Genome-wide methylation analysis identifies a core set of hypermethylated genes in CIMP-H colorectal cancer

BMC Cancer

Research Article

Tyler McInnes¹  Anthony E. Reeve¹  Parry J. Guilford¹  Michael A. Black¹  Donghui Zou¹  Dasari S. Rao²  Vicky L. Phillips³  Francesca M. Munro³  John L. McCall³ 

[1] Cancer Genetics Laboratory, Centre for Translational Cancer Research (Te Aho Matatū), Department of Biochemistry, University of Otago, 9054, Dunedin, New Zealand;Cancer Genetics Laboratory, Centre for Translational Cancer Research (Te Aho Matatū), Department of Biochemistry, University of Otago, 9054, Dunedin, New Zealand;Department of Surgical Sciences, Dunedin School of Medicine, University of Otago, 9054, Dunedin, New Zealand;Department of Surgical Sciences, Dunedin School of Medicine, University of Otago, 9054, Dunedin, New Zealand;

关键词: Epigenome;    Methylation;    Colorectal cancer;    CIMP;   

DOI  :  10.1186/s12885-017-3226-4

 received in 2016-06-14, accepted in 2017-03-23,  发布年份 2017

来源: Springer

【 摘 要 】

BackgroundAberrant DNA methylation profiles are a characteristic of all known cancer types, epitomized by the CpG island methylator phenotype (CIMP) in colorectal cancer (CRC). Hypermethylation has been observed at CpG islands throughout the genome, but it is unclear which factors determine whether an individual island becomes methylated in cancer.MethodsDNA methylation in CRC was analysed using the Illumina HumanMethylation450K array. Differentially methylated loci were identified using Significance Analysis of Microarrays (SAM) and the Wilcoxon Signed Rank (WSR) test. Unsupervised hierarchical clustering was used to identify methylation subtypes in CRC.ResultsIn this study we characterized the DNA methylation profiles of 94 CRC tissues and their matched normal counterparts. Consistent with previous studies, unsupervized hierarchical clustering of genome-wide methylation data identified three subtypes within the tumour samples, designated CIMP-H, CIMP-L and CIMP-N, that showed high, low and very low methylation levels, respectively. Differential methylation between normal and tumour samples was analysed at the individual CpG level, and at the gene level. The distribution of hypermethylation in CIMP-N tumours showed high inter-tumour variability and appeared to be highly stochastic in nature, whereas CIMP-H tumours exhibited consistent hypermethylation at a subset of genes, in addition to a highly variable background of hypermethylated genes. EYA4, TFPI2 and TLX1 were hypermethylated in more than 90% of all tumours examined. One-hundred thirty-two genes were hypermethylated in 100% of CIMP-H tumours studied and these were highly enriched for functions relating to skeletal system development (Bonferroni adjusted p value =2.88E-15), segment specification (adjusted p value =9.62E-11), embryonic development (adjusted p value =1.52E-04), mesoderm development (adjusted p value =1.14E-20), and ectoderm development (adjusted p value =7.94E-16).ConclusionsOur genome-wide characterization of DNA methylation in colorectal cancer has identified 132 genes hypermethylated in 100% of CIMP-H samples. Three genes, EYA4, TLX1 and TFPI2 are hypermethylated in >90% of all tumour samples, regardless of CIMP subtype.

【 授权许可】

CC BY   
© The Author(s). 2017