Six OSCC cell lines SCC-4 (Cat. no. CRL-1624), SCC-15 (Cat. no. CRL-1623), CAL-27 (Cat. no. CRL-2095) [all from American Type Culture Collection (ATCC), Manassas, VA, USA], HSC-3 (RCB1975), HSC-2 (RCB1945) (both from Riken Cell Bank, Tsukuba, Japan) and Ca9-22 (JCRB0625) [from the Japanese Collection of Research Bioresources (JCRB), Osaka, Japan] were used in this study. All cells were cultured in Dulbecco's modified Eagle's medium (DMEM). The medium was supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO, USA), 100 U/ml penicillin and 100 μg/ml streptomycin at 37°C in a 5% CO₂ atmosphere. A non-tumorigenic immortalized oral keratinocyte line (HOK‑16B, generous gift from Dr No-Hee Park, University of California, Los Angeles, CA, USA) was maintained in oral keratinocyte medium, supplemented with 1% keratinocyte growth factor plus epithelial growth factor mixture (Invitrogen, Carlsbad, CA, USA). A total of 42 freshly frozen OSCC tissues, as well as 42 matched controls were obtained from the Department of Oral and Maxillofacial Surgery, Shenzhen Hospital of Southern Medical University, Shenzhen, China. None of the patients with OSCC had received radiotherapy or chemotherapy prior to surgery. This study protocol conformed to the Ethics Committee of Shenzhen Hospital of Southern Medical University (Shenzhen, China). All human materials were obtained with informed consent and approved by the Ethics Committee of Shenzhen Hospital of Southern Medical University.

The micro-array data was downloaded from the open GEO database (https://www.ncbi.nlm.nih.gov/geo/) and the GEO accession number is GSE3524 (24). These microarray expression data were analyzed by GEO2R bioinformatics software (http://www.ncbi.nlm.nih.gov/geo/geo2r/), which can analyze any GEO series. The adjusted P-values (adj. P-value) using the Benjamini and Hochberg (BH) false discovery rate (FDR) method by default were applied to correct for the occurrence of false‑positive results. An adj. P<0.05 and a |logFC|≥1 were set as the cut-off criteria. A heatmap of the 49 lncRNAs which the most significant differences in expression was generated using the online tool Morpheus (https://software.broadinsti-tute.org/morpheus/).

For lncRNA analysis, total RNA was isolated from the cells and tissues using TRIzol reagent (Invitrogen) according to the manufacturer's instructions and reverse transcribed using the Superscript III first strand synthesis system (Life Technologies, Carlsbad, CA, USA). For miRNA analysis, RNA was extracted from the liver tissues using the miRNeasy mini kit (Qiagen, West Sussex, UK) according to the manufacturer's instructions. The RNA was then reverse transcribed into cDNA. Amplifications were carried out on an ABI 7500 Real-Time PCR system (Life Technologies) using SYBR-Green according to the manufacturer's instructions. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or U6 snRNA were used as endogenous controls. Data were analyzed using 7500 software v.2.0.1 (Applied Biosystems, Foster City, CA, USA), and calculated using the 2^−∆∆Cq method (25). All experiments were performed in triplicate.

The miR-138 mimic, miR-138 inhibitor, mimics negative control (mimics NC) and inhibitor NC were purchased from RiboBio Co., Ltd. (Guangzhou, China). A small interfering RNA against H19 (si-H19) and si-Scramble were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Plasmid cDNA-H19 was constructed by introducing a BamHI-EcoRI fragment containing the H19 cDNA into the same site in the pcDNA3.1 plasmid (Invitrogen). Cell transfections with si-H19, pcDNA-H19 and miR-138 mimic and inhibitor were performed as previously described (26). The HSC-2 and Ca9-22 cells were transfected with H19 siRNA, miR-138 mimic or miR-138 inhibitor using the Lipofectamine^® RNAiMAX kit (Thermo Fisher Scientific, Waltham, MA, USA). The CAL-27 cells were transfected with pcDNA-H19 or pcDNA-H19 plus miR-138 mimic using Lipofectamine^® 2000 (Life Technologies) according to the manufacturer's instructions. The cells were harvested after 48 h and used for further analysis.

Lentiviral constructs carrying shRNA targeting H19 (Lv-shRNA), and an empty negative control vector (LV-GFP) were obtained from GenePharma (Shanghai, China). The 293 cells (CRL-1573, ATCC) were co-transfected with Lenti-Pac HIV Expression Packaging Mix and the lentiviral vectors (or the control lenti-virus vectors) using Lipofectamine 2000 (Life Technologies). After 48 h, lentiviral particles in the supernatant were harvested and ultra-centrifuged to concentrate the lentiviral particles. Subsequently, the HSC-2 and Ca9-22 cells grown on 6-well plates were transduced with the lentiviruses of 10 transduction units (TU) per cell. The expression of H19 in the transduced cells was examined by real-time PCR analysis.

Cell proliferation was examined by MTT assay according to the manufacturer's instructions (Roche Applied Science, Basel, Switzerland). Briefly, approximately 1×10³ cells were seeded in a 96-well culture plate for 24 h. The cells were then transfected with si-H19, si-Scramble, miR-138 inhibitor or miR-control. At various time points, 0.5 mg/ml MTT solution was added to each well. The absorbance was then recorded at 490 nm on a Bio-Rad model 680 microplate reader, (Bio-Rad Laboratories, Hercules, CA, USA). For the analysis of apoptosis, the cells were stained with Annexin V-FITC and propidium iodide (PI) (TACS Annexin V-FITC, Trevigen Inc., Gaithersburg, MD, USA) and then analyzed with double‑label flow cytometry on a flow cytometer (FACSCanto II; BD Biosciences, San Jose, CA, USA). For cell cycle analysis, the cells were resuspended in PBS, stained with PI containing RNase A for 30 min at 37°C, and analyzed by flow cytometry. All the assays were conducted in triplicate.

All animal procedures were performed according to national guidelines and approved by the Animal Care Ethics Committee of Shenzhen Hospital of Southern Medical University. Twenty female BALB/c nude mice (weighing 20±2 g; 4 weeks old, Laboratory Animal Center of Shanghai, Academy of Science) were used in this study. The HSC-2 and Ca9-22 cells transfected with Lv-shRNA or Lv‑Control were injected into the left flanks of the nude mice (2×10⁶ cells/mouse). After 5 weeks, the mice were sacrificed and tumor tissues were dissected. The tumor weight was measured and the tumor volume was calculated according to the following formula: (length x width²)/2. The largest tumor size and volume in the mice of different groups was 1.30 g and 1,500 mm³ (Lv-Control-transfected HSC-2 cells), 1.25 g and 1,700 mm³ (Lv-Control transfected Ca9-22 cells), 0.68 g and 1,150 mm³ (Lv-siRNA transfected HSC-2 cells), 0.73 g and 1,300 mm³ (Lv-siRNA transfected Ca9-22 cells).

In silico prediction of miRNA binding sites within the H19 3′UTR was performed using TargetScan (www.targetscan.org) and PicTar (http://www.pictar.org).

The 3′-UTR of H19, with wild-type or mutant (Mut) binding sites for miR-138, was amplified and cloned into the pGL3 vector (Promega, Madison, WI, USA) to generate the plasmid pGL3-WT-H19-1 (wt-H9-1) or pGL3-WT-H19-2 (wt-H9-2). The putative binding site of miR-138 in H19 was mutated by using a QuikChange Site Directed Mutagenesis Kit (Agilent, Santa Clara, CA, USA) to synthetize mutant type pGL3‑mut‑H19‑1 (Mut‑H19‑1) or pGL3-mut-H19-2 vector (Mut-H19-2). For the luciferase reporter assay, the 293 cells were co-transfected with the luciferase reporter vectors and miR-138 mimics or corresponding negative control (GenePharma) using Lipofectamine 2000 reagent (Life Technologies). The pRL-TK plasmid (Promega) was used as a normalizing control. After 48 h of incubation, luciferase activity was analyzed using the Dual‑Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions.

The cells were plated in 6-well plates and transfected when cultured to 95% confluence. The cell layers were then scratched using a 10 μl plastic pipette tip to produce wounds. The wounds were photographed at 0 and 48 h under an inverted phase contrast microscope (IX71; Olympus Corp., Tokyo, Japan). Three random fields were marked and measured. All the assays were carried out in triplicate.

For invasion assays, a total of 3×10⁴ HSC-2 and Ca9-22 cells in 150 μl serum-free medium at post-transfection were seeded into the upper chamber (24‑well insert, pore size 8 μm; Corning, NY, USA) pre‑coated with 30 μg/well Matrigel solution (BD Biosciences), and the lower chambers were filled with 500 μl of 10% FBS medium. Following incubation at 37°C for 48 h, the membranes were fixed with 4% polyoxymethylene and stained with 0.1% crystal violet (Sigma-Aldrich). Five pre-determined fields were counted under a microscope (Olympus Corp., original magnification, ×200). All assays were performed in triplicate.

Total proteins were extracted from cells using radioimmunoprecipitation assay (RIPA) lysis buffer (Sigma‑Aldrich) and quantified with a Bicin Choninic Acid (BCA) protein assay kit (Pierce, Rockford, IL, USA). A total of 40 μg of protein were subjected to 10% SDS-PAGE, and subsequently transferred onto a polyvinylidene difluoride membranes (Millipore, Billerica, MA, USA). The blots were incubated with the primary antibodies specific for cleaved caspase-3 (1:500; Cat. no. 9661), cleaved poly(ADP-ribose) polymerase (PARP; 1:1,000; Cat. no. 5625), Bax (1:1,000; Cat. no. 5023), EZH2 (1:1,000; Cat. no. 5246), zinc finger E‑box‑binding homeobox 1 (ZEB1; 1:1,000; Cat. no. 3396), E-cadherin (1:1,000; Cat. no. 14472), vimentin (1:1,000; Cat. no. 5741), N-cadherin (1:1,000; Cat. no. 13116) and β-actin (1:2,000; Cat. no. 4970). All antibodies were obtained from Cell Signaling Technology, Inc. (Danvers, MA, USA). Following 3 washes in TBST, the membranes were incubated with corresponding horseradish peroxidase (HRP)-conjugated secondary antibody (1:10,000; Santa Cruz Biotechnology, Inc.) for 2 h at room temperature, and washed with TBST 3 times. The protein bands were visualized by ECL detection reagent (GE Healthcare Life Sciences, Piscataway, NJ, USA). The intensity of the protein fragments was quantified with the Quantity One software (4.5.0 basic; Bio-Rad).

Statistical analyses were performed with GraphPad Prism 5 software (GraphPad Software, Inc., La Jolla, CA, USA). All data are presented as the means ± SD. Differences were analyzed by a Student's t‑test between two groups or one-way analysis of variance (ANOVA), followed by Tukey's multiple comparison tests between multiple groups. Spearman's analysis was used in correlation analysis. Survival analysis under the circumstance of Kaplan-Meier method. A P-value <0.05 was considered to indicate a statistically significant difference.

To identify the lncRNAs involved in the development and progression of OSCC, a GSE dataset was obtained from the GEO database under the accession number GSE3524 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE3524). Analysis of these data revealed that 24 lncRNAs were downregulated and 25 miRNAs were upregulated in the tumor group, compared with the normal group. Among the aberrantly expressed lncRNAs, lncRNA H1 was the most significantly upregulated lncRNA in GSE dataset (Fig. 1A). To validate the microarray analysis finding, we detected H19 expression in a cohort of 42 paired tumor tissues and normal tissues. The detailed patient clinical data are presented in Table I. As shown in Fig. 1B, H19 expression was increased in tumor tissues compared with normal tissues. These data suggested that H19 may be involved in the process of carcinogenesis.

To determine whether H19 expression was associated with the grade of malignancy and nodal invasion in OSCC, we examined the expression level of H19. We demonstrated that H19 expression was positively associated with the pathological grades of OSCC and nodal invasion (Fig. 1C and D). Based on the relative expression ratios of <0.5, the 42 clinical cases were divided into 2 groups as follows: the H19 low expression group (n=17) and the H19 high expression group (n=25). We then assessed whether the expression of H19 correlated with the post-operative survival time of patients with OSCC. Kaplan-Meier survival analysis revealed that patients with a high H19 expression had a poorer overall survival than those with a low H19 expression (Fig. 1E). Taken together, these findings indicate that H19 may be used as a biomarker for the diagnosis and prognosis of OSCC.

To examine the effect of H19 on cell growth, we first measured the expression levels of H19 in 6 OSCC cell lines (HSC-2, HSC-3, SCC-4, SCC-15, CAL-27 and Ca9-22) and a normal oral mucosa cell line (HOK-16B). As shown in Fig. 2A, a higher expression of H19 was observed in the OSCC cell lines compared with the normal oral mucosa cell line, particularly in the HSC-2 and Ca9-22 cells. Subsequently, the HSC-2 and Ca9-22 cells were transiently transfected with si-H19 and the endogenous level of H19 was effectively decreased (Fig. 2B). As demonstrated by MTT assays, we found that H19 silencing significantly decreased the viability of the HSC‑2 and Ca9‑22 cells (Fig. 2C and D). Furthermore, the proportion of the cell population undergoing apoptosis was increased after knockdown of H19 in the HSC-2 and Ca9-22 cells (Fig. 2E–G). Flow cytometry also revealed a significant promotion of cells in the G0/G1 phase of the cell cycle in the HSC-2 and Ca9-22 cells transfected with si-H19 (Fig. 2H and I). In addition, western blot analysis revealed that the expression levels of apoptosis-related proteins, including cleaved caspase-3 and cleaved PARP and Bax were markedly increased after H19 knockdown (Fig. 2J–L). Our results thus revealed that H19 knockdown inhibited several malignancy-related parameters of OSCC in vitro.

To evaluate the functional roles of H19 in vivo, we established a xenograft mouse mode in which Lv-shRNA or Lv-Control-transfected HSC-2 and Ca9-22 cells were transplanted into the flanks of BALB/c nude mice. Consistent with the results obtained in vitro, after 5 weeks, tumor volumes in the mice injected with cells from the si-H19 group were markedly smaller compared with those in the mice injected with cells from the Lv-Control (Fig. 3A and B). Similarly, tumor weights in the mice injected with cells from the Lv‑shRNA group were significantly lower compared with those in the mice injected with cells from the Lv-Control group (Fig. 3C and D). These results indicate that the knockdown of H19 expression inhibits tumor growth in vivo.

Based on the above-mentioned results in that the expression of H19 was higher in metastatic tissues, we hypothesized that H19 may be associated with the metastasis of OSCC. Thus, wound healing and Transwell assays were performed to examine the effects of H19 on OSCC cell metastasis. As shown in Fig. 4A, H19 silencing markedly inhibited the migration of monolayer-cultured HSC-2 and Ca9-22 cells. In addition, the numbers of invaded cells were markedly attenuated in the cells in which H19 was knocked down compared with the control cells (Fig. 4B). It is well known that EMT plays a critical role in the invasion and metastasis of OSCC cells (27). In this study, we thus assessed the effect of H19 on the expression of EMT related-genes. Our results demonstrated that H19 silencing significantly decreased the expression levels of vimentin and N-cadherin (mesenchymal markers), but increased the expression levels of ZEB1 and E-cadherin (epithelial markers) (Fig. 4C). These data suggest that the knockdown of H19 suppresses OSCC cell metastasis by inhibiting EMT.

Recently, studies have confirmed that lncRNAs can function as ceRNAs or molecular sponges that modulate miRNAs in cancers (28,29). In this study, we thus performed bioinformatics analysis using TargetScan and PicTar, and found that miR-138 contains a binding site for H19. The predicted binding sites for H19 in the miR-138 sequence are illustrated in Fig. 5A. To examine whether H19 directly targets miR-138, we conducted a luciferase assay. As shown in Fig. 5B, luciferase activity was significantly inhibited when wt‑H19‑1 or wt-H19-2 was co-transfected with miR-138 mimics compared with that after mimic NC co-transfection, whereas the inhibitory effect was abolished when the H19 3′-UTR was mutated. This indicated that miR-138 probably interacted with H19.

It has been found that miR-138 functions as a tumor suppressor in human OSCC (23). Thus, we first measured miR-138 expression in 6 OSCC cell lines (HSC-2, HSC-3, SCC-4, SCC-15, CAL-27 and Ca9-22) and a normal oral mucosa cell line (HOK-16B). Consistent with the findings previous of that study, the expression of miR-138 was also expressed at low levels in the OSCC cell lines, particularly in the HSC-2 and Ca9-22 cells (Fig. 5C). To further determine whether H19 affects OSCC cell proliferation and metastasis by regulating miR-138 expression, we measured miR-138 expression after the silencing or overexpression of H19 in HSC-2 and CAL-27 cells, and the results of RT-qPCR indicated that miR-138 expression was upregulated after the knockdown of H19 in the HSC-2 cells (Fig. 5D), whereas the miR-138 level was downregulated after H19 was overexpressed in the CAL-27 cells (Fig. 5E). Furthermore, we detected miR-138 expression in 42 pairs of OSCC tissues and normal tissues. Consistent with the results obtained with the OSCC cell lines, the expression of miR-138 was downregulated in the OSCC tissues (Fig. 5F), and an inverse correlation was observed between the expression of miR-138 and the expression of H19 (Fig. 5G). All these data suggest that H19 negatively regulates the expression of miR-138 in OSCC.

Recent findings have demonstrated that EZH2 acts as an oncogene and correlates with the malignant potential and a poor prognosis in a wide range of cancer types, including OSCC (30–32). Importantly, EZH2 has been identified as a target of miR‑138 in several cancer cells (33–35). Therefore, we sought to determine whether H19 functions as a ceRNA for miR-138 to regulate the expression of EZH2 in OSCC. Western blot analysis revealed that the knockdown of H19 in the HSC‑2 cells significantly decreased the expression levels of EZH2, while transfection with miR‑138 inhibitor restored EZH2 expression in the OSCC cells in which H19 was knocked down (Fig. 6A). By contrast, the overexpression of H19 in the CAL-27 cells increased the expression levels of EZH2, while transfection with miR‑138 mimic inhibited the promotion of EZH2 expression in the cells overexpressing H19 (Fig. 6B). These results suggest that H19 regulates the expression of the oncogene EZH2 by competing with miR-138 in OSCC cells.

In order to analyze the importance of miR‑138 in H19-mediated OSCC proliferation, apoptosis and invasion, we knocked down the expression of miR-138 in the HSC-2 cells transfected with si-H19 and overexpressed its expression in the CAL-27 cells transfected with pcDNA-H19, separately. The results revealed that miR-138 knockdown blocked the inhibitory effects of H19 on cell proliferation, apoptosis and invasion (Fig. 6C–J), suggesting that the effects of H19 on OSCC growth and invasion are partially mediated by miR-138. Taken together, these data indicated that the effects of H19 on the growth and metastasis of OSCC are partially mediated by regulating the expression of miR-138.