Integrative multi-omics Analysis Proposes a Metabolic Classification of Gliomas: Distinct Metabolic States, Immune Infiltration, and Prognosis

Integrative multi-omics Analysis Proposes a Metabolic Classification of Gliomas: Distinct Metabolic States, Immune Infiltration, and Prognosis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Integrative multi-omics Analysis Proposes a Metabolic Classification of Gliomas: Distinct Metabolic States, Immune Infiltration, and Prognosis Qiang Zhu, Wanxiang Niu, Maolin Mu, Chengkun Ye, Chaoshi Niu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7196875/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 30 Dec, 2025 Read the published version in Journal of Translational Medicine → Version 1 posted 4 You are reading this latest preprint version Abstract Background The tumor microenvironment (TME) of glioma harbors diverse cell types; however, cell metabolic heterogeneity remains to be explored. This study aims to characterize the metabolic features of different cell types in the TME by integrating multiple datasets, including genomics, bulk and single-cell transcriptomics, and metabolomics. Methods Unsupervised machine learning was used to construct an energy metabolic classifier based on the metabolic pathways identified from bulk RNA-seq of gliomas in the TCGA dataset. The classifier was externally validated using multiple datasets, including genomics, bulk RNA-seq, snRNA-seq, and the metabolomics data. Furthermore, metabolic heterogeneity associated with the classifier was further characterized at single-cell resolution. Results The energy metabolism-based classifier stratified patients into two prognostic clusters: patients in cluster 1 were characterized by high pathway activity of glycolysis, the pentose phosphate pathway (PPP), and fatty acid oxidation (FAO), whereas patients in cluster 2 exhibited higher activity in glutaminolysis. This metabolic classifier revealed both intratumoral and intertumoral metabolic heterogeneity, and the complexity was further validated by the metabolomics profiling and snRNA-seq data from the CPTAC dataset. Notably, OSMR, highly expressed in cluster 1, showed significant co-expression with key glycolytic enzyme genes. The OSM/OSMR/JAK1/STAT3 axis potently drives malignant progression of glioma cells, specially enhancing their invasive and migratory capabilities. Single-cell resolution analyses demonstrated that tumor metabolic heterogeneity is primarily driven by malignant cells rather than non-malignant components, while tumor microenvironment (TME) factors were also found to modulate malignant cell metabolism. Significantly, glycolytic activity in glioma cells increased during the phenotypic transition from PN (proneural) to MES (mesenchymal), with cluster 1 metabolic phenotypes predominating in the tumor core. Compared to cluster 2, cluster 1 patients exhibited higher mRNA expression of immunosuppressive checkpoint genes, which correlated with pronounced immunosuppression in the TME. Furthermore, various immune cells demonstrated distinct metabolic preferences at single-cell resolution. Conclusions This study developed an energy metabolic-based classifier for gliomas with prognostic and therapeutic potential. Metabolic reprogramming was linked with the PN-to-MES transition of glioma cells and immunosuppression in the tumor microenvironment. Multi-omics data, especially snRNA-seq, offered insights into metabolism heterogeneity at single-cell resolution, enabling personalized treatment strategies. Gliomas metabolic reprogramming metabolic heterogeneity unsupervised classification immunity metabolism Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Background Gliomas are the most common and aggressive primary intracranial malignancies, with a poor prognosis despite multimodal treatment, such as surgery, chemotherapy and radiotherapy. Recent advances in molecular research and the rapid development of single-cell sequencing have uncovered remarkable heterogeneity in the tumor microenvironment[1, 2]. Professor Neftel, leveraging single-cell sequencing, proposed glioblastoma cell subtypes based on the transcriptional classification established by Professor Verhaak[2, 3]. Different types of glioblastoma cells exhibit variations in their genetic, epigenetic, and microenvironmental factors, endowing them with distinct capabilities for proliferation and invasion. Energy metabolism reprogramming is a newly recognized hallmarks of tumors[4]. Tumor cells adopt metabolic reprogramming to support their own proliferation and progression [5]. In clinical specimens, Professor Melin’s team has reported distinct metabolic phenotypes exist according to adult glioma subtypes[6]. Recent studies suggest that the plasticity and flexibility of tumor cell metabolism offer potential therapeutic opportunities[7-9]. Single-cell sequencing enables precise characterization of diverse cell types within the tumor microenvironment, including clonal diversity, metabolic profiles, and intercellular interactions[10]. Previous studies have reported that ovarian and hepatocellular tumor cells can be categorized into distinct metabolic clusters to predict the patient's treatment response and overall prognosis[11, 12]. However, the mechanisms of glioma cell metabolism remain poorly understood, and metabolic classifiers for gliomas are still lacking. Immunotherapy has become a crucial component of cancer treatment, making it increasingly important to understand the metabolic heterogeneity of infiltrating immune cells and tumor cells in depth. An increasing numbers of studies have demonstrated that tumor metabolism not only plays a vital role in promoting the development and survival of tumor cells, but can also modulate anti-tumor immune responses by releasing metabolites that influence immune molecule expression, such as lactate, PGE2, and arginine.[13, 14]. In glioblastoma, Professor Veglia reveals that PERK-driven glucose metabolism in monocyte-derived macrophages (MDM) promotes MDM immunosuppressive activity via histone lactylation, indicating that immune cell metabolism modulates their anti-tumor activity[15]. Additionally, tumor cells can inhibit anti-tumor immune responses by competing for and consuming vital nutrients, thereby impairing the metabolic fitness of tumor-infiltrating immune cells[16]. Understanding the metabolic heterogeneity and characteristics of immune cells in the tumor microenvironment facilitates the identification of therapeutic targets within metabolic pathways. This approach holds promise for enhancing cancer immunotherapies. This study established a glioma energy metabolism-based classifier based on bulk RNA-seq, and validated it using multi-omics data, including genomics, bulk and single-cell transcriptomics, and metabolomics. Single-cell analysis characterized metabolic reprogramming during the PN-to-MES transition in glioma cells. Additionally, it elucidated the metabolic preferences of various immune cell types. Overall, by leveraging multi-omics data, this study uncovered glioma metabolic heterogeneity, providing novel insights and potential therapeutic targets. Methods 1. Data accessibility This study utilized several glioma datasets, including the TCGA dataset, the CGGA dataset, the CPTAC dataset[17], the IvyGAP dataset[18], and the GSE182109 dataset[19]. These datasets contain genomic data, bulk RNA-seq, single-cell or single-nucleus RNA-seq, and metabolome data. The GSE182109 dataset and IvyGAP datasets provide transcriptome data from different anatomical regions within the same patients. In the CPTAC database, 16 patients were subjected to both bulk RNA-seq and single-nucleus RNA seq. The sequencing data are available for download from the following websites (https://portal.gdc.cancer.gov/, https://glioblastoma.alleninstitute.org, https://www.ncbi.nlm.nih.gov/geo/). 2. Construction of energy metabolic classifiers Based on previous studies[20, 21] and MSigDB[22], we identified four critical metabolic pathways in gliomas: glycolysis, the pentose phosphate pathway, fatty acid oxidation, and glutaminolysis. We identified key genes associated with these metabolic pathways (Table S1). To assess the activity of these pathways in individual patient, we utilized the ssGSEA algorithm (GSVA package in R)[23]. Unsupervised classification is a key machine learning approach. In this study, we used the K-means method to ascertain the optimal classification based on glioma metabolic profiles. Prior to clustering, samples were scaled to group them based on metabolic pathway patterns. We then reordered the samples based on metabolic clusters and the scaled ssGSEA values for heatmap visualization (using the Pheatmap package in R). Additionally, we applied the consensus clustering algorithm (ConsensusClusterPlus package in R) with 1000 iterations and 80% resampling to confirm the number of clusters (K). Principal component analysis (PCA) was also performed to highlight significant differences between clusters. 3. Calculation of immune signature scores To evaluate cell proliferation, we utilized 31 validated genes associated with cell cycle progression (CCP)[24]. Immune infiltration levels were assessed by several immune signatures related to tumor-infiltrating lymphocytes (TILs), immune cytolytic activity (CYT), and interferon (IFN) responses[24]. ssGSEA scores were calculated to quantify the abundance of these gene sets. Tumor immune scores were estimated using the ESTIMATE algorithm[25]. The Microenvironment Cell Populations-Counter (MCP-counter) algorithm was used to quantify the populations of eight immune cell types and two stromal cell types[26]. The cluster score was computed by averaging the expression levels of all pathways within each cluster and subtracting the mean expression of four metabolic pathways. During this calculation, the values for each metabolic pathway were scaled across all samples. For scRNA-seq data from the CPTAC and GSE182109 datasets, tumor cells were classified into metabolic cluster 1 if their metabolic cluster 1 score exceeded that of metabolic cluster 2, and vice versa. 4. Preprocessing of scRNA-seq data The preprocessing of the raw data for the CPTAC and GSE182109 datasets was described in detail in the publications[17, 19]. The Seurat package v5.0[1] was used for downstream scRNA analysis. A series of quality filters were performed to remove the low-quality cells: too few total transcript counts (<300); possible debris with too few genes detected (<200) and too few UMIs (10%). Each sample was normalized and scaled by ‘SCTransform’ function in the Seurat package to correct for batch effects (parameters: vars.to.regress = c (“nCount_RNA”, “percent.mito”), variable.features.n = 3000). The Harmony algorithm was applied to integrate all samples for downstream analysis[27]. To reduce dimensions, principal component analysis (PCA) was initially conducted with 30 principal components (npcs =30), followed by t-distributed stochastic neighbor embedding (t-SNE) using the first 20 dimensions (dims = 1:20). The Wilcoxon rank-sum test was used to identify the specific genes of each cluster through the FindAllMarkers function (logfc.threshold = 0.25). 5. Cell annotation Using singleR[28], cellmarker[29], panglaodb[30] and marker genes reported in the literature[17, 31], all cell types were annotated through a combination of automated and manual methods. 6. Metabolic activity evaluation The metabolic pathway activity of annotated cells was calculated as described in the previous article[32]. ssGSEA values were used to compare metabolic profiles across various cell types. Metabolic pathways with a p-value < 0.05 in GSEA were considered statistically significant. The OXPHOS gene set was extracted from the KEGG database. The gene sets of hypoxia and angiogenesis were retrieved from the MSigDB library. 7. Inference of CNV To differentiate malignant cells from non-malignant cells, copy number variation (CNV) was calculated using the R package inferCNV (https://github.com/broadinstitute/inferCNV). Genes expressed in fewer than 10 cells and with a median expression below 0.1 were excluded from the analysis to focus on robustly expressed genes. Subsequently, genes were annotated based on their chromosomal positions, and CNV scores were calculated using moving averages across sets of 100 genes. Hierarchical clustering was applied to separate non-malignant cells from malignant cells based on distinct chromosomal deletions or amplifications. 8. Trajectory analysis Trajectory analysis was facilitated by importing data from Seurat into the Monocle2 package in R[33] . Initially, genes with expression levels below 0.5 and cells expressing fewer than 200 genes were excluded from the dataset. The remaining data were reordered based on significantly differentially expressed genes identified with a q-value threshold of 1e-40. Subsequently, a DDRTree dimensional reduction algorithm was applied to interpret the data structure. Pseudotime analysis was then performed to map the temporal progression of cellular differentiation, pinpointing pathways that significantly changed along trajectories. Cells within the same branch were considered to reside at similar differentiation stages. 9. Glioma cell line and cell culture The glioma cell lines of U87, LN229 and GL261 were purchased from ATCC and authenticated by STR assay. The glioma cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% antibiotics in a humidified incubator maintained at 37°C with 5% CO₂. For transient knockdown experiments, siRNA was transfected into glioma cells using lipofectamine 8000 (Beyotime Biotechnology, China), followed by further experiments. The plasmids were obtained from Nanjing Corues Biotechnology Co, Ltd. The detail information of plasmids is listed in Table S2. The total RNA and protein were extracted 48 hours and 72 hours after transfection respectively. 10. Real-time quantitative PCR (RT-qPCR) analysis Total RNA was extracted from glioma cell lines using TRIzol reagent, and reverse transcription was carried out using Roche reagent. cDNA was amplified by RT-PCR using Roche reagents. Genes expression levels were normalized to GAPDH expression. Primers used for target genes amplification are listed in Table S3. 11. Western blot Total protein was extracted from glioma cells using pre-chilled RIPA buffer (Beyotime Biotechnology, China) containing a mixture of protease and phosphatase inhibitors, and protein concentration was quantified using a BCA protein assay kit (BL1784B, biosharp). Proteins were separated by gel electrophoresis and transferred onto a polyvinylidene fluoride (PVDF) membrane (Merck, Germany). The membrane was blocked in 5% non-fat milk for 1 hour and incubated with primary antibodies against OSMR (10982-1-AP, Proteintech), HK-2(22029-1-AP, Proteintech), LDHA (sc-137244, Santa Cruz), JAK1 (66466-1-lg, Proteintech), p-JAK1 (ab138005, abcam), STAT3 (ab68152, abcam), p-STAT3 (ab267373, abcam), β-actin (66009-1-Ig, Proteintech) overnight at 4°C. Following three washes with 1×TBST, the PVDF membrane was incubated with a horseradish peroxidase (HRP)-conjugated secondary antibody (Sangon Biotech, China) at room temperature for 1 hour. The membrane was washed three more times with 1×TBST, followed by exposure and analysis. 12. Cell migration analysis Cell migration was assessed using wound healing assay. For the wound healing assay, treated cells were evenly seeded into a 6-well plate and allowed to reach full confluence. A scratch was introduced in the cell monolayer, and the cells were then cultured in serum-free medium. Cell migration was monitored under a microscope, and images were captured and analyzed at 0, 24, and 48 hours. 13. Cell invasion analysis Cell invasion ability was evaluated using Transwell chambers (8-μm pore size, Corning, USA) coated with Matrigel (BD Biosciences, USA). Briefly, 1×10 4 cells in 200 μL serum-free medium were seeded into the upper chamber, while 500 μL complete medium containing 10% fetal bovine serum (FBS) was added to the lower chamber as a chemoattractant. After incubation for 48 h at 37°C in a humidified atmosphere with 5% CO 2 , non-invaded cells on the upper surface of the membrane were carefully removed with a cotton swab. Invaded cells on the lower surface of the membrane were fixed with 4% paraformaldehyde for 15 min, stained with 0.1% crystal violet for 20 min, and washed with PBS. The number of invaded cells was counted under an inverted microscope in five randomly selected fields at 200× magnification. 14. tumor xenograft model Female C57BL/6J mice (4 weeks old, body weight 19-21g) were purchased from GemPharmatech Co, Ltd. (Jang Su, China) and housed in laminar airflow cabinets under specific pathogen-free (SPF) conditions. For orthotopic tumor implantation, 1×10⁵ GL261 cells stably transfected with sh-OSMR or sh-NC were stereotactically injected into the brain of each mouse. Intracranial tumor growth was evaluated by bioluminescence imaging. All animal studies were approved by the Institutional Animal Care and Use Committee of the First Affiliated Hospital of the University of Science and Technology of China. 15. Statistical analysis The mean (standard deviation) was used for continuous variables with normal distribution, and the median and interquartile range (IQR) were used for other data. When comparing two groups of continuous variables, the t-test or the Mann–Whitney rank-sum test was used based on the distribution of data. The chi-square test was used for categorical variables. A P -value < 0.05 was considered statistically significant. The Kaplan-Meier survival curve was created using GraphPad Prism 9 software. A P -value < 0.05 from the log-rank test was considered indicative of a significant difference in survival times. Results Construction of energy metabolic classifier based bulk RNA-seq A flowchart was created to systematically illustrate the process of establishing the energy metabolic classifier (Figure S1). Based on previous studies of tumor metabolism, we identified four metabolic pathways: glycolysis, PPP, FAO, and glutaminolysis to describe the metabolic status of tumors (Figure 1a). The ssGSEA algorithm was used to evaluate the activity of those metabolic pathways, and values were scaled using ‘scale’ function in RStudio. The heatmap revealed that in the TCGA dataset, patients in metabolic cluster 1 exhibited higher activity in the glycolysis, PPP, and FAO pathways, while patients in cluster 2 showed increased activity in the glutamine metabolism pathway (Figure 1b). This same metabolic pattern was also observed in the CGGA dataset (Figure 1c). The heatmaps demonstrated the robustness of consensus clustering (K=2) compared to other numbers of clusters in both the TCGA and CGGA datasets, based on the K-means unsupervised classification (Figure 1d, Figure 1e, Figure S2a-S2d). In terms of tumor grade, patients in metabolic cluster 1 had a higher proportion of high-grade gliomas in the TCGA and CGGA datasets (Figure 1f and 1g). Univariate Cox analysis revealed that patients in cluster 1 had a significantly worse prognosis (Figure 1h, the TCGA dataset: P <0.001; Figure 1i, the CGGA dataset: P <0.001). Based on these results, two metabolic clusters were identified from the bulk transcriptome of gliomas: (1) dependent on glycolysis, PPP, and FAO; (2) dependent on glutaminolysis. Principal component analysis showed significant heterogeneity between the two clusters using this metabolic classifier, and the classification into two metabolic clusters was demonstrated to be robust (Figure S2e and S2f). Moreover, there were significant differences in the mRNA expression of metabolism-related genes between the two metabolic clusters (Figure S2g). Clinical, genomic, and immune characteristics associated with metabolic classifier To further analyze the clinical characteristics associated with the energy metabolic classifier, this study included several important clinical characteristics from the TCGA dataset (Figure 2a). It is obvious that the age of patients in cluster 1 is significantly higher than that in cluster 2. In terms of molecular pathology, cluster 1 had a significantly higher proportion of IDH wild-type patients compared to cluster 2, and patients in cluster 2 have a higher proportion of MGMT methylation (Figure 2a). Multivariate survival analysis revealed that the metabolic classifier was a significant risk factor in both the TCGA dataset (HR=1.85[1.09, 3.13], P =0.025) (Figure S3a). Similar results were observed in the CGGA dataset (HR=1.67[1.30, 2.13], P <0.001) (Figure S3b). Oncogene mutation events can stimulate cell-autonomous metabolic reprogramming. In this study, we found that several metabolic pathway-related genes were significantly amplified or deleted based on the CNV data (Figure 2b), and the mRNA expression levels of these genes in the clusters were consistent with the amplification or deletion event in the genomic data (Figure S4a and S4b). In addition, the overall copy number variation events in cluster 1 were significantly more frequent than those in cluster 2 (Figure 2b). Considering the significant difference in overall survival between patients in the two clusters, this study explored the classical oncogenic pathways in the two clusters. Based on previous literature reports, we identified ten classic tumor core pathways and quantified pathway activity using the GSVA algorithm[34]. In cluster 1, the activities of the cell cycle, Hippo, TGF-β, NRF2, and NOTCH were significantly higher than those in cluster 2, while the activities of PI3K, RAS, and WNT pathways were higher in patients in cluster 2 (Figure 2c). Similar results were observed in the CGGA dataset (Figure S5a). Metabolic reprogramming has been shown to significantly influence tumor immunity. In the TCGA dataset, multiple targetable immune checkpoint genes were highly expressed in cluster 1 (Figure 2d). A similar tendency was observed in the CGGA dataset (Figure S5b). Several immune signatures, such as TILs, CYT, and IFN response, were significantly elevated in cluster 1 (Figure 2e), and a similar tendency was also found in the CGGA dataset (Figure S5c). Furthermore, the ESTIMATE algorithm[25] revealed significantly higher immune ( P <0.001) and stromal scores ( P <0.001) in metabolic cluster 1compared to metabolic cluster 2 (Figure 2f). This result was further validated in the CGGA dataset (Figure S5d). To further explore the relationships between tumor metabolism and immunity, we assessed the abundance of all ten immune and stromal cell types using the MCP-counter algorithm. In both the TCGA and CGGA datasets, metabolic cluster 1 exhibited significantly higher abundances of CTLs, CD8+ T cells, monocytes, and myeloid cells compared to cluster 2 (Figures S6a and S6b). Overall, this study demonstrated significant differences between the two metabolic clusters in clinical characteristics, genomic mutations, molecular pathology, and tumor immunity, highlighting intrinsic heterogeneity between the two patient groups. Multi-omics data further validated the energy metabolic classifier The robustness of the metabolic classifier was further validated using the CPTAC dataset. Patients were assigned to corresponding metabolic clusters based on the established classifier. The heatmap revealed consistent metabolic differences in the CPTAC dataset compared to the TCGA and CGGA datasets (Figure 3a). Metabolic cluster 1 exhibited higher activities in glycolysis, PPP, and FAO pathways, whereas cluster 2 showed elevated activity in glutamine metabolism (Figure 3a). The heatmap of K-means unsupervised classification also suggested robust classification (K=2) (Figure 3b, Figure S7a and S7b). PCA revealed significant heterogeneity between the two clusters (Figure S7c). Although no significant difference in overall survival was observed between the two clusters, patients in metabolic cluster 1 tended to have a worse prognosis compared to those in cluster 2 ( P =0.259) (Figure S7d). In addition to bulk RNA-seq, metabolite levels were also quantified in the CPTAC dataset. The levels of lactate, alanine, and palmitic acid in metabolic cluster 1 were significantly higher than those in metabolic cluster 2 (Figure 3c). The levels of glycerol-3-phosphate and glutamate in metabolic cluster 2 were significantly higher than those in metabolic cluster 1 (Figure 3c). The observed metabolite levels difference aligned with the metabolic pathway activity difference. Furthermore, the Hippo, TP53, PI3K, and RAS pathways exhibited consistent differences with those observed in the TCGA and CGGA datasets (Figure 3d). In terms of the immune profile of the tumor microenvironment, metabolic cluster 1 exhibited higher mRNA levels of immunosuppressive checkpoint genes (Figure 3e). Meanwhile, several immune features, including IFN, TILs and CYT, were significantly elevated in metabolic cluster 1 compared to cluster 2 (Figure 3f). Moreover, the ESTIMATE algorithm revealed significantly higher immune and stromal scores in metabolic cluster 2 compared to cluster 2 (Figure 3g). The MCP algorithm indicated significantly higher abundance of monocytes, myeloid cells, and neutrophils in metabolic cluster 1 (Figure S7e), consistent with the findings in the TCGA and CGGA datasets. Further validation using multi-omics data from the CPTAC dataset, confirmed the robustness of the energy metabolic classifier, which accurately reflects glioma metabolic heterogeneity and highlights the significant impact of metabolic reprogramming on tumor immunity. The role of the OSM/OSMR in glioma metabolism To further explore whether a specific transcriptomic program is associated with metabolic phenotypes in metabolic cluster 1, differential gene analysis was performed between the two metabolic clusters, and results were visualized using a volcano plot (Figure 4a). To identify the top upregulated genes in metabolic cluster 1, the top 100 highly expressed genes from the TCGA, CGGA, and CPTAC datasets were intersected, identifying OSMR as a key gene (Figure 4b). OSMR, the receptor for OSM (Oncostatin M), is primarily secreted by macrophages and T cells and belongs to the interleukin-6 cytokine family. It has been reported that OSM/OSMR play a key role in the proliferation and differentiation of various tumors. In gliomas, it has been reported that OSM acts through OSMR to enhance glioma cell migration. Additionally, there are report suggesting that M2-type macrophages secrete OSM, which acts on OSMR to facilitate the proneural-to-mesenchymal transition (PMT) in glioblastoma. We found that OSMR gene was highly expressed in mesenchymal subtype and high-grade gliomas in the TCGA cohort (Figure 4c and 4d). Similar trends were also observed in the CGGA cohort (Figure S8a and S8b). However, the role of OSMR in glioma metabolism remains poorly understood. Survival analysis revealed that high OSMR expression is significantly associated with worse overall survival in glioma patients in the TCGA dataset (LGGs: P =0.0013, GBMs: P =0.0097) (Figure 4e and 4f). In CGGA datasets, high expression of OSMR showed significant worse overall survival in LGGs ( P <0.0001) (Figure S8c). For GBM, although high OSMR expression is not significantly associated with overall survival, there is still a tendency in that direction ( P =0.137) (Figure S8d). We further explored the correlation between OSMR and glycolysis-related genes. Correlation analysis revealed significant positive associations between OSMR and both mRNA and protein levels of hexokinase 2 (HK2) and lactate dehydrogenase A (LDHA) in the CPTAC dataset (mRNA level: Pearson’s r =0.60, P <0.001, Pearson’s r =0.60, P <0.001; protein level: Pearson’s r =0.54, P <0.001, Pearson’s r =0.29, P =0.002) (Figure 4g and 4h; Figure 4I and 4J). In the TCGA dataset, OSMR also showed strong positive correlation with HK2 (Pearson’s r =0.58, P <0.001) and LDHA (Pearson’s r =0.71, P <0.001) at mRNA level (Figure S8e and S8f). According to the published literature[35, 36], OSM was added to the culture medium at a concentration of 10 ng/ml for the subsequent experiments. Additionally, siRNA was used to knocked down OSMR in glioma cell lines to explore the biological function of OSMR (Figure S8g) and observed a significant decrease in the invasion and migration ability of the U87 cell line (Figure 4k-4n). In terms of metabolism, OSMR knockdown significantly reduced the protein levels of HK2 and LDHA through western blotting in U87 cell line (Figure 4o). As key enzymes in the aerobic glycolytic pathway, HK2 and LDHA suggest that OSM/OSMR plays a critical role in aerobic glycolytic metabolic reprogramming in glioma. The OSM/OSMR/JAK1/STAT3 axis drives malignant progression of glioma Pathway enrichment analysis revealed significant activation of the JAK/STAT pathway in glioma patients with high OSMR expression (Figure 5a). Given JAK1’s predominant expression in the nervous system and previous reports on STAT3 with OSMR, we evaluated its phosphorylation status and that of downstream STAT3 via Western blotting. OSM treatment significantly upregulated p-JAK1 (Tyr1034/1035) and p-STAT3 (Tyr705) levels, indicating pathway activation, whereas pre-treatment with a JAK-STAT3 pathway inhibitor (Ganoderic acid A (GA-A), 1.5 uM) abolished this effect, confirming JAK1-STAT3 pathway dependency (Figure 5b). To functionally validate the axis, we performed rescue experiments in U87 cells (representing metabolic cluster 1). GA-A significantly attenuated OSM-induced cell invasion ( P <0.001) (Figure 5c and 5d) and migration ( P =0.001) in U87 cell line (Figure 5e and 5f), phenocopying OSMR knockdown effects. Complementary gain-of-function studies were conducted in LN229 cells (metabolic cluster 2) stably overexpressing OSMR, which exhibited similar results with U87 cell line (Figure 5g-5j). In vivo validation utilized a syngeneic orthotopic model, where 1×10⁵ luciferase-labeled GL261 cells (sh-OSMR vs. sh-NC) were stereotactically injected into the brain of C57BL/6J mice (n=5/group). Bioluminescence imaging revealed a significant reduction in tumor burden ( P =0.0039) (Figure 5l and 5m) and prolonged median survival in sh-OSMR group ( P =0.0011) (Figure 5m). Collectively, these results establish OSMR as a critical mediator of glioma progression through JAK1/STAT3-dependent metabolic reprogramming, highlighting its potential as a therapeutic target. Metabolic profiles of malignant cells at single-cell level This study aimed to explore metabolism at single-cell resolution using single cell RNA sequencing to better understand metabolic heterogeneity in the glioma microenvironment. Single-nucleus sequencing and bulk RNA sequencing were performed on 16 patients from the CPTAC dataset. The snRNA-seq data from 16 patients were quality-checked and integrated using the Harmony algorithm. Subsequent dimensionality reduction and clustering analyses were performed. The t-SNE projection showed excellent integration performance (Figure S9a). As described in the methods, cells were annotated both automatically and manually, including tumor cells, immune cells, and brain-resident cells. The t-SNE visualization displayed the annotated cell types (Figure 6a). A dot plot further illustrated the expression of cell type-specific marker genes (Figure 6b). The inferCNV algorithm was used to identify malignant or non-malignant cells, confirming accurate annotation of glioma cells (Figure S9b). For these 16 patients, they were assigned to metabolic clusters based on bulk RNA-seq and malignant cells from snRNA-seq data using the metabolic classifier (Figure 6c and 6d). Notably, classifications based on bulk RNA-seq and snRNA-seq consistently grouped patients into the same metabolic clusters. The metabolic score accurately reflected the energy metabolic activity of each patient (Figure 6e). As shown in Figure 6f, the metabolic classification of most tumor cells was consistent with the metabolic clusters in the bulk RNA-seq data. Based on the above results, tumor metabolic heterogeneity is primarily driven by malignant cells. In addition, we sought to investigate the impact of the tumor microenvironment on tumor metabolism. Given that environmental conditions cannot be directly integrated into the metabolic analyses, hypoxia and angiogenesis signatures were used as proxies for oxygen and nutrients supply within tumor microenvironment. Consistent with published research, the glycolysis pathway exhibits a significant positive correlation with hypoxia and angiogenesis (hypoxia, Pearson’s r =0.67, P <0.001; angiogenesis, Pearson’s r =0.27, P <0.001) (Figure 6g). The fatty acid oxidation pathway is also positively correlated with hypoxia and angiogenesis (hypoxia, Pearson’s r =0.22, P <0.001; angiogenesis, Pearson’s r =0.20, P <0.001) (Figure 6g). Glycolysis, free fatty acid oxidation, and glutamine metabolism were all significantly positively correlated with oxidative phosphorylation (glycolysis, Pearson’s r =0.43, P <0.001; fatty acids, Pearson’s r =0.36, P <0.001, glutamine metabolism, Pearson’s r =0.34, P <0.001) (Figure S10a). Glioma cell lines from the CCLE database were used to explore the metabolism of glioma cells without nonmalignant cells in vitro. Although the limited number of glioma cell lines may introduce some statistical variability, intriguingly, certain metabolic patterns in these cell lines contrast with those observed in tumor tissues. Fatty acid oxidation tends to show a negatively correlation with hypoxia, though this did not reach statistical significance (Pearson’s r =-0.22, P =0.42) (Figure 6h, top). Glutamine metabolism was significantly negatively correlated with angiogenesis (Pearson’s r =-0.54, P =0.037) (Figure 6h, bottom), and also had a negative correlation trend with hypoxia (Pearson’s r =-0.46, P =0.082) (Figure S10b). This result suggests that nonmalignant cells in the tumor microenvironment may have a certain impact on the metabolism of tumor cells. In summary, tumor metabolic heterogeneity is primarily driven by malignant cells, our results underscore the significant roles of cell-cell interactions and the tumor microenvironment in shaping tumor metabolism. Metabolic characteristics of malignant and nonmalignant cells in tumor microenvironment To further elucidate the metabolic characteristics of different cell types within the tumor microenvironment, we assess the metabolic activities of five annotated cell types based on prior studies. Tumor cells exhibited the highest metabolic activity (Figure 7a), and demonstrated elevated across the majority of metabolic pathways (Figure 7b). In contrast, tumor-infiltrating lymphocytes had the lowest metabolic activity (Figure 7a). To further investigate the relationship between the metabolic classifier and cell properties, we compared the metabolic pathway activities of malignant and nonmalignant cells. Among malignant cells, tumor cells from patients in metabolic cluster 1 exhibited significantly higher cluster 1 scores ( P <0.001) (Figure 7c, right), while those in metabolic cluster 2 showed higher cluster 2 scores ( P <0.001) (Figure 7c, left). In contrast, nonmalignant cells displayed no statistical differences in cluster 1 scores or cluster 2 scores between metabolic clusters ( P =0.067 (left), P =0.078 (right)) (Figure 7d). The findings further demonstrate that the metabolic heterogeneity of tumor tissue is primarily driven by malignant cells, and our metabolic classifier effectively captures this heterogeneity at single-cell resolution. We further explored the relationship between tumor metabolic heterogeneity and spatial location using the IvyGAP and GSE182109 datasets. Samples were divided into metabolic cluster 1 and metabolic cluster 2 based on the energy-related metabolism classifier. In the IvyGAP dataset, tumor tissue in the marginal regions, such as infiltrating and leading edge areas, exhibited higher cluster 1 score. Conversely, cellular tumors associated with hyperplastic blood vessels and microvascular proliferation were more likely to exhibit metabolic cluster 2 status. In contrast, cellular tumors in regions with pseudopalisading cells around necrosis and the perinecrotic zone tended to display metabolic cluster 1 status (Figure 7e). In the GSE182109 dataset, samples were collected from multiple regions, including the core area, the MRI-enhancing area, and the marginal area, for scRNA-seq analysis (Figure7 f). Based on the cell annotations provided in the original study, we calculated metabolic cluster scores for malignant cells using our classifier. Although statistical testing was not feasible due to the limited number of samples from the marginal areas (n=2), we observed that the proportion of metabolic cluster 1 malignant cells was significantly lower in the tumor edge area compared to the core and MRI-enhancing areas (Figure 7f). These results suggest that malignant cells are more likely to exhibit the metabolic cluster 1 phenotype in regions closer to the tumor core. Collectively, these findings highlight the spatial heterogeneity of glioma metabolism and demonstrate that our metabolic cluster accurately delineates the spatial metabolic characteristics of malignant cells at the single-cell resolution. This underscores the utility of our classifier in capturing both inter- and intra-tumoral embolic diversity. Based on the intrinsic transcriptional characteristics of tumor cells and previous literature, glioma cells are typically classified into four subtypes: OPC-like, NPC-like, AC-like, and MES-like cells[2]. Trajectory analysis revealed a prominent transition starting from OPC- and NPC-like malignant cells, which correspond to the proneural (PN) subtype of GBM as classified by the TCGA dataset, to MES-like cells. This finding aligns well with previous findings (Figure 7g and 7h). Additionally, glycolytic activity in glioma cells was observed to increase progressively during the PN-to-MES transition (Figure 7i). Furthermore, hypoxia and angiogenesis within the tumor microenvironment were also found to be associated with the malignant transformation of tumor cells (Figure S11a and S11b), suggesting that tumor metabolism plays a critical role in driving the mesenchymal transformation of glioma cells. Based on these findings, we propose that metabolic reprogramming in tumor cells, coupled with in influence of the tumor microenvironment, significantly contributes to the malignant transformation of glioma cells. Metabolic characteristics of immune cells at single-cell level In this study, we further explored the metabolic profiles of diverse immune cell populations. Utilizing singleR and leveraging insights from previous literature, we performed further dimensionality reduction and clustering analysis on myeloid cells (Figure 8a) and lymphocytes (Figure 8b). The marker genes of various immune cell subtypes were found to be specifically and highly expressed in their corresponding cell types (Figure 8c and 8d). Based on the cell annotation, this study revealed distinct metabolic characteristics across various immune cells types. Among myeloid cells, monocytes and macrophages exhibited significantly higher glycolytic activity compared to microglia (monocytes vs microglia: P =0.002, microglia vs macrophages: P =0.03 (top)) (Figure 8e). Neutrophils and DC displayed relatively high glycolytic activity but low oxidative phosphorylation activity (Figure 8e). Among lymphocytes, T regulatory cells (Tregs) demonstrated higher glycolytic activity than CD4+ T cells (Figure 8f, P =0.034). While CD8+ T cells showed significantly higher oxidative phosphorylation activity compared to CD4+ T cells ( P =0.002) (Figure 8f). Additionally, plasma cells exhibited markedly higher oxidative phosphorylation and fatty acid oxidation activities than other lymphocyte subtypes (Figure 8f). To further investigate the metabolic profiles of T cells, we conducted metabolic enrichment analysis of CD4+ and CD8+ T cells. CD4+ T cells displayed significantly elevated activity in lysine degradation, sphingolipid metabolism, glyceride metabolism, and glycerophospholipid metabolism (Figure 8g). In contrast, CD8+ T cells exhibited higher activity in oxidative phosphorylation and cytochrome P450 metabolism. Overall, these findings highlight that different immune cell types exhibit distinct metabolic preferences (Figure 8g). Discussion The metabolic heterogeneity of glioma is likely driven by multiple factors. Advances in sequencing technology have enabled the generation of large-scale genomic, single-cell, transcriptomic, and metabolic sequencing data for gliomas, providing an unprecedented opportunity to explore the complexity of glioma metabolism. In this study, the bulk-seq sequencing data of gliomas in the TCGA dataset were used to construct an energy-related metabolic classifier based on the activity of core metabolic pathways. Patients were divided into two prognostic clusters: those in cluster 1 were exhibited high pathway activity in glycolysis, the pentose phosphate pathway (PPP), and fatty acid oxidation (FAO), whereas those in cluster 2 displayed elevated activity in glutaminolysis. The robustness of the metabolic classifier was externally validated using the CGGA dataset. Additionally, the CPTAC dataset was employed to verify the rationality and robustness of the metabolic classifier across bulk RNA-seq, scRNA-seq, and tissue metabolite levels. Furthermore, we evaluated the metabolic classifier-related differences among the genome, clinical characteristics, molecular pathological composition, and immune infiltration. Our findings revealed that metabolic heterogeneity is primarily driven by tumor cells, although the tumor microenvironment also significantly influences the metabolic reprogramming of glioma cells. Collectively, this study enhances our understanding of metabolic heterogeneity in gliomas and provides a foundation for developing personalized therapeutic strategies for glioma patients. Increasingly, it is becoming clear that glioma cells have metabolic heterogeneity and preferences[ 37 ]. Understanding the metabolic reprogramming and progression of glioma cells is crucial for targeting glioma metabolism for treatment. For instance, Professor Evangelista's team classified pancreatic ductal adenocarcinoma into three metabolic subtypes based on tumor metabolites and demonstrated that the glycolytic subtype was strongly associated with the mesenchymalization of tumor cells[ 38 ]. Similarly, Professor Schaeffer and colleagues identified metabolic heterogeneity in high-grade ovarian cancer, primarily driven by variation in oxidative phosphorylation activity. Compared with previous studies, our research provides several novel insights: 1. By integrating multi-omics data (genomics, transcriptomics, and metabolomics), we established an energy-related metabolic classifier for gliomas using the TCGA dataset and validated its robustness across multiple external datasets. 2. Clinical characteristics, classic oncogenic pathways, and immune profiles between metabolic clusters were deeply explored. 3. We investigated the underlying causes of metabolic heterogeneity in gliomas at single-cell resolution and elucidated the relationship between glioma cells progression and metabolic reprogramming. 4. We further characterized the metabolic status of diverse immune cell populations at single-cell resolution. Genome instability and mutations have long been recognized as fundamental hallmarks of the cancer[ 39 ]. The Cancer Genome Atlas (TCGA) research network has constructed a comprehensive catalog of genomic alterations driving tumorigenesis[ 3 , 40 ]. Previous studies have further demonstrated that mutations in metabolism-related genes constitute a core component of metabolic reprogramming across various cancer types[ 8 , 41 ]. The hypoxia conditions prevalent in glioma core regions induce significant adaptive changes in tumor cells. Consistent with published literature, hypoxia has been shown to promote genomic instability and epigenetic modifications, ultimately contributing to tumor progression, malignant transformation, and chemoresistance[ 42 , 43 ]. Building upon these findings, our study revealed that patients in metabolic cluster 1 exhibited a higher mutation burden, and genes involved in metabolic pathways such as glycolysis underwent significantly more genomic mutations in the TCGA dataset. These observations suggest that metabolic gene mutations may represent an intrinsic driver of metabolic heterogeneity in gliomas. Further exploration identified OSMR as the top upregulated gene in cluster 1 through differential gene analysis. While previous studies have reported that the OSMR would promote the proliferation and the differential of the certain tumors. As for glioblastomas, macrophage-secreted OSM acts on OSMR in glioblastoma cells to promote proneural–mesenchymal transformation[ 36 , 44 ]. This study found that a decreased level of OSMR in the presence of OSM significantly reduced the expression of HK2 and LDHA, key genes in the aerobic glycolytic pathway, thereby affecting glycolytic activity. Furthermore, this study demonstrated that OSM/OSMR axis promotes malignant progression of glioma cells via JAK1/STAT3 signaling pathway. This finding is consistent with the prior reports showing that OSM binding to OSMR activates the JAK/STAT pathway, which plays a critical role in regulating glycolysis[ 45 , 46 ]. In orthotopic xenograft mouse models, knockdown of OSMR significantly prolonged survival compared to sh-NC group. Notably, patients in metabolic cluster 1 exhibited distinct immunosuppression in their immune profiles, indirectly suggesting that OSMR may reshape the immune microenvironment through metabolic reprogramming. The specific mechanisms by which OSMR mediates reprogramming and modulates the immune microenvironment in warrant further investigation. Metabolism reprogramming, characterized by the adaptation of energy metabolism to support rapid cell growth and proliferation, has emerged as a critical hallmark of cancer[ 4 ]. In 1964, Professor Warburg first reported that tumor cells use glycolysis for energy under aerobic conditions (the Warburg effect)[ 47 ]. The heterogeneity of tumor cells leads to a complex metabolic pattern. In fact, PPP, FAO, and glutaminolysis pathways are also used to fuel proliferation[ 48 , 49 ]. Recent researches have indicated that the activation of oncogenic pathways can lead to upregulation of metabolic pathways[ 8 , 50 ]. For instance, under nutrient-deprived conditions (e.g., limited glucose, or glutamine availability), tumor cells activate the c-Myc pathway to modulate key metabolic enzymes (PHGDH, PSAT1 and PSPH) in the serine synthesis pathway, thereby maintaining cellular survival[ 51 ]. In our study, several classical oncogenic pathways were significantly activated in patients in cluster 1, such as pathways related to glycolysis metabolism. For example, Hippo and c-Myc exhibit significantly higher activity in cluster 1. We extended this study by leveraging single-cell transcriptomics from the CPTAC and GSE182109 datasets. First, we found that tumor metabolic heterogeneity was primarily driven by malignant cells, with trajectory analysis showing a gradual increase in glycolytic activity during the transition from OPC-like to MES-like phenotypes. Second, analysis of the IvyGAP and GSE182109 datasets revealed an increasing prevalence of metabolic cluster 1 status in tumor cells located nearer to the tumor core. To further explore tumor microenvironmental influences on metabolism, we assessed the relationships between metabolic activity and microenvironment factors using hypoxia and angiogenesis signatures as proxies for oxygen and nutrient availability. Comparison with CCLE glioma cell line data revealed significant differences between in vivo and in vitro metabolic profiles of glioma cells. We proposed several possible reasons behind this observed phenomenon: (1) A comprehensive understanding of tumor cell biology in vivo is required. The metabolic vulnerabilities of tumor cells in vivo cannot be fully replicated in cultured cell models. Tumor cells rely significantly on a rapid influx of nutrients, which can vary based on their location within the glioma microenvironment. Therefore, it is necessary to establish a model system recapitulating tumor microenvironments to explore tumor metabolism. (2) A deeper understanding of the impact of cellular interactions in the tumor microenvironment on tumor growth and progression is crucial. There is increasing evidence that immune responses are associated with changes in tissue metabolism, including nutrient consumption, increased oxygen consumption, and the production of reactive nitrogen and oxygen intermediates[ 52 – 54 ]. In this study, the level of lactate in cluster 1 was significantly higher than that in cluster 2. Correspondingly, patients in cluster 1 expressed greater immune signatures and immune checkpoint genes. In a phase 3 clinical trial, the PD-1 inhibitor Nivolumab demonstrated no significant increase in side effects when combined with radiotherapy and Temozolomide. While it showed some anti-tumor activity, further studies are necessary to determine its clinical benefits[ 55 ]. Furthermore, this study found that patients in cluster 1 had a higher proportion of IDH wild-type and mesenchymal patients, indicating more severe immunosuppression in their tumor microenvironment, which suggests that these patients may benefit more from combined immunotherapy. Similar to tumor cells, immune cells demonstrate distinct metabolic heterogeneity. Under homeostatic conditions, immune cells maintain a quiescent state. However, during pathogenic challenges (e.g., infections, inflammation) or tumor development, immune cells undergo rapid activation and functional adaptation[ 56 ]. T cells exhibit dynamic metabolic reprogramming across different activation states. Naïve T cells in their resting state show limited proliferative activity and rely primarily on oxidative phosphorylation (OXPHOS) for energy generation. Following antigen stimulation, they differentiate into effector T cells and shift to a glycolytic activity, indicative of metabolic activation[ 57 ]. Activated neutrophils, M1-polarized macrophages, and iNOS-expressing dendritic cells (DCs) predominantly rely on glycolysis to meet energy demands[ 58 ]. Therefore, understanding immune cell metabolic profiles and their functional consequences may enable the development of targeted therapies to modulate immunosuppressive microenvironment and improve clinical outcomes. Conclusions This study developed an energy metabolism-based classifier for gliomas with both prognostic and therapeutic potential. Metabolic reprogramming showed significant associations with both OPC-to-MES transition in glioma cells and immunosuppression within the tumor microenvironment. Multi-omics data, especially single-cell transcriptomes, provide us with a perspective to understand metabolism of gliomas at the single cell resolution, so as to better customize individualized treatment strategies for patients. Abbreviations TME: tumor microenvironment; TCGA: the cancer genome atlas; CGGA: Chinese glioma genome atlas; PPP: the pentose phosphate pathway; FAO: fatty acid oxidation (FAO); MES: mesenchymal; MDM: monocyte-derived macrophages; CPTAC: Clinical Proteomic Tumor Analysis Consortium; Ivy GAP: Ivy glioblastoma atlas project; MSigDB: Molecular Signatures Database; ssGSEA: single sample Gene Set Enrichment Analysis; PCA: principal Component Analysis; CCP : cell cycle progression; TILs : tumor-infiltrating lymphocytes; CYT : immune cytolytic activity; IFN: interferon; CCLE: Cancer Cell Line Encyclopedia; KEGG: Kyoto Encyclopedia of Genes and Genomes; CNV: copy number variation; WHO: world health organization; IDH: isocitrate dehydrogenase; MGMT promoter methylation: O6-methylguanine-DNA methyltransferase promoter methylation; IQR: interquartile range: DCs: Dendritic cells: Oligs: oligodendrocytes; GLUT: glucose transporters; G-6-P: Glucose 6-phosphate; MCTs: Monocarboxylate transporters; LDH: lactate dehydrogenase; TCA: tricarboxylic acid cycle; α-KG: α-Ketoglutaric acid; G-6-P: glucose-6-phosphatase ; R-5-P: Ribose 5-phosphate; NADPH: nicotinamide adenine dinucleotide phosphate; Lac: lactate; TAMs: tumor associated microglia or macrophages; SMCs: smooth muscle cells; AC-like: astrocyte like; MES-like: mesenchymal like; NPC-like: neural progeneitor; OPC-like: Oligdendrocytes like CThbv: cellular tumor of hyperplastic blood vessels; CTmvp: cellular tumor of microvascular proliferation; CTpan: cellular tumor of pseudopalisading cell around necrosis; CTpnz: cellular tumor of perinecrotic zone; IT: infiltrating area; LE: leading edge; CTLs: Cytotoxic T lymphocytes. Declarations Ethics approval and consent to participate: All animal studies were approved by the Institutional Animal Care and Use Committee of the First Affiliated Hospital of the University of Science and Technology of China. Consent for publication: All authors consent to the publication of this manuscript. Availability of data and material: Not applicable. Competing interests: The authors declare that they have no competing interests. Funding This study was supported in part by grants from the National Natural Science Foundation of China [82273281]. This study was supported in part by grants from the National Natural Science Foundation of China [82203693]. Authors’ contributions Qiang Zhu and Chaoshi Niu conceived and designed the study. Qiang Zhu and Wanxiang Niu were responsible for data acquisition and analysis. Qiang Zhu, Chengkun Ye, and Wanxiang Niu undertook the interpretation of data. Qiang Zhu prepared the original draft of the manuscript, which was further reviewed and edited by Qiang Zhu, Maolin Mu, and Chaoshi Niu. Chaoshi Niu provided supervision throughout the study. Acknowledgements: Not applicable References Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, Satija R: Dictionary learning for integrative, multimodal and scalable single-cell analysis . Nat Biotechnol 2024, 42 (2):293-304. Neftel C, Laffy J, Filbin MG, Hara T, Shore ME, Rahme GJ, Richman AR, Silverbush D, Shaw ML, Hebert CM et al : An Integrative Model of Cellular States, Plasticity, and Genetics for Glioblastoma . Cell 2019, 178 (4):835-849.e821. Verhaak RGW, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, Miller CR, Ding L, Golub T, Mesirov JP et al : Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1 . Cancer Cell 2010, 17 (1). Hanahan D: Hallmarks of Cancer: New Dimensions . Cancer discovery 2022, 12 (1):31-46. Peng X, Chen Z, Farshidfar F, Xu X, Lorenzi PL, Wang Y, Cheng F, Tan L, Mojumdar K, Du D et al : Molecular Characterization and Clinical Relevance of Metabolic Expression Subtypes in Human Cancers . Cell reports 2018, 23 (1):255-269.e254. Björkblom B, Wibom C, Eriksson M, Bergenheim AT, Sjöberg RL, Jonsson P, Brännström T, Antti H, Sandström M, Melin B: Distinct metabolic hallmarks of WHO classified adult glioma subtypes . Neuro Oncol 2022, 24 (9):1454-1468. Upadhyayula PS, Higgins DM, Mela A, Banu M, Dovas A, Zandkarimi F, Patel P, Mahajan A, Humala N, Nguyen TTT et al : Dietary restriction of cysteine and methionine sensitizes gliomas to ferroptosis and induces alterations in energetic metabolism . Nat Commun 2023, 14 (1):1187. Jin N, Bi A, Lan X, Xu J, Wang X, Liu Y, Wang T, Tang S, Zeng H, Chen Z et al : Identification of metabolic vulnerabilities of receptor tyrosine kinases-driven cancer . Nat Commun 2019, 10 (1):2701. Jiang Z, Liu Z, Li M, Chen C, Wang X: Increased glycolysis correlates with elevated immune activity in tumor immune microenvironment . EBioMedicine 2019, 42 :431-442. LeBlanc VG, Trinh DL, Aslanpour S, Hughes M, Livingstone D, Jin D, Ahn BY, Blough MD, Cairncross JG, Chan JA et al : Single-cell landscapes of primary glioblastomas and matched explants and cell lines show variable retention of inter- and intratumor heterogeneity . Cancer Cell 2022, 40 (4). Gentric G, Kieffer Y, Mieulet V, Goundiam O, Bonneau C, Nemati F, Hurbain I, Raposo G, Popova T, Stern M-H et al : PML-Regulated Mitochondrial Metabolism Enhances Chemosensitivity in Human Ovarian Cancers . Cell metabolism 2019, 29 (1):156-173.e110. Bidkhori G, Benfeitas R, Klevstig M, Zhang C, Nielsen J, Uhlen M, Boren J, Mardinoglu A: Metabolic network-based stratification of hepatocellular carcinoma reveals three distinct tumor subtypes . Proc Natl Acad Sci USA 2018, 115 (50):E11874-E11883. Karayama M, Masuda J, Mori K, Yasui H, Hozumi H, Suzuki Y, Furuhashi K, Fujisawa T, Enomoto N, Nakamura Y et al : Comprehensive assessment of multiple tryptophan metabolites as potential biomarkers for immune checkpoint inhibitors in patients with non-small cell lung cancer . Clinical & translational oncology : official publication of the Federation of Spanish Oncology Societies and of the National Cancer Institute of Mexico 2021, 23 (2):418-423. Yan Y, Chang L, Tian H, Wang L, Zhang Y, Yang T, Li G, Hu W, Shah K, Chen G, Guo Y: 1-Pyrroline-5-carboxylate released by prostate Cancer cell inhibit T cell proliferation and function by targeting SHP1/cytochrome c oxidoreductase/ROS Axis . Journal for immunotherapy of cancer 2018, 6 (1):148. De Leo A, Ugolini A, Yu X, Scirocchi F, Scocozza D, Peixoto B, Pace A, D'Angelo L, Liu JKC, Etame AB et al : Glucose-driven histone lactylation promotes the immunosuppressive activity of monocyte-derived macrophages in glioblastoma . Immunity 2024, 57 (5):1105-1123.e1108. Hurley HJ, Dewald H, Rothkopf ZS, Singh S, Jenkins F, Deb P, De S, Barnes BJ, Fitzgerald-Bocarsly P: Frontline Science: AMPK regulates metabolic reprogramming necessary for interferon production in human plasmacytoid dendritic cells . Journal of leukocyte biology 2021, 109 (2):299-308. Wang LB, Karpova A, Gritsenko MA, Kyle JE, Cao S, Li Y, Rykunov D, Colaprico A, Rothstein JH, Hong R et al : Proteogenomic and metabolomic characterization of human glioblastoma . Cancer Cell 2021, 39 (4):509-528.e520. Puchalski RB, Shah N, Miller J, Dalley R, Nomura SR, Yoon JG, Smith KA, Lankerovich M, Bertagnolli D, Bickley K et al : An anatomic transcriptional atlas of human glioblastoma . Science 2018, 360 (6389):660-663. Abdelfattah N, Kumar P, Wang C, Leu JS, Flynn WF, Gao R, Baskin DS, Pichumani K, Ijare OB, Wood SL et al : Single-cell analysis of human glioma and immune cells identifies S100A4 as an immunotherapy target . Nat Commun 2022, 13 (1):767. De Berardinis RJ, Chandel NS: Fundamentals of cancer metabolism . Science advances 2016, 2 (5). Karasinska JM, Topham JT, Kalloger SE, Jang GH, Denroche RE, Culibrk L, Williamson LM, Wong HL, Lee MKC, O'Kane GM et al : Altered Gene Expression along the Glycolysis-Cholesterol Synthesis Axis Is Associated with Outcome in Pancreatic Cancer . Clin Cancer Res 2020, 26 (1):135-146. Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P: The Molecular Signatures Database (MSigDB) hallmark gene set collection . Cell systems 2015, 1 (6):417-425. Hänzelmann S, Castelo R, Guinney J: GSVA: Gene set variation analysis for microarray and RNA-Seq data . BMC Bioinformatics 2013, 14 . Kumar A, Coleman I, Morrissey C, Zhang X, True LD, Gulati R, Etzioni R, Bolouri H, Montgomery B, White T et al : Substantial interindividual and limited intraindividual genomic diversity among tumors from men with metastatic prostate cancer . Nat Med 2016, 22 (4):369-378. Yoshihara K, Shahmoradgoli M, Martínez E, Vegesna R, Kim H, Torres-Garcia W, Treviño V, Shen H, Laird PW, Levine DA et al : Inferring tumour purity and stromal and immune cell admixture from expression data . Nat Commun 2013, 4 :2612. Becht E, Giraldo NA, Lacroix L, Buttard B, Elarouci N, Petitprez F, Selves J, Laurent-Puig P, Sautès-Fridman C, Fridman WH, de Reyniès A: Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression . Genome Biol 2016, 17 (1):218. Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh P-r, Raychaudhuri S: Fast, sensitive and accurate integration of single-cell data with Harmony . Nat Methods 2019, 16 (12):1289-1296. Aran D, Looney AP, Liu L, Wu E, Fong V, Hsu A, Chak S, Naikawadi RP, Wolters PJ, Abate AR et al : Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage . Nature immunology 2019, 20 (2):163-172. Hu C, Li T, Xu Y, Zhang X, Li F, Bai J, Chen J, Jiang W, Yang K, Ou Q et al : CellMarker 2.0: an updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data . Nucleic Acids Res 2023, 51 (D1):D870-d876. Franzén O, Gan L-M, Björkegren JLM: PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data . Database 2019, 2019 . Xiong A, Zhang J, Chen Y, Zhang Y, Yang F: Integrated single-cell transcriptomic analyses reveal that GPNMB-high macrophages promote PN-MES transition and impede T cell activation in GBM . EBioMedicine 2022, 83 :104239. Xiao Z, Dai Z, Locasale JW: Metabolic landscape of the tumor microenvironment at single cell resolution . Nat Commun 2019, 10 (1):3763. Trapnell C, Cacchiarelli D, Grimsby J, Pokharel P, Li S, Morse M, Lennon NJ, Livak KJ, Mikkelsen TS, Rinn JL: The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells . Nat Biotechnol 2014, 32 (4):381-386. Sanchez-Vega F, Mina M, Armenia J, Chatila WK, Luna A, La KC, Dimitriadoy S, Liu DL, Kantheti HS, Saghafinia S et al : Oncogenic Signaling Pathways in The Cancer Genome Atlas . Cell 2018, 173 (2):321-337.e310. Jahani-Asl A, Yin H, Soleimani VD, Haque T, Luchman HA, Chang NC, Sincennes M-C, Puram SV, Scott AM, Lorimer IAJ et al : Control of glioblastoma tumorigenesis by feed-forward cytokine signaling . Nat Neurosci 2016, 19 (6):798-806. Hara T, Chanoch-Myers R, Mathewson ND, Myskiw C, Atta L, Bussema L, Eichhorn SW, Greenwald AC, Kinker GS, Rodman C et al : Interactions between cancer cells and immune cells drive transitions to mesenchymal-like states in glioblastoma . Cancer Cell 2021, 39 (6). Kim J, DeBerardinis RJ: Mechanisms and Implications of Metabolic Heterogeneity in Cancer . Cell metabolism 2019, 30 (3):434-446. Daemen A, Peterson D, Sahu N, McCord R, Du X, Liu B, Kowanetz K, Hong R, Moffat J, Gao M et al : Metabolite profiling stratifies pancreatic ductal adenocarcinomas into subtypes with distinct sensitivities to metabolic inhibitors . Proc Natl Acad Sci U S A 2015, 112 (32):E4410-4417. Drews RM, Hernando B, Tarabichi M, Haase K, Lesluyes T, Smith PS, Morrill Gavarró L, Couturier DL, Liu L, Schneider M et al : A pan-cancer compendium of chromosomal instability . Nature 2022, 606 (7916):976-983. Network CGAR: Comprehensive genomic characterization defines human glioblastoma genes and core pathways . Nature 2008, 455 (7216):1061-1068. Vander Heiden MG, DeBerardinis RJ: Understanding the Intersections between Metabolism and Cancer Biology . Cell 2017, 168 (4):657-669. Garofano L, Migliozzi S, Oh YT, D'Angelo F, Najac RD, Ko A, Frangaj B, Caruso FP, Yu K, Yuan J et al : Pathway-based classification of glioblastoma uncovers a mitochondrial subtype with therapeutic vulnerabilities . Nature cancer 2021, 2 (2):141-156. Bhandari V, Li CH, Bristow RG, Boutros PC, Aaltonen LA, Abascal F, Abeshouse A, Aburatani H, Adams DJ, Agrawal N et al : Divergent mutational processes distinguish hypoxic and normoxic tumours . Nat Commun 2020, 11 (1):737. Chen Z, Hambardzumyan D: Macrophage-tumor cell intertwine drives the transition into a mesenchymal-like cellular state of glioblastoma . Cancer Cell 2021, 39 (6):743-745. Masjedi A, Hajizadeh F, Beigi Dargani F, Beyzai B, Aksoun M, Hojjat-Farsangi M, Zekiy A, Jadidi-Niaragh F: Oncostatin M: A mysterious cytokine in cancers . International Immunopharmacology 2021, 90 :107158. Li YJ, Zhang C, Martincuks A, Herrmann A, Yu H: STAT proteins in cancer: orchestration of metabolism . Nat Rev Cancer 2023, 23 (3):115-134. Warburg O: On the origin of cancer cells . Science 1956, 123 (3191):309-314. Boroughs LK, DeBerardinis RJ: Metabolic pathways promoting cancer cell survival and growth . Nature cell biology 2015, 17 (4):351-359. Altman BJ, Stine ZE, Dang CV: From Krebs to clinic: glutamine metabolism to cancer therapy . Nat Rev Cancer 2016, 16 (10):619-634. Martinez-Outschoorn UE, Peiris-Pagés M, Pestell RG, Sotgia F, Lisanti MP: Cancer metabolism: a therapeutic perspective . Nature reviews Clinical oncology 2017, 14 (1):11-31. Sun L, Song L, Wan Q, Wu G, Li X, Wang Y, Wang J, Liu Z, Zhong X, He X et al : cMyc-mediated activation of serine biosynthesis pathway is critical for cancer progression under nutrient deprivation conditions . Cell research 2015, 25 (4):429-444. Terry S, Engelsen AST, Buart S, Elsayed WS, Venkatesh GH, Chouaib S: Hypoxia-driven intratumor heterogeneity and immune evasion . Cancer Lett 2020, 492 :1-10. Huang B, Song BL, Xu C: Cholesterol metabolism in cancer: mechanisms and therapeutic opportunities . Nature metabolism 2020, 2 (2):132-141. Chen B, Gao A, Tu B, Wang Y, Yu X, Wang Y, Xiu Y, Wang B, Wan Y, Huang Y: Metabolic modulation via mTOR pathway and anti-angiogenesis remodels tumor microenvironment using PD-L1-targeting codelivery . Biomaterials 2020, 255 :120187. Omuro A, Reardon DA, Sampson JH, Baehring J, Sahebjam S, Cloughesy TF, Chalamandaris AG, Potter V, Butowski N, Lim M: Nivolumab plus radiotherapy with or without temozolomide in newly diagnosed glioblastoma: Results from exploratory phase I cohorts of CheckMate 143 . Neuro-oncology advances 2022, 4 (1):vdac025. Hao S, Yan KK, Ding L, Qian C, Chi H, Yu J: Network Approaches for Dissecting the Immune System . iScience 2020, 23 (8):101354. Pearce EL, Poffenberger MC, Chang CH, Jones RG: Fueling immunity: insights into metabolism and lymphocyte function . Science 2013, 342 (6155):1242454. Pearce EL, Pearce EJ: Metabolic pathways in immune cell activation and quiescence . Immunity 2013, 38 (4):633-643. Supplementary Files supplementary.docx Cite Share Download PDF Status: Published Journal Publication published 30 Dec, 2025 Read the published version in Journal of Translational Medicine → Version 1 posted Reviewers agreed at journal 02 Sep, 2025 Reviewers invited by journal 27 Jul, 2025 Editor assigned by journal 26 Jul, 2025 First submitted to journal 23 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7196875","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":491586565,"identity":"0cc4de2c-2fe3-439e-902d-6a52f61a5727","order_by":0,"name":"Qiang Zhu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCElEQVRIiWNgGAWjYDACCSBOAGI29uaDDz4gCRLWws9zLNlwBlQ1YS0gIDkjx0yahxgt8rN7DG88qLljt+FGgoG0bVtdHX8D88HbPAx2ebi0MM45Y2yRcOxZ8oYzDxKMc9sOS0gcYEu25mFILsalhVkix0wige1wssHxhAPJuW0HJAwYeEAuPJDYgEMLG1jLP6AWoJrDlm11QC383/Bq4QFpSWw7bCfZkczYzNjGDLKFDa8WCYm0YovEvsMJwEBmZuw5d1hyxmE2Y8s5Bsk4tcjPSN5488e3w/Zs7P3ff/woq+Pnb29+eONNhR1OLWCbgBhJATOIMMCjHqrFHr+SUTAKRsEoGNEAANAVVH1k9ChwAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-8253-1811","institution":"The First Affiliated Hospital of USTC: Anhui Provincial Hospital","correspondingAuthor":true,"prefix":"","firstName":"Qiang","middleName":"","lastName":"Zhu","suffix":""},{"id":491586566,"identity":"eea6ec30-711a-47a6-ad87-73a8074e9e78","order_by":1,"name":"Wanxiang Niu","email":"","orcid":"","institution":"The First Affiliated Hospital of USTC: Anhui Provincial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Wanxiang","middleName":"","lastName":"Niu","suffix":""},{"id":491586567,"identity":"769a1932-beeb-4990-891e-eaf0ac9b6f00","order_by":2,"name":"Maolin Mu","email":"","orcid":"","institution":"The First Affiliated Hospital of USTC: Anhui Provincial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Maolin","middleName":"","lastName":"Mu","suffix":""},{"id":491586568,"identity":"94434cb4-0212-4747-b26f-025a1edda8b9","order_by":3,"name":"Chengkun Ye","email":"","orcid":"","institution":"The First Affiliated Hospital of USTC: Anhui Provincial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Chengkun","middleName":"","lastName":"Ye","suffix":""},{"id":491586569,"identity":"eaea049a-8f3f-4799-9685-85ba1beb63aa","order_by":4,"name":"Chaoshi Niu","email":"","orcid":"","institution":"The First Affiliated Hospital of USTC: Anhui Provincial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Chaoshi","middleName":"","lastName":"Niu","suffix":""}],"badges":[],"createdAt":"2025-07-23 13:25:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7196875/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7196875/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12967-025-07602-z","type":"published","date":"2025-12-30T15:58:07+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":87884317,"identity":"6b00fca3-87ef-4c53-9545-d4ac6d82dc94","added_by":"auto","created_at":"2025-07-30 04:58:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":207948,"visible":true,"origin":"","legend":"\u003cp\u003eGliomas exhibited significant metabolic heterogeneity between the metabolic clusters\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a) \u003c/strong\u003eSchematics of the metabolic pathways of tumors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b)\u003c/strong\u003e Heatmap showing metabolic pathway activity between metabolic clusters in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c)\u003c/strong\u003e Heatmap showing metabolic pathway activity between metabolic clusters in the CGGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d)\u003c/strong\u003e Heatmap showing the robust classification of consensus clustering (K=2) in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e)\u003c/strong\u003e Heatmap showing the robust classification of consensus clustering (K=2) in the CGGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f)\u003c/strong\u003e Proportion of gliomas with different WHO grades between two metabolic clusters in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g) \u003c/strong\u003eProportion of gliomas with different WHO grades between two metabolic clusters in the CGGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(h) \u003c/strong\u003eKaplan-Meier curves of the metabolic classifier in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(i) \u003c/strong\u003eKaplan-Meier curves of the metabolic classifier in the CGGA dataset. Log-rank test \u003cem\u003eP\u003c/em\u003e-values are shown.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/38d1ebe6749e27b1daafde4b.png"},{"id":87887014,"identity":"cba21750-9560-4f99-9045-60cda2c85562","added_by":"auto","created_at":"2025-07-30 05:32:09","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":241767,"visible":true,"origin":"","legend":"\u003cp\u003eClinical characteristics, genomic, oncogenic and immune differences between metabolic clusters.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a)\u003c/strong\u003e Heatmap showing the distributions of age, sex, tumor grade, MGMT promoter methylation, and molecular subtypes between metabolic clusters.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b)\u003c/strong\u003e Comparison of CNVs between two metabolic clusters. Significant amplifications or deletions of metabolism-related genes are indicated by arrows.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c \u003c/strong\u003eQuantitative analysis of ten classic oncogenic pathways between the metabolic clusters.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d)\u003c/strong\u003e Comparison of the mRNA expression of immune checkpoint inhibitor genes between the metabolic clusters in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e)\u003c/strong\u003e Comparison of immune signatures between the metabolic clusters in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f)\u003c/strong\u003e Comparison of scores calculated with the ESTIMATE algorithm between the metabolic clusters. Boxplots display the median and interquartile range (IQR), with whiskers extending the maximum and minimum values within 1.5 times the IQR. The outliers are marked as individual points. Statistical significance was determined using the Wilcoxon rank-sum test, with *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05, **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01, ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001, and \"ns\" indicating non-significant results.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/54fa256420c217d84e603697.png"},{"id":87885312,"identity":"828c2110-e4a9-4b9b-ae02-46f853b9a070","added_by":"auto","created_at":"2025-07-30 05:06:44","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":235951,"visible":true,"origin":"","legend":"\u003cp\u003eFurther validation of multi-omics data from the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a)\u003c/strong\u003e Heatmap of metabolic pathway activity between metabolic clusters.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b)\u003c/strong\u003e Heatmap showing the robust classification of consensus clustering (K=2) in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c)\u003c/strong\u003e Comparison of relative metabolites involved in metabolic pathways between the metabolic clusters in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d)\u003c/strong\u003e Quantitative analysis of ten classic oncogenic pathways between the metabolic clusters in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e)\u003c/strong\u003e Comparison of the mRNA levels of immune checkpoint genes between the metabolic clusters in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f)\u003c/strong\u003e Comparison of diverse immune signatures between the metabolic clusters in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g)\u003c/strong\u003e Comparison of scores calculated with the ESTIMATE algorithm between the metabolic clusters in the CPTAC dataset.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/4699bbde45a102744fd48662.png"},{"id":87884305,"identity":"152440b6-2056-48c5-8c98-8d155174ad9c","added_by":"auto","created_at":"2025-07-30 04:58:43","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":387982,"visible":true,"origin":"","legend":"\u003cp\u003eThe role of the OSMR in glioma metabolism\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a)\u003c/strong\u003e Volcano plots of differentially expressed genes between metabolic subtypes in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b)\u003c/strong\u003e Venn diagram showing OSMR as candidate genes across the TCGA, CGGA, and CPTAC datasets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c)\u003c/strong\u003e The scaled mRNA expression of OSMR across transcriptome subtypes of gliomas in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d)\u003c/strong\u003e The scaled mRNA expression of OSMR across normal brain and all WHO grades of gliomas in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e) \u003c/strong\u003eKaplan-Meier curve analysis of low grade gliomas (LGGs) in TCGA dataset stratified by OSMR expression level.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f) \u003c/strong\u003eKaplan-Meier curves showing worse overall survival with high OSMR expression in glioblastoma (GBM) patients in the TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g, h) \u003c/strong\u003eThe correlation analysis of OSMR mRNA expression with HK2 and LDHA mRNA levels in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(i, j) \u003c/strong\u003eThe correlations analysis of OSMR protein levels of OSMR with HK2 and LDHA protein levels in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(k, l) \u003c/strong\u003eThe invasion ability of the U87 cell line with si-OSMR-2 was significantly decreased compared with that of the control group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(m, n) \u003c/strong\u003eThe migration ability of the U87 cell line with si-OSMR-2 was significantly decreased compared with that of the control group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(o) \u003c/strong\u003eProtein levels of the aerobic glycolysis enzymes HK2 and LDHA following the knockdown of OSMR in the U87 cell line.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/4d74f5dde8fdcf588e44a641.png"},{"id":87885304,"identity":"930c77d5-04ae-4b77-84c4-5eff65dd0efc","added_by":"auto","created_at":"2025-07-30 05:06:44","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":522546,"visible":true,"origin":"","legend":"\u003cp\u003eThe OSM/OSMR/JAK1/STAT3 axis drives progression of glioma\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a) \u003c/strong\u003eEnrichment analysis of different expressed genes between the high vs low expression group of OSMR in TCGA dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b) \u003c/strong\u003eProtein levels of the pathway of JAK1/STAT3 following the knockdown of OSMR in the U87 cell line.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c, d) \u003c/strong\u003eThe invasion ability of the U87 cell line treated with GA-A was significantly lower compared to the control group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e, f)\u003c/strong\u003e The migration ability of the U87 cell line treated with GA-A was significantly lower compared to the control group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g, h)\u003c/strong\u003e The invasion ability of the LN229 cell line treated with GA-A was significantly lower compared to the control group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(i, j) \u003c/strong\u003eThe migration ability of the LN229 cell line treated with GA-A was significantly lower compared to the control group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(k)\u003c/strong\u003e The protocol of the orthotopic tumor implantation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(l)\u003c/strong\u003e Bioluminescence imaging of intracranial tumor growth.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(m)\u003c/strong\u003e Radiance of bioluminescence imaging of intracranial tumor.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(n)\u003c/strong\u003e Kaplan-Meier curves showing better overall survival with sh-OSMR group of xenograft mouse compared to the sh-NC group.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/c459ab307c9aa75aabc7567b.png"},{"id":87884307,"identity":"4aa8b90b-7da3-441d-9f75-de6dbebdb167","added_by":"auto","created_at":"2025-07-30 04:58:44","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":368608,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of the metabolic classifier at the single-cell resolution of the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a)\u003c/strong\u003e tSNE plot of snRNA-seq data from the CPTAC dataset with the corresponding cell types marked.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b)\u003c/strong\u003e Dot plot displaying the expression of marker genes across various cell types. The dot size represents the proportion of cells within each type, while the color indicates the average expression level.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c) \u003c/strong\u003eMetabolic classification based on bulk RNA-seq data of 16 patients in the CPTAC dataset who underwent both bulk RNA and snRNA sequencing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d) \u003c/strong\u003eMetabolic classification based on mixed malignant cells of snRNA-seq for 16 patients in the CPTAC dataset using the metabolic classifier.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e) \u003c/strong\u003eThe robustness of the metabolic cluster scores in defining the metabolic classifier. The cluster score is calculated as the average of two cluster-specific pathways minus the average of four metabolic pathways.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f) \u003c/strong\u003eComparison of metabolic patterns between mixed malignant cells and bulk tumors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g) \u003c/strong\u003eCorrelation of glycolysis, FAO, and glutaminolysis with hypoxia, and angiogenesis of malignant cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(h) \u003c/strong\u003eCorrelation of FAO, and glutaminolysis with hypoxia, and angiogenesis of glioma cells from the CCLE database.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/a4cb68cb837763146cb5484b.png"},{"id":87884352,"identity":"c6aa5470-cae2-4c83-af75-305d6171653d","added_by":"auto","created_at":"2025-07-30 04:58:45","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":319228,"visible":true,"origin":"","legend":"\u003cp\u003eMetabolic profiles of various cell types at the single-cell level.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a\u003c/strong\u003e) Activity of metabolic pathways across five distinct cell types in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b) \u003c/strong\u003ePerformance of five cell types in metabolic pathways extracted in KEGG in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c) \u003c/strong\u003eCluster scores of two metabolic clusters in malignant cells in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d) \u003c/strong\u003eCluster scores of two metabolic clusters in non- malignant cells in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e) \u003c/strong\u003eThe numbers of two metabolic clusters at different locations for patients in the IVY GAP dataset. (CThbv: cellular tumor of hyperplastic blood vessels, CTmvp: cellular tumor of microvascular proliferation, CTpan: cellular tumor of pseudopalisading cell around necrosis, CTpnz: cellular tumor of perinecrotic zone, IT: infiltrating area, LE: leading edge, CTLs: Cytotoxic T lymphocytes.)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f) \u003c/strong\u003eProportion of metabolic cluster 1 and cluster 2 cells at single-cell level at different locations in the GSE182109 dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g) \u003c/strong\u003ePseudotime analysis revealed the plasticity and dynamic transition of glioma cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(h) \u003c/strong\u003eTrajectory analysis demonstrated a significant transition starting from OPC- and NPC-like tumor cells in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(i) \u003c/strong\u003eThe glycolytic activity of glioma cells increased with the PN-MES transition.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/abc443e569947f76167356c0.png"},{"id":87884347,"identity":"6b2d7bb7-3f52-4446-9046-bbbc5a656059","added_by":"auto","created_at":"2025-07-30 04:58:45","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":317653,"visible":true,"origin":"","legend":"\u003cp\u003eMetabolic profile of immune cells at the single-cell level in the CPTAC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(a)\u003c/strong\u003e tSNE projections of myeloid cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b) \u003c/strong\u003etSNE projections of lymphoid cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(c) \u003c/strong\u003eDot plot showing marker genes of different myeloid cell types.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(d) \u003c/strong\u003eDot plot showing marker genes of different lymphoid cell types.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(e) \u003c/strong\u003eViolin plot showing the metabolic activities of glycolysis, oxidative phosphorylation, and FAO in myeloid cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(f) \u003c/strong\u003eViolin plot showing metabolic the activities of glycolysis, oxidative phosphorylation, and FAO in lymphoid cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(g) \u003c/strong\u003ePerformance of metabolic pathways extracted from KEGG between CD4+ T cells and CD8+ T cells.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/d1040f6e1da70a6e97c07c21.png"},{"id":99545550,"identity":"2c2b5ef9-7317-43db-a224-88d0e744dacb","added_by":"auto","created_at":"2026-01-05 16:08:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5555519,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/902a8853-9262-4fe1-8edb-90cc22b50e2b.pdf"},{"id":87884312,"identity":"4f35d430-5d23-423e-9e79-b2ee6a438d8a","added_by":"auto","created_at":"2025-07-30 04:58:44","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3812544,"visible":true,"origin":"","legend":"","description":"","filename":"supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-7196875/v1/00b5c7ff374c41d571388619.docx"}],"financialInterests":"","formattedTitle":"Integrative multi-omics Analysis Proposes a Metabolic Classification of Gliomas: Distinct Metabolic States, Immune Infiltration, and Prognosis","fulltext":[{"header":"Background","content":"\u003cp\u003eGliomas are the most common and aggressive primary intracranial malignancies, with a poor prognosis despite multimodal treatment, such as surgery, chemotherapy and radiotherapy. Recent advances in molecular research and the rapid development of single-cell sequencing have uncovered remarkable heterogeneity in the tumor microenvironment[1, 2]. Professor Neftel, leveraging single-cell sequencing, proposed glioblastoma cell subtypes based on the transcriptional classification established by Professor Verhaak[2, 3]. Different types of glioblastoma cells exhibit variations in their genetic, epigenetic, and microenvironmental factors, endowing them with distinct capabilities for proliferation and invasion. Energy metabolism reprogramming is a newly recognized hallmarks of tumors[4]. Tumor cells adopt metabolic reprogramming to support their own proliferation and progression [5]. In clinical specimens, Professor Melin\u0026rsquo;s team has reported distinct metabolic phenotypes exist according to adult glioma subtypes[6]. Recent studies suggest that the plasticity and flexibility of tumor cell metabolism offer potential therapeutic opportunities[7-9].\u003c/p\u003e\n\u003cp\u003eSingle-cell sequencing enables precise characterization of diverse cell types within the tumor microenvironment, including clonal diversity, metabolic profiles, and intercellular interactions[10]. Previous studies have reported that ovarian and hepatocellular tumor cells can be categorized into distinct metabolic clusters to predict the patient\u0026apos;s treatment response and overall prognosis[11, 12]. However, the mechanisms of glioma cell metabolism remain poorly understood, and metabolic classifiers for gliomas are still lacking.\u003c/p\u003e\n\u003cp\u003eImmunotherapy has become a crucial component of cancer treatment, making it increasingly important to understand the metabolic heterogeneity of infiltrating immune cells and tumor cells in depth. An increasing numbers of studies have demonstrated that tumor metabolism not only plays a vital role in promoting the development and survival of tumor cells, but can also modulate anti-tumor immune responses by releasing metabolites that influence immune molecule expression, such as lactate, PGE2, and arginine.[13, 14]. In glioblastoma, Professor Veglia reveals that PERK-driven glucose metabolism in monocyte-derived macrophages (MDM) promotes MDM immunosuppressive activity via histone lactylation, indicating that immune cell metabolism modulates their anti-tumor activity[15]. Additionally, tumor cells can inhibit anti-tumor immune responses by competing for and consuming vital nutrients, thereby impairing the metabolic fitness of tumor-infiltrating immune cells[16]. Understanding the metabolic heterogeneity and characteristics of immune cells in the tumor microenvironment facilitates the identification of therapeutic targets within metabolic pathways. This approach holds promise for enhancing cancer immunotherapies.\u003c/p\u003e\n\u003cp\u003eThis study established a glioma energy metabolism-based classifier based on bulk RNA-seq, and validated it using multi-omics data, including genomics, bulk and single-cell transcriptomics, and metabolomics. Single-cell analysis characterized metabolic reprogramming during the PN-to-MES transition in glioma cells. Additionally, it elucidated the metabolic preferences of various immune cell types. Overall, by leveraging multi-omics data, this study uncovered glioma metabolic heterogeneity, providing novel insights and potential therapeutic targets.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003e1. Data accessibility\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study utilized several glioma datasets, including the TCGA dataset, the CGGA dataset, the CPTAC dataset[17], the IvyGAP dataset[18], and the GSE182109 dataset[19]. These datasets contain genomic data, bulk RNA-seq, single-cell or single-nucleus RNA-seq, and metabolome data. The GSE182109 dataset and IvyGAP datasets provide transcriptome data from different anatomical regions within the same patients. In the CPTAC database, 16 patients were subjected to both bulk RNA-seq and single-nucleus RNA seq. The sequencing data are available for download from the following websites (https://portal.gdc.cancer.gov/, https://glioblastoma.alleninstitute.org, https://www.ncbi.nlm.nih.gov/geo/).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2. Construction of energy metabolic classifiers\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBased on previous studies[20, 21] and MSigDB[22], we identified four critical metabolic pathways in gliomas: glycolysis, the pentose phosphate pathway, fatty acid oxidation, and glutaminolysis. We identified key genes associated with these metabolic pathways (Table S1). To assess the activity of these pathways in individual patient, we utilized the ssGSEA algorithm (GSVA package in R)[23].\u003c/p\u003e\n\u003cp\u003eUnsupervised classification is a key machine learning approach. In this study, we used the K-means method to ascertain the optimal classification based on glioma metabolic profiles. Prior to clustering, samples were scaled to group them based on metabolic pathway patterns. We then reordered the samples based on metabolic clusters and the scaled ssGSEA values for heatmap visualization (using the Pheatmap package in R). Additionally, we applied the consensus clustering algorithm (ConsensusClusterPlus package in R) with 1000 iterations and 80% resampling to confirm the number of clusters (K). Principal component analysis (PCA) was also performed to highlight significant differences between clusters.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3. Calculation of immune signature scores\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate cell proliferation, we utilized 31 validated genes associated with cell cycle progression (CCP)[24]. Immune infiltration levels were assessed by several immune signatures related to tumor-infiltrating lymphocytes (TILs), immune cytolytic activity (CYT), and interferon (IFN) responses[24]. ssGSEA scores were calculated to quantify the abundance of these gene sets. Tumor immune scores were estimated using the ESTIMATE algorithm[25]. The Microenvironment Cell Populations-Counter (MCP-counter) algorithm was used to quantify the populations of eight immune cell types and two stromal cell types[26]. The cluster score was computed by averaging the expression levels of all pathways within each cluster and subtracting the mean expression of four metabolic pathways. During this calculation, the values for each metabolic pathway were scaled across all samples. For scRNA-seq data from the CPTAC and GSE182109 datasets, tumor cells were classified into metabolic cluster 1 if their metabolic cluster 1 score exceeded that of metabolic cluster 2, and vice versa.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4. Preprocessing of scRNA-seq data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe preprocessing of the raw data for the CPTAC and GSE182109 datasets was described in detail in the publications[17, 19]. The Seurat package v5.0[1] was used for downstream scRNA analysis. A series of quality filters were performed to remove the low-quality cells: too few total transcript counts (\u0026lt;300); possible debris with too few genes detected (\u0026lt;200) and too few UMIs (\u0026lt;1000); possible dead cells or sign of cellular stress and apoptosis with too high a proportion of mitochondrial gene expression over the total transcript count (\u0026gt;10%). Each sample was normalized and scaled by \u0026lsquo;SCTransform\u0026rsquo; function in the Seurat package to correct for batch effects (parameters: vars.to.regress = c (\u0026ldquo;nCount_RNA\u0026rdquo;, \u0026ldquo;percent.mito\u0026rdquo;), variable.features.n = 3000). The Harmony algorithm was applied to integrate all samples for downstream analysis[27]. To reduce dimensions, principal component analysis (PCA) was initially conducted with 30 principal components (npcs =30), followed by t-distributed stochastic neighbor embedding (t-SNE) using the first 20 dimensions (dims = 1:20). The Wilcoxon rank-sum test was used to identify the specific genes of each cluster through the FindAllMarkers function (logfc.threshold = 0.25).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5. Cell annotation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUsing singleR[28], cellmarker[29], panglaodb[30] and marker genes reported in the literature[17, 31], all cell types were annotated through a combination of automated and manual methods.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e6. Metabolic activity evaluation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe metabolic pathway activity of annotated cells was calculated as described in the previous article[32]. ssGSEA values were used to compare metabolic profiles across various cell types. Metabolic pathways with a p-value \u0026lt; 0.05 in GSEA were considered statistically significant. The OXPHOS gene set was extracted from the KEGG database. The gene sets of hypoxia and angiogenesis were retrieved from the MSigDB library.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e7. Inference of CNV\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo differentiate malignant cells from non-malignant cells, copy number variation (CNV) was calculated using the R package inferCNV (https://github.com/broadinstitute/inferCNV). Genes expressed in fewer than 10 cells and with a median expression below 0.1 were excluded from the analysis to focus on robustly expressed genes. Subsequently, genes were annotated based on their chromosomal positions, and CNV scores were calculated using moving averages across sets of 100 genes. Hierarchical clustering was applied to separate non-malignant cells from malignant cells based on distinct chromosomal deletions or amplifications.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e8. Trajectory analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTrajectory analysis was facilitated by importing data from Seurat into the Monocle2 package in R[33] . Initially, genes with expression levels below 0.5 and cells expressing fewer than 200 genes were excluded from the dataset. The remaining data were reordered based on significantly differentially expressed genes identified with a q-value threshold of 1e-40. Subsequently, a DDRTree dimensional reduction algorithm was applied to interpret the data structure. Pseudotime analysis was then performed to map the temporal progression of cellular differentiation, pinpointing pathways that significantly changed along trajectories. Cells within the same branch were considered to reside at similar differentiation stages.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e9. Glioma cell line and cell culture\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe glioma cell lines of U87, LN229 and GL261 were purchased from ATCC and authenticated by STR assay. The glioma cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% antibiotics in a humidified incubator maintained at 37\u0026deg;C with 5% CO₂. For transient knockdown experiments, siRNA was transfected into glioma cells using lipofectamine 8000 (Beyotime Biotechnology, China), followed by further experiments. The plasmids were obtained from Nanjing Corues Biotechnology Co, Ltd. The detail information of plasmids is listed in Table S2. The total RNA and protein were extracted 48 hours and 72 hours after transfection respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e10. Real-time quantitative PCR (RT-qPCR) analysis\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTotal RNA was extracted from glioma cell lines using TRIzol reagent, and reverse transcription was carried out using Roche reagent. cDNA was amplified by RT-PCR using Roche reagents. Genes expression levels were normalized to GAPDH expression. Primers used for target genes amplification are listed in Table S3.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e11. Western blot\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTotal protein was extracted from glioma cells using pre-chilled RIPA buffer (Beyotime Biotechnology, China) containing a mixture of protease and phosphatase inhibitors, and protein concentration was quantified using a BCA protein assay kit (BL1784B, biosharp). Proteins were separated by gel electrophoresis and transferred onto a polyvinylidene fluoride (PVDF) membrane (Merck, Germany). The membrane was blocked in 5% non-fat milk for 1 hour and incubated with primary antibodies against OSMR (10982-1-AP, Proteintech), HK-2(22029-1-AP, Proteintech), LDHA (sc-137244, Santa Cruz), JAK1 (66466-1-lg, Proteintech), p-JAK1 (ab138005, abcam), STAT3 (ab68152, abcam), p-STAT3 (ab267373, abcam), \u0026beta;-actin (66009-1-Ig, Proteintech) overnight at 4\u0026deg;C. Following three washes with 1\u0026times;TBST, the PVDF membrane was incubated with a horseradish peroxidase (HRP)-conjugated secondary antibody (Sangon Biotech, China) at room temperature for 1 hour. The membrane was washed three more times with 1\u0026times;TBST, followed by exposure and analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e12. Cell migration analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCell migration was assessed using wound healing assay. For the wound healing assay, treated cells were evenly seeded into a 6-well plate and allowed to reach full confluence. A scratch was introduced in the cell monolayer, and the cells were then cultured in serum-free medium. Cell migration was monitored under a microscope, and images were captured and analyzed at 0, 24, and 48 hours.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e13. Cell invasion analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCell invasion ability was evaluated using Transwell chambers (8-\u0026mu;m pore size, Corning, USA) coated with Matrigel (BD Biosciences, USA). Briefly, 1\u0026times;10\u003csup\u003e4\u003c/sup\u003e cells in 200 \u0026mu;L serum-free medium were seeded into the upper chamber, while 500 \u0026mu;L complete medium containing 10% fetal bovine serum (FBS) was added to the lower chamber as a chemoattractant. After incubation for 48 h at 37\u0026deg;C in a humidified atmosphere with 5% CO\u003csub\u003e2\u003c/sub\u003e, non-invaded cells on the upper surface of the membrane were carefully removed with a cotton swab. Invaded cells on the lower surface of the membrane were fixed with 4% paraformaldehyde for 15 min, stained with 0.1% crystal violet for 20 min, and washed with PBS. The number of invaded cells was counted under an inverted microscope in five randomly selected fields at 200\u0026times; magnification.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e14. tumor xenograft model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFemale C57BL/6J mice (4 weeks old, body weight 19-21g) were purchased from GemPharmatech Co, Ltd. (Jang Su, China) and housed in laminar airflow cabinets under specific pathogen-free (SPF) conditions. For orthotopic\u0026nbsp;tumor implantation, 1\u0026times;10⁵ GL261 cells stably transfected with sh-OSMR or sh-NC were stereotactically\u0026nbsp;injected into the brain of each mouse. Intracranial tumor growth was evaluated by bioluminescence imaging. All animal studies were approved by the Institutional Animal Care and Use Committee of the First Affiliated Hospital of the University of Science and Technology of China.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e15. Statistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe mean (standard deviation) was used for continuous variables with normal distribution, and the median and interquartile range (IQR) were used for other data. When comparing two groups of continuous variables, the t-test or the Mann\u0026ndash;Whitney rank-sum test was used based on the distribution of data. The chi-square test was used for categorical variables. A \u003cem\u003eP\u003c/em\u003e-value \u0026lt; 0.05 was considered statistically significant. The Kaplan-Meier survival curve was created using GraphPad Prism 9 software. A \u003cem\u003eP\u003c/em\u003e-value \u0026lt; 0.05 from the log-rank test was considered indicative of a significant difference in survival times.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eConstruction of energy metabolic classifier based bulk RNA-seq\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA flowchart was created to systematically illustrate the process of establishing the energy metabolic classifier (Figure S1). Based on previous studies of tumor metabolism, we identified four metabolic pathways: glycolysis, PPP, FAO, and glutaminolysis to describe the metabolic status of tumors (Figure 1a). The ssGSEA algorithm was used to evaluate the activity of those metabolic pathways, and values were scaled using \u0026lsquo;scale\u0026rsquo; function in RStudio. The heatmap revealed that in the TCGA dataset, patients in metabolic cluster 1 exhibited higher activity in the glycolysis, PPP, and FAO pathways, while patients in cluster 2 showed increased activity in the glutamine metabolism pathway (Figure 1b). This same metabolic pattern was also observed in the CGGA dataset (Figure 1c).\u003c/p\u003e\n\u003cp\u003eThe heatmaps demonstrated the robustness of consensus clustering (K=2) compared to other numbers of clusters in both the TCGA and CGGA datasets, based on the K-means unsupervised classification (Figure 1d, Figure 1e, Figure S2a-S2d). In terms of tumor grade, patients in metabolic cluster 1 had a higher proportion of high-grade gliomas in the TCGA and CGGA datasets (Figure 1f and 1g). Univariate Cox analysis revealed that patients in cluster 1 had a significantly worse prognosis (Figure 1h, the TCGA dataset: \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001; Figure 1i, the CGGA dataset: \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001). Based on these results, two metabolic clusters were identified from the bulk transcriptome of gliomas: (1) dependent on glycolysis, PPP, and FAO; (2) dependent on glutaminolysis. Principal component analysis showed significant heterogeneity between the two clusters using this metabolic classifier, and the classification into two metabolic clusters was demonstrated to be robust (Figure S2e and S2f). Moreover, there were significant differences in the mRNA expression of metabolism-related genes between the two metabolic clusters (Figure S2g).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical, genomic, and immune characteristics associated with metabolic classifier\u0026nbsp;\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo further analyze the clinical characteristics associated with the energy metabolic classifier, this study included several important clinical characteristics from the TCGA dataset (Figure 2a). It is obvious that the age of patients in cluster 1 is significantly higher than that in cluster 2. In terms of molecular pathology, cluster 1 had a significantly higher proportion of IDH wild-type patients compared to cluster 2, and patients in cluster 2 have a higher proportion of MGMT methylation (Figure 2a). Multivariate survival analysis revealed that the metabolic classifier was a significant risk factor in both the TCGA dataset (HR=1.85[1.09, 3.13], \u003cem\u003eP\u003c/em\u003e=0.025) (Figure S3a). Similar results were observed in the CGGA dataset (HR=1.67[1.30, 2.13], \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) (Figure S3b).\u003c/p\u003e\n\u003cp\u003eOncogene mutation events can stimulate cell-autonomous metabolic reprogramming. In this study, we found that several metabolic pathway-related genes were significantly amplified or deleted based on the CNV data (Figure 2b), and the mRNA expression levels of these genes in the clusters were consistent with the amplification or deletion event in the genomic data (Figure S4a and S4b). In addition, the overall copy number variation events in cluster 1 were significantly more frequent than those in cluster 2 (Figure 2b).\u003c/p\u003e\n\u003cp\u003eConsidering the significant difference in overall survival between patients in the two clusters, this study explored the classical oncogenic pathways in the two clusters. Based on previous literature reports, we identified ten classic tumor core pathways and quantified pathway activity using the GSVA algorithm[34]. In cluster 1, the activities of the cell cycle, Hippo, TGF-\u0026beta;, NRF2, and NOTCH were significantly higher than those in cluster 2, while the activities of PI3K, RAS, and WNT pathways were higher in patients in cluster 2 (Figure 2c). Similar results were observed in the CGGA dataset (Figure S5a).\u003c/p\u003e\n\u003cp\u003eMetabolic reprogramming has been shown to significantly influence tumor immunity. In the TCGA dataset, multiple targetable immune checkpoint genes were highly expressed in cluster 1 (Figure 2d). A similar tendency was observed in the CGGA dataset (Figure S5b). Several immune signatures, such as TILs, CYT, and IFN response, were significantly elevated in cluster 1 (Figure 2e), and a similar tendency was also found in the CGGA dataset (Figure S5c). Furthermore, the ESTIMATE algorithm[25] revealed significantly higher immune (\u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) and stromal scores (\u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) in metabolic cluster 1compared to metabolic cluster 2 (Figure 2f). This result was further validated in the CGGA dataset (Figure S5d). To further explore the relationships between tumor metabolism and immunity, we assessed the abundance of all ten immune and stromal cell types using the MCP-counter algorithm. In both the TCGA and CGGA datasets, metabolic cluster 1 exhibited significantly higher abundances of CTLs, CD8+ T cells, monocytes, and myeloid cells compared to cluster 2 (Figures S6a and S6b). Overall, this study demonstrated significant differences between the two metabolic clusters in clinical characteristics, genomic mutations, molecular pathology, and tumor immunity, highlighting intrinsic heterogeneity between the two patient groups.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMulti-omics data further validated the energy metabolic classifier\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe robustness of the metabolic classifier was further validated using the CPTAC dataset. Patients were assigned to corresponding metabolic clusters based on the established classifier. The heatmap revealed consistent metabolic differences in the CPTAC dataset compared to the TCGA and CGGA datasets (Figure 3a). Metabolic cluster 1 exhibited higher activities in glycolysis, PPP, and FAO pathways, whereas cluster 2 showed elevated activity in glutamine metabolism (Figure 3a). The heatmap of K-means unsupervised classification also suggested robust classification (K=2) (Figure 3b, Figure S7a and S7b). PCA revealed significant heterogeneity between the two clusters (Figure S7c). Although no significant difference in overall survival was observed between the two clusters, patients in metabolic cluster 1 tended to have a worse prognosis compared to those in cluster 2 (\u003cem\u003eP\u003c/em\u003e=0.259) (Figure S7d). In addition to bulk RNA-seq, metabolite levels were also quantified in the CPTAC dataset. The levels of lactate, alanine, and palmitic acid in metabolic cluster 1 were significantly higher than those in metabolic cluster 2 (Figure 3c). The levels of glycerol-3-phosphate and glutamate in metabolic cluster 2 were significantly higher than those in metabolic cluster 1 (Figure 3c). The observed metabolite levels difference aligned with the metabolic pathway activity difference. Furthermore, the Hippo, TP53, PI3K, and RAS pathways exhibited consistent differences with those observed in the TCGA and CGGA datasets (Figure 3d).\u003c/p\u003e\n\u003cp\u003eIn terms of the immune profile of the tumor microenvironment, metabolic cluster 1 exhibited higher mRNA levels of immunosuppressive checkpoint genes (Figure 3e). Meanwhile, several immune features, including IFN, TILs and CYT, were significantly elevated in metabolic cluster 1 compared to cluster 2 (Figure 3f). Moreover, the ESTIMATE algorithm revealed significantly higher immune and stromal scores in metabolic cluster 2 compared to cluster 2 (Figure 3g). The MCP algorithm indicated significantly higher abundance of monocytes, myeloid cells, and neutrophils in metabolic cluster 1 (Figure S7e), consistent with the findings in the TCGA and CGGA datasets. Further validation using multi-omics data from the CPTAC dataset, confirmed the robustness of the energy metabolic classifier, which accurately reflects glioma metabolic heterogeneity and highlights the significant impact of metabolic reprogramming on tumor immunity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe role of the OSM/OSMR in glioma metabolism\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo further explore whether a specific transcriptomic program is associated with metabolic phenotypes in metabolic cluster 1, differential gene analysis was performed between the two metabolic clusters, and results were visualized using a volcano plot (Figure 4a). To identify the top upregulated genes in metabolic cluster 1, the top 100 highly expressed genes from the TCGA, CGGA, and CPTAC datasets were intersected, identifying OSMR as a key gene (Figure 4b). OSMR, the receptor for OSM (Oncostatin M), is primarily secreted by macrophages and T cells and belongs to the interleukin-6 cytokine family. It has been reported that OSM/OSMR play a key role in the proliferation and differentiation of various tumors. In gliomas, it has been reported that OSM acts through OSMR to enhance glioma cell migration. Additionally, there are report suggesting that M2-type macrophages secrete OSM, which acts on OSMR to facilitate the proneural-to-mesenchymal transition (PMT) in glioblastoma. We found that OSMR gene was highly expressed in mesenchymal subtype and high-grade gliomas in the TCGA cohort (Figure 4c and 4d). Similar trends were also observed in the CGGA cohort (Figure S8a and S8b). However, the role of OSMR in glioma metabolism remains poorly understood. Survival analysis revealed that high OSMR expression is significantly associated with worse overall survival in glioma patients in the TCGA dataset (LGGs: \u003cem\u003eP\u003c/em\u003e=0.0013, GBMs: \u003cem\u003eP\u003c/em\u003e=0.0097) (Figure 4e and 4f). In CGGA datasets, high expression of OSMR showed significant worse overall survival in LGGs (\u003cem\u003eP\u003c/em\u003e\u0026lt;0.0001) (Figure S8c). For GBM, although high OSMR expression is not significantly associated with overall survival, there is still a tendency in that direction (\u003cem\u003eP\u003c/em\u003e=0.137) (Figure S8d). We further explored the correlation between OSMR and glycolysis-related genes. Correlation analysis revealed significant positive associations between OSMR and both mRNA and protein levels of hexokinase 2 (HK2) and lactate dehydrogenase A (LDHA) in the CPTAC dataset (mRNA level: Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.60, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.60, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001; protein level: Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.54, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.29, \u003cem\u003eP\u003c/em\u003e=0.002) (Figure 4g and 4h; Figure 4I and 4J). In the TCGA dataset, OSMR also showed strong positive correlation with HK2 (Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.58, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) and LDHA (Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.71, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) at mRNA level (Figure S8e and S8f). According to the published literature[35, 36], OSM was added to the culture medium at a concentration of 10 ng/ml for the subsequent experiments. Additionally, siRNA was used to knocked down OSMR in glioma cell lines to explore the biological function of OSMR (Figure S8g) and observed a significant decrease in the invasion and migration ability of the U87 cell line (Figure 4k-4n). In terms of metabolism, OSMR knockdown significantly reduced the protein levels of HK2 and LDHA through western blotting in U87 cell line (Figure 4o). As key enzymes in the aerobic glycolytic pathway, HK2 and LDHA suggest that OSM/OSMR plays a critical role in aerobic glycolytic metabolic reprogramming in glioma.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe OSM/OSMR/JAK1/STAT3 axis drives malignant progression of glioma\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePathway enrichment analysis revealed significant activation of the JAK/STAT pathway in glioma patients with high OSMR expression (Figure 5a). Given JAK1\u0026rsquo;s predominant expression in the nervous system and previous reports on STAT3 with OSMR, we evaluated its phosphorylation status and that of downstream STAT3 via Western blotting. OSM treatment significantly upregulated p-JAK1 (Tyr1034/1035) and p-STAT3 (Tyr705) levels, indicating pathway activation, whereas pre-treatment with a JAK-STAT3 pathway inhibitor (Ganoderic acid A (GA-A), 1.5 uM) abolished this effect, confirming JAK1-STAT3 pathway dependency (Figure 5b). To functionally validate the axis, we performed rescue experiments in U87 cells (representing metabolic cluster 1). GA-A significantly attenuated OSM-induced cell invasion (\u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) (Figure 5c and 5d) and migration (\u003cem\u003eP\u003c/em\u003e=0.001) in U87 cell line (Figure 5e and 5f), phenocopying OSMR knockdown effects. Complementary gain-of-function studies were conducted in LN229 cells (metabolic cluster 2) stably overexpressing OSMR, which exhibited similar results with U87 cell line (Figure 5g-5j). In vivo validation utilized a syngeneic orthotopic model, where\u0026nbsp;1\u0026times;10⁵ luciferase-labeled GL261 cells (sh-OSMR vs. sh-NC) were stereotactically injected into the brain of C57BL/6J mice (n=5/group).\u0026nbsp;Bioluminescence imaging revealed a significant reduction in tumor burden (\u003cem\u003eP\u003c/em\u003e=0.0039) (Figure 5l and 5m) and prolonged median survival in sh-OSMR group (\u003cem\u003eP\u003c/em\u003e=0.0011) (Figure 5m). Collectively, these results establish OSMR as a critical mediator of glioma progression through JAK1/STAT3-dependent metabolic reprogramming, highlighting its potential as a therapeutic target.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetabolic profiles of malignant cells at single-cell level\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study aimed to explore metabolism at single-cell resolution using single cell RNA sequencing to better understand metabolic heterogeneity in the glioma microenvironment. Single-nucleus sequencing and bulk RNA sequencing were performed on 16 patients from the CPTAC dataset. The snRNA-seq data from 16 patients were quality-checked and integrated using the Harmony algorithm. Subsequent dimensionality reduction and clustering analyses were performed. The t-SNE projection showed excellent integration performance (Figure S9a). As described in the methods, cells were annotated both automatically and manually, including tumor cells, immune cells, and brain-resident cells. The t-SNE visualization displayed the annotated cell types (Figure 6a). A dot plot further illustrated the expression of cell type-specific marker genes (Figure 6b). The inferCNV algorithm was used to identify malignant or non-malignant cells, confirming accurate annotation of glioma cells (Figure S9b). For these 16 patients, they were assigned to metabolic clusters based on bulk RNA-seq and malignant cells from snRNA-seq data using the metabolic classifier (Figure 6c and 6d). Notably, classifications based on bulk RNA-seq and snRNA-seq consistently grouped patients into the same metabolic clusters. The metabolic score accurately reflected the energy metabolic activity of each patient (Figure 6e). As shown in Figure 6f, the metabolic classification of most tumor cells was consistent with the metabolic clusters in the bulk RNA-seq data. Based on the above results, tumor metabolic heterogeneity is primarily driven by malignant cells.\u003c/p\u003e\n\u003cp\u003eIn addition, we sought to investigate the impact of the tumor microenvironment on tumor metabolism. Given that environmental conditions cannot be directly integrated into the metabolic analyses, hypoxia and angiogenesis signatures were used as proxies for oxygen and nutrients supply within tumor microenvironment. Consistent with published research, the glycolysis pathway exhibits a significant positive correlation with hypoxia and angiogenesis (hypoxia, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.67, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001; angiogenesis, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.27,\u003cem\u003e\u0026nbsp;P\u003c/em\u003e\u0026lt;0.001) (Figure 6g). The fatty acid oxidation pathway is also positively correlated with hypoxia and angiogenesis (hypoxia, Pearson\u0026rsquo;s\u003cem\u003e\u0026nbsp;r\u003c/em\u003e=0.22, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001; angiogenesis, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.20, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) (Figure 6g). Glycolysis, free fatty acid oxidation, and glutamine metabolism were all significantly positively correlated with oxidative phosphorylation (glycolysis, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.43, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001; fatty acids, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.36,\u003cem\u003e\u0026nbsp;P\u003c/em\u003e\u0026lt;0.001, glutamine metabolism, Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=0.34, \u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) (Figure S10a). Glioma cell lines from the CCLE database were used to explore the metabolism of glioma cells without nonmalignant cells in vitro. Although the limited number of glioma cell lines may introduce some statistical variability, intriguingly, certain metabolic patterns in these cell lines contrast with those observed in tumor tissues. Fatty acid oxidation tends to show a negatively correlation with hypoxia, though this did not reach statistical significance (Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=-0.22, \u003cem\u003eP\u003c/em\u003e=0.42) (Figure 6h, top). Glutamine metabolism was significantly negatively correlated with angiogenesis (Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=-0.54, \u003cem\u003eP\u003c/em\u003e=0.037) (Figure 6h, bottom), and also had a negative correlation trend with hypoxia (Pearson\u0026rsquo;s \u003cem\u003er\u003c/em\u003e=-0.46, \u003cem\u003eP\u003c/em\u003e=0.082) (Figure S10b). This result suggests that nonmalignant cells in the tumor microenvironment may have a certain impact on the metabolism of tumor cells. In summary, tumor metabolic heterogeneity is primarily driven by malignant cells, our results underscore the significant roles of cell-cell interactions and the tumor microenvironment in shaping tumor metabolism.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetabolic characteristics of malignant and nonmalignant cells in tumor microenvironment\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo further elucidate the metabolic characteristics of different cell types within the tumor microenvironment, we assess the metabolic activities of five annotated cell types based on prior studies. Tumor cells exhibited the highest metabolic activity (Figure 7a), and demonstrated elevated across the majority of metabolic pathways (Figure 7b). In contrast, tumor-infiltrating lymphocytes had the lowest metabolic activity (Figure 7a).\u003c/p\u003e\n\u003cp\u003eTo further investigate the relationship between the metabolic classifier and cell properties, we compared the metabolic pathway activities of malignant and nonmalignant cells. Among malignant cells, tumor cells from patients in metabolic cluster 1 exhibited significantly higher cluster 1 scores (\u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) (Figure 7c, right), while those in metabolic cluster 2 showed higher cluster 2 scores (\u003cem\u003eP\u003c/em\u003e\u0026lt;0.001) (Figure 7c, left). In contrast, nonmalignant cells displayed no statistical differences in cluster 1 scores or cluster 2 scores between metabolic clusters (\u003cem\u003eP\u003c/em\u003e=0.067 (left), \u003cem\u003eP\u003c/em\u003e=0.078 (right)) (Figure 7d). The findings further demonstrate that the metabolic heterogeneity of tumor tissue is primarily driven by malignant cells, and our metabolic classifier effectively captures this heterogeneity at single-cell resolution. We further explored the relationship between tumor metabolic heterogeneity and spatial location using the IvyGAP and GSE182109 datasets. Samples were divided into metabolic cluster 1 and metabolic cluster 2 based on the energy-related metabolism classifier. In the IvyGAP dataset, tumor tissue in the marginal regions, such as infiltrating and leading edge areas, exhibited higher cluster 1 score. Conversely, cellular tumors associated with hyperplastic blood vessels and microvascular proliferation were more likely to exhibit metabolic cluster 2 status. In contrast, cellular tumors in regions with pseudopalisading cells around necrosis and the perinecrotic zone tended to display metabolic cluster 1 status (Figure 7e). In the GSE182109 dataset, samples were collected from multiple regions, including the core area, the MRI-enhancing area, and the marginal area, for scRNA-seq analysis (Figure7 f). Based on the cell annotations provided in the original study, we calculated metabolic cluster scores for malignant cells using our classifier. Although statistical testing was not feasible due to the limited number of samples from the marginal areas (n=2), we observed that the proportion of metabolic cluster 1 malignant cells was significantly lower in the tumor edge area compared to the core and MRI-enhancing areas (Figure 7f). These results suggest that malignant cells are more likely to exhibit the metabolic cluster 1 phenotype in regions closer to the tumor core. Collectively, these findings highlight the spatial heterogeneity of glioma metabolism and demonstrate that our metabolic cluster accurately delineates the spatial metabolic characteristics of malignant cells at the single-cell resolution. This underscores the utility of our classifier in capturing both inter- and intra-tumoral embolic diversity.\u003c/p\u003e\n\u003cp\u003eBased on the intrinsic transcriptional characteristics of tumor cells and previous literature, glioma cells are typically classified into four subtypes: OPC-like, NPC-like, AC-like, and MES-like cells[2]. Trajectory analysis revealed a prominent transition starting from OPC- and NPC-like malignant cells, which correspond to the proneural (PN) subtype of GBM as classified by the TCGA dataset, to MES-like cells. This finding aligns well with previous findings (Figure 7g and 7h). Additionally, glycolytic activity in glioma cells was observed to increase progressively during the PN-to-MES transition (Figure 7i). Furthermore, hypoxia and angiogenesis within the tumor microenvironment were also found to be associated with the malignant transformation of tumor cells (Figure S11a and S11b), suggesting that tumor metabolism plays a critical role in driving the mesenchymal transformation of glioma cells. Based on these findings, we propose that metabolic reprogramming in tumor cells, coupled with in influence of the tumor microenvironment, significantly contributes to the malignant transformation of glioma cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetabolic characteristics of immune cells at single-cell level\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, we further explored the metabolic profiles of diverse immune cell populations. Utilizing singleR and leveraging insights from previous literature, we performed further dimensionality reduction and clustering analysis on myeloid cells (Figure 8a) and lymphocytes (Figure 8b). The marker genes of various immune cell subtypes were found to be specifically and highly expressed in their corresponding cell types (Figure 8c and 8d).\u003c/p\u003e\n\u003cp\u003eBased on the cell annotation, this study revealed distinct metabolic characteristics across various immune cells types. Among myeloid cells, monocytes and macrophages exhibited significantly higher glycolytic activity compared to microglia (monocytes vs microglia: \u003cem\u003eP\u003c/em\u003e=0.002, microglia vs macrophages:\u003cem\u003e\u0026nbsp;P\u003c/em\u003e=0.03 (top)) (Figure 8e). Neutrophils and DC displayed relatively high glycolytic activity but low oxidative phosphorylation activity (Figure 8e). Among lymphocytes, T regulatory cells (Tregs) demonstrated higher glycolytic activity than CD4+ T cells (Figure 8f, \u003cem\u003eP\u003c/em\u003e=0.034). While CD8+ T cells showed significantly higher oxidative phosphorylation activity compared to CD4+ T cells (\u003cem\u003eP\u003c/em\u003e=0.002) (Figure 8f). Additionally, plasma cells exhibited markedly higher oxidative phosphorylation and fatty acid oxidation activities than other lymphocyte subtypes (Figure 8f). To further investigate the metabolic profiles of T cells, we conducted metabolic enrichment analysis of CD4+ and CD8+ T cells. CD4+ T cells displayed significantly elevated activity in lysine degradation, sphingolipid metabolism, glyceride metabolism, and glycerophospholipid metabolism (Figure 8g). In contrast, CD8+ T cells exhibited higher activity in oxidative phosphorylation and cytochrome P450 metabolism. Overall, these findings highlight that different immune cell types exhibit distinct metabolic preferences (Figure 8g).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe metabolic heterogeneity of glioma is likely driven by multiple factors. Advances in sequencing technology have enabled the generation of large-scale genomic, single-cell, transcriptomic, and metabolic sequencing data for gliomas, providing an unprecedented opportunity to explore the complexity of glioma metabolism. In this study, the bulk-seq sequencing data of gliomas in the TCGA dataset were used to construct an energy-related metabolic classifier based on the activity of core metabolic pathways. Patients were divided into two prognostic clusters: those in cluster 1 were exhibited high pathway activity in glycolysis, the pentose phosphate pathway (PPP), and fatty acid oxidation (FAO), whereas those in cluster 2 displayed elevated activity in glutaminolysis. The robustness of the metabolic classifier was externally validated using the CGGA dataset. Additionally, the CPTAC dataset was employed to verify the rationality and robustness of the metabolic classifier across bulk RNA-seq, scRNA-seq, and tissue metabolite levels. Furthermore, we evaluated the metabolic classifier-related differences among the genome, clinical characteristics, molecular pathological composition, and immune infiltration. Our findings revealed that metabolic heterogeneity is primarily driven by tumor cells, although the tumor microenvironment also significantly influences the metabolic reprogramming of glioma cells. Collectively, this study enhances our understanding of metabolic heterogeneity in gliomas and provides a foundation for developing personalized therapeutic strategies for glioma patients.\u003c/p\u003e\u003cp\u003eIncreasingly, it is becoming clear that glioma cells have metabolic heterogeneity and preferences[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Understanding the metabolic reprogramming and progression of glioma cells is crucial for targeting glioma metabolism for treatment. For instance, Professor Evangelista's team classified pancreatic ductal adenocarcinoma into three metabolic subtypes based on tumor metabolites and demonstrated that the glycolytic subtype was strongly associated with the mesenchymalization of tumor cells[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Similarly, Professor Schaeffer and colleagues identified metabolic heterogeneity in high-grade ovarian cancer, primarily driven by variation in oxidative phosphorylation activity. Compared with previous studies, our research provides several novel insights: 1. By integrating multi-omics data (genomics, transcriptomics, and metabolomics), we established an energy-related metabolic classifier for gliomas using the TCGA dataset and validated its robustness across multiple external datasets. 2. Clinical characteristics, classic oncogenic pathways, and immune profiles between metabolic clusters were deeply explored. 3. We investigated the underlying causes of metabolic heterogeneity in gliomas at single-cell resolution and elucidated the relationship between glioma cells progression and metabolic reprogramming. 4. We further characterized the metabolic status of diverse immune cell populations at single-cell resolution.\u003c/p\u003e\u003cp\u003eGenome instability and mutations have long been recognized as fundamental hallmarks of the cancer[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. The Cancer Genome Atlas (TCGA) research network has constructed a comprehensive catalog of genomic alterations driving tumorigenesis[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Previous studies have further demonstrated that mutations in metabolism-related genes constitute a core component of metabolic reprogramming across various cancer types[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. The hypoxia conditions prevalent in glioma core regions induce significant adaptive changes in tumor cells. Consistent with published literature, hypoxia has been shown to promote genomic instability and epigenetic modifications, ultimately contributing to tumor progression, malignant transformation, and chemoresistance[\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Building upon these findings, our study revealed that patients in metabolic cluster 1 exhibited a higher mutation burden, and genes involved in metabolic pathways such as glycolysis underwent significantly more genomic mutations in the TCGA dataset. These observations suggest that metabolic gene mutations may represent an intrinsic driver of metabolic heterogeneity in gliomas.\u003c/p\u003e\u003cp\u003eFurther exploration identified OSMR as the top upregulated gene in cluster 1 through differential gene analysis. While previous studies have reported that the OSMR would promote the proliferation and the differential of the certain tumors. As for glioblastomas, macrophage-secreted OSM acts on OSMR in glioblastoma cells to promote proneural\u0026ndash;mesenchymal transformation[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. This study found that a decreased level of OSMR in the presence of OSM significantly reduced the expression of HK2 and LDHA, key genes in the aerobic glycolytic pathway, thereby affecting glycolytic activity. Furthermore, this study demonstrated that OSM/OSMR axis promotes malignant progression of glioma cells via JAK1/STAT3 signaling pathway. This finding is consistent with the prior reports showing that OSM binding to OSMR activates the JAK/STAT pathway, which plays a critical role in regulating glycolysis[\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. In orthotopic xenograft mouse models, knockdown of OSMR significantly prolonged survival compared to sh-NC group. Notably, patients in metabolic cluster 1 exhibited distinct immunosuppression in their immune profiles, indirectly suggesting that OSMR may reshape the immune microenvironment through metabolic reprogramming. The specific mechanisms by which OSMR mediates reprogramming and modulates the immune microenvironment in warrant further investigation.\u003c/p\u003e\u003cp\u003eMetabolism reprogramming, characterized by the adaptation of energy metabolism to support rapid cell growth and proliferation, has emerged as a critical hallmark of cancer[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. In 1964, Professor Warburg first reported that tumor cells use glycolysis for energy under aerobic conditions (the Warburg effect)[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. The heterogeneity of tumor cells leads to a complex metabolic pattern. In fact, PPP, FAO, and glutaminolysis pathways are also used to fuel proliferation[\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]. Recent researches have indicated that the activation of oncogenic pathways can lead to upregulation of metabolic pathways[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. For instance, under nutrient-deprived conditions (e.g., limited glucose, or glutamine availability), tumor cells activate the c-Myc pathway to modulate key metabolic enzymes (PHGDH, PSAT1 and PSPH) in the serine synthesis pathway, thereby maintaining cellular survival[\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. In our study, several classical oncogenic pathways were significantly activated in patients in cluster 1, such as pathways related to glycolysis metabolism. For example, Hippo and c-Myc exhibit significantly higher activity in cluster 1.\u003c/p\u003e\u003cp\u003eWe extended this study by leveraging single-cell transcriptomics from the CPTAC and GSE182109 datasets. First, we found that tumor metabolic heterogeneity was primarily driven by malignant cells, with trajectory analysis showing a gradual increase in glycolytic activity during the transition from OPC-like to MES-like phenotypes. Second, analysis of the IvyGAP and GSE182109 datasets revealed an increasing prevalence of metabolic cluster 1 status in tumor cells located nearer to the tumor core. To further explore tumor microenvironmental influences on metabolism, we assessed the relationships between metabolic activity and microenvironment factors using hypoxia and angiogenesis signatures as proxies for oxygen and nutrient availability. Comparison with CCLE glioma cell line data revealed significant differences between in vivo and in vitro metabolic profiles of glioma cells.\u003c/p\u003e\u003cp\u003eWe proposed several possible reasons behind this observed phenomenon: (1) A comprehensive understanding of tumor cell biology in vivo is required. The metabolic vulnerabilities of tumor cells in vivo cannot be fully replicated in cultured cell models. Tumor cells rely significantly on a rapid influx of nutrients, which can vary based on their location within the glioma microenvironment. Therefore, it is necessary to establish a model system recapitulating tumor microenvironments to explore tumor metabolism. (2) A deeper understanding of the impact of cellular interactions in the tumor microenvironment on tumor growth and progression is crucial. There is increasing evidence that immune responses are associated with changes in tissue metabolism, including nutrient consumption, increased oxygen consumption, and the production of reactive nitrogen and oxygen intermediates[\u003cspan additionalcitationids=\"CR53\" citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. In this study, the level of lactate in cluster 1 was significantly higher than that in cluster 2. Correspondingly, patients in cluster 1 expressed greater immune signatures and immune checkpoint genes. In a phase 3 clinical trial, the PD-1 inhibitor Nivolumab demonstrated no significant increase in side effects when combined with radiotherapy and Temozolomide. While it showed some anti-tumor activity, further studies are necessary to determine its clinical benefits[\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e]. Furthermore, this study found that patients in cluster 1 had a higher proportion of IDH wild-type and mesenchymal patients, indicating more severe immunosuppression in their tumor microenvironment, which suggests that these patients may benefit more from combined immunotherapy.\u003c/p\u003e\u003cp\u003eSimilar to tumor cells, immune cells demonstrate distinct metabolic heterogeneity. Under homeostatic conditions, immune cells maintain a quiescent state. However, during pathogenic challenges (e.g., infections, inflammation) or tumor development, immune cells undergo rapid activation and functional adaptation[\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e]. T cells exhibit dynamic metabolic reprogramming across different activation states. Na\u0026iuml;ve T cells in their resting state show limited proliferative activity and rely primarily on oxidative phosphorylation (OXPHOS) for energy generation. Following antigen stimulation, they differentiate into effector T cells and shift to a glycolytic activity, indicative of metabolic activation[\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e]. Activated neutrophils, M1-polarized macrophages, and iNOS-expressing dendritic cells (DCs) predominantly rely on glycolysis to meet energy demands[\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e]. Therefore, understanding immune cell metabolic profiles and their functional consequences may enable the development of targeted therapies to modulate immunosuppressive microenvironment and improve clinical outcomes.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study developed an energy metabolism-based classifier for gliomas with both prognostic and therapeutic potential. Metabolic reprogramming showed significant associations with both OPC-to-MES transition in glioma cells and immunosuppression within the tumor microenvironment. Multi-omics data, especially single-cell transcriptomes, provide us with a perspective to understand metabolism of gliomas at the single cell resolution, so as to better customize individualized treatment strategies for patients.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eTME: tumor microenvironment;\u003c/p\u003e\n\u003cp\u003eTCGA: the cancer genome atlas;\u003c/p\u003e\n\u003cp\u003eCGGA: Chinese glioma genome atlas;\u003c/p\u003e\n\u003cp\u003ePPP: the pentose phosphate pathway;\u003c/p\u003e\n\u003cp\u003eFAO: fatty acid oxidation (FAO);\u003c/p\u003e\n\u003cp\u003eMES: mesenchymal;\u003c/p\u003e\n\u003cp\u003eMDM: monocyte-derived macrophages;\u003c/p\u003e\n\u003cp\u003eCPTAC: Clinical Proteomic Tumor Analysis Consortium;\u003c/p\u003e\n\u003cp\u003eIvy GAP: Ivy glioblastoma atlas project;\u003c/p\u003e\n\u003cp\u003eMSigDB: Molecular Signatures Database;\u003c/p\u003e\n\u003cp\u003essGSEA: single sample Gene Set Enrichment Analysis;\u003c/p\u003e\n\u003cp\u003ePCA: principal Component Analysis;\u003c/p\u003e\n\u003cp\u003eCCP : cell cycle progression;\u003c/p\u003e\n\u003cp\u003eTILs : tumor-infiltrating lymphocytes;\u003c/p\u003e\n\u003cp\u003eCYT : immune cytolytic activity;\u003c/p\u003e\n\u003cp\u003eIFN: interferon;\u003c/p\u003e\n\u003cp\u003eCCLE: Cancer Cell Line Encyclopedia;\u003c/p\u003e\n\u003cp\u003eKEGG: Kyoto Encyclopedia of Genes and Genomes;\u003c/p\u003e\n\u003cp\u003eCNV: copy number variation;\u003c/p\u003e\n\u003cp\u003eWHO: world health organization;\u003c/p\u003e\n\u003cp\u003eIDH: isocitrate dehydrogenase;\u003c/p\u003e\n\u003cp\u003eMGMT promoter methylation: O6-methylguanine-DNA methyltransferase promoter methylation;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIQR: interquartile range:\u003c/p\u003e\n\u003cp\u003eDCs: Dendritic cells:\u003c/p\u003e\n\u003cp\u003eOligs: oligodendrocytes;\u003c/p\u003e\n\u003cp\u003eGLUT: glucose transporters;\u003c/p\u003e\n\u003cp\u003eG-6-P: Glucose 6-phosphate;\u003c/p\u003e\n\u003cp\u003eMCTs: Monocarboxylate transporters;\u003c/p\u003e\n\u003cp\u003eLDH: lactate dehydrogenase;\u003c/p\u003e\n\u003cp\u003eTCA: tricarboxylic acid cycle;\u003c/p\u003e\n\u003cp\u003e\u0026alpha;-KG: \u0026alpha;-Ketoglutaric acid;\u003c/p\u003e\n\u003cp\u003eG-6-P:\u0026nbsp;\u003ca href=\"https://www.bing.com/ck/a?!\u0026\u0026p=a79a198089e21ab3JmltdHM9MTcxOTQ0NjQwMCZpZ3VpZD0yYWEwMjA3ZS00YWFkLTZkMmMtMmY5Mi0zNDMzNGI3ZjZjNDgmaW5zaWQ9NTIxMw\u0026ptn=3\u0026ver=2\u0026hsh=3\u0026fclid=2aa0207e-4aad-6d2c-2f92-34334b7f6c48\u0026psq=6-PG%e7%9a%84%e5%85%a8%e7%a7%b0\u0026u=a1aHR0cHM6Ly9wdWJjaGVtLm5jYmkubmxtLm5paC5nb3YvY29tcG91bmQvNi1QaG9zcGhvZ2x1Y29uaWMtYWNpZA\u0026ntb=1\" target=\"_blank\"\u003eglucose-6-phosphatase\u0026nbsp;\u003c/a\u003e;\u003c/p\u003e\n\u003cp\u003eR-5-P: Ribose 5-phosphate;\u003c/p\u003e\n\u003cp\u003eNADPH: nicotinamide adenine dinucleotide phosphate;\u003c/p\u003e\n\u003cp\u003eLac: lactate;\u003c/p\u003e\n\u003cp\u003eTAMs: tumor associated microglia or macrophages;\u003c/p\u003e\n\u003cp\u003eSMCs: smooth muscle cells;\u003c/p\u003e\n\u003cp\u003eAC-like: astrocyte like;\u003c/p\u003e\n\u003cp\u003eMES-like: mesenchymal like;\u003c/p\u003e\n\u003cp\u003eNPC-like: neural progeneitor;\u003c/p\u003e\n\u003cp\u003eOPC-like: Oligdendrocytes like\u003c/p\u003e\n\u003cp\u003eCThbv: cellular tumor of hyperplastic blood vessels;\u003c/p\u003e\n\u003cp\u003eCTmvp: cellular tumor of microvascular proliferation;\u003c/p\u003e\n\u003cp\u003eCTpan: cellular tumor of pseudopalisading cell around necrosis;\u003c/p\u003e\n\u003cp\u003eCTpnz: cellular tumor of perinecrotic zone;\u003c/p\u003e\n\u003cp\u003eIT: infiltrating area;\u003c/p\u003e\n\u003cp\u003eLE: leading edge;\u003c/p\u003e\n\u003cp\u003eCTLs: Cytotoxic T lymphocytes.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u0026nbsp;\u003c/strong\u003eAll animal studies were approved by the Institutional Animal Care and Use Committee of the First Affiliated Hospital of the University of Science and Technology of China.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u0026nbsp;\u003c/strong\u003eAll authors consent to the publication of this manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and material:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u0026nbsp;\u003c/strong\u003eThe authors declare that they have no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported in part by grants from the National Natural Science Foundation of China [82273281].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis study was supported in part by grants from the National Natural Science Foundation of China [82203693].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eQiang Zhu and Chaoshi Niu conceived and designed the study. Qiang Zhu and Wanxiang Niu were responsible for data acquisition and analysis. Qiang Zhu, Chengkun Ye, and Wanxiang Niu undertook the interpretation of data. Qiang Zhu prepared the original draft of the manuscript, which was further reviewed and edited by Qiang Zhu, Maolin Mu, and Chaoshi Niu. Chaoshi Niu provided supervision throughout the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u003c/strong\u003e Not applicable\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eHao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, Satija R: \u003cstrong\u003eDictionary learning for integrative, multimodal and scalable single-cell analysis\u003c/strong\u003e. \u003cem\u003eNat Biotechnol \u003c/em\u003e2024, \u003cstrong\u003e42\u003c/strong\u003e(2):293-304.\u003c/li\u003e\n\u003cli\u003eNeftel C, Laffy J, Filbin MG, Hara T, Shore ME, Rahme GJ, Richman AR, Silverbush D, Shaw ML, Hebert CM\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAn Integrative Model of Cellular States, Plasticity, and Genetics for Glioblastoma\u003c/strong\u003e. \u003cem\u003eCell \u003c/em\u003e2019, \u003cstrong\u003e178\u003c/strong\u003e(4):835-849.e821.\u003c/li\u003e\n\u003cli\u003eVerhaak RGW, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, Miller CR, Ding L, Golub T, Mesirov JP\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIntegrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2010, \u003cstrong\u003e17\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eHanahan D: \u003cstrong\u003eHallmarks of Cancer: New Dimensions\u003c/strong\u003e. \u003cem\u003eCancer discovery \u003c/em\u003e2022, \u003cstrong\u003e12\u003c/strong\u003e(1):31-46.\u003c/li\u003e\n\u003cli\u003ePeng X, Chen Z, Farshidfar F, Xu X, Lorenzi PL, Wang Y, Cheng F, Tan L, Mojumdar K, Du D\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eMolecular Characterization and Clinical Relevance of Metabolic Expression Subtypes in Human Cancers\u003c/strong\u003e. \u003cem\u003eCell reports \u003c/em\u003e2018, \u003cstrong\u003e23\u003c/strong\u003e(1):255-269.e254.\u003c/li\u003e\n\u003cli\u003eBj\u0026ouml;rkblom B, Wibom C, Eriksson M, Bergenheim AT, Sj\u0026ouml;berg RL, Jonsson P, Br\u0026auml;nnstr\u0026ouml;m T, Antti H, Sandstr\u0026ouml;m M, Melin B: \u003cstrong\u003eDistinct metabolic hallmarks of WHO classified adult glioma subtypes\u003c/strong\u003e. \u003cem\u003eNeuro Oncol \u003c/em\u003e2022, \u003cstrong\u003e24\u003c/strong\u003e(9):1454-1468.\u003c/li\u003e\n\u003cli\u003eUpadhyayula PS, Higgins DM, Mela A, Banu M, Dovas A, Zandkarimi F, Patel P, Mahajan A, Humala N, Nguyen TTT\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eDietary restriction of cysteine and methionine sensitizes gliomas to ferroptosis and induces alterations in energetic metabolism\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e(1):1187.\u003c/li\u003e\n\u003cli\u003eJin N, Bi A, Lan X, Xu J, Wang X, Liu Y, Wang T, Tang S, Zeng H, Chen Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIdentification of metabolic vulnerabilities of receptor tyrosine kinases-driven cancer\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2019, \u003cstrong\u003e10\u003c/strong\u003e(1):2701.\u003c/li\u003e\n\u003cli\u003eJiang Z, Liu Z, Li M, Chen C, Wang X: \u003cstrong\u003eIncreased glycolysis correlates with elevated immune activity in tumor immune microenvironment\u003c/strong\u003e. \u003cem\u003eEBioMedicine \u003c/em\u003e2019, \u003cstrong\u003e42\u003c/strong\u003e:431-442.\u003c/li\u003e\n\u003cli\u003eLeBlanc VG, Trinh DL, Aslanpour S, Hughes M, Livingstone D, Jin D, Ahn BY, Blough MD, Cairncross JG, Chan JA\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSingle-cell landscapes of primary glioblastomas and matched explants and cell lines show variable retention of inter- and intratumor heterogeneity\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2022, \u003cstrong\u003e40\u003c/strong\u003e(4).\u003c/li\u003e\n\u003cli\u003eGentric G, Kieffer Y, Mieulet V, Goundiam O, Bonneau C, Nemati F, Hurbain I, Raposo G, Popova T, Stern M-H\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ePML-Regulated Mitochondrial Metabolism Enhances Chemosensitivity in Human Ovarian Cancers\u003c/strong\u003e. \u003cem\u003eCell metabolism \u003c/em\u003e2019, \u003cstrong\u003e29\u003c/strong\u003e(1):156-173.e110.\u003c/li\u003e\n\u003cli\u003eBidkhori G, Benfeitas R, Klevstig M, Zhang C, Nielsen J, Uhlen M, Boren J, Mardinoglu A: \u003cstrong\u003eMetabolic network-based stratification of hepatocellular carcinoma reveals three distinct tumor subtypes\u003c/strong\u003e. \u003cem\u003eProc Natl Acad Sci USA \u003c/em\u003e2018, \u003cstrong\u003e115\u003c/strong\u003e(50):E11874-E11883.\u003c/li\u003e\n\u003cli\u003eKarayama M, Masuda J, Mori K, Yasui H, Hozumi H, Suzuki Y, Furuhashi K, Fujisawa T, Enomoto N, Nakamura Y\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eComprehensive assessment of multiple tryptophan metabolites as potential biomarkers for immune checkpoint inhibitors in patients with non-small cell lung cancer\u003c/strong\u003e. \u003cem\u003eClinical \u0026amp; translational oncology : official publication of the Federation of Spanish Oncology Societies and of the National Cancer Institute of Mexico \u003c/em\u003e2021, \u003cstrong\u003e23\u003c/strong\u003e(2):418-423.\u003c/li\u003e\n\u003cli\u003eYan Y, Chang L, Tian H, Wang L, Zhang Y, Yang T, Li G, Hu W, Shah K, Chen G, Guo Y: \u003cstrong\u003e1-Pyrroline-5-carboxylate released by prostate Cancer cell inhibit T cell proliferation and function by targeting SHP1/cytochrome c oxidoreductase/ROS Axis\u003c/strong\u003e. \u003cem\u003eJournal for immunotherapy of cancer \u003c/em\u003e2018, \u003cstrong\u003e6\u003c/strong\u003e(1):148.\u003c/li\u003e\n\u003cli\u003eDe Leo A, Ugolini A, Yu X, Scirocchi F, Scocozza D, Peixoto B, Pace A, D\u0026apos;Angelo L, Liu JKC, Etame AB\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eGlucose-driven histone lactylation promotes the immunosuppressive activity of monocyte-derived macrophages in glioblastoma\u003c/strong\u003e. \u003cem\u003eImmunity \u003c/em\u003e2024, \u003cstrong\u003e57\u003c/strong\u003e(5):1105-1123.e1108.\u003c/li\u003e\n\u003cli\u003eHurley HJ, Dewald H, Rothkopf ZS, Singh S, Jenkins F, Deb P, De S, Barnes BJ, Fitzgerald-Bocarsly P: \u003cstrong\u003eFrontline Science: AMPK regulates metabolic reprogramming necessary for interferon production in human plasmacytoid dendritic cells\u003c/strong\u003e. \u003cem\u003eJournal of leukocyte biology \u003c/em\u003e2021, \u003cstrong\u003e109\u003c/strong\u003e(2):299-308.\u003c/li\u003e\n\u003cli\u003eWang LB, Karpova A, Gritsenko MA, Kyle JE, Cao S, Li Y, Rykunov D, Colaprico A, Rothstein JH, Hong R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eProteogenomic and metabolomic characterization of human glioblastoma\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2021, \u003cstrong\u003e39\u003c/strong\u003e(4):509-528.e520.\u003c/li\u003e\n\u003cli\u003ePuchalski RB, Shah N, Miller J, Dalley R, Nomura SR, Yoon JG, Smith KA, Lankerovich M, Bertagnolli D, Bickley K\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAn anatomic transcriptional atlas of human glioblastoma\u003c/strong\u003e. \u003cem\u003eScience \u003c/em\u003e2018, \u003cstrong\u003e360\u003c/strong\u003e(6389):660-663.\u003c/li\u003e\n\u003cli\u003eAbdelfattah N, Kumar P, Wang C, Leu JS, Flynn WF, Gao R, Baskin DS, Pichumani K, Ijare OB, Wood SL\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSingle-cell analysis of human glioma and immune cells identifies S100A4 as an immunotherapy target\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e(1):767.\u003c/li\u003e\n\u003cli\u003eDe Berardinis RJ, Chandel NS: \u003cstrong\u003eFundamentals of cancer metabolism\u003c/strong\u003e. \u003cem\u003eScience advances \u003c/em\u003e2016, \u003cstrong\u003e2\u003c/strong\u003e(5).\u003c/li\u003e\n\u003cli\u003eKarasinska JM, Topham JT, Kalloger SE, Jang GH, Denroche RE, Culibrk L, Williamson LM, Wong HL, Lee MKC, O\u0026apos;Kane GM\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAltered Gene Expression along the Glycolysis-Cholesterol Synthesis Axis Is Associated with Outcome in Pancreatic Cancer\u003c/strong\u003e. \u003cem\u003eClin Cancer Res \u003c/em\u003e2020, \u003cstrong\u003e26\u003c/strong\u003e(1):135-146.\u003c/li\u003e\n\u003cli\u003eLiberzon A, Birger C, Thorvaldsd\u0026oacute;ttir H, Ghandi M, Mesirov JP, Tamayo P: \u003cstrong\u003eThe Molecular Signatures Database (MSigDB) hallmark gene set collection\u003c/strong\u003e. \u003cem\u003eCell systems \u003c/em\u003e2015, \u003cstrong\u003e1\u003c/strong\u003e(6):417-425.\u003c/li\u003e\n\u003cli\u003eH\u0026auml;nzelmann S, Castelo R, Guinney J: \u003cstrong\u003eGSVA: Gene set variation analysis for microarray and RNA-Seq data\u003c/strong\u003e. \u003cem\u003eBMC Bioinformatics \u003c/em\u003e2013, \u003cstrong\u003e14\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eKumar A, Coleman I, Morrissey C, Zhang X, True LD, Gulati R, Etzioni R, Bolouri H, Montgomery B, White T\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSubstantial interindividual and limited intraindividual genomic diversity among tumors from men with metastatic prostate cancer\u003c/strong\u003e. \u003cem\u003eNat Med \u003c/em\u003e2016, \u003cstrong\u003e22\u003c/strong\u003e(4):369-378.\u003c/li\u003e\n\u003cli\u003eYoshihara K, Shahmoradgoli M, Mart\u0026iacute;nez E, Vegesna R, Kim H, Torres-Garcia W, Trevi\u0026ntilde;o V, Shen H, Laird PW, Levine DA\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eInferring tumour purity and stromal and immune cell admixture from expression data\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2013, \u003cstrong\u003e4\u003c/strong\u003e:2612.\u003c/li\u003e\n\u003cli\u003eBecht E, Giraldo NA, Lacroix L, Buttard B, Elarouci N, Petitprez F, Selves J, Laurent-Puig P, Saut\u0026egrave;s-Fridman C, Fridman WH, de Reyni\u0026egrave;s A: \u003cstrong\u003eEstimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression\u003c/strong\u003e. \u003cem\u003eGenome Biol \u003c/em\u003e2016, \u003cstrong\u003e17\u003c/strong\u003e(1):218.\u003c/li\u003e\n\u003cli\u003eKorsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh P-r, Raychaudhuri S: \u003cstrong\u003eFast, sensitive and accurate integration of single-cell data with Harmony\u003c/strong\u003e. \u003cem\u003eNat Methods \u003c/em\u003e2019, \u003cstrong\u003e16\u003c/strong\u003e(12):1289-1296.\u003c/li\u003e\n\u003cli\u003eAran D, Looney AP, Liu L, Wu E, Fong V, Hsu A, Chak S, Naikawadi RP, Wolters PJ, Abate AR\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eReference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage\u003c/strong\u003e. \u003cem\u003eNature immunology \u003c/em\u003e2019, \u003cstrong\u003e20\u003c/strong\u003e(2):163-172.\u003c/li\u003e\n\u003cli\u003eHu C, Li T, Xu Y, Zhang X, Li F, Bai J, Chen J, Jiang W, Yang K, Ou Q\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eCellMarker 2.0: an updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data\u003c/strong\u003e. \u003cem\u003eNucleic Acids Res \u003c/em\u003e2023, \u003cstrong\u003e51\u003c/strong\u003e(D1):D870-d876.\u003c/li\u003e\n\u003cli\u003eFranz\u0026eacute;n O, Gan L-M, Bj\u0026ouml;rkegren JLM: \u003cstrong\u003ePanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data\u003c/strong\u003e. \u003cem\u003eDatabase \u003c/em\u003e2019, \u003cstrong\u003e2019\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eXiong A, Zhang J, Chen Y, Zhang Y, Yang F: \u003cstrong\u003eIntegrated single-cell transcriptomic analyses reveal that GPNMB-high macrophages promote PN-MES transition and impede T cell activation in GBM\u003c/strong\u003e. \u003cem\u003eEBioMedicine \u003c/em\u003e2022, \u003cstrong\u003e83\u003c/strong\u003e:104239.\u003c/li\u003e\n\u003cli\u003eXiao Z, Dai Z, Locasale JW: \u003cstrong\u003eMetabolic landscape of the tumor microenvironment at single cell resolution\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2019, \u003cstrong\u003e10\u003c/strong\u003e(1):3763.\u003c/li\u003e\n\u003cli\u003eTrapnell C, Cacchiarelli D, Grimsby J, Pokharel P, Li S, Morse M, Lennon NJ, Livak KJ, Mikkelsen TS, Rinn JL: \u003cstrong\u003eThe dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells\u003c/strong\u003e. \u003cem\u003eNat Biotechnol \u003c/em\u003e2014, \u003cstrong\u003e32\u003c/strong\u003e(4):381-386.\u003c/li\u003e\n\u003cli\u003eSanchez-Vega F, Mina M, Armenia J, Chatila WK, Luna A, La KC, Dimitriadoy S, Liu DL, Kantheti HS, Saghafinia S\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eOncogenic Signaling Pathways in The Cancer Genome Atlas\u003c/strong\u003e. \u003cem\u003eCell \u003c/em\u003e2018, \u003cstrong\u003e173\u003c/strong\u003e(2):321-337.e310.\u003c/li\u003e\n\u003cli\u003eJahani-Asl A, Yin H, Soleimani VD, Haque T, Luchman HA, Chang NC, Sincennes M-C, Puram SV, Scott AM, Lorimer IAJ\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eControl of glioblastoma tumorigenesis by feed-forward cytokine signaling\u003c/strong\u003e. \u003cem\u003eNat Neurosci \u003c/em\u003e2016, \u003cstrong\u003e19\u003c/strong\u003e(6):798-806.\u003c/li\u003e\n\u003cli\u003eHara T, Chanoch-Myers R, Mathewson ND, Myskiw C, Atta L, Bussema L, Eichhorn SW, Greenwald AC, Kinker GS, Rodman C\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eInteractions between cancer cells and immune cells drive transitions to mesenchymal-like states in glioblastoma\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2021, \u003cstrong\u003e39\u003c/strong\u003e(6).\u003c/li\u003e\n\u003cli\u003eKim J, DeBerardinis RJ: \u003cstrong\u003eMechanisms and Implications of Metabolic Heterogeneity in Cancer\u003c/strong\u003e. \u003cem\u003eCell metabolism \u003c/em\u003e2019, \u003cstrong\u003e30\u003c/strong\u003e(3):434-446.\u003c/li\u003e\n\u003cli\u003eDaemen A, Peterson D, Sahu N, McCord R, Du X, Liu B, Kowanetz K, Hong R, Moffat J, Gao M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eMetabolite profiling stratifies pancreatic ductal adenocarcinomas into subtypes with distinct sensitivities to metabolic inhibitors\u003c/strong\u003e. \u003cem\u003eProc Natl Acad Sci U S A \u003c/em\u003e2015, \u003cstrong\u003e112\u003c/strong\u003e(32):E4410-4417.\u003c/li\u003e\n\u003cli\u003eDrews RM, Hernando B, Tarabichi M, Haase K, Lesluyes T, Smith PS, Morrill Gavarr\u0026oacute; L, Couturier DL, Liu L, Schneider M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA pan-cancer compendium of chromosomal instability\u003c/strong\u003e. \u003cem\u003eNature \u003c/em\u003e2022, \u003cstrong\u003e606\u003c/strong\u003e(7916):976-983.\u003c/li\u003e\n\u003cli\u003eNetwork CGAR: \u003cstrong\u003eComprehensive genomic characterization defines human glioblastoma genes and core pathways\u003c/strong\u003e. \u003cem\u003eNature \u003c/em\u003e2008, \u003cstrong\u003e455\u003c/strong\u003e(7216):1061-1068.\u003c/li\u003e\n\u003cli\u003eVander Heiden MG, DeBerardinis RJ: \u003cstrong\u003eUnderstanding the Intersections between Metabolism and Cancer Biology\u003c/strong\u003e. \u003cem\u003eCell \u003c/em\u003e2017, \u003cstrong\u003e168\u003c/strong\u003e(4):657-669.\u003c/li\u003e\n\u003cli\u003eGarofano L, Migliozzi S, Oh YT, D\u0026apos;Angelo F, Najac RD, Ko A, Frangaj B, Caruso FP, Yu K, Yuan J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ePathway-based classification of glioblastoma uncovers a mitochondrial subtype with therapeutic vulnerabilities\u003c/strong\u003e. \u003cem\u003eNature cancer \u003c/em\u003e2021, \u003cstrong\u003e2\u003c/strong\u003e(2):141-156.\u003c/li\u003e\n\u003cli\u003eBhandari V, Li CH, Bristow RG, Boutros PC, Aaltonen LA, Abascal F, Abeshouse A, Aburatani H, Adams DJ, Agrawal N\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eDivergent mutational processes distinguish hypoxic and normoxic tumours\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2020, \u003cstrong\u003e11\u003c/strong\u003e(1):737.\u003c/li\u003e\n\u003cli\u003eChen Z, Hambardzumyan D: \u003cstrong\u003eMacrophage-tumor cell intertwine drives the transition into a mesenchymal-like cellular state of glioblastoma\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2021, \u003cstrong\u003e39\u003c/strong\u003e(6):743-745.\u003c/li\u003e\n\u003cli\u003eMasjedi A, Hajizadeh F, Beigi Dargani F, Beyzai B, Aksoun M, Hojjat-Farsangi M, Zekiy A, Jadidi-Niaragh F: \u003cstrong\u003eOncostatin M: A mysterious cytokine in cancers\u003c/strong\u003e. \u003cem\u003eInternational Immunopharmacology \u003c/em\u003e2021, \u003cstrong\u003e90\u003c/strong\u003e:107158.\u003c/li\u003e\n\u003cli\u003eLi YJ, Zhang C, Martincuks A, Herrmann A, Yu H: \u003cstrong\u003eSTAT proteins in cancer: orchestration of metabolism\u003c/strong\u003e. \u003cem\u003eNat Rev Cancer \u003c/em\u003e2023, \u003cstrong\u003e23\u003c/strong\u003e(3):115-134.\u003c/li\u003e\n\u003cli\u003eWarburg O: \u003cstrong\u003eOn the origin of cancer cells\u003c/strong\u003e. \u003cem\u003eScience \u003c/em\u003e1956, \u003cstrong\u003e123\u003c/strong\u003e(3191):309-314.\u003c/li\u003e\n\u003cli\u003eBoroughs LK, DeBerardinis RJ: \u003cstrong\u003eMetabolic pathways promoting cancer cell survival and growth\u003c/strong\u003e. \u003cem\u003eNature cell biology \u003c/em\u003e2015, \u003cstrong\u003e17\u003c/strong\u003e(4):351-359.\u003c/li\u003e\n\u003cli\u003eAltman BJ, Stine ZE, Dang CV: \u003cstrong\u003eFrom Krebs to clinic: glutamine metabolism to cancer therapy\u003c/strong\u003e. \u003cem\u003eNat Rev Cancer \u003c/em\u003e2016, \u003cstrong\u003e16\u003c/strong\u003e(10):619-634.\u003c/li\u003e\n\u003cli\u003eMartinez-Outschoorn UE, Peiris-Pag\u0026eacute;s M, Pestell RG, Sotgia F, Lisanti MP: \u003cstrong\u003eCancer metabolism: a therapeutic perspective\u003c/strong\u003e. \u003cem\u003eNature reviews Clinical oncology \u003c/em\u003e2017, \u003cstrong\u003e14\u003c/strong\u003e(1):11-31.\u003c/li\u003e\n\u003cli\u003eSun L, Song L, Wan Q, Wu G, Li X, Wang Y, Wang J, Liu Z, Zhong X, He X\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ecMyc-mediated activation of serine biosynthesis pathway is critical for cancer progression under nutrient deprivation conditions\u003c/strong\u003e. \u003cem\u003eCell research \u003c/em\u003e2015, \u003cstrong\u003e25\u003c/strong\u003e(4):429-444.\u003c/li\u003e\n\u003cli\u003eTerry S, Engelsen AST, Buart S, Elsayed WS, Venkatesh GH, Chouaib S: \u003cstrong\u003eHypoxia-driven intratumor heterogeneity and immune evasion\u003c/strong\u003e. \u003cem\u003eCancer Lett \u003c/em\u003e2020, \u003cstrong\u003e492\u003c/strong\u003e:1-10.\u003c/li\u003e\n\u003cli\u003eHuang B, Song BL, Xu C: \u003cstrong\u003eCholesterol metabolism in cancer: mechanisms and therapeutic opportunities\u003c/strong\u003e. \u003cem\u003eNature metabolism \u003c/em\u003e2020, \u003cstrong\u003e2\u003c/strong\u003e(2):132-141.\u003c/li\u003e\n\u003cli\u003eChen B, Gao A, Tu B, Wang Y, Yu X, Wang Y, Xiu Y, Wang B, Wan Y, Huang Y: \u003cstrong\u003eMetabolic modulation via mTOR pathway and anti-angiogenesis remodels tumor microenvironment using PD-L1-targeting codelivery\u003c/strong\u003e. \u003cem\u003eBiomaterials \u003c/em\u003e2020, \u003cstrong\u003e255\u003c/strong\u003e:120187.\u003c/li\u003e\n\u003cli\u003eOmuro A, Reardon DA, Sampson JH, Baehring J, Sahebjam S, Cloughesy TF, Chalamandaris AG, Potter V, Butowski N, Lim M: \u003cstrong\u003eNivolumab plus radiotherapy with or without temozolomide in newly diagnosed glioblastoma: Results from exploratory phase I cohorts of CheckMate 143\u003c/strong\u003e. \u003cem\u003eNeuro-oncology advances \u003c/em\u003e2022, \u003cstrong\u003e4\u003c/strong\u003e(1):vdac025.\u003c/li\u003e\n\u003cli\u003eHao S, Yan KK, Ding L, Qian C, Chi H, Yu J: \u003cstrong\u003eNetwork Approaches for Dissecting the Immune System\u003c/strong\u003e. \u003cem\u003eiScience \u003c/em\u003e2020, \u003cstrong\u003e23\u003c/strong\u003e(8):101354.\u003c/li\u003e\n\u003cli\u003ePearce EL, Poffenberger MC, Chang CH, Jones RG: \u003cstrong\u003eFueling immunity: insights into metabolism and lymphocyte function\u003c/strong\u003e. \u003cem\u003eScience \u003c/em\u003e2013, \u003cstrong\u003e342\u003c/strong\u003e(6155):1242454.\u003c/li\u003e\n\u003cli\u003ePearce EL, Pearce EJ: \u003cstrong\u003eMetabolic pathways in immune cell activation and quiescence\u003c/strong\u003e. \u003cem\u003eImmunity \u003c/em\u003e2013, \u003cstrong\u003e38\u003c/strong\u003e(4):633-643.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-translational-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jtrm","sideBox":"Learn more about [Journal of Translational Medicine](http://translational-medicine.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jtrm/default.aspx","title":"Journal of Translational Medicine","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Gliomas, metabolic reprogramming, metabolic heterogeneity, unsupervised classification, immunity metabolism","lastPublishedDoi":"10.21203/rs.3.rs-7196875/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7196875/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe tumor microenvironment (TME) of glioma harbors diverse cell types; however, cell metabolic heterogeneity remains to be explored. This study aims to characterize the metabolic features of different cell types in the TME by integrating multiple datasets, including genomics, bulk and single-cell transcriptomics, and metabolomics.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e\u003cp\u003eUnsupervised machine learning was used to construct an energy metabolic classifier based on the metabolic pathways identified from bulk RNA-seq of gliomas in the TCGA dataset. The classifier was externally validated using multiple datasets, including genomics, bulk RNA-seq, snRNA-seq, and the metabolomics data. Furthermore, metabolic heterogeneity associated with the classifier was further characterized at single-cell resolution.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe energy metabolism-based classifier stratified patients into two prognostic clusters: patients in cluster 1 were characterized by high pathway activity of glycolysis, the pentose phosphate pathway (PPP), and fatty acid oxidation (FAO), whereas patients in cluster 2 exhibited higher activity in glutaminolysis. This metabolic classifier revealed both intratumoral and intertumoral metabolic heterogeneity, and the complexity was further validated by the metabolomics profiling and snRNA-seq data from the CPTAC dataset. Notably, OSMR, highly expressed in cluster 1, showed significant co-expression with key glycolytic enzyme genes. The OSM/OSMR/JAK1/STAT3 axis potently drives malignant progression of glioma cells, specially enhancing their invasive and migratory capabilities. Single-cell resolution analyses demonstrated that tumor metabolic heterogeneity is primarily driven by malignant cells rather than non-malignant components, while tumor microenvironment (TME) factors were also found to modulate malignant cell metabolism. Significantly, glycolytic activity in glioma cells increased during the phenotypic transition from PN (proneural) to MES (mesenchymal), with cluster 1 metabolic phenotypes predominating in the tumor core. Compared to cluster 2, cluster 1 patients exhibited higher mRNA expression of immunosuppressive checkpoint genes, which correlated with pronounced immunosuppression in the TME. Furthermore, various immune cells demonstrated distinct metabolic preferences at single-cell resolution.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis study developed an energy metabolic-based classifier for gliomas with prognostic and therapeutic potential. Metabolic reprogramming was linked with the PN-to-MES transition of glioma cells and immunosuppression in the tumor microenvironment. Multi-omics data, especially snRNA-seq, offered insights into metabolism heterogeneity at single-cell resolution, enabling personalized treatment strategies.\u003c/p\u003e","manuscriptTitle":"Integrative multi-omics Analysis Proposes a Metabolic Classification of Gliomas: Distinct Metabolic States, Immune Infiltration, and Prognosis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-30 04:58:38","doi":"10.21203/rs.3.rs-7196875/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2025-09-02T15:54:52+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-27T22:07:16+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-26T16:08:51+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Translational Medicine","date":"2025-07-23T09:22:44+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-translational-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jtrm","sideBox":"Learn more about [Journal of Translational Medicine](http://translational-medicine.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jtrm/default.aspx","title":"Journal of Translational Medicine","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"29341dd9-78f8-4b1c-a39b-24382527cae0","owner":[],"postedDate":"July 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-05T16:05:59+00:00","versionOfRecord":{"articleIdentity":"rs-7196875","link":"https://doi.org/10.1186/s12967-025-07602-z","journal":{"identity":"journal-of-translational-medicine","isVorOnly":false,"title":"Journal of Translational Medicine"},"publishedOn":"2025-12-30 15:58:07","publishedOnDateReadable":"December 30th, 2025"},"versionCreatedAt":"2025-07-30 04:58:38","video":"","vorDoi":"10.1186/s12967-025-07602-z","vorDoiUrl":"https://doi.org/10.1186/s12967-025-07602-z","workflowStages":[]},"version":"v1","identity":"rs-7196875","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7196875","identity":"rs-7196875","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}