Gene expression analysis
Gene expression profiling was performed on RNA samples from parental and resistant cell lines using microarray (U87MG) or RNA-seq (A172). The list of 1174 mitochondria-related genes was constructed according to MitoCarta2.0 [17] and Mitochondria RT2 Profiler PCR Array (Qiagen, Denmark). The mitochondria-related genes that were significantly different between parental and TMZ-resistant cells were identified by at least a 1.5-fold difference and a P-value< 0.05. The heat map was generated using Multi-Experiment Viewer (http://mev.tm4.org/) according to the log2 (fold change) values of selected genes.
Analysis of the Cancer genome atlas (TCGA) clinical datasets
For gene expression analysis, the GBM clinical transcriptome next-generation sequencing (NGS) data were obtained from TCGA database (https://portal.gdc.cancer.gov/). This included a total of 169 specimens, which consisted of 156 primary and 13 recurrent tumors. The data of fragments per kilobase of transcript per million mapped reads upper quantile (FPKM-UQ) were downloaded for further processing. The expression fold change and significance level (t-test) of mitochondria-related gene expression between primary and recurrent GBM NGS data were calculated. For survival, a publicly available cancer microarray database, SurvExpress, was used for analysis of the TCGA datasets [18].
Culture of the GBM cell lines and derivation of the TMZ-resistant cells
The human GBM cell lines U87MG and A172 were purchased from American Type Culture Collection (Manassas, VA, USA). DMEM (Thermo Fisher Scientific, Waltham, MA, USA) with supplementation of 10% serum and antibiotics was used as medium. The resistant cells were derived from long-term co-incubation with 100 μM TMZ (Sigma-Aldrich, St. Louis, MO, USA) which significantly reduced cellular proliferation and survival in the beginning, but recovered eventually. Random single-cell clones were then cultured, with majority having SOD2 expression [14]. Analysis of the clones showed an association between SOD2 and the TIC biomarkers (Additional file 1: Figure S1A). We selected r#10 of U87MG (U87MG-r#10) and r#6 of A172 (A172-r#6) for the following studies. Co-incubation with TMZ was kept for regular maintenance of the resistant cells.
Construction of patient-derived xenograft models for primary tumor study
Samples were obtained directly from the tumor tissue, which were surgically resected in a naïve GBM patient (GBM#4) and a recurrent GBM patient with prior multiple treatment (GBM#1). They were then minced and implanted into subcutaneous area of immunocompromised NOD-SCID mice (BioLASCO, Taipei, Taiwan) in less than 6 h. After tumor formation, the mice were sacrificed for extraction of the tumor, annotated as P0. The tumor was minced and successively implanted into another mouse for the first passage, annotated as P1. The passage continued for maintenance of the primary cells. For cell study or for cryopreservation, tumors of only three or fewer passages were applied to keep the tumor features [19]. The tumor samples were minced and incubated in a papain dissociation kit (#130–095-942, Miltenyi Biotec, Bergisch Gladbach, Germany) for treatment with gentleMACS™ dissociators. The GBM cells were then isolated by negative magnetic selection using the Mouse Ablation Kit (#130–104-694, Miltenyi Biotec). The processed cells can be used for sphere cell cultures (in serum-free medium: DMEM/F12 (Thermo Fisher Scientific), 1% penicillin/streptomycin, 2% B27 (Thermo Fisher Scientific), 10 ng/ml basic fibroblast growth factor (Cell Guidance Systems, Cambridge, United Kingdom), and 10 ng/ml epidermal growth factor (ProSpec, East Brunswick, NJ, USA)), cell sorting and animal experiments. We also cultured the cells in low-serum (1–2%) containing medium for short-term duration to observe cell morphology and growth [20, 21].
Tumor spheroid formation assays
Spheroid cells were cultured with serum-free medium containing 0.3% methylcellulose (Sigma-Aldrich) in ultra-low adherent plates [22]. For the tumor spheroid formation assay, low numbers of cells (e.g., 1, 5, 10, 20, and 50 cells) were applied. After 2 weeks, the formation of spheres and their diameters was assessed. The frequency of initiation capacity was then calculated using Extreme Limiting Dilution Analysis (ELDA, http://bioinf.wehi.edu.au/software/elda/).
Western blot analysis
The cell lysate was separated via SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Bio-Rad, Hercules, CA, USA). The membranes were first blocked with 5% nonfat milk and were then incubated overnight with primary antibodies against SOD2 (1:3000, Cell Signaling, Danvers, MA, USA), CD133 (1:1000, Proteintech, Rosemont, IL, USA), Bmi-1 (1:1000, GeneTex, Irvine, CA, USA), SOX2 (1:1000, GeneTex), Oct4 (1:5000, GeneTex), caspase 3 (1:1000, Cell Signaling), Oct3/4 (1:1000, Santa Cruz, Dallas, TX, USA), vimentin (1:1000, GeneTex), MGMT (1:1000, BD, Franklin Lakes, NJ, USA), and beta-actin (1:5000, Millipore, Burlington, MA, USA). After the membranes were washed, they were incubated with secondary antibodies. Finally, after eliciting signals with chemiluminescence substrate, Amersham Hyperfilm ECL (GE Healthcare, Chicago, IL, USA) was used to detect the expression intensity. The density was quantified by ImageQuant (GE Healthcare).
Immunohistochemistry (IHC)
All paraformaldehyde-fixed, paraffin-embedded tissue sections were prepared from xenograft mouse or human archival tissue (Pathology Department of National Cheng Kung University Hospital). The method of staining was previously described [14]. The primary antibodies were described in the western blot analysis but at a dilution of 1:200. The staining were automatically identified and assessed by ImageJ (http://rsbweb.nih.gov/ij/).
Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was isolated by TRIzol (Invitrogen, Carlsbad, CA, USA) following a standard procedure and was subjected to qRT-PCR with SuperScript II reagent (Invitrogen). The product was mixed with SYBR® Green Master Mix (Applied Biosystems, Foster City, CA, USA); the specific primers used were as follows: (SOD2, F:5′-GGCCTACGTGAACAACCTGAA, R:5′-CTGTAACATCTCCCTTGGCCA; CD133, F:5′-TCCACAGAAATTTACCTACATTGG, R:5′-CAGCAGAGAGCAGATG.
ACCA; Bmi-1, F:5′-TGGAGAAGGAATGGTCCACTTC, R:5′-GTGAGGAAACTGT.
GGATGAGGA; SOX2, F:5′-AAATGGGAGGGGTGCAAAAGAGGAG,R:5′-CAGCT.
GTCATTTGCTGTGGGTGATG; GAPDH, F:5′-GAAGGTGAAGGTCGGAGTC, R:5′-GAAGATGGTGATGGGATTC). The expression was detected using an ABI 7000 Sequence Detection System (Applied Biosystems) and was normalized to GAPDH using the 2-ΔΔCT formula.
Clonogenic assay and cell density assay
For clonogenic assay, 400 cells/well were cultured in 6-well plate with treatment given on the next day. The cells were washed after three days and incubated in treatment-free medium to allow them to grow into colonies. They were then stained and fixed with 50% ethanol containing 0.5% methylene blue for 90 min, and the numbers of the colonies were counted. For cell density assay, 5000~20,000 cells/well were cultured in 6-well plate with treatment given on the next day. They were allowed for proliferation for three days. The cells were then stained and fixed with the aforementioned solution, and was dissolved in 1% N-lauroyl-sarcosine followed by measurement of the optical density at 570 nm.
Detection of surface CD133 in cells and mitochondrial ROS expression by flow cytometry
Cells were dissociated and labeled with APC-anti-CD133 (Miltenyi Biotec) to detect the stemness feature or MitoSOX® (Invitrogen) to detect mitochondria-specific ROS expression. The staining procedure followed the manufacturers’ protocols with minimal adjustments. Fluorescence-activated cell sorting (FACS) was applied using a FACSCalibur system (BD) and CellQuest software (BD) for data collection and analysis, including determination of mean fluorescent intensity (MFI). For sorting, a FACSAria™ III (BD) was used to isolate the CD133 specific cells. The sorted CD133+ and CD133− cells were collected and cultured in the serum-free medium and the serum-containing medium, respectively.
Measurement of SOD2 activity
The activity of SOD2 was detected using an Amplex Red Hydrogen Peroxidase Assay (Invitrogen) and a Superoxide Dismutase Assay Kit (Cayman Chemical, Ann Arbor, MI, USA) according to the manufacturers’ instructions. Briefly, the cells were cultured in plates with or without treatment and were then transferred to a microplate. The indicated volume of working solution was pipetted into each well to initiate the reaction. To detect SOD2 enzyme activity, potassium cyanide was added simultaneously to block SOD1 and SOD3 reactions [23]. A microplate reader with the indicated excitation/emission wavelengths was used for data collection and analysis.
RNA-based gene modulation of SOD2
Lipofectamine® RNAiMAX reagent (Invitrogen) and LTX with Plus™ reagent (Invitrogen) were used following manufacturer’s protocol for transient transfection of SOD2 siRNA (S13268, Ambion, Austin, TX, USA) and pBI-EGFP-MnSOD (#16612, Addgene, Cambridge, MA, USA), pBI-EGFP (kindly provided by Dr. Hsiao-Sheng Liu, National Cheng Kung University, Taiwan), respectively. For stable knockdown, the cells were infected with SOD2-lentiviral short hairpin RNA (shRNA) or empty vector (both from RNAi Core, Academia Sinica, Taiwan). The next day, the infected cells were then selected for stable clones in antibiotic-containing medium for weeks, which was followed by confirmation of the knock-down efficiency and selection (Additional file 1: Figure S1B).
Xenograft mouse model for tumor growth assessment or survival studies
Male NOD-SCID mice 5~6 weeks of age were used in this study. For the tumor growth assessment, cells (2 × 106) were inoculated into the subcutaneous area of the right flank. The tumor volume was measured twice a week according to the following National Cancer Institute formula: length×width2 × 3.14/6. When the tumors reached 200 mm3 in size, the animals were randomly assigned for treatment. For the survival studies, burr holes were generated in the right frontal brain area of the skull. Then, using a stereotactic instrument, the cells (2.5 × 105) were injected through an ultrafine needle at a location 1.5 mm anterior to the bregma, 2.5 mm lateral to the midline, and 3.5 mm ventral to the surface of the dura mater. Treatment was initiated after 5 days. Drug administration consisted of TMZ (5 mg/kg) via oral gavage and/or the SOD inhibitor sodium diethyldithiocarbamate trihydrate (DETC, 100 mg/kg, Sigma-Aldrich) via intraperitoneal injection. TMZ was administered 3–4 h after the DETC injection.
In vivo assessment of the TIC tumorigenic potential
A titrated number of tumor cells was injected subcutaneously into the NOD-SCID mice. The tumor volume was measured regularly, and the frequency of stem cell initiation was analyzed using ELDA. The tumor was then extracted and subsequently isolated for serial transplantation into another mouse. Gene expression during each passage was assessed by qPCR after tumor extraction.
Statistics
Data were statistically analyzed using Prism 7 (GraphPad, La Jolla, CA, USA). The differences in continuous variables were calculated by unpaired, two-tailed Student’s t-test. The survival data were plotted by Kaplan-Meier curve, and the difference was calculated using the Log-Rank test. Significance was set at P ≤ 0.05.