Human OSCC lines (FaDu and SAS) were purchased from American Type Culture Collection (ATCC; Manassas, VA, USA). SAS cells were grown in Dulbecco’s modified Eagle’s medium (DMEM), and FaDu cells were grown in RPMI-1640. All culture media were supplemented with 5% fetal bovine serum (FBS, Gibco, Grand Island, NY, USA) and penicillin/streptomycin (Gibco), and were grown at 37°C in a 5% CO₂ atmosphere.

Small hairpin RNA vectors for PODXL silencing (5′-GTCGTCAAAGAAATCACTATT-3′) were obtained from the National RNAi Core Facility (Academia Sinica, Taiwan). To generate stable PODXL-knockdown cell lines, HEK293T packaging cells were co-transfected with a packaging plasmid (pCMV-ΔR8.91), and envelope (pMDG) and hairpin pLKO-RNAi vectors using a PolyJET Transfection kit (SignaGen Laboratories, Ijamsville, MD, USA). At 48-h post-transfection, virus-containing supernatants were collected, mixed with fresh media containing polybrene (8 μg/ml), and incubated with target cells for another 48 h. Transduced cells were selected with puromycin (2 μg/ml) for 7 days.

Cells (10⁵) were seeded in a Transwell insert (8-μm filters, Corning, New York, NY, USA) coated with or without Matrigel (BD Biosciences, La Jolla, CA, USA) for 8 (for the migration assay) and 24 h (for the invasion assay). After incubation, cells were fixed with 4% paraformaldehyde for 10 min. Cells that had not invaded were removed with a cotton swab; invaded cells were stained with 4′,6-diamino-2-phenylindole (DAPI), imaged under an inverted fluorescent microscope (Zeiss), and quantified using ImageJ software.

Cells (2×10⁴) were seeded in a chamber with complete growth media, and monitored under an inverted light microscope (Zeiss HAL100 reflected-light microscope) with a temperature and CO₂ control system. Images were captured every 10 min for 6 h. The migration distance was defined as the movement of the cell center per unit time, as measured by MetaMorph software.

Differences in the proliferation of SAS/LKO and SAS/shPODXL cells were evaluated by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay. Briefly, 2,000 cells were seeded in a 96-well plate with complete media for 1–5 days, and cells were incubated in 50 μl of 0.5 mg/ml MTT for 3 h. DMSO was added to dissolve the crystals, which were measured on a microplate reader at 540-nm absorbance. Data are presented as the percentage of growth on the day after seeding. To test the effect of PODXL on colony formation, 2000 SAS/LKO and SAS/shPODXL cells were seeded in 6-well plates and incubated for 7 days. Surviving colonies were stained with crystal violet after methanol fixation. Visible colonies (≥50 cells) were counted.

Cells were lysed in RIPA buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecylsulfate (SDS), and 50 mM Tris-HCl at pH 7.4) containing a protease and phosphatase inhibitor cocktail (Roche, Basel, Switzerland). For immunoprecipitation, cell lysates were precleared with agarose-protein G for 1 h and incubated overnight at 4°C with the appropriate protein G-conjugated primary antibodies. Beads were washed three times with RIPA buffer and boiled in sample buffer (50 mM Tris-HCl at pH 6.8, 2% SDS, 0.1% bromophenol blue, and 10% glycerol). Equal amounts of proteins were separated on SDS-polyacrylamide gel electrophoresis (PAGE) and then transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA, USA). Membranes were blocked with 1% bovine serum albumin (BSA)/TBST and incubated overnight with specific primary antibodies against PODXL (1:500, Santa Cruz Biotechnology, Santa Cruz, CA, USA), FAK (1:2000, Millipore), phospho-FAK (1:2000, Millipore), paxillin (1:2000, Millipore), phospho-paxillin (1:500, Millipore), or α-tubulin (1:5000, Sigma, St. Louis, MO, USA). Membranes were then incubated with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (1:10000, Jackson Immunoresearch) for 1 h at room temperature, and proteins were detected using an ECL kit (Millipore, Temecula, CA, USA).

Cells were grown onto cover slide and fixed with 4% paraformaldehyde for 10 min, followed by blocking with 3% bovine serum albumin (BSA) for 1 h. The cover slide was incubated with FITC-conjugated cortactin (Merck Millipore) and rhodamine-conjugated phalloidin (Invitrogen) for 1 h at room temperature. The fluorescent images were observed under a confocal microscope (TCS-SP5, Leica).

A paraffin-embedded human colon cancer tissue microarray (US Biomax Inc.) was immersed in xylene for 30 min and then rehydrated in graded ethanol. The slide was immersed with 0.01 M sodium citric buffer (pH 6.0) for antigen retrieval, followed by soaking with 3% hydrogen peroxide and blocking with normal horse serum (ABC kit, Vector). The slide was incubated with anti-PODXL antibody (10 μg/ml, Sigma) for 1 h at room temperature. After washing three times with PBST, the Super Enhancer reagent (Super Sensitive Polymer HRP detection system, BioGenex) was added and incubated for 20 min. DAB Chromogen was then added for 3 min, and the reaction was stopped by PBS washing. The slide was stained by hematoxylin for 3 min and mounted with a xylene-based mounting solution.

The methylation status in PODXL was analyzed by a methylation-specific polymerase chain reaction (MSP) assay using an EZ DNA Methylation-Direct assay kit (Zymo Research). Briefly, the genomic DNA of SAS and FaDu cells was extracted and modified by sodium bisulfate treatment, which converted all of the unmethylated cytosines to uracils while leaving methylated cytosines unchanged. Bisulfate-converted DNA was subjected to subsequent PCR amplification. The specific MSP primers for PODXL (methylated, PODXL-M: 5′-TCGTCGGGTTTAT TTAGAAGTATTC-3′ and 5′-TATATATACGCGAAAA CCAAAACG-3′) and (unmethylated, PODXL-U: 5′-TGT TGGGTTTATTTAGAAGTATTTGG-3′ and 5′-TATATATA CACAAAAACCAAAACAA-3′) were designed from the MethPrimer database. Methylated and unmethylated genomic DNAs (Zymo Research) were used as an experimental control. To analyze the methylated CpG site in the PODXL promoter region, specific PCR primers designed from the MethPimer database were used for amplification, and PCR products were cloned into the pGEM-T vector for further sequencing analysis. Four clones were sequenced to assess the level of methylation of each CpG site.

Total RNA was extracted using an RNeasy Plus Mini kit (Qiagen) and reverse-transcribed using SuperScrip III reverse transcriptase (Invitrogen). A quantitative PCR was performed using the resulting cDNA, LightCycler 480 SYBR Green I Master Mix (Roche) and a LightCycler 480 System (Roche). Results are expressed as multiples of change relative to the control sample, using the method of ΔΔC_T. GAPDH or 18SrRNA was used as internal control for normalization. Primer sequences are listed in Table I.

Total RNA was extracted using an RNeasy Plus Mini kit (Qiagen), and RNA quality was analyzed on an Agilent 2100 Bioanalyzer (RNA6000 nanochip). A human cDNA microarray was analyzed according to the protocol of the Agilent 2 color system (Agilent G4845A, 4×44K). Experiments were performed and analyzed by the Microarray Core Facility of the Institute of Molecular Biology, and the Core Facility of the Institute of Cellular Organismic and Biology, Academia Sinica. Genetic networks were analyzed using Pathway Studio 7 and Sub-network Enrichment Analysis (SNEA) software (Ariadne Genomics, Rockville, MD, USA). Individual cancer datasets were downloaded from Oncomine (Compendia Bioscience). p-values are given for the medium-rank analysis.

The data were derived from at least three independent experiments. Values are expressed as the mean ± standard error of the mean (SEM). Significant differences were determined using a two-tailed, unpaired t-test, unless otherwise specified. A p-value of <0.05 or <0.01 was considered significant.

In order to investigate the association of PODXL with cancer, we used the Oncomine Cancer Profiling database to analyze expression levels of PODXL in cancerous vs. normal tissues. Five types of cancer were selected, and five array datasets were analyzed in each cancer type including colon (21–25), gastric (26–28), liver (29–32), esophageal (33–37), and head and neck cancers (38–42). Results showed that PODXL was overexpressed in all types of cancerous tissues, compared to normal ones (Fig. 1), suggesting that PODXL expression may be associated with cancer. Assessment of the PODXL messenger RNA level by real-time PCR analyses showed similar results in colon cancer specimens (Fig. 2A). Moreover, results of the immunohistochemical assay revealed that the PODXL protein level was upregulated in lymph node metastatic colon tumor tissues, compared to matched primary tumor sites (n=33, p=0.022) (Fig. 2B).

We further examined the correlation of PODXL expression with tumor migration in two human oral squamous cancer cell lines, SAS and FaDu. Results of real-time PCR analysis showed a higher expression level of PODXL in SAS cells that exhibited potential migratory ability compared to FaDu cells (Fig. 3A), suggesting that PODXL expression might be associated with tumor motility. We used shRNA to silence PODXL mRNA in SAS cells, and results of the Transwell analyses showed that the knockdown of PODXL significantly diminished tumor migration and invasion (Fig. 3B). Moreover, time-lapse photographic observations revealed that suppression of PODXL impacted the cell migratory velocity (Fig. 3C). We further found that inhibition of PODXL effectively reduced tumor proliferation and colony formation in SAS cells (Fig. 3D and E), indicating that PODXL expression was associated with tumor aggressiveness.

Given that the activation of focal adhesion kinase (FAK) participates in promoting cell migration and invasion, we next assessed whether PODXL regulated FAK signaling. Results showed that the phosphorylation of FAK and its downstream molecule, paxillin, was diminished in PODXL-knockdown SAS cells (Fig. 4A). FAK and paxillin play important roles in modulating F-actin reorganization, which results in enhanced cell polarity. We next examined F-actin expression using phalloidin staining, and the results showed that suppression of PODXL reduced the expression levels of filopodia at the membrane edge (closed arrow) and podosome-like structures in the perinuclear region (open arrows), compared to mock-transduced cells (Fig. 4B). To further analyze the effect of PODXL on invadopodia formation, immunofluorescent co-staining with F-actin and cortactin, which were used to identify structures of invadopodia, were performed. Results showed that knockdown of PODXL effectively reduced F-actin and cortactin colocalization. In mock-transduced cells, most of the cortactin and F-actin had co-localized at the membrane edge, indicating the invasive front. Some of the colocalization was observed in the perinuclear region, which indicated the invadopodia precursor (43). However, suppression of PODXL significantly inhibited the colocalization of cortactin and F-actin at both the membrane edge and in perinuclear regions (Fig. 4C), suggesting that inhibition of PODXL impacted focal kinase activation and invadopodia formation.

Epigenetic regulation including histone modification and DNA methylation often participates in modulating gene transcription. We next assessed whether the PODXL expression was associated with methylation of its DNA. The MethPrimer database was used to predict a CpG island located from downstream 143 to upstream 517 of the PODXL gene, which was related to the transcriptional start site (Fig. 5A). Data of the methylate-specific PCR (MSP) assay showed that PODXL displayed totally unmethylated expression in SAS cells, whereas it showed only partially unmethylated status in FaDu cells (Fig. 5B upper panel). Totally unmethylated and methylated genomic DNA was used as experimental controls. Further analysis of the precise methylated level of each CpG site by a bisulfate sequencing assay showed that PODXL DNA in SAS cells was 17% methylated, whereas 42% methylation of the CpG island of PODXL was detected in FaDu cells (Fig. 5B lower panel). These data were consistent with findings of the real-time PCR and MSP analyses. Moreover, treatment of FaDu cells with 5-aza-deoxycytidine (5-aza-dC) for 72 h, a DNA methyltransferase inhibitor, significant increased the PODXL mRNA level (Fig. 5C), indicating that DNA methylation may participate in regulating PODXL expression.

To further explore PODXL-mediated gene expression levels, the cDNA of two biological repeats from PODXL-knockdown SAS and MDA-MB-231 cells was subjected to microarray analyses to examine gene changes after silencing of PODXL. We found that PODXL suppression markedly attenuated genes specifically associated with ECM organization (COL13, COL17 and ITGB3), cell adhesion and the EMT (CDH1, CDH3, LOX and LOX4), and pro-metastasis cytokines (interleukin (IL)1β, IL8 and IL24) (Fig. 6A). The expression levels of these genes were further confirmed by real-time PCR analysis (Fig. 6B). Cellular functional grouping by a Gene Set Enrichment Analysis (GSEA) illustrated that suppression of PODXL significantly affected wound-healing, cell adhesion, ECM polymerization and degradation, and adherent junction assembly (Fig. 6C and D). We further analyzed the hit entities affected by PODXL and linked to the tentative intracellular kinase cascade underlying PODXL using sub-network enrichment analysis (SNEA) algorithm software. The hit entities selected by the microarray data were analyzed by an SNEA algorithm, and the results showed that the pivotal effectors involved in the downstream signaling of PODXL responsible for regulating gene expression, might include Rac1, protein kinase C (PKC), mitogen-activated protein kinase (MAPK), transforming growth factor (TGF)-β, and activator protein (AP)-1 pathways (Fig. 6E). These data indicated that PODXL plays a crucial role in promoting tumor metastasis.