Expression and prognostic role of E2F transcription factors in high‐grade glioma

Abstract Introduction Patients with high‐grade glioma (HGG) suffered poor survival due to inherent or acquired therapeutic resistance and refractory recurrence. The outcome of HGG patients has improved little during the past decade. Therefore, molecular signatures are urgently needed for improving diagnosis, survival prediction and identification of therapeutic targets for HGG. E2F transcription factors (E2Fs), a family of transcription factors recognized as master regulators of cell proliferation, have been found to be involved in the pathogenesis of various tumor types. Aims To investigate the expression of E2Fs and their prognosis value in high‐grade glioma (HGG). Results Expression of E2Fs was analyzed in 394 HGG samples from TCGA dataset. E2Fs were generally expressed in HGG. Except for E2F3 and E2F5, expression of E2Fs was significantly upregulated and linked with grade progression. E2F1, E2F2, E2F7, and E2F8 were highly correlated with aggressive proliferation oncogenes, as well as potential therapeutic resistance oncogenes. Elevated E2Fs (not E2F3) were associated with adverse tumor features and poorer outcome. E2F7 and E2F8 exhibited superior outcome prediction performance compared with other E2Fs. Additionally, E2F7 and E2F8 independently predicted poorer survival in HGG patients. Gene set enrichment analysis identified a variety of critical oncogenic pathways that were tightly associated with E2F7 or E2F8, including epithelial‐mesenchymal transition, NFκB, STAT3, angiogenesis pathways. Furthermore, elevated expression of E2F7 indicated worse therapeutic response of HGG to irradiation and silencing of E2F7 conferred higher cell‐killing effect when combined with irradiation treatment. Mechanically, E2F7 directly regulates the transcriptional activity of EZH2 via binding at the corresponding promoter area. Conclusions E2Fs (except for E2F3 and E2F5) are highly expressed in HGG and indicate adverse outcome. E2F7 and E2F8 were identified as novel potential prognostic markers in HGG. E2F7 was further validated to be closely associated with radioresistance of HGG and a critical transcriptional regulator of EZH2.


| INTRODUC TI ON
Glioma is one of the most frequent and malignant primary brain tumors, accounting for more than 70% of adult brain tumors. 1 Highgrade glioma (HGG) comprises anaplastic glioma and glioblastoma. 2 The hallmarks of HGG are composed of aggressive infiltrative pattern, high proliferation rate, therapeutic resistance, and fast recurrence speed. The current available clinical treatment strategy includes maximum surgical resection, radiotherapy (IR), and temozolomide (TMZ) chemotherapy, 2 while the survival outcome of HGG patients has improved little, owing the therapeutic resistance of HGG. 3 Current study shows that pathological classification is limited, and molecular signatures are urgently needed for improving diagnosis and survival prediction for HGG. 4 E2F transcription factors (E2Fs) are a family of members that play crucial role in coordinating the balance of cell cycle via a transcriptional axis. 5 E2Fs can be generally sorted into three groups based on their structure and identified function, 6 activators (E2F1-3), canonical repressors (E2F4-6), and atypical repressors (E2F7-8). The expression of activators increases at the G1-S phase transition, while the atypical repressors peak in the late S phase. Other members are generally expressed during all phases. 5 The DNA binding ability of activators and canonical repressors is mediated by the protein complex with a member of the transcription factor dimerization partner family (TFDP1, TFDP2, or TFDP3). 7 However, E2F7 and E2F8 are quite unique and they can bind E2F consensus sequences independent of the dimerization proteins. 8,9 Currently, multiple types of regulation of E2Fs have been identified. Transcriptionally, E2Fs are largely self-regulated, 10 and studies found that E2Fs' subcellular localization was controlled by a variety of factors, like CRM1. [11][12][13] In addition, a series of modifications to E2Fs have been identified, affecting the function, location, or stability of E2F family. 14,15 Accumulating evidence has implied that E2Fs were closely linked with tumorigenesis in a variety of cancer types. [16][17][18] The elevated expression of E2Fs was found to be closely associated with poor prognosis of cancer patients, like pancreatic tumors. 19 Among the E2Fs, E2F1 is the most widely investigated member in a huge panel of cancers. However, interestingly, in contrary to the previous defined function groups, some studies revealed that the atypical repressors also played a critical role in conferring aggressiveness to tumor cells.
For example, Qing et al found that E2F8 transcriptionally promoted the expression of CCND1 via completely binding at the corresponding promoter area. 18 This raise a possibility that the function of E2Fs in tumor cells might be hijacked by oncogenic factor and confer to distinct or opposite function as we current know. Studies revealed that the oncogenic role of E2Fs was not confined to aggressive cell cycle. Weijts et al found that E2F7 and E2F8 induced angiogenesis by interacting with hypoxia-inducible factor 1 (HIF1) and promoting the expression of vascular endothelial growth factor A (VEGFA), 20 two critical factors in HGG. In addition, E2F1 and E2F2 were reported to be highly associated with angiogenesis genes and cell interaction in breast cancer, conferring a more invasive phenotype. 21 Interestingly, the study found that E2F1 regulated the glycolysis process in cancer cells, another hallmark of HGG. 22 Via recruiting histone deacetylase and chromatin regulators, E2F7 promoted the DNA repair process in a transcription-independent manner. 23 However, even though the E2Fs have been reported to tightly linked with malignant behavior of multiple cancer cells, the expression pattern and prognosis role of E2Fs have not been fully characterized in HGG.
In the present study, we investigated the expression profile of E2Fs, as well as the survival outcome in HGG patients via the analysis of TCGA dataset. Gene set enrichment analysis was utilized to identify alternated oncogenic pathways. The therapeutic resistance role of E2Fs was explored, and we further validated the critical role of E2F7 in promoting radioresistance.

| Ethics
In this study, the usage of cell lines and experimental animals (nude mice) was approved by the Scientific Ethics Committee of Xi'an Jiaotong University, Xi'an, China.

| In vitro cell cultures
Glioma cell lines (U87 and U373) were provided by Xi'an Jiaotong University. Cell lines were cultivated in DMEM/F12 medium containing 10% FBS supplement (vol%), 1% Penicillin-Streptomycin solution and the culture medium was changed every 2-5 days.

| RNA isolation and quantitative real-time PCR
mRNA was isolated by Trizol (Thermo scientific) according to the manufacturer's protocol. cDNA was synthesized by using iScript

| Cell viability assay
Viability of tumor cells was determined using AlamarBlue reagent (Thermo scientific). Cells were seeded at 1000 cells per well in a 96well plates, after indicated period of time AlamarBlue reagent was added into each well and 6 hours later fluorescence was measured (Excitation 515-565 nm, Emission 570-610 nm) using Synergy HTX multi-mode reader (BioTek).

| In vivo bioluminescent imaging
GBM tumor cells were transduced with lentiviral particles (pHAGE PGK-GFP-IRES-LUC-W) for coexpression of GFP and luciferase.
Animals were administrated intraperitoneally with 2.5 mg/100 µL solution of XenoLight D-luciferin (PerkinElmer) and anesthetized with isoflurane for the imaging analysis. The luciferase images were captured by using an IVIS 100 imaging system (PerkinElmer).

| Statistical analysis
All data are presented as mean ± SD. Statistical differences between two groups were evaluated by two-tailed t test. The comparison

| mRNA expression dynamics of E2Fs in HGG
To determine the expression profile of E2Fs in HGG, we performed data mining and analyzed the publicly available dataset, TCGA dataset. 25 A total 394 cases ( We initially analyzed the mRNA expression profile of E2Fs. As shown in Figure 1A, all E2Fs were generally determined in HGG samples. Compared with LGG, except for E2F3 and E2F5, all E2Fs were highly expressed in HGG samples and correlated with grade progression, with highest expression attributed to glioblastoma (GBM, grade IV) ( Figure 1B). To verify our finding, we referred to another dataset, Gravendeel dataset, 26 which contains 276 cases with corresponding pathology information. The result was similar with TCGA dataset and showed that all E2Fs but E2F5 were elevated in HGG ( Figure S1).

| Correlation analysis of E2Fs in HGG with critical oncogenes
As E2Fs were reported to participate in the cell cycle of tumor cells, we next analyzed the association between E2Fs and well- The unsupervised clustering analysis showed that the correlation signature of E2F1, E2F2, E2F7, and E2F8 were quite similar with each other ( Figure S2). Notably, the result demonstrated that E2F1, E2F2, E2F7, and E2F8 were positively correlated with cell cycle and proliferation markers ( Figure 2A). However, none of E2Fs were identified to be highly correlated with stemness associated genes. Interesting, we found that E2F1, E2F2, E2F7, and E2F8 were tightly associated with therapeutic resistance genes, including EZH2 ( Figure 2B), CHEK1, and AUKRA. In addition, we found E2F7 and E2F8 demonstrated much higher correlation with VIM, a crucial epithelial-mesenchymal transition (EMT) marker in tumor cells ( Figure 2C). All results in this aspect were shown in Table 2.
Collectively, the result revealed that higher expression of E2Fs might be responsible for aggressive proliferation and therapeutic resistance of HGG tumor cells.  Given the bigger median survival difference and higher AUC value of E2F7 and E2F8, we further explore their link with clinical signature in HGG. As shown in Figure   For bold values, the statistically significant P values are P < .001.

| Survival outcome analysis for HGG patients
promoter (P < .001). Interestingly, elevated expression of E2F7 was shown to be associated with older age of HGG patients (P < .001).
The analysis revealed similar result for E2F8 ( Figure S3B). Next, we utilized logistic regression to determine the association of E2F7 or E2F8 expression with prognostic clinicopathologic variables.
However, no significant differences were identified on sex subgroups. These results showed that E2F7 and E2F8 were closely associated with clinicopathologic features of HGG.
Lastly, independent prognostic factors were investigated via cox proportional hazard regression models ( For bold values, the statistically significant P values are P < .001. P = .041) and E2F8 (HR = 1.886; 95% CI, 1.049-3.390; P = .0340) were significant and independent predictors for poor outcome in HGG patients (Table 6). In addition, the multivariate cox regression analyses also revealed that IDH mutation status and patient age were significant and independent predictors. Collectively, the survival analysis revealed the novel role of E2F7 and E2F8 in HGG.

| Bioinformatics analysis of oncogenic role of E2F7 and E2F8 in HGG
Given the better performance of E2F7 and E2F8 in predicting patient outcome, we further explored the oncogenic role of E2F7  (Table S2). As shown in Figure 4, significant differences (normalized P < .05, FDR < 0.25) were observed in the enrichment result. Ranked as the top one pathway, epithelial-mesenchymal transition (EMT) was found to be highly enriched in E2F7 or E2F8 high-expression group in HGG ( Figure 4A,B), a malignant phenotype transition that confers higher radioresistance to tumor cells. 39 Besides, E2F7 and E2F8 were significantly involved with a variety of signaling pathways promoting tumor initiation and progression in HGG, evidenced by the enrichment of NFκB, STAT3, angiogenesis, hypoxia, and glycolysis pathways in high-expression phenotype ( Figure 4A,B).

| E2F7 promotes radioresistance of HGG tumor cells
Based on the high correlation with multiple oncogenes (Figure 2) and the enriched oncogenic pathways (Figure 4), we posited that E2F7 or E2F8 might be linked with irradiation therapeutic resistance in HGG, which has been rarely investigated before. Thus, the patients in TCGA dataset underwent clinical treatment (IR or temozolomide (TMZ)) were sorted out and enrolled in the Kaplan-Meier analysis (subgrouped by median expression of E2F7 or E2F8). The result showed that higher expression of E2F7 was closely associated with worse response to IR treatment ( Figure 5A,B). However, the expression of E2F7 and E2F8 did not show significant link with TMZ treatment response ( Figure S4A,B). In addition, we found that the expression of E2F7 remarkablely increased after IR treatment ( Figure S4C). To further confirm our finding, we utilized

| E2F7 is linked with aggressive oncogenic process and transcriptionally regulates expression of EZH2
To determine the potential underline mechanism for the role of E2F7 in promoting radioresistance, we performed additional bioinformatics analysis. Differential expressed genes (DEGs, 2729 genes in total) in TCGA dataset were first identified and selected oncogenes were plotted with heatmap ( Figure 6A). Gene ontology analysis showed that E2F7 was highly associated with aggressive cell proliferation, EMT signature (cell migration, adhesion, and motility), and DNA repair ( Figure 6B). KEGG analysis further highlighted the critical role  Figure 6C). EZH2 has been previously identified as a crucial regulator factor in promoting radioresistance of HGG, 34 we posited that E2F7 might regulate the transcription activity of EZH2 and promote radioresistance of tumor cells. Via qRT-PCR, we found that silencing of E2F7 induced significant reduction of EZH2 mRNA expression in U87 and U373 tumor cells ( Figure 6D). ChIP-sequencing data from ENCODE (Encyclopedia of DNA Elements) revealed clear binding peaks for E2F7 at promoter area of EZH2 in K562 tumor cells ( Figure 6E). Finally, ChIP-PCR was performed to determine the occupancy of E2F7 at promoter area of EZH2 and the result showed significant binding pattern of E2F7, compared with IgG ( Figure 6F). Collectively, these data suggest E2F7 is involved with multiple oncogenic process and a potentially key regulator of EZH2 transcriptional activity in HGG tumor cells.

| D ISCUSS I ON
Glioma accounts for the vast majority of adult malignant brain tumors, highly infiltrative pattern, aggressive proliferation, therapeutic resistance, and fast recurrence speed as hallmarks.  Moreover, due to the independence of dimerization proteins, E2F7 serves as a recruitment factor and affects the target gene expression. However, all the above hypotheses warrant further functional validation.
In conclusion, this study systemically analyzed the expression pattern and overall outcome of HGG patients. E2F7 and E2F8 were identified as novel hub genes for aggressiveness of HGG. Pathway alternations highlighted the irradiation protection role of E2F7 and E2F8 in HGG. E2F7 promoted radioresistance of HGG via multiple oncogenic process, including transcriptional regulation of EZH2. Our findings might provide potential experimental evidence for future clinical treatment of HGG.

ACK N OWLED G M ENT
This study was supported by the National Natural Science Foundation of China (no. 81802502).

CO N FLI C T O F I NTE R E S T
No conflicts of interest exists in the submission of the manuscript and the manuscript is approved by all authors for publication.