Abstract
It is estimated that the rate of epigenetic changes may be orders of magnitude higher than that of genetic changes and that purely epigenetic mechanisms may explain why cancers arise with few or no recurrent mutations. However, supporting evidence remains limited, partly due to the cost of experimentally studying genome-wide epigenetic dysregulation. Since genome modification enzymes are recruited by long noncoding RNAs (lncRNAs) to specific genomic sites, analyzing differentially expressed genes and differentially methylated regions (DMRs) at the DNA binding sites of differentially expressed lncRNAs is important for uncovering epigenetic dysregulation. We performed RNA-seq and MeDIP-seq on a set of colorectal cancer (CRC) and normal colon samples and developed an analysis pipeline for combined analyses of gene expression, DNA methylation, and lncRNA/DNA binding. The genes identified in our data and important for CRC agree with widely reported findings. We found that aberrantly transcribed noncoding transcripts may epigenetically dysregulate genes, that correlated gene expression is significantly determined by epigenetic dysregulation, that differentially expressed noncoding transcripts and their epigenetic targets form distinct modules in different cancer cells, and that many hub lncRNAs in these modules are primate-specific. These results suggest that lncRNA-mediated epigenetic dysregulation greatly determines aberrant gene expression and that epigenetic dysregulation is highly species-specific. The analysis pipeline can effectively unveil cancer- and cell-specific modules of epigenetic dysregulation, and such modules may provide novel clues for identifying diagnostic, therapeutic, and prognostic targets for epigenetic dysregulation.