Cell lines and culture conditions
Human GICs (MZGC1, MZGC2, MZGC3, MZGC4, MZGC5, MZGC6, MZGC7 and MZGC8 cells) were isolated and established as cell lines from the surgical tissues of patients treated in our institution from March 2015 to April 2021. GICs were cultured in Dulbecco’s modified Eagle’s medium/nutrient mixture F12 Ham’s liquid supplemented with SUBSER-ESrP® (Cell Science & Technology Institute, Inc., Sendai, Japan), CELRENA® medium (Cell Science & Technology Institute, Inc.), in 2-hydroxyethyl methacrylate -coated T75 flasks in a humidified incubator at 37 °C under an atmosphere of 5% CO2 and 95% air. For the methionine depletion studies, cells were plated on laminin-coated 10-cm culture dishes and cultured with complete CELRENA® medium or methionine deficient CELRENA® medium (custom medium CST1 MFG No. cn2243a0) for 48–72 h. Cells were dispersed onto laminin-coated 96-well plates, replaced with treated media on day 1, and time-lapse photography was performed with Incucyte® (Sartorius, Osaka, Japan). Simvastatin and FDI-6 were purchased from Selleck (Tokyo, Japan).
Dye exclusion test
Dye exclusion testing was performed to analyze cell viability. Fixed amounts of cells were seeded and cultured in medium with (control) or without methionine for 3, 5 and 7 days. Harvested cells were suspended in phosphate buffered saline-containing trypan blue and then examined to determine the percentage of cells with blue clear cytoplasm (nonviable cells) versus total cells using a cell counter (LUNA, Logos Biosystems, Annandale, VA, USA). The relative rate against the control sample was calculated.
Cell cycle analysis and detection of surface marker of GICs
Cells with or without treatment of methionine depletion/statin addition were collected using Accutase® solution (Sigma, USA). The Cell Cycle Phase Determination Kit (Cayman Chemical, Ann Arbor, MI, USA) was used for cell cycle analysis. In brief, collected cells were rinsed twice with buffer, then fixed at − 20 °C overnight. Cells were washed twice with ice-cold phosphate buffered saline, and stained with propidium iodide/RNase staining buffer solution in the dark for 30 min at room temperature. Then, cells were analyzed with a flow cytometer (Guava® EasyCyte™ Mini, Luminex Japan, Tokyo, Japan). A histogram of the cell cycle distribution was generated from 5000 events per sample and data were analyzed using Guava® Cell Cycle software. To detect CD133, the putative marker of GICs, by flow cytometry (FCM), 1 × 106 cells derived from tumorspheres were analyzed using fluorescent (FITC) labelled antibodies (CD133/1 (AC133)-VioBright FITC, Miltenyi Biotec, Germany) and BD FACSCalibur (BD Biosciences, USA).
RNA extraction including microRNA (miRNA), cDNA synthesis, and quantitative PCR (qPCR) analysis
RNA from cultured tumor cells was extracted with the mirVana miRNA Isolation Kit (Ambion, Thermo Fisher Scientific K.K., Tokyo, Japan) or RNeasy Plus Mini Kit (QIAGEN, Germantown, MD, USA). For cDNA synthesis, RNA was reverse transcribed from random hexamers using the SuperScript™ VILO™ cDNA Synthesis Kit (Invitrogen, Life Technologies, Thermo Fisher Scientific, Waltham, MA, USA). Real-time qPCR was then performed in triplicate on the StepOne Plus or Quant3 (Applied Biosystems, Thermo Fisher Scientific) using SYBR™ Green Realtime PCR Master Mix (Applied Biosystems, Thermo Fisher Scientific) to determine the mRNA levels. PCR was performed using a 20 μl volume containing 2 µl cDNA, 300 µM of each primer, and 10 µl of 2 × PCR master mix under the following conditions: 95 °C for 10 min followed by 40 cycles of 95 °C for 15 s and annealing/extension at 60 °C for 1 min. The data were normalized to the amount of human 18S rRNA, and the values are represented as the mean ± SD of 2−ΔΔCt in a triplicate assay.
The primers used in this study were described in Supplementary Table S2.
Colony formation assay
Cells were seeded into laminin-coated 12-well plates in triplicate at a density of 500 cells/well in 2 ml of medium with or without treatment medium. After 120 h, formed colonies were trypsinized, reseeded and cultured for further 72 h. After 192 h, the cell colonies were stained for 15 min with a solution containing 0.5% crystal violet and 25% methanol, followed by three rinses with tap water to remove excess dye. Each well was observed with Olympus IX71 microscope (Olympus, Tokyo, Japan).
Gene and miRNA expression analysis
Gene and miRNA expression was analyzed with a GeneChip™ System with a Human Genome Clariom™ D Array and GeneChip™ miRNA 4.0 Array (Thermo Fisher Scientific) according to the manufacturer’s instructions. mRNA expression analysis used amplification and biotin labeling of fragmented cDNA with the biotin labeling system (GeneChip™ WT PLUS Reagent Kit, Thermo Fisher Scientific) in duplicate. miRNA expression analysis used biotin-labeled total RNA 1 µg including miRNA from tissue using the FlashTag™ Biotin HSR RNA Labeling Kit (Affymetrix®, Thermo Fisher Scientific). Labeled probes were hybridized to the Human Genome Clariom™ D Array (Thermo Fisher Scientific) or GeneChip™ miRNA 4.0 Array (Thermo Fisher Scientific) with the GeneChip™ Scanner 3000 7G (Affymetrix®, Thermo Fisher Scientific). Expression data of mRNA were extracted from image files using GeneSpring GX 14.9.1 (Agilent Technologies, Santa Clara, CA, USA). Normalization and expression value calculations were performed using the GeneSpring system. Signal values of miRNA were calculated using the Transcriptome Analysis Console software (Affymetrix®, Thermo Fisher Scientific). Statistical analysis was performed using two-tailed, unpaired t-tests. Any change ≥ twofold with P < 0.05 was considered statistically significant.
Enrichment analysis used Metascape (http://metascape.org/) and Gene Set Enrichment Analysis (GSEA) (http://www.broad.mit.edu/GSEA). Significance of biological processes was determined through P-values calculated on a hypergeometric distribution (log10). Metascape was used to conduct meta-analysis with generation of heat maps of gene ontology terms that hierarchically cluster together under experimental conditions. Statistical criteria for a differentially expressed gene included a change greater than 1.5 fold, and an overall false discovery rate smaller than 5% (change ≥ ± 1.5 fold and false discovery rate ≤ 0.05) with Benjamini-Hochberg correction (q) to account for multiple comparisons. GSEA using MSigDB (v7.4) was also used to determine the significance of a pre-defined gene set by comparing the correlation between expression and class distinction with other random situations. The significance threshold was set at nominal P value < 0.05 or false discovery rate q value < 0.25.
Measurement of cholesterol, S-adenosyl-methionine (SAM), and S-adenosylhomocysteine (SAH) in tissue using ELISA
Measurement of cholesterol, SAM, and SAH used glioblastoma stem cells cultured with CELRENA medium or methionine-deprived CELRENA medium for 72 h and 10 × 106 cells/cell pellets were homogenized by sonication in ice-cold phosphate buffered saline followed by centrifugation at 10,000 g for 15 min. Intracellular cholesterol, SAM, and SAH concentrations were measured using the Total Cholesterol Assay Kit (Colorimetric, Cell Biolabs, Inc., San Diego, CA, USA) and SAM and SAH ELISA Combo Kit (Cell Biolabs, Inc.) in accordance with the manufacturer's instructions under the indicated conditions. The standards were generated using the cholesterol standard and SAH/SAM-bovine serum albumin conjugate supplied in the kit. After the reaction was stopped, the OD450 and OD620 values were read using a SpectraMax iD3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA). SAH/SAM ratios were calculated as the ratio of SAM to average SAH using the intracellular concentration (µg/mL) to determine the relative methylation potential.
Reduced Representation Bisulfite Sequencing (RRBS) and data analysis
RRBS libraries with single MspI digestion were constructed for glioblastoma stem cells. Briefly, 1 ug of genomic DNA was digested with 20U of MspI enzymes in 18 ul reaction mixtures at 37 °C for 2–3 h. After purification, the digested products were blunt-ended, and then dA added, followed by methylated-adapter ligation using a NEXTflex Bisulfite-Seq Ligation Kit. To obtain DNA fractions of > 100 bp, the MspI-digested products were purified using Agencourt AMPure XP (Beckman Coulter). Bisulfite conversion was conducted using an EX DNA Methylation Gold Kit following the manufacturer's instructions. The final libraries were generated by PCR amplification using NEXTflex Bisulfite-Seq Ligation Kit and sequenced using an Illumina HiSeqX to generate 150 bp pair-end reads (Supplementary Table S1).
Filtering sequence reads for both poor quality and adapters were performed via Trim Galore v0.6.4 (https://github.com/FelixKrueger/TrimGalore). Alignment to the human reference assembly hg19 and methylation calling was performed Bismark v0.21.0 (PMID:21,493,656).
CpG sites in the resulting RRBS data were then interrogated for methylation patterns and differential methylation (q value < 0.05 and methylation percentage difference of at least 25%) using the methylKit. The differential methylation data were then queried for differentially methylated regions (DMRs) using annotatr (https://bioconductor.org/packages/release/bioc/html/annotatr.html) and GeneDMRs, R package for Gene-based DMR analysis (https://www.biorxiv.org/content/10.1101/2020.04.11.037168v1.full). The resulting DMR outputs were also visualized in the genome browser or in the Integrative Genomics Viewer (IGV) program (https://software.broadinstitute.org/software/igv/). To establish the relationship between histone mark, insulators, transcription factors, and DMRs, the RRBS data conjunction with the data acquired from the Database of ChIP-Atlas Peak browser (https://chip-atlas.org/peak_browser) were projected on to the IGV.
In silico analysis
Gene expression data obtained by the RNA sequencing technique involved in cholesterol metabolism in different anatomical regions (infiltrating tumor, pseudopalisading cells) of the glioblastoma multiform were identified using the Ivy Glioblastoma Atlas Project (Ivy GAP) database (http://glioblastoma.alleninstitute.org/).
Statistical analyses were performed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA). Data are expressed as means ± SD of experiments performed in triplicate. One-way analysis of variance with Tukey’s post-hoc test or unpaired Student t-test was performed for comparisons of the quantitative data between groups. P < 0.05 was considered to indicate a statistically significant difference.