Expression and prognostic characteristics of m5C regulators in low‐grade glioma

Abstract Glioma is the most common intracranial malignant tumour. A clear diagnosis and molecular targeted therapy are of great significance for improving the survival time and quality of life of patients with low‐grade glioma. 5‐methylcytosine methylation is one of the ways of RNA modification, but there are limited studies on the role of m5C methylation of low‐grade glioma. Single‐nucleotide variant, RNA expression matrix and corresponding clinical data of low‐grade glioma came from public database. The single‐nucleotide variant and expression of m5C regulators were estimated. A prognostic model based on m5C regulators was constructed by Cox regression. Potential functions of these molecules were assessed by gene set enrichment analysis. DNMT3A mutation was the most frequent among the m5C regulators in low‐grade glioma. NSUN3, TET2, TRDMT1, ALYREF, DNMT3B, DNMT1, NOP2 and NSUN2 were up‐regulated. One prognostic model was constructed which had a strong predictive power for the overall survival of low‐grade glioma. We studied the expression and prognostic characteristics of m5C regulators in low‐grade glioma, supplied biomarkers for the diagnosis and prognosis and provided the foundation for the study of the pathogenesis of low‐grade glioma.


C methylation (ALYREF)
is involved in the stabilization and recognition process of m 5 C methylation. 11 TET2 is the 'eraser' of m 5 C methylation. m 5 C regulates varieties process such as RNA output, RNA stability and translation process. 12 However, there are limited studies on the role of m 5 C methylation of low-grade glioma.
This study intends to analyse the expression and prognostic characteristics of m 5 C regulators in low-grade glioma based on high-throughput sequencing databases, identify biomarkers and construct a prognostic model, and provide a target for the research and treatment of low-grade glioma.
Meanwhile, the clinical data of low-grade glioma in TCGA were extracted form cBioportal (http://www.cbiop ortal.org/) website, including basic information basic information (gender, age, etc), survival data (survival time and survival status) and molecular status (IDH, 1p/19q, TERT mutation, ATRX mutation and MGMT methylation, etc) Genes were annotated by the Ensemble human genome browser (http://asia.ensem bl.org/index.html). All data were corrected using 'limma' package of R language. Besides, 'sva' package was used to merge data of CGGA datasets (mRNAseq_693 and mRNAseq_325).

| Construction of a prognostic model based on m 5 C regulators
The TCGA dataset was used as the training cohort to construct a prognostic model, and the CGGA dataset was used as the validation cohort in this study. First, univariate Cox regression was used to identify the m 5 C regulators related to the prognosis of low-grade glioma.
Furthermore, multivariate Cox regression was used to construct a prognostic model and a nomogram of low-grade glioma based on m 5 C. The risk score was calculated by ∑β gene n × Expression gene n .
Kaplan-Meier survival curve was introduced to compare the OS (overall survival) of different risk groups divided by the median risk score. In addition, receiver operating curve (ROC) of 1, 3 and 5 years were used to test the predictive ability of the model for low-grade glioma.

| Gene set enrichment analysis (GSEA)
We estimated the Pearson correlation coefficients between genes with m 5 C regulators to find m 5 C regulator-related factors. The genes with correlation coefficient > 0.4 (or < −0.4) and P < .05 were defined as m 5 C regulator-related factors, and GSEA was performed based on these factors by 'clusterProfiler' package of R language.

| Statistical analysis
All statistics and graphics in this study were implemented based on R language (4.0.0). We used the Wilcoxon rank-sum test rather than the t test to determine whether difference in two groups was statistically significant in this study, because the Wilcoxon test minimizes the influence of outliers compared with t test. Log-rank test was used to compare different risk groups in the Kaplan-Meier curve. It was considered statistically different when P < .05.

| Mutation characteristics of m 5 C regulators in low-grade glioma
The 'maftools' package of R language was used to analyse the sin-

| Differentially expressed m 5 C regulators
The basic characteristics of the patients of TCGA and CGGA datasets were shown in Table S1. We analysed the differentially expressed m 5 C regulators of the TCGA dataset and drew a heatmap to display the expression of m 5 C regulators in each low-grade glioma sample, as shown in Figure 2. Eight differentially expressed m 5 C regulators (NSUN3, TET2, TRDMT1, ALYREF, DNMT3B, DNMT1, NOP2 and NSUN2) were identified. Interestingly, compared with normal brain tissue, these differentially expressed molecules were all upregulated in low-grade glioma. The expressions of all m 5 C regulators between normal and tumour samples were shown in Figure 3.
Since different IDH and 1p/19q co-deletion status has an impact on the prognosis of low-grade glioma patients (survival outcomes F I G U R E 1 Summary of low-grade glioma mutation profiles. A, Missense mutation, SNP and C > T were more common in low-grade glioma mutation. B, Overview of mutation profiles of m 5 C regulators (DNMT3A, TET2, DNMT3B, DNMT1 and NSUN5) in each sample in the IDH wild-type group were worse compared with IDH mutant-type group 13 and survival outcomes in 1p/19q non-co-deletion group were worse compared with 1p/19q co-deletion group 14,15 ), we evaluated the expression of these 8 differentially expressed genes in different IDH and 1p/19q co-deletion status subgroups. The expression difference of TRDMT1 in each subtype was the most significant. TRDMT1 expression was higher in IDH wild-type without 1p/19q co-deletion group, while IDH wild-type with 1p/19q co-deletion group and IDH mutant group had lower TRDMT1 expression.
Only ALYEF had no significant difference in expression in IDH and 1p/19q co-deletion status subgroups. The detailed expression status of 8 differentially expressed molecules in each group was shown in  Figure S1). In addition, gene expression levels of NSUN3 and TET2 were diverse in different MGMT promoter status ( Figure S2). Furthermore, gene expression levels of NOP2, NSUN2, TRDMT1 and TET2 were diverse in different ATRX status ( Figure   S3).

| Construction of a prognostic model based on m 5 C regulators
At first, univariate Cox regression was applied to find prognosis-related m 5 Table 1. The risk score and survival status of each sample were shown in Figure 5A. Meanwhile, Figure 5B presented the nomogram of the prognostic model for low-grade glioma.
In order to verify the prognostic ability of the model for the overall survival of low-grade glioma, Kaplan-Meier curve and ROC curve were constructed, as shown in Figure 6. In training cohort, patients in low-risk group had better prognosis compared with those in high-risk group. (Figure 6A

| GSEA
In order to explore the possible functions of m 5 C regulators of low-grade glioma, we performed GSEA on the m 5 C F I G U R E 2 Heatmap of m 5 C regulators in low-grade glioma. NSUN3, TET2, TRDMT1, ALYREF, DNMT3B, DNMT1, NOP2 and NSUN2 were up-regulated in low-grade glioma F I G U R E 3 Expression of m 5 C regulators between low-grade glioma and normal brain tissue

| D ISCUSS I ON
This study analysed the mutation and expression characteristics of This study still has some limitations. First, results obtained based on sequencing datasets still need to be examined by basic experiments. Second, more potential m 5 C regulators have yet to be discovered.
We studied the expression and prognostic characteristics of m 5 C regulators in low-grade glioma, supplied biomarkers for the diagnosis and prognosis and provided the foundation for the study of the pathogenesis of low-grade glioma.

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C A L A PPROVA L
No more ethical approval was needed since all data of this study were acquired from public database.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data were available from corresponding author on reasonable request.