YD-10B human OSCC cells, which were derived from tongue cancer patient tissues, were obtained from the Department of Oral Pathology, College of Dentistry, Yonsei University (Seoul, Korea). Cells were grown in DMEM/F12 (3:1 ratio) medium supplemented with 10% fetal bovine serum (FBS), 1×10⁻¹⁰ M cholera toxin, 0.4 mg/ml hydrocortisone, 5 µg/ml insulin, 5 µg/ml apo-transferrin and 2×10⁻¹¹ M triiodothronine (T3) in a humidified atmosphere of 5% CO₂ at 37^°C. Normal gingival fibroblasts (NFs) were obtained from a patient who had wisdom teeth extracted and who did not have oral mucosal disease. Cancer-associated fibroblasts (CAFs) were obtained from a surgical specimen of a patient with oral squamous cell carcinoma (OSCC) and were selected from explanted cancer tissues in Versene solution [200 mg EDTA and 1 mg glucose in 1 liter phosphate-buffered saline (PBS) buffer] (6). Briefly, the tissue was washed with betadine, cut into small pieces with scissors and washed three times with PBS. Fibroblasts were selected in Versene solution and characterized by immunohistochemical staining with anti-vimentin (dilution 1:100; cat. no. M0725; Dako, Carpinteria, CA, USA) and anti-α-smooth muscle actin (α-SMA) (dilution 1:100; cat. no. ab5694; Abcam, Cambridge, MA, USA) antibodies. Early passages (<9) of the fibroblasts were subjected to analysis. Informed consent was provided by the patients whose CAFs and NFs were used in this study and approval was granted by the Institutional Review Board of Yonsei University College of Dentistry. The isolated fibroblasts were maintained in DMEM/F12 (3:1 ratio) complete medium containing 1% penicillin/streptomycin. Dimethyl sulfoxide (DMSO) [0.1% (v/v)] was used in cell culture for control analysis.

Dulbecco's modified Eagle's medium (DMEM), Ham's F-12 nutrient mixture, FBS, antibiotic-antimycotic (100X), PBS and 0.25% trypsin-EDTA (1X) were purchased from Gibco-BRL (Thermo Fisher Scientific, Inc., Waltham, MA, USA). Cholera toxin, hydrocortisone, insulin, apo-transferrin, T3 and DMSO were obtained from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany). Lysophosphatidic acid (LPA) was purchased from Enzo Life Sciences (Plymouth Meeting, PA, USA). Recombinant human transforming growth factor (TGF-β1), epidermal growth factor (EGF), and interleukin (IL)-1β were obtained from EMD Millipore (Billerica, MA, USA). SB431542, AG1478 and IL-1R antagonist were purchased from Calbiochem (Merck KGaA). The following antibodies were purchased from their respective sources: fascin (cat. no. M3567) and E-cadherin (dilution 1:1,000; cat. no. M0725) (Dako); total (cat. no. 9242)/phosphor (cat. no. 9246) forms of IκBα (dilution 1:1,000; both from Cell Signaling Technology, Danvers, MA, USA); p50 (cat. no. sc-166588) and p65 (cat. no. sc-71675) subunit of NF-κB (dilution 1:1,000; both from Santa Cruz Biotechnology, Santa Cruz, CA, USA); β-actin (cat. no. A5441) and GAPDH (cat. no. G9545) (dilution 1:1,000; both from Sigma-Aldrich; Merck KGaA). Alexa Fluor 568 phalloidin and Oregon Green 488 gelatin were purchased from Molecular Probes (Thermo Fisher Scientific, Inc.).

Fascin 1-specific shRNA (h) lentiviral particles were transduced in cultured cells with 5 µg/ml Polybrene according to the manufacturer's protocol (Santa Cruz Biotechnology). Continual selection was followed with 1 mg/ml puromycin to establish the fascin-depleted stable cell line. Control shRNA (h) lentiviral particles-A (Santa Cruz Biotechnology) were also used as a negative control. The extent of fascin depletion was evaluated by western blot analysis.

Protein (50 µg) was separated on 10% SDS-polyacrylamide gel and were transferred to a polyvinylidene difluoride (PVDF) membrane (EMD Millipore). The membrane was blocked with 10% skim milk in PBS containing 0.1% Tween-20 (PBS-T) and subsequently incubated overnight with a 1:1,000 dilution of the primary antibody against its specific protein at 4^°C. The blots were then incubated with a 1:3,000 dilution of their respective horseradish peroxidase-conjugated secondary antibodies for 2 h at room temperature and were washed with PBS-T. The targeted proteins were visualized using an enhanced chemiluminescence detection kit (Amersham Life Science, Arlington Heights, IL, USA) according to the manufacturer's instructions. Nuclear extracts were prepared using a Nonidet P-40 lysis method. Band intensities were measured with the ImageJ 1.52a software program (National Institutes of Health, Bethesda, MD, USA)

NFs and CAFs (5×10⁴ cells) in a 100-mm Petri dish were cultured in 6 ml of 1% FBS media for 48 h. The media were centrifuged at 10,000 × g for 5 min and the supernatant was used as CM.

The conditioned medium was collected, and the protein concentration was determined by the Bradford method (Bio-Rad Laboratories, Hercules, CA, USA). Equal amounts of protein (20 µg) were used for analysis on 8% sodium dodecyl sulfate (SDS)-polyacrylamide gel containing 0.1% (w/v) gelatin. The gel was washed with zymogram renaturation buffer (Bio-Rad Laboratories) for 1 h and incubated with zymogram development buffer (Bio-Rad Laboratories) for 16 h at 37°C in a shaking incubator. The gel was stained with 0.2% Coomassie Brilliant Blue and gelatinolytic activity of MMP-9 and MMP-2 were detected as clear bands in a dark blue background (7).

Polycarbonate nucleopore filter inserts (8-µm pore-sized) in a 24-well Transwell chamber (Corning Costar, Cambridge, MA, USA) were coated with Matrigel (30 µg/well; BD Biosciences, Franklin Lakes, NJ, USA). Cells (5×10⁴ cells) were added into the upper chamber, and complete medium was added to the bottom chamber and maintained for 48 h in a 37°C incubator. Invaded cells on the lower surface of the membrane was fixed with ethanol and non-invasive cells were removed with a cotton swab (8). Then, cells were stained with hematoxylin. Invaded cells from five fields were counted under a fluorescence microscope (EVOS™ XL Cell Imaging System; Life Technologies; Thermo Fisher Scientific, Inc.). CM was diluted 1:1 with complete media and added to the bottom of the Transwell chamber. Specific inhibitors were also added in the bottom of Transwell chamber for inhibitor studies. For fascin expression, cells in the upper and lower chamber were collected at appropriate assay times and were analyzed.

FITC-conjugated gelatin-coated coverslips were prepared as previously described (9). Cells (3×10³ cells) were plated on Oregon Green 488 gelatin-coated coverslips in 12-well plates and cultured for 16 h. Cells were fixed with 4% paraformaldehyde followed by permeabilization with 0.5% Triton X-100/PBS and stained for actin with Alexa Fluor 568 phalloidin. Areas of matrix degradation was identified by a loss of fluorescence using an EVOS FL monochrome microscope (Thermo Fisher Scientific Inc.). To quantify invadopodium-mediated ECM degradation, black and white images of gelatin degradation were analyzed using ImageJ 1.52a software (National Institutes of Health).

The cancer cells were cultured on a dermal equivalent that was generated with a Type I-A collagen mixture (Nitta Gelatin Inc., Osaka, Japan) with eight volumes of ice-cold collagen solution, one volume of 10X reconstitution solution (0.022 g/ml NaHCO₃, 0.0477 g/ml HEPES and 0.05 N NaOH) and one volume 10X DMEM. Gingiva fibroblast (1.5×10⁵ cells) suspensions in culture medium were then added. This mixture was poured onto polycarbonate filter inserts (3-µm pore size, 12-mm diameter; EMD Millipore) and placed in 6-well plates (Corning Costar). After a 24-h incubation at 37°C, 3 ml of medium was added to the 6-well plates. Cancer cells (1×10⁶ cells) from each cell line were seeded onto the dermal equivalent. After 48 h, the media in the 6-well plates were changed, and culture media were added to the epidermal equivalent. After 48 h, the cultures were exposed to air by removing the medium from the epithelial layer to generate an air-liquid interface microenvironment. The culture medium was then changed every 2–3 days for 2 weeks. Each culture was performed independently three times and then formalin-fixed, paraffin-embedded and histologically examined. To measure invasive areas and depth, the culture tissue was stained with hematoxylin and eosin (H&E) (6).

Transcriptional activity of NF-κB was measured by luciferase reporter assay using the pNF-κB-Luc reporter plasmid (Clontech Laboratories, Palo Alto, CA, USA). Mock and the fascin-depleted cell line (fascin^dep) at 70–80% confluence in 6-well plate were co-transfected with 1 µg of NF-κB reporter constructs and 0.5 µg pSV-β-galactosidase for 8 h in serum- and antibiotic-free Opti-MEM (Gibco-BRL; Themo Fisher Scientific, Inc.) with Lipofectamine 2000 reagent (Invitrogen; Themo Fisher Scientific, Inc.). Luciferase and β-galactosidase activities were assayed according to the manufacturer's protocol (Promega, Madison, WI, USA), using a microplate spectrofluorometer (Molecular Devices, Palo Alto, CA, USA). Luciferase activity was normalized by β-galactosidase activity in cell lysate and expressed as an average of three independent experiments. pTAL construct was used as negative control.

All animal studies were performed in accordance with the experimental protocols that were approved by the Animal Ethics Committee of Eulji University. Male Balb/C athymic nude mice (5 weeks of age, 10 g of body weight; provided by the Central Animal Laboratory, Seoul, Korea) were maintained at 20–22°C on a 12-h light/dark cycle. Mock (n=8) and fascin-depleted cells (fascin^dep) (n=8) (5×10⁴ cells/0.1 ml PBS) were submucosally injected into the tongues of mice under anesthesia using a 0.5 ml insulin syringe (n=5/group). Growth of the tumor xenografts in mice was observed for 5 weeks. Tissues were collected and fixed in formalin and processed for paraffin embedding for histopathological studies. The tumor areas in each tongue were acquired with a fluorescence microscope (EVOS™ XL Cell Imaging system; Life Technologies; Thermo Fisher Scientific, Inc.) with a ×40 magnification and digitally quantified using ImageJ 1.52a software (National Institutes of Health).

The analysis for statistical purposes was conducted using InStat GraphPad Prism version 5.01 statistical software (GraphPad Software, Inc., San Diego, CA, USA). One-way ANOVA and Tukey's honestly significant difference (HSD) post-hoc test were applied. The non-parametric Mann-Whitney test was used to evaluate the western blot analysis. Results are expressed as mean ± standard deviation (SD). A P-value <0.05 was considered to indicate a statistically significant result.

In a previous study, we performed array-comparative genomic hybridization (CGH) analysis by separating the surgical tissue of oral squamous cell carcinoma (OSCC) patients into carcinoma, dysplasia (marginal tissue) and adjacent normal epithelial cells (10). In the array-CGH data, fascin showed a higher expression level in cancer specimens than that in normal specimens and dysplasia (Fig. 1). To confirm the role of fascin in OSCC, fascin depletion was performed with lentiviral short hairpin RNA (shRNA) against fascin mRNA and a stable cell line (fascin^dep) was established (Fig. 2A). The fascin^dep cell line had reduced invadopodia spot formation and FITC-gelatin degradation activity (Fig. 2B). This result indicates that fascin is involved in invadopodium formation and extracelluar matrix degradation activity. In the Matrigel-coated Transwell system, fascin^dep cells showed reduced invasion activity compared to wild-type Mock cells (Fig. 2C), respectively. Invasive areas into the dermal equivalents were also determined using three-dimensional (3D) cultures. Fascin^dep cells possessed low invasion activity compared to the Mock (Fig. 2D). In the zymography assay, gelatinolytic activities of pro-MMP-9 and active-MMP-2 in conditioned media (CM) from fascin^dep cells were significantly lower than those in the Mock (Fig. 2E). No statistically significant changes were observed in the active-MMP-9 and pro-MMP-2 activity.

The effect of fascin on tumor growth was tested in nude mice with orthotopic tongue tumors. Submucosally inoculated Mock and fascin^dep cells were subjected to growth for 5 weeks. As shown in Fig. 3A, tumor growth was significantly increased in mice inoculated with Mock cells. However, the tumor area (mm²) was significantly lower in the mice inoculation with the fascin^dep cells. The difference in tumor area between the two groups was statistically significant (Fig. 3B). Fascin expression was strongly observed in the invasive tumor front in the tissue of the Mock group. In the fascin^dep group, fascin expression was relatively low and encapsulated tumor was observed (Fig. 3C).

To verify changes in fascin expression during the infiltration process, a Matrigel-coated Transwell invasion assay was performed. In Transwell infiltration without Matrigel, there was no significant change in fascin expression. However, expression of fascin was increased during Matrigel-coated Transwell infiltration (Fig. 4A). In addition, unlike the reaction induced by conditioned media (CM) from normal fibroblasts (NFs), the expression of fascin was significantly increased by CM from cancer-derived fibroblasts (CAFs) (Fig. 4B). These results indicate that the expression of fascin which occurs during invasion is regulated by the microenvironment surrounding the cancer. In order to investigate the mechanism involved in fascin expression, various molecules derived from the tumor microenvironment were used to treat cancer cells and fascin expression was examined (Fig. 4C). Lysophosphatidic acid (LPA), a phospholipid derivative that acts as a signaling molecule, was not involved in fascin expression. However, stimulation by TGF-β1, EGF, or IL-1 β significantly increased fascin expression. Selective inhibitor SB431542 (TGF- β RI kinase inhibitor), AG1478 (EGF receptor tyrosine kinase inhibitor), or IL-1 receptor (IL-1R) antagonist inhibited Matrigel-induced-fascin expression (Fig. 4D). CAF stimulated-fascin expression was also diminished by treatment with SB431542, AG1478, or IL-1 receptor antagonist (Fig. 4E).

The RhoA-NF-κB signaling axis plays a pivotal role in cancer invasion and motility (11). RhoA, a small GTPase protein in the Rho family, affects the invasive behavior of tumor cells by stimulating actin polymerization at the leading edge (12). Constitutively active RhoA activates NF-κB activity during invasion (13). The YD-10B OSCC cell line used in this study is a highly invasive cancer cell line with originally very low E-cadherin levels (6). In a stable cell line with dominant-negative RhoA N19, fascin expression was relatively suppressed compared to that in the Mock while the expression of epithelial phenotype marker E-cadherin was increased (Fig. 5A). In addition, IκBα an inhibitory molecule of NF-κB showed low phosphorylation levels in the fascin^dep cells (Fig. 5B). Consistent with these results, nuclear translocation levels of the p50 and p65 NF-κB subunit were decreased in the fascin^dep cell line compared to those in the Mock (Fig. 5C). Relatively low activity of NF-κB was found in the fascin^dep cells using a NF-κB luciferase reporter (Fig. 5D). These results indicate that RhoA and IκBα/NF-κB signaling are involved in fascin expression in OSCC cells.