Cell cycle-related biomarker lineage promotes the malignant progression of recurrent IDH wild-type glioma

Abstract

Background

Glioma is the most common primary brain tumor in adults, malignant progression and recurrence are inevitable. Although mechanisms of glioma development and progression have been demonstrated, the underlying driver pathway activation and the related therapeutic targets of recurrent gliomas, notably recurrent isocitrate dehydrogenase (IDH) wild-type gliomas, are rudimentary.

Methods

We analyzed the differential expression genes (DEGs) between two subgroups of 80 IDH wild-type recurrent gliomas from the Chinese Glioma Genome Atlas (CGGA). We then constructed and validated an 11-gene signature based on cell cycle-related DEGs. SPSS and R languages ​​were used for statistical analysis and graphical work.

Results

We proposed a molecular classification based on gene expression profiles and divided them into two subtypes that differed in their biological features and prognoses. We then constructed and validated an 11-gene signature based on cell cycle-related DEGs to better explore the association between cell cycle and tumor malignant progression. Additionally, the risk score could predict the sensitivity of patients to radiotherapy or chemotherapy. Functional analysis demonstrated that genes associated with the high-risk group were involved in various aspects of glioma cell cycle regulation. Moreover, the knockdown of SYCE2 blocked cell proliferation and improved cell sensitivity to temozolomide (TMZ).

Conclusion

This study developed a cell cycle-related signature for predicting the prognosis of recurrent IDH wild-type glioma, and identified a novel biomarker SYCE2, which could be used to assess the response of patients to TMZ.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04121-7.

Introduction

Glioma is the most common and malignant brain tumor, which is characterized by obvious genetic heterogeneity, high recurrence rate, high mortality rate, chemotherapy resistance and poor clinical prognosis [1]. According to the 2021 World Health Organization classification of Central Nervous System tumors, the adult diffuse gliomas can be simplified into three types: astrocytoma, IDH mutant; oligodendroglioma, IDH mutant and 1p/19q codeletion; and glioblastoma (GBM), IDH wild-type [2]. The new classification still underscores the importance of IDH mutation status for gliomas. According to reports, there are obvious molecular differences between IDH mutant and IDH wild-type glioma, and IDH wild-type glioma has a worse prognosis than IDH mutant glioma [3]. Owing to incomplete surgery or resistance to therapy, glioma inevitably relapse, and recurrent gliomas tend to be more aggressive and more malignant, particularly IDH wild-type recurrent gliomas [4, 5]. For the management of recurrent IDH wild-type gliomas, the treatment criteria are currently not definite. Consequently, further stratification of tumor prognosis and the exploration of novel and reliable biomarkers for predicting glioma may be beneficial for the management and targeted therapy of IDH wild-type recurrent glioma.

The increasing understanding of oncogenes and tumor suppressor genes in oncogenic somatic has led to the development of innovative treatment strategies. Despite significant advancements in cancer treatment, the challenge of cancer heterogeneity remains [6]. Through in-depth research into the cell cycle regulation system, we gradually realized that tumors were a disease closely linked to the cell cycle. Cell cycle dysregulation was the basis of abnormal cell proliferation and played a crucial role in the occurrence and progression of tumors [7]. The cell cycle has a tightly regulated process that proceeds in an orderly manner. This regulation is mainly accomplished by four kinds of substances: cyclins, cyclin-dependent kinases (CDKs), retinoblastoma proteins and CDKs inhibitors [8, 9]. In addition, there were some key transition points in the evolution of the cell cycle, known as cell cycle checkpoints. They restrict cell cycle transitions when stimulated by external pressure. Cell cycle checkpoints mainly include G1/S phase checkpoints, S phase checkpoints and G2/M phase checkpoints, which play an important role in the life activities of eukaryotes as a protective mechanism [10]. Defects in checkpoint function may lead to the proliferation of cells with inherited genetic mutations and chromosomal structural abnormalities that contribute to tumorigenesis and response to chemoradiation [11–13].

Uncontrolled cell proliferation is one of the characteristics of tumors, and dysregulation of the cell cycle is the biological basis of abnormal cell proliferation. There is growing evidence that cell cycle regulatory pathways are linked to other features of cancer, including metabolic remodeling and immune escape [14–16]. As a result, targeted therapies that target key sites in the cell cycle regulation mechanism can not only inhibit cancer cell proliferation, but also reverse cancer metabolism and restore cancer immune surveillance. With in-depth understanding of the mechanisms of cell cycle regulation, CDKs that facilitate the cell cycle transition are considered key therapeutic targets. Pestell et al. found that cyclinD1 can increase the recruitment of macrophages (CD11b+) in the breast tumor microenvironment and promote angiogenesis and the occurrence of breast tumors [17]. Li et al. found that Palbociclib improved the resistance of GBM to TMZ through lncRNASNHG15/CDK6/miR-627, making it more sensitive to TMZ treatment, and reduced the M2 polarization in glioma-associated microglia [18]. Sun et al. confirmed the effective anti-tumor effect of Palbociclib in a mouse glioma model [19]. Morris-Hanon et al. found that Palbociclib effectively prevented the proliferation of patient-derived glioma stem cells and induced cells to exhibit a senescence-like phenotype [20]. These results suggest that targeting CDKs inhibitors may improve the prognosis of glioma patients. Gobin et al. also demonstrated that DNA repair and cell-cycle gene expression were closely related to the prognosis of recurrent glioblastoma. And tumor stratification by gene signatures could guide targeted therapy in recurrent glioblastoma [21]. Thus, the precise regulation of the cell cycle plays a critical role in the normal survival, reproduction, differentiation and inheritance of organisms. However, its association with IDH wild-type recurrent glioma has been rarely reported.

In this study, we performed molecular subtyping on 80 IDH wild-type recurrent gliomas from the CGGA. The results showed that patients could be divided into two subtypes with different biological characteristics and prognoses. We then developed a cell cycle-related biomarker lineage to predict prognosis in patients and validated it in the Glioma Longitudinal Analysis (GLASS) dataset. Using Kaplan-Meier (K-M) survival curve and Cox regression analysis, we discovered that risk score was an independent prognostic biomarker. Bioinformatics analysis revealed that the risk score was strongly correlated with cell cycle-related functional annotations and checkpoints. This model could also predict the sensitivity of glioma patients to radiotherapy or chemotherapy. In addition, we identified an important gene (SYCE2) that affects the proliferation function of tumor cells in vitro and improves sensitivity to TMZ. These results provide new ideas for understanding the mechanism of malignant progression and individualized treatment of IDH wild-type recurrent gliomas.

Materials and methods

Study design

The flow chart of our research is shown in Fig. S1.

Data collection and pre-processing

In this study, 80 IDH wild-type recurrent glioma samples were selected from the CGGA microarray data as the discovery cohort. Download transcriptional RNA-sequencing data and corresponding clinical information from the CGGA database (http://www.cgga.org.cn) [22, 23]. Similarly, 81 IDH wild-type recurrent glioma samples were obtained from the GLASS database (http://www.synapse.org/glass) and used as the validation cohort [24]. We obtained glioma samples with RNA-seq data and corresponding clinical information from The Cancer Genome Atlas (TCGA) database (http://cancergenome.nih.gov/), and used them to verify the correlation between SYCE2 and chemotherapy and radiotherapy. Patient inclusion criteria and overall survival (OS) were defined as follows: (1) recurrent IDH wild-type glioma; (2) patients with missing OS or censor information were excluded; (3) histopathological diagnosis verified as 1p/19q co-deleted tumors were excluded; (4) OS was defined as from the date of initial recurrence to the date of last follow-up, second recurrence or death. Next, genes with the same hazard ratio (HR) value direction ​​were screened in the GLASS and CGGA RNA seq datasets to eliminate potential differences between the two datasets. Table 1 summarizes the clinical features of the two cohorts of patients.

Table 1.

Clinical characteristics of patients in this study

Characteristic	CGGA cohort	GLASS cohort
Total cases	80	81
Gender
Male	45	29
Female	35	52
Age
≤ 40	24	8
> 40	56	73
Grade
II	5	0
III	14	1
IV	61	80
MGMT promoter
Methylation	27	11
Unmethylation	33	20
NA	21	51
Subtype
Neural	19	0
Proneural	44	9
Classical	2	23
Mesenchymal NA Histology Gliomblastoma Astrocytoma NOS	15 0 61 5 14	36 13 80 1 0

Consensus clustering

We first analyzed the remaining genes by univariate Cox regression to obtain genes significantly associated with the OS in the CGGA cohort (P < 0.05). Then, 896 candidate genes with highly variable expression [median absolute deviation (MAD) >0.5] were selected. Unsupervised clustering of candidate genes was performed using the R package Consensusclusterplus for robust clustering. The optimal K was evaluated using cumulative distribution function and consensus heatmap [25]. We then trained a partition around the central point classifier in the discovery cohort [26]. According to the Pearson correlation between samples and centroids, each sample in the validation cohort was assigned to the gene expression centroid with the highest correlation, and the corresponding subtypes were obtained. Principal components analysis (PCA) was employed to distinguish expression differences between the two subtypes using the R package princomp.

Functional analysis

DEGs between subgroup 1 (Sub1) and subgroup 2 (Sub2) patients in the CGGA cohort were obtained by the R package limma. Metascape (http://metascape.org) was used for functional enrichment analysis of the DEGs between Sub1 and Sub2 [27], and the results were exhibited by the R package ggplot2.

Construction of a prediction model

We first identified the DEGs between Cluster 1 (C1) and Cluster 2 (C2) through the R package limma. Univariate Cox regression analysis was then performed to identify DEGs with prognostic significance (P < 0.01). Subsequently, the survival-associated DEGs were incorporated into a least absolute shrinkage and selection operator (LASSO)-Cox regression model, a linear regression method, to obtain gene signature and corresponding coefficients. The regression coefficients were used to calculate the risk score for each sample in the discovery and validation cohorts. The risk score for each glioma patient in the CGGA and GLASS cohorts was calculated according to the following formula:

Risk score = (expr_gene1 × β1) + (expr_gene2 × β2) +… + (expr_gene11 × β11).

To determine whether the risk score is an independent prognostic factor, we executed univariate and multivariate Cox regression analyses. A nomogram was then constructed based on the risk score and clinical characteristics using the R software package rms.

Bioinformatic analysis

Pearson correlation analysis was performed to determine genes positively and negatively associated with risk scores in the CGGA and GLASS cohorts. Then, these genes (R >0.6, P < 0.05) were uploaded to the Metascape website for functional enrichment analysis. Gene set enrichment analysis (GSEA) software (www.gsea-msigdb.org) was used to validate the functional enrichment analysis related to risk score [28]. And gene set variation analysis (GSVA) was carried out to identify the association between biological processes and risk score.

Immunohistochemistry (IHC)

We obtained tissues from clinical specimens for embedding and frozen section (10 μm section thickness). Frozen sections were fixed with cold acetone for 1 min and then baked at 55℃ for 30 min. Subsequently, brain tumor sections were quenched, blocked, incubated in primary antibody (anti-SYCE2, NBP1-90637, Novus, USA) and secondary antibody (ZSGB-Bio, China), stained using the DAB reagent, counterstaining, sealing. Six fields of each section were selected for quantitative analysis, and stained cells were counted by two investigators independently. Antibodies used for immunohistochemical analyses were optimized using both positive and negative controls.

Cell culture and transfection

Human glioma cell lines such as H4, U87, LN229, SNB19, A172, U118 and U251 were acquired from the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. The GBM patient-derived cell lines, N33, have been authenticated after stable passage [29]. All cell lines were cultured in DMEM medium (only N33 cells were cultured in DMEM / F12 medium) containing 10% fetal bovine serum (FBS, Gibco) and 1% penicillin–streptomycin (Gibco) in a humidified 5% CO₂ atmosphere at 37℃. All experiments in this study were performed with mycoplasma-free cells. LN229 and U251 cell lines (50%-70% density) were transfected with lentiviral vectors at a multiplicity of infection (MOI) of 5 for 10–12 hours. They were then cultured in a puromycin-containing medium to select for stable SYCE2-knockdown cells. The levels of SYCE2 protein and knockdown efficiency were measured by Western blot. The details of the shRNA sequences were as follows (sh-SYCE2-1 (110800-1): 5’-GCGTGGAGACTGTGTACAAAG-3’ and sh-SYCE2-2 (110802-1): 5’-ATGCACTCATGACCAACTTCA-3’).

Western blot

Total cellular proteins were extracted by RIPA lysis buffer (Solarbio, China), and quantified by BCA protein assay reagent (Thermo, USA). The extracted proteins were separated by 12.5% SDS-PAGE and then transferred onto polyvinylidene fluoride membranes (Millipore, USA). After blocking in 5% nonfat milk for 1 h at room temperature, the membranes were then incubated with antibodies specific for SYCE2 (NBP2-85849, Novus, USA) or Alpha Tubulin (11224-1-AP, Proteintech, USA) at 4℃ overnight. Later, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at RT, and then the protein band signal was detected on the machine (Bio-Rad, USA). Alpha Tubulin was used as a loading control to quantify relative protein levels.

Cell growth and colony formation assays

For cell growth assay, glioma cells were seeded in 96-well plates at a density of 1000 per well. Cell Counting Kit-8 (CCK-8) assays were performed on days 0, 1, 2, 3, 4, and 5 after cell attachment. Add 10ul of CCK-8 reagent (Dojindo, Japan) to each well, incubate at 37℃, 5% CO₂ for 1 h, and the optical density (OD) values were determined by measuring the absorbance at 450 nm. The growth curve of each group was plotted according to the average OD values. To assess the clonogenic capacity, 200 glioma cells were plated in 6-well plates. After 10–14 days of culture, the cells were fixed for 15 min in 4% paraformaldehyde for 15 min and then stained with 0.1% crystal violet for 10 min. The number of cell colonies was counted under the microscope.

DNA content quantitation assay (cell cycle)

1 × 10⁶ tumor cells were collected from each group, washed with PBS and fixed overnight at 4℃ in 70% ethanol solution. After rewashing with PBS, RNase A (37℃ water bath for 30 min) and propidium iodide (4℃ incubation for 30 min) were added in turn for staining. Cells were then analyzed using a CytoFLEX Flow Cytometer (Beckman Coulter) and the results were analyzed using Cytexpert software.

Statistical analysis

In our study, all data statistics and analysis were performed using R language v4.1.2 and SPSS v22.0. Chi-square tests were used to determine differences in clinical characteristics between the two subtypes. The survival differences between groups were assessed using K-M survival curve analyses with log-rank tests. P < 0.05 was considered statistically significant.

Results

Consensus clustering divided IDH wild-type recurrent glioma into two subgroups

To characterize the intrinsic heterogeneity of IDH wild-type recurrent glioma patients, we conducted a cluster analysis on 80 samples from the CGGA cohort using the R package ConsensusClusterPlus. By consensus clustering of 80 samples, patients could be divided into two different subtypes (Fig. 1A and Fig. S2A-C). Furthermore, we applied PCA to the gene expression profiles of 80 samples and verified significant differences within these two subtypes (Fig. 1B). Notably, we observed an obvious difference in prognosis between the two subtypes, with shorter overall survival for Sub1 than Sub2 (Fig. 1C).

Fig. 1.

Identification of distinct subtypes in recurrent IDH wide-type glioma. A Heatmaps of two subtypes via consensus clustering in CGGA cohort. B Principal component analysis (PCA) of two subtypes using candidate genes. C Survival analyses of patients in 2 subtypes based on overall survival (OS). D Functional enrichment analysis for Sub1 by Metascape

Subsequently, we investigated potential biological functional differences among the two subtypes. We first analyzed the gene expression profiles to identify the DEGs between Sub1 and Sub2 in the CGGA cohort, and then performed functional enrichment analysis with DEGs (FDR < 0.05) of each subtype (Fig. S2D). Metascape analyses showed that Sub1 (669 up-regulated genes) was primarily enriched in signaling pathways including cell cycle, DNA conformation change, DNA repair and DNA metabolic process (Fig. 1D). Unlike those enriched in Sub1, genes enriched in Sub2 (123 up-regulated genes) were related to cation transmembrane transport, small molecule biosynthetic process and metabolism of steroids (Fig. S2E). These results revealed that cell cycle-related biological processes were strongly associated with poor prognosis in IDH wild-type recurrent glioma patients.

We further explored the relationship between cell cycle-related genes and IDH wild-type recurrent glioma. Among the pathways enriched in Sub1, we selected the pathway items related to the cell cycle, combined with the corresponding gene sets, and obtained a total of 334 genes. We then used this gene set to stratify our sample in the CGGA cohort. The samples were still divided into two clusters according to the top horizontal line of the clustering tree (Fig. S3A). We demonstrated significant differences between these two subtypes based on PCA and K-M survival curve analysis (Fig. S3B-C). To validate the above findings, we assessed the reproducibility of subtypes in the validation cohort using the R package pamr. Then, each sample in the validation cohort was assigned to the corresponding cluster based on the Pearson correlation between samples and centroids. As expected, similar results were observed in the GLASS cohort, with the obtained subtypes showing similar expression patterns and significant prognostic values as the CGGA cohort (Fig. S3). The clinical characteristics of these two clusters of patients in CGGA (Table S1) and GLASS (Table S2) were summarized in Table S1 and S2.

Construction of a cell cycle-related prognostic signature for IDH wild-type recurrent glioma

We first identified 330 DEGs between the two clusters in the CGGA cohort by R package limma (FDR < 0.05). We then selected 84 genes strikingly correlated with patients’ OS in the CGGA cohort by univariate Cox regression analysis (P < 0.01), and 11 genes were screened via the lasso regression algorithm (Fig. 2A-C). Subsequently, the risk score for each case was calculated using the formula mentioned in the Materials and methods section. And use a similar approach to calculate the risk score for the validation cohort.

Fig. 2.

Construction of a cell cycle-related prognostic signature in the CGGA cohort. A Cross-validation for tuning parameter screening in the LASSO regression model. B LASSO coefficient profiles of the common genes. C Coefficient values of each signature gene. D Distribution of the risk score, overall survival (OS) and expression level of 11 genes in the risk score. E-G Kaplan-Meier (K-M) survival analysis of the risk score. H-K Kaplan-Meier (K-M) survival analysis (OS and PFS) of the risk score in glioma patients stratified by received chemotherapy or radiotherapy

To comprehensively describe this risk score, we evaluated the association between risk score and patient survival, as well as the relationship to the expression level of each gene in CGGA (Fig. 2D) and GLASS (Fig. S4A) cohorts. The results showed that survival time tended to shorten as the risk score increased, and the mortality rate (ratio of red dots to blue dots) in the high-risk group was higher than that in the low-risk group. The expression levels of genes were positively correlated with risk scores, all of which were associated with poor prognosis. Patients were stratified into high-risk and low-risk groups based on the median value of their risk scores. K-M survival curve analysis indicated that patients in the high-risk group had a significantly shorter OS and progression-free survival (PFS) time than those in the low-risk group in the CGGA cohort (Fig. 2E). In addition, this risk score could be used to predict survival in GBM patients rather than LGG patients (Fig. 2F-G). Similarly, the high-risk group was also observed to be associated with poor prognosis in the GLASS cohort (Fig. S4B). Although the prognostic difference between high-risk and low-risk groups in GBM patients was not very significant, the high-risk group showed a poorer prognosis in the GLASS cohort (p = 0.065; Fig. S4C). We then stratified the patients who received chemotherapy (TMZ) or radiotherapy separately and observed that there was a significant difference in the prognosis between the high-risk group and the low-risk group (Fig. 2H-K). However, the results were not statistically significant due to the small sample size of patients in the GLASS cohort who received radiotherapy or chemotherapy (Fig. S4D-E). These results suggest that risk score may serve as a biomarker of a patient’s sensitivity to chemotherapy (TMZ) or radiotherapy.

An individualized prediction model based on the risk score

We then performed univariate and multivariate Cox regression analyses. The results showed that grade and risk score were independent prognostic factors in the CGGA cohort of patients (Fig. 3A). Subsequently, AUC analysis was performed to estimate the specificity and sensitivity of the risk score to predict survival. We calculated AUC curves at different time points and the risk score had a highly time-dependent AUC of the CGGA cohort (1 year: 72.6%, 2 year: 80.6%, 3 years: 81.3%) (Fig. 3B). Later, a nomogram model was constructed with independent prognostic parameters for patient OS, with C-index of 0.70 in the CGGA cohort (Fig. 3C). The calibration curve of 1-, 2- and 3-year probability of survival implied that the nomogram exhibited concordance and calibration well (Fig. 3D). In short, our results confirmed that the high predictive value of risk scores for IDH wild-type recurrent glioma patients.

Fig. 3.

Prognostic validity of the risk score in IDH wide-type recurrent glioma. A Univariate and multivariate Cox regression analyses of risk score and clinic pathological factors. B The ROC curve analyses to predict 1-year, 2-year, and 3-year OS according to the risk score. C A nomogram was developed by integrating the signature risk score with the clinicopathologic features. D Calibration curves of the nomogram for predicting OS at 1-year (blue line), 2-year (red line) and 3-year (black line)

Biological functions analysis of the risk score

To explore the underlying biological function of risk score, we used Pearson correlation analysis to identify genes significantly correlated with risk score (R > 0.6, p < 0.05) and annotated their functions using Metascape. The results revealed that the positively related genes (844 genes) were mainly enriched in cell cycle, DNA replication, chromatin organization and regulation of cell cycle G2/M phase transition (Fig. 4A). And GSEA analysis showed consistent results. The high-risk group was enriched in E2F targets, G2M checkpoint and mitotic spindle (Fig. 4B). In addition, by assessing the relationship between risk scores and cell cycle checkpoint-related signaling pathways, high-risk groups were found to be considerably enriched in these features (Fig. 4C). To further investigate the association between risk scores and cell cycle, genes associated with cell cycle (S, G0/G1 or G2/M phase) were included in the analysis. The analysis showed that the risk score was positively correlated with the expression levels of cell cycle-related genes in the CGGA cohort (Fig. 4D). And similar results were observed in the Glass cohort (Fig. S5). These results indicated that the cell cycle-related biomarker lineage plays a significant role in predicting the malignant progression of IDH wild-type recurrent gliomas.

Fig. 4.

Functional analyses related to the 11-gene signature in the CGGA cohort. A Functional annotation of genes positively correlated with the risk score by Metascape. B GSEA analysis was used to explore the biological functions that were significantly enriched in patients with the high-risk score. C The relationship between risk score and cell cycle checkpoint-related pathways. D Relationship between risk score distribution and expression of cell cycle-related genes (S; G1/S; G2/M phase)

SYCE2 is an independent predictor of response to chemotherapy in patients with IDH wild-type recurrent gliomas

To explore the main reasons why the risk score predicts the sensitivity of glioma patients to chemotherapy, we analyzed the 11 genes that make up the signature. Patients were stratified into high and low groups based on the best cutoff value in the survival analysis. Only the genes SYCE2, RFX3 and MEN1 showed significant prognostic value in both the recurrent and primary cohorts for patients receiving chemotherapy (Table S3). Subsequently, we focused on the SYCE2 gene, which is rarely reported in the tumor-related literature and has never been studied in the field of glioma.

We first assessed whether the expression level of SYCE2 affects the clinical prognosis of patients via K-M survival curve analysis. The results showed that patients with high SYCE2 expression had shorter OS and PFS in the CGGA and GLASS cohorts (Fig. 5A and Fig. S6A). We then aimed to investigate whether SYCE2 could serve as a predictive biomarker for patients’ sensitivity to adjuvant chemotherapy or radiotherapy. Patients in the SYCE2 high group who received only ion radiotherapy (IR) or chemotherapy (TMZ) had shorter OS and PFS compared to the SYCE2 low group in the CGGA cohort (Fig. 5B-C). Similar results were observed in GLASS and TCGA (primary cohort) cohorts (Fig. S6B-E).

Fig. 5.

SYCE2 is an independent predictor of response to TMZ or IR in IDH wide-type recurrent glioma. A The expression level of SYCE2 was correlated to the prognosis of patients in the CGGA cohort. B-C Survival analysis SYCE2 expression level was associated with survival in patients treated with IR alone or TMZ alone in the CGGA cohort. IR, ion radiotherapy; TMZ, temozolomide. D Kaplan-Meier (K-M) survival analysis (OS) of the four subgroups stratified by MGMT promoter methylation status and the SYCE2 expression in CGGA and GLASS cohorts

In clinical treatment of glioma, the MGMT promoter (MGMTp) methylation status correlates with the response of patients to TMZ. Thus, we investigated the relationship between SYCE2 and MGMTp. Patients were divided into four subgroups based on the level of expression of SYCE2 and the methylation status of the MGMTp. As shown, SYCE2 could well distinguish patient outcomes, and a significant reduction in survival was observed in the high SYCE2 and MGMTp methylation group (Fig. 5D). However, there was no significant difference in the unmethylated MGMTp group. These results suggested that SYCE2 may be a potential treatment target for glioma with TMZ resistance.

SYCE2 promotes malignant progression of glioma

To investigate the potential biological function of SYCE2 in glioma, we carried out relevant experiments in vitro. We first examined the expression of SYCE2 in human glioma tissues and glioma cell lines. The results showed that SYCE2 was highly expressed in high-grade gliomas compared with low-grade gliomas (Fig. 6A). In addition, among the eight cell lines, the protein level of SYCE2 was highly expressed in H4, U251 and LN229 cell lines (Fig. 6B). We then transfected human SYCE2 shRNA into LN229 and U251 cell lines to knock down the expression of SYCE2 and identified the knockdown effect by western blot (Fig. 6C). Subsequently, we investigated the effect of SYCE2 on the proliferation of glioma cell lines. CCK-8 assay revealed that the cell proliferation rate was considerably repressed after SYCE2 silencing (Fig. 6D). Similarly, In the colony formation assay, SYCE2 knockout significantly inhibited the clonogenic capacity of U251 and LN229 cells compared to negative controls (Fig. 6E). In addition, we conducted cell cycle analysis to determine whether SYCE2 promotes cell proliferation by altering the cell cycle. The flow cytometry results showed that SYCE2 knockout blocked the cell cycle at the G0/G1 phase, leading to the evident accumulation of cells in this phase (Fig. 6F). Furthermore, we measured the IC50 value ​​of tumor cells (without any treatment) in response to TMZ (Fig. S7). The combination treatment of SYCE2 suppression and TMZ intervention greatly reduced the colony formation compared to treatment with only TMZ or SYCE2 silencing (Fig. 6D-G).

Fig. 6.

The knockdown of SYCE2 inhibits the proliferation of glioma cells. A SYCE2 expression in different grades of gliomas was detected by IHC analysis. B The expression of SYCE2 protein in the glioma cell lines. C Western blotting analysis of SYCE2 protein expression in LN229 and U251 cell lines after applying with control shRNA or shSYCE2. D CCK-8 detected the growth curve of glioma cells treated with control shRNA or shSYCE2. E Colony formation detected the clonogenic capacity of glioma cells treated with control shRNA or shSYCE2. F Flow cytometry analyses for cell cycle of glioma cells transduced with control shRNA or shSYCE2. G Colony formation assay showing the sensitizing effects of SYCE2 on LN229 (IC50:337.37umol/ml) and U251 (IC50:262.73umol/ml) cells after TMZ treatment. * P < 0.05, ** P < 0.01, *** P < 0.001

Discussion

Recurrence is one of the characteristics of the tumor. Recurrent gliomas are more aggressive and malignant, generally developing from low-grade to high-grade gliomas. Among the numerous molecular genetic alterations in glioma, the prognostic effect of IDH mutations has been commonly accepted. In the present study, we identified two subtypes of IDH wild-type recurrent glioma with distinct biological features and prognosis via gene expression profiles in the CGGA cohort. Sub1 has a poor prognosis, and its functional annotation items were predominantly related to cell cycle-related biological processes. It mainly incorporated pathways such as cell cycle, DNA replication, chromatin organization, histone modification, Fanconi pathway and cell cycle phase transition. These pathways were primarily involved in fundamental biological processes that affected tumor growth, proliferation, differentiation and survival, thereby promoting glioma progression and recurrence. Meanwhile, Sub2 was enriched in oxidative phosphorylation, cation transmembrane transport, small molecule biosynthetic process and metabolism of steroids. It indicated that Sub2 was closely related to molecular synthesis, transport and metabolism. Sub2 is more similar to maintaining the normal life activities of the organism compared to Sub1.

We then focused on the analysis of cell cycle-related DEGs to explore their relationship with the malignant progression of IDH wild-type recurrent gliomas. Based on cell cycle-related DEGs, we established an 11-gene signature that could classify patients into high and low risk. The prognostic value of this risk score was fully verified by K-M curve analysis and Cox regression analysis. These results implied that the risk score has a high predictive value for IDH wild-type recurrent glioma patients. Moreover, we observed significant prognostic differences between high and low-risk groups of patients receiving radiotherapy or chemotherapy based on OS and PFS survival analyses. This indicated that the 11-gene signature could predict the sensitivity of patients to radiotherapy or chemotherapy. Given these significant results, the 11-gene signature may be beneficial in stratifying IDH wild-type recurrent glioma patients for adjuvant therapy and provide some insights into the clinical management of patients.

For bioinformatics analysis of the 11-gene signature, functional annotation analysis was performed for significantly associated genes (R >0.6; P < 0.05) via Metascape. The results showed that the positively related genes were mainly enriched in cell cycle, DNA replication, chromatin organization and regulation of cell cycle G2/M phase transition. And GSEA analysis demonstrated that the high-risk group was enriched in E2F targets, G2M checkpoint and mitotic spindle. These results suggested that high-risk groups may influence the malignant progression of glioma by affecting cycle-related biological pathways. The disorder of the cell cycle regulation mechanism leads to abnormal cell proliferation and differentiation, which is one of the main reasons for the occurrence and development of tumors. It has been reported that cell cycle arrest reflects a survival mechanism that provides tumor cells with an opportunity to repair their own damaged DNA [30]. Thus, the precise regulation of the cell cycle plays a crucial role in the normal survival, reproduction and differentiation of organisms.

Most of the genes in the signature were also associated with malignant tumor progression. CBS protein catalyzes the conversion of homocysteine to cystathionine to maintain glutathione levels. Knockdown of CBS in vitro increases HIF-2α protein levels and HIF-2 target gene expression, thereby promoting glioma tumorigenesis [31]. CCDC88C activates Gi-α and enhances non-canonical Wnt signaling by FZDRs. Although inhibits the β-catenin/TCF/LEF pathway and tumorigenesis, it enhances PI3K-Akt and Rac1 signaling and tumor cell invasiveness [32]. Tumor suppressor MEN1 plays a role in histone modification and epigenetic gene regulation. Wang et al. demonstrated that MEN1 was strongly correlated with poor outcomes in glioma patients. Importantly, MEN1 inhibitors effectively reduced the proliferation of adult glioma cells [33]. KDM1A is highly expressed in glioma stem cells (GSCs). KDM1A silencing could reduce the stemness and induce the differentiation of GSCs and improve the resistance to chemotherapy/radiotherapy [34, 35]. MDC1, as the mediator of the DNA damage checkpoint, interacts with H2AX to promote the recruitment of repair proteins to the sites of DNA breaks in response to DNA damage [36, 37]. The TNKS1BP1 protein has been reported to regulate actin cytoskeleton and cancer cell invasion [38]. In addition, TNKS1BP1 affects the cell cycle distribution and regulates DNA double-strand breaks (DSBs) in the homologous recombinant pathway via suppression of RAD51 foci formation. And its overexpression may affect the outcome of chemotherapy and radiotherapy in cancer patients [39]. XPO5 is one of the key proteins responsible for the transport of precursor miRNAs from the nucleus to the cytoplasm. Shigeyasu et al. confirmed that XPO5 knockdown resulted in decreased cell proliferation, cell cycle arrest, and downregulation of key oncogenic miRNAs in colorectal cancer cells [40].

The SYCE2 protein is part of the synaptonemal complex (SC) formed between homologous chromosomes during meiotic prophase. Although the components of the SC were initially thought to be expressed only in the germ line, it is reported that SYCE2 was expressed at varying levels in somatic tumors, such as tumors of the ovary, thyroid, stomach, kidney, etc [41, 42]. Hosoya et al. observed elevated SYCE2 expression after treatment of DLD1 (human colorectal adenocarcinoma) and HT1080 (human fibrosarcoma) cell lines with the demethylating agent 5-azacytidine, suggested that SYCE2 expression was regulated by a demethylation-dependent process [43]. They also verified the relationship between SYCE2 protein and HP1α, displaying that SYCE2 directly binds to and dissociates HP1α from H3K9me3. This results in enhanced ATM-dependent DNA repair activity and increased total DNA repair activity when exogenous DNA damage is induced. In other words, SYCE2 activates an ATM-dependent DNA damage response mechanism in cancer cells, ultimately inducing resistance to DNA damage. In addition, previous studies reported that SYCE2 silencing in mice reduces the efficiency of DSB repair in the germ line [44]. This suggested that SYCE2 was closely related to DSBs. SYCE2 confers tumor cells with resistance to radiation and DNA cross-linking chemotherapeutic agents by increasing DSBs repair activity. Thus, SYCE2 is a potential target for the sensitization of tumor cells to DNA-damaging adjuvant therapy. Glioma has strong genetic heterogeneity, high mortality and chemotherapy resistance. In this research, bioinformatic analysis showed that SYCE2 correlates with the MGMTp methylation status. It could identify the prognosis of patients with MGMTp methylation, but not non-methylated patients. The methylation status of MGMTp is an important prognostic factor for the clinical treatment of glioma patients. When the MGMT (DNA repair protein which could repair the killing effect of chemotherapy on tumor cells) promoter is methylated, gene function will be inhibited [45, 46]. Thus, it may have a better response to alkylation chemotherapy (TMZ). In addition, the analysis revealed that SYCE2 could predict the sensitivity of IDH wild-type recurrent glioma patients to TMZ therapy. Combined with subsequent robust experimental results, we confirmed that SYCE2 was a poor prognostic factor in gliomas and may be a potential treatment target for glioma patients with TMZ resistance.

Other genes such as HCFC1, RFX3 and PLCG1 have been reported in many cancers, including glioma, kidney and cervical cancer [47–49].

Conclusions

In conclusion, we have identified a molecular classification based on gene expression profiles in IDH wild-type recurrent glioma. We then constructed a cell cycle-related signature for predicting the prognosis of IDH wild-type recurrent glioma and identified a novel biomarker SYCE2, which could be used to evaluate the response of patients to TMZ. Our findings provide new insights that this signature may be a potential prognostic biomarker for personalizing clinical treatment of IDH wild-type recurrent glioma patients that could predict patient sensitivity to chemo/radiotherapy. As a novel oncogene, SYCE2 could be used to assess malignant progression and response to TMZ in glioma patients. However, we acknowledge several limitations of our study. First, the cell lines used in this study are standard GBM lines (LN229 and U251 cell lines), not specifically derived from recurrent glioblastomas. There are currently no commercialized recurrent glioblastoma cell lines, and it is difficult to obtain corresponding samples. And, while our study was performed in vitro, future in vivo experiments are essential to confirm the relevance of our observations. Similarly, future studies with larger cohorts and deeper experiments are warranted to validate these findings.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J.B.X. and Y.Z. involved in data analysis, article writing and performed most of the experiments. W.H.F. and W.P.M. participated in data downloading and preliminary analysis. W.P.M. involved in specimen collection and IHC assay. H.J.W. and Z.S.B involved in conception, supervision, and design of all manuscripts. All authors have read and approved the final manuscript.

Funding

This study was supported by funds from the National Natural Science Foundation of China (81972337); Natural Science Foundation of Beijing Municipality (JQ20030, 7222018); Excellent young talents project of Capital Medical University(B2101); Postdoctoral Scientific Research Developmental Fund of Heilongjiang Province of China (LBH-Q20042); Beijing Postdoctoral Research Foundation(2021-ZZ-022); the China Postdoctoral Science Foundation (2021M702308).

Data availability

All data generated or analyzed during this study can be obtained directly by contacting the corresponding author (H.J.W.). This study analyzed publicly available datasets, including CGGA, GLASS, and TCGA. These data can be found here: http://www.cgga.org.cn/; http://www.synapse.org/glass; https:// cancergenome.nih.gov/.

Declarations

Ethics approval and consent to participate

This study was conducted with the approval of the Ethics Committee of Beijing Tiantan Hospital. All methods were carried out in accordance with the institutional guidelines and regulations. Written informed consent was obtained from all patients.

Consent for publication

The authors declare that they have no competing interests.

Competing interests

The authors declare no competing interests.