Long non-coding RNAs as pan-cancer master gene regulators of associated protein-coding genes: a systems biology approach

Abstract

Despite years of research, we are still unraveling crucial stages of gene expression regulation in cancer. On the basis of major biological hallmarks, we hypothesized that there must be a uniform gene expression pattern and regulation across cancer types. Among non-coding genes, long non-coding RNAs (lncRNAs) are emerging as key gene regulators playing powerful roles in cancer. Using TCGA RNAseq data, we analyzed coding (mRNA) and non-coding (lncRNA) gene expression across 15 and 9 common cancer types, respectively. 70 significantly differentially expressed genes common to all 15 cancer types were enlisted. Correlating with protein expression levels from Human Protein Atlas, we observed 34 positively correlated gene sets which are enriched in gene expression, transcription from RNA Pol-II, regulation of transcription and mitotic cell cycle biological processes. Further, 24 lncRNAs were among common significantly differentially expressed non-coding genes. Using guilt-by-association method, we predicted lncRNAs to be involved in same biological processes. Combining RNA-RNA interaction prediction and transcription regulatory networks, we identified E2F1, FOXM1 and <i>PVT1</i> regulatory path as recurring pan-cancer regulatory entity. <i>PVT1</i> is predicted to interact with <i>SYNE1</i> at 3'-UTR; <i>DNAJC9, RNPS1</i> at 5'-UTR and <i>ATXN2L, ALAD, FOXM1</i> and <i>IRAK1</i> at CDS sites. The key findings are that through E2F1, FOXM1 and <i>PVT1</i> regulatory axis and possible interactions with different coding genes, <i>PVT1</i> may be playing a prominent role in pan-cancer development and progression.