Forty-two OSCC tissues and paired normal tissues were obtained from Peking Union Medical College Hospital. The informed consents of OSCC patients were collected before the experiment. All OSCC patients have not received chemotherapy or radiotherapy before undergoing surgery. This study was approved by the Institutional Ethics Committee of Peking Union Medical College Hospital, approval number was 2017PUMC26. All patients provided written informed consents.

OSCC cell line SCC-4 (ZKCC-X1937) and Immortalized Human Oral Mucosal Epithelial Cells hTERT-OME (ATCC^® PCS-200-014™) were obtained from Beijing Zhongke Quality Inspection Biotechnology Co., Ltd. and American Type Culture Collection (ATCC). These cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS and cultured in an incubator with 5% CO₂ at 37°C.

SNHG5 and FZD4 overexpression vectors or siRNAs, and miR-655-3p mimics (B01001) or inhibitor (B03001) were obtained from GenePharma. For transfection, cells were seeded into 6-well plates at a density of 2×10⁵ cells/well. Cells were transfected with miR-655-3p mimics or inhibitor (100 pmol), SNHG5 and FZD4 siRNA (100 pmol) or vectors (4 µg) using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) when 60–70% confluence was achieved, in accordance with the manufacturer's protocol. Subsequent to transfection for 6–8 h, cell culture medium was replaced with DMEM without antibiotics and incubated at 37°C with 5% CO₂.

TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.) was used to extract total RNA. RNA was reversely transcribed to complementary DNA (cDNA) using a Reverse Transcription Kit (Takara). RT-qPCR assay was performed using Real-time PCR Mixture assays (Takara) and corresponding primers. SNHG5 and miR-655-3p expression were normalized to U6, when FZD4 was normalized to GAPDH. Their expressions were quantified by the 2^−∆∆cq method. The primers used were: SNHG5 forward 5′-CGAGTAGCCAGTGAAGATAATG-3′; SNHG5 reverse 5′-CACACAACAGTCAAGTAAACC-3′; miR-655-3p forward: 5′-CAATCCTTACTCCAGCCAC-3′ and reverse, 5′-GTGTCTTAAGGCTAGGCCTA-3′; U6-forward: 5AAAT-3′ and reverse, 5′-TTGCGTGTCAT-3′; FZD4-forward: 5′-GGTGGCTCCCCTCTTTACTT-3′ and reverse, 5′-ATCACACACGTTGCAGAAC-3′; GAPDH forward: 5′-ACAACTTTGGTATCGTGGAAGG-3′, and reverse, 5′-GCCATCACGCCACAGTTTC-3′.

Prepared transfected SCC-4 cells (2×10³ cells/well) were put in a 96-well plate. Next, the SCC-4 cells were incubated in DMEM medium for 24, 48, 72 or 96 h. Then, the cells were incubated with 10 µl MTT solution for 4 h. MTT solution was then aspirated. And Formazan solution was added to completely dissolve the crystals. The absorbance at 490 nm was examined with a microplate reader (Olympus Corp.).

Matrigel matrix (1:8; 50 µl/well; BD Biosciences) in the upper chamber was used to detect cell invasion. After 30 min, the SCC-4 cell suspension (2×10³ cells/well) was added to the Transwell upper chamber. The lower chamber was added with DMEM medium (10% FBS). After 48 h of incubation, crystal violet (Beyotime Institute of Biotechnology) was used to fix and stain the cells on the lower surface of the membrane for 30 min. Cell migration assay was performed without Matrigel. The other steps are the same as cell invasion. An optical microscope was used to count the number of stained cells for five randomly selected fields.

The 3′-UTR of wild-type and mutant SNHG5 (wt-SNHG5 and mut-SNHG5) or FZD4 (wt-FZD4 and mut-FZD4) were inserted into the pmiR-GLO vectors (Promega Corporation). The above reporter plasmids and miR-655-3p mimics were transfected into SCC-4 cells. After 48 h, the activity of firefly and Renilla luciferase was measured by the dual-luciferase reporter gene system (Promega Corporation).

Protein samples were extracted by RIPA lysis buffer (Beyotime Institute of Biotechnology). Protein concentration was measured using Bicinchoninic Acid protein assay kit (Sigma-Aldrich; Merck KGaA). Next, 25 µg protein was separated by 10% SDS-PAGE and transferred to PVDF membranes. And the protein was blocked at room temperature for 2 h with 5% non-fat milk. Then, protein samples were incubated with FZD4 (ab83042, 1:1,000 dilution, Abcam, rabbit polyclonal antibodies) and GAPDH (ab8245, 1:1,000 dilution, Abcam, mouse monoclonal antibody) primary antibodies overnight at 4°C. After washing with TBST, protein samples were incubated with corresponding horseradish peroxidase-conjugated secondary antibodies (ab205719, 1:5,000 dilution; Abcam). Protein bands were visualized by ECL kit (Beyotime Institute of Biotechnology) and were quantified with Image Lab Software (Bio-Rad Laboratories, Inc.).

All statistical analyses were conducted by using SPSS 19.0 or Graphpad Prism 6. All data were presented as the mean ± standard deviation from at least three separate experiments. Differences between two groups were estimated by unpaired two-tailed Student t-test. Differences between multiple groups were compared using one-way analysis of variance followed by Tukey's post hoc test. The correlations between miR-655-3p expression and lncRNA SNHG5 or FZD4 expression in OSCC tissues were analyzed using Spearman's rank test. P<0.05 indicates a statistically significant difference.

The role of lncRNA SNHG5 was explored in OSCC. First, SNHG5 expression was detected in OSCC tissues and cells. Compared to normal tissues, the expression of SNHG5 in OSCC tissues was increased (Fig. 1A). In addition, higher expression of SNHG5 was also found in SCC-4 cells than that in the hTERT-OME cells (Fig. 1B). Next, SNHG5 siRNA or overexpression vector was transfected into SCC-4 cells. We found that the expression of SNHG5 was reduced by its siRNA and was enhanced by its overexpression vector in SCC-4 cells (Fig. 1C). In addition, upregulation of SNHG5 promoted cell proliferation, while knockdown of SNHG5 restrained cell proliferation in SCC-4 cells (Fig. 1D). Meanwhile, SCC-4 cell migration was promoted by SNHG5 overexpression and was inhibited by SNHG5 downregulation (Fig. 1E). The same effect of SNHG5 on cell invasion was also examined in SCC-4 cells (Fig. 1F). Briefly, upregulation of lncRNA SNHG5 promotes cell proliferation, migration and invasion in OSCC.

lncRNA SNHG5 was found to have a binding site with miR-655-3p in the starBase database (http://starbase.sysu.edu.cn/, Fig. 2A). Dual luciferase reporter was used to verify the relationship between them. It was found that miR-655-3p mimics reduced the luciferase activity of wt-SNHG5, but had little effect on mut-SNHG5 luciferase activity in SCC-4 cells (Fig. 2B). In addition, we found that SNHG5 was negatively correlated with the expression of miR-655-3p in OSCC tissues (Fig. 2C). Next, SNHG5 siRNA or vector and miR-655-3p mimics or inhibitor were transfected into SCC-4 cells, respectively. RT-qPCR showed that miR-655-3p expression was reduced by SNHG5 upregulation and enhanced by SNHG5 downregulation in SCC-4 cells (Fig. 2D). Meanwhile, miR-655-3p mimics reduced SNHG5 expression, while miR-655-3p inhibitor promoted SNHG5 expression in SCC-4 cells (Fig. 2E). These results indicate that SNHG5 is a competitive miRNA of miR-655-3p in OSCC.

Next, the abnormal expression and function of miR-655-3p was investigated in OSCC development. Compared to normal tissues, miR-655-3p expression in OSCC tissues was found to be downregulated (Fig. 3A). And compared with hTERT-OME cells, downregulation of miR-655-3p was detected in SCC-4 cells (Fig. 3B). To explore the function of miR-655-3p in OSCC, miR-655-3p mimics or inhibitor were transfected into SCC-4 cells. RT-qPCR showed that miR-655-3p mimics increased its expression, while miR-655-3p inhibitor reduced its expression in SCC-4 cells (Fig. 3C). In addition, this increased expression of miR-655-3p was reduced by SNHG5 upregulation (Fig. 3C). More importantly, cell proliferation was restrained by miR-655-3p overexpression and promoted by miR-655-3p downregulation in SCC-4 cells (Fig. 3D). Similarly, miR-655-3p mimics also inhibited SCC-4 cell migration and invasion, miR-655-3p inhibitors also promoted the migration and invasion of SCC-4 cells (Fig. 3E and F). In addition, upregulation of SNHG5 weakened the inhibitory effect of miR-655-3p on cell proliferation, migration and invasion in SCC-4 cells (Fig. 3D-F). In conclusion, miR-655-3p acts as a tumor suppressor in OSCC by competitively binding to SNHG5.

Then, the TargetScan database (http://www.targetscan.org) predicts that FZD4 is the target of miR-655-3p (Fig. 4A). Luciferase reporter assay was performed to verify the prediction. It was found that miR-214-3p mimics reduced the luciferase activity of wt-FZD4, but had no effect on mut-FZD4 luciferase activity (Fig. 4B). In addition, FZD4 was negatively correlated with the expression of miR-655-3p in OSCC tissues (Fig. 4C). At the same time, miR-655-3p mimics reduced FZD4 expression, while miR-655-3p inhibitors promoted FZD4 expression in SCC-4 cells (Fig. 4 and E). Besides that, we also found that lncRNA SNHG5 was positively correlated with the expression of FZD4 in OSCC tissues (Fig. 4F). These results indicate that FZD4 is a direct target of miR-655-3p.

To elucidate the regulatory mechanism of lncRNA SNHG5, miR-655-3p and FZD4, the SNHG5 vector or miR-655-3p inhibitor was transfected into SCC-4 cells with FZD4 siRNA (si-FZD4). First, upregulation of FZD4 was examined in OSCC tissues and cells (Fig. 5A and B). And we found that si-FZD4 reduced its expression in SCC-4 cells. However, upregulation of SNHG5 or downregulation of miR-655-3p restored this decrease in FZD4 expression (Fig. 5C and D). Functionally, FZD4 silencing inhibited SCC-4 cell proliferation. The SNHG5 vector or miR-655-3p inhibitor weakened the inhibitory effect of si-FZD4 on cell proliferation in SCC-4 cells (Fig. 5E). Moreover, knockdown of FZD4 suppressed the migration and invasion of SCC-4 cells. And upregulation of SNHG5 or downregulation of miR-655-3p eliminated the inhibitory effect of si-FZD4 on SCC-4 cell migration and invasion (Fig. 5F and G). Therefore, FZD4 promotes the progression of OSCC by interacting with the SNHG5/miR-655-3p axis.