Cell lines and primary tissue
Cell lines were purchased from Japan Health Science Research Resource Bank (IM95, NUGC3, and MKN1), ATCC (SNU16, AGS) and Korean Cell Line Bank (NCC59, NCC24, SNU1967, SNU719, SNU484, NCC19, and SNU1750). YCC10 and YCC11 (Yonsei Cancer Centre in Seoul, South Korea), GES1 (Dr. Alfred Cheng, Chinese University of Hong Kong), and HFE145 (Dr. Hassan Ashktorab, Howard University) were kind gifts. Cell lines were authenticated using Short Tandem Repeat profiling using ANSI/ATCC ASN-0002-2011 guidelines and tested Mycoplasma negative according to the MycoAlert Mycoplasma Detection Kit (Lonza). Normal human stomach antrum and fundus tissue slides were purchased from Novus Biologicals (NBP2-30203, NBP2-30204). Human tissue microarray slides containing matched normal and gastric cancer cases, and human control tissue sections (liver and muscle) were provided and processed by SingHealth Advanced Molecular Pathology Laboratory, Singapore.
GC cell line panel and subtyping
EBV and MSI positive and negative cell lines (n = 12) for the GC discovery cell line panel were selected based on prior literature [65,66,67,68,69]. For the DNA methylation subtyping of GC cell lines (n = 62), raw intensity data from the Infinium HumanMethylation450 beadchip array were analyzed in R using the minfi package to obtain beta values (ratio of methylation). Samples were normalized using the functional normalization function. CpG sites mapping to sex chromosomes or SNP regions were excluded. Similar to the criteria used in the gastric TCGA study, variations (quantified by standard deviations) of the methylation levels at CpG sites within CpG islands across all samples were calculated, and the top 10,000 sites (Additional file 1: Figure S1) were used for an unsupervised hierarchical clustering using Ward’s linkage method and Euclidean distance metrics which identified 4 major branches. We categorized these 4 branches into 3 major methylation groups, which we termed “CIMP-high” (branch 1), “CIMP-low” (branches 3 and 4), and “non-CIMP” (branch 2). Infinium HumanMethylation450 beadchip array data and mutational status of select cancer genes for the GC cell lines was accessed through the Singapore Gastric Cancer Consortium (http://www.sgcc.sg/research/unique-resources).
MeDIP-sequencing and analysis
DNA was sonicated using COVARIS S2 and peak fragment distribution between 100 and 500 bp was verified on an Agilent Bioanalyzer (Agilent Technologies) using the DNA1000 chip. Fragmented DNA was end-repaired, dA-tailed, and adapter ligated using NEBNext® DNA Library Prep Master Mix Set for Illumina (E6040). Samples were then spiked with control DNAs that were unmethylated, methylated, and hydroxymethylated (Diagenode C02040010) as a quality control measure. For each sample, input DNA that was not exposed to the primary antibody was included. Adapter-ligated DNA was subjected to immunoprecipitation with a primary monoclonal antibody against 5-methyl cytosine (Diagenode C15200081) using a previously published protocol [70]. Real-time PCR using primers against the spiked DNA controls were performed to verify successful and specific enrichment of methylated DNA (data not shown). Immunoprecipitated samples were amplified using Phusion® High-Fidelity DNA Polymerase (M0530) and NEBNext® Multiplex Oligos for Illumina® (E7335) for 10 cycles. Amplified libraries were run on the Agilent Bioanalyzer using the High sensitivity DNA kit prior to Illumina sequencing using a single-end 100 base pair configuration. MeDIP short reads were aligned to the h19 reference genome and duplicates were removed using SAM tools. Average library size of the 14 MeDIP libraries were ~ 80 million reads. Peaks were called using MACS2 using input control. Differential peak analysis between cell lines was performed using DiffBind (FDR ≤ 0.1) of CIMP vs. non-CIMP and non-malignant cell lines which generated a list of 595 unique genes with gain of promoter methylation (+/− 2 kb TSS) and annotated using GenCODE.
RNA sequencing and analysis
Total RNA quality check was done using the RNA 6000 Lab Chip Kit on the Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA). Libraries were prepared using Illumina Tru-Seq Stranded Total RNA with Ribo-Zero Gold kit protocol, according to the manufacturer’s instructions (Illumina, San Diego, CA, USA). Library fragment size was determined using the DNA 1000 Kit on the Agilent Bioanalyzer (Agilent Technologies). Libraries were quantified by qPCR using the KAPA Library Quantification Kit (KAPA Biosystems). Libraries were pooled in equimolar and cluster generation was performed on the Illumina cBOT system (Illumina). Sequencing (150 bp pair-end) was performed on the Illumina HiSeq 3000 system at the Duke-NUS Genome Biology Facility, according to the manufacturer’s protocol. The paired end reads were aligned to the human reference genome (GRCh37) using Hisat2. Gene expression profile (FPKM or RPKM) for each sample was determined.
Proteomics and analysis
Each cell line was grown and extracted in quadruplicates using RIPA buffer (Sigma) according to the manufacturer’s instructions. In total, 200 μg of protein was used for MS sample preparation. Samples were boiled at 95 °C prior to separation on a 12% NuPAGE Bis-Tris precast gel (Thermo Fisher Scientific) for 15 min at 170 V in 1× MOPS buffer. The gel was fixed using the Colloidal Blue Staining Kit (Thermo Fisher Scientific), and each lane was divided into 2 equal fractions. For in-gel digestion, samples were destained in destaining buffer (25 mM ammonium bicarbonate; 50% ethanol), reduced in 10 mM DTT for 1 h at 56 °C followed by alkylation with 55 mM iodoacetamide (Sigma) for 45 min in the dark. Tryptic digest was performed in 50 mM ammonium bicarbonate buffer with 2 μg trypsin (Promega) at 37 °C overnight. Peptides were desalted on StageTips and analyzed by nanoflow liquid chromatography on an EASY-nLC 1200 system coupled to a Q Exactive HF mass spectrometer (Thermo Fisher Scientific). Peptides were separated on a C18-reversed phase column (25 cm long, 75 μm inner diameter) packed in-house with ReproSil-Pur C18-AQ 1.9 μm resin (Dr Maisch). The column was mounted on an Easy Flex Nano Source and temperature controlled by a column oven (Sonation) at 40 °C. A 215-min gradient from 2 to 40% acetonitrile in 0.5% formic acid at a flow of 225 nl/min was used. Spray voltage was set to 2.4 kV. The Q Exactive HF was operated with a TOP20 MS/MS spectra acquisition method per MS full scan. MS scans were conducted with 60,000 at a maximum injection time of 20 ms and MS/MS scans with 15,000 resolution at a maximum injection time of 50 ms. The raw files were processed with MaxQuant version 1.5.2.8 using the LFQ quantification option [71] on unique+razor peptides with at least 2 ratio counts. Carbamidomethylation was set as fixed modification while methionine oxidation and protein N-acetylation were considered as variable modifications. Search results were filtered with a false discovery rate of 0.01. Known contaminants, proteins groups only identified by site, and reverse hits of the MaxQuant results were removed. LFQ intensities can be found in Supplementary data file (GC Proteomics). Differential analysis (log base2 fold change < 2, adjusted p value < 0.05) revealed 64 downregulated proteins in GC CIMP [n = 6] vs non-CIMP and non-malignant cell lines [n = 8] (Additional files 4, 5).
Gene expression and pathway analysis
Differential gene expression of GC CIMP [n = 6] vs non-CIMP and non-malignant cell lines [n = 8] was performed using Deseq2 (log base2 fold change ≤ 2, adjusted p value < 0.05). For the pan-cancer analysis, 22 cancers (TCGA abbreviations: LIHC, HNSC, ESCA, THYM, BLCA, UCEC, ACC, LGG, BRCA, CESC, CHOL, COAD-READ, GBM, KIRP, LUAD, MESO, PAAD, PRAD, SARC, SKCM, TGCT, and UVM) were analyzed for differential gene expression analysis of CBS in most “hypermethylated/CIMP” vs. other methylation categories (Wilcoxon rank sum test, P < 0.05), as defined by TCGA (The Cancer Genome Atlas) or by Karpinski et al (for UCEC) [52] via MEXPRESS [72]. For UCEC, the TCGA-CIMP sample labels were not publicly available, and therefore, we used CIMP sample labels from Karpinski et al (a pan-cancer study of CIMP). For the GES1 CBS-deficient cells (P15) and HFE145 CBS-deficient cells (P23), a log base2 fold change was generated for each clone compared to parental and CRISPR control cells and further filtered with a minimum difference of 1 FPKM (fragments per kilobase of exon model per million reads mapped). Only genes with a linear fold difference (upregulated [1.5 fold change]) and downregulated [0.6 fold change]) were used for the gene set overlap analysis. For TCGA and CCLE (Cancer Cell Line Encyclopedia) data, genes positively and negatively correlated with CBS expression were obtained from cbioportal and filtered using q-values < 0.05. MSigDB overlap was computed with hallmark gene sets (v6.2) with q-values < 0.01. For the mouse samples, a log base2 fold change was generated for each gender in the Tg-hCBS Cbs-/- category compared to Cbs+/+and further filtered with a minimum difference of 0.5 FPKM. Differential analysis revealed 413 upregulated (1.5 fold change) and 527 downregulated genes (0.6 fold change) common to both genders, which was used for the PANTHER Go-Slim Biological process analysis (FDR < 0.05). For the MSigDB analysis of mouse samples, a log base2 fold change was generated for each gender in the Tg-hCBS Cbs-/- category compared to Cbs+/+and further filtered with a minimum difference of 1 FPKM. Differential analysis and conversion of mouse to human gene identifiers by the MSigDB tool (v7.2) indicated the following. For upregulated genes (1.5 fold change), 412/1032 converted in female and 555/1249 converted in male whereas for downregulated genes (0.6 fold change), 962/1768 converted in female and 558/1191 converted in male. Overlap with hallmark gene sets (v7.2) with q-values < 0.01 was performed with the converted IDs.
Identification of candidate epimutations in CIMP group
For the discovery GC cell line panel, we overlapped promoter hypermethylated genes (n = 595), with those downregulated at RNA (n = 1464 genes) and protein levels (n = 64) in CIMP vs. non-CIMP groups using omic platforms as described in the relevant section. For the recurrence analysis of the candidate epimutations in the expanded GC cell line cohort, RPKM, and methylation values were accessed through the Singapore Gastric Cancer Consortium (http://www.sgcc.sg/research/unique-resources). For the independent analysis of primary GCs, promoter associated hypermethylated genes in CIMP (as defined by TCGA and clustering of Singapore cohort methylation data [4]) vs. non-CIMP group (β-value difference ≥ 0.3) were overlapped with downregulated genes in CIMP vs non-CIMP groups (q-value < 0.05, fold difference ≤ 0.6, median RPKM value in non-CIMP is ≥ 1). We used a stringent β-value difference of ≥ 0.3 since we were concomitantly assessing the effect of hypermethylation on gene expression. TCGA methylation and gene expression data were derived from Infinium HumanMethylation450 beadchip and RNA sequencing platforms respectively. SG cohort methylation and gene expression data were derived from Infinium HumanMethylation27 beadchip and Affymetrix Human Genome U133 Genechips platforms respectively. Except for FERMT2 in SG cohort, CpG probes pertaining to the other candidate epimutations were represented in both cohorts.
RNA extraction, reverse transcription, and quantitative PCR
Total RNA was extracted using the Qiagen RNAeasy mini kit according to the manufacturer’s instructions. RNA was converted to cDNA using Improm-IITM Reverse Transcriptase (Promega). Quantitative PCR was performed in triplicate using Quantifast SYBR Green PCR kit (Qiagen) on an Applied Biosystems HT7900 Real Time PCR System for 40 cycles using the following primers: CBS Exon 10/11 F—5′ TGCGGAACTACATGACCAAG 3′ , CBS Exon 10/11 R—5′ TGAGGTCCTCCTCCTTCAGA 3′, CBS Exon 16/17 F—5′ GCAGATCCAGTACCACAGCA 3′, CBS Exon 16/17 R—5′ CGAAGGAGAAGTGGGCAAG 3′, ACTB F—5′ TCCCTGGAGAAGAGCTACG 3′, ACTB R—5′ GTAGTTTCGTGGATGCCACA 3′. Delta Ct was calculated as a difference in Ct number between experimental and reference gene.
Western blotting
Protein lysates were prepared using ice-cold RIPA buffer (Thermo Fisher Scientific, 89900) containing protease inhibitor cocktail (Nacalai Tesque). Protein concentration was determined using Pierce BCA protein assay (Thermo Scientific, 23225). Laemmli sample buffer with reducing agent was added to the lysates and boiled at 100 °C for 5 min. Samples were loaded in each well of a 4–15% Mini-Protean TGX gel (Bio-Rad) and transferred to the PVDF membrane using a semi-dry blotting system (Bio-Rad). Membranes were probed with monoclonal antibodies against CBS (1 in 500 dilution, Abnova H00000875-M01) or ACTB (1 in 2000 dilution, Sigma A5316) in 5% milk-PBST [phosphate-buffered saline with Tween] at 4 °C overnight. Secondary anti-mouse antibody (LNA931, Amersham) was used at a dilution of 1 in 2000 for 1 h at room temperature. Membranes were developed using Amersham ECL Prime Western Blotting Detection Reagent and imaged using Chemidoc Imaging system (Bio-Rad). Bands were quantified using Image Lab software (Bio-Rad).
CBS genetic data
Somatic mutation and copy number data for the stomach adenocarcinoma TCGA 2014 dataset was obtained from cbioportal. FATHMM scores for the single-nucleotide CBS somatic variants were derived from COSMIC database. Germline exomes for the stomach TCGA cohort were aligned and variant called as previously described [73]. Briefly, variants with an ExAC MAF < 0.05 were predicted to be deleterious if they were HIGH impact (via Variant Effect Predictor) or MODERATE impact with a CADD score > 25. An additional pathogenicity score for rs117687681 (R369C) was computed using CScape (0.89). CBS promoter methylation values for the 6 mutated cases were calculated as an average of 2 TSS probes, cg22633722 and cg09622447.
H3K27me3 chromatin immunoprecipitation [ChIP]
H3K27me3 ChIP was performed by Zymo Research (ab6002, Abcam). After recovery of ChIP and input DNA, whole-genome amplification was performed using the WGA4 kit (Sigma-Aldrich) and BpmI-WGA primers. Amplified DNAs were purified using PCR purification columns (QIAGEN) and digested with BpmI (New England Biolabs) to remove WGA adapters. Thirty nanograms of amplified DNA was used for each sequencing library preparation (New England Biolabs). Libraries were sequenced on Hiseq2500 (Illumina) to an average depth of 20–30 million reads per library. Sequence reads were mapped against the human reference genome (hg19) using Burrows-Wheeler Aligner (BWA-MEM, version 0.7.10). Reads from histone ChIP-seq were trimmed at the first and last 10 bp prior to alignment. Reads with low mapping quality (MAQP< 10) and PCR duplications were removed using Samtools. Histone ChIP-enriched peaks were detected using CCAT. Positive controls for H3K27me3 enrichment were selected from a prior study [74].
Transfection experiments
SNU1750 cells were transfected with 3 individual siRNAs for CBS (Integrated DNA Technologies, hs.Ri.CBS.13.1/13.2/13.3]) or negative control (Integrated DNA Technologies, 51-01-14-04) at 5 nM using Lipofectamine RNAiMAX (Thermo Fisher Scientific, 13778150). RNA and protein were harvested at 48 h. For the NF-κB reporter assay [75], GES1 and HFE145 CBS-deficient cell lines were transfected with 500 ng of 6X-NF-κB-Luciferase reporter or promoter-less pGL3-Basic vector using Lipofectamine 2000 (Thermo Fisher Scientific). Luciferase activity (Promega, E1500) was measured at 48 h post transfection for basal measurements and at 24 h post transfection for the GYY 4137 experiment. Fold change in luciferase activity was normalized to the viability of the cells using the CellTiter-Glo® Luminescent cell Viability Assay for all samples.
Cell line treatment with drug
CIMP cell lines were treated with 1uM Azacytidine [Sigma-Aldrich, A3656] or DMSO for 72 h prior to RNA harvest. GES1 and HFE145 CBS-deficient cell lines were treated with 0.5 mM of H2S donor GYY 4137 (Cayman Chemical, 13345) or DMSO for 24 h.
Immunohistochemistry
CBS monoclonal antibody (Abnova H00000875-M01) was initially tested for immunohistochemistry application in different buffer and antigen retrieval conditions using human liver sections as a positive control (data not shown), after which the following automated method was selected on the Leica Bond III machine. In brief, antigen retrieval was performed with Bond Epitope Solution 2, with primary antibody dilution of 1 in 400 and incubation time of 20 min. Leica Bond Refine kit was used for detection. Imaging was performed on a Nikon Eclipse Ti. The gastric tissue microarray slides were analyzed by a pathologist and cases were scored “positive” if cytoplasmic staining was found in > 5% of gastric epithelial cells and “negative” if staining was entirely absent or present in < 5% of gastric epithelial cells.
Generation of CBS CRISPR clones
GES1 and HFE145 cells were transiently transfected with CRISPR Cas single format vector with a backbone of U6-gRNA/CMV-Cas9-GFP (Sigma-Aldrich) and containing guide RNAs targeting exon 3 of CBS gene (HS0000002142 [clone1], HS0000002144 [clone 2], Sigma-Aldrich) or control vector (CRISPR universal negative control 1, Sigma-Aldrich) using Lipofectamine 2000. After 48 h, ~ 960 GFP positive single cells from each transfection were sorted by the Duke-NUS Flow core facility. After ~ 3 weeks, at least ten clones from each vector were cultured and screened and knockouts verified by loss of CBS protein using Western blotting. For each cell line model, 2 CBS-deficient clones, 1 CRISPR control, and parental cells were utilized for all downstream experiments.
Metabolic analysis of CBS CRISPR clones
GES1 parental, CRISPR control cells, and CBS-deficient clone 1 and clone 2 were cultured and harvested together at P16–P19. HFE145 parental, CRISPR control cells, and CBS-deficient clone 1 and clone 2 were cultured and harvested together at P23–P24. Each cell line was harvested at least in duplicate. Cells were collected and quenched using 150 mM cold sodium chloride (Merck, Germany) solution. Following centrifugation at 3400g, the cell supernatant was removed. The cell pellets obtained were then processed using a two-phase liquid–liquid extraction protocol based on the modified method of Bligh and Dyer. Briefly, polar and lipid metabolites were extracted by sequential addition of methanol (Optima grade, Fisher Scientific, USA), 3.8 mM tricine (Merck) solution and chloroform (gradient grade, Merck) (1:0.5:1 v/v/v, total 2 mL) to the sample. The mixture was vortexed for 1 min following each addition of solvent. The mixture was then centrifuged at 4 °C, 14,000g for 20 min. This resulted in separation of the sample into two fractions—the top methanol-tricine solution layer contained the polar metabolites while the bottom chloroform layer contained the lipid species. The top polar layer was collected and the remaining mixture was re-extracted using a mixture of methanol and tricine solution (9:10 v/v), followed by centrifugation at 4 °C, 14,000g for another 10 min. The resulting polar layer was combined together with the first polar extract and the bottom chloroform layer containing the lipid metabolites was collected separately. Both layers were stored at – 80 °C before LC-MS analysis. Each polar extract was analyzed in triplicate using an ultra-performance liquid chromatography system (UPLC, Acquity, Waters, USA) in tandem with a mass spectrometer (Q Exactive, Thermo Scientific, USA). A C18 UPLC column (Acquity UPLC HSS T3 column, 2.1 × 100 mm, 1.8 μm, Waters) was used for separation and the mobile phase comprised of two solvents. “A” being water with 0.1% formic acid (VWR, USA) and “B” being methanol with 0.1% formic acid. The UPLC program is as follows: the column was first equilibrated for 0.5 min at 0.1% B. The gradient was then increased from 0.1% B to 50% B over 8 min before being held at 98% B for 3 min. The column was washed for a further 3 min with 98% acetonitrile (Merck) with 0.1% formic acid and finally equilibrated with 0.1% B for 1.5 min. The solvent flow rate was set at 0.4 mL/min; a column temperature of 30 °C was used. The eluent from the UPLC system was directed into the MS. High-resolution mass spectrometry was then performed in both positive and negative electrospray ionization (ESI) modes, with a mass range of 70 to 1050 m/z and a resolution of 70,000. Sheath and auxiliary gas flow was set at 30.0 and 20.0 (arbitrary units) respectively, with a capillary temperature of 400 °C. The spray voltages were 1.25 kV for positive and 1.5 kV negative mode ionization. Mass calibration was performed using standard calibration solution (Thermo Scientific) prior to injection of the samples. A quality control (QC) sample comprising of equal aliquots of each sample was run at regular intervals during the batch LC-MS runs. Lipid extracts were analyzed in triplicate using an ultra-performance liquid chromatography system (UPLC, Acquity, Waters) in tandem with a mass spectrometer (Xevo G2 QToF, Waters). A C18 UPLC column (Acquity UPLC CSH column, 1.0 × 50 mm, 1.7 μm, Waters) was used for separation and the mobile phase comprised of two solvents: “A” comprising of acetonitrile, methanol, and water (2:2:1) with 0.1% acetic acid (Merck) and 0.025% ammonia solution (VWR, USA), and “B” comprising of isopropanol (Fisher Scientific) with 0.1% acetic acid and 0.025% ammonia solution. The UPLC program is as follows: the column was first equilibrated for 1 min at 1% B with a flow rate of 0.1 mL/min. The gradient was then increased from 1% B to 82.5% B over 9 min before B was further increased to 99% for a 5 min wash at a flow rate of 0.15 mL/min. The column was re-equilibrated for 2.2 min at 1% B. Column temperature was maintained at 45 °C and eluent from the LC system was directed into the MS. High-resolution mass spectrometry was then performed in both positive and negative ESI modes with a mass range of 100 to 1800 m/z and a resolution of ≥ 10,000. Cone and desolvation gas flows was set at 40.0 and 600.0 (L/h) respectively, with a desolvation temperature of 600 °C. The ESI capillary voltages were 2.0 kV for positive mode ionization, and 1.0 kV for negative mode ionization. Mass calibration was performed using sodium formate prior to injection of the samples. Similarly, a quality control (QC) sample comprising of equal aliquots of each sample was run at regular intervals during the batch LC-MS runs. The raw LC-MS data obtained from both polar and lipid extracts was processed using a XCMS-based peak finding algorithm. The QC samples were first used to adjust for instrumental drift and subsequently the experimental data was normalized based on the total ion intensity. Metabolite identities were confirmed based on mass spectral comparison with available metabolite standards. Both univariate (Student’s t test with Welch correction) and multivariate (Partial Least Squares Discriminant Analysis, PLS-DA) (SIMCA version 13.0.3, Umetrics, Sartorius, Sweden) approaches were used to identify statistically significant differences in metabolite levels between parental and CRISPR control cells vs. CBS-deficient clone 1 and clone 2 (univariate—p value < 0.05; multivariate—variable importance for projection, VIP > 1).
For the spent media sample preparation, GES1 parental, CRISPR control cells, and CBS-deficient clone 1 were cultured and harvested together at P20. RPMI media containing no cells was used as a control. Twenty microliters of 0.5 M dithiothreitol (DTT) (Bio-Rad, CA, USA) was added to 400 μL of medium [30, 76]. The mixture was vortexed for 1 min and incubated at 37 °C for 15 min. After incubation, the mixture was centrifuged at 4 °C, 12,000g for 5 min. The treated medium was transferred into a vial and used immediately for homocysteine detection using LC-MS/MS analysis. For the detection of SAM, SAH, and methionine, the spent medium was analyzed separately without any treatment. The treated medium was analyzed in 4 replicates using an ultra-performance liquid chromatography system (UPLC, Acquity, Waters Corp) in tandem with a triple-quadrupole mass spectrometer (Xevo TQ-S, Waters Corp) operating in ESI-positive mode. The source temperature and desolvation temperature were set at 150 °C and 500 °C, respectively. The cone gas flow was 150 l h–1 and the desolvation gas flow was 700 l h–1. The capillary voltage was 2.90 kV. Compound-dependent MS parameters and multiple-reaction monitoring transitions for the analytes are shown in the table below. A C18 UPLC column (Acquity UPLC BEH column, 2.1 × 100 mm, 1.7 μm, Waters) was used for chromatographic separation and the mobile phase comprised of two solvents: “A” comprising of water with 0.1% formic acid (VWR, USA) and “B” comprising of methanol (Fisher Scientific, USA) with 0.1% formic acid. The UPLC program is as follows: the column was first equilibrated for 0.5 min at 0.1% B with a flow rate of 0.4 mL/min. The gradient was then increased from 0.1% B to 10% B over 2.5 min before B was further increased to 50% for 3 min at a flow rate of 0.3 mL/min. The column was re-equilibrated at 0.4 mL/min for 1.5 min at 1% B. Column temperature was maintained at 30 °C and eluent from the LC system was directed into the ESI source of the mass spectrometer. The injection volume was 4 μl. Chromatographic peak integration was performed with Targetlynx software (version 4.1 SCN810, Waters Corp). Student’s t test was used to assess statistical significance between control and CBS-deficient groups for homocysteine, SAM, SAH, and methionine measurements.
Optimized compound-dependent MS parameters using Xevo TQ-S mass spectrometer
Analyte
|
Precursor ion mass (m/z)
|
Fragment ion mass (m/z)
|
Dwell time (s)
|
Cone voltage (V)
|
Collision energy (V)
|
---|
Homocysteine
|
136.0
|
90.1
|
0.025
|
15
|
16
|
SAM
|
399.2
|
136.2
|
0.025
|
62
|
42
|
SAH
|
385.1
|
134.0
|
0.025
|
66
|
33
|
Methionine
|
150.1
|
104.7
|
0.025
|
40
|
15
|
H2S synthesis enzyme activity
The cellular H2S production rate was measured following previously described fluorescent methods with minor modifications [77]. Briefly, cultured cells were homogenized in an ice-cold buffer (50 mM Tris-HCl, pH 8.0, 50 mM NaCl) by passing through a 29G-needle syringe 20 times, followed by centrifugation at 14,000g for 15 min at 4 °C. The protein concentration was determined by Nanodrop with correction to remove DNA/RNA contamination. The reaction mixture contained cell extract, 50 mM Tris-HCl (Sigma-Aldrich Corp.), 1 mM L-cysteine (Sigma- Aldrich Corp.), 1 mM homocysteine (Sigma-Aldrich Corp.), 50 μM pyridoxal-5′-phosphate (MP Biomedicals, Santa Ana, CA, USA), and 10 μM H2S probe 7-azido-4-methylcoumarin (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Reactions were incubated for 2 h at 37 °C and fluorescence was measured using ex 365 nm, em 450 nm with Varioskan Flash microplate reader. A linear standard curve was generated from reactions with sodium hydrosulfide.
Methylation array profiling and analysis
GES1 parental, CRISPR control cells (P8, P17, P21, P26, and P31), and CBS-deficient clone 1 (P8, P15, P19, P24, and P29) and clone 2 (P8, P14, P18, P23, and P28) were cultured and harvested together at indicated time points. HFE145 parental, CRISPR control cells (P5 and P23), and CBS-deficient clone 1 (P5 and P23) and clone 2 (P5 and P23) were cultured and harvested together at indicated time points. Genomic DNA was hybridized on Infinium MethylationEPIC Beadchip according to the manufacturer’s specifications. Array normalization was performed using the preprocessFunnorm function from the minfi package [10.1093/bioinformatics/btw691]. Probes where the β-value consistently deviated more than 0.2 compared to both the CRISPR negative control and parental cell line of the same passage were considered differentially methylated. Since the global methylation analysis of the CBS-deficient cell lines were across all CpG contexts, we adopted a more stringent β-value difference of ≥ 0.2. Enrichment analysis for PRC2 related-factors was performed using ReMap 2018 v 1.2 [78]. Gastric CIMP- and EBV-CIMP-specific methylation patterns were analyzed using TCGA Infinium HumanMethylation450 beadchip data, comparing samples in either CIMP subgroup to non-CIMP samples in the same dataset. Comparison of GC methylation subtypes (gastric CIMP, EBV-CIMP, non-CIMP) to GIM methylation subtypes (“Methylation high” or CIMP, “Methylation low”’ or non-CIMP as previously published [15]) were performed at the CGI level. For this analysis, we chose a more relaxed β-value difference ≥ 0.1 as we were comparing at the CGI level, which is usually devoid of methylation. In the CpG context analysis, proportion of probes represented the number of hyper- or hypomethylated CpG sites divided by total number of hyper- or hypomethylated CpG sites, respectively in each sample group. Overlap between DNA methylation changes in CBS-deficient cells with DNA methylation changes observed in TCGA gastric CIMP and EBV-CIMP tumors was performed on CGI level, considering any CGIs that were affected on the probe level in both groups. For the pan-cancer analysis, CBS promoter methylation was calculated as an average of 7 CpG probes in the promoter CGI based on TCGA Infinium HumanMethylation450 beadchip data accessed through MEXPRESS [72]. CIMP groups were identified based on TCGA defined clusters (LIHC, ESCA, BLCA, THYM, HNSC) or from Karpinski et al. [52] for UCEC. Differential methylation analysis of CIMP vs non-CIMP groups, and linear regression to gene expression was performed using Wilcoxon rank sum test, and Spearman’s correlation analysis, respectively.
Tg-hCBS Cbs
−/− mouse work
Tg-hCBS Cbs−/− mice were generated as previously described [79], which lack the endogenous mouse Cbs gene but have a wild-type human CBS-encoding cDNA under the control of the Zinc-inducible metallothionein promoter. Tg-hCBS Cbs−/− offspring (6 males and 5 females; no zinc supplementation) was obtained from an incross of Tg-hCBS Cbs−/− x Tg-hCBS Cbs−/−, and euthanized for tissue collection at different ages (ranging from 2.1–11.1 months). Mouse stomachs were opened along the greater curvature, washed 3× in cold 1× PBS-10% FBS solution, fixed in 4% formaldehyde and embedded in paraffin for hematoxylin and eosin staining and/or RNA extraction (Qiagen RNeasy FFPE Kit). Cbs+/+mouse stomachs (3 males and 3 females; ranging from 2.8 to 8.1 months) from C57BL/6J mice (genetic background of Tg-hCBS Cbs−/− mice) were purchased from InVivos, Singapore, and processed alongside the Tg-hCBS Cbs−/− stomachs.