Huang Runzhi, Li Zhenyu, Li Chen, Wang Guanghua, Yan Penghui, Peng Li, Wang Jiaqi, Zhu Xiaolong, Hu Peng, Zhang Junfang, Chang Zhengyan, Huang Zongqiang, Cheng Liming, Zhang Jie
Shanghai First Maternity and Infant Hospital, Tongji University School of Medicine, Shanghai, China.
Division of Spine, Department of Orthopedics, Tongji Hospital affiliated to Tongji University School of Medicine, Shanghai, China.
Front Genet. 2020 Feb 4;10:1380. doi: 10.3389/fgene.2019.01380. eCollection 2019.
Alternative splicing (AS) modifies 92-94% human genes, abnormal splicing events might relate to tumor development and invasion. Glioblastoma Multiforme (GBM) is a fatal, invasive, and malignant tumor in nervous system. The recurrence and development leads to poor prognosis. However, few studies have focused on AS in GBM.
RNA-seq and Alternative splicing events (ASEs) data of GBM samples were downloaded from The Cancer Genome Atlas (TCGA) and TCGASpliceSeq databases, respectively. Firstly, the Cox regression analysis was utilized to identify the overall survival splicing events (OS-SEs). Secondly, a multivariable model was applied to access the prognostic value of risk score. Then, we constructed a co-expressed network between splicing factors (SFs) and overall survival alternative splicing events (OS-SEs). Additionally, to explore the relationship between the potential prognostic signaling pathways and OS-SEs, we constructed a network between these pathways and OS-SEs. Ultimately, to better explain the results, validations from multi-dimension platforms were applied.
In the first step, 1,062 OS-SEs were selected by Cox regression. Then, 11 OS-SEs were integrated in a multivariate model by Lasso regression. The area under the curve (AUC) of receiver operator characteristic (ROC) curve was 0.861. In addition, the risk score generated from the multivariate model was confirmed to be an independent prognostic factor (P < 0.001). What's more, in the network of SFs and ASEs, CELF5 significantly regulated GSG1L|35696|AP and GSG1L|35698|AP (P < 0.001, R = 0.511 and = -0.492). Additionally, GSG1L|35696|AP (P = 0.006) and GSG1L|35698|AP (P = 0.007) showed a significant relationship with cancer status. Eventually, KEGG pathways related to prognosis of GBM were selected by GSVA. The primary bile acid synthesis (P < 0.001, R = 0.420) was the significant pathway co-expressed with Germ Cell-Specific Gene 1-Like Protein (GSG1L).
Based on the comprehensive bioinformatics analysis, we proposed that aberrant splicing factor CUGBP Elav-like family member 5 (CELF5) significantly, positively and negatively, regulated ASE of GSG1L, and the primary bile acid synthesis pathway might play an important role in tumorigenesis and prognosis of GBM.
可变剪接（AS）可修饰92%-94%的人类基因，异常剪接事件可能与肿瘤的发生发展及侵袭有关。多形性胶质母细胞瘤（GBM）是神经系统中一种致命、侵袭性强的恶性肿瘤。其复发和发展导致预后不良。然而，很少有研究关注GBM中的AS。
分别从癌症基因组图谱（TCGA）和TCGASpliceSeq数据库下载GBM样本的RNA测序和可变剪接事件（ASE）数据。首先，利用Cox回归分析来识别总生存剪接事件（OS-SEs）。其次，应用多变量模型评估风险评分的预后价值。然后，我们构建了剪接因子（SFs）与总生存可变剪接事件（OS-SEs）之间的共表达网络。此外，为了探究潜在的预后信号通路与OS-SEs之间的关系，我们构建了这些通路与OS-SEs之间的网络。最后，为了更好地解释结果，采用了多维度平台进行验证。
第一步，通过Cox回归选择了1062个OS-SEs。然后，通过Lasso回归将11个OS-SEs整合到多变量模型中。受试者工作特征（ROC）曲线的曲线下面积（AUC）为0.861。此外，多变量模型生成的风险评分被证实是一个独立的预后因素（P<0.001）。而且，在SFs与ASEs的网络中，CELF5对GSG1L|35696|AP和GSG1L|35698|AP有显著调控作用（P<0.001，R=0.511和=-0.492）。此外，GSG1L|35696|AP（P=0.006）和GSG1L|35698|AP（P=0.007）与癌症状态显示出显著关系。最终，通过基因集变异分析（GSVA）选择了与GBM预后相关的KEGG通路。初级胆汁酸合成（P<0.001，R=0.420）是与生殖细胞特异性基因1样蛋白（GSG1L）共表达的显著通路。
基于综合生物信息学分析，我们提出异常剪接因子CUGBP Elav样家族成员5（CELF5）对GSG1L的ASE有显著的正向和负向调控作用，并且初级胆汁酸合成途径可能在GBM的肿瘤发生和预后中起重要作用。
