A total of 6 OS cell lines, U-2 OS, SW-1353, ZK-58, HOS, Saos-2 and MG-63, were used in the present study. All were commercially purchased from the American Type Culture Collection (ATCC; Manassas, MA, USA). A control human fetal osteoblast (hFOB) cell line was also purchased from the ATCC. All cell lines were cultured in RPMI-1640 medium (Qiagen China Co., Ltd., Shanghai, China). The RPMI-1640 medium was supplemented with 5% fetal bovine serum (FBS; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) in humidified 5% CO₂ at room temperature. The OS samples were obtained from surgical archives and were taken from patients registered at Xiangya Hospital, Central South University (Changsha, China) between May 2014 and July 2015. A total of 68 patients were included in this study (47–79 years; 38 male, 30 female). All patients provided informed, written consent. The protocols of experimental procedures using human specimens were approved by the Human Research Ethics Committee of Xiangya Hospital, Central South University (Changsha, China; approval no. 2014F0027).

In the present study, a lentiviral system was used to ectopically overexpress miR-336 in the OS cell lines HOS and MG-63. The pcDNA3.1 plasmids containing miR-336 mimic (3′-AGUACGUCAAGGCUCA-5′) or negative control were synthesized by and purchased from Tiangen Biotech Co., Ltd. (Beijing, China). The Sox-2 plasmid was synthesized by Tiangen Biotech Co., Ltd and inserted into a pcDNA3.1 plasmid. An empty pcDNA3.1 vector served as the negative control. Viral transfection into OS cell lines HOS and MG-63 were performed using a Lipofectamine 2000 system (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) according to the manufacturer's protocol. Following transfection for 36 h, the culture medium was replenished with fresh medium. All plasmids were experimentally verified by sequencing.

Total RNA was harvested from OS cell lines and human specimens with TRIzol reagent (Thermo Fisher Scientific, Inc.). The cDNA (a total of 10 µg) generated by reverse transcription was obtained using the miScript II RT kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer's protocol. A well-established TaqMan microRNA RT-qPCR kit (Applied Biosystems; Thermo Fisher Scientific, Inc.) was used to monitor miR-336 expression. To quantify Sox-2 mRNA expression levels, the SYBR-Green PCR Master kit (Takara Bio, Inc.) was used. The thermocycling conditions used were as follows: 50°C for 2 min, 95°C for 10 min followed by 30 cycles of 95°C for 15 sec and 60°C for 1 min. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control. Reactions were performed with the ABI PRISM^® 7000 Sequence Detection System (Applied Biosystems; Thermo Fisher Scientific, Inc.) following the manufacturer's protocol. Expression levels of miR-336 and Sox-2 were determined by the 2^−ΔΔCq method (17). The experiments were performed with ≥3 replicates.

The primer sequences used were as follows: miR-336, forward 5′-GATGCGACGTGAGTAAG-3′ and reverse 5′-CTGAGCCGGGTCCGAGGT-3′; GAPDH, forward 5′-CTCGATTGCATCGACATATCGT-3′ and reverse 5′-ACGCTTCGCGATCGTGCGTGAT-3′; Sox-2, forward 5′-GCTACAATAGCTCACCCTGAT-3′ and reverse 5′-CAATCTCCTGCCTACGAGAG-3′.

HOS and MG-63 cells were plated in a 96-well plate for 48 h (10⁴ cells/well) in RPMI-1640 medium (Qiagen China Co., Ltd.) supplemented with 5% FBS (Sigma-Aldrich; Merck KGaA). The miR-336 and negative control plasmids were transfected with luciferase reporter plasmids into the indicated OS cells, using the aforementioned procedure. The 3′-UTR for Sox-2 with binding site for miR-336 (5′ UGAUACUAUUUGAGCUAUU3′) was cloned into the Xbal site downstream of the Renilla luciferase reporter plasmid phRL-TK (Promega Corporation, Madison, WI, USA) leading to wild-type Sox-2 luciferase plasmids (Sox-2 3′-UTR WT). The mutated Sox-2 3′-UTR (5′ UGAUACUAUUUGUUUGCAU 3′) was similarly integrated into phRL-TK to generate a mutant control plasmid (Sox-2 3′-UTR mut). The luciferase activities were monitored with a Dual-Luciferase reporter system (Promega Corporation) as relative luciferase units according to the manufacturer's protocol.

The upper chamber of the Transwell plate was coated with Matrigel (Invitrogen; Thermo Fisher Scientific, Inc.) overnight. HOS and MG-63 cells were resuspended 24 h following transfection and loaded into the upper chamber (10⁴ cells/well) in RPMI-1640 medium. The lower chambers were supplemented with RPMI-1640 medium with additional 3% FBS. Following 24 h incubation, the upper chambers were removed and cells migrating into the lower chambers were fixed with 5% polytetrafluoroethylene (PFA) and stained with crystal violet. The Transwell assay results were assessed under a Leica fluorescent microscope (DM-IRB; Leica Microsystems GmbH, Wetzlar, Germany). A total of five fields were assessed and the experiments were performed in triplicate. Invasion was evaluated and normalized to the number of cells counted under control conditions.

A similar protocol was performed as stated for the invasion assay, in the absence of Matrigel. The final concentration of cells was ~10⁴ cells in each Transwell chamber. The lower chamber was loaded with RPMI-1640 medium supplemented with 5% FBS. Following incubation for 24 h, the cells were loaded into 5% PFA and stained with crystal violet.

The Cell Counting kit-8 (CCK-8; Dojindo Molecular Technologies, Inc., Kumamoto, Japan) was used to measure the proliferation of HOS and MG-63 cells. Following transfection for 36 h, as stated above, HOS and MG-63 cells were re-suspended and seeded into a 96-well plate (10⁴ cells/well) for 120 h. 25 µl MTT solution was added into the culture with final concentration of 10 mg/ml for 4 h. The solution was shaken for 5 min, leading to complete solubilization. The proliferation was evaluated once a day, for a total of 5 days. Crystalline formazan was dissolved in 200 µl 10% sodium dodecyl sulfate (SDS) solution for 24 h and the optical density at 490 nm was evaluated using the Spectramax M5 microplate monitor (Molecular Devices, LLC, Sunnyvale, CA, USA) following the manufacturer's protocol.

HOS and MG-63 cells were loaded into a 12-well plate (10⁴ cells/well, BD Biosciences, Franklin Lakes, NJ, USA) with Dulbecco's modified Eagle's medium supplemented with 5% FBS (Sigma-Aldrich; Merck KGaA) and washed with cold PBS solution 48 h post transfection. The apoptosis of tumor cells was quantified using a double staining (Annexin V-fluorescein isothiocyanate/propidium iodide) apoptosis detection toolkit (Sigma-Aldrich, Merck KGaA), following the manufacturer's protocol. A Becton Dickinson FACSCalibur (BD Biosciences) cytometer was used to analyze the data. Samples were processed with CellQuest software version 3.3 (BD Biosciences).

The human OS cell lines HOS and MG-63 were harvested using lysis buffer (15% glycerol and 3% NP-40) obtained from Qiagen China Co., Ltd. Samples were centrifuged at 15,000 × g at 4°C for 20 min. The protein extracts (100 µg in each lane) were electrophorised using 10% SDS-PAGE and migrated to nitrocellulose membranes (Bio-Rad Laboratories, Inc., Hercules CA, USA). The blot was blocked with 4% fat-free milk for 1 h at 20°C. The membrane was coated with rabbit anti-PAX7 antibody (1:1,000; catalog no. AV32393; Qiagen China Co., Ltd.) and anti-GAPDH antibody (catalog no. G9545; Qiagen China Co., Ltd.) at 4°C overnight. The horseradish peroxidase-conjugated secondary antibodies (mouse anti-human, 1:1,000; catalog no. SAB5201369; Sigma-Aldrich; Merck KGaA) were then added and incubated at 20°C for 2 h. Immunoblot signals were visualized using an ImageQuant^™ LAS 4000 mini-biomolecular imager (Fujifilm, Tokyo, Japan). The blots were quantified automatically using ImageJ software version 1.48 (National Institutes of Health, Bethesda, MD, USA). The experiment was performed in triplicate.

A total of 10⁶ HOS or MG-63 cells expressing miR-336 or negative controls were injected submucosally into BALB/c nude mice. The animal research was approved by the Ethics Committee of Xiangya Hospital, Central South University. A total of 12 mice (4–6 weeks; average weight, 15.5 g; 6 male and 6 female) were used in the study. Mice were housed at 20–22°C, 55–60% humidity with a light/dark cycle of 10–12 h. Ad libitum access to food and water was provided. The tumor volume was measured each week for a total of 5 weeks (the tumor volume ranged from ~500 to 1,600 mm³ at 5 weeks). At the end of the implantation study, all mice were sacrificed by overdose of sodium pentobarbital (5%, 300 mg/kg via intraperitoneal injection; catalogue no. 1507002; Sigma-Aldrich; Merck KGaA) and Ki-67 immunostaining was performed. Briefly, tumor samples were fixed in formalin, parraffin-embedded and cut into 5 µm sections with a microtome. Following deparaffinization and rehydration, antigens were retrieved with 1x Cytomation target retrieval solution (DakoCytomation GmbH, Hamburg, Germany) in a decloaker chamber at 95°C for 15 min and then at 20°C for 10 min. Slides were then incubated with hydrogen peroxide for 2 min. Following rinsing twice with TBS-0.2% Tween 20, slides were then analyzed using the Ki-67 ELISA kit (catalog no. Ek-M21211; Huayun Biotech, Guangzhou, China Co., Ltd.).

All statistical results were analyzed by SPSS 16.0 (SPSS, Inc., Chicago, IL, USA) using the Student's t-test. Pearson's correlation coefficient was used to assess the significance of any correlation between variables. All experiments were performed in at least triplicate. P<0.05 was considered to indicate a statistically significant difference.

As previously reported, Sox-2 expression levels are increased in OS cells (18). To confirm this, the expression level of Sox-2 protein was measured by western blot. The results revealed that, compared with hFOB cells, the expression of Sox-2 was visibly elevated in all OS cell lines (Fig. 1A). HOS and MG-63 cells were selected for further study because the Sox-2 levels were the highest (Fig. 1A). To further identify novel microRNAs involved in OS cell regulation, the online databases MIRDB (www.mirdb.org) (19), PicTar (pictar.mdc-berlin.de) and TargetScan (release 6.2, www.targetscan.org) (20) were used to predict the potential targets (Fig. 1B). miR-429 and miR-336 were predicted to be plausible candidates by means of overlapping results. miR-429 has been demonstrated to antagonize Sox-2 in colorectal cancer previously (21). However, few studies have identified the involvement of miR-336 in OS. Therefore, miR-336 was selected for further investigation. A significant negative correlation between miR-336 expression and Sox-2 mRNA expression levels in human OS samples was identified (R=−0.4996, P=0.0004; Fig. 1C). The negative correlation was also significant at the protein level of Sox-2 in human OS samples (R=−0.5845, P=0.0002; Fig. 1D). miR-336 expression levels were downregulated in human OS specimens compared with normal controls, whereas opposite expression patterns were observed in Sox-2 (Fig. 1C and D). A dual-luciferase reporter assay was also prepared with either wild type Sox-2 3′-UTR or a mutated miR-336 binding site (Fig. 1E). The results revealed that miR-336 significantly downregulated luciferase activities in both HOS and MG-63 cells transfected with wild type Sox-2 3′-UTR compared with the control (P<0.01; Fig. 1F and G, the control denotes HOS and MG-63 cells transfected with the empty pcDNA3.1 vector). However, miR-336 had no significant effect if HOS and MG-63 cells were transfected with plasmids containing mutated Sox-2 3′-UTR (Fig. 1F and G, respectively). These results suggested that miR-336 expression was downregulated in OS cells and was negatively correlated with Sox-2 expression.

To further establish the functional link between Sox-2 and miR-336, miR-336 was ectopically overexpressed using an miR-336 mimic in HOS and MG-63 cells. The RT-qPCR results demonstrated that miR-336 expression was significantly increased in cells transfected with miR-336 mimic compared with cells transfected with the empty pcDNA3.1 control vector (P<0.01; Fig. 2A). The results of the 5 day MTT assay revealed that miR-336 significantly inhibited the viability of HOS and MG-63 cells at 3, 4 and 5 days compared with the cells transfected with the empty pcDNA3.1 control vector (P<0.05; Fig. 2B). Further investigation revealed that ectopic miR-336 expression significantly suppressed migration in HOS and MG-63 cells compared with the cells transfected with the empty pcDNA3.1 control vector (P<0.01; Fig. 2C). In addition, the invasive capacity was also reduced in HOS and MG-63 cells transfected with a miR-336 mimic compared with the cells transfected with the empty pcDNA3.1 control vector (P<0.01; Fig. 2D). The apoptosis rate of OS cells was also demonstrated to be affected by miR-336 overexpression: miR-336 overexpression increased the overall apoptosis rate in HOS and MG-63 cells compared with cells transfected with the empty pcDNA3.1 control vector (Fig. 2E). These results suggested that overexpression of miR-336 inhibited tumor progression, potentially through inducing apoptosis.

As miR-336 was confirmed to inhibit OS cell malignancy through targeting Sox-2, overexpression of Sox-2 was assessed to determine whether it would reverse the adverse phenotype of OS cells. HOS and MG-63 cells were transfected with miR-336 plasmids, miR-336 plasmids+pcDNA3.1 or miR-336+Sox-2. Transfection of the empty pcDNA3.1 vector was used as a control for the Sox-2 plasmid transfection group. Transfection with an empty pcDNA3.1 vector control with a miR-336 mimic did not affect the expression of Sox-2 with miR-336 precursor cotransfection (Fig. 3A). However, transfection with Sox-2 visibly increased intrinsic Sox-2 levels in HOS and MG-63 cells, even with miR-336 cotransfection, compared with the miR-336 and miR-336 + pcDNA3.1 groups (Fig. 3A). The quantification results revealed that Sox-2 transfection increased Sox-2 protein expression levels by ~5 fold compared with the miR-336 and miR-336 + pcDNA3.1 groups (P<0.01; Fig. 3B). The migration and invasion of OS cell lines when Sox-2 levels were restored was then investigated. The results revealed that Sox-2 induction significantly promoted migration in HOS and MG-63 cells compared with the miR-336 and miR-336 + pcDNA3.1 groups (P<0.01; Fig. 3C). The invasion of HOS and MG-63 cells transfected with Sox-2 plasmids was also significantly increased compared with the miR-336 and miR-336 + pcDNA3.1 groups (P<0.01; Fig. 3D). These results suggested that restoring Sox-2 expression counteracted the effect of miR-336 in OS cell lines.

The aforementioned results demonstrated the in vitro effect of miR-336 in OS cell lines. However, whether miR-336 has a similar tumor suppressive function in vivo remains unclear. HOS and MG-63 cells stably transfected with an miR-336 mimic or the empty pcDNA3.1 vector were subcutaneously injected into nude mice. All mice were sacrificed according to formal protocols 5 weeks following implantation, and the volume of solid tumors was measured. miR-336-transfected HOS and MG-63 cells transfection produced smaller solid tumors compared with empty pcDNA3.1 vector controls (Fig. 4A). The difference in tumor volume became significant as early as 3 weeks following implantation in mice injected with HOS and MG-63 cells compared with controls (P<0.05; Fig. 4B). Ki-67 staining, which is indicative of proliferation in cells, was also visibly decreased following transfection with an miR-336 mimic compared with empty pcDNA3.1 vector controls (Fig. 4C). The RT-qPCR results confirmed that miR-336 expression was significantly upregulated in these solid tumors resulting from HOS and MG-63 cells transfected with an miR-336 mimic compared with empty pcDNA3.1 vector controls (P<0.01 and P<0.01, respectively; Fig. 4D). Sox-2 levels were also significantly downregulated in solid tumors resulting from HOS and MG-63 cells transfected with an miR-336 mimic compared with empty pcDNA3.1 vector controls (P<0.01 and P<0.01, respectively; Fig. 4E). These results suggested that miR-336 functions as a tumor suppressor in vivo.