E2Fs were highly expressed in DLBCL in the mRNA level
To call DEGs of E2Fs family in DLBCL, we used the GEPIA dataset to compare the mRNA expression of DLBCL patient samples from TCGA tumors (n=47) and GTEx healthy tissues (n=337). We used log2 (TPM + 1) to quantity mRNA expression data for E2Fs and found that E2F1/4/5/6/8 were highly-expressed in DLBCL, while E2F2/3/7 were not significantly different. (Fig.1)
Expression of E2Fs translation factors in lymphoma cell lines
After determining the expression of E2Fs in clinical lymphoma patients, we analyzed E2Fs’ mRNA expression in cell lines through the CCLE and EMBL-EBI database. By assembling CCLE, we found that E2F1/2/4/5/8 were relatively highly expressed in DLBCL, comparing the average expression base line of all tumors (FigS1). Moreover, EMBL-EBI was employed to test mRNA level of E2F translation factors in DLBCL cell lines(n=21). E2F1/2/3/4/5/8 increased in most DLBCL cell lines (Fig.2), especially in SU-DHL-5, SU-DHL-10, WSU-NHL. Collectively, with the aim to further elucidate the major factors of E2Fs in DLCBL, three databases were searched for intersection of the predicted results. We found that E2F1/4/5/8 overexpress in the GEPIA, CCLE and EMBL-EBI. (Fig. 3)
Prognostic Values of E2F1/4/5/8 in Patients With DLBCL
Next, we initially assessed the E2F1/4/5/8 prognosis values in 471 DLBCL patients (GSE31312). It was determined that the higher expression of E2F1 mRNA predicts extended OS and PFS((p=0.0015,p=0.02, respectively). Elevated E2F5/8 were significantly correlated with shortened OS (p=0.0051,0.015, respectively) and PFS (p=0.019,0.15, respectively). (Fig.4). Next, multivariate Cox’s hazard regression analysis results indicated E2F1 (HR = 1.45, 95%CI = 1.141–1.841, P = 0.002), E2F5 (HR = 0.74, 95% CI = 0.584-0.938, P = 0.013), E2F8 (HR = 1.607, 95%CI = 1.103-2.343, P = 0.014) expression level are independent prognostic factors for overall survival in DLBCL patients (Table1).
Functional enrichment analysis of E2F1/4/5/8 and co-expression genes in patients with DLBCL
To reveal the function and potential mechanism of E2F1/4/5/8, we constructed a network of E2F1/4/5/8 and their neighboring genes (TFDP1/2, RBL1/2, E2F2) by the GeneMANIA database(Fig5A). By analyzing Metascape, we found the E2F1/4/5/8 and their neighboring genes were mainly enriched in G1/S cell cycle. These genes may form protein-DNA complex.Then they enhanced the interaction between RNA polymerase and specific promoters to promote targeted gene expression by changing the structure of DNA.(Fig.5B and Table2).
The top KEGG pathways showed that the cell cycle signaling pathway and TGF-beta signaling pathway were significantly found to be involved in the development of multiple tumor and participated in the tumorigenesis and pathogenesis of lymphoma.(Fig.5C and Table 3.) In addition, to further understand the role of E2Fs in lymphoma, we conducted a relevant protein–protein interaction (PPI) analysis(Fig.5D). The results showed that E2F2/3/4/5, RBL1/2, TFDP1/2, CDK2 mainly formed transcription factor complex, regulated the maligant proliferation of tumor through G1/S transition of mitotic cell cycle and TGF-beta signaling pathway.
E2Fs expression is negatively regulated by DNA methylation
In diffuse large B-cell lymphoma, DNA methylation is the most widely studied epigenetic process leading to chemotherapy resistance21. In order to explore the transcriptional regulation of E2Fs family in DLBCL, we analyzed the 450 K methylation array data using SMARTAPP Databases to verify whether E2Fs expression may be regulated by their DNA methylation status. By comparing the Correlation between gene expression and methylation in 47 DLBCL patients (TCGA), The method for Correlation Coefficient is Spearman, we found that E2F1/3/5 expression is negative with methylation sites (p=0.00013,0.026,0.0011, respectively) (Fig.6)
Gene–gene interaction network of the E2F1/3/5 with the DNA methylation associated genes
Above, we found that E2F1/3/5 were hypomethylated in the TCGA databases. Previous studies indicated that human DNA methylation is catalyzed by DNA transmethylase including DNMT1/3A/3B 22. Therefore, we used the GeneMANIA database to predict the interaction relationship of the E2F1/3/5 and DNMT1/3A/3B (Fig.7). The 6 central nodes representing E2F1/3/5, DNMT1/3A/3B were surrounded by 20 nodes representing genes,which have similar functions(Fig.7A). E2F2/4, TFDP1/2 was correlated with central nodes, participating in the same reaction within a pathway. (Fig.7B). E2F2/4/6, TFDP1/2, TRDMT1, DNMT3L, DMAP1, ZBTB18 was correlated with central nodes, predicting functional relationships between protein interactions(Fig.7C). As for co-localization, E2F4 was correlated with E2F3/5, DNMT1(Fig.7D). In addition, HDGF, TFDP1/2 was correlated with E2F1/3, DNMT1 in terms of co-expression(Fig.7E).
In GEPIA Dataset, E2F1, E2F4, E2F5, E2F6, E2F8, TFDP1, HDGF and DNMT1 expression is higher in DLBCL patients (Fig.1, FigS2). Next, we found that the methylation of HDGF, PWWP2A, TFDP1, TRDMT1, E2F1, E2F3, E2F5 and DNMT1 were significantly negative with gene expression in SMART APP Databases. In conclusion, elevated E2F1, E2F5, TFDP1, DNMT1 and HDGF were negative with DNA methylation in DLBCL patients (Fig6,FigS3). In GSE31312, correlation analysis showed that E2F5 is positive correlation with TFDP1(R=0.18, P<0.001) and HDGF (R=0.13, P<0.01), DNMT1 is negative correlation with TFDP1(R=-0.16, P<0.001) and HDGF (R=-0.16, P<0.001) (Fig.8) in DLBCL patients.