GEPIA2 (Gene Expression Profiling Interactive Analysis, http://gepia2.cancer-pku.cn/#index) is a web resource based om tumor and normal samples from the TCGA and the GTEx databases, which provides differential gene expression profiling, correlation analysis, and patient survival analysis (21, 29). In this study, we utilized the “single gene analysis” module of GEPIA2 to compare mRNA expression differences between GBM and normal brain tissue. Student’s t test was used to generate p values for mRNA expression, and p < 0.05 was considered statistically significant.

UALCAN (http://ualcan.path.uab.edu/index.html) is a convenient, interactive web-portal to perform analyses of TCGA gene expression data, which provides analyses of relative expression of genes across tumor and normal samples, estimation of the effect of gene expression level and clinicopathologic features on patient survival, and identification of the top up and down regulated genes in individual cancer types (22). In this research, the expression levels of the AlkB family were obtained through the “TCGA Gene analysis” module of the website. Differences in mRNA expression were calculated by Student’s t test, and p < 0.05 was considered statistically significant.

The Human Protein Atlas website (http://www.proteinatlas.org/) is an information database of protein expression patterns in normal human tissues and in cancer, which can identify clinically useful biomarkers by applying antibodies and protein expression data as tools (30). In this study, we obtained the protein expression profiles of the AlkB family members in normal brain and GBM tissues through this database.

PrognoScan (http://dna00.bio.kyutech.ac.jp/PrognoScan/PrognoScan.html) is a database offering a tool for assessing the relationship between gene expression and tumor patient prognosis, which employs the minimum P-value approach for grouping patients so as to find the optimal cut point in continuous gene expression measurement (23). In our study, we utilized this useful database to evaluate the relationship between the AlkB family gene expression and GBM patient overall survival.

The cBioPortal (cBio Cancer Genomics Portal, http://cbioportal.org) is a free resource for interactive exploration of multidimensional cancer genomics, through which we can get access to molecular profiles and clinical attributes from large-scale cancer genomics projects (24). In this study, we acquired the genetic alteration map of the AlkB family members in GBM tissues by searching this database.

The STRING (https://cn.string-db.org/) is an open-access database providing integration of all known and predicted protein-protein interactions (PPIs), including physical interactions and functional associations, by automated text mining of the scientific literature, databases of interaction experiments, computational interaction predictions and systematic transfers of interaction evidence from one organism to another (25, 31). In our study, we utilized this database to analyze the interaction of altered proteins in tumor samples with mutation of at least one gene of AlkB family.

Cytoscape is an open software which can visually integrate biomolecular interaction networks with high-throughput expression data and other molecular states into a unified conceptual framework (32). In the study, we performed functional integration of 38 altered proteins in tumor samples with mutation of at least one gene of AlkB family.

WebGestalt (WEB-based Gene SeT AnaLysis Toolkit, http://www.webgestalt.org/option.php) is a functional enrichment analysis web tool, which supports three methods for enrichment analysis, including Over-Representation Analysis (ORA), Gene Set Enrichment Analysis (GSEA), and Network Topology-based Analysis (NTA) (26). In our study, we performed Gene ontology (GO) enrichment analysis by applying the WebGestalt website.

TIMER web server (https://cistrome.shinyapps.io/timer/) is a comprehensive resource for systematical analysis of immune infiltrates across diverse cancer types (27, 33). In this study, we utilized “gene module” to explore the correlation between AlkB family gene expression and abundance of immune infiltrates. And the “survival module” was applied to assess the association of clinical outcome with immune cell infiltration and the AlkB family gene expression. The multivariable cox proportional hazard model was used as the statistical method.

LinkedOmics (http://www.linkedomics.org/login.php) is publicly available website which includes multi-omics data from all 32 TCGA Cancer types and 10 Clinical Proteomics Tumor Analysis Consortium (CPTAC) cancer cohorts (28). In our study, we applied the “LinkFinder” module to explore the correlated significant genes of AlkB family genes, and the “LinkInterpreter” module to perform the GO pathway enrichment analysis based on the association results.

5 cases of normal brain and 20 cases of GBM formalin-fixed, paraffin-embedded (FFPE) tumor tissue blocks were submitted for immunohistochemistry (IHC). IHC was performed following standard protocols. The slides were dewaxed, boiled in 0.01 M citrate buffer (pH 6.0) for 30 minutes, and then incubated with 3% hydrogen peroxide for 15 minutes. After washing with PBS, the slides were incubated with 5% normal bovine serum albumin (BSA) for 1 hour, followed by incubation with rabbit polyclonal antibody recognizing ALKBH8 (A7142, 1: 100, ABclonal Technology, Wuhan, China) overnight. After incubating with secondary antibody, the slides were then incubated with 3,3′ diaminobenzidine (DAB) (PV-6000D, ZSGB-BIO, Beijing, China) for staining. The slides were then counterstained with hematoxylin, dehydrated, cleared, and mounted. The study was approved by the Ethics Committee of Xiangya Hospital (Ethics approval No. 202201015).

ALKBH8 expression profile difference between normal brain and GBM tissues was analyzed using Wilcoxon rank testing. P-value < 0.05 was considered to be statistically significant.

We firstly searched the GEPIA2 database to explore the mRNA expression levels of the AlkB family (ALKBH1-8 and FTO) in GBM and normal brain tissue. The results revealed that the expression levels of ALKBH1/2/3/4/5/7/8/FTO were elevated, while the expression level of ALKBH6 was reduced compared with normal brain tissues (Figure 1A). We also utilized GEPIA2 to analyze the relative expression level of the AlkB family in GBM, and the result illustrated that ALKBH5/7/FTO had relatively higher expression levels among the family members (Figure 1B). Besides the GEPIA2 database, we applied UALCAN to explore the mRNA expression levels of the AlkB family in GBM and normal brain tissue. The results showed that the expression levels of ALKBH1/2/8 were significantly up-regulated and the expression levels of ALKBH5/6/FTO were significantly down-regulated in GBM compared with normal tissues (Figure 2A). After description of the mRNA expression levels of AlkB family, we applied the Human Protein Atlas to explore the protein expression levels of AlkB family in GBM. The results revealed that ALKBH2/4/7/8/FTO were highly or mediumly expressed, while ALKBH3/6 were lowly or not detected in GBM tissues (Figure 2B). Taken together, these results indicated the potential biological roles of up-regulated ALKBH2/8 in GBM pathogenesis.

We next managed to investigate whether the mRNA expression levels of AlkB family genes were associated with GBM patients’ survival. After searching the PrognoScan database, we found intriguingly that patients with higher transcription levels of ALKBH2 and ALKBH8 possessed significant shorter overall survival (OS) time based on the GEO GSE4271 dataset (34) (Figure 3). Conversely, patients with higher transcription level of FTO displayed significant longer OS time.

To furtherly explore the biological function of AlkB Family in GBM, we first evaluated the genetic alterations of the AlkB family genes by searching the cBioPortal database. The results presented that the genetic alteration rates of ALKBH1-8/FTO in GBM tumor samples were 2.9%, 1.9%, 1.7%, 10%, 1.5%, 2.4%, 2.2%, 1% and 3%, respectively (Figure 4A). We also found that “mRNA high” was the most common alteration mode for most of the AlkB family members. After confirming the genetic alteration profiling of AlkB family in GBM, we explored the altered proteins and genes in GBM samples with mutation of at least one gene of AlkB family by utilizing the cBioPortal database, and identified 38 proteins (Supplementary Table 1) and 183 genes (Supplementary Table 2) most related to AlkB family genetic alterations. Then we conducted protein interaction analysis by applying the STRING website and Cytoscape software to depict the PPIs among these 38 altered proteins based on the corresponding nodes and combined scores (Supplementary Table 3). We found that a variety of proteins displayed the critical functions determining cell fate. Specifically, a series of DNA damage repair related proteins, including ATM, CHEK2 and MSH2, were closely associated with the functions of AlkB family in GBM. Moreover, we also identified other proteins playing pivotal roles in controlling metabolism and immunity (Figure 4B). After analyzing the altered proteins, we performed GO annotation and pathway enrichment analysis by applying the WebGestalt website based on the 183 altered genes. The GO annotation results suggested that the altered genes were mainly associated with biological regulation, response to stimulus and metabolic process. Regarding cellular components, the altered genes were mostly enriched in membrane and endomembrane system. In terms of molecular function categories, the altered genes were mostly related to protein binding, ion binding and nucleic acid binding (Figure 5A). The corresponding enriched GO pathways included leukocyte migration, extracellular structure organization and T cell activation (Figure 5B).

On account of the GO pathway enrichment results, we became interested in the correlation between AlkB family gene expression and abundance of immune infiltrates in GBM tissues. By searching the TIMER web server, we found that ALKBH1 expression level was positively correlated with CD8+ T cells, macrophages and neutrophils infiltration (Figure 6A). Regarding ALKBH2, its expression level was positively correlated with B cells and CD8+ T cells, and negatively correlated with CD4+ T cells, neutrophils and dendritic cells infiltration (Figure 6B). For ALKBH3, the only significant immune cells were macrophages, which was negatively correlated with the gene expression level (Figure 6C). Regarding ALKBH4, its expression level was positively correlated with CD8+ T cells, CD4+ T cells, macrophages and neutrophils (Figure 6D). For ALKBH5, its expression level was positively correlated with dendritic cells, and negatively correlated with B cells and CD8+ T cells infiltration (Figure 6E). Regarding ALKBH6, its expression level was not significantly tied to immune cell infiltration (Figure 6F). For ALKBH7, its expression level was positively correlated with B cells and CD8+ T cells, and negatively correlated with CD4+ T cells and dendritic cells infiltration (Figure 6G). Regarding ALKBH8, its expression level was not significantly related to immune cell infiltration (Figure 6H). For FTO, its expression level was positively correlated with CD8+ T cells, CD4+ T cells, macrophages and neutrophils, which suggested active anti-tumor microenvironment and might explain patients with higher transcription level of FTO possessed significant longer OS time (Figure 6I). After analyzing the correlation between AlkB family gene expression and immune infiltrates in GBM tissues, we furtherly assessed the association of clinical outcome with immune cell infiltration and the AlkB family gene expression by applying the multivariable Cox proportional hazard model. The results revealed that CD4+ T cells, dendritic cells and ALKBH4/6/8 were significantly associated with patient survival (Table 2).

After identifying ALKBH2/8/FTO as significant prognostic biomarkers based on the mRNA expression levels, we proceeded to explore the correlated significant genes of ALKBH2/8, and to perform the GO pathway enrichment analysis based on the association results by taking advantage of the LinkedOmics website. The positively and negatively correlated significant genes were exhibited in Figures 7A, B. We noticed that the correlated genes consisted of DNA damage repair related genes such as NTHL1, PCNA and ERCC8, and cell metabolism related genes such as MRPS26, EIF5A and EEF1G, which was consistent with the previous results. Concerning the GO pathway enrichment results, we found that mRNA processing was positively correlated with the expression level of both ALKBH2 and ALKBH8, and immune response was negatively correlated with ALKBH2/8 expression level, which might partly explain the worse survival of GBM patients with higher transcription levels of ALKBH2/8 (Figures 8A, B). The high expression of ALKBH8 in GBM was verified by IHC.

After analyzing the online databases, we performed the ALKBH8 staining in 5 normal brain tissues and 20 GBM tissues. The IHC results illustrated that ALKBH8 was highly expressed in GBM tissue compared with normal brain, which is consistent with the results obtained from the databases (Figures 9A, B).