Gene expression data and complete clinical annotations were obtained from the CGGA (cgga.org.cn/) and TCGA (https://portal.gdc.cancer.gov/) databases. Patient data included the age, sex, radiotherapy and chemotherapy statuses of patients, complete follow-up information, histopathological classification and primary/recurrent status of World Health Organization (WHO) malignant gliomas. In the present study, the mRNAseq_693 and mRNAseq_325 glioma cohorts were analyzed. Tumor-immune system interactions and drug bank database (https://www.drugbank.ca/) analysis was used to determine the correlation between target genes and lymphocytes, immune regulators and immune checkpoints in gliomas.

The expression levels of FBXL6 in gliomas were analyzed using data from the TCGA database. The ‘limma’ package in the R platform was used to analyze the differential expression of FBXL6 between normal and tumor tissues in gliomas. Low-grade gliomas typically refer to WHO grade I or II tumors, which tend to grow slowly, have distinct borders and invade surrounding tissues to a lesser extent. Patients with low-grade gliomas generally have a better prognosis (25). Conversely, high-grade gliomas usually refer to WHO grade III or IV tumors and are characterized by rapid growth, indistinct borders and high invasiveness. Patients with high-grade gliomas typically have a poorer prognosis (26). Visualization analysis was performed using the ‘ggplot2’ and ‘ggpubr’ packages in R. Additionally, the prognostic value of FBXL6 in gliomas was analyzed using the TCGA and CGGA databases.

The CGGA database provided detailed clinical data, including patient age, sex, radiotherapy and chemotherapy status, complete follow-up information, histological classification and primary/recurrent status of WHO malignant gliomas. TCGA clinical data included sex, age and tumor grade. Therefore, a detailed analysis of the correlation between FBXL6 expression and clinical features using was performed using bioinformatic tools. Receiver operating characteristic (ROC) curves were used to determine the accuracy of FBXL6 expression in predicting 1-, 3- and 5-year survival rates in patients with glioma. To construct a nomogram using both FBXL6 expression data and clinical data, the ‘rms,’ ‘survival,’ and ‘regplot’ packages in R were used to assess the impact of age, sex and tumor grade on patients with glioma. Clinical decision curve analysis was used to evaluate the accuracy of the 1-, 3- and 5-year survival rate predictions in patients with glioma. Additionally, the reliability of the nomogram was assessed using calibration curves.

To investigate the specific mechanisms underlying FBXL6 expression in gliomas, gene correlation and enrichment analyses were performed. For the correlation analysis, the filtering criteria were R≥0.6 and P<0.001. To explore the biological processes and pathways related to the correlated genes in gliomas, enrichment analyses were performed using the Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Set Enrichment Analysis (GSEA) databases.

The role of FBXL6 expression in immune cell infiltration was investigating by using the CiberSort algorithm to study the distribution of 22 tumor-infiltrating immune cells in high and low FBXL6 expression groups. Furthermore, Spearman correlation analysis was conducted to explore the strength of the association between FBXL6 expression and the 22 types of immune-infiltrating cells. Immune and stromal scores were calculated using estimation methods to reflect the relationship between FBXL6 expression and the immune microenvironment. TIMER2 (http://timer.cistrome.org/) analysis FBXL6 expression in different correlation between immune cells and glioma subtypes.

With the advancement of precision medicine, the demand for personalized treatment has increased. The sensitivity of gene expression and drugs is a critical factor in personalized treatment (27). Robust prediction of in vivo chemotherapy responses by collecting pretreatment baseline gene expression levels and drug sensitivity data from cancer cell lines has been a long-standing and controversial issue in pharmacogenomics. Cancer cell lines in labs may not accurately reflect patient tumor complexity, leading to possible mismatches between predicted and real responses to chemotherapy. Patient-specific genetic variations and tumor environments also affect treatment outcomes, challenging the applicability of cell line data to predict patient responses effectively. Drug response information and drug-targeting pathways were collected from the Genomics of Drug Sensitivity in Cancer database (https://www.cancerrxgene.org/). Spearman correlation analysis was performed to identify drugs related to the risk score.

The U251 cell (cat. no. TCHu 58) line was obtained from the China Center for Typical Culture Collection. LN229 (cat. no. iCell-h124) and HEB (cat. no. XY-XB-1640) cell lines were purchased from Shanghai Xuan Ya Biotechnology Co., Ltd. U251 and LN229 are high-grade glioma cell lines with overexpression of the TP53 and EGFR genes, which serve crucial roles in the pathogenesis and progression of gliomas. Additionally, U251 cells also exhibit specific gene expressions such as neurofibromin, cyclin-dependent kinase inhibitor 2A and phosphatidylinositol 3,4,5-triphosphate 3-phosphatase and dual-specificity protein phosphatase, which are associated with the pathogenesis of gliomas. HEB is a normal human brain glial cell line that exhibits a polygonal shape when adhered to the matrix, with a more uniform size and a patchy growth pattern (28). Compared to HEB cells, U251 cells have a higher apoptosis rate and a faster proliferation rate. Cells were cultured in DMEM supplemented with 10% FBS (Gibco; Thermo Fisher, Inc.), 100 units/ml penicillin and 100 mg/ml streptomycin (Gibco; Thermo Fisher, Inc.). To prepare cell samples for western blot, cells were lysed with RIPA lysis buffer (Beyotime Institute of Biotechnology). The mass of protein/lane is 150 µg. Proteins were separated by SDS-PAGE using a XCell SureLock Mini-Cell (Invitrogen; Thermo Fisher Scientific, Inc.) and transferred to a PVDF membrane (Invitrogen; Thermo Fisher, Inc.). The concentration of the separating gel was between 6% and 8%. Membranes were blocked in 5% non-fat milk in PBS (Invitrogen; Thermo Fisher, Inc.) at room temperature for 1 h before incubation with primary antibodies overnight at 4°C. The primary antibodies used were anti-FBXL6 antibodies (cat. no. PA5-64927; 1:200; Thermo Fisher, Inc). The membranes were then incubated with the appropriate HRP-conjugated secondary antibodies (cat. no. C510051; 1:5,000; Shanghai Shenggong Biology Engineering Technology Service, Ltd.) at 37°C on a shaker for 2 h. Anti-GAPDH antibodies (cat. no. b181602; 1:2,000; Abcam) were used for the reference protein. Immunoreactive bands (cat. no. WBKLS0500; Millipore Immobilon Western, ECL; EMD Millipore, Billerica, MA, USA) were quantified using ImageJ software (v.1.48; National Institutes of Health, Bethesda, MD, USA).

RNA was extracted from glioma cells using TRIzol^® reagent (Invitrogen; Thermo Fisher, Inc.) and quantified using a 725 spectrophotometer (Shanghai Sunny Hengping Scientific Instrument Co., Ltd.). Oligo dT (Roche Applied Science, 10814270001) was used to prime cDNA synthesis. RT-qPCR was performed using the SYBR Green Premix Ex Taq (Takara Biotechnology Co., Ltd) on an Illumina Eco (Illumina, Inc.). Total RNA was extracted from tissue and cells using Tripure Isolation reagent (Roche Applied Science; Penzberg; Germany), according to the manufacturer's instructions. The Transcription First Strand cDNA Synthesis kit (Roche Applied Science) was used to synthesize cDNA from 1 µg total RNA at room temperature at 24°C for 50 min. Differences in gene expression were calculated using the 2^−ΔΔCq method (29). The thermocycling conditions were as follows: Initial denaturation at 94°C for 5 min, followed by 30 cycles at 94°C for 45 sec, 59°C for 45 sec and 72°C for 45 sec, followed by 72°C for 45 sec and final extension at 72°C for 10 min. Experiments were performed in triplicate with SYBR Green I Master mix (Roche Applied Science); GAPDH was used as the internal control. The primers used were as follows: FBXL6 forward (F), 5′-CATCAACCGTAATAGCATTCCCC-3′ and reverse (R), 5′-CACATCAGGTTCAACAGCCG-3′ and GAPDH F, 5′-GGAGCGAGATCCCTCCAAAAT-3′ and R, 5′-GGCTGTTGTCATACTTCTCATGG-3′.

In the present study, paraffin-embedded glioma specimens and matched adjacent non-tumor tissues were surgically excised from patients at the Department of Neurosurgery and forwarded to the Department of Pathology for the pathological diagnosis of glioma. Sample collection took place from January 2020 to June 2023. Upon confirmation of glioma diagnosis, paraffin-embedded specimens were stored within the Department of Pathology. This was a retrospective study and informed consent for this study was waived because the patients had previously provided written informed consent for the use of their post-operative samples and information in future clinical research. The study protocol, including the use of these tissue samples, was approved by the Ethics Committee of the Ninth People's Hospital affiliated with Shanghai Jiao Tong University School of Medicine (approval no. H9H-2023-T489-1; Shanghai, China). Clinical samples were fixed in 4% paraformaldehyde at room temperature for 24 h, subsequently embedded in paraffin, and sectioned to a uniform thickness of 3 micrometers. Immunostaining was performed using the two-step Elivision Plus kit system (Dako; Agilent Technologies, Inc.). The sections were dewaxed in xylene, rehydrated with a series of ethanol solutions (100, 95, 80 and 70%), and then boiled in ethylenediaminetetraacetic acid buffer (pH 9.0) for 20 min in an autoclave. Next, 0.3% H₂O₂ was used to block endogenous peroxidase activity at room temperature for 15 min, and the sections were incubated with normal goat serum (1:20; Beyotime Institute of Biotechnology) for 20 min at room temperature to reduce non-specific binding. Tissue sections were incubated with the anti-FBXL6 (cat. no. PA5-64927; 1:200; Thermo Fisher) for 50 min at room temperature. The secondary antibody was applied using the Envision Detection kit (SM802; Dako; Agilent Technologies; Ready-to-use type) for 20 min at room temperature. Slides were stained for 2 min with diaminobenzidine tetrahydrochloride (DAB) and then counterstained 2 min with hematoxylin at room temperature. The completed sections were mounted using a neutral resin (K-0212, Shanghai Jiehao Biotechnology Co., Ltd.). The stained tumor cells were assessed with a Nikon conventional optical microscope in 10 independent fields at magnification, ×400. Immunohistochemical image results were quantified using ImageJ software (v.1.48; National Institutes of Health, Bethesda, MD, USA). All the sections were scored by two independent pathologists who were blinded to the type of sample. Specifically, the intensity of positive cytoplasmic staining was rated on a scale of 0–3 (0, negative; 1, light brown; 2, medium brown; and 3, dark brown). The corresponding percentages of positively stained cells were set as: 1, 1–25%; 2, 26–50%; 3, 51–75%; and 4, 76–100%.

All statistical analyses were performed using R (version 4.1.2; RStudio, Inc.). The results are presented as the median ± interquartile range or mean ± standard deviation (SD). Western blot data were analyzed using GraphPad Prism (version 5; Dotmatics). One way ANOVA was employed to compare the means ± SD between two groups followed by the Least Significant Difference posts hoc test. The Chi-square test and Fisher's exact test were utilized for analyzing qualitative data. For comparisons between two groups, the Mann-Whitney U test (non-parametric) was used, while the Kruskal-Wallis test followed by Dunn's test was applied for comparisons involving >2 groups. The log-rank test and Kaplan-Meier survival analysis were used to compare prognosis among the different risk groups. Univariate and multivariate Cox regression analyses were performed to evaluate the prognostic value of risk models. The significance of the correlation was assessed using the Spearman's rank correlation test. All experiments were validated with three repeated trials. P<0.05 was considered to indicate a statistically significant difference.

The data collection and analysis performed in the present study (Fig. 1) identified key biomarkers in glioma. Multicenter screening and validation were performed and FBXL6 was identified as a key marker. Compared with normal tissues, pan-cancer analysis demonstrated that FBXL6 was highly expressed not only in GBM, but also in certain other cancers, including bladder, breast and colorectal cancers (Fig. 2A). TCGA-glioma analysis demonstrated consistently higher FBXL6 expression in tumor tissues compared with normal tissues in the TIME2 database (Fig. 2B). Kaplan-Meier curve analysis of TCGA-glioma data demonstrated that high FBXL6 expression was associated with decreased overall survival and progression-free survival compared with low FBXL6 expression, an unfavorable outcome for patients with glioma (Fig. 2C-D). Moreover, the CGGA database showed the same relationship between FBXL6 expression and prognosis (Fig. 2E). Western blot was performed to examine the FBXL6 protein levels in glioma U251 and normal HEB cells. The protein expression level of FBXL6 was higher in U251 cells compared with HEB cells (Fig. 3A). The RT-qPCR results demonstrated that FBXL6 mRNA expression levels were significantly higher in U251 compared with HEB cells (Fig. 3B). Given the crucial role of FBXL6 in cancer (22,30), its expression levels were investigated across different stages of glioma. A tissue microarray comprising samples from 64 glioma patients was constructed for this purpose. Patients were stratified into two groups based on their IHC Score (IRS): i) Low-expression group, IRS≤7; ii) and High-expression group, IRS≥8. Subsequently, the expression level of FBXL6 in the glioma tissue microarray was assessed via IHC. Compared with low-grade glioma, the expression of FBXL6 protein increases in high-grade gliomas. Furthermore, the protein expression level of FBXL6 was increased in patients with glioma at WHO Grade IV compared with patients at WHO Grade II stage (Fig. 3C,D; Table I). These findings suggested that the high expression of FBXL6 in glioma may affect its biological function.

CGGA, univariate (Fig. 4A) and multivariate (Fig. 4B) COX regression analyses demonstrated that FBXL6 expression was associated with poor survival and was considered an independent prognostic factor in each dataset. FBXL6 was also associated with IDH mutations, Primary, Recurrent and Secondary status, WHO grade, chemotherapy regime and 1p19q co-deletion. Therefore, FBXL6 may have a high predictive value in patients with gliomas and could potentially be used as a prognostic factor for gliomas. In addition, the expression level of FBXL6 was significantly higher in patients with glioma undergoing chemotherapy compared with those not undergoing chemotherapy (Fig. 4C). According to the WHO classification, the higher the glioma grade, the higher the expression level of FBXL6 (Fig. 4D). In various subtypes of pathological tissues, the expression of FBXL6 was correlated with multiple subtypes of glioma and compared with other subtypes, FBXL6 had the highest expression level in glioblastoma (Fig. 4E). There was no statistical significance observed between FBXL6 expression levels and the IDH1 mutation or 1p19q codeletion status (Fig. S1A-B). These results suggested that FBXL6 expression promoted the occurrence and development of glioblastoma and could potentially serve as a potential molecular marker.

Given the significant correlation between FBXL6 expression and clinical factors such as tumor grade and age (Fig. 5A), the survival rates of patients with glioblastoma over 1-, 3- and 5-year periods were predicted using FBXL6 expression levels. The ROC curve demonstrated that FBXL6 expression was strongly predictive of the survival rates of patients with glioblastoma, with AUC values of 0.573, 0.639 and 0.631 for the 1-, 3- and 5-year survival rates, respectively (Fig. 5B). A nomogram was constructed to predict the correlation between FBXL6 expression and age, sex and tumor grade, which demonstrated a significant clinical predictive value for FBXL6 expression according to age and grade. Confidence in the validity of the nomogram was further reinforced by clinical calibration curve data (Fig. 5C,D).

To explore the specific biological functions of FBXL6 in glioblastoma, gene correlation analysis was performed with a correlation threshold of 0.6 and P<0.001. The results demonstrated that FBXL6 expression was positively correlated with mitochondrial RHO GTPase 2, aarF domain-containing protein kinase 5, uridine-cytidine kinase-like 1, AP-4 complex accessory subunit Tepsin, zinc finger protein GLI4 and signal peptide peptidase-like 2B and significantly negatively correlated with solute carrier family 2, facilitated glucose transporter member 12, growth hormone-inducible transmembrane protein, phosphatidylethanolamine-binding protein 4 and TLC domain-containing protein 4 (Fig. 6A). The general control of nucleotide synthesis 5 (GCN5) is an enzyme that serves a critical role in the modification of histones, influencing gene expression by adding acetyl groups to the histone proteins, which impacts the structure and function of chromatin. Gene correlation analysis demonstrated a significant correlation between FBXL6 and GCN5, which was consistent with previous research (Fig. S2). Furthermore, the expression of FBXL6 was positively correlated with IDH1 and negatively correlated with MGMT. However, the expression of GCN5 demonstrated no significant correlation with IDH1 and was negatively correlated with MGMT (Fig. S3). Gene Expression Profiling Interactive Analysis 2 database (gepia2.cancer-pku.cn/#index) investigates the relationship between the expression of FBXL6 and GCN5 and their correlation with isocitrate dehydrogenase 1 (IDH1) and methylated DNA-protein-cysteine methyltransferase (MGMT). These findings indicated that FBXL6 and these correlated genes may influence the progression of glioblastoma.

Gene correlation analysis identified FBXL6 as positively associated with SPPL2B, GLI4, and ADCK5, and negatively with TLCD4, PEBP4, and GHITM (Fig. 6A,B). The GO analysis revealed significant connections to biological processes like ‘organelle image’, ‘nuclear division’, and ‘chromosome aggregation’, underscoring FBXL6′s involvement in key cellular activities and structures. In terms of molecular function, FBXL6 expression was associated with ‘channel activity’, ‘passive transmembrane transport’ and ‘transporter activity’ (Fig. 6C). GSEA enrichment analysis demonstrated that the calcium signaling pathway, cell adhesion molecule cascades and long-term potentiation were differentially enriched in patients with high FBXL6 expression (Fig. 6D).

FBXL6 expression was correlated with immunomodulators in gliomas including lymphocytes, immunoinhibitory and immunostimulatory molecules and major histocompatibility complex (MHC) molecules. The association between FBXL6 expression and the expression of immunomodulatory genes in gliomas was also examined. In lymphocytes, FBXL6 expression was positively correlated with CD56dim, whereas it showed significant negative correlations with effector memory CD4 T cells and interstitial dendritic cells (Fig. 7A). Furthermore, FBXL6 expression was significantly positively correlated with the expression of immune-inhibiting genes, adenosine receptor 2a and lymphocyte activation gene 3 protein, whereas it was negatively correlated with TGF-β receptor 1 expression (Fig. 7B). Considering immunostimulatory factors, FBXL6 expression in glioblastomas demonstrated a positive correlation with tumor Necrosis Factor Receptor Superfamily Member 25), whereas it showed a negative correlation with IL-6 receptor expression (Fig. 7C). However, for MHC molecules, the expression of FBXL6 demonstrates a significant positive correlation with TAPBP (TAP Binding Protein; Fig. 7D). The analysis of the TIMER2 database indicated that in GBM and low-grade glioma subtypes, the expression of FBXL6 was negatively correlated with CD8 T cells and positively correlated with B cells (Fig. S4). These results suggested that FBXL6 may contribute to the malignant progression of glioblastoma through its multifaceted involvement in the tumor immune interaction system.

In addition, the relevance of FBXL6 expression in the tumor immune microenvironment was assessed by estimating the stromal and immune scores of the high- and low-expression groups. Low expression of FBXL6 was significantly associated with lower stromal and immune cores (Fig. 8A), which may contribute to tumor immune escape and suppression, thereby promoting glioblastoma progression (30). Similarly, in immune-infiltrating cells, low expression of FBXL6 was significantly correlated with CD4+ memory cells and monocytes, which was consistent with our previous results and may contribute to immune evasion and promote malignant progression in patients with glioblastoma (30). Additionally, a high expression level of FBXL6 was significantly associated with M0 macrophages (Fig. 8B), whereas FBXL6 expression predominantly correlated with memory resting CD4 T cells, memory activated CD4 T cells and monocytes (Fig. 8B). The association between immune cells and FBXL6 expression using CIBERSORT was further explored and it was demonstrated that M2 and M0 macrophages were significantly positively correlated with FBXL6 expression and significantly negatively correlated with monocytes and memory resting CD4 T cells (Fig. 8C-F).

With the continued advancement of precision medicine and increasing demand for personalized treatment, the relationship between FBXL6 expression and the IC₅₀ of certain drugs in glioma treatments was investigated to elucidate their possible application in the individualized treatment of glioma. The R package ‘oncoPredict’ was used to predict the relationship between FBXL6 expression and drugs with a screening condition of P<0.001. Statistically significant differences in the sensitivity of eight anti-cancer drugs between the high and low groups of FBXL6 expression were found. A total of nine of these drugs (BMS345541, gefitinib, axitinib, bexarotene, foretinib, BHG712, BX-795, BX-912 and phenformin) had a higher IC50 in the low FBXL6 expression group compared with the high FBXL6 expression group, which demonstrated that the patients with high levels of FBXL6 expression may be more sensitive to treatment with these particular anticancer drugs (Fig. 9). These results suggested that these drugs may serve a potential role in the future treatment of gliomas with high FBXL6 expression.