GSE124814 and GSE85217 datasets were retrieved from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). The platform for GSE124814 included the transcriptome sequencing data of 1350 MB samples and 291 normal cerebellum samples. The platform for GSE85217 included 763 MB samples of transcriptome sequencing data and matched clinical information. Box plot and scatter plot were used to compare the expression levels of CENPE in MB samples and normal cerebellum tissues.

The median expression of CENPE was selected to divided the patients into CENEP^high expression group and CENPE^low expression group. The R software package survminer was used to analyze the correlation between CENPE expression and the overall survival of patients with MB. The relationship between CENPE expression and prognosis of MB patients was analyzed by Spearman and logistic regression analysis, and visualized by R package ggplot2.

We used the R package rms to integrate data on age, gender, tumor subtypes, and survival time. Using Cox proportional hazard regression analysis established a nomogram to evaluate the prognostic significance of these features in 599 samples. We evaluated the diagnostic and predictive capabilities of CENPE by drawing Receiver operating characteristic (ROC) curves. The area under the curve (AUC) was calculated, AUC>0.7 indicates satisfactory predictive ability.

We selected the TIMER method in the R package IOBR to evaluate the infiltration level of immune cells in non-WNT/non-SHH MB sample in dataset GSE85217. Spearman correlation analysis was used to analyze the correlation between CENPE expression and the infiltration of immune cells, which was visualized by R package ggplot2 (4.1.3).

We analyzed the correlation between CENPE and all genes in the GSE85217 dataset through the R package tidyverse, and after ranking according to correlation, the 100 genes with the strongest positive correlation and the 100 genes with the strongest negative correlation with CENPE were selected. These 200 genes were subjected to Gene Ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) analysis using the R package clusterProfiler. Gene set enrichment analysis (GSEA) was performed using the GSEA software (version 3.0) derived from GSEA (http://software.broadinstitute.org/gsea/index.jsp) to investigate the differences in biological functions and KEGG pathways between the CENEP^high and CENPE^low non-WNT/non-SHH MB groups (cutoff value was 50%). To evaluate relevant pathways and molecular mechanisms, based on gene expression profiling and phenotype grouping, with a minimum gene set of 5 and a maximum gene set of 5000, with 1000 resampling, P < 0.05 and FDR < 0.25 were considered statistically significant.

The protein-protein interaction (PPI) network of CENPE co-expressed genes in non-WNT/non-SHH MB was analyzed using the search tool (STRING database, https://string-db.org/, version 11.0b) for the retrieval of interacting genes. The Betweenness score of each node analyzed by Cytoscape 3.7.2, and the top 50 with the highest score were screened to draw the skeleton network and correlation heat map.

The non-WNT/non-SHH MB cell lines D283 and D341 were purchased from ATCC and cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Waltham, MA, USA) containing 10% fetal bovine serum (FBS) (Gibco, Waltham, MA, USA) at 37°C (16). The cells were grown to ~50% confluency in six-well plates before transfection. We transfected the small interfering RNA (siRNA) sequences with Lipofectamine 3000 (Invitrogen, San Diego, CA, USA) according to the manufacturer’s instructions. The siRNA sequences of CENPE in D283 and D341 cells were used as follows: siRNA-CENPE (5′-GGCUGUAAUAUAAAUCGAA-3′), and siRNA negative control (NC) (5′-TTCTCCGAACGTCACGT-3′).

D283 and D341 cells were lysed with protease inhibitor, and RIPA cleavage buffer (Thermo Fisher Scientific, Shanghai, China). The protein lysates were resolved via sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and imprinted on polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA, USA) for analysis. Anti-CENEP (1:1,000 dilution, DF7745; Affinity Biosciences, OH, USA), anti-P53 (1:500 dilution, AF0879; Affinity Biosciences, OH, USA), anti-P21 (1:500 dilution, AF6290; Affinity Biosciences, OH, USA), anti-CDK1 (1:500 dilution, DF6024; Affinity Biosciences, OH, USA) and anti-β-Actin (1:5,000 dilution, AF7018; Affinity Biosciences, OH, USA) were incubated overnight. Goat Anti-Rabbit IgG (H+L) HRP-conjugated secondary antibodies (1:5,000 dilution; S0001, Affinity Biosciences, OH, USA) were added at room temperature for 2 h. The membrane was detected with BeyoECL Plus developer (Beyotime, Shanghai, China).

D283 and D341 cells were cultured in six-well plates with 2 × 10⁵ cells for 24 h and were transfected with CENPE siRNA for 48 h. The cells were collected, and then fixed with 70% ethanol at 4°C overnight. The cells were stained with RNase A-containing PI buffer (C1052; Beyotime, Shanghai, China) at 37°C for 30 min. The cell cycle distribution was detected by flow cytometry (Beckman Coulter Quanta SC System).

D283 and D341 cells transfected with CENPE siRNA were cultured in a 96-well plate at 1 × 10³/well and allowed to adhere overnight. 10 μL CCK-8 solution was added to each well and incubated for 1.5 h. A microplate reader (Thermo Fisher Scientific, Shanghai, China) was used to detected the absorbance at 450 nm every day.

D283 and D341 cells transfected with CENPE siRNA were fixed with 4% paraformaldehyde for 15 min at room temperature, and permeabilized with 0.2% TritonX-100 for 15 min. After blocking in 10% goat serum for 30 min, the cells were incubated with anti-ki67 (1:200 dilution, AF0198; Affinity Biosciences, OH, USA) antibody at 4°C overnight. Then the cells were incubated with AlexaFluor 488 (goat anti-rabbit IgG, Abcam, Cambridge, UK, 1:1,000 dilution) at room temperature for 40 min. At last, the cells were stained with Dapi (1 μg/ml) for 10 min in dark place at room temperature and viewed with an inverted IX71 microscope system (Olympus, Tokyo, Japan). The mean intensity was measured by ImageJ software.

R (4.1.3) software and GraphPad Prism (8.0.0) software were used for statistical analyses, the main R packages used in the study were as follows: “survminer”, “tidyverse”, “IOBR”, “clusterProfiler”, “rms” and “ggplot2”. All in experiments were repeated three times. A t-test was used in comparing the means of two groups, p < 0.05 means statistical significance.

Our results showed that the expression of CENPE was significantly higher in MB than in normal cerebellum tissue ( Figure 1A ). Moreover, the expression of CENPE was also elevated in different subtypes of MB ( Figure 1B ). Kaplan-Meier survival analysis showed that the expression of CENPE in MB was not statistically significant with the survival prognosis of MB patients ( Figure 1C ). However, the expression of CENPE in the non-WNT/non-SHH subtype and SHH subtype was negatively correlated with the survival prognosis of patients. The Hazard Ratio (HR) value of CENPE in survival of non-WNT/non-SHH MB patients was higher than that in SHH subtype ( Figures 1D–F ).

To analyze the diagnostic value of CENPE expression in MB, we performed ROC curve and nomogram analysis on the CENPE expression data of GSE85217 database to evaluate the diagnostic value of the gene. The area under the ROC curve (AUC) was 0.726 ( Figure 2A ), indicating the good predictive ability. We combined the expression level of CENPE with clinical information to construct a nomogram to predict 1-, 3-, and 5-year survival in patients with MB. The nomogram showed that the subtypes of MB and the expression of CENPE were good prognostic indicators ( Figure 2B ). We further constructed a nomogram among the non-WNT/non-SHH subtypes with the worst prognosis in MB to predict 1-, 3-, and 5-year survival of patients. The results indicated that prognostic prediction of the expression level of CENPE in non-WNT/non-SHH MB was better than other traditional clinical features ( Figure 2C ). Moreover, the results of Figures 1C–F indicated that among all subtypes of MB patients, high expression of CENPE was most strongly correlated with poor prognosis in non-WNT/non-SHH MB. Therefore, we chose to further investigate the role of CENPE in the occurrence and development of non-WNT/non-SHH MB.

To explore the correlation between the expression level of CENPE and tumor immune response, we first used CIBRSORT to evaluate the differences in immune cell infiltration in non-WNT/non-SHH MB with high and low expression of CENPE. The results showed that CD4+ central memory T cells (Tcm), Class-switched memory B cells, Eosinophils, NKT, regulatory T cells (Tregs) were highly infiltrated in the low CENPE expression group. In contrast, the expression of Activated dendritic cells (aDC), B cells, CD4+ memory T cells (Tem), CD8+ Tcm, DC, conventional DC, immature DC, Macrophages, Macrophages M1, Mast cells, NK cells, pro B-cells, Tgd cells, Th1 cells, Th2 cells were highly infiltrated in the high CENPE expression group ( Figures 3A, B ). We further analyzed the correlation between the expression level of CENPE and immune infiltration in non-WNT/non-SHH MB. The results showed that CENPE expression correlated positively with the infiltration of Tergs (r=0.118, p=0.010), NKT(r=0.106, p=0.022), but negatively with Tgd cells (r=-0.149, p=0.001), pro B cells (r=-0.110, p=0.0017), Th2 cells (r=-0.186, p<0.001), CD4+ Tem (r=-0.131, p=0.004) ( Figures 3C–H ).

We then used the R package tidyverse to further explore the biological function of CENPE. 5840 genes were positively correlated with the expression of CENPE, and 5272 genes were negatively correlated with the expression of CENPE ( Figure 4A ). Heat map showed the top 50 genes positively and negatively correlated with CENPE expression ( Figures 4B, C ). The top 200 co-expressed genes of CENPE were annotated by GO and KEGG functional enrichment analysis. The KEGG analysis showed that co-expression of CENPE was mainly involved Cell cycle, Human T-cell leukemia virus 1 infection, Cellular senescence, p53 signaling pathway, Fanconi anemia pathway ( Figure 4D ). GO functional annotations showed that CENPE co-expressed genes were mainly involved in the organelle fission (GO-BP) ( Figure 4E ), chromosomal region (GO-CC) ( Figure 4F ), tubulin binding (GO-MF) ( Figure 4G ). PPI network and correlated analysis were used to identify the interactions between the top 50 proteins related to CENPE. We found that most proteins in the network have strong positive or negative correlations with each other. ( Figures 5A, B ).

To further characterize the function of CENPE, we identified 8404 differentially expressed genes (DEGs) between the CENPE^high and CENPE^low non-WNT/non-SHH MB samples, including 4517 upregulated and 3887 downregulated genes ( Figure 6A ). The heat map showed the top 50 upregulated/downregulated genes ( Figure 6B ). Then the differentially expressed genes were used for KEGG analysis by GSEA. We ranked signal pathways based on standardized enrichment scores ( Figure 7A ). The eight KEGG-annotated pathways positively associated with high CENPE expression were in Cell cycle, DNA replication, Ribosome, Fanconi anemia pathway, Homologous recombination, Oocyte meiosis, Cellular senescence, p53 signaling pathway ( Figures 7B–I ). The above research results indicated that in non-WNT/non-SHH MB, the genes co expressed with CENPE and the DEGs between high and low CENPE groups both included cell cycle and P53 signaling pathway in their gene function enrichment results.

To further verify the relationship between CENPE and cell cycle, we analyzed the correlation between CENPE and cell cycle regulatory genes in non-WNT/non-SHH MB. The cycle regulatory genes BUB1B, ESPL1, PTTG1, PCNA, PKMYT1, CDC45, PLK1, MCM2, MCM4, MCM6, E2F1, CCNA2, CCNB1, CCNB2, CCNE1, CDC6, CDC20, CDC25A, CDC25C and CDK1 were positively correlated with CENPE ( Figure 8 ).

To explore the regulatory effect of CENPE on the cell cycle and p53 pathway, CENPE siRNA was used to interfere with CENPE expression. The protein expression levels of CENPE, p53, p21, and CDK1 were detected by Western blotting. The results showed that after transfection with CENPE, the expression of CENPE significantly decreased and the expression of p53 increased ( Figures 9A–C ). The cell cycle analysis revealed that knocking down CENPE caused accumulation of cells in G2/M phase in both non-WNT/non-SHH MB cell lines ( Figures 9F–H ). The expression of key factors p21 and CDK1 downstream of p53 responsible for cell cycle regulation (G2/M) had also changed. Compared with the control group, knocking down CENPE resulted in an increase in p21 expression ( Figures 9A, D ) and a decrease in CDK1 expression ( Figures 9A, E ). These data indicated that CENPE may regulate cell cycle through the p53 pathway.

CCK-8 assays were performed to investigate the effect of CENPE on cell proliferation. The results showed that silencing CENPE significantly inhibited the proliferation of both cell lines at 24 h, 48 h, and 72 h after seeding ( Figures 10A–C ). In addition, immunofluorescence evaluation found that the average fluorescence intensity of cell proliferation protein Ki67 in the CENPE knockdown group was lower than that in the control group ( Figures 10D, E ). Based on the analysis of cell cycle, the inhibitory effect of knocking down CENPE on the proliferation of non-WNT/non-SHH MB cell lines may be partially mediated by cell cycle arrest.