The normal human fetal osteoplastic cell line (hFOB) and human osteosarcoma cell lines SaOS2, 143B, MG63 and U2OS were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). The cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific) at 37°C in a humidified 5% CO₂ atmosphere. Twenty paired osteosarcoma and adjacent normal tissues were received from patients in the Second Affiliated Hospital of Xi'an Jiaotong University from January to March, 2017 (Xi'an, China). All patients provided informed written consent and the study was approved by the Institutional Research Ethics Committee of Xi'an Jiaotong University.

MG63 and U2OS cells were seeded into 6-well plates (1×10⁵ cells/well) and grown for 24 h. Subsequently, MG63 and U2OS cells were transfected with 50 µM of GPNMB siRNAs, siGPNMB-1 and siGPNMB-2 or negative control siRNAs, siNC-1 and siNC-2 (all from Thermo Fisher Scientific, Inc.) for 48 h using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific). The transfection efficiency was detected using qPCR and western blot analysis.

The effect of GPNMB on cell viability was assessed by an MTT assay, as previously reported (21). The cells were seeded into 96-well plates at 1×10⁴ cells/well. Subsequently, 20 µl of MTT (5 mg/ml; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) was added at 24, 48 and 72 h and incubated for 4 h. The supernatant was aspirated. Dimethyl sulfoxide (DMSO; 200 µl; Beijing Dingguo Changsheng Biotechnology Co., Ltd., Beijing, China) was added to dissolve formazan crystals. Absorbance at 490 nm was detected using a microplate reader (Bio-Tek Instruments, Inc., Winooski, VT, USA).

A 5′-bromo-2′-deoxyuridine (BrdU) assay was performed to assess the effect of GPNMB on cell proliferation, as previously described (22). Cells (1×10⁵ cells/ml) grown on coverslips were incubated with BrdU (Sigma-Aldrich; Merck KGaA) for 40 min and stained with anti-BrdU antibody (1:200; cat. no. sc-70443; Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 2 h. Images were captured under an MC 100 microscope (Carl Zeiss, Oberkochen, Germany), assessing the percentages of BrdU in six random fields.

To prevent proliferation, the cells were cultured in 10 g/ml of mitomycin C (Sigma-Aldrich; Merck KGaA) for 2 h (23). Then, a Transwell invasion assay was performed using an invasion chamber coated with Matrigel (BD Biosciences, San Jose, CA, USA). The lower chamber was filled with 600 µl of medium. Cells (1×10⁵ cells/well) were plated into the upper Transwell chambers in serum-free medium for 24 h at 37°C. The cells on the surface of the upper chamber were wiped with a cotton swab. After fixing with methanol for 20 min and staining with 0.1% crystal violet for 30 min, the invasive cells on the surface of the bottom membrane were determined by counting five random 100X fields under an MC 100 microscope (Carl Zeiss). The migration assay was the same as the invasion assay, with the exception of the upper chamber which was not coated with Matrigel. Each experiment was performed in triplicate.

Total RNA was extracted using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.). The synthesis of cDNA was performed with the PrimeScript^™ RT reagent kit (Takara Biotechnology, Co., Ltd., Dalian, China). The qPCR was carried out using SYBR-Green Premix (Takara Biotechnology, Co., Ltd.). The Bio-Rad CFX96 Touch qPCR system (Bio-Rad Laboratories, Hercules, CA, USA) was used to analyze the signal. The primers for qRT-PCR were as follows: GPNMB forward, 5′-ACAAGGAATACAACCCAATA-3′ and reverse, 5′-ATAGCCACTCCAGCACA-3′; GAPDH forward, 5′-GACTCATGACCACAGTCCATGC-3′ and reverse, 5′-AGAGGCAGGGATGATGTTCTG-3′. The expression of mRNAs was normalized to GAPDH. The relative expression was calculated using the 2^−ΔΔCt method.

Proteins were resolved in 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and then the separated proteins were transferred to polyvinylidene fluoride membranes (Invitrogen; Thermo Fisher Scientific, Inc.). Subsequently, the membranes were blocked in 5% bovine serum albumin (BSA) for 2 h and probed with primary antibodies, including rabbit anti-GPNMB (1:1,000; cat. no. ab98856), rabbit anti-p-Akt (1:1,000; ab38449), rabbit anti-Akt (1:500; cat. no. ab8805), rabbit anti-p-mTOR (1:1,000; cat. no. ab109268) and rabbit anti-mTOR (1:2,000; cat. no. ab2732; all from Abcam, Cambridge, MA, USA) overnight at 4°C. Subsequently, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (1:5,000; cat. no. ab6721; Abcam). The immunoblots were visualized with an enhanced chemiluminescence system (Pierce Biotechnology; Thermo Fisher Scientific, Inc.). Densitometry analysis was performed using Image-Pro Plus 6.0 software (Media Cybernetics Inc., Rockville, MD, USA).

The data were analyzed using SPSS 22.0 software (IBM Corp., Armonk, NY, USA) and are presented as the mean ± SD. Statistical analysis was performed using one-way ANOVA with subsequent SNK-q testing for multiple comparisons. P<0.05 was considered to indicate a statistically significant difference.

The expression of GPNMB in human osteosarcoma tissues and human osteosarcoma cell lines was determined by qPCR and western blot analysis. The human osteosarcoma tissues derived from 20 patients demonstrated significantly higher GPNMB mRNA levels than the adjacent non-cancerous tissues (Fig. 1A). The mRNA and protein expression of GPNMB in the osteosarcoma cell lines SaOS2, 143B, MG63 and U2OS was upregulated in comparison with normal human fetal osteoplastic cells (hFOBs) (Fig. 1B and C). The MG63 and U2OS cells indicated higher GPNMB mRNA and protein expression than the SaOS2 and 143B cells. Therefore, these two cell lines were chosen for further study.

To investigate the functions of GPNMB in relation to osteosarcoma, we separately silenced its expression in the MG63 and U2OS cells via GPNMB siRNA transfection. A high inhibitory GPNMB mRNA and protein level was found in the MG63 and U2OS cells transfected with GPNMB siRNA (P<0.05; Fig. 1D and E). Of the two GPNMB siRNAs (siGPNMB-1 and siGPNMB-2), siGPNMB-2 was the most efficient and thus was chosen for further experiments.

Knockdown of GPNMB in the MG63 and U2OS cells resulted in reduced cell proliferation (Fig. 2A and B). The Transwell assay indicated that GPNMB silencing notably inhibited the migration (Fig. 2C and D) and invasion (Fig. 2E and F) of the MG63 and U2OS cells. These results indicated that GPNMB may promote the progression of osteosarcoma.

The abnormal activation of the PI3K/Akt/mTOR signaling pathway plays a critical role in osteosarcoma pathogenesis (24,25). Therefore, we further evaluated whether GPNMB affected the PI3K/Akt/mTOR signaling pathway in osteosarcoma cells. The protein levels of p-PI3K, PI3K, p-AKT, AKT, p-mTOR and mTOR were obviously upregulated in the MG63 and U2OS cells (Fig. 3A), whereas the protein levels were significantly downregulated by GPNMB siRNA in both MG63 (Fig. 3B) and U2OS cells (Fig. 3C). These results indicated that the GPNMB silencing inhibited the activation of the PI3K/Akt/mTOR signaling pathway in osteosarcoma cells.

To identify whether the effect of GPNMB silencing on osteosarcoma cells was achieved by regulating the PI3K/Akt/mTOR signaling pathway, the MG63 and U2OS cells were cultured with IGF-1 (3 ng/ml) (26), an agonist of PI3K. IGF-1 activated the PI3K/Akt/mTOR signaling and abolished the inhibition of PI3K/Akt/mTOR signaling induced by siGPNMB-2, as evidenced by the upregulation of the protein expression of p-PI3K, PI3K, p-AKT, AKT, p-mTOR and mTOR (Fig. 4A and B).

Furthermore, the inhibitory effect of GPNMB silencing on the proliferation (Fig. 4C and F), migration (Fig. 4D and G) and invasion (Fig. 4E and H) of the MG63 and U2OS cells was significantly reversed by IGF-1. These results demonstrated that GPNMB silencing inhibited the progression of osteosarcoma cells by blocking the PI3K/Akt/mTOR signaling pathway.