Data Downloading and Normalization
We utilized the publicly available transcriptome data and clinical information of 763 ccRCC patients from TCGA database (https://portal.gdc.cancer.gov/repository), CPTAC database (https://cptac-data-portal.georgetown.edu/studysummary/S050), ICGC database (https://icgc.org/), and EMBL database (https://www.ebi.ac.uk/arrayexpress/), including TCGA-KIRC, CPTAC-3, RECA-EU, and E-MTAB-3267 cohorts. The expression levels of the RNA-seq samples were transformed from fragments per kilobase of transcript per million mapped reads (FPKM) to transcripts per million (TPM), and log2(TPM + 1) was taken. Batch effects were corrected using the combat algorithm of the “sva” package in R software. In addition, copy number variations (CNVs) files, somatic mutation data, and tumor mutation burden (TMB) data were obtained from the TCGA database. The detailed baseline clinical data of ccRCC patients was summarized in Table S1.
Collection of Glutamine Metabolism Related Genes (GMRGs)
We acquired the GMRGs set from the Molecular Signatures Database (MSigDB; https://www.gsea-msigdb.org/gsea/msigdb/human/search.jsp). We selected ‘GOBP_GLUTAMINE_FAMILY_AMINO_ACID_METABOLIC_PROCESS’ gene set, and intersected the same genes between the gene set and the transcriptome data of above integrated cohort for further analyses (Table S2).
Unsupervised Clustering Analysis
We applied unsupervised clustering to separate patients into distinct molecular subtypes with the “ConsensusClusterPlus” package in R software19. The consensus matrix, cumulative distribution function (CDF), and relative change in area under the CDF curve were employed to select the best cluster number.
Exploration of the Immune Landscape in Distinct Subgroups
We applied CIBERSORT algorithm to quantify the proportions of immune cells in the tumor microenvironment (TME) of each ccRCC sample20. The expressions of immune checkpoints (ICPs) were analyzed and compared among distinct clusters. Besides, we identified the abundance of stromal and immune compositions of ccRCC samples with the ESTIMATE algorithm21.
Identification of Differential Expression Genes (DEGs) and Functional Enrichment Analysis
We identified the DEGs among distinct GMRCluster with the “limma” package in R software, and the criteria was set at adjusted P value < 0.001 and |log2-fold change (FC)| > 0.822. Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were conducted with the “clusterProfiler” package in R software, premised on the p-value and q-value thresholds of <0.0523. We performed the gene set variation analysis (GSVA) among distinct clusters with the KEGG gene set (c2.cp.kegg.v2023.1.Hs.symbols), and the criteria was set at |log2-fold change (FC)| > 0.1 and adjusted P value < 0.0524 .
Construction and Validation of a Glutamine Metabolism Related Predictive Signature
We performed the univariate Cox regression analysis with the DEGs to identify the prognosis-related genes. Statistically significant DEGs (P < 0.001) were then used for the subsequent least absolute shrinkage and selector operation (LASSO) analysis and multivariate Cox analysis to construct the predictive signature (GMRScore). The GMRScore was computed with the RiskGenes(SMTNL2, MIOX, TMEM27, SLC16A12, HRH2,and SAA1) as described: GMRScore = h0(t)* exp (∑expression of RiskGene* corresponding coefficient). Patients were stratified into a low or high GMRGroup using the median GMRScore.
We evaluated the prognostic significance of the signature with Kaplan–Meier analysis, and stratified survival analyses for the patients in different clinical subgroups were conducted. Tumor immune dysfunction and exclusion (TIDE) with default parameters was employed to predict the clinical response to ICB. Patients with high TIDE scores were predicted to be non-responders, while patients with low TIDE scores were considered to be responders25. Besides, two cohort of patients treated with anti-PD1 ICB were used to validate the immunotherapy response of the GMRScore. The David Liu cohort consisted of 121 metastatic melanoma patients who were treated with nivolumab or pembrolizumab26. The David A. Braun cohort consisted of 181 advanced clear cell renal cell carcinoma patients treated with nivolumab27.
Establishment and Validation of a Nomogram
We depicted a nomogram combining the GMRScore and clinical characteristics to predict the 1-, 3-, and 5-year overall survival (OS) for ccRCC patients using the “rms”, “regplot”, and “survival” packages in R software. The ROC curves and calibration curves were performed to access the reliabilities of the nomogram.
External Validation of P5CS Protein Expression
The proteomic profiling of 232 ccRCC and paired adjacent samples obtained from Fudan University Shanghai Cancer Center (FUSCC) were analyzed to validate the expression of P5CS (ALDH18A1) in ccRCC. All patients provided consent for the examination and signed an informed consent form. The Helsinki Declaration II was followed for the design of the study and the testing techniques. Further, the Fudan University Shanghai Cancer Center's ethical committee approved the study methods utilized in this research (FUSCC, Shanghai, China).
Cell Culture and Reagents
Human ccRCC 786-O and 769-P cells were purchased from the Type Culture Collection Cell Bank, Chinese Academy of Sciences. The cells were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. The cells were cultured at 37 °C with 5% CO2.
Cell Transfection
A day prior to transfection, cells were seeded in a 10-cm dish to achieve a 50% density. Once the cells reached a 70% density, transfection could commence. The negative control and P5CS siRNA were blended with Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) as per the manufacturer's instructions. Following a 15-minute incubation at room temperature, the mixtures were introduced to the cell culture dish. Cells were collected 48 hours post-transfection for subsequent analysis.
Western Blotting
Cells were harvested by scraping into an SDS sample buffer containing a cocktail of protease inhibitors and PhosSTOP Phosphatase Inhibitor (Roche, Pleasanton, CA, USA). Cell extracts were boiled in loading buffer, and equal amounts of cell extracts were separated on 10% SDS-PAGE gels. Western blotting was conducted according to the standard procedure. We investigated the abundance of P5CS (17719-1-AP,1:1000, Proteintech), N-cadherin (22018-1-AP, Proteintech), SNAI1 (13099-1-AP, Proteintech),Vimentin (10366-1-AP, Proteintech), and GAPDH (60004-1-Ig, Proteintech).
Cell Counting Kit (CCK)-8 Assay
The cells was plated in a 96-well plate at a density of 5000 cells per well. After incubation for 1, 2, 3, 4, and 5 days, the cells were detected the cell viability by using the Cell Counting Kit-8 cell viability kit according to the manufacturer’s instructions. The absorbance at 450 nm was measured using a microplate reader (BioTek, VT, USA).
Transwell Assay
In summary, 5 × 103 cells that were stably infected were resuspended in a serum-free medium and planted into each top compartment of a 24-well transwell plate. A layer of Matrigel was spread on the upper surface of the Transwell filter. The bottom compartment received 800 μL of medium enriched with 20% FBS. Following a 24-hour incubation period at 37 °C and 5% CO2, the chambers underwent three PBS washes, each lasting five minutes. Cells in the lower chambers were then stabilized with 4% paraformaldehyde for half an hour, while the non-invasive cells above were removed using cotton balls. Next, a 20-minute staining process with 1% crystal violet was performed at room temperature. After another series of three 5-minute PBS washes, the cells in the lower section were examined and photographed using an inverted microscope. The ImageJ software was utilized to tally the number of invaded cells in each viewed area.
Cell apoptosis analysis using flow cytometry
After cells were transfected, the cells were trypsinized without EDTA. After washing twice with pre-chilled PBS, 1-10×105 cells were re-suspended in 500ul 1X binding buffer provided in the FITC Annexin V Apoptosis Detection kit (cat#:AP101-100-kit; Multi Sciences) to 1-10×105 cells/ml. Next, 5 µl Annexin V-FITC and 10 µl PI staining solution was added and the samples were incubated at room temperature for 5 min in the dark. The cells were subjected to flow cytometry analysis with a FACS Calibur flow cytometer.
Statistical Analysis
All statistical analyses and graph visualizations were conducted using R v4.3.2 and GraphPad Prism v8.0.2. The Wilcoxon rank-sum test was applied to compare the differences between the two groups. The Kaplan-Meier method was utilized to perform survival analysis, and log-rank test was used to determine the significance. A P-value of less than 0.05 was considered statistically significant unless stated otherwise.