β-elemene (98% purity) was obtained from Jingang Pharmaceutical Co. (Dalian, China). TMZ and 4′,6-diamidino-2-phenylindole (DAPI) were obtained from Sigma-Aldrich Co., LLC. (St. Louis, MO, USA). The antibodies against CD133, ABCG2, GFAP, neuron specific enolase (NSE), and myelin basic protein (MBP), used in western blots and immunofluorescence assays, were purchased from Boster Co., Ltd. (Wuhan, China). Epidermal growth factor (EGF), basic fibroblast growth factor (bFGF), leukaemia inhibitory factor (LIF), N2, Accutase Cell Dissociation Reagent and Alexa Fluor 594 goat anti-rabbit IgG were from Invitrogen Corp. (Carlsbad, CA, USA). The antibody against GAPDH was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Anti-CD133 and IgG1 isotype control antibodies used in flow cytometry analysis were purchased from Miltenyi Biotec (Bergish Gladbach, Germany). The Cell Counting Kit-8 (CCK-8) was obtained from Dojindo Molecular Technologies, Inc. (Japan). Nude mice were provided by the Experimental Animal Center of The Academy of Military Medical Sciences. Primary glioblastoma cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (Hyclone, UT, USA) supplemented with 10% foetal calf serum, 50 IU/ml penicillin and 50 mg/ml streptomycin and grown at 37°C in a humidified atmosphere with 5% CO₂.

Tumour tissues originated from 20 patients with glioblastoma (10 men and 10 women, 42±11.7 years, World Health Organization/WHO grade III–IV) undergoing surgical resection in the Department of Neurosurgery of the General Hospital of Shenyang Military. The cells of two of these patients (case 1, WHO grade III; and case 2, WHO grade IV) were cultured as primary glioblastoma cells. The study was approved by the Ethics Committee of the General Hospital of Shenyang Military and abided by the Declaration of Helsinki, with informed consent obtained from all study participants. Tumour samples were stored in sterile serum-free DMEM/F12 and processed within 0.5 h after resection. Tissues were cut into 1-mm² pieces, washed with PBS and digested with 0.25% trypsin at 37°C for 15 min. Cells were cultured in serum-containing DMEM/F12 after passage through a 70-μm strainer (BD Pharmingen, San Diego, CA, USA). Primary glioblastoma cells were resuspended in NSC medium (NSCM) containing DMEM/F12 medium, 20 ng/ml EGF, 20 ng/ml bFGF, 10 ng/ml LIF and N2 (1:100) for culturing GSCs.

Paraffin sections of glioblastoma tissues were obtained and deparaffinized. Antigen retrieval was performed with citrate buffer, pH 6.0 (Invitrogen). Non-specific sites were blocked by incubating sections with 5% BSA in a humidified chamber for 1 h at room temperature. The samples were incubated with 0.3% H₂O₂ for 15 min to block endogenous peroxidase activity and then labelled with different primary antibodies at proper concentrations (1:100 for CD133, ABCG2 and GFAP) for 1 h at room temperature. In subsequent steps, we used the Vectastain ABC kit (Vector Laboratories, Burlingame, CA, USA) and DAB as a chromogen (Changdao Biotech, China). The nuclear counterstaining of sections was performed with hematoxylin. In every tissue, additional staining without primary antibody was performed in parallel as a negative control. The integrated optical density (IOD) was used to semi-quantitatively estimate the expression of CD133, ABCG2 and GFAP in Image-Pro Plus 6.0.

Cells were digested into single cells with Accutase Cell Dissociation Reagent and labelled with anti-CD133 and control IgG isotype monoclonal antibodies (Miltenyi Biotec, San Diego, CA, USA). Flow cytometry analysis was conducted on a FACSCalibur flow cytometer (Becton-Dickinson, San Jose, CA, USA). Data acquisition and analysis were performed with CellQuest software (Becton-Dickinson).

Primary glioblastoma cells were cultured in NSCM and formed cell spheres after 3 days. After forming spheres of at least 100 cells (culturing for approximately 5 days), cell spheres were digested to single cell suspension with Accutase Cell Dissociation Reagent and subcultured with NSCM in ultra-low attachment culture flasks (Corning, Inc., NY, USA).

In each well of 24-well plates, 2×10⁵ dissociated cells or 20 cell spheres were cultured on poly-L-lysine-coated cover slips for various times. Cells on cover slips were washed with PBS, fixed with methanol and then treated with 0.1% Triton X-100. After blocking with 5% BSA, cells were incubated with proper dilutions of anti-CD133, GFAP, MBP or NSE antibodies at 4°C overnight. The primary antibodies were detected with Alexa Fluor 594 goat anti-rabbit IgG (dilution 1:500) for 40 min at 37°C in the dark. Cell nuclei were stained with DAPI, and immunofluorescent images were examined under a fluorescent microscope (Leica Microsystems, Germany).

Cell viability was evaluated with the CCK-8 assay after cells in exponential growth were cultured in a 96-well plate. Trypan blue staining confirmed >80% viability, and cells were treated according to the study design. Then, 10 μl of CCK-8 was added to each well, and the mixture was incubated for 4 h at 37°C. The optical density (OD) of each well was measured at 450 nm using a spectrophotometric microplate reader (BioTek Instruments, Inc., Winooski, VT, USA). Five replicate wells were used for each cell sample.

The cells were lysed with RIPA buffer [50 mM Tris-HCl (pH 7.4), 1.0% NP-40, 0.25% Na-deoxycholate, 1 mM EDTA, 150 mM NaCl, 1 mM aprotinin, 1 mg/ml PMSF, 1 μg/ml pepstatin and 1 μg/ml leupeptin]. The concentrations of total protein in the cellular extracts were measured using the BCA assay kit from Keygen Biotech, Co., Ltd. (Nanjing, China). After separation in 10–12% sodium dodecyl sulphate polyacrylamide (SDS-PAGE) gels, the proteins were transferred to nitrocellulose filter membranes (Bio-Rad, Hercules, CA, USA). The membranes were blocked with 5% BSA in Tris-buffered saline with Tween-20 at 4°C overnight and probed with various primary antibodies at 4°C overnight, followed by incubation with horseradish peroxidase-conjugated secondary antibodies at 37°C for 3 h. The membranes were exposed to an ECL system (Amersham, Sweden), and fluorescence was detected by exposing the membrane to X-ray film. The results were scanned with the Image Quant 5.2 software (Amersham).

GSC spheres were digested into single cells, and 1000 cells/animal (suspending in 0.1 ml DMEM/F12) were subcutaneously injected into the right shoulder region of 4-week-old female nude mice. The use of animals was approved by the Institutional Animal Care and Use Committee. Beginning with the 7th day after transplantation, various drugs were intraperitoneally injected. The volumes of tumours were measured every 3 days and calculated according to the formula: V = 1/2 × largest diameter × smallest diameter² (21). Tumours were weighed on the 21st day after transplantation.

The values are reported as the means ± standard deviation (SD) of at least three independent experiments. Statistical analysis was performed using Student’s t-test. Statistical significance was accepted at the level of P<0.05 between different groups, and P<0.01 was considered highly significant. Statistical analyses were performed with SPSS software (SPSS, Inc., Chicago, IL, USA).

To investigate the distribution of CD133⁺ cells in human glioblastoma tissues, immunohistochemistry was performed with an anti-CD133 antibody. To examine the proportion of CD133⁺ primary glioblastoma cells, we cultured these cells from fresh glioblastoma tissues and performed flow cytometry using an anti-CD133 antibody. We found that CD133⁺ cells were not only assembled in some vascular walls but were also sparsely localized in other zones of glioblastoma tissues (Fig. 1A–D). Primary glioblastoma cells obtained from two glioblastoma samples (case 1 and case 2) had obvious cellular atypia and pathological nuclear fission (Fig. 1E and F). The proportions of CD133⁺ cells were 1.3% in case 1 (Fig. 1G) and 3.5% in case 2 (Fig. 1H).

To obtain GSC spheres, we cultured primary glioblastoma cells in NSCM and observed cell sphere formation with an inverted microscope. The results showed that primary glioblastoma cells derived from case 1 and case 2 could form GSC spheres in NSCM. The second-generation GSC spheres formed on the 3rd day after passage and then grew gradually with a denser structure and higher refractivity (Fig. 2).

To test the expression of CD133 and the multipotent differentiation capacity in the second-generation GSC spheres, we cultured GSC spheres in DMEM/F12 with 10% foetal calf serum and examined CD133, GFAP (astrocyte marker), MBP (oligodendrocyte marker) and NSE (neuron marker) expression by immunofluorescence assays 0, 48 and 96 h after plating. We found that the second-generation GSC spheres strongly expressed CD133 with little expression of GFAP, MBP and NSE (Fig. 3, 0 h). During 48 h of adherent growth conditions in serum-containing DMEM/F12, GSC spheres gradually spread out. The expression of CD133 decreased, whereas the expression of GFAP, MBP and NSE increased with time. After 96 h, GFAP, MBP and NSE showed strong expression (Fig. 3). These results suggest that the GSC spheres we obtained possessed stem cell phenotypes and multipotent differentiation potential.

To evaluate the effect of β-elemene on GSC sphere formation, cell spheres were digested into single cells. The cells were then plated at a density of 100 cells/well in 96-well plates and simultaneously treated with 100 μg/ml β-elemene for 72 h (untreated cells as control group). Finally, the number of cell spheres (containing at least 10 cells/sphere) was counted. To examine the effect of β-elemene on proliferation, 1000 GSCs and 10000 primary glioblastoma cells per well were independently plated in 96-well plates. These cells were then treated with 0 (control group), 50, 100, 150 or 200 μg/ml β-elemene for 24 h, and the CCK-8 assay was used to examine cell viability. To test the ability of β-elemene to dissociate GSC spheres, we added 200 μg/ml β-elemene to cell spheres for 12 h and observed sphere dispersion with an inverted microscope. The results showed that the number of GSC spheres that formed in the β-elemene group was less than in the control groups in both cases (Fig. 4A, P<0.01). The proliferation of both GSCs and primary glioblastoma cells (GC) (Fig. 4B) decreased dose-dependently in the presence of β-elemene. The degree of this viability loss was less for GSCs than for primary glioblastoma cells (Fig. 4B), which suggested that GSCs possessed stronger resistance to β-elemene than glioblastoma cells. β-elemene can disperse GSC spheres, and this can result in the fragmentation and death of some cells (Fig. 4C–F).

To investigate the effect of β-elemene on the expression of GSC markers CD133 and ABCG2 and differentiation marker GFAP, GSC spheres were treated with 100 and 200 μg/ml β-elemene for 24 h, and western blot analysis was performed using specific antibodies. The results were semi-quantitatively estimated using Gel-Pro analyzer 4.0 software and illustrated graphically. We found that the expression levels of CD133 and ABCG2 were significantly downregulated, while the expression of GFAP was increased by β-elemene in a dose-dependent manner (Fig. 5). These results suggested that β-elemene impairs the stemness of GSC spheres and promotes their differentiation.

To evaluate the effect of β-elemene on GSC tumours in vivo, we subcutaneously injected GSCs into the flank of nude mice and then intraperitoneally injected NaCl or 50 mg/kg β-elemene for 1 week. The tumour volumes were measured every 3 days. Points are the means ± SD of tumour volume in each group (n=6). The results showed that a significant inhibition of tumour volume was observed in the β-elemene-treated group compared with the control group from the 9th day after transplantation (the 3rd day after intraperitoneal injection) to the 21st day, as shown in Fig. 6A and B. On the 21st day, tumours were dissected and measured. The tumour weights of the β-elemene group were significantly lower than those of the control group, as shown in Fig. 6C and D. These results suggested that the development of tumours was suppressed by β-elemene in GSC-bearing nude mice.

To evaluate the effect of β-elemene on the expression of stem cell markers and differentiation markers in vivo, immunohistochemistry was performed using antibodies against CD133, ABCG2 and GFAP on the aforementioned tumour tissues of GSC-transplanted nude mice in β-elemene and NaCl groups (Fig. 7A). Image-Pro Plus 6.0 software was applied to calculate the IOD of each group (the same selected areas and magnifications in all images), and then statistical analysis was performed. The IOD of each group is illustrated in a vertical bar chart (Fig. 7B and C). We found in this assay that the expression levels of CD133 and ABCG2 in the β-elemene group were significantly lower than in the NaCl group, and the expression level of GFAP in the β-elemene group was higher than in the NaCl group in both cases 1 and 2. These results suggest that β-elemene decreased the expression of CD133 and ABCG2 but increased the expression of GFAP in vivo.

As reported, GSCs were usually more resistant to TMZ cytotoxicity than normal glioblastoma cells (21). To determine whether treatment with β-elemene sensitizes GSCs to TMZ, GSC spheres were digested into single cells and then plated at a density of 5×10³ cells/well in 96-well plates. Cells were organized into the following three groups: control (untreated), TMZ (treated with 6 μg/ml TMZ for 24 h) and combination (treated with 6 μg/ml TMZ and 100 μg/ml β-elemene for 24 h). After treatment with the drugs, CCK-8 assays were performed to examine cell viability. Additionally, 18 nude mice were injected with GSCs to create tumours and divided into the following three groups (6 mice/group): NaCl (negative control, intraperitoneally injected with 50 mg/kg NaCl for 1 week), TMZ (intraperitoneally injected with 33 mg/kg TMZ for 1 week) and combination (intraperitoneally injected with 33 mg/kg TMZ and 50 mg/kg β-elemene for 1 week). The volumes of tumours were measured and calculated according to the formula: V = 1/2 × largest diameter × smallest diameter², every 3 days, and tumours were weighed on the 21st day. The results showed that cell viability was lower in the combination group than either the control or the TMZ groups (Fig. 8A and B). In the in vivo experiments, we found that both the volumes (Fig. 8C and D) and the weights (Fig. 8E and F) of tumours were less in the combination group than in either the control or TMZ groups. These findings suggest that β-elemene increased the sensitivity of GSCs to TMZ-induced cytotoxicity.