基于整合生物信息学分析识别骨肉瘤干性相关基因

Identification of stemness-related genes in osteosarcoma based on integrated bioinformatics analysis

关键词:  骨肉瘤  癌症干细胞  干细胞指数  单细胞RNA测序

Key words:  osteosarcoma  cancer stem cells  stemness index  single cell RNA sequencing

摘要:

目的 本研究通过整合生物信息学分析骨肉瘤（osteosarcoma， OS）干性相关基因（stemness-related genes， SRGs）及预后价值。方法 从TARGET数据库中获取OS转录组测序数据。采用“edgeR”包对基因表达谱进行标准化。基于一类逻辑回归机器学习算法计算干性指数（mRNA expression-based stemness index， mRNAsi）。采用皮尔逊相关性分析识别SRGs。采用“survival”包分析SRGs的预后价值。从GEO（Gene Expression Omnibus）数据库下载数据集GSE152048，利用主成分分析、降维和细胞类型注释来分析OS细胞的分子特征及异质性，并通过“iTALK”包预测OS细胞间通信。结果 OS样本间mRNAsi有显著差异，死亡样本显著高于存活样本；转移样本显著高于无转移样本。通过对皮尔逊相关性分析结果（5 131个SRGs）、Cox回归分析结果（1 640个风险相关基因）和K-M生存分析结果（7 066个生存相关基因）取交集，共识别了658个关键基因。NDUFB9与mRNAsi的正相关性最强，EHD2与mRNAsi的负相关性最强。通过降维分析将OS肿瘤细胞划分为10个细胞亚群。NDUFB9在有干性特征的细胞亚群中高表达，包括间充质干细胞和成骨细胞等；而EHD2在各肿瘤细胞亚群中的表达量极低或不表达。结论 本研究刻画了OS干性特征和单细胞转录组图谱，识别了预后相关SRGs，为揭示OS干性特征调控机制及发掘潜在治疗靶点提供了理论依据。

Abstract:

Objective To identify stemness-related genes(SRGs) in osteosarcoma(OS) by bioinformatics analysis and to explore their clinical implication. Methods RNA sequencing(RNA-seq) data were obtained from TARGET database, and normalized using “edgeR” package. The mRNA expression-based stemness index(mRNAsi) was obtained by using a one-class logistic regression machine learning algorithm based on RNA-seq data of OS patients. SRGs were identified with Pearson correlation analysis. GSE152048 dataset were downloaded from the Gene Expression Omnibus(GEO) database, which were used to explore and tumor heterogeneity of OS and to validate the expression of key genes. The potential cellular communication patterns and cell cycle status were predicted. Results The mRNAsi was significantly different among OS samples. Specifically, mRNAsi of the death group was significantly higher than the survival group; mRNAsi of the metastatic group was significantly higher than the non-metastatic group. By taking the intersection of the results of Cox regression analysis(1 640 risk-related genes), Kaplan-Meier survival analysis(7 066 survival-related genes), and Pearson correlation analysis(5 131 SRGs), finally 658 key genes were identified. NDUFB9 showed the strongest positive correlation with mRNAsi, while EHD2 showed the strongest negative correlation with mRNAsi.OS cells were divided into 10 cell subsets by dimension reduction analysis. NDUFB9 was highly expressed in stem cell-like cell subsets, including mesenchymal stem cells and osteoblasts, whereas the expression of EHD2 in OS cell subsets was extremely low. Conclusion This study investigated the stemness features and single cell transcriptome landscape of OS, and identified prognostic SRGs, which provided information for revealing the regulatory mechanism of OS stemness and potential therapeutic targets for OS.