Tumor Tissue Specimens
Our study was reviewed and approved by the medical ethics committee of Stomatological hospital, Medical School of Nanjing University, and carried out according to the recommendations of the Declaration of Helsinki. Matched pairs of tumor and normal samples (8 pairs, n = 16) were collected from Stomatological hospital, Medical School of Nanjing University from 2018, another batch of independent fresh tissues samples (n = 20) were collected for verify metabolomics results. All fresh tissues were snap-freezed in liquid nitrogen within 30 min after surgery and preserved at −80°C till for GC-MS analysis. All tumor tissues and normal tissues were confirmed by HE staining of frozen section (Supplementary Figure 1). In key enzyme validated group, another 81 patients with oral squamous cell carcinoma and matched normal tissues were recruited, all patients underwent radical resection at Stomatological Hospital, Medical School of Nanjing University from 2013.
Chemicals and reagents
Reagents required for sample handling were of analytical grade, such as ammonium formate and formic acid (ROE Scientiﬁc Inc, Newark, USA), isopropyl alcohol and methyl tert-butyl ether (MTBE) (TEDIA Inc, Fairﬁeld, USA), acetonitrile Methanol (Merck KGaA Inc, Darmstadt, Germany). Milli-Q water was used throughout the course of the experiment (Millipore, Billerica, MA, USA). Methanol, choloroform and pyridine were purchased from Tedia (Tedia way, Fairfield, USA), and myristic-d 27 acid was obtained from Sigma-Aldrich (St. Louis, MO, USA, respectively). Rabbit polyclonal GLUD1 and GLS1 antibody were purchase from Proteintech (Wuhan, China). Enzy ChromTM Glutamine Assay Kit were purchased from BioAssay Systems (EGLN–100, USA), Glutamate Colorimetric Assay Kit were purchased from Biovision (Catalog #K629–100, USA).
50 mg of frozen tumor and normal tissues were transferred into 2 mL Eppendorf tube and 400 μ L ice-cold 75% methanol, 10 μL tinidazole (internal standard, 50 μg/mL) as an internal standard was added and mixed well with an oscillating mill (Retsch MM 400, 25 Hz, 5 min) and 5 min of vortexing, add 1mL ice-cold MTBE and 200 μ L water. The mixture centrifuged at 15,000g for 10 min at 4 °C resultant clear supernatant was collected and dried in vacuum 45 °C for 3 h. The dried lower phase aliquots were derivatized subsequently by adding 30 μL methoxylamine hydrochloride in pyridine (10 mg/L) and 1 μ L 1,2–13C 2 -myristic acid (internal standard, 5 mg/mL), vortexed and left for 90 min at 30 °C with the Thermo Mixer C (Eppendorf). Then 30mL BSTFA was added with 1% TCMS and shaken at 37 °C for 30 min. The derivatized samples were centrifuged and analyzed within 24 hours.
Gas chromatography-mass spectrometry
Derivatized samples were analyzed by A Trace 1310 Gas Chromatograph equipped with an AS 1310 auto sampler connected to a TSQ 8000 triple quadrupole mass spectrometer (Thermo Scientiﬁc, Waltham, MA) was used to perform GC-MS analysis. TG–5MS GC column was used to separate1μL of derivatized of tumor and normal tissues samples for GC–MS analysis (Thermo, 0.25 mm x 30 m, 0.25 um) in split mode with a 20:1 ratio of above samples. The carrier gas (such as helium) maintained at a constant ﬂow of 1.2 mL/min. Initial temperature of GC oven was 60 °C and kept for 1min, gradually increase the temperature to 320°C and keep it for 5 minutes. The transfer line temperature was set to 250 °C, and the EI ion source temperature was 280 °C, the voltage was 70 eV. A scan range of mass spectra was 50–500m/z and a time range of 3.5–19 min.
TRIzol reagent was used to isolate total RNA (Invitrogen, USA)according to the manufacturer’s instructions. Real-time PCR was performed with Thermal Cycler Dice Real Time System (TaKaRa, Japan). The primer sequences of GLS1 were: 5’-AGGGTCTGTTACCTAGCTTGG–3’ and 5’-ACGTTCGCAATCCTGTAGATTT–3’, GLUD1
were: 5’-CTCCAGACATGAGCACAGGTGA–3’ and 5’-CCAGTAGCAGAGATGCGTCCAT–3’.The primer sequences of β-actin were: 5’-CCTGGCACCCAGCACAAT–3’ and 5’-GGGCCGGACTCGTCATACT–3’. The relative quantity of GLSI and GLUD1 mRNA level was calculated based on the standard ∆∆CT methods.
Immunohistochemistry was used to detect Rabbit polyclonal GLUD1 and GLS1 antibody (Proteintech, China) in 67 patients with early OSCC and matched normal tissues per the company’ sprotocol, PBS was used as negative control. The staining result was determined by counting 1000 tumor cells in three 100x magnification fields by two independent pathologists and further classified as low expression (the percentage of positive rate, ＜25%) and high expression (the percentage of positive rate ≥25%).
Metabolomics statistical analysis
After removed the artiﬁcial peaks due to derivatization, raw data ﬁles of GC-MS analysis determined by NIST 2014 standard mass spectral databases built-in Xcalibur 2.2 software (Thermo Scientiﬁc, Waltham, MA). Sandardized using internal parameters, the peak area of metabolites were calculated with Xcalibur 2.2 software. We use Metaboanalyst1 3.0 perform statistical analyses, which contain the R package of statistical computing software. Total spectral intensity and additionally Pareto scaled
were used to normalized initial experimental data.
Univariate analysis was by ANOVA, such as fold change, T test, Volcano Plot. Multivariate analysis was via unsupervised Principal Component Analysis (PCA) followed by Partial Least Squares Discriminant Analysis (PLS-DA). Hierarchical cluster analysis was carried out using Dendrogram and Heatmap. We also applied KEGG for further metabolite pathway analysis. Immunohistochemical results were analyzed by SPSS 17.0 software package. The relationships between GLS1 protein expression and the clinicopathological parameters were determined by Chi-square tests. We estimated survival curves using the Kaplan-Meier method and compared them using a two-sided log-rank test p.