Skip to main content
Download PDF

m6A regulator-based molecular classification and hub genes associated with immune infiltration characteristics and clinical outcomes in diffuse gliomas

BMC Medical Genomics volume 18, Article number: 37 (2025) Cite this article

Abstract

Background

m6A methylation modification is a new regulatory mechanism involved in tumorigenesis and tumor-immunity interaction. However, its impact on glioma immune microenvironment and clinical outcomes remains unclear.

Methods

Comprehensive expression profiles of 18 m6A regulators were used to identify molecular subtypes exhibiting distinct m6A modification patterns in 1673 glioma samples sourced from public datasets. A multi-genes signature was constructed for predicting clinical outcomes and response to immunotherapy in glioma patients. Immunohistochemistry and cellular experiments were performed for validation.

Results

Two m6A subtypes of gliomas were identified. The m6A-low-risk subtype was characterized by paucity of immune infiltrates; While the m6A-high-risk subtype had higher abundances of multiple immune cells including lymphocyte and macrophage as well as increased expression of PD-L1, corresponding to an immunosuppressive phenotype. The m6A-high-risk subtype had poorer survival than the m6A-low-risk subtype in both the glioblastoma and lower grade gliomas cohorts. Eight m6A-related hub genes of high prognostic significances were identified and selected for developing a scoring signature termed as m6Ascore. Elevated m6Ascore indicated worse survival for glioma patients under standard care, but showed enhanced response to immunotherapy. Moreover, we demonstrated that overexpression of FTO, a m6A demethylase, inhibited the expressions of m6A-related hub genes (PTX3, SPAG4), impaired glioma cell viability and reduced macrophage chemotaxis.

Conclusion

This work develops an immune- and clinical-relevant m6A subtyping and a scoring model, which enhances our understanding of the role of m6A modification in regulating immune infiltration microenvironment in gliomas and helps to identify patients who are more likely to benefit from immunotherapy.

Peer Review reports

Introduction

Diffuse gliomas represent the most frequent malignant primary tumors in the central nervous system (CNS) in adult, which are classified into astrocytoma, oligodendroglioma or glioblastoma by the microscopic morphology of glial cells or their precursors. Then a malignant grade is assigned according to the mitotic activity and presence of microvascular proliferation and necrosis. Traditionally, such histopathological manifestation is the basis for glioma diagnosis and treatment. However, gliomas demonstrate significant heterogeneity. Notable difference in survival can be observed in patients with the same histological type of glioma. Over the last couple of decades, the discovery of genetic alterations, such as isocitrate dehydrogenase (IDH) mutations and chromosome 1p/19q co-deletion, as well as the identification of transcriptome profiling-based subtypes and multi-genes signatures have greatly improved molecular understanding of gliomas, ultimately leading to incorporate molecular characteristics into the diagnosis of gliomas and provide more objective and precise stratifications for prognostication and even individualized therapy [1,2,3].

Despite the substantial advances in cancer genetics and molecular pathology, diffuse gliomas are still universally intractable. The median survival of patients with glioblastoma (GBM, WHO grade 4), the most common and aggressive form of gliomas, is less than 20 months even under the current standard treatment including surgical resection, chemoradiotherapy and tumor-treating fields. Lower grade gliomas (LGGs, WHO grade 2 and 3) have relatively favorable prognosis; however, they remain invasive and inevitably relapse. Immunotherapy, represented by immune checkpoint inhibitor (ICI) against programmed cell death-1 (PD-1), has revolutionized the treatment paradigm of various cancers and shows promising efficacy in the pre-clinical research of gliomas [4]. Unfortunately, several large randomized phase III trials of ICI failed to demonstrate survival benefits for unselected patients [5,6,7]. Survival gain was only observed in less than 10% of the enrolled patients receiving ICI treatment. The uniqueness of the immune landscape in brain tumors and inter-tumor heterogeneity at the immunological level may account for the unsatisfactory outcomes in glioma immunotherapies. Gliomas are generally considered as immunologically ‘cold’ tumors with less lymphocyte infiltration, but lots of immunosuppressive components. Dissection of the intrinsic properties of tumor immune microenvironment (TIME) is essential for identifying effective biomarkers for immunotherapy, and developing novel targets to counter the immunosuppressive microenvironment in gliomas.

The post-transcriptional modification of RNA has emerged in recent years as a new regulatory mechanism controlling gene expression in eukaryotes. N6-methyladenine (m6A) is the most prevalent form of epigenetic modification in messenger RNA. m6A modification sites have been found in approximately one-third of the mammalian transcriptome and most of which are evolutionarily conserved [8]. m6A modification can be reversibly and dynamically regulated by methyltransferases (also known as ‘writers’) and demethylases (also known as ‘erasers’), then recognized by binding proteins (also known as ‘readers’) to influence RNA fates involving nuclear export, alternative splice, stability and translation. Dysregulation of the cross-talk among m6A regulators (writers, erasers and readers) is frequently observed in various cancers, including gliomas, playing a crucial role in tumorigenesis and progression [8,9,10]. The role of m6A modification in tumor-immunity interaction is also being elucidated. A recent study found that demethylase ALKBH5 facilitated glioma-associated macrophage recruitment and generated an immunosuppressive microenvironment by a m6A-dependent manner [11]. Yet, the comprehensive effect of m6A regulators-mediated modification network in regulating glioma immune microenvironment and response to immunotherapy remains unclear. In the present study, we integrated genomic and transcriptomic information of 1637 samples from public datasets to determine m6A modification patterns in gliomas, and identified clinically relevant subtypes of gliomas featured by abnormalities in m6A regulators.

Methods

Data source and processing

Genomic, transcriptomic and clinicopathological information of 656 diffuse glioma samples were downloaded from UCSC Xena browser TCGA Pan-Cancer cohort (Toil RNA-seq recomputed project; https://xenabrowser.net) [12]. Two independent datasets from the Chinese Glioma Genome Atlas (CGGA) database including 981 samples of gliomas (http://www.cgga.org.cn) were employed for external validation [13]. Transcriptomic and clinical data of patients with GBM underwent anti-PD-1 therapy was provided by Dr. Zhao [14] for prediction of responsiveness to ICI immunotherapy. Responders were defined as: (1) radiologically stable or shrinking for at least six months; (2) very few to no tumor cells in surgical samples after immunotherapy.

Only protein-coding genes were extracted for RNA-seq expression analysis using the normalized transcripts per million (TPM) value. Genes with null expression or TPM less than 0.01 in more than 75% of samples were deleted, followed by removal of genes with median TPM expression < 0.1. The filtered TPM values were then log transformation and normalization to the median expression of 11 housekeeper genes [15]. Tumor mutation burden (TMB) was calculated by the sum of the number of non-synonymous mutation. Neoantigen counts of TCGA samples were obtained from Thorsson’s study [16].

Differential expression and gene set enrichment analyses

RNA-seq read counts were employed for identification of differential expression genes (DEGs) by using R package Deseq2. The fold change and p-value created by Deseq2 DEG output was applied to generate a ranked list of genes with formula log10(P) × sign(log2(fold change)) [17] for gene set enrichment analysis (GSEA) and visualization using clusterProfiler package [18]. Gene sets, including hallmark genes, C2: canonical pathway and C5: gene ontology, downloaded from the MSigDB database (vision 7.5.1) were conducted for functional annotations [19].

Fold change to median TPM value of each gene was calculated, then log2 transformation, and used for performing single-sample gene set enrichment analysis (ssGSEA) by GSVA package [20]. Gene sets comprised of transcripts with characteristic of an established lymphocyte signature [21] and 20 immune cell types [22] were employed for calculation of ssGSEA enrichment score, representing the abundance of intratumoral infiltrates. CIBERSORT deconvolution method and signature matrix “LM22” were used to validate the fractions of various immune cells [23]. The level of total immune infiltration was estimated by ESTIMATE algorithm [24]. A machine learning algorithm termed ImmunoPhenoScore (IPS) developed by Charoentong et al. [22] was used for prediction of response to ICI immunotherapy.

Protein-protein interaction, unsupervised clustering and principal component analysis

Principle component analysis (PCA) was performed using package ade4 with the normalized TPM values. Unsupervised clustering was implemented using ConsensuClusterPlus package and k-means method with 80% item resampling and 1000 times repetition [25].Consensus heatmap and cumulative distribution function were employed to determine the optimal K-value. Protein-protein interaction (PPI) network was analyzed by STRING online tool (https://string-preview.org) and visualized by Cytoscape software (Vision 3.7.2) [26].

Generation of a m6A-related risk scoring model

We used a classical penalized regression model to develop a multi-genes signature for risk prediction [27]. glmnet package was used to perform the least absolute shrinkage and selection operator (LASSO) on a cox proportional hazard regression model to remove redundant genes and identify genes with best prognostic significances. The linear combination of the expressions of filtered genes weighted by estimated regression coefficients in multivariate analysis was applied to calculate the risk score of each patient. X-tile software (version 3.6.1) [28] was employed to generate an optimal cutoff value in the training dataset to separate patients into two groups.

Immunohistochemistry

Paraffin-embedded specimens of 181 gliomas of different histological types were collected from Sun Yat-sen University Cancer Center (SYSUCC) at the duration of 2010 to 2016 and assembled in tissue microarrays (TAM). Informed consent was obtained from the patients and the study was approved by the Ethics Committee of Sun Yat-sen University Cancer Center. TAM was stained with GDF15 (Abcam, Cambridge, UK), PTX3 (Proteintech, Rosemont, IL), SPAG4 (Proteintech) antibodies and immune cell markers including CD4, CD8 and CD163 (Absin, Shanghai, China) by immunohistochemistry (IHC) method as previously described [29]. All the samples were evaluated independently by two neuropathologists.

Cell culture and plasmid transfection

Glioma cell lines U87, T98 and microglia cell line HMC3 were obtained from the State Key Laboratory of Oncology in South China. These cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Corning Inc., Corning, NY) with 10% fetal bovine serum (Gibco, ThermoFisher, Hampton, NH) and antibiotics (Gibco) at 5% CO2 and 37 °C. Plasmid clones expressing FTO (FTO-OE) and control empty vector were purchased from OBiO Biological Inc. (Shanghai, China) and transfected into glioma cells using Lipofectamine 2000 kit (Invitrogen, Waltham, MA) following the manufacturer’s protocol.

Immunoblotting

Cell lysates were subjected to SDS–polyacrylamide gel electrophoresis, then transferred to polyvinylidene difluoride membranes (Merck Millipore, Billerica, MA). The membranes were incubated with primary antibodies at 1:1000 dilution and 4 °C overnight (anti-FTO, anti-SPAG4, anti-PTX3: Proteintech; anti-GDF15: Abcam). After three washes with 1× Tris-buffered saline containing Tween 20 (TBST) for 3 times, the membrane was incubated with secondary antibody labeled with horseradish peroxidase for chemiluminescence detection. Band densities were normalized against the loading control GAPDH (CST, Danvers, MA) and TUBULIN (Proteintech) using ImageJ software.

Cell proliferation assay

For CCK8 assay, cells were seeded at concentration of 1000 per well in 96-well plates with fresh medium. Cell viability was tested using Cell Counting Kit-8 (CCK8, Dojindo, Kumamoto, Japan) at the time in 0, 24, 48 and 72 h. The microplates were incubated at 37°C for additional 4 h. Absorbance was read at 450nm using a microplate reader (ThermoFisher). For Edu (5-Ethynyl-2’-deoxyuridine) assay, 1 × 105 logarithmic growth stage cells were seeded in 24-well plates with corresponding concentration of Edu reagent for 2 h. Cells were washed and incubated with 4% Paraformaldehyde for 30 min, then permeated with 0.3% TritonX-100 in PBS and dyed with reaction solution (C0075S, Beyotime, Shanghai, China). The images were collected using an upright fluorescence microscope.

Migration assay

1 × 104 glioma cells were suspended in 900µL complete medium and plated in lower chamber of Transwell inserts at the first day. 5 × 103 HMC3 cells (a human-derived macrophage cell line) suspended in 100µL serum-free medium were seeded in the upper chamber (pore size: 8 μm; FALCON, Corning Inc.) at the next 12 h. After incubation for 24 h at 37 °C, the non-migrated microglia cells in the insert were wiped away with a cotton swab, whereas the microglia cells migrating the porous membrane were fixed with 4% paraformaldehyde for 10 min and stained with DAPI for 5 min. After the inserts were completely dried, 10 random fields of the migrated cells were counted under an upright fluorescence microscope.

Statistical analysis

Normalized expression values and calculated scores between two groups were compared by Wilcoxon tests. Kaplan-Meier analyses with log-rank tests were conducted to compare survival differences. Univariate and multivariate Cox regression were performed to calculate hazard ratio (HR). Stepwise backward method was chosen in multivariate analysis. Time-dependent receiver operating characteristic (ROC) curve was used for assessment of the specificity and sensitivity of prognostic factors using survivalROC package. Spearman correlations between variables were analyzed using corrplot package. R (vision 3.6.2) and GraphPad Prism (version 8.0.1) software were used for statistical analyses. P < 0.05 was considered statistically significant.

Results

Identification of clinically relevant m6A subtypes in gliomas

A total of 18 well-known m6A regulators (7 writers, 2 erasers and 9 readers) was included for analysis in this study [30]. The majority of the m6A regulators were up-regulated in gliomas compared to normal brain tissue in TCGA dataset (supplementary Figure S1). The expression levels of the m6A regulators were also associated with survival outcomes of glioma patients. For instance, higher levels of METTEL14, ZC3H13, FTO and YTHDC1, along with lower levels of WTAP, YTHDF2 and IGF2BP2, indicated a more favorable overall survival (OS, supplementary Figure S2). Significant correlations of the expression levels among the m6A regulators were observed in gliomas (supplementary Figure S3). Additionally, PPI analysis revealed dense interactions and functional connections among these m6A regulators (Fig. 1A). These findings suggest that m6A regulators-mediated modification network could potentially have an impact on glioma tumorigenesis and progression. In order to categorize glioma subtypes with distinct m6A modification patterns, a consistent unsupervised analysis based on the comprehensive expression profiles of the 18 m6A regulators was performed on 656 samples from TCGA dataset. Four clusters were eventually identified (Fig. 1B, C and supplementary Figure S4A, B). The expression levels of the m6A regulators were notably different among these four m6A clusters (Fig. 1C). Moreover, the PCA result depicted a diversity in the expression portraits within the four m6A clusters, confirming the effectiveness of the unsupervised clustering in gliomas (Fig. 1D).

Fig. 1
figure 1

m6A modulation patterns and clinical relevance in diffuse gliomas. A, Protein-protein interaction between m6A regulators. The circle size represented the connection strength of each node. Writers were colored by red in the circle; erasers were colored by blue; readers were colored by purple in the circle. B, Consistent clustering based on the comprehensive expression profiles of 18 m6A regulators. C, Expressions of the m6A regulators among different m6A clusters. The subtypes of gliomas were derived from previous publications by TCGA Research Network (Ceccarelli, et al. Cell, 2016). D, Principle component analysis for the transcriptome profiles of the four m6A clusters. E, The proportions of different m6A clusters in GBM and LGG. F, The survival differences of the four m6A clusters. G, The proportions of combination of chromosome 7 gain and chromosome 10 loss in the m6A-high-risk and m6A-low-risk subtypes. H-I, The frequencies of genetic alteration events in different m6A subtypes from LGG (H) and GBM (I) cohorts. J-K, Comparisons of survival between different m6A subtypes from LGG (J) and GBM (K) cohorts

Full size image

Then, we investigated the clinical relevance of the m6A clusters. We found that gliomas categorized into m6A-cluster 1 and 2 accounted for 90% of the GBM samples from TCGA dataset, whereas gliomas belonging to cluster 3 and 4 constituted nearly 85% of the LGG samples (Fig. 1E). Furthermore, patients with gliomas classified into cluster 1 demonstrated comparable survival outcome to those in cluster 2, while the outcomes for patients in cluster 3 and 4 were similar (Fig. 1F). Hence, we combined cluster 1 and 2 into one subtype (termed as m6A-high-risk subtype), while amalgamating cluster 3 and 4 into a separate group (termed as m6A-low-risk subtype) for further analyses. The expression of most m6A regulators (except for WTAP and IGF2BP2) was down-regulated in m6A-high-risk subtype compared to m6A-low-risk subtype (supplementary Figure S4C). Genomic analyses revealed that in both LGG and GBM cohorts, m6A-high-risk subtype exhibited higher frequencies of molecular alterations associated with inferior survival, including the combined chromosome 7 gain and chromosome 10 loss (7+/10-), EGFR amplification and CDKN2A/B loss. Conversely, IDH1 mutation, which has been linked to a favorable outcome, was more likely to appear in m6A-low-risk subtype (Figure G-I). More importantly, gliomas of m6A-high-risk subtype showed significantly worse OS than low-risk subtype in both the GBM and LGG cohorts (Figure J, K). The above findings supported the application of m6A regulators-based clustering methodology as a novel molecular classification approach for identifying an aggressive subtype in gliomas, complementing histological classification and grading.

Functional annotations and immune infiltration characteristics in different m6A subtypes

The differential expression in genes between the m6A-high-risk and m6A-low-risk subtypes of gliomas was analyzed for exploring biological behaviors mediated by m6A modification patterns. This exploration has the potential to elucidate the survival disparities observed in m6A subtyping. Given that the m6A-high-risk subtype is primarily identified in GBM, while the majority of m6A-low-risk subtype is presented in LGG, we separately examined the differential expression of genes in these two m6A subtypes within their respective GBM and LGG cohorts to mitigate potential confounding effects stemming from inherent disparities between GBM and LGG. As shown in the Venn plot (Fig. 2A), 211 up-regulated genes and 78 down-regulated genes were identified in m6A-high-risk subtype versus m6A-low-risk subtype in both the GBM and LGG cohorts. We regarded these 289 DEGs as hub genes closely associated with m6A modification network, and were subsequently used for annotating biological functions. These hub DEGs were found to be enriched in numerous pathways involved in the regulation of tumor immunology, such as lymphocyte chemotaxis, monocyte chemotaxis, cytokine interactions and cancer immunotherapy by PD-1 blockade (Fig. 2B). Consistently, the results of GSEA showed that interferon-γ, interleukins and T-cell receptor signaling, as well as antigen processing and cell killing pathways, were enriched in the m6A-high-risk subtype versus the m6A-low-risk subtype (Fig. 2C; supplementary Table S1 and S2), indicating the potential role of m6A modification in skewing immune infiltration and response in gliomas. Subsequently, we calculated immune score of each sample using the ESTIMATE algorithm that reflects the purity of tumor and the level of immune infiltrates. m6A-high-risk subtype exhibited a higher immune score than m6A-low-risk subtype (Fig. 2D-E). ssGSEA enriched scores were then used for profiling the abundances of various immune cells infiltrating into the tumor. Significant increases in multiple immune components, including CD8+ T cell, regulatory T cell (Treg), suppressor macrophage (M2) and myeloid-derived suppressor cell (MDSC) were observed in m6A-high-risk subtype compared to m6A-low-risk subtype in both the GBM and LGG cohorts (Fig. 2F). CIBERSORT deconvolution method was employed to validating the estimation of intratumoral immune infiltrates. In comparison to m6A-low-risk subtype, the abundances of CD8+ lymphocyte and M2 macrophage were significantly elevated in m6A-high-risk subtype (supplementary Figure S5). Chemokines are increasingly recognized for their chemotactic effects to recruit immune cells into tumors [31]. We found that multiple chemokines, such as CCL2 and CCL5, were significantly up-regulated in gliomas of m6A-high-risk subtype (Fig. 2G).

Fig. 2
figure 2

Immune infiltration characteristics in different m6A subtypes. A, Venn plot showed overlap of differential expression genes (DEGs) between the m6A-high-risk and m6A-low-risk subtypes. B, Functional annotations of the overlapping DEGs. C, Gene sets enriched in m6A-high-risk subtype versus m6A-low-risk subtype in GBM cohort. D-E, Comparisons of immune score calculated by ESTIMATE algorithm between the m6A-high-risk and m6A-low-risk subtype in GBM (D) and LGG (E) cohort. F, Comparisons of abundances of various immune cells estimated by ssGSEA enriched score between the two m6A subtypes. G, Comparisons of the expressions of multiple chemokines between the two m6A subtypes. ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001

Full size image

Establishment of a m6A-related multi-genes signature with prognostic significance

To enhance the utility of the m6A subtyping for predicting clinical outcomes, we sought to establish a quantitative model using the expression of m6A hub genes to predict the OS of glioma patients. LASSO penalized Cox proportional hazards regression was employed to screen genes with best prognostic contribution among the 289 hub genes (Fig. 3A). A set of eight genes were obtained (Fig. 3B), and their expressions were highly correlated with m6A regulators (Fig. 3C). A risk score, termed as m6Ascore, was further calculated on each patient using a formula based on the expression values of these eight genes weighted by their regression coefficients in multivariate analysis as follow:

Fig. 3
figure 3

Development of a multi-genes signature with prognostic significance. A, LASSO penalized regression identified eight genes with optimal prognostic contributions. B, hazard ratios of the eight prognostic genes in univariate Cox proportional hazard regression analyses. C, The correlations between the expressions of prognostic genes and m6A regulators. The genes labeled by red indicated risk genes; The genes labeled by blue indicated protected genes; The genes labeled by black indicated m6A regulators. The color of dot represented correlation coefficient; The size of dot represented p-value. D, ROC curve analysis of the m6Ascore for predictiing overall survival. E, The correlations of m6Asocre and immune socre calculated by ESTIMATE algorithm. F-H, Survival differences between high and low m6Ascore groups in all gliomas (H), GBM (I) and LGG (J) cohort. I and J, Survival differences between high and low m6Ascore groups of gliomas from the CGGA cohort 1 (I) and CGGA cohort 2 (J)

Full size image

m6Ascore = (0.173 * expression value of ARL9) - (0.532 * expression value of COMTD1) + (0.135 * expression value of DDIT4L) + (0.150 * expression value of GDF15) + (0.240 * expression value of MCUB) + (0.122 * expression value of PTX3) + (0.243 * expression value of SPAG4) - (0.114 * expression value of ZC3H12B).

A higher m6Ascore was observed in GBM compared to LGG, and in Glioma belonging to the m6A-high-risk subtype compared to the m6A-low-risk subtype (supplementary Figure S6). The area under curve (AUC) of m6Ascore in ROC analysis for predicting 3- and 5-year OS of glioma patients was 0.90 and 0.87, respectively, showing a remarkable accuracy (Fig. 3D). In subgroup analysis, the AUC of prediction of 5-year survival is 0.86 for GBM patients and 0.77 for LGG patients. In addition, significant correlations were observed between the m6Ascore and the immune score calculated using the ESTIMATE algorithm in both the GBM and LGG cohorts (Fig. 3E). Immune molecules, including PD-L1, B7-H3, GATA3 and LGALS3, have been found to be associated with prognosis in GBM [32, 33]. Significant correlations were also observed between the m6Ascore and the expression levels of these immune-related genes (supplementary Figure S7). These findings highlight the strong connection between m6A modification networks and glioma immunity and progression. Subsequently, X-tile software was applied to generate the optimal cutoff value (-2.70) of the m6Asocre for stratifying TCGA patients into the low and high m6Ascore groups. Glioma patients in the high m6Ascore group possessed worse survival than those in the low m6Ascore group (Fig. 3F). A higher m6Ascore indicated poorer prognosis in both the GBM and LGG cohorts (Fig. 3G, H). Furthermore, the prognostication value of the m6Ascore was validated in two additional datasets from the CGGA database (Fig. 3I, J). When evaluated as a continuous variable, m6Ascore still showed a significant association with OS, and emerged as an independent prognostic factor in multivariate analyses (Table 1).

Table 1 Univariate and multivariate Cox regression analyses for overall survival of glioma patients in public datasets
Full size table

Experimental validation of the m6A-related hub genes

GDF15, PTX3 and SPAG4, three of the eight component genes of the m6Asocre were selected for experimental validation. Representative IHC images of GDF15, PTX3 and SPAG4 expressed in glioma samples from the SYSUCC cohort were shown in Fig. 4A. Patients with positive expression of GDF15, PTX3 or SPAG4 had significantly improved survival than those with negative expression (Fig. 4C). Besides, we detected the expression of T lymphocyte markers CD4, CD8 and suppressive macrophage marker CD163 (Fig. 4B). In comparison to gliomas with negative expression of GDF15, PTX3 and SPAG4, positive expression groups showed significant enrichment of lymphocyte and macrophage (Fig. 4D, E).

Fig. 4
figure 4

Immunohistochemistry validation of the prognostic significance and immune relevance of the m6Ascore. A-B, Representative images of the expressions of m6Asocre component genes (A) and immue cell markers (B) in glioma samples from the SUSYCC cohort. C, Survival comparisons of gliomas between positive and negative expression group. D-E, Comparions of infiltrating lymphocyte (D) and macrophage (E) count between positive and negative expression group. **, P < 0.01; ***, P < 0.001

Full size image

Significant negative correlations were identified in the expression of FTO with GDF15, PTX3 and SPAG4 in gliomas from TCGA dataset (Fig. 3D and supplementary Figure S8A-C). As a m6A demethylase, FTO may participate in the regulation of GDF15, PTX3 and SPAG4 expression, thus affecting glioma immune infiltration. To verify this hypothesis, we enhanced the expression of FTO in glioma cell lines U87 and T98 by transfection of FTO plasmid. The overexpression of FTO significantly inhibited the expression of PTX3 and SPAG4 compared to control groups (Fig. 5A). In addition, increased FTO significantly impaired the proliferation of glioma cells (Fig. 5B-D) and reduced macrophage chemotaxis toward glioma cells in a co-culture model (Fig. 5E-F). In TCGA dataset, the expression of FTO in gliomas was also negatively correlated with intratumoral macrophage abundances (supplementary Figure S8D, E). These findings indicated the roles of FTO and the underling m6A modification mechanism in regulating glioma immune infiltration microenvironment.

Fig. 5
figure 5

The role of FTO in regulating glioma proliferation and macrophage chemotaxis. A, Overexpression of FTO inhibited the expressions of PTX3 and SPAG4. B-D, Overexpression of FTO inhibited the proliferation of glioma cells in CCK8 (B) and Edu (C-D) assay. E-F, Overexpression of FTO reduced macrophage migration in transwell assay. The white bar represents 100 μm. *, P < 0.05; **, P < 0.01

Full size image

m6Ascore predict response to immunotherapy

As mentioned above, higher m6Ascore was associated with elevated abundances of immune infiltrates in gliomas, we thus investigated the role of m6Ascore to predict response to immunotherapy in patients with malignant gliomas (WHO grade 3 and grade 4). By using TCGA data, a significant elevation of PD-L1 expression, lymphocyte infiltration score, IPS, TMB and neoantigen counts were observed in gliomas stratified to the high m6Ascore group as compared to the low m6Ascore group (Fig. 6A-E). These immune and genomic markers have been widely accepted as candidates for predicting response to immunotherapy in multiple solid cancers. Then, we analyzed transcriptomic and clinical data from a published cohort of glioma patients receiving anti-PD-1 treatment. We found that patients in the high m6Ascore group had a higher proportion to achieve long-term response (Fig. 6F), and correspondingly manifested a significant improvement in OS compared to the low m6Ascore group (Fig. 6G). These findings have clinical implication for utilizing the m6Ascore to identified responders to anti-PD-1 immunotherapy.

Fig. 6
figure 6

m6Ascore predict the response to immunotherapy. A-E, Comparisons of PD-L1 expression (A), lymphocyte infiltration signature score (B), immunophenoscore (IPS; C), tumor mutation burden (TMB; D) and neoantigen count (E) between the high and low m6Ascore groups of malignant gliomas in TCGA dataset. F, The proportion of glioma patients responded to anti-PD-1 treatment. G, Survival difference between the high and low m6Ascore groups of glioma patients receiving anti-PD-1 treatment. *, P < 0.05; ***, P < 0.001

Full size image

Discussion

Genomic and epigenomic profiles of cancers have been utilized for molecular classification. Using gene expression analysis and unsupervised clustering methodology, TCGA research network initially classified gliomas into four molecular subtypes: proneural, mesenchymal, classical, and neural [34]. The proneuronal subtype is common in IDH-mutant glioma and tends to survive longer, but remains controversial [3]. Subsequently, TCGA proposed a new epigenetic classification for gliomas based on DNA methylation profiles, which associated with IDH status and clinical outcomes [35]. Despite this advancement, the utilization of methylation sequencing technique and complex bioinformatic analysis remains challenging for widespread application in clinical practice. Here, we identified two subtypes of gliomas with distinct m6A methylation modification patterns, and developed a m6A-related scoring modeltermed m6Ascore to quantify m6A modification patterns. These subtyping and scoring models are based on the expression of a dozen genes that can estimate the feature of TIME and predict the prognosis in gliomas, thereby providing a simpler approach for clinical application.

It has been known that dysregulation of cross-talk among m6A regulators reprograms the epitranscriptome and affects global gene expressions. Aberrantly epigenetic modification on m6A-containing transcripts with oncogenic or tumor-suppressive functions may contribute to cancer initiation, progression and therapy resistance [8, 10]. Therefore, the comprehensive expression profiles of m6A regulators can be used to identify tumors with different m6A modification patterns. In the present study, we identified two subtypes of gliomas with distinct m6A modification patterns and clinical outcomes, which were termed as m6A high-risk and m6A low-risk subtype. Patients with gliomas classified as the m6A high-risk subtype had worse survival compared to the m6A low-risk subtype. Although prior studies have reported the clinical relevance of m6A clustering in gliomas [36,37,38], they have ignored the fact that these two m6A subtypes are respectively enriched in GBM and LGG. We are the first to demonstrate that the prognostic significance of m6A subtyping was independent of the malignant grade of gliomas, as evidenced by stratifying the samples into GBM and LGG subgroup for survival analyses. Recent research elucidates that IDH-wildtype astrocytoma with one of the following molecular alterations: chromosome 7+/10-, EGFR amplification and TERT promoter mutation has similar prognosis with histologically diagnosed GBM, which can be defined as molecular GBM [39]. In addition, appearance of CDKN2A/B homozygous deletion in LGG should be reclassified as glioma of WHO grade 4 [40]. We found that such molecular markers were more prevalent in gliomas classified as the m6A high-risk subtype in both the GBM and LGG cohorts, indicating that m6A phenotypes can provide a more accurate reflection of the molecular nature of gliomas compared to WHO grading. Moreover, our findings have practical implications for using the m6A subtyping to identify aggressive subtypes in gliomas following histological diagnosis. The high-risk patients require intensified follow-up and consideration of a more rigorous therapeutic regimen.

Immune dysfunction is a crucial factor facilitating tumor progression. Takashima et al. [32, 33] found that a low Th2 balance, low activity of the PD-L1/PD-1 axis and the expression of immunosuppressive genes, including CD276 and LGALS3, were associated with the progression and prognosis in gliomas. In the current study, gene sets enrichment analysis and experimental validation indicate that m6A epigenetics regulates glioma progression via the modulation of the TIME. Glioma classified as the m6A-low risk subtype was characterized by paucity of immune infiltrates, corresponding to immune-ignorant phenotype in an immunological classification proposed by Chen and colleagues [41]. Relatively, the m6A high-risk subtype exhibited higher abundances of multiple immune infiltrates, including CD8+ lymphocyte and immunosuppressive cells (e.g., Treg, M2 macrophage and MDSC) as well as increased expression of PD-L1, corresponding to the immune-exhausted phenotype. Immunosuppressive cells, specifically macrophages, and co-inhibitory checkpoint molecules like PD-1/PD-L1 are crucial factors in the generation and maintenance of a suppressive TIME in gliomas. Glioma-associated macrophages can dampen T cell-mediated antitumor response by various approaches, such as secretion of suppressive cytokines, induction of checkpoint molecules and inhibition of antigen presenting cell. Meanwhile, macrophage is able to promote tumor progression by directly supporting cancer stem cell, angiogenesis and metastasis [42]. The immunosuppressive and pro-tumoral microenvironment in glioma classified as the m6A high-risk subtype explains its worse clinical outcome.

The m6A-related scoring model comprises eight genes: ARL9, COMTD1, DDIT4L, GDF15, MCUB, ZC3H12B, PTX3, and SPAG4. All of these genes have been found to be involved in immune regulation and inflammation. Tan et al. [43] found that the low expression or methylation of ARL9 was related with enriched CD8+ T cells and better OS in LGG. COMTD1 is a mitochondrial-related genes involved in the pathogenesis of systemic lupus erythematosus [44]. DDIT4L modulates cellular metabolic stress responses and innate immune activity during development [45]. Roth et al. found that GDF-15 contributes to proliferation and immune escape of malignant gliomas [45]. GDF-15 also emerges as a regulator of T cell extravasation into the tumor microenvironment [46]. MCUB is required for the suppression of proinflammatory metabolism and for the acquisition of an anti-inflammatory phenotype in macrophages [47]. ZC3H12B participates in macrophage-mediated immune scape in colon cancer [48]. PTX3 plays a pivotal role in innate immune response and is a prognostic marker for gliomas [48, 49]. SPAG4 has been identified as being associated with immune cell infiltration in periodontitis [50]. In this study, we found that overexpression of FTO significantly reduced the expression of PTX3 and SPAG4, attenuated macrophage chemotaxis toward glioma cells and inhibited tumor proliferation. Moreover, the correlations of expression of FTO with PTX3 and SPAG4 as well as intratumoral macrophage infiltration were confirmed at the histological level. Although FTO is the first identified m6A demethylase in epigenetic research and is highly expressed in the brain, its potential contribution to glioma remains controversial [9, 51]. Jiang et al. [52] and Sarah et al. [53] found that deletion of FTO increased the overall m6A levels at transcriptome and impaired glioma proliferation. Conversely, other literatures revealed that decreased expression of FTO was associated with higher glioma grade and worse clinical outcome. Knockdown of FTO promoted the growth of glioma cells in vitro and in vivo [51]. To our knowledge, this is the first experimental research to demonstrate the effect of FTO in regulating glioma-associated macrophages. As a key enzyme in m6A modification, FTO may reduce macrophage infiltration by inhibiting the secretion of inflammatory factors, such as PTX3 and SPAG4, in a m6A-dependent manner. Furthermore, these genes may serve as novel targets for glioma immunotherapy.

Cancers with distinct TIME characteristics require different immunotherapeutic strategies. Immune-ignorant cancers should consider immune enhancement therapies such as engineered T cells and cancer vaccines. Whereas, immune-exhausted cancers may benefit from immunotherapies targeting the PD-1/PD-L1 pathway, which is one of the earliest and most well-characterized mechanisms contributing to intratumoral T cell dysfunction and tumor immune evasion [41]. Actually, exhausted T cells are not fully tolerance but retain the expansion capacity and cytotoxic potential. Blocking the PD-1/PD-L1 signaling selectively restores a natural immune effect against tumors without systemic immune overactivation [54]. Hallmarks of intratumoral immune exhaustion, including lymphocyte infiltration and PD-L1 expression, have been accepted as biomarkers for predicting response to anti-PD-1 treatment [55]. Immune score represents a novel method to identifying responders. Using TCGA transcriptomic data and machine learning algorithms, Charoentong et al. [22] proposed a scoring scheme, termed IPS, to determine pan-cancer immunophenotypes and predict responses to ICI. Higher IPS indicated better response. Nevertheless, these biomarkers have not been widely validated in their use for gliomas due to the lack of available tumor samples and sequencing data in high-quality clinical research. The identification of biomarkers for clinical response remains a major obstacle hindering advancements in glioma immunotherapy. In our study, the m6Ascor can effectively reflect the abundance of immune infiltrates and PD-L1 expression in gliomas. Additionally, m6Ascore was highly correlated with IPS calculated in patients with malignant gliomas, suggesting its potential to predict response to immunotherapy. Another piece of direct evidence is that elevated m6Asocre was predominantly observed in patients who responded to anti-PD-1 immunotherapy, correlating with favorable survival outcome in a previously published patient cohort. Our data provided preliminary evidence for using the m6Ascore to identify glioma patients who are more likely to benefit from ICI therapy; however, further validation with larger clinical samples is warranted.

Conclusions

This work identified two subtypes of gliomas with distinct m6A methylation modification patterns and showed the potential of epigenetic modification on regulating glioma immune infiltration environment. We also developed a m6A-related scoring model termed m6Ascore to quantify m6A modification patterns as well as predict response to ICI immunotherapy and clinical outcomes. We believe that our findings will enhance the understanding of the heterogeneity and plasticity of glioma immune microenvironment and guide more individualized therapeutic strategies.

Data availability

Publicly available datasets and online tools used in this study have been described in the methods section. The data of histological and cellular experiments is available from the corresponding authors on reasonable request.

Abbreviations

CNS:

Central nervous system

IDH:

Isocitrate dehydrogenase

GBM:

Glioblastoma

LGG:

Lower grade glioma

ICI:

Immune checkpoint inhibitor

PD-1:

Programmed cell death-1

TIME:

Tumor immune microenvironment

m6A:

N6-methyladenine

CGGA:

Chinese Glioma Genome Atlas

TCGA:

The Cancer Genome Atlas

TMB:

Tumor mutation burden

DEG:

Differential expression gene

GSEA:

Gene set enrichment analysis

PCA:

Principle component analysis

PPI:

Protein-protein interaction

IHC:

Immunohistochemistry

LASSO:

Least absolute shrinkage and selection operator

HR:

Hazard ratio

References

  1. Eckel-Passow JE, Lachance DH, Molinaro AM, Walsh KM, Decker PA, Sicotte H, et al. Glioma groups based on 1p/19q, IDH, and TERT promoter mutations in tumors. N Engl J Med. 2015;372(26):2499–508.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Louis DN, Perry A, Wesseling P, Brat DJ, Cree IA, Figarella-Branger D, et al. The 2021 WHO classification of tumors of the Central Nervous System: a summary. Neurooncology. 2021;23(8):1231–51.

    CAS  Google Scholar 

  3. Behnan J, Finocchiaro G, Hanna G. The landscape of the mesenchymal signature in brain tumours. Brain. 2019;142(4):847–66.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Waldman AD, Fritz JM, Lenardo MJ. A guide to cancer immunotherapy: from T cell basic science to clinical practice. Nat Rev Immunol. 2020;20(11):651–68.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Reardon DA, Brandes AA, Omuro A, Mulholland P, Lim M, Wick A, et al. Effect of Nivolumab vs Bevacizumab in patients with recurrent glioblastoma: the CheckMate 143 phase 3 Randomized Clinical Trial. JAMA Oncol. 2020;6(7):1003–10.

    Article  PubMed  Google Scholar 

  6. Omuro A, Brandes AA, Carpentier AF, Idbaih A, Reardon DA, Cloughesy T et al. Radiotherapy combined with nivolumab or temozolomide for newly diagnosed glioblastoma with unmethylated MGMT promoter: an international randomized phase III trial. Neurooncology. 2022.

  7. Weller M, Lim M, Idbaih A, Steinbach J, Finocchiaro G, Raval R, et al. CTIM-25. A randomized phase 3 study of nivolumab or placebo combined with radiotherapy plus temozolomide in patients with newly diagnosed glioblastoma with methylated mgmt promoter: checkmate 548. Neurooncology. 2021;23(Supplement6):vi55–6.

    Google Scholar 

  8. Huang H, Weng H, Chen J. M(6)a modification in Coding and non-coding RNAs: roles and therapeutic implications in Cancer. Cancer Cell. 2020;37(3):270–88.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Zhang Y, Geng X, Li Q, Xu J, Tan Y, Xiao M, et al. m6A modification in RNA: biogenesis, functions and roles in gliomas. J Experimental Clin cancer Research: CR. 2020;39(1):192.

    Article  CAS  PubMed Central  Google Scholar 

  10. Panneerdoss S, Eedunuri VK, Yadav P, Timilsina S, Rajamanickam S, Viswanadhapalli S, et al. Cross-talk among writers, readers, and erasers of m(6)a regulates cancer growth and progression. Sci Adv. 2018;4(10):eaar8263.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Dong F, Qin X, Wang B, Li Q, Hu J, Cheng X, et al. ALKBH5 facilitates Hypoxia-Induced Paraspeckle Assembly and IL8 Secretion to generate an immunosuppressive Tumor Microenvironment. Cancer Res. 2021;81(23):5876–88.

    Article  CAS  PubMed  Google Scholar 

  12. Vivian J, Rao AA, Nothaft FA, Ketchum C, Armstrong J, Novak A, et al. Toil enables reproducible, open source, big biomedical data analyses. Nat Biotechnol. 2017;35(4):314–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  13. Zhao Z, Zhang K-N, Wang Q, Li G, Zeng F, Zhang Y, et al. Chinese Glioma Genome Atlas (CGGA): a Comprehensive Resource with functional genomic data from Chinese glioma patients. Genom Proteom Bioinform. 2021;19(1):1–12.

    Article  CAS  Google Scholar 

  14. Zhao J, Chen AX, Gartrell RD, Silverman AM, Aparicio L, Chu T, et al. Immune and genomic correlates of response to anti-PD-1 immunotherapy in glioblastoma. Nat Med. 2019;25(3):462–9.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Eisenberg E, Levanon EY. Human housekeeping genes, revisited. Trends Genet. 2013;29(10):569–74.

    Article  CAS  PubMed  Google Scholar 

  16. Thorsson V, Gibbs DL, Brown SD, Wolf D, Bortone DS, Ou Yang TH, et al. The Immune Landscape of Cancer. Immunity. 2019;51(2):411–2.

    Article  CAS  PubMed  Google Scholar 

  17. Reimand J, Isserlin R, Voisin V, Kucera M, Tannus-Lopes C, Rostamianfar A, et al. Pathway enrichment analysis and visualization of omics data using g:Profiler, GSEA, Cytoscape and EnrichmentMap. Nat Protoc. 2019;14(2):482–517.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Yu G, Wang L-G, Han Y, He Q-YJO. clusterProfiler: an R package for comparing biological themes among gene clusters. 2012;16(5):284–7.

  19. Liberzon A, Subramanian A, Pinchback R, Thorvaldsdóttir H, Tamayo P, Mesirov JP. Molecular signatures database (MSigDB) 3.0. Bioinformatics (Oxford, England). 2011;27(12):1739-40.

  20. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Calabrò A, Beissbarth T, Kuner R, Stojanov M, Benner A, Asslaber M, et al. Effects of infiltrating lymphocytes and estrogen receptor on gene expression and prognosis in breast cancer. Breast Cancer Res Treat. 2009;116(1):69–77.

    Article  PubMed  Google Scholar 

  22. Charoentong P, Finotello F, Angelova M, Mayer C, Efremova M, Rieder D, et al. Pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade. Cell Rep. 2017;18(1):248–62.

    Article  CAS  PubMed  Google Scholar 

  23. Newman AM, Steen CB, Liu CL, Gentles AJ, Chaudhuri AA, Scherer F, et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol. 2019;37(7):773–82.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Yoshihara K, Shahmoradgoli M, Martinez E, Vegesna R, Kim H, Torres-Garcia W, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4:2612.

    Article  PubMed  Google Scholar 

  25. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinf (Oxford England). 2010;26(12):1572–3.

    CAS  Google Scholar 

  26. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Tang X-R, Li Y-Q, Liang S-B, Jiang W, Liu F, Ge W-X, et al. Development and validation of a gene expression-based signature to predict distant metastasis in locoregionally advanced nasopharyngeal carcinoma: a retrospective, multicentre, cohort study. Lancet Oncol. 2018;19(3):382–93.

    Article  CAS  PubMed  Google Scholar 

  28. Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. 2004;10(21):7252–9.

    Article  CAS  PubMed  Google Scholar 

  29. Hu WM, Zhang J, Sun SX, Xi SY, Chen ZJ, Jiang XB, et al. Identification of P4HA1 as a prognostic biomarker for high-grade gliomas. Pathol Res Pract. 2017;213(11):1365–9.

    Article  CAS  PubMed  Google Scholar 

  30. Li Y, Xiao J, Bai J, Tian Y, Qu Y, Chen X, et al. Molecular characterization and clinical relevance of m(6)a regulators across 33 cancer types. Mol Cancer. 2019;18(1):137.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Fan W, Wang D, Li G, Xu J, Ren C, Sun Z, et al. A novel chemokine-based signature for prediction of prognosis and therapeutic response in glioma. CNS Neurosci Ther. 2022;28(12):2090–103.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Takashima Y, Kawaguchi A, Hayano A, Yamanaka R. CD276 and the gene signature composed of GATA3 and LGALS3 enable prognosis prediction of glioblastoma multiforme. PLoS ONE. 2019;14(5):e0216825.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Takashima Y, Kawaguchi A, Kanayama T, Hayano A, Yamanaka R. Correlation between lower balance of Th2 helper T-cells and expression of PD-L1/PD-1 axis genes enables prognostic prediction in patients with glioblastoma. Oncotarget. 2018;9(27):19065–78.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. 2010;17(1):98–110.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Ceccarelli M, Barthel FP, Malta TM, Sabedot TS, Salama SR, Murray BA, et al. Molecular profiling reveals biologically discrete subsets and pathways of progression in diffuse glioma. Cell. 2016;164(3):550–63.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Chai RC, Wu F, Wang QX, Zhang S, Zhang KN, Liu YQ, et al. M(6)a RNA methylation regulators contribute to malignant progression and have clinical prognostic impact in gliomas. Aging. 2019;11(4):1204–25.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Lin S, Xu H, Zhang A, Ni Y, Xu Y, Meng T, et al. Prognosis analysis and validation of m(6)a signature and Tumor Immune Microenvironment in Glioma. Front Oncol. 2020;10:541401.

    Article  PubMed  PubMed Central  Google Scholar 

  38. Xu S, Tang L, Dai G, Luo C, Liu Z. Expression of m6A regulators correlated with Immune Microenvironment predicts therapeutic efficacy and prognosis in Gliomas. Front cell Dev Biology. 2020;8:594112.

    Article  Google Scholar 

  39. Stichel D, Ebrahimi A, Reuss D, Schrimpf D, Ono T, Shirahata M, et al. Distribution of EGFR amplification, combined chromosome 7 gain and chromosome 10 loss, and TERT promoter mutation in brain tumors and their potential for the reclassification of IDHwt astrocytoma to glioblastoma. Acta Neuropathol. 2018;136(5):793–803.

    Article  PubMed  Google Scholar 

  40. Appay R, Dehais C, Maurage CA, Alentorn A, Carpentier C, Colin C, et al. CDKN2A homozygous deletion is a strong adverse prognosis factor in diffuse malignant IDH-mutant gliomas. Neurooncology. 2019;21(12):1519–28.

    CAS  Google Scholar 

  41. Sanmamed MF, Chen L. A paradigm shift in Cancer Immunotherapy: from enhancement to normalization. Cell. 2018;175(2):313–26.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Quail DF, Joyce JA. The Microenvironmental Landscape of Brain tumors. Cancer Cell. 2017;31(3):326–41.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Tan Y, Zhang S, Xiao Q, Wang J, Zhao K, Liu W, et al. Prognostic significance of ARL9 and its methylation in low-grade glioma. Genomics. 2020;112(6):4808–16.

    Article  CAS  PubMed  Google Scholar 

  44. Sun Y, Geng Y, Zhao M, Wang F. Mitochondrial-Related Pathogenic Genes in Systemic Lupus Erythematosus: A Multi-Omics Mendelian Randomization Study. Available at SSRN 4986348.

  45. Michalski C, Cheung C, Oh JH, Ackermann E, Popescu CR, Archambault AS et al. DDIT4L regulates mitochondrial and innate immune activities in early life. JCI Insight. 2024;9(5).

  46. Haake M, Haack B, Schäfer T, Harter PN, Mattavelli G, Eiring P, et al. Tumor-derived GDF-15 blocks LFA-1 dependent T cell recruitment and suppresses responses to anti-PD-1 treatment. Nat Commun. 2023;14(1):4253.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Feno S, Munari F, Reane DV, Gissi R, Hoang DH, Castegna A, et al. The dominant-negative mitochondrial calcium uniporter subunit MCUb drives macrophage polarization during skeletal muscle regeneration. Sci Signal. 2021;14(707):eabf3838.

    Article  CAS  PubMed  Google Scholar 

  48. Ma YS, Wu TM, Ling CC, Yu F, Zhang J, Cao PS, et al. M2 macrophage-derived exosomal microRNA-155-5p promotes the immune escape of colon cancer by downregulating ZC3H12B. Mol Ther Oncolytics. 2021;20:484–98.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Garlanda C, Bottazzi B, Magrini E, Inforzato A, Mantovani A. PTX3, a Humoral Pattern Recognition Molecule, in Innate Immunity, tissue repair, and Cancer. Physiol Rev. 2018;98(2):623–39.

    Article  CAS  PubMed  Google Scholar 

  50. Gao X, Jiang C, Yao S, Ma L, Wang X, Cao Z. Identification of hub genes related to immune cell infiltration in periodontitis using integrated bioinformatic analysis. J Periodontal Res. 2022;57(2):392–401.

    Article  CAS  PubMed  Google Scholar 

  51. Tao B, Huang X, Shi J, Liu J, Li S, Xu C et al. FTO interacts with FOXO3a to enhance its transcriptional activity and inhibits aggression in gliomas. Signal Transduct Target Therapy. 2020;5(1).

  52. Xiao L, Li X, Mu Z, Zhou J, Zhou P, Xie C, et al. FTO Inhibition enhances the Antitumor Effect of Temozolomide by targeting MYC-miR-155/23a Cluster-MXI1 Feedback Circuit in Glioma. Cancer Res. 2020;80(18):3945–58.

    Article  CAS  PubMed  Google Scholar 

  53. Huff S, Kummetha IR, Zhang L, Wang L, Bray W, Yin J, et al. Rational design and optimization of m(6)A-RNA demethylase FTO inhibitors as Anticancer agents. J Med Chem. 2022;65(16):10920–37.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Woroniecka KI, Rhodin KE, Chongsathidkiet P, Keith KA, Fecci PE. T-cell dysfunction in Glioblastoma: applying a New Framework. Clin Cancer Res. 2018;24(16):3792–802.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  55. Lynes JP, Nwankwo AK, Sur HP, Sanchez VE, Sarpong KA, Ariyo OI et al. Biomarkers for immunotherapy for treatment of glioblastoma. J Immunother Cancer. 2020;8(1).

Download references

Acknowledgements

Not applicable.

Funding

This work was supported by the Basic and Applied Basic Research Foundation of Guangdong Province (2022A1515111118) and the Scientific and Technological Planning Project of Guangzhou City (202201010515, 2023A04J1782).

Author information

Author notes

  1. Jie Lu and Siyu Chen contributed equally to this work.

Authors and Affiliations

  1. Department of Neurosurgery/Neuro-oncology, Sun Yat-sen University Cancer Center, 651 Dongfeng Road East, Guangzhou, 510060, P. R. China

    Jie Lu, Siyu Chen, Ke Sai & Depei Li

  2. State Key Laboratory of Oncology in South China, Guangdong Provincial Clinical Research Center for Cancer, Sun Yat-Sen University Cancer Center, Guangzhou, 510060, P. R. China

    Jie Lu, Siyu Chen, Wanming Hu, Ke Sai & Depei Li

  3. Department of Pathology, Sun Yat-sen University Cancer Center, Guangzhou, P. R. China

    Wanming Hu

  4. Department of Traditional Chinese Medicine, The First Affiliated Hospital of Guangdong Pharmaceutical University, 19 Nongling Road, Guangzhou, 510060, P. R. China

    Pingping Jiang

Authors
  1. Jie Lu
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Siyu Chen
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Wanming Hu
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Ke Sai
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Depei Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Pingping Jiang
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Public data collection and analysis were performed by Pingping Jiang. Immunohistochemistry analysis was performed by Wanming Hu and Ke Sai. Cellular experiments were completed by Jie Lu and Siyu Chen. The first draft of the manuscript was written by Jie Lu and Depei Li. Ke Sai revised the manuscript.

Corresponding authors

Correspondence to Depei Li or Pingping Jiang.

Ethics declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki and approved by the Ethics Committee of Sun Yat-sen University Cancer Center. Informed consent was obtained from the patients.

Consent to publish

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

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

Electronic supplementary material

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Lu, J., Chen, S., Hu, W. et al. m6A regulator-based molecular classification and hub genes associated with immune infiltration characteristics and clinical outcomes in diffuse gliomas. BMC Med Genomics 18, 37 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12920-025-02104-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12920-025-02104-9

Keywords

BMC Medical Genomics

ISSN: 1755-8794

Contact us