Co-localized genomic regulation of miRNA and mRNA via DNA methylation affects survival in multiple tumor types
Co-localized genomic regulation of miRNA and mRNA via DNA methylation affects survival in multiple tumor types
复制标题
DOI：
10.1016/j.cancergen.2016.09.001
复制
发表时间：
2016-10-01
期刊：
影响因子：
1.9
通讯作者：
Baggerly, Keith
中科院分区：
文献类型：
--
作者：
Doecke, James D.;Wang, Ying;Baggerly, Keith
癌症中基因表达异常的部分原因是 DNA 甲基化和 miRNA 靶基因下调的不规则模式。利用癌症基因组图谱 (TGCA) 的数据，我们研究了 15 种癌症类型的共定位 mRNA、miRNA 和 DNA 甲基化数据，重点关注直接调控证据有力的案例。为了限制对 miRNA 标记在相应 mRNA 转录本中共定位的区域的关注，我们检查了 2839 个样本中 354 个 mRNA、389 个 miRNA 和 13,809 个 DNA 甲基化探针的表达数据，以确定相关性是否大于绝对值 0.6。我们鉴定了 32 个基因、34 个 miRNA 和 143 个 DNA 甲基化位点探针，其中包含 180 个“三联体组合”，它们共同提供了癌症中共定位基因组调控的证据。在 HOXC5、PDE2A、SH3TC2 和 TP63 四个基因中发现了在不同组织类型中表现出最高患病率的五种三联体组合。在总共 32 个基因中，两种肿瘤类型（肾透明细胞癌 (KIRC, 4) 和低级别胶质瘤 (LGG, 4)）中有 8 个基因与生存时间显着相关 (p < 0.002)。总之，本文提供的数据为我们的主要假设提供了证据，即与甲基化水平密切相关的基因和 miRNA 更有可能与癌症结果相关。
Aberrant gene expression in cancer is due in part to irregular patterns of DNA methylation and miRNA target gene down regulation. Using data from The Cancer Genome Atlas (TGCA), we investigated co-localized mRNA, miRNA and DNA methylation data across 15 cancer types, focusing on cases where evidence for direct regulation was strong. Restricting attention to regions where miRNA markers co-localize within a corresponding mRNA transcript, we checked expression data from 2839 samples for 354 mRNAs, 389 miRNAs and 13,809 DNA methylation probes for correlations greater than an absolute 0.6. We identified 32 genes, 34 miRNAs and 143 DNA methylation site probes comprising 180 "triplet combinations" that together provide evidence of co-localized genomic regulation in cancer. The five triplet combinations showing the highest prevalence across tissue types were found in four genes, HOXC5, PDE2A, SH3TC2 and TP63. Of the total 32 genes, eight among two tumor types (Kidney Renal Clear Cell Carcinoma (KIRC, 4) and Low Grade Glioma (LGG, 4)) were significantly associated with survival time (p < 0.002). Together, the data presented in this paper provide evidence toward our primary hypothesis, that both genes and miRNAs strongly correlated with methylation level are more likely to be associated with cancer outcomes.