Skip to main content
NCBI home page
Springer logoLink to Springer
. 2026 Jul 9;26(1):184. doi: 10.1007/s10142-026-01973-2

Integrative bulk and single-cell analyses implicate ASL in MIF-associated myeloid programs in glioblastoma

Wenjing Zhao 1, Shenglan Li 1, Wenbin Li 1,✉
PMCID: PMC13346236  PMID: 42420626

Abstract

Arginine metabolism is implicated in glioblastoma (GBM) progression and immune regulation, but prognostic biomarkers with defined cellular context remain insufficiently characterized. Eighteen arginine metabolism-related genes were analyzed in TCGA GBM. LASSO Cox regression with tenfold cross-validation and subsequent univariate/multivariate Cox models were used to identify prognostic genes. Patients were stratified by optimal cutoffs for Kaplan–Meier analysis, with validation in CGGA and GEO datasets. Predictive performance was evaluated by ROC and decision curve analysis (DCA). scRNA-seq data from five GBM samples were processed using Seurat, arginine-metabolism activity was quantified at single-cell resolution. Myeloid cells were subsetted for subclustering, with state dynamics examined by pseudotime, RNA velocity, and CellChat. ASL, FAH, and NAGS were prioritized as prognostic candidates, with ASL characterized in the TCGA cohort as a prognostic indicator explicitly associated with high-risk molecular subtypes such as IDH-wildtype and 1p/19q non-codeletion.. High expression of ASL predicted shorter overall survival in TCGA and was further validated in CGGA and GEO datasets. ASL showed the best clinical prognosis discrimination and net benefit in DCA. In scRNA-seq, 38,263 high-quality cells were used for downstream analyses. ASL expression and arginine-metabolism activity were enriched in monocyte/macrophage populations. Trajectory analyses placed MKI67⁺ monocytes upstream and showed ASL decreasing along pseudotime. RNA velocity supported directional transitions. CellChat highlighted macrophage migration inhibitory factor (MIF)-, MHC-II-, and SPP1-associated signaling within myeloid networks, consistent with Spearman correlation patterns between ASL and MIF-axis genes. ASL is identified as an arginine metabolism-associated prognostic signal that is particularly tied to high-risk molecular subgroups of GBM, and exhibits strong correlations with myeloid-centered immune programs and MIF-related communication features in the tumor microenvironment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10142-026-01973-2.

Keywords: Arginine metabolism, Glioma, Glioblastoma, ASL, Monocyte, Immune microenvironment

Introduction

Primary brain tumors predominantly manifest as gliomas, accounting for 80–85% of cases in adults (Ostrom et al. 2018). Among these, lower-grade gliomas exhibit highly variable clinical trajectories that often defy histological classification, with frequent progression to glioblastoma (GBM) (Cancer Genome Atlas Research 2015). This malignant transformation is driven by several key risk factors, especially IDH wild-type status, CDKN2A/B homozygous deletion, and TERT promoter mutations, which are associated with significantly shorter progression-free survival (PFS) (Louis et al. 2021). Additionally, clinical parameters such as advanced age at diagnosis, suboptimal extent of resection, and spontaneous growth rate on serial imaging further predispose these tumors to high-grade progression (Ius et al. 2012). GBM represents the most aggressive and prevalent primary malignant brain tumor in adults, characterized by its highly invasive nature and poor prognosis (Ostrom et al. 2018; Paolillo et al. 2018). Despite the introduction of tumor treating fields (TTF), the standard therapy—comprising maximal safe surgical resection followed by concurrent chemoradiotherapy with temozolomide—has remained largely unchanged for two decades (Tan et al. 2020; Stupp et al. 2005). Consequently, the median overall survival remains dismally low at approximately 15 months (Subramanian et al. 2005; Chinot et al. 2014; Taal et al. 2014), underscoring the urgent need for novel therapeutic strategies. The therapeutic resistance and high recurrence rate of GBM are primarily attributed to its highly invasive nature, which precludes complete surgical resection, and to the dynamic remodeling of the tumor microenvironment (TME).

The TME of malignant brain tumors comprises diverse cellular components that exhibit both pro- and anti-tumorigenic properties (Whiteside 2008). Within the complex cellular composition of the GBM microenvironment, tumor-associated myeloid cell and microglia constitute the predominant immune cell population, comprising up to 50% of viable cells within the tumor mass (Xuan et al. 2021). The prognostic significance of immune cell infiltration in GBM remains controversial, with some studies reporting no correlation or even negative associations with survival (Whiteside 2008; Strauss et al. 2007 Strauss et al. 2007), while others suggest potential anti-tumor immune responses (Zitvogel et al. 2006). This divergence particularly manifests in tumor-associated bone marrow-derived cell. These cells, primarily consisting of monocytes and monocyte-derived macrophages (MDMs), have been predominantly polarized to an anti-inflammatory phenotype and implicated in promoting tumor progression, invasion, and therapeutic resistance (Buonfiglioli and Hambardzumyan 2021; Chen et al. 2020; Wu and Watabe 2017). Therefore, therapeutic strategies targeting the phenotypic plasticity of bone marrow-derived cells represent a promising avenue for modulating the immune microenvironment and potentially improving clinical outcomes.

Arginine metabolism has emerged as a critical regulator of tumor biology (Rabinovich et al. 2015; Hou et al. 2022; Hajji et al. 2022). This semi-essential amino acid is primarily synthesized through the intestinal-renal axis via the sequential actions of argininosuccinate synthetase 1 (ASS1) and argininosuccinate lyase (ASL). Additionally, cells expressing ASS1 and ASL can synthesize arginine locally from citrulline and aspartate (Haines et al. 2011; Husson et al. 2003) (Fig. 1). In 2019, a Phase I clinical trial involving patients with ASS1-deficient, recurrent, and heavily pretreated GBM demonstrated the safety of PEGylated arginine deiminase (PEG-ADI) therapy, with 80% of participants achieving stable disease (Hall et al. 2019). These findings underscore the significance of targeting arginine metabolism as a pivotal strategy in the advancement of therapeutic approaches for GBM. ASL, which catalyzes the conversion of argininosuccinate to arginine and fumarate, has garnered attention as a potential therapeutic target. Its crucial role in maintaining cancer cell homeostasis has been documented across multiple malignancies, including breast cancer, hepatocellular carcinoma, colorectal cancer, and GBM (Huang et al. 2013, 2015; Syed et al. 2013). Therefore, targeting ASL-associated metabolism could offer a promising avenue for therapeutic intervention in patients with GBM.Our study investigated the prognostic significance of arginine metabolism-related genes in GBM and their impact on the immune microenvironment. Through integrative bioinformatic analyses, we identified a prognostically significant gene associated with increased monocyte infiltration and tumor progression. These findings provide novel insights into GBM pathobiology and highlight potential therapeutic strategies for improving patient outcomes.

Fig. 1.

Fig. 1

Open in a new tab

Systematic and local arginine metabolism in GBM. Arginine is primarily synthesized through the intestinal-renal axis, where citrulline from the small intestine is converted into arginine in the kidneys by ASS1 and ASL. This systemic arginine is then distributed via the bloodstream. GBM cells meet their high metabolic demands through both the uptake of circulating arginine and local de novo synthesis from citrulline, a process facilitated by the expression of ASS1 and ASL within the tumor microenvironment

Material and methods

Data acquisition and sample collection

Bulk RNA sequencing data was retrieved from The Cancer Genome Atlas (TCGA) (https://xenabrowser.net/) (RRID:SCR_003193) and Chinese Glioma Genome Atlas (CGGA) dataset (https://www.cgga.org.cn/) (RRID:SCR_018844). Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) data and matched clinical information of TCGA GBM patients were downloaded from the UCSC Xena website. From CGGA, the mRNAseq_693 dataset was used, samples with complete expression and clinical annotation were retained, and those with a histopathological diagnosis of GBM were specifically selected for downstream analysis. For TCGA and CGGA, FPKM was transformed as log₂(FPKM + 1) for downstream analysis with the limma package and Cox regression.

Three additional GBM cohorts from the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) (RRID:SCR_005019) - GSE4412, GSE13041, and GSE7696 - were downloaded as series-matrix files. Expression values were log₂-transformed as provided and analyzed. For GSE13041, samples from different platforms were merged and batch-corrected with ComBat.

The scRNA sequencing data (GSE271379) was obtained from GEO database, which contains scRNA sequencing data from 5 GBM patients.

Gene sets related to arginine metabolism were gained from the Molecular Signatures Database (MSigDB) database (https://www.gsea-msigdb.org/gsea/msigdb). Specifically, three canonical gene sets were retrieved: 'GOBP_ARGININE_METABOLIC_PROCESS', 'WP_ARGININE_METABOLISM_AND_BYPRODUCTS', and 'KEGG_ARGININE_AND_PROLINE_METABOLISM'. These gene sets were subsequently merged, and redundant genes were removed to generate a comprehensive and non-redundant arginine metabolism-related gene list for downstream analysis (Table S1).

Screening of arginine-associated genes

By using “limma”, “survival” and “survminer” R packages, LASSO (Least Absolute Shrinkage and Selection Operator) Cox regression analysis was performed to develop a prognostic signature of GBM patients based on 18 arginine-related genes retrieved from MSigDB database. The optimal value of the penalty parameter lambda was determined through tenfold cross-validation using the minimum criteria. Then, uni- and multi-variate Cox Regression Analysis were used to identify survival-associated genes with a significance threshold of P < 0.05, using “forestplot”, “survival” and “survminer” R packages.

Correlation with clinical characters

Clinical information on GBM was collected from TCGA, including primary-recurrent-secondary (PRS) type (primary vs recurrent), isocitrate dehydrogenase (IDH) mutation status (mutant vs wildtype), 1p19q codeletion status (codeletion vs non-codeletion) and O6-methylguanine-DNA methyltransferase (MGMT) methylation type (methylation vs non-methylation). The expression level of genes was measured in each group and plotted into violin plots using the “limma” and “ggpubr” package.

Prognostic model construction and validation

To further explore the effect of arginine related genes to GBM, patients were divided into two groups according to the optimal cutoff of hub genes using the "surv_cutpoint" of "survminer" R package. The optimal cutoff was determined using maximally selected rank statistics, which provides a more statistically robust and biologically meaningful threshold by maximizing the log-rank statistic compared to an arbitrary median split. Moreover, the K-M survival curves and log-rank test were portrayed using the “survminer” and “survival” R packages. Besides, the GEO validation datasets and the CGGA cohort were utilized as an independent external validation program to elucidate the accuracy of the prognostic model.

To evaluate the predictive performance of arginine-related genes, we generated time-dependent receiver operating characteristic (ROC) curve analysis using the 'survivalROC' package in R software. The area under the curve (AUC) of GBM patients was calculated to assess the prognostic accuracy at 1, 2 and 3 years of overall survival. The decision curve analysis (DCA) was performed to evaluate the net benefit of molecular markers as predictive tools across different threshold probabilities. In this study, the DCA analysis was further conducted to assess the clinical utility of these arginine related genes, using “rmda”, “ggDCA” and “caret” R packages.

Enrichment analysis with functional annotation

Gene Ontology (GO) enrichment analysis of all 18 arginine related genes was performed using the 'clusterProfiler' package to identify enriched biological processes (BP), molecular functions (MF), and cellular components (CC). KEGG pathway analysis was conducted to identify significantly enriched pathways. For both analyses, we set the significance threshold at adjusted p-value < 0.05. The gene ratio, defined as the ratio of enriched genes to the total number of genes in each term, was calculated to measure enrichment strength and plotted into dot plot.

As an unsupervised analysis method, Gene Set Enrichment Analysis (GSEA) was used to perform the enrichment analysis and provide more detailed functional annotations. Based on the MSigDB hallmark gene sets (v2024.1), the normalized enrichment score (NES) and adjusted p-value of hub genes were calculated to determine the significance of enrichment in this study.

Single-cell RNA sequencing analysis

Single-cell RNA sequencing (scRNA-seq) data processing and analysis were performed using the Seurat package (v5.2.1) (RRID:SCR_016341). Raw count matrices were generated from 10X Genomics data. Initial quality control filtered out cells with fewer than 300 detected genes and cells with > 20% mitochondrial content. Cells with fewer than 1,000 unique molecular identifiers (UMIs) were also excluded. Doublet detection was performed using Scrublet, a widely adopted computational tool for identifying cell doublets in scRNA-seq data. Data were normalized using the LogNormalize method, and the top 2,000 highly variable genes (HVGs) were identified using the vst method. Principal component analysis (PCA) was applied for dimensionality reduction, and the top 9 principal components were selected based on elbow plot inspection. Batch effects across the five patient samples were corrected using Harmony, with patient ID specified as the primary covariate for integration. Harmony-corrected embeddings were subsequently used for downstream graph-based clustering and trajectory analysis. The UMAP was used for visualization, and graph-based clustering was performed using the Louvain algorithm (resolution = 0.4). Cell type annotation was conducted using the SingleR package with the Human Primary Cell Atlas as reference, and canonical marker genes were examined to validate cell-type assignments. Differential expression analysis between clusters was performed using the Wilcoxon rank-sum test with adjusted p-value < 0.05.

For monocyte/macrophage subpopulation analysis, cells were subset based on annotated identities, re-normalized, re-embedded, and reclustered. Module scores for monocyte and macrophage signatures were calculated using AddModuleScore. Correlation analyses between genes of interest were performed using the Spearman correlation coefficient.

Trajectory analysis and pseudotime inference

To investigate the developmental trajectory and dynamic gene expression patterns of monocyte cells, we performed pseudotime analysis using Monocle3 (version 1.3.7) (RRID:SCR_018685). The analysis was conducted on previously identified monocyte populations from scRNA data. The expression matrix was preprocessed using principal component analysis with 50 dimensions. UMAP was employed for dimensionality reduction. Trajectory inference was performed using Monocle3's learn_graph function with partition-based graph abstraction. Pseudotime ordering was determined based on the learned trajectory graph, with root states selected according to the expression of canonical monocyte markers. Gene expression dynamics along the pseudotime trajectory were analyzed using graph_test function with principal_graph neighbor settings. The expression patterns of ASL were specifically examined along the pseudotime trajectory. Visualization was performed using “ggplot2”.

To ensure the robustness of our findings, we employed Slingshot (v. 2.12.0) as an independent validation method. For Slingshot analysis, we used UMAP coordinates and specified MKI67-expressing monocytes as the starting point. Both methods were applied to the same preprocessed single-cell expression matrix to ensure comparability.

RNA velocity analysis

RNA velocity analysis was performed using the “scVelo” package (RRID:SCR_018168) in Python3. The GENCODE human genome annotation (release 44) was used as reference. Raw spliced and unspliced count matrices were generated from the single-cell RNA sequencing data. Quality control and preprocessing steps included filtering of low-quality cells and normalization of the count data. The preprocessed data was integrated with previously computed UMAP embeddings and cell type annotations. Velocity vectors were estimated using the dynamical model implemented in scVelo, which accounts for transcriptional dynamics by modeling the balance between unspliced and spliced mRNA abundances. The resulting velocity field was visualized using both grid and stream representations overlaid on UMAP projections.

Cell communication analysis

Cell–cell communication analysis was performed using the “CellChat” package (v1.6.1) (RRID:SCR_021946) in R software. CellChat analysis was based on the human ligand-receptor interaction database. Significant cell–cell communications were identified by filtering interactions using a communication probability threshold (0.05) and subsequent network aggregation at the pathway level. The interaction strength was determined by considering both expression level of signaling genes and the proportion of cells expressing them. The cell–cell communication network was visualized using circle plots and bubble plots. Signaling pathways were ranked by their overall communication strength, and the top pathways were selected for detailed analysis. For each significant pathway, specific ligand-receptor pairs and their expression patterns were examined across different cell types.

Statistical analysis

All statistical analysis above was conducted using R software (version 4.4.2) (RRID:SCR_001905) and Python3(RRID:SCR_018685). Comparisons between two different groups were carried out using the Wilcoxon test. Correlation analysis was evaluated using Spearman's test. Statistical significance set at P < 0.05, ns > 0.05, and indicated by asterisks (* P < 0.05, ** P < 0.01, *** P < 0.001, and **** P < 0.0001).

Results

Identification of arginine related genes with prognostic significance

To investigate the clinical implications of arginine metabolism in GBM, we systematically analyzed 18 arginine metabolism-associated genes curated from the MSigDB database. The LASSO Cox regression analysis was performed to identify prognostic signatures, with the optimal lambda value determined through cross-validation (Fig. 2A and B). To establish an accurate prognostic model, we further refined the gene set using univariate and multivariate Cox proportional hazards regression analyses (Table S7). The univariate Cox analysis prioritized three pivotal genes - ASL, FAH, and NAGS - as candidate prognostic markers (Table S2). However, further refinement using multivariate Cox proportional hazards regression provided a more critical test of their independent predictive values. Among all genes, only ASL retained its status as a prominent prognostic factor (HR = 1.904, 95% CI: 1.241–2.922, p = 0.0032), whereas FAH and NAGS failed to maintain significance in the multivariate context. Given these findings, ASL was prioritized as the primary candidate for all downstream analysis.

Fig. 2.

Fig. 2

Open in a new tab

Identification of hub arginine metabolism genes and their associations with PRS/IDH status. A LASSO Cox regression coefficient profiles of 18 arginine metabolism-related genes plotted against log(λ). Numbers on the top axis indicate the number of non-zero coefficients at each λ. B Ten-fold cross-validation for selecting the optimal λ in the LASSO Cox model. Red dots represent the mean cross-validated partial likelihood deviance and gray bars indicate ± 1 standard error (SE). Vertical dashed lines denote the selected λ values (λ_min and λ_1se). C Violin plots comparing expression levels of ASL, FAH, and NAGS between primary and recurrent GBM (PRS type). Adjusted p-values were calculated using the Wilcoxon rank-sum test and are shown in each panel. D Violin plots comparing expression levels of ASL, FAH, and NAGS between IDH-mutant and IDH-wildtype GBM. Adjusted p-values were calculated using the Wilcoxon rank-sum test

Validation of arginine biomarkers for GBM prognosis

To evaluate the prognostic performance of the three identified genes, we conducted Kaplan–Meier survival analyses. Patient stratification into high- and low-expression groups was performed using optimal cutoff values determined by the maximally selected rank statistics method implemented in the 'survminer' R package. For ASL, the resulting optimal cutpoint was 10.047 (TCGA). The survival analyses revealed that elevated expression of all three genes was significantly associated with unfavorable clinical outcomes, suggesting their potential oncogenic roles in GBM progression. Specifically, patients with high ASL expression demonstrated significantly reduced overall survival compared to those with low expression (p = 0.00026). Similar adverse prognostic associations were observed for both FAH (p = 0.00019) and NAGS (p = 0.0058) (Fig. 3A).

Fig. 3.

Fig. 3

Open in a new tab

Prognostic value of ASL/FAH/NAGS in TCGA and GEO cohorts and predictive performance evaluation. A Kaplan–Meier overall survival (OS) curves for ASL, FAH, and NAGS in the TCGA GBM cohort. Patients were stratified into high and low expression groups using an optimal cutoff. P-values were derived from the log-rank test. Numbers at risk are shown beneath each plot. B Kaplan–Meier OS curves for ASL in three independent GEO datasets (GSE4412, GSE13041, and GSE7696). P values were calculated with the log-rank test. Numbers at risk are shown below each plot. C ROC analysis of prognostic discrimination. Left: ROC curves of 1-year overall survival prediction comparing ASL, FAH, and NAGS in the TCGA GBM cohort, AUC values are indicated in the legend. Right: Time-dependent ROC curves for ASL at 1, 2, and 3 years of follow-up, AUC values are shown in the legend. D DCA comparing standardized net benefit of ASL, FAH, and NAGS across a range of threshold probabilities, with “All” and “None” representing treat-all and treat-none strategies

ASL was further validated in three independent GEO datasets (GSE4412, GSE13041, and GSE7696) using the same approach, the resulting optimal cutpoints were 10.586 (GSE4412), 6.590 (GSE13041) and 6.944 (GSE7696). It shows that high ASL expression was associated with poorer survival in each cohort (p < 0.0001, p = 0.031, and p = 0.047, Fig. 3B). The corresponding analysis of ASL was found in the CGGA cohort for comparison (p = 0.048, cutpoint = 7.044), reinforcing the robustness of our findings (Figure S1C). The baseline clinical characteristics of patients in the CGGA cohort are shown in Table S8.

To assess the predictive accuracy of these arginine metabolism-related genes, we performed time-dependent ROC curve analysis for 1-year overall survival prediction. All three genes demonstrated moderate discriminative ability: ASL (AUC = 0.643), FAH (AUC = 0.684), and NAGS (AUC = 0.660), all with statistical significance (Fig. 3C). Additionally, ASL was evaluated with landmark ROC analyses at one, two, and three years (AUC = 0.643, 0.592, and 0.578, Fig. 3C). To further evaluate the clinical applicability of these molecular markers, DCA was conducted across various threshold probabilities. The DCA demonstrated that all three genes provided positive net benefits, with ASL consistently outperforming the default strategies of treating all patients or treating none (Fig. 3D), indicating its potential value as a prognostic biomarker for clinical decision-making in GBM patients.

Integration of arginine metabolism signatures with established molecular prognostic factors in glioblastoma

To further explore clinical relevance, we investigated the association between these candidate genes and established prognostic markers in GBM. After Benjamini–Hochberg correction, the results revealed a significant correlation for ASL expression with PRS classification (p.adj = 1.94e − 06, Fig. 2C), IDH mutation status (p.adj = 1.20e − 11, Fig. 2D), and 1p/19q codeletion status (p.adj = 1.65e − 06, Fig. 4A). Notably, no significant association was observed between ASL expression and MGMT promoter methylation (p.adj = 5.32e − 01, Figure S1B), suggesting its prognostic value is independent of this conventional marker. Considering the significant associations with IDH status and 1p/19q codeletion, we performed multivariate Cox regression to further evaluate its independence (Table S7). In the CGGA cohort, ASL was not an independent prognostic factor when all molecular variables were included, suggesting its prognostic significance is primarily driven by its enrichment in high-risk molecular subgroups, such as IDH-wildtype and 1p/19q non-codeletion GBM. Similar trends were observed for FAH and NAGS. Comparison with normal brain samples further confirmed the upregulation trend of three genes in GBM. Although the elevation of ASL did not reach formal statistical significance (p.adj = 7.43e − 02, Figure S1A), the consistent upward trend across all three genes supports their potential role in glioma genesis.

Fig. 4.

Fig. 4

Open in a new tab

Clinical correlation and functional enrichment of arginine metabolism-related genes. A Violin plots showing expression of ASL, FAH, and NAGS across 1p/19q codeletion status (codeleted vs non-codeleted). Adjusted p-values were calculated using the Wilcoxon rank-sum test. B Gene Ontology (GO) enrichment analysis of the 18 arginine metabolism-related genes. Dot size indicates gene count and dot color represents adjusted P-value, the x-axis denotes gene ratio. C KEGG pathway enrichment analysis of the 18 arginine metabolism-related genes. Dot size indicates gene count and dot color represents adjusted P-value, the x-axis denotes gene ratio. D Gene set enrichment results (hallmark gene sets) for ASL, FAH, and NAGS high-expression phenotypes. Enrichment curves are shown with key pathways listed, normalized enrichment score (NES) and adjusted P-values are displayed in each panel

Enrichment analysis related to arginine related genes

GO enrichment analysis identified distinct functional categories across three domains. In biological processes (BP), the top enriched terms were primarily associated with arginine metabolic process, amino acid metabolic process and alpha-amino acid metabolic process. Cellular component (CC) analysis revealed enrichment in mitochondrial matrix, membrane raft and membrane microdomain while molecular function (MF) analysis showed significant enrichment in organic acid binding, carboxylic acid binding and amino acid binding (Fig. 4B). KEGG pathway analysis identified eight significantly enriched pathways (adjusted p < 0.05). The most significantly enriched pathways included arginine biosynthesis, biosynthesis of amino acids, arginine and proline metabolism (Fig. 4C).

To elucidate the biological implications of these arginine metabolism genes, we performed differential expression analysis between high- and low-expression groups stratified by ASL, FAH, and NAGS expression levels. Using stringent criteria (adjusted p < 0.05, |log2FC|> 1), we identified distinct transcriptional signatures associated with each gene's expression status. GSEA was subsequently conducted to delineate the functional characteristics of these expression patterns. Notably, in the ASL-high expression group, we observed significant enrichment of multiple immune-related and inflammatory pathways from the MSigDB hallmark gene sets. The most significantly enriched pathways included interferon-gamma response (normalized enrichment score [NES] = 3.32, adjusted p < 0.001), inflammatory response (NES = 3.22, adjusted p < 0.001), allograft rejection (NES = 3.22, adjusted p < 0.001), TNF-α signaling via NF-κB (NES = 3.18, adjusted p < 0.001), and interferon-alpha response (NES = 3.02, adjusted p < 0.001) (Fig. 4D). These findings suggest a potential link between ASL expression and immune modulation in the glioblastoma microenvironment.

Profiling tumor microenvironment heterogeneity in glioblastoma

To delineate the cellular heterogeneity within the GBM microenvironment, we analyzed single-cell RNA sequencing data from tumor specimens of five patients. After quality control and initial filtering, 40,957 cells were retained. Following downstream preprocessing for dimensionality reduction and clustering, we performed further filtering to remove potential doublets and clusters exhibiting low-quality transcriptomic profiles, resulting in a finalized set of 38,263 high-quality cells included for UMAP embedding and subsequent analyses. We identified 2,000 highly variable genes (HVGs) and performed PCA, selecting the top 9 principal components based on elbow plot evaluation.

UMAP clustering of the 38,263 cells revealed 16 distinct transcriptional states (Fig. 5A, Table S3). Cluster-specific gene signatures were identified and visualized as a heatmap (Fig. 5G). Cell type annotation using SingleR, together with canonical marker validation, identified five major populations: T lymphocytes (n = 2,063), astrocytes (n = 18,097), endothelial cells (n = 1,722), monocytes/macrophages (n = 15,413), and tissue stem cells (n = 968) (Fig. 5B). The robustness of this classification was validated through examination of lineage-specific marker expression patterns (Fig. 5D, Table S4).

Fig. 5.

Fig. 5

Open in a new tab

Single-cell atlas of GBM microenvironment and ASL/arginine metabolism distribution. A UMAP visualization of all single cells colored by unsupervised clusters. B UMAP visualization colored by annotated major cell types (astrocytes, endothelial cells, monocytes, T cells, and tissue stem-like cells). C Violin plot showing ASL expression across major cell types. D Feature plots of canonical marker genes used to support cell type annotation. E AUCell-based activity score of the arginine metabolism gene set projected onto the UMAP embedding. F Feature plot of ASL expression projected onto the UMAP embedding. G Heatmap of representative cluster marker genes across the identified clusters. H Cell type composition across clusters shown as a stacked bar chart (proportion of each cluster within each major cell type)

To investigate the relevance of arginine metabolism in the GBM microenvironment, we employed AUCell analysis, which revealed significant enrichment of arginine metabolism-associated gene signatures across the tumor ecosystem (Fig. 5E). Notably, focused analysis of ASL, a pivotal gene in arginine metabolism, demonstrated preferential expression within the monocyte/macrophage compartment (Fig. 5C and F), suggesting a potential cell type-specific role in GBM pathophysiology.

Myeloid cell representing metabolic heterogeneity of tumor microenvironment

Given the prominent representation of myeloid cells in the GBM microenvironment, we performed focused analysis of the monocyte/macrophage compartment. Following subset isolation, we conducted refined transcriptional analysis including normalization, principal component analysis, and UMAP dimensionality reduction. Utilizing the top 5 principal components for clustering analysis revealed 12 distinct myeloid subpopulations (Fig. 6A, Table S5).

Fig. 6.

Fig. 6

Open in a new tab

Myeloid subclustering and ASL dynamics along monocyte-state trajectories. A UMAP of the monocyte/macrophage subset colored by Seurat clusters. B UMAP with annotated myeloid subpopulations (monocyte- and macrophage-related states labeled). C Proportional composition of clusters within macrophage and monocyte groups. D Arginine metabolism activity score (AUCell) projected onto a low-dimensional embedding of myeloid cells. E Violin plot comparing ASL expression between macrophages and monocytes. F ASL expression overlaid on the myeloid UMAP with inferred trajectory/lineage structure. G Scatter plot of ASL expression along pseudotime with a smoothed trend line. H Violin plot of ASL expression across annotated monocyte sub-states. I UMAP colored by pseudotime values. J UMAP colored by annotated myeloid sub-states

To characterize these subpopulations, we performed differential expression analysis, identifying the top 10 discriminative genes for each cluster based on detection percentage metrics (pct.1 and pct.2). Systematic evaluation of canonical lineage markers was conducted to delineate monocyte versus macrophage identity. Specifically, we assessed the expression patterns of monocyte-associated genes (CD14, CCR2, and LYZ) and macrophage-specific markers (CD68, CD163, and MRC1). Integration of these expression signatures with established literature-derived markers enabled precise annotation of all 12 myeloid clusters (Fig. 6 and C).

Arginine metabolic activity and interplay in monocyte immune dynamics

To further investigate the relationship between arginine metabolism and myeloid cell states, we performed targeted AUCell analysis within the monocyte/macrophage subset. This analysis revealed significant enrichment of arginine metabolism-associated gene signatures across myeloid populations (Fig. 6D). Notably, differential expression analysis demonstrated significantly elevated ASL expression in monocytes compared to macrophages (Fig. 6E), suggesting cell state-specific regulation of arginine metabolism within the monocyte.

Multi-method trajectory analysis identifies MKI67 + monocytes as progenitor population in GBM-associated monocytes

To elucidate the developmental dynamics within myeloid populations, we performed comprehensive trajectory inference analyses (Table S6). Initial analysis using Monocle3 revealed a continuous developmental landscape encompassing all identified subpopulations, with MKI67-expressing monocytes emerging as the putative developmental origin (Fig. 6I and J). The reconstructed trajectory demonstrated distinct directional progression from early progenitor states through terminal differentiation stages. Notably, ASL expression exhibited a gradual increase during the early trajectory phase, reaching its peak at pseudotime ≈ 5, followed by a progressive decline toward terminal differentiation stages, suggesting its potential regulatory role in early myeloid development (Fig. 6F-H).

To validate these developmental trajectories, we employed an orthogonal approach using Slingshot algorithm. Despite its more stringent cell inclusion criteria (47.8% coverage), Slingshot analysis corroborated the principal developmental trajectories identified by Monocle3, particularly positioning MKI67 + monocytes as the trajectory initiation point (Fig. 7A) and their elevated ASL expression (Fig. 7B). The concordance between these methodologically distinct approaches substantiates the robustness of our trajectory inference.

Fig. 7.

Fig. 7

Open in a new tab

Multi-method trajectory validation, RNA velocity, and CellChat-based communication networks in myeloid cells. A Slingshot trajectory inference shown on UMAP embedding (left: lineage curves, right: pseudotime gradient). B ASL expression along Slingshot pseudotime with a fitted trend line, points are colored by pseudotime. C RNA velocity stream plot projected onto the embedding, illustrating directional state transitions among myeloid subpopulations. D Circle plot summarizing inter-subpopulation communication inferred by CellChat. E Heatmap showing the number of inferred interactions between myeloid subpopulations. F Bubble plot of representative ligand-receptor interactions across sender/receiver groups (dot size reflects significance, color reflects communication probability). G Alluvial/Sankey plot summarizing incoming communication patterns, signaling pathways, and associated interaction patterns in target cell groups. H-J Circle plots of selected signaling pathways inferred by CellChat, including MHC-II, MIF, and SPP1 pathways

To further characterize the dynamic transcriptional states, we performed RNA velocity analysis, which provides insights into future cell states based on splicing dynamics. The velocity vector field, projected onto the UMAP embedding, revealed coherent patterns of transcriptional flux, with pronounced directional flows emanating from Myeloid (Endothelial-enriched) (Fig. 7C). The grid-based visualization of RNA velocity vectors illuminated both local cellular dynamics and global developmental patterns. While some divergence was observed between velocity-based trajectories and pseudotime ordering, these differences likely reflect the complementary biological features captured by each analytical approach, collectively providing a comprehensive view of monocyte state transitions.

Cell–cell communication analysis suggests potential MIF-associated signaling networks in myeloid population

To dissect the intercellular signaling networks within the myeloid compartment, we performed comprehensive cell–cell communication analysis. Circular visualization and hierarchical clustering of the interaction networks identified candidate directional communication patterns among monocyte subpopulations (Fig. 7D and E). Quantitative analysis of ligand-receptor pairs identified significant interactions, with particularly strong enrichment of CD44-CD74, CD44-CXCR4, and SPP1-CD44 signaling axes between MKI67 + and S100A12 + monocyte populations (Fig. 7F).

Pathway enrichment analysis of the intercellular communication networks highlighted the predominance of MHC II presentation, macrophage migration inhibitory factor (MIF) signaling, and SPP1-mediated pathways within the monocyte microenvironment (Fig. 7G-J). Notably, networks correlated with the MIF-axis emerged as a central communication hub associated with MKI67-expressing monocytes.

To further explore the clinical relevance of ASL-associated intercellular communication, we performed correlation analyses between ASL and key molecular mediators identified in the CellChat inference. ASL expression demonstrated significant negative correlation with MKI67 (R = -0.37, p.adj = 1.23e − 06), while exhibiting strong positive correlations with MIF (R = 0.54, p.adj = 4.39e − 13), CD44 (R = 0.4, p.adj = 1.66e − 07), CD74 (R = 0.4, p.adj = 1.66e − 07), and CXCR4 (R = 0.4, p.adj = 1.66e − 07) (Spearman correlation) (Figure S2). These correlation patterns are consistent with the CellChat-inferred ligand-receptor communications, providing complementary evidence for the potential coordination between ASL and the identified molecular mediators.

Discussion

Metabolic reprogramming represents a fundamental hallmark of malignant transformation, characterized by profound alterations in cellular bioenergetics and biosynthetic processes that sustain neoplastic growth (Hirschey et al. 2015). Cancer cells exhibit distinctive metabolic dependencies that distinguish them from their non-transformed counterparts, particularly in their heightened nutrient acquisition and utilization patterns. This metabolic adaptation frequently manifests as selective nutrient dependencies, notably amino acid auxotrophy. Of particular significance is the metabolism of arginine, a semi-essential amino acid that serves as a crucial mediator in cellular proliferation and diverse metabolic cascades (Morris 2002). In the context of GBM, current evidence indicates a significant upregulation of arginine transport mechanisms, as demonstrated by the substantial accumulation of arginine metabolic intermediates (Kobayashi et al. 2008; Chinnaiyan et al. 2012). This suggests arginine metabolism as a functional and hyperactive metabolic network, indicating that therapeutic strategies targeting arginine availability might represent a viable intervention point in glioblastoma treatment.

In the present investigation, we conducted a comprehensive analysis of arginine metabolism-associated genes to identify potential prognostic indicators in GBM. Through systematic integration of bulk-RNA sequencing data and single-cell transcriptomic profiling, ASL emerged as a pivotal mediator in GBM progression among the arginine metabolic network components. Our findings revealed that elevated ASL expression is significantly associated with specific immune landscape alterations within the GBM microenvironment, particularly linked to distinct monocyte infiltration patterns. Through comprehensive computational analyses, we identified ASL as a potential regulator linked to monocyte infiltration and malignant progression in GBM.

ASS1 and ASL constitute critical enzymatic components of the arginine-citrulline cycle, exhibiting dual functionality in both renal arginine biosynthesis and hepatic urea cycle metabolism (Moren et al. 2018). ASS1, a cytosolic enzyme, catalyzes an ATP-dependent condensation reaction between citrulline and aspartate, yielding argininosuccinate while concurrently hydrolyzing ATP to AMP and pyrophosphate. This reaction is followed by ASL-mediated catalysis, wherein argininosuccinate undergoes cleavage to generate arginine and fumarate. The resultant arginine enters multiple metabolic pathways: it can be recycled to citrulline via NOS, or serve as a substrate for the biosynthesis of various metabolites including ornithine, agmatine, and guanidinoacetate. Notably, while multiple cellular enzymes utilize arginine as a substrate, ASL maintains a unique position in mammalian cellular metabolism as the sole enzyme capable of facilitating endogenous arginine synthesis (Caldwell et al. 2015; Keshet and Erez 2018).

ASL, a key enzyme in the arginine biosynthesis pathway, has been implicated in the progression of various malignancies (Huang et al. 2013, 2015; Sun et al. 2024; Khare et al. 2021; Bednarz-Misa et al. 2021). Within the central nervous system (CNS), the enzymatic activity of ASL is indispensable for the biosynthesis of NO, a critical regulator of cerebral microvascular endothelial cell homeostasis and blood–brain barrier (BBB) integrity (Kho et al. 2023). While its role in GBM remains largely unexplored, our study demonstrates that ASL overexpression significantly correlates with poor prognosis of GBM patients, particularly in high-risk molecular subgroups. Through comprehensive survival analysis, we established that elevated ASL expression is associated with reduced overall survival rates. This finding was further validated by ROC curve analysis and DCA, suggesting the potential of ASL as a prognostic biomarker in GBM. Nevertheless, multivariate analysis revealed that the prognostic significance of ASL was not independent of established molecular markers, suggesting that elevated ASL expression primarily reflects enrichment in high-risk molecular subgroups such as IDH-wildtype and 1p/19q non-codeletion GBM, rather than conferring an independent adverse effect. Notably, ASL expression showed significant associations with PRS classification, IDH mutation status, and 1p/19q codeletion status, implying that ASL may contribute to GBM aggressiveness partly through pathways overlapping with these molecular features, potentially involving NO-mediated vascular regulation and BBB disruption.

Myeloid cells constitute a fundamental component of the innate immune system, exhibiting critical functions in antimicrobial defense, tissue homeostasis, and the pathogenesis of various inflammatory conditions, including cancer. The hematopoietic process in the bone marrow maintains a continuous generation of common myeloid progenitors (CMPs), which subsequently differentiate into granulocyte-monocyte progenitors (GMPs), ultimately yielding mature monocytes and neutrophils through distinct developmental pathways (Friedmann-Morvinski and Hambardzumyan 2023). Despite monocytes relatively low abundance in peripheral circulation, they emerge as the predominant infiltrating immune cell population within the GBM microenvironment (Chen et al. 2023). Monocytes infiltration predominantly demonstrate pro-tumorigenic properties, with their infiltration density serving as a negative prognostic indicator (Chen and Hambardzumyan 2018). Previous studies have identified MKI67 + monocytes as progenitors, showing features such as EMT, G2/M checkpoints, and inflammatory responses (Wang et al. 2023). In this study, the extensive infiltration of monocytes in the GBM tumor microenvironment was portrayed. In addition, we further extracted monocyte subpopulations, and found that MKI67 + monocytes played a predominant role in the early stage of monocyte differentiation.

Under conditions of compromised immune function, including viral infections, bacterial infections, and neoplastic processes, the metabolic fate of L-arginine is predominantly directed toward NO synthesis. This metabolic pathway establishes a complex regulatory network that modulates multiple cellular metabolic cascades, thereby orchestrating the activation and functional polarization of monocytes/macrophages (Holzmuller et al. 2018). Furthermore, arginine metabolism serves as a critical modulator of inflammatory responses within the monocytic compartment (Lechowski et al. 2013). Our investigation revealed elevated expression of arginine metabolic pathways in monocytes, with particular emphasis on the remarkable overexpression of ASL in MKI67 + monocytic populations. This elevated ASL expression is associated with distinct myeloid states, as it marks the trajectory origin shared with MKI67 + monocytes, suggesting a potential link between arginine metabolism and early monocyte state transitions.

Our trajectory analysis revealed that ASL is highly expressed at the MKI67 + monocyte origin (pseudotime 0–5) but negatively correlates with MKI67 at the single-cell level. This paradox could be interpreted in light of the graded nature of Ki67 expression within the MKI67 + compartment (Miller et al. 2018). The elevated ASL at the trajectory origin implies that early-stage MKI67 + cells likely reside in a relatively lower-Ki67 substate, whereas the progressive decline of ASL along the trajectory may reflect a transition toward a higher-Ki67 state—a pattern that warrants further validation. Regardless of this transitional dynamic, our data collectively support that ASL drives glioma malignancy primarily through metabolic reprogramming and NF-κB activation rather than by accelerating the cell cycle (Wang et al. 2025).

MIF demonstrates consistently upregulated expression patterns across diverse malignancies, establishing its potential utility as a prognostic molecular signature for tumor invasiveness and disease recurrence (Mora Barthelmess et al. 2023; Huang et al. 2013). Integrative analyses revealed that MIF represents a potential central node associated with multiple downstream signaling programs. These include the activation of NF-κB pathway, ERK1/2 signaling axis, and AP-1 transcriptional complex. Additionally, MIF exerts its biological effects through engagement with the CD74/CD44 receptor complex, triggering subsequent cellular responses and signal transduction events (Fingerle-Rowson et al. 2009). Previous investigations across diverse neoplastic entities have demonstrated that the MIF-CD74 constitute a significant signaling axis that orchestrates multiple tumor-promoting processes, including proliferation enhancement, apoptotic resistance, neoangiogenesis stimulation, and immunosuppressive microenvironment establishment (Wirtz et al. 2021; Wang et al. 2021; Domingues et al. 2013; Arora et al. 2023). Through comprehensive intercellular communication network analysis, our study has elucidated the potential pivotal role of MIF signaling pathway in GBM pathogenesis. Correlation analyses supported coordinated expression patterns between ASL and components of the MIF signaling axis, however, experimental validation will be required to determine causality.

Through the integrative analysis of bulk and single-cell RNA-sequencing datasets, our study identifies ASL as a key prognostic marker in GBM. The single-cell transcriptomic evidence, including trajectory and RNA velocity analyses, suggests that ASL expression characterizes early-stage MKI67-positive monocyte populations and aligns with their differentiation path. Furthermore, the correlation between ASL and MIF-axis components indicates a potential association between arginine metabolism and myeloid-centered communication features in the tumor microenvironment.

The promising results of ASL in our study aligns with the emerging translational focus on arginine-targeting agents, such as ADI-PEG20 (Hajji et al. 2022). Our findings suggest that profiling these arginine-related metabolic enzymes may not only serve as a predictive biomarker for patient survival but also help identify candidates who could most benefit from arginine-depletion strategies combined with standard therapy.

Limitations and future perspectives

While providing key insights, our bioinformatic approach has several limitations. The small scRNA-seq cohort may not fully capture the inter-patient heterogeneity of GBM, potentially limiting the generalizability of our myeloid subpopulation findings. As findings are derived exclusively from transcriptomics, the links between ASL and monocyte differentiation remain inherently correlative. Without protein-level validation or functional perturbations, ASL's regulatory role in the MIF signaling axis remains hypothetical. Additionally, our trajectory and communication models offer a predictive roadmap but do not account for the spatial complexity of the tumor microenvironment. Future research incorporating larger clinical cohorts, spatial transcriptomics, and mechanistic experiments using in vivo or in vitro models will be essential to validate these candidates and establish the causal impact of ASL on glioblastoma progression.

Conclusions

In summary, through comprehensive single-cell transcriptomic profiling and bulk RNA sequencing analyses of GBM specimens, we systematically characterized the complex heterogeneity within the GBM tumor microenvironment. Our findings revealed a previously unrecognized role of ASL in GBM pathogenesis and patient outcomes. Crucially, our study underscores that ASL serves as a prognostic indicator intimately associated with high-risk molecular subtypes including IDH-wildtype and 1p/19q non-codeletion. Our integrative analyses revealed a strong positive association between ASL and the MIF signaling axis, which is closely linked to the differentiation programs of MKI67 + monocyte populations. This ASL-associated monocyte differentiation program is strongly linked to GBM progression. These molecular insights not only advance our understanding of GBM biology but also reveal ASL as a promising therapeutic target tailored for patients with high-risk molecular profiles, potentially opening new avenues for targeted intervention strategies in GBM treatment.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (ZIP 1016 KB) (1,016.2KB, zip)

Abbreviations

GBM

Gioblastoma

TTF

Tumor treating fields

TME

Tumor microenvironment

MDMs

Monocyte-derived macrophages

MIF

Macrophage migration inhibitory factor

ASS1

Argininosuccinate synthetase 1

ASL

Argininosuccinate lyase

PEG-ADI

PEGylated arginine deiminase

TCGA

The cancer genome atlas

CGGA

Chinese glioma genome atlas

MSigDB

The molecular signatures database

FPKM

Fragments per kilobase of exon model per million mapped fragments

LASSO

Least absolute shrinkage and selection operator

PRS

Primary-recurrent-secondary

IDH

Isocitrate dehydrogenase

MGMT

O6-methylguanine-DNA methyltransferase

ROC

Receiver operating characteristic

DCA

Decision curve analysis

GO

Gene ontology

BP

Biological processes

MF

Molecular functions

CC

Cellular components

GSEA

Gene set enrichment analysis

NES

Normalized enrichment score

UMIs

Unique molecular identifiers

UMAP

Uniform manifold approximation and projection

HVGs

Highly variable genes

PCA

Principal component analysis

CNS

Central nervous system

BBB

Blood-brain barrier

CMPs

Common myeloid progenitors

GMPs

Granulocyte-monocyte progenitors

Author contributions

Wenjing Zhao, Shenglan Li contributed to the conception and design of the study. Wenjing Zhao conducted the statistical analysis. All authors wrote the paper Shenglan Li and Wenbin Li revised the paper. All authors contributed to the manuscript and approved the submitted version.

Funding

This work was financially supported by the Talent Introduction Fund of Beijing Tiantan Hospital (RCY1-2020–2025-LWB), the Beijing Clinical Key Specialty Project (2-1-2-038), the National Natural Science Foundation of China General Program (32471036), the National Natural Science Foundation of China (82303822) and the Exploratory Study on the Mechanism of Action of Oncolytic Virus (OH2) in the Treatment of Glioma (HX-A-048(2021)).

Data availability

Bulk transcriptomic profiles and corresponding clinical annotations were retrieved from the UCSC Xena platform (TCGA-GBM cohort; https://xenabrowser.net/) and the CGGA (mRNAseq_693 cohorts; https://www.cgga.org.cn/) dataset. Additional independent cohorts were obtained from the GEO datasets (https://www.ncbi.nlm.nih.gov/geo/) repository under accession numbers GSE271379 (single-cell RNA sequencing data), GSE4412, GSE13041, and GSE7696 (microarray profiling datasets).

Declarations

Ethics approval

Not applicable.

Consent for publication

Not applicable.

Conflicts of interest

The authors declare no competing interests.

Competing interests

The authors declared no competing interest.

Clinical trial number

Not applicable.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. Arora M et al (2023) Rapid adaptation to CDK2 inhibition exposes intrinsic cell-cycle plasticity. Cell 186(12):2628-2643 e21. 10.1016/j.cell.2023.05.013 [DOI] [PubMed] [Google Scholar]
  2. Bednarz-Misa I et al (2021) Altered L-arginine metabolic pathways in gastric cancer: potential therapeutic targets and biomarkers. Biomolecules. 10.3390/biom11081086 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Buonfiglioli A, Hambardzumyan D (2021) Macrophages and microglia: the cerberus of glioblastoma. Acta Neuropathol Commun 9(1):54. 10.1186/s40478-021-01156-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Caldwell RB et al (2015) Arginase: an old enzyme with new tricks. Trends Pharmacol Sci 36(6):395–405. 10.1016/j.tips.2015.03.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cancer Genome Atlas Research N. et al. (2015) Comprehensive, Integrative Genomic Analysis of Diffuse Lower-Grade Gliomas. N Engl J Med 372(26): 2481–98.10.1056/NEJMoa1402121 [DOI] [PMC free article] [PubMed]
  6. Chen Z, Hambardzumyan D (2018) Immune microenvironment in glioblastoma subtypes. Front Immunol 9:1004. 10.3389/fimmu.2018.01004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen Z et al (2020) Genetic driver mutations introduced in identical cell-of-origin in murine glioblastoma reveal distinct immune landscapes but similar response to checkpoint blockade. Glia 68(10):2148–2166. 10.1002/glia.23883 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chen Z et al (2023) Monocyte depletion enhances neutrophil influx and proneural to mesenchymal transition in glioblastoma. Nat Commun 14(1):1839. 10.1038/s41467-023-37361-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chinnaiyan P et al (2012) The metabolomic signature of malignant glioma reflects accelerated anabolic metabolism. Cancer Res 72(22):5878–88. 10.1158/0008-5472.CAN-12-1572-T [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chinot OL et al (2014) Bevacizumab plus radiotherapy-temozolomide for newly diagnosed glioblastoma. N Engl J Med 370(8):709–22. 10.1056/NEJMoa1308345 [DOI] [PubMed] [Google Scholar]
  11. Domingues PH et al (2013) Association between inflammatory infiltrates and isolated monosomy 22/del(22q) in meningiomas. PLoS ONE 8(10):e74798. 10.1371/journal.pone.0074798 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fingerle-Rowson G et al (2009) A tautomerase-null macrophage migration-inhibitory factor (MIF) gene knock-in mouse model reveals that protein interactions and not enzymatic activity mediate MIF-dependent growth regulation. Mol Cell Biol 29(7):1922–32. 10.1128/MCB.01907-08 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Friedmann-Morvinski D, Hambardzumyan D (2023) Monocyte-neutrophil entanglement in glioblastoma. J Clin Invest. 10.1172/JCI163451 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Haines RJ, Pendleton LC, Eichler DC (2011) Argininosuccinate synthase: at the center of arginine metabolism. Int J Biochem Mol Biol 2(1):8–23 [PMC free article] [PubMed] [Google Scholar]
  15. Hajji N et al (2022) Arginine deprivation alters microglial polarity and synergizes with radiation to eradicate non-arginine-auxotrophic glioblastoma tumors. J Clin Invest 132(6):e142137. 10.1172/JCI142137 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hall PE et al (2019) A phase i study of pegylated arginine deiminase (Pegargiminase), cisplatin, and pemetrexed in argininosuccinate synthetase 1-deficient recurrent high-grade glioma. Clin Cancer Res 25(9):2708–2716. 10.1158/1078-0432.CCR-18-3729 [DOI] [PubMed] [Google Scholar]
  17. Hirschey MD et al (2015) Dysregulated metabolism contributes to oncogenesis. Semin Cancer Biol 35 Suppl:S129–S150. 10.1016/j.semcancer.2015.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Holzmuller P et al (2018) Trypanosomatid infections: how do parasites and their excreted-secreted factors modulate the inducible metabolism of l-arginine in macrophages? Front Immunol 9:778. 10.3389/fimmu.2018.00778 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hou X et al (2022) Targeted arginine metabolism therapy: a dilemma in glioma treatment. Front Oncol 12:938847. 10.3389/fonc.2022.938847 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Huang Q et al (2013) Increased co-expression of macrophage migration inhibitory factor and matrix metalloproteinase 9 is associated with tumor recurrence of meningioma. Int J Med Sci 10(3):276–285. 10.7150/ijms.5185 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Huang FW et al (2013) Highly recurrent TERT promoter mutations in human melanoma. Science 339(6122):957–959. 10.1126/science.1229259 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Huang HL et al (2013) Attenuation of argininosuccinate lyase inhibits cancer growth via cyclin A2 and nitric oxide. Mol Cancer Ther 12(11):2505–2516. 10.1158/1535-7163.MCT-12-0863 [DOI] [PubMed] [Google Scholar]
  23. Huang HL et al (2015) Argininosuccinate lyase is a potential therapeutic target in breast cancer. Oncol Rep 34(6):3131–3139. 10.3892/or.2015.4280 [DOI] [PubMed] [Google Scholar]
  24. Husson A et al (2003) Argininosuccinate synthetase from the urea cycle to the citrulline-NO cycle. Eur J Biochem 270(9):1887–99. 10.1046/j.1432-1033.2003.03559.x [DOI] [PubMed] [Google Scholar]
  25. Ius T et al (2012) Low-grade glioma surgery in eloquent areas: volumetric analysis of extent of resection and its impact on overall survival. A single-institution experience in 190 patients: clinical article. J Neurosurg 117(6):1039–1052. 10.3171/2012.8.JNS12393 [DOI] [PubMed] [Google Scholar]
  26. Keshet R, Erez A (2018) Arginine and the metabolic regulation of nitric oxide synthesis in cancer. Dis Model Mech 11(8):033332. 10.1242/dmm.033332 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Khare S et al (2021) ASS1 and ASL suppress growth in clear cell renal cell carcinoma via altered nitrogen metabolism. Cancer Metab 9(1):40. 10.1186/s40170-021-00271-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kho J et al (2023) Argininosuccinate lyase deficiency causes blood-brain barrier disruption via nitric oxide-mediated dysregulation of claudin expression. JCI Insight. 10.1172/jci.insight.168475 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kobayashi K et al (2008) Enhanced tumor growth elicited by L-type amino acid transporter 1 in human malignant glioma cells. Neurosurgery 62(2):493–503. 10.1227/01.neu.0000316018.51292.19 [DOI] [PubMed] [Google Scholar]
  30. Lechowski S et al (2013) Combined arginine and glutamine decrease release of de novo synthesized leukotrienes and expression of proinflammatory cytokines in activated human intestinal mast cells. Eur J Nutr 52(2):505–12. 10.1007/s00394-012-0353-1 [DOI] [PubMed] [Google Scholar]
  31. Louis DN et al (2021) The 2021 WHO classification of tumors of the central nervous system: a summary. Neuro Oncol 23(8):1231–1251. 10.1093/neuonc/noab106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Miller I et al (2018) Ki67 is a graded rather than a binary marker of proliferation versus quiescence. Cell Rep 24(5):110 e5-11125. 10.1016/j.celrep.2018.06.110 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mora Barthelmess R, Stijlemans B, Van Ginderachter JA (2023) Hallmarks of cancer affected by the MIF cytokine family. Cancers (Basel) 15(2):395. 10.3390/cancers15020395 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Moren L et al (2018) Metabolomic profiling identifies distinct phenotypes for ASS1 positive and negative GBM. BMC Cancer 18(1):167. 10.1186/s12885-018-4040-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Morris SM Jr (2002) Regulation of enzymes of the urea cycle and arginine metabolism. Annu Rev Nutr 22:87–105. 10.1146/annurev.nutr.22.110801.140547 [DOI] [PubMed] [Google Scholar]
  36. Ostrom QT et al (2018) CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2011–2015. Neuro Oncol 20(suppl_4):1–86. 10.1093/neuonc/noy131 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Paolillo M, Boselli C, Schinelli S (2018) Glioblastoma under Siege: an overview of current therapeutic strategies. Brain Sci 8(1):15. 10.3390/brainsci8010015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Rabinovich S et al (2015) Diversion of aspartate in ASS1-deficient tumours fosters de novo pyrimidine synthesis. Nature 527(7578):379–383. 10.1038/nature15529 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Strauss L et al (2007) A unique subset of CD4+CD25highFoxp3+ T cells secreting interleukin-10 and transforming growth factor-beta1 mediates suppression in the tumor microenvironment. Clin Cancer Res 13(15 Pt 1):4345–54. 10.1158/1078-0432.CCR-07-0472 [DOI] [PubMed] [Google Scholar]
  40. Strauss L et al (2007) The frequency and suppressor function of CD4+CD25highFoxp3+ T cells in the circulation of patients with squamous cell carcinoma of the head and neck. Clin Cancer Res 13(21):6301–11. 10.1158/1078-0432.CCR-07-1403 [DOI] [PubMed] [Google Scholar]
  41. Stupp R et al (2005) Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 352(10):987–96. 10.1056/NEJMoa043330 [DOI] [PubMed] [Google Scholar]
  42. Subramanian A et al (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102(43):15545–15550. 10.1073/pnas.0506580102 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Sun W et al (2024) FOXO3a-regulated arginine metabolic plasticity adaptively promotes esophageal cancer proliferation and metastasis. Oncogene 43(3):216–223. 10.1038/s41388-023-02906-0 [DOI] [PubMed] [Google Scholar]
  44. Syed N et al (2013) Epigenetic status of argininosuccinate synthetase and argininosuccinate lyase modulates autophagy and cell death in glioblastoma. Cell Death Dis 4(1):e458. 10.1038/cddis.2012.197 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Taal W et al (2014) Single-agent bevacizumab or lomustine versus a combination of bevacizumab plus lomustine in patients with recurrent glioblastoma (BELOB trial): a randomised controlled phase 2 trial. Lancet Oncol 15(9):943–953. 10.1016/S1470-2045(14)70314-6 [DOI] [PubMed] [Google Scholar]
  46. Tan AC et al (2020) Management of glioblastoma: State of the art and future directions. CA Cancer J Clin 70(4):299–312. 10.3322/caac.21613 [DOI] [PubMed] [Google Scholar]
  47. Wang Y et al (2021) MIF is a 3’ flap nuclease that facilitates DNA replication and promotes tumor growth. Nat Commun 12(1):2954. 10.1038/s41467-021-23264-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Wang D et al (2023) Single-cell transcriptomics reveals heterogeneity and intercellular crosstalk in human intervertebral disc degeneration. iScience 26(5):106692. 10.1016/j.isci.2023.106692 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wang S et al (2025) Screening of targets related to amino acid metabolism in glioma to identify the tumor-promoting effects of its core gene ASL. Sci Rep 15(1):38161. 10.1038/s41598-025-22105-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Whiteside TL (2008) The tumor microenvironment and its role in promoting tumor growth. Oncogene 27(45):5904–12. 10.1038/onc.2008.271 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wirtz TH et al (2021) Macrophage migration inhibitory factor exerts pro-proliferative and anti-apoptotic effects via CD74 in murine hepatocellular carcinoma. Br J Pharmacol 178(22):4452–4467. 10.1111/bph.15622 [DOI] [PubMed] [Google Scholar]
  52. Wu SY, Watabe K (2017) The roles of microglia/macrophages in tumor progression of brain cancer and metastatic disease. Front Biosci (Landmark Ed) 22(10):1805–1829. 10.2741/4573 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Xuan W et al (2021) Context-dependent glioblastoma-macrophage/microglia symbiosis and associated mechanisms. Trends Immunol 42(4):280–292. 10.1016/j.it.2021.02.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Zitvogel L, Tesniere A, Kroemer G (2006) Cancer despite immunosurveillance: immunoselection and immunosubversion. Nat Rev Immunol 6(10):715–27. 10.1038/nri1936 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary file1 (ZIP 1016 KB) (1,016.2KB, zip)

Data Availability Statement

Bulk transcriptomic profiles and corresponding clinical annotations were retrieved from the UCSC Xena platform (TCGA-GBM cohort; https://xenabrowser.net/) and the CGGA (mRNAseq_693 cohorts; https://www.cgga.org.cn/) dataset. Additional independent cohorts were obtained from the GEO datasets (https://www.ncbi.nlm.nih.gov/geo/) repository under accession numbers GSE271379 (single-cell RNA sequencing data), GSE4412, GSE13041, and GSE7696 (microarray profiling datasets).

Articles from Functional & Integrative Genomics are provided here courtesy of Springer

RESOURCES