PLoS ONE (Jan 2012)

Integrated epigenome profiling of repressive histone modifications, DNA methylation and gene expression in normal and malignant urothelial cells.

  • Ewa Dudziec,
  • Andreas Gogol-Döring,
  • Victoria Cookson,
  • Wei Chen,
  • James Catto

DOI
https://doi.org/10.1371/journal.pone.0032750
Journal volume & issue
Vol. 7, no. 3
p. e32750

Abstract

Read online

Epigenetic regulation of gene expression is commonly altered in human cancer. We have observed alterations of DNA methylation and microRNA expression that reflect the biology of bladder cancer. This common disease arises by distinct pathways with low and high-grade differentiation. We hypothesized that epigenetic gene regulation reflects an interaction between histone and DNA modifications, and differences between normal and malignant urothelial cells represent carcinogenic events within bladder cancer. To test this we profiled two repressive histone modifications (H3K9m3 and H3K27m3) using ChIP-Seq, cytosine methylation using MeDIP and mRNA expression in normal and malignant urothelial cell lines. In genes with low expression we identified H3K27m3 and DNA methylation each in 20-30% of genes and both marks in 5% of genes. H3K9m3 was detected in 5-10% of genes but was not associated with overall expression. DNA methylation was more closely related to gene expression in malignant than normal cells. H3K27m3 was the epigenetic mark most specifically correlated to gene silencing. Our data suggest that urothelial carcinogenesis is accompanied by a loss of control of both DNA methylation and H3k27 methylation. From our observations we identified a panel of genes with cancer specific-epigenetic mediated aberrant expression including those with reported carcinogenic functions and members potentially mediating a positive epigenetic feedback loop. Pathway enrichment analysis revealed genes marked by H3K9m3 were involved with cell homeostasis, those marked by H3K27m3 mediated pro-carcinogenic processes and those marked with cytosine methylation were mixed in function. In 150 normal and malignant urothelial samples, our gene panel correctly estimated expression in 65% of its members. Hierarchical clustering revealed that this gene panel stratified samples according to the presence and phenotype of bladder cancer.