After obtaining written informed consent, human oral cancer tissues (n=61) were collected from patients undergoing standard surgical procedures at The Affiliated Stomatological Hospital of Sun Yat-sen University (Guangzhou, China) between January 2008 and December 2012. The diagnosis was based on pathological evidence, and the specimens were immediately snap-frozen in liquid nitrogen. All patients received cisplatin-based neoadjuvant chemotherapy before surgical excision. The study protocol was approved by the Ethics Committee of The Affiliated Stomatological Hospital of Sun Yat-sen University.

Human oral cancer UM1, UM2, Cal27 and MD1386Ln cell lines were provided by Dr Zhou, University of Illinois at Chicago (Chicago, IL, USA). SCC9, SCC15 and SCC25 cell lines were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). The Tca8113 cell line was obtained from the Chinese Oral Tissue Culture and Collection Center (Shanghai, China). The cells were maintained in Dulbeccos modified Eagles medium (DMEM)/F12 (Life Technologies, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Gibco, Grand Island, NY, USA) in a humid atmosphere containing 5% CO₂ at 37°C.

cis-Diammineplatinum (II) dichloride was purchased from Sigma-Aldrich (St. Louis, MO, USA; P/N: P4394). Antibodies against β-catenin, GSK3β and β-tubulin were purchased from Cell Signaling Technology, Inc. (Danvers, MA, USA). The antibody against PPP2AB56α was purchased from ProteinTech (Chicago, IL, USA) (P/N: 12675-2-AP). Antibodies against MRP1, ABCG2, P-glycoprotein and topoisomerase IIβ were obtained from Abcam (Cambridge, MA, USA). The antibody against EZH2 was purchased from OriGene (Rockville, MD, USA).

miR-218 mimic (P/N: miR10000275) and inhibitor (anti-miR-218; P/N: miR20000275), PPP2R5A siRNA (P/N: stQ0004381-1) and β-catenin siRNA (P/N: stQ0002532-1) were obtained from RiboBio (Guangzhou, China). Cells were plated and cultured for 24 h and transfected with miR-218 mimic (20 nM), anti-miR-218 (50 nM) or siRNAs (50 nM). The transfection was performed using Lipofectamine RNAi Max (Invitrogen, Carlsbad, CA, USA). Lentiviral PPP2R5A (LV-PPP2R5A) and empty lentiviral vector (LV-NC) were designed and constructed by GenePharma Co., Ltd. (Shanghai, China), and they were transfected into tumor cells according to the manufacturer's instructions.

Total RNA was isolated from oral cancer tissues or cells using TRIzol reagent (Invitrogen). miR-218 was quantified using the mirVana™ qRT-PCR miRNA detection kit (Ambion) as previously described (Liu et al, 2009) according to the manufacturers instructions. Human U6 rRNA served as an internal control. To determine the relative mRNA levels of PPP2R5A and GAPDH (used as an internal control), 1 µg of total RNA was subjected to reverse transcription using a PrimeScript RT reagent kit (Takara Bio, Shiga, Japan). Real-time PCR was performed using a quantitative 2-step RT-PCR assay (Roche Diagnostics, Basel, Switzerland) according to the standardized protocol. The sequences of the specific primers for mRNA amplification were as follows: PPP2R5A forward primer, 5′-CTAGTCCCTCCTGACCCA-3′ and reverse primer, 5′-TAAATGCTGCTCAATCCC-3′; GAPDH forward primer, 5′-AACGGATTTGGTCGTATTG-3′ and reverse primer, 5′-GGAAGATGGTGATGGGATT-3′. The thermal cycling parameters were as follows: pre-incubation at 95°C for 5 min, followed by 40 cycles of amplification (95°C for 10 sec, 60°C for 20 sec, and 72°C for 20 sec), melting curve (95°C for 5 sec, 65°C for 1 min, and 97°C), and cooling (40°C for 10 sec). The relative expression level was computed using the 2^−ΔΔCt analysis method. All qRT-PCR reactions were evaluated in triplicate. All primers were ordered from Invitrogen.

Cell viability was measured using a tetrazolium salt-based Cell Counting Kit-8 assay (CCK-8; Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer's protocol. Briefly, the cells (5×10³/well) were seeded in 96-well plates and allowed to attach overnight. They were then cultured in fresh medium containing different concentrations of cisplatin (0, 0.25, 0.5, 1, 2, 4, 8, 16 and 32 µg/ml). After incubation for another 72 h, 10 µl of CCK-8 solution was added to each well and incubated at 37°C for 2 h. The absorbance of each well was assessed on a microplate reader (Tecan GENios, Zurich, Switzerland) at 450 nm. Cell viability was calculated as the ratio of each treatment group to the untreated control group. The IC₅₀ values were determined as the concentration resulting in a 50% reduction in cell viability compared with the control. All experiments were performed 3 times in quadruplicate.

Cells were grown in 6-well plates, transfected with the desired reagents for 48 h and treated with cisplatin after transfection. After treatment for 48 h, the cells were collected, washed with ice-cold phosphate-buffered saline (PBS) twice and gently resuspended in 1X binding buffer. After addition separately of 5 µl of Annexin V-FITC and 10 µl of propidium iodide, the cells were gently vortexed and incubated for 5 min in the dark. Flow cytometry (FACScan; Beckman Coulter, Brea, CA, USA) was used to assess apoptotic cells. Apoptotic cells were quantified using CellQuest software.

Total cell lysates were prepared using RIPA lysis buffer (50 mM Tris-HCl pH 7.4, 10 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate and 0.1% SDS). The protein concentration was measured using the BCA method (Beyotime, Beijing, China). Proteins were resolved by SDS-PAGE (10% gel) and subjected to immunoblot analysis using antibodies against MRP1 (1:1,000), ABCG2 (1:1,000), P-glycoprotein (1:500), topoisomerase IIβ (1:1,000), EZH2 (1:1,000), β-catenin (1:1,000), GSK3β (1:1,000), β-tubulin (1:2,000) or PPP2AB56α (1:1,000). The membrane was further probed with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:1,000 dilution; Beyotime) or goat anti-mouse IgG (1:1,000; Cell Signaling Technology, Inc.), and the bands were visualized using a Novex ECL HRP chemiluminescent substrate reagent kit (Invitrogen). The protein bands were quantified using Alpha Innotech imaging software (San Leandro, CA, USA).

A 55-bp fragment from the 3′ untranslated region (3′UTR) of the PPP2R5A gene (position 2586 to 2640; NM_006243.3) containing the miRNA-218 binding site was cloned into the XbaI site of the pGL3 firefly luciferase reporter vector (Promega, Madison, WI, USA). A corresponding mutant construct was created by mutating the seed region of the miR-218 binding site. The constructs were then verified by sequencing. Cells were transfected with the reporter constructs containing the targeting sequence from the PPP2R5A 3′UTR (named pGL-PPP2R5A WT) or its mutant (named pGL-PPP2R5A Mut) using Lipofectamine 2000 (Invitrogen). The pRL-TK vector (Promega) was co-transfected as an internal control for normalization of the transfection efficiency. The luciferase activities were then determined as previously described (30) using a Lumat LB 9507 luminometer (Berthold Technologies GmbH & Co. KG, Bad Wildbad, Germany).

All numerical data are presented as the mean ± standard deviation (SD), and all error bars represent the SD of the mean. The Student's t-test and one-way analysis of variance (ANOVA) were used to determine differences. The Chi-square or Fisher's exact tests were used to analyze the association between miR-218 and PPP2AB56α expression. All statistical analyses were performed using SPSS 16.0 software. A P-value <0.05 was considered significant.

First, we determined whether upregulation of miR-218 expression was a common feature of cisplatin-resistant oral cancer cells. UM1 and Cal27 cells, both of which are sensitive to cisplatin (cisplatin IC₅₀ values of 0.75 and 0.9 µg/ml, respectively), were continuously exposed to a gradient concentration of cisplatin for ~6 months until the cells acquired cisplatin resistance. Cell viability was determined using the CCK-8 assay in cell lines exposed to increasing concentrations of CDDP for 48 h. As illustrated in Fig. 1A, UM1cis and Cal27cis cells were more resistant to cisplatin than the parental cell lines, with IC₅₀ values of 5.23 and 8.42 µg/ml, respectively, which were ~7- and 9-fold higher than the IC₅₀ values for UM1 and Cal27. miR-218 expression was determined by quantitative real-time RT-PCR in cisplatin sensitive and resistant UM1 and Cal27 cells. As shown in Fig. 1B, cisplatin-sensitive cells expressed lower levels whereas cisplatin-resistant cell lines exhibited markedly increased levels of miR-218. Next, we investigated the endogenous miR-218 expression in an oral cancer cell line panel. As expected, the levels of miR-218 were significantly lower in CDDP-sensitive cell lines compared with CDDP-resistant cell lines (Fig. 1C and D). To confirm these results, we used oral cancer tissues from chemosensitive and non-sensitive patients and compared the expression profiles of miR-218. Chemosensitive or non-sensitive was determined based on the clinical assessment of patients treated with cisplatin-based neoadjuvant chemotherapy; those with a good therapeutic outcome were classified as sensitive, others were non-sensitive. It was found that miR-218 was increased in the non-sensitive oral cancers (4.16±0.51 vs. 1.38±0.25; P<0.05, data not shown).

To determine whether the modulation of miR-218 levels influences the response to cisplatin, we suppressed miR-218 expression in cisplatin-resistant cell lines by transfecting cells with anti-miR-218 (Fig. 2A). Both UM1 and Cal27 cells became more sensitive to cisplatin compared with the negative control group (Fig. 2B and C). In addition, a marked increase in apoptosis was observed in the anti-miR-218-transfected cells (Fig. 2D and E). No significant difference between the mock and the negative control group were found (data not shown). Furthermore, we also detected the expression levels of chemoresistance-related proteins. As expected, increased protein levels of MRP1, ABCG2, P-glycoprotein, topoisomerase IIβ (TopoIIβ) and EZH2 were found in the UM1cis cell line compared to the parental cell line UM1 (Fig. 2F and G). However, levels of these proteins were reduced when UM1cis cells were transfected with anti-miR-218 compared with levels in the anti-NC-transfected cells, whereas levels were increased in the UM1 cells transfected with the miR-218 mimic (Fig. 2F and G). These results indicate that miR-218 expression induces oral cancer resistance to cisplatin treatment.

To explore the potential target through which miR-218 induces cisplatin resistance in oral cancer cells, we predicted putative miR-218 targets using miRWalk, PicTar, TargetScan and miRanda. Our analysis identified PPP2R5A, a key subunit of PP2A, as a candidate miR-218 target (Fig. 3A). The putative targeted sequence found in the PPP2R5A 3′UTR [wild-type, (WT)] or the mutant sequence [mutant type (MUT)] was cloned into the luciferase reporter plasmid, respectively. Our results showed that miR-218 reduced the luciferase activity in the reporter construct containing wild-type PPP2R5A, but not the mutant 3′UTR in UM1 cells (Fig. 3B). Furthermore, the luciferase activity of the construct containing the wild-type 3′UTR, but not the mutant 3′UTR was increased in the UM1cis cells transfected with anti-miR-218 (Fig. 3C). The cells were further transfected with miR-218 or anti-miR-218 and were examined for PPP2R5A expression by qRT-PCR and western blot analysis. As shown in Fig. 3D and F, miR-218 transfection led to an obvious decrease in PPP2R5A mRNA and protein expression. Conversely, anti-miR-218 transfection significantly increased the expression level of PPP2R5A (Fig. 3E and F). These data confirmed that PPP2R5A is a direct target of miR-218 and may contribute to the effect of miR-218 on oral cancer chemoresistance.

To determine whether the expression levels of PPP2R5A influence the cisplatin response of oral cancer cells, we investigated the endogenous PPP2R5A expression in an oral cancer cell line panel. As shown in Fig. 4A, the levels of PPP2R5A were significantly higher in the CDDP-sensitive cell lines compared with levels in the CDDP-resistant cell lines. Moreover, a negative correlation between expression of PPP2R5A and miR-218 was identified (P=0.0149; Fig. 4B). Then, the UM1 and Cal27 cells were transfected with an siRNA against PPP2R5A or its control (siNC) and PPP2R5A protein (PPP2AB56α) was knocked down (Fig. 4C). Alternatively, UM1cis and Cal27cis cells were infected with lentiviruses expressing PPP2R5A (LV-PPP2R5A) or its control (LV-control) and PPP2AB56α was indeed overexpressed (Fig. 4D). The treated UM1, UM1cis, Cal27 and Cal27cis cells were further exposed to cisplatin to assess cell apoptosis. The results showed that suppression of PPP2R5A inhibited apoptosis in the sensitive cells (Fig. 4E), and overexpression of PPP2R5A induced apoptosis in the resistant cells (Fig. 4F).

To further explore whether miR-218 plays a role as an oncomiR through PPP2R5A, we evaluated the expression of PPP2R5A protein and its correlation with miR-218 expression in human oral cancer tissues. As expected, PPP2AB56α was upregulated in chemosensitive oral cancer tissues (Fig. 5A and B). Furthermore, a reverse correlation between PPP2AB56α and miR-218 expression was noted (Fig. 5C). Next, we performed rescue experiments to validate that PPP2R5A targeting is involved in miR-218-mediated drug resistance. To achieve this aim, UM1 and Cal27 cells were infected with miR-218 mimics or combined with LV-PPP2R5A, and cell viability was analyzed using the CCK-8 assay. It was found that miR-218 enhanced cell viability, and LV-PPP2R5A coexpression partially abrogated miR-218-induced cell growth (Fig. 6A). In contrast, cell viability results showed that anti-miR-218 could restore cisplatin sensitivity in a concentration-dependent manner in vitro, which could be partially reversed by the knockdown of PPP2R5A (Fig. 6B). Western blot analysis confirmed that PPP2AB56α in these cells was modulated by miR-218 (Fig. 6C). Taken together, our results revealed that miR-218 acts as an oncomiR and induces cisplatin resistance in oral cancers by targeting PPP2R5A.

Given that PPP2R5A is a repressor of Wnt signaling, we postulated that miR-218 overexpression in oral cancer cells may play a role in Wnt signaling. In UM1 and Cal27 cells, in which miR-218 is expressed at low levels and Wnt signaling remains inactive, we found that miR-218 overexpression resulted in increased levels of β-catenin and GSK3β proteins. When the cells were simultaneously transfected with miR-218 and LV-PPP2R5A, β-catenin protein levels were not affected (Fig. 7A). To determine whether miR-218 impairs cisplatin sensitivity specifically through the Wnt signaling pathway, we transfected UM1 and Cal27 cells with miR-218, miR-218 + LV-PPP2R5A, or a control prior to cisplatin treatment. Results from the flow cytometric analysis showed increased apoptosis of cells transfected with miR-218 + LV-PPP2R5A compared with those transfected with miR-218 (Fig. 7B). Finally, we transfected siRNA against β-catenin to validate the miR-218-induced Wnt activation. As shown in Fig. 7C and D, miR-218 activated Wnt signaling, whereas the siRNA against β-catenin abrogated the miR-218-promoting activation of Wnt signaling and cell survival. Collectively, these data indicate that miR-218 may increase cisplatin resistance via the Wnt signaling pathway in oral cancer cells by suppressing PPP2R5A expression.