cadherin, N-cadherin and Slug as previously described (18).

Statistical analysis was conducted with GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA). One-way ANOVA was used to compare between multiple groups. All results are expressed as mean ± SD. p<0.05 and p<0.01 were considered to indicate statistically significant differences.

In the CCK-8 assays, nobiletin at 5, 10 and 15 µM exhibited no significant cytotoxic effect to U87 and U251 cells (Fig. 1A). Based on our previous data, these concentrations intervene with neither metabolism nor function of mitochondria, so they were then used in the following experiments. Wound healing models were used to assess the effect of nobiletin on the migratory of glioma cells. It can be seen from Fig. 1B that the migratory potential was remarkably inhibited by nobiletin. Twenty-four hours after the scratch, cells in the vehicle group almost filled the scratched area, whereas an obvious gap was observed in the wound in nobiletin-treated cells. The quantification by scratch width showed that the inhibition rate of nobiletin (10 and 15 µM) was approximately 25 and 45% in U251 cells and 33 and 50% in U87 cells, respectively. Next, the regulatory effect of nobiletin on the invasion of U87 and U251 cells were validated via matrigel transmembrane assay. The results showed that cells in the control group invaded through the matrigel faster than nobiletin-treated cells, and the inhibitory rate of nobiletin (15 µM) reached approximately 65 and 60% in U87 and U251 cells, respectively (Fig. 1C).

E-cadherin, occluding, N-cadherin, and fibronectin play an essential role during the progress of cancer metastasis. In western blotting assays, the expression of EMT-related proteins in glioma cells were examined. In Fig. 1D-E, the epithelial markers occludin and E-cadherin were increased while mesenchymal markers such as fibronectin and N-cadherin were reduced in both cell lines. As is known, Snail, Slug and Twist1, are potent repressors of E-cadherin. Herein, transcriptional factors involved in EMT were further tested to prove the efficacy of nobiletin. We found nobiletin considerably inhibited Slug expression while displaying negligible effect on the levels of Snail and Twist1.

To uncover the molecular mechanisms of nobiletin, essential and crucial pathways manipulating cancer invasion are examined, such as ERK/p38, PI3K/AKT and Wnt/β-catenin axis. The ERK/p38 signaling was only subtly affected by nobiletin treatment at 15 µM. At the concentration of 30 and 60 µM, nobiletin displayed a potent inhibitory effect on ERK and p38 phosphorylation. Moreover, PI3K/AKT pathway was remarkably suppressed at 15 µM. Decrement in the phosphorylation of AKT, GSK3β and β-catenin proteins was observed in nobiletin-treated U87 and U251 cells (Fig. 2A). Combination of nobiletin and LY294002 (a specific PI3K inhibitor) synergistically repressed the activation of AKT pathway in glioma cells (Fig. 2B). As a potent inhibitor of GSK3β, AZD1080 was also used in the study. Nobiletin reversed AZD1080-triggered GSK-3β phosphorylation in both U87 and U251 cells. SKL2001, a small molecule inhibitor of β-catenin degradation, was chosen to promote β-catenin stabilization (19). SKL2001 (40 µM) enhanced cell invasion, and the increment was partially abrogated by nobiletin (Fig. 2D). In summary, these results imply that the AKT/GSK-3β/β-catenin axis plays a critical role in the function of nobiletin on glioma cell migration.

Amplified secretion of TGF-β by carcinoma cells, and augmented TGF receptor expression, leading to autocrine TGF-β loop, are regarded to drive or to be requisite for EMT progression in tumor cells. Matrigel invasion and wound-healing assay was performed to assess whether nobiletin inhibits invasion stimulated by TGF-β. In the CCK-8 analysis, medium containing TGF-β (10 ng/ml) and nobiletin (10, 15 µM) exerted no distinguished cytotoxicity on cell viability (Fig. 3A). Therefore, TGF-β (10 ng/ml) and nobiletin (10, 15 µM) were used for following assays. After exposure to TGF with or without nobiletin for 24 h, U87 cells were harvested and reseeded into a Transwell-coated with matrigel. After 24 h, cells that penetrated through the inserts was stained with crystal violent and assessed under microscopy. TGF-β enhanced both invasion and migration, but the tendency was reversed by nobiletin (Fig. 3B and C). Cells that had undergone EMT in response to TGF-β accompanied with fibronectin and N-cadherin increment and E-cadherin and occludin decrement. In line with inhibitory effect on basal level of mesenchymal markers, nobiletin reversed TGF-β-induced epithelial-mesenchymal transition (Fig. 4A).

TGF/Smads regulates the activities of EMT transcription factors such as Slug (Snai2), which suppresses E-cadherin transcription. TGF-β also activates non-Smad pathway, such as AKT and GSK-3β. The role of above-mentioned pathways in regulating TGF-β-stimulated EMT was assessed. Consistently, nobiletin blunted the acitivity of AKT/GSK-3β axis induced by TGF-β, however, showing little effect on TGF-β-activated Smad3 phosphorylation (Fig. 4B). Collectively, these data indicated that nobiletin did not retard EMT via directly targeting Smad3 signaling.

By interacting with E-cadherin, β-catenin is important in EMT progression. On the other hand, active GSK-3β will form a complex with β-catenin to hinder the translocation of β-catenin into nucleus. Phosphorylated GSK-3β is an inactive form of GSK-3β, which would abolish the interaction with β-catenin, subsequently promoting nuclear translocation and inhibiting β-catenin degradation. As revealed in the immunoprecipation assay, an increase was observed in the β-catenin/GSK-3β complex in TGF-β treated U87 cells (Fig. 5A). Western blot assays for β-catenin ubiquitination displayed that TGF-β considerably reduced β-catenin proteasome degradation, and the decrement was greatly alleviated by nobiletin treatment (Fig. 5B). Furthermore, nucleus translocation of β-catenin was also upregulated due to TGF stimuli. Nevertheless, the function was abolished after nobiletin treatment (Fig. 5C), as demonstrated in the immunofluorescence staining. Fractionation followed by western blot also showed similar results (Fig. 5D). ChIP tests were carried out to explore the binding of β-catenin to the promoter of Slug. Noteworthy, nobiletin successfully prevented the binding of β-catenin to the Slug promoter, induced by TGF (Fig. 5E). It can be concluded that nobiletin might influence EMT via regulating GSK-3β and the downstream effector β-catenin.

To confirm the essential role of Slug in the anti-metastasis effect of nobiletin, a non-Slug-expressing glioma cell line was applied to compare the effect of nobiletin on migration and EMT. According to preliminary research, these three cell lines have divergent Slug protein levels. The protein expression of Slug was rather low, almost undetectable in U343 cells, and high level of Slug was observed in both U87 and U251 cells (Fig. 6A). Nobiletin had little effect on the invasion of U343 cells at 10 and 15 µM. However, at the same concentration, the inhibitory rate reached 45 and 60% in U87 and U251 cell, respectively (Fig. 6B). Moreover, nobiletin failed to change Slug protein expression in U343 cells. Notably, nobiletin treatment brought down Slug protein expression in Slug plasmid-transfected U343 cells (Fig. 6C). The phenomenon that nobiletin enlarged E-cadherin expression and inhibited the level of N-cadherin in both U87 and U251 cells was not observed in U343 cells (Fig. 6D). EMT-related proteins remained unaffected in U343 cells. These data implied that Slug, either at transcriptional or post-translational level, is greatly involved in the regulatory effect of nobiletin.

To further assess the antitumor efficiency of nobiletin in vivo, athymic nude mouse model bearing U87-Luc implanted xenografts were used. After 24-day treatment, the bioluminescence imaging results demonstrated impressive repression of tumor volume in nobiletin-treated nude mice (p<0.01, versus the vehicle group). Tumor sections were then subjected to IHC analysis for detection of E-cadherin, N-cadherin and Slug. E-cadherin expression was enhanced while N-cadherin and Slug was inhibited in nobiletin-treated group (Fig. 7). All these date indicated that nobiletin prevented EMT and growth of glioma cancer cells in vivo.