Nojima Masanori, Matsui Takeshi, Tamori Akihiro, Kubo Shoji, Shirabe Ken, Kimura Koichi, Shimada Mitsuo, Utsunomiya Tohru, Kondo Yasuteru, Iio Etsuko, Naito Yutaka, Ochiya Takahiro, Tanaka Yasuhito
Institute of Medical Science Hospital, Center for Translational Research, the University of Tokyo, Tokyo, Japan.
Department of Gastroenterology, Teine Keijinkai Hospital, Sapporo, Japan.
Mol Cancer. 2016 Apr 30;15(1):31. doi: 10.1186/s12943-016-0514-6.
While hepatitis B and C viral infection have been suppressed, non-B non-C hepatocellular carcinoma (NBNC-HCC) is considered to be rising in incidence terms in some developed countries where prevalence of those viral infections among HCC patients had been very high (such as Japan, Korea, and Italy). To elucidate critical molecular changes in NBNC-HCC, we integrated three large datasets relating to comprehensive array-based analysis of genome-wide DNA methylation (N = 43 pairs) and mRNA/miRNA expression (N = 15, and 24 pairs, respectively) via statistical modeling.
Hierarchical clustering of DNA methylation in miRNA coding regions clearly distinguished NBNC-HCC tissue samples from relevant background tissues, revealing a remarkable tumor-specific hypomethylation cluster. In addition, miRNA clusters were extremely hypomethylated in tumor samples (median methylation change for non-clustered miRNAs: -2.3%, clustered miRNAs: -24.6%). The proportion of CpGs hypomethylated in more than 90% of the samples was 55.9% of all CpGs within miRNA clusters, and the peak methylation level was drastically shifted from 84% to 39%. Following statistical adjustment, the difference in methylation levels within miRNA coding regions was positively associated with their expression change. Receiver operating characteristic (ROC) analysis revealed a great discriminatory ability in respect to cluster-miRNA methylation. Moreover, miRNA methylation change was negatively correlated with corresponding target gene expression amongst conserved and highly matched miRNA sites.
We observed a drastic negative shift of methylation levels in miRNA cluster regions. Changes in methylation status of miRNAs were more indicative of target gene expression and pathological diagnosis than respective miRNA expression changes, suggesting the importance of genome-wide miRNA methylation for tumor development. Our study dynamically summarized global miRNA hypomethylation and its genome-wide scale consequence in NBNC-HCC.
尽管乙型和丙型肝炎病毒感染已得到抑制,但在一些发达国家,非B非C型肝细胞癌(NBNC-HCC)的发病率仍被认为在上升,这些国家的肝细胞癌患者中这些病毒感染的患病率曾经非常高(如日本、韩国和意大利)。为了阐明NBNC-HCC中的关键分子变化,我们通过统计建模整合了三个大型数据集,这些数据集分别涉及基于芯片的全基因组DNA甲基化综合分析(N = 43对)以及mRNA/miRNA表达分析(分别为N = 15对和24对)。
miRNA编码区域DNA甲基化的层次聚类清楚地将NBNC-HCC组织样本与相关背景组织区分开来,揭示了一个显著的肿瘤特异性低甲基化簇。此外,miRNA簇在肿瘤样本中极度低甲基化(非聚类miRNA的甲基化变化中位数:-2.3%,聚类miRNA:-24.6%)。在超过90%的样本中低甲基化的CpG比例占miRNA簇内所有CpG的55.9%,并且甲基化峰值水平从84%急剧转变为39%。经过统计调整后,miRNA编码区域内的甲基化水平差异与它们的表达变化呈正相关。受试者工作特征(ROC)分析显示聚类miRNA甲基化具有很强的鉴别能力。此外,在保守且高度匹配的miRNA位点中,miRNA甲基化变化与相应靶基因表达呈负相关。
我们观察到miRNA簇区域甲基化水平出现急剧的负向变化。miRNA甲基化状态的变化比各自的miRNA表达变化更能指示靶基因表达和病理诊断,这表明全基因组miRNA甲基化对肿瘤发展具有重要意义。我们的研究动态总结了NBNC-HCC中全局miRNA低甲基化及其全基因组规模的后果。