Human osteosarcoma cell lines HOS, Saos-2 and MG-63 were purchased from American Type Culture Collection (ATCC, Rockville, MD, USA). For separation of cancer stem cells from OS cell lines, HOS, Saos-2 and MG-63 cells were incubated with fluorescein isothiocyanate (FITC) conjugated CD133 antibody (Miltenyi Biotec, Bergisch Gladbach, Germany) for 20 min at room temperature. Then, cells were washed with cold PBS twice followed by sorting the CD133-positive OS cells and CD133-negative OS cells on a FACS vantage (FACSCalibur, BD Biosciences, Franklin Lakes, NJ, USA). CD133-positive OS cells were considered as the OS stem cells, whereas the CD133-negative OS cells were identified as the OS non-stem cells (9). All of these cells were cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS, Gibco-Invitrogen, Carlsbad, CA, USA).

Total RNAs were extracted from HOS, Saos-2 and MG-63 OS cells using Trizol reagent (Invitrogen). cDNA of these cells was synthesized using M-MLV Reverse Transcriptase (Invitrogen) by taking total RNAs as templates. For detecting the expression of PKM2, Real-time PCR was performed using SYBR Premix Ex Taq (Takara, Shiga, Japan) on ABI PRISM 7900 sequence detection system (Applied Biosystems, Carlsbad, CA, USA). GAPDH gene was used as the internal reference to determine the relative expression of PKM2.

To investigate the role of PKM2 in OS stem cells, we overexpressed or knocked down the PKM2 by transfecting with PKM2 plasmid or small interfering RNA (siRNA), respectively. For conducting PKM2 eukaryotic expression plasmid, open reading frame of PKM2 gene was amplified by PCR followed by linking the PCR products to pcDNA3.1 plasmid (Invitrogen). For knockdown of PKM2, siRNA targets to PKM2 was purchased from Genechem Co., Ltd. (Shanghai, China). For transfection, 2 μg/ml PKM2 plasmid or 50 pmol/ml PKM2 siRNA were transfected into the OS stem cells by using Lipofectamine 2000 (Invitrogen).

OS cells (1×10⁴) were seeded in 96-well plates and cultured at 37°C. Cells were transfected with PKM2 vector or PKM2 siRNA. At 24 h after transfection, the cells were treated with 2 mM metformin and 8 μM cisplatin for 48 h, cell viability was measured by Cell Counting Kit-8 (CCK-8, Dojindo, Kumamoto, Japan) assays. The absorbance was measured at 450 nm by using an ELISA microplate reader (Sunrise microplate reader; Tecan, Mannedorf, Switzerland). Half maximal inhibitory concentration (IC₅₀) was calculated according to the results of CCK-8 assays.

Total proteins were extracted from OS stem cells and OS non-CSCs using radio immunoprecipitation assay (RIPA) lysis buffer (Cell Signaling Technology, Inc., Danvers, MA, USA). The proteins in cell lysates were separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins on the gels were then transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA, USA). The membranes were blocked by 5% non-fat dry milk for 2 h followed by incubating with primary antibodies against PKM2, pro-caspase-9, pro-caspase-3, cleaved caspase-9, cleaved caspase-3 and β-actin (Cell Signaling Technology, Inc.) overnight. Then, the proteins on PVDF were incubated with appropriate HRP-conjugated secondary antibodies followed by detecting the protein signals by using an enhanced chemiluminescent substrate (Thermo Fisher Scientific, Inc., Waltham, MA, USA).

HOS CSCs were transfected with PKM2 vector followed by treating with 2 mM metformin and 8 μM cisplatin. The cells were stained using the Annexin V-FITC apoptosis detection kit (Sigma-Aldrich, St. Louis, MO, USA) and subsequently analyzed using flow cytometry (Becton-Dickinson, San Jose, CA, USA).

HOS CSCs were transfected with PKM2 vector followed by treating with 2 mM metformin and 8 μM cisplatin. Proliferation of HOS CSCs was measured by using ³H-thymidine incorporation assay.

HOS CSCs were transfected with PKM2 vector followed by treating with 2 mM metformin and 8 μM cisplatin. Relative glucose uptake was measured using an Amplex Red Glucose/Glucose Oxidase assay kit (Molecular Probes, Camarillo, CA, USA) according to the manufacturer's instructions.

HOS CSCs were transfected with PKM2 vector followed by treating with 2 mM metformin and 8 μM cisplatin. Production of lactate in HOS CSCs was detected by using a Lactate assay kit (BioVision, Milpitas, CA, USA) according to the manufacturer's protocol.

HOS CSCs were transfected with PKM2 vector followed by treating with 2 mM metformin and 8 μM cisplatin. After treatment, cells were harvested and washed with PBS twice. NP40 solution (1%) was used to split the cells. ATP in the cell lysates were measured by using ATP Colorimetric/Fluorometric Assay kit (BioVision).

Four-week-old female immunodefcient nude BALB/c mice were purchased from Shanghai Super-B&K Laboratory Animal Corp., Ltd. (Shanghai, China). For xenografts, mice were injected subcutaneously into the right armpit with 5×10⁶ HOS CSCs. Cisplatin (5 mg/kg) and metformin (200 mg/kg) were administered by intraperitoneal injection twice a week. Tumor size was measured every five days. Tumor volume was calculated according to the following formula: volume (V) = 1/2 × length × width². For detection of PKM2 expression in tumor tissues, collagenase type III (Worthington Biochemical Corp., Lakewood, NJ, USA) was used to digest the tissues. The animal care and experimental protocols were approved by the Animal Care Committee of The Affiliated Zhongshan Hospital of Dalian University and complied with the recommendations of the Chinese guidelines for the care and use of laboratory animals. The study was approved by ethics committee of The Affiliated Zhongshan Hospital of Dalian University.

Data are presented as mean ± SD and analyzed by using SPSS 14.0. Two-tail Student's t-test and ANOVA were performed to determine the differences. A value of P<0.05 was considered to indicate a statistically significant difference. All the experiments were independently repeated at least 3 times.

To investigate the cisplatin sensitivity of OS stem cells and OS non-stem cells, we first separated them from the HOS, Saos-2 and MG-63 OS cell lines. As shown in Fig. 1A, the separation efficiency was evaluated by flow cytometry. CD133-positive OS cells were considered as the OS stem cells. Whereas, the CD133-negative OS cells were identified as the OS non-stem cells. Results of CCK-8 assays showed that the sensitivity of OS stem cells (HOS CSCs, Saos-2 CSCs and MG-63 CSCs) to cisplatin was significantly lower than the OS non-stem cells (HOS non-CSCs, Saos-2 non-CSCs and MG-63 non-CSCs) (Fig. 1B). IC₅₀ of cisplatin to OS CSCs (28.4 μM to HOS CSCs, 32.9 μM to Saos-2 CSCs and 26.3 μM to MG-63 CSCs) was significantly higher than their corresponding OS non-CSCs (6.1 μM to HOS non-CSCs, 8.8 μM to Saos-2 non-CSCs and 7.9 μM to MG-63 non-CSCs) (Fig. 1C). Taken together, we demonstrated that osteosarcoma stem cells showed resistance to cisplatin treatment.

PKM2 plays an important role in tumor growth, development and chemoresistance (19,20). In our study, we found that the expression of PKM2 was significantly overexpressed in OS stem cells compared with that in their corresponding OS non-stem cells at mRNA (Fig. 2A) and protein levels (Fig. 2B). To investigate the role of the overexpressed PKM2, we transfected the HOS, Saos-2 and MG-63 CSCs with PKM2 siRNA. Effects of PKM2 siRNA on OS stem cells are shown in Fig. 2C. Noteworthy, we observed that knockdown of PKM2 by the specific siRNA restored the sensitivity of OS stem cells to cisplatin treatment (Fig. 2D). These results emphasized the key role of PKM2 in cisplatin-resistance in OS stem cells.

To investigate the potential role of metformin in the sensitivity of OS stem cells to cisplatin, we treated the HOS CSCs and non-CSCs with 8 μM cisplatin and 2 mM metformin for evaluating their synergistic effects. We found that metformin markedly enhanced the cytotoxicity of cisplatin to HOS CSCs, whereas the effect of metformin was slight on enhancing the cytotoxicity of cisplatin on HOS non-CSCs (Fig. 3A). Combination with metformin decreased the IC₅₀ of cisplatin to HOS CSCs by 74.9%. In contrast, the reduction of cisplatin IC₅₀ to HOS non-CSCs was 31.2% (Fig. 3B). These data demonstrated that metformin was able to enhance the antitumor effect of cisplatin, and HOS CSCs were more sensitive to metformin-promoted cell death than their corresponding HOS non-CSCs. Mechanically, results of western blot analysis showed that metformin treatment significantly downregulated the expression of PKM2 in HOS CSCs (Fig. 3C). We therefore inferred that synergistic effect of metformin on cisplatin-induced cell death was dependent on the inhibition of PKM2. To prove this inference, we transfected the HOS CSCs with PKM2 expression vector. We found that enforced expression of PKM2 'rescued' the HOS CSCs from the combination treatment with cisplatin and metformin (Fig. 3D). These results indicated that metformin had the ability to resensitize OS stem cells to cisplatin by targeting PKM2.

To investigate whether metformin enhanced the antitumor effect of cisplatin on OS CSCs in vivo, we established the in vivo model of OS by using HOS CSCs. We observed that single administration with cisplatin slightly reduced the tumor sizes in mice. Of note, although metformin single treatment did not reduce the tumor volume obviously, it strongly enhanced the antitumor effect of cisplatin on HOS CSC model (Fig. 4A and B). After analyzing the expression of PKM2 in tumor tissues formatted by HOS CSCs, we found that metformin significant decreased the expression of PKM2 in HOS CSC in vivo model (Fig. 4C). These results suggested that metformin have to the ability to sensitize the OS CSCs to cisplatin by inhibiting the expression of PKM2 in vivo.

As PKM2 is a key regulator of glucose metabolism in tumor cells (21), we next studied the effect of metformin on glucose metabolism in HOS CSCs. As shown in Fig. 5A and B, cisplatin treatment did not obviously influence the glucose metabolism in HOS CSCs. However, metformin treatment significantly decreased the consumption of glucose as well as increased the production of lactate in HOS CSCs. In addition, transfection with PKM2 vector restored the glucose uptake and lactate production in HOS CSCs treated with metformin. These results demonstrated that metformin had the ability to weaken the glucose metabolism by inhibiting the expression of PKM2. Furthermore, as the results of glucose metabolism reduction, metformin decreased the production of ATP in HOS CSCs (Fig. 5C).

In tumor cells, glucose metabolism and intracellular ATP regulate cell proliferation and chemoresistance (22,23). In our study, we found that combination with metformin and cisplatin induced drastic proliferation inhibition in HOS CSCs. However, enforced expression of PKM2 restored the proliferation of HOS CSCs which were co-treated with metformin and cisplatin (Fig. 6A). Furthermore, we observed that metformin treatment reduced the resistance of HOS CSCs to cisplatin-induced apoptosis. On the other hand, transfection with PKM2 vector impaired the effect of cisplatin plus metformin (Fig. 6B and C). Taken together, we demonstrated that metformin promoted cisplatin-induced proliferation inhibition and apoptosis in HOS CSCs by downregulating the expression of PKM2.

Multidrug resistance is a major obstacle for chemotherapy of OS. We therefore treated OS stem cells (HOS CSCs, Saos-2 CSCs and MG-63 CSCs) with three kinds of chemotherapeutic drugs (cisplatin, doxorubicin and 5-fluorouracil). As shown in Fig. 7A, co-treatment with metformin significantly increased the sensitivity of HOS CSCs, Saos-2 CSCs and MG-63 CSCs to cisplatin, doxorubicin and 5-fluorouracil. Metformin treatment obviously decreased the IC₅₀ of cisplatin, doxorubicin and 5-fluorouracil to HOS, Saos-2 and MG-63 OS stem cells (Fig. 7B). We demonstrated that metformin can increase the sensitivity of OS stem cells to chemotherapy.