Aberrant DNA methylation defines isoform usage in cancer, with functional implications

• Published in: PLoS computational biology 2019-07, Vol.15 (7), p.e1007095-e1007095
• Main Authors: Chen, Yun-Ching, Elnitski, Laura
• Format: Article
• Language: English
• Subjects: Alternative Splicing; Binding sites; Bioinformatics; Biology and life sciences; Cancer; Cancer genetics; Cell growth; Computational Biology; Correlation; CpG Islands; Deoxyribonucleic acid; DNA; DNA Methylation; Endometrial cancer; Female; Gene expression; Gene Expression Profiling; Gene Expression Regulation, Neoplastic; Genes; Genetic aspects; Genomes; Genomics; Head & neck cancer; Health aspects; Humans; Isoforms; Male; Medicine and Health Sciences; Methylation; Motor vehicle drivers; Mutation; Neoplasm Proteins - genetics; Neoplasm Proteins - metabolism; Neoplasms - classification; Neoplasms - genetics; Neoplasms - metabolism; Protein Isoforms - genetics; Protein Isoforms - metabolism; Proteins; Ribonucleic acid; RNA; RNA polymerase; RNA, Neoplasm - genetics; RNA, Neoplasm - metabolism; Sequence Analysis, RNA; Splicing; Splicing factors; Squamous cell carcinoma; Suppressors; Transcription; Transcription (Genetics); Transcription Initiation Site; Transcription Termination, Genetic; Tumorigenesis; Tumors
• ISSN: 1553-7358; 1553-734X; 1553-7358
• DOI: 10.1371/journal.pcbi.1007095

Alternative transcript isoforms are common in tumors and act as potential drivers of cancer. Mechanisms determining altered isoform expression include somatic mutations in splice regulatory sites or altered splicing factors. However, since DNA methylation is known to regulate transcriptional isoform activity in normal cells, we predicted the highly dysregulated patterns of DNA methylation present in cancer also affect isoform activity. We analyzed DNA methylation and RNA-seq isoform data from 18 human cancer types and found frequent correlations specifically within 11 cancer types. Examining the top 25% of variable methylation sites revealed that the location of the methylated CpG site in a gene determined which isoform was used. In addition, the correlated methylation-isoform patterns classified tumors into known subtypes and predicted distinct protein functions between tumor subtypes. Finally, methylation-correlated isoforms were enriched for oncogenes, tumor suppressors, and cancer-related pathways. These findings provide new insights into the functional impact of dysregulated DNA methylation in cancer and highlight the relationship between the epigenome and transcriptome.