Screening of targets related to amino acid metabolism in glioma to identify the tumor-promoting effects of its core gene ASL

Abstract

Gliomas, notably high-grade variants, dominate the spectrum of central nervous system (CNS) malignancies, characterized by aggressive behavior and diffuse invasion. Despite advances in tumor immunology, patient outcomes are stagnant. Amino acid metabolism is pivotal in glioma progression, driving the quest for metabolic targets. Bioinformatics allows deep dives into large-scale patient data from TCGA, CGGA, and GEO. Comparative studies on glioma amino acid metabolism have identified genes associated with tumor characteristics and patient survival. This yields an amino acid metabolism-based risk score model, which elucidates key biological processes and signaling pathways. Our holistic strategy clarifies amino acid metabolism’s role in glioma onset, paves the way for targeted therapies. Precise analysis and strategic targeting of metabolic pathways hold great promise for improving glioma treatment, offering hope to patients battling this relentless CNS malignancy.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-22105-z.

Introduction

Gliomas, the most prevalent class of primary central nervous system (CNS) tumors, constitute approximately 30% of all primary CNS neoplasms and account for approximately 81% of all malignant CNS tumors^1,2. Notwithstanding their relatively low incidence, gliomas are infamous for their high lethality³. Originating from glial cells or their precursors, gliomas encompass a heterogeneous spectrum of subtypes, including astrocytomas, oligodendrogliomas and ependymomas, spanning a continuum from indolent to highly malignant presentations that challenge therapeutic interventions and survival rates^4,5.

Tumor metabolic reprogramming is a hallmarker of malignancy⁶, epitomized by the heightened metabolic demands of neoplastic and immune cells. These cells exhibit a heightened metabolic activity, pivotal for proliferation and function, which underscores the aggressiveness of tumors and the dynamic activity of immune cells within the tumor microenvironment⁷. Tumor cells adapt to nutrient deprivation and hypoxia by enhancing aerobic glycolysis, reducing oxygen reliance, and sustaining growth under adverse conditions^8,9. Amino acid metabolism, which is central to cellular bio-synthesis¹⁰ and energy production¹¹, supports cell proliferation by providing nitrogen for macromolecules and maintaining cellular redox homeostasis.

Amino acid metabolism profoundly influences tumor behavior¹² and the tumor microenvironment¹³, promoting invasiveness and immunosuppression, and correlating with worse clinical outcomes. Overexpressed amino acid transporters in glioma cells¹⁴ facilitate nutrient scavenging, especially under conditions where endogenous synthesis is insufficient to meet demand¹⁵. The glutamate-glutamine cycle is exploited by glioma cells for glutamine scavenging, fueling mitochondrial metabolism and tumor growth¹⁶. Disruptions in methionine metabolism affect cellular methylation patterns¹⁷, potentially activating oncogenic pathways, while dysregulated tryptophan metabolism triggers immunosuppression via the aryl hydrocarbon receptor (AhR)¹⁸. Accumulations of cysteine and its metabolites, like cysteine sulfinate, serve as biomarkers for glioblastoma, reflecting both metabolic flexibility¹⁹ and reduced oxidative phosphor-rylation²⁰, supporting aggressive proliferation. The impact of Amino acid metabolism on the tumor microenvironment and immune modulation is substantial, shaping immune cell function²¹, tumor progression, and therapeutic response.

Recent advancements in bioinformatics have facilitated the comprehensive analysis of extensive patient datasets, yielding key insights into the molecular characteristics and heterogeneity of gliomas²², revolutionizing our understanding of this complex disease landscape. By analyzing gene expression profiles from TCGA, CGGA, and GEO, we identified genes significantly associated with glioma prognosis. Integrating hazard ratios with survival data, we developed a predictive model for amino acid metabolism risk in glioma, pinpointing potential therapeutic targets²³. Core genes representative of amino acid metabolism in glioma were selected for functional analysis to elucidate mechanistic insights into glioma pathogenesis and treatment responsiveness. Our research aims to translate these findings into innovative therapeutic strategies, targeting amino acid metabolism to improve survival and outcomes of patients with glioma.

Results

Screening of genes related to amino acid metabolism and model building

Capitalizing on the gene set pertinent to amino acid metabolism acquired from the Gene Set Enrichment Analysis (GSEA) database, coupled with survival curves generated through the Gene Expression Profiling Interactive Analysis (GEPIA) platform, we identified six genes with elevated expression levels that significantly influence the survival of patients with glioma (P < 0.0001) (Fig. 1A–F). The identified genes include: argininosuccinate lyase (ASL), a critical enzyme in the urea cycle; solute carrier family 6 member 6 (SLC6A6), a transporter for amino acids; prolyl 4-hydroxylase subunit alpha 2 (P4HA2), essential for collagen stabilization; prolyl 4-hydroxylase subunit beta (P4HB), another component of the prolyl hydroxylase complex; procollagen-lysine, 2-oxoglutarate 5-dioxy- genase 1 (PLOD1), a modulator of collagen structure; and phosphoserine amino- transferase 1 (PSAT1), a key player in serine biosynthesis.

Fig. 1.

Expression and the prognosis of amino acid metabolism-related genes. (A–F)The Overall survival of ASL (A), SLC6A6 (B), P4HA2 (C), P4HB (D), PLOD1 (E) PSAT1 (F) in TCGA All-grade gliomas; (G) Univariate cox regression analysis in TCGA-seq cohort; (H) Multivariate cox regression analysis in TCGA-seq cohort; (I) The nomogram of the prediction of 1-5year survival in GBMLGG patients; (J) The calibration plot of TCGA-seq dataset; (K) The expression of genes related to amino acid metabolism among different grade gliomas in TCGA; (L) The expression of genes involved in amino acid metabolism between different IDH status gliomas in TCGA. (M) The expression of genes related to amino acid metabolism in different molecular subtypes in TCGA.

To translate these genomic insights into a clinically applicable prognostic tool, we developed a risk score model based on amino acid metabolism within the high-grade glioma (HGG) cohort, drawing upon the comprehensive data available from The Cancer Genome Atlas (TCGA). The model employs the following formula²³: Risk Score = ∑[ln(HR) × expression] where ln(HR) × expression represents the natural logarithm of the hazard ratio for each gene, and expression signifies the level of mRNA expression. For HGG, the risk score model is defined as:

Utilizing the advanced features of the Sangerbox 3.0 web-based platform, we visualized the association between the risk score and the prognosis of these genes across all grade gliomas in the TCGA cohort. Patients were divided into high-risk and low-risk groups according to the amino acid metabolic risk score. There was a significant difference in survival between the high-risk and low-risk groups in both GBM and LGG, and the results were statistically significant (Fig. 2A–D). These findings were verified in CGGA325, CGGA693 and GSE16011 datasets (Figure S1E–H). Subsequently, univariate and multivariate COX regression analyses were conducted using additional clinical variables, including WHO grade, gender, age, and IDH mutation, demonstrated that the risk score serves as an independent adverse prognostic factor, distinct from other clinical features (Fig. 1G,H). By establishing a nomogram model, the one-year, three-year, and five-year survival rates were predicted, resulting in a patient survival prediction score. The predicted survival rates were found to be largely consistent with the actual survival rates (Fig. 1I,J).The heatmap elucidates that the five genes associated with poor prognosis exhibit elevated expression levels in the high-risk score group. Conversely, PSAT1, whose decreased expression indicates a worse prognosis, exhibits lower expression in the high-risk group, corroborating the concordance between the risk score model and the intrinsic survival patterns associated with these genes (Fig. 2E).

Fig. 2.

Expression Profiles and the prognosis of the amino acid metabolism- related risk score model. (A–D)Survival curves of the amino acid metabolism-related risk score in high-grade gliomas in TCGA (A), CGGA325 (B), CGGA693 (C), and GSE16011 (D) datasets; (E) The distribution heatmap of the risk score, patients survival status and the expression of amino acid metabolism related genes in TCGA all grade gliomas; (F–H)The risk score varied among different molecular subtypes (F), WHO grades (G), as well as IDH mutation statuses (H) in TCGA gliomas; (I–K)The risk score varied among different molecular subtypes, WHO grades and IDH mutation statuses in the CGGA325 dataset; (L–N)The risk score varied among different molecular subtypes, WHO grades and IDH mutation statuses in the CGGA693 dataset; (O–Q) The risk score varied among different molecular subtypes, WHO grades and IDH mutation statuses in the GSE16011 dataset; (R) The heatmap shows the association between the expression of amino acid metabolism-related genes and various clinical parameters in TCGA all grade gliomas.

Moreover, the survival statuses of patients stratified by their risk scores delineate a clear dichotomy: “death” predominates in the high-risk group, while “alive” is more frequently encountered in the low-risk group. Patients in the high-risk group exhibit significantly shorter overall survival compared to their low-risk counterparts. This observation underscores the model’s predictive accuracy and its clinical utility in the prognostication of patients with glioma, providing a valuable tool for guiding therapeutic decisions and patient management.

High-risk score predicts worse clinical phenotypes in glioma

To elucidate the intricate relationship between amino acid metabolism and glioma progression, we integrated the expression data of six critical genes with core clinical parameters, including patient demographics, WHO grade, and IDH mutational status. Heatmap analysis (Fig. 2R) highlighted that HGG gliomas and IDH1 Wildtype gliomas were disproportionately represented in the high-risk score group. GBM, the most aggressive glioma subtype, predominated in high-risk group, whereas oligodendroglioma was more prevalent in the low-risk group. Older patients were overrepresented in the high-risk group. Gene expression trends across grades (Fig. 1K), subtypes (Fig. 1M), and IDH mutation statuses (Fig. 1L) demonstrated that the five poor-prognosis genes exhibited progressively higher expression with increasing glioma grade, with the mesenchymal subtype showing the greatest upregulation. IDH1 Wild-type was associated with increased gene expression. In contrast, high PSAT1 expression linked with IDH1 mutant status and lower grade, suggesting a potential protective role. On the other hand, the model showed consistent prognostic value across all molecular subtypes (Fig. 2F,I,O), suggesting that metabolic reprogramming may be the “common basis” for glioblastoma plasticity.

Survival analysis (Fig. 2A–D) confirmed that higher risk scores were associated with worse outcomes. The risk score increased with glioma grade (Fig. 2G), peaked in patients with the mesenchymal subtype (Fig. 2F), and was higher in IDH1 Wild-type patients (Fig. 2H). External validation across three datasets (CGGA325, CGGA693, and GEO) yielded statistically significant results (p < 0.05), confirming the reliability of the amino acid metabolic risk score model (Fig. 2I–Q). The model exhibited a gradient across glioma grades and subtypes, with scores increasing with tumor severity, showing statistically significant differences. The mesenchymal subtype exhibited higher risk scores, aligning with its worse prognosis. IDH1 Wild-type patients were associated with higher scores, possibly indicating increased therapeutic resistance. The model’s applicability was confirmed in low-grade glioma, showing consistent associations between higher risk scores and worse prognosis (Fig. S1E–H). These findings underscore its potential as a universal prognostic marker to inform personalized treatment strategies.

In summary, the capacity of the amino acid metabolic-based risk score to differentiate WHO grades, subtypes, and IDH1 status highlights its clinical value. Identifying patients with high-risk scores can expedite targeted interventions and personalized treatments, aiming to improve survival and quality of life in patients with glioma.

High amino acid metabolism risk score indicated a more complex tumor immune microenvironment

Using the Limma toolkit, we analyzed transcriptomic differences between the low and the high-risk score patients, identifying distinct gene expression patterns (Fig. 3F). Top 200 differentially expressed genes were selected for functional annotation analyses using Metascape, which revealed enrichment of pathways related to inflammation, cytokine signaling, immune regulation, and angiogenesis (Fig. 3A), and summarized in a circular diagram (Fig. 3E). Gene Set Enrichment Analysis (GSEA) confirmed that cytokine interactions, leukocyte migration, and macrophage differentiation enriched in high-risk patients (Fig. 3G,H), highlighting the association between elevated risk scores and the more complex tumor immune microenvironment. Cross-validation in CGGA325, CGGA693, and GEO datasets yielded consistent results (Fig. 3B–D; Fig. S3F–H), strengthening the robustness of our findings. First, the correlation between the amino acid metabolism risk score and the tumor purity, as well as the stroma and the immune score, was investigated. It was found that in the TCGA dataset, the risk score showed a significant negative correlation with tumor purity, but positive correlations with the other two scores (Fig. 3L–N). This relationship was also confirmed in the CGGA325, CGGA693, and GSE16011 datasets (Fig. 3O–T). These findings suggest that a higher risk score indicates a lower tumor purity and a more complex tumor immune microenvironment. X-Cell analysis of the TCGA microarray cohort showed high-risk patients exhibited increased M2 macrophages infiltration and decreased activated NK cells and M1 macrophages proportion (Fig. 3I). This pattern was consistently reproduced in CGGA325 and CGGA693 datasets (Fig. 3J,K), confirming the impact of the risk score on the immune landscape of high-grade gliomas. Furthermore, we validated three immunoinfiltration methods—CIBERSORT, EPIC, and MCP-counter (Supplementary Fig. 5A–I)—and found that high-risk groups exhibited significantly higher M2 macrophage infiltration scores compared to low-risk groups, while activated NK cell subpopulations showed markedly lower infiltration levels. These findings confirm the robustness of our observation regarding an immunosuppressive microenvironment in high-risk patients, characterized by M2 macrophage enrichment and reduced NK cell counts, which was consistently validated across three computational approaches (X-CELL and CIBERSORT). This multi-algorithm validation strongly supports our core proposition: The amino acid metabolism risk score not only accurately predicts patient prognosis but also identifies tumor immune characteristics marked by myeloid-derived immunosuppression and innate immune dysfunction.

Fig. 3.

Functional enrichment analysis of the amino acid metabolism-related risk score in multiple independent datasets. (A–D) GO and KEGG enrichment analysis of amino acid metabolism-related risk score in TCGA, CGGA325, CGGA693 and GSE16011; (E) Chord diagram of KEGG pathway enrichment analysis based on the TCGA cohort; (F) Identification of differentially expressed genes between the high and the low-risk score group in TCGA dataset; (G,H) gene set enrichment analysis (GSEA) results comparing the high and the low-risk group in TCGA; (I–K)Analysis of immune cell infiltration in high-grade gliomas between the high and the low-risk score group in TCGA, CGGA325, and CGGA693 datasets; (L–T) Correlations between the amino acid metabolism-related risk score and the immune score, the stroma score and the tumor purity in TCGA, CGGA325, and CGGA693 datasets. P values were calculated using the Wilcoxon rank-sum test; *p < 0.05, **p < 0.01, *p < 0.001, ****p < 0.0001. (L–T) Correlations between the amino acid metabolism-related risk score and the immune score, the stroma score and the tumor purity in TCGA, CGGA325, and CGGA693 datasets.

These results emphasize the interplay between amino acid metabolism and immune dynamics in glioma, reinforcing the utility of the risk score as a predictive biomarker for tumor-immune interactions and clinical outcomes, while providing novel sights into glioma progression and potential therapeutic targets.

Identification of ASL as the core gene and validation of its high expression in clinical samples suggested a poor prognosis

Using Sangerbox 3.0, we generated ROC curves for six amino acid metabolism-related genes in the TCGA dataset (Fig. 4A–F), identifying ASL had the strongest predictive capacity (Fig. 4A). This finding highlighted ASL’s central role in glioma metabolism. We assessed ASL protein expression in various glioma cell lines compared with normal astrocytes, observed significantly higher ASL levels in glioma cells (Fig. 4G), and implicated ASL in glioma pathobiology. Further analysis via the Human Protein Atlas confirmed ASL upregulation in glioma cells compared to normal brain tissue, with high-grade gliomas showing more pronounced ASL expression than low-grade tumors (Fig. 4H). Additional immunohistochemical analyses of glioma samples across different grades consistently demonstrated increased ASL expression with advancing tumor grade (Fig. 4I–K).

Fig. 4.

Identification of core genes and validation of its biological effects. (A–F) The ROC curves illustrate the predictive performance of amino acid metabolism-related genes in the TCGA cohort. The area under the curve (AUC) values are indicated for each gene; (G) ASL expression in different glioma cell lines; (H) Immunohistochemical sections of the ASL in different grades of gliomas; (I–K)The clinical validation of WHO II (I), WHO III (J) and WHO IV (K) grade glioma samples.

Collectively, these findings underscore the critical role of ASL in glioma genesis and metabolic disruption, suggesting its potential avenues for exploring ASL’s mechanistic role and its potential viability as a diagnostic marker and a therapeutic target in neuro-oncology.

ASL knockdown suppresses glioma malignancy

To investigate the role of ASL in glioma progression, we selected U251 and LN229 cell lines with high endogenous ASL expression. Transfection with siASL significantly reduced ASL protein levels, as confirmed by western blotting (Fig. 5K,L).

Fig. 5.

Functional and mechanistic studies of core gene ASL in promoting the malignant phenotype of glioma cells. (A,C) Changes of the proliferation capacity in LN229 and U251 ASL knockdown cells; (B,D) Changes of the migration and invasion ability in LN229 and U251 ASL knockdown cells; (E) Cell proliferation curves following NF-κB inhibitor treatment in ASL-overexpressing cells; (F) Changes of the migration and invasion ability in ASL overexpression cells following NF-κB inhibitor treatment; (G–J) GSEA enrichment pathway analyses between the high and the low ASL expression group in TCGA, CGGA and GSE16011 datasets; (K,L) Western blots verified the efficacy of ASL knockdown in LN229 and U251 glioma cells; (M) Protein expression of NF-κB signaling pathway in ASL knockdown cells; (N) Protein expression of NF-κB signaling pathway based ASL overexpression.

ASL knockdown markedly attenuated the proliferative capacity of both U251 and LN229 cells over time, as measured by MTS assay (Fig. 5A,C). Furthermore, Transwell migration assays demonstrated that ASL suppression significantly impaired glioma cell motility (Fig. 5B,D). These results indicate that ASL is critical for maintaining the malignant phenotype of glioma cells.

ASL overexpression enhances glioma cell migration and invasion

We next overexpressed ASL in glioma cells using plasmid transfection, which was verified by western blotting (Fig. S4G). Transwell assays revealed that ASL overexpression significantly enhanced both migration and invasion capabilities compared to control cells (Fig. S4H). These findings demonstrate that ASL potently promotes glioma cell dissemination.

ASL activates NF-κB signaling, and NF-κB inhibition attenuates ASL-induced glioma cell proliferation and migration

Gene set enrichment analysis (GSEA) of TCGA, CGGA, and GEO datasets showed significant enrichment of NF-κB pathway genes in gliomas with high ASL expression (Fig. 5G–J). Western blot analysis confirmed that ASL knockdown reduced phosphorylated NF-κB (p-NF-κB) levels, while ASL overexpression increased p-NF-κB expression (Fig. 5M,N).

Treatment with the NF-κB inhibitor BAY 11-7082 suppressed proliferation (Fig. 5E,F) and mitigated the enhanced migration and invasion induced by ASL overexpression. These results indicate that ASL promotes glioma malignancy primarily through activation of the NF-κB signaling pathway.

Discussion

Glioma, the leading primary malignancy of the central nervous system, remains a daunting challenge in neuro-oncology. Despite advances in multimodal therapies, patient outcomes are dismal, necessitating novel strategies targeting tumor-specific vulnerabilities. While metabolic alterations like the Warburg effect are well-established, recent studies highlight amino acid metabolism as a critical hub for glioma progression. Our work builds upon foundational discoveries by Garofano et al.²⁴, who identified mitochondrial subtypes in GBM, and Peng et al., who demonstrated branched-chain amino acid dysregulation in tumors²⁵. However, the mechanistic links between specific amino acid genes and glioma immune evasion remain poorly understood.

Amino Acid Metabolism in Glioma: Bridging the Gap. Our integrated bioinformatics approach identified six key genes (ASL, SLC6A6, P4HA2, P4HB, PLOD1, PSAT1) linking amino acid metabolism to glioma prognosis. While previous studies reported ASL’s epigenetic roles in autophagy sensitivity and TERT activation via fumarate²⁶, we reveal its novel function as a metabolic-immune modulator through NF-κB signaling (Fig. 5G–J). This aligns with Strickland et al., on metabolic reprogramming²⁷ but contrasts with Kobayashi et al., who focused solely on ASL’s cell-autonomous effects without addressing microenvironment crosstalk²⁸.

While our data firmly establish a functional link between ASL and NF-κB activation, the precise molecular intermediate warrants discussion. As the enzyme that cleaves argininosuccinate, ASL sits at a critical metabolic node controlling the production of both arginine and fumarate. We propose that ASL overexpression leads to an accumulation of its product, fumarate. This oncometabolate has been shown to inhibit α-KG-dependent dioxygenases, leading to hypermethylation and altered gene expression²⁹. Furthermore, fumarate can inhibit succinate dehydrogenase (SDH), causing succinate accumulation and the generation of mitochondrial reactive oxygen species (mROS)³⁰. mROS are potent activators of the NF-κB pathway through IKK activation, providing a direct mechanistic conduit linking ASL activity to the pro-inflammatory signaling we observed³¹.

Alternatively, ASL-driven arginine metabolism could influence NF-κB through nitric oxide (NO) synthesis. However, the role of NO in cancer is complex and context-dependent, often being anti-tumorigenic at high levels. It is more plausible that ASL overexpression in glioma cells creates a local arginine depletion, which in itself is a signal that activates NF-κB in stromal and immune cells to promote a pro-tumorigenic environment³². While the relative contributions of fumarate, ROS, and arginine depletion remain to be fully dissected, our discovery of the ASL-NF-κB axis provides a strong foundation for future metabolomic and mechanistic studies aimed at identifying the key signaling intermediates."

Translational differentiation from prior ASL studies

While ASL’s roles in epigenetic regulation and TERT activation are well-documented, our study unveils its actionable potential as a metabolic-immune interface regulator. Crucially, the ASL-NF-κB axis: Exploits metabolic vulnerabilities: Unlike epigenetic modifiers, ASL’s enzymatic activity is pharmacologically targetable (e.g., via arginine depletion or small-molecule inhibitors). Addresses adaptive resistance: By correlating ASL with M2 polarization (Fig. 3I–K), we provide a rationale for combining metabolic and immunotherapies—a strategy absent in prior ASL studies. Overcomes heterogeneity: Our risk model’s consistency across subtypes (Fig. 2) suggests broader applicability than single-gene epigenetic markers."

Our finding that ASL expression correlates with M2 macrophage infiltration (Fig. 3I–K) and regulates NF-κB signaling (Fig. 5) is further supported by emerging paradigms in immunometabolism. Beyond its canonical role in the urea cycle, ASL is a pivotal regulator of cellular arginine pools. Arginine is not only a precursor for nitric oxide (NO) synthesis—a key effector of macrophage function—but also for polyamines and proline, which are essential for cell proliferation and collagen deposition, processes characteristic of tumor-promoting environments³³. Recent studies have shown that altered arginine metabolism in tumor cells can starve infiltrating T cells, impairing anti-tumor immunity³⁴. It is plausible that ASL overexpression in glioma cells creates an arginine-depleted niche that favors the polarization of macrophages towards an M2, pro-tumorigenic phenotype.

Furthermore, the accumulation of urea cycle intermediates, such as fumarate due to ASL dysregulation, can influence immune responses through multiple mechanisms. Fumarate has been shown to inhibit anti-tumor T cell function by inducing oxidative stress and suppressing cytokine production³⁵. It also acts as an oncometabolite that can drive epigenetic reprogramming by inhibiting α-KG-dependent dioxygenases, including histone and DNA demethylases³⁶. This epigenetic remodeling can stabilize the M2 transcriptional program in macrophages. Thus, ASL may promote immune suppression not only through NF-κB but also by creating a metabolically and epigenetically hostile microenvironment, positioning it as a central node linking tumor metabolism to immune escape.

ASL-NF-κB axis: a new therapeutic paradigm

Our discovery that ASL regulates NF-κB pathway activity and correlates with an immunosuppressive cytokine milieu extends beyond a simple linear pathway and aligns with the emerging concept of ‘metabolic tuning’ of immune responses. It is well-established that the metabolic state of TAMs profoundly influences their function via NF-κB. For instance, a shift to glycolysis, a hallmark of M2-like polarization, is not merely a consequence of activation but is actively required to fuel the energetic and biosynthetic demands of sustained cytokine production and secretion²⁶.

More directly, specific metabolites can act as signaling molecules to regulate NF-κB. Succinate, accumulated in activated macrophages, stabilizes HIF-1α and can enhance IL-1β production through NF-κB³⁷. Conversely, itaconate, another TCA cycle-derived immune metabolite, has been shown to alkylate KEAP1 and suppress NF-κB-driven inflammation, illustrating the precise metabolic control over this pathway³⁸.

In the context of our study, we postulate that ASL overexpression in glioma cells alters the local availability of arginine, fumarate, and other urea cycle intermediates. This metabolically reprogrammed niche could then be ‘sensed’ by infiltrating macrophages. For example, arginine depletion itself is a potent signal that can shift macrophage polarization towards an M2 phenotype³⁹. Furthermore, fumarate—a known product of ASL activity—can have paracrine effects; it can be released from tumor cells, taken up by TAMs, and inhibit key epigenetic enzymes like KDM5 histone demethylases, thereby potentially stabilizing a pro-tumorigenic genetic program that is reinforced by NF-κB32⁴⁰. Thus, ASL likely contributes to a feed-forward loop where tumor cell-intrinsic NF-κB activation and metabolite secretion collaborate to instruct and maintain an immunosuppressive macrophage population via both direct and metabolic mechanisms.

The discovery that ASL regulates NF-κB activity (Fig. 5M,N) extends beyond known urea cycle functions. Specifically:Mechanistic novelty: Unlike T Crook et al. who linked ASL to methylation, we show its control of NF-κB-driven cytokine networks (Fig. 3G,H), explaining the M2 macrophage dominance (Fig. 3I–K).Therapeutic implications: Our findings complement Campesato et al. on immunosuppressive tryptophan metabolism but propose ASL inhibition as a strategy to reverse NF-κB-mediated immune evasion⁴¹.

The therapeutic strategy of inhibiting ASL to reprogram the immune microenvironment aligns with a major frontier in oncology: metabolic immunotherapy. This approach is founded on the principle that small molecule inhibitors targeting key metabolic enzymes can dismantle immunosuppressive networks and overcome resistance to checkpoint blockade. Pioneering efforts in this field have focused on enzymes like IDO1, which catabolizes tryptophan to kynurenine, creating an immunosuppressive milieu. Although clinical outcomes with IDO1 inhibitors were mixed, they provided crucial proof-of-concept that targeting tumor metabolism can modulate immunity⁴².

More directly relevant to ASL are strategies targeting arginine metabolism. Pegargiminase (ADI-PEG20), an arginine-depleting enzyme, is showing promising clinical activity in mesothelioma and other arginine-auxotrophic cancers by starving tumor cells and, importantly, by reprogramming tumor-associated macrophages towards an immunostimulatory phenotype⁴³. Similarly, inhibition of glutaminase, a key enzyme in glutamine metabolism, has been shown to synergize with anti-PD-1 therapy by reducing the suppression of cytotoxic T cells⁴⁴.

In this context, our findings position ASL as a compelling new metabolic immune target. By demonstrating that ASL inhibition suppresses glioma malignancy and alters the immune landscape (e.g., reducing M2 macrophages), our work provides a strong rationale for developing ASL inhibitors. We envision that such agents could be used in combination with existing immunotherapies, akin to the strategies employed for IDO1 or arginine depletion, to reverse the profound immunosuppression that characterizes glioblastoma and ultimately enhance the efficacy of immune checkpoint blockade.

Limitations and Future Directions: While our risk model was validated in TCGA/CGGA cohorts, its applicability to pediatric gliomas requires testing, given reported age-related metabolic differences. Additionally, the ASL-NF-κB link warrants in vivo validation using transgenic models, building on methods by Ma et al⁴⁵.

Conclusion

By contextualizing ASL within amino acid metabolic networks and the immune microenvironment, our study advances the paradigm of “metabolic checkpoint” targeting in glioma. This bridges the gap between seminal work on arginine metabolism and emerging immunometabolism concepts, offering a roadmap for combined metabolic-immune therapies.

Materials and methods

Bioinformatics analysis website and public data sources

RNA sequence data and clinical information were downloaded from The Cancer Genome Atlas (TCGA) portal (https://portal.gdc.cancer.gov/projects), Chinese glioma genome atlas (CGGA) (http://www.cgga.org.cn/) and Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/).Gene Set Enrichment Analysis (GSEA) (https://www.gsea-msigdb.org/gsea/index.jsp) contains a large number of gene sets, and amino acid metabolism related gene sets are obtained from this platform. GEPIA website (http://gepia.cancer-pku.cn/detail.php?gene=&clicktag=survival) to quickly obtain the prognosis of amino acid metabolism genes. Sangerbox 3.0 (http://vip.sangerbox.com/home.html) is a cloud-based bioinformatics analysis platform, which makes bioinformatics analysis more popular and easily accessible. In this study, relevant student analysis was conducted on this platform. Metascape (https://metascape.org/gp/index.html#/main/step1) is an online bioinformatics analysis platform designed for biologists, where we performed enrichment analysis and gene lists for functional annotation.

Collection of related genes and model establishment of amino acid metabolism

With "amino acid metabolism" as the search term, the amino acid metabolism related gene sets were collected on the GSEA website. The four amino acid metabolism related gene sets were downloaded with 548 genes (Gene set names and specific genes are listed in the Supplementary Material), duplicates were excluded, and the remaining 346 genes were obtained. These 346 genes were entered into the GEPIA website survival analysis module to plot the survival curve of genes related to amino acid metabolism in glioma patients in TCGA database and select 6 genes that were significantly associated with glioma prognosis (p < 0.0001): ASL, SLC6A6, P4HA2, P4HB, PLOD1 and PSAT1. On Sangerbox 3.0, the "heatmap and prognostic score relationship tool" was used to draw the heatmap to intuitively observe the prognostic score relationship of these six genes in glioma. By summing the expression levels of these six genes * corresponding to HR, the gene risk score related to amino acid metabolism can be obtained to construct the amino acid metabolism risk score model in glioma.

Bioinformatics-related figure drawing

The Sangerbox 3.0 multiple boxplots (violin plots) were used to obtain the differential boxplots of the amino acid metabolic risk score model and six genes in different WHO grades, molecular subtypes and IDH 1 mutation status in glioma. Using Sangerbox 3.0 Kaplan–Meier survival curve drawing tool, the significance of group prognosis difference was evaluated by log rank test method, drawing the survival curve of amino acid metabolism related genes and amino acid metabolism score model in glioma. The risk score of the amino acid metabolism model was divided into two groups according to the value. The Limma rapid differential analysis tool in Sangerbox 3.0 was input to obtain the differentially expressed genes. The logFC value of the genes was ranked from large to small, and the top 200 genes were selected into the Metascape website for enrichment analysis. Then, biological function annotation and gene molecular pathway annotation of differential genes between different groups were performed using Sangerbox 3.0 online platform GO and KEGG one-click enrichment analysis tools. The grouped data from the enrichment analysis were imported into the Sangerbox 3.0 online platform GSEA simple analysis and visualization tool, and C2: KEGG gene sets and C5: GO biological processes functional and pathway enrichment analysis were performed on the different genes between different groups. The survival time, survival status and risk score of amino acid metabolism related genes in high-grade glioma patients in the TCGA database were input into the Sangerbox 3.0 online platform ROC simple drawing tool, with time points set at 1 and 3 years, to output the ROC curve of amino acid metabolism related genes in high-grade gliomas in the TCGA database.

Tumor-microenvironment analysis

The X-cell, Estimate, CIBERSORT, EPIC and MCP-counter algorithm was used to calculate the proportion of cells in the TCGA, CGGA325, and CGGA693 microarray cohorts to elucidate the relationship between amino acid metabolism and the composition of the tumor immune microenvironment.

Cell lines and the source of the specimens

Human glioma cell lines LN 229, U251, U373, and T98 were purchased from Shanghai Cell Bank Company of Chinese Academy of Sciences, and the normal human astrocyte cell line NHA was obtained from the Institute of Neurosurgery, Beijing, China. According to the Affiliated Hospital of China Medical University The protocol was approved by the ethics committee of the First Hospital, Fresh clinical glioma specimens were obtained from the Department of Neurosurgery of the First Affiliated Hospital of China Medical University, and informed consent was obtained from the relevant patients and their families.

Immunohistochemistry and Western blot

After cutting the tissue section, the slices were baked in a 60℃ warm box for 30 min, Dewaxing of xylene and ethanol hydration at gradient concentration, Wash in distilled water for 10min; Antigen repair in citrate buffer for 2 min; Circircles around the sliced tissue with a PAP pen, Add endogenous peroxidase blocking agent and incubated for 20 min at room temperature; The sections were washed with PBS, Add 10% goat serum for sealing; Remove the fluid from the sections, The specific histochemical antibodies were incubated for 4℃ overnight; After washing with PBS the following day, On sections with biotin-labeled sheep anti-rabbit IgG secondary antibody, Incubate at room temperature for 30 min; Slice were washed again with PBS, Biotin-resistant protein-peroxidase of Streptomyces sp., Incubate at room temperature for 20min; dropwise DAB, Color development time for tens of seconds ~ 2min, After the color in PBS; Nuclear counterstaining in hematoxylin for 5min, 1% hydrochloric acid alcohol differentiation for 10s; Differentiated nuclei in tap water for 10min, The sections were washed with distilled water for 10min, Dehydrated, transparent; Neutral tree gum seal sheet, Microscopes were observed and photographed after overnight room temperature.

Protein quantification: BCA quantification on ice, determined the absorbance of standard protein and sample protein at 562nm wavelength, drew the curve between the standard protein absorbance and concentration, calculated from standard curve sample protein concentration, sample protein concentration to 2 ug / ul with 5X loading buffer and super clean water, and denatured protein heated with 100℃ metal bath for 10min for subsequent WB experiment. Western Blot: 10% electrophoresis gel was prepared by using ase dispensing kit, After the gel is formed, 20 ug of protein sample was added to the well, Less than 20 ug with a 1X Loading Buffer complement, The loading gel was placed at 120V for electrophoresis, The blue loading buffer migrated to the bottom of the gel with voltage, The end of the electrophoresis, After rotating the membrane for 2h with a constant current of 100 mA, Under room temperature shaker with 5% nonfat milk powder for 1 h, Following an incubation with the primary antibody overnight in a 4℃ refrigerator, The membranes was washed three times in TBST the next day, With secondary antibody for 1 h at room temperature, TBST was washed three more times; Final luminescence was performed with an ECL luminometer.

Cells transfected

After the glioma cell line U251 and LN 229 cells were seeded and cultured overnight, they were transfected using siASL synthesized by Biotechnology Technology Co., Ltd. and overexpression plasmid synthesized by Heyuan Biotech (Shanghai) Co., Ltd., and cells were extracted for functional experiments 48 h after transfection siRNA. The sequence is as follows: SiASL-1 Sense: GUGGAUGUUCAAGGCAGCAAATT;Anti-sense:UUUGCUGCCUUGAACAUCCACTT; SiASL-2 Sense: GCCUAUUA CCUGGUCCGCAAATT; Anti-sense: UUUGCGGACCAGGUAAUAGGCTT. Over-expressionplasmid:PCDNA3.1-CVM-ASL-3xFLAG- hGHpolay-EF1a-EGFP.

Cell proliferation capacity was measured by MTS assay

The proliferative capacity of the glioma cells was determined by the MTS assay. The knockdown/overexpression transient cell lines after siRNA/overexpression plasmid treatment were seeded in 100ul per well containing 1000 cells in 96-well plates in a 37℃, 5% CO2 incubator for 5 days. From the day of inoculation, 20ul MTS of solution was added to the wells and incubated at constant temperature for 3 h to determine the absorbance value of the medium at 490nm. Knockdown cell lines 1000 cells per well, overexpressing cell lines 500 cells per well,

Cell migration and invasion capacity were determined by transwell

Cell migration ability was detected by placing the Transwell chamber in 24-well plate, the cells with 1% serum medium were seeded with 5 * 10⁴ cells per upper chamber. 1ml of 30% serum medium was added to the lower chamber for 10%. After incubation for 16h, the chambers were washed three times with PBS and the cells were fixed with 4% paraformaldehyde for 5min and then stained with 0.1% crystal violet for 100 times for cell counting. In addition, we also tested cell invasion ability. The method was roughly the same as the migration experiment, but the cells in the upper chamber needed to be adjusted to 1 * 10⁵ and spread 80 ul 8:8 diluted matrix glue, and the rest was the same as the migration experiment.

Statistical methods

This paper mainly uses R 3.6.1, Sangerbox 3.0, GraphPad Prism 8, omicshare, GENE DENOVO, GEPIA and Metascape websites for statistical analysis and drawing. Two-tailed independent sample t-test and one-way analysis of variance (ANOVA) were used to determine differences between groups, and the correlation between the two variables was performed with Spearman’s linear correlation. For the GSEA enrichment analysis, NES is the enrichment score, and FDR q-value is the corrected significance. Statistical results were considered statistically significant at p-value < 0.05 and were considered significant at FDR q-value < 0.25 in the enrichment analysis.

Ethics statement

This study was approved by the Ethics Committee of the First Affiliated Hospital of China Medical University. All methods were performed in accordance with relevant guidelines and regulations, including the Declaration of Helsinki. Informed consent was obtained from all participants or their legal guardians.

Abbreviations

GBM

Glioblastoma

TCGA

The cancer genome atlas

CGGA

Chinese glioma genome atlas

GEO

Gene expression omnibus

GO

Gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

GSEA

Gene set enrichment analyses

ROC

Receiver operating characteristic curve

Author contributions

Zou JY, Zhu C and Wang SS made substantial contributions to the conception and design. Zou JY, Lu J and Zhu C download and curate data. Yu ZX and Li JY made contributions to methodology. Wang SS, Zhu C and Zou JY analyzed and made contributions to interpretation of data. Yan ZH, Chen L were major contributor in writing the manuscript and revision. All authors read and approved the final manuscript.

Data availability

Data is provided within the supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.

Ethics declarations

The use of all human specimens was approved by the Ethics Committee of the First Affiliated Hospital of China Medical University, and informed consent was obtained from the patients and their family members.