Discovery and Investigation of Long Non-Coding RNAs in Cancer

Abstract

Cancer is a leading cause of death worldwide, and its etiology and mechanistic underpinnings have been the target of extensive investigation. As a disease resulting from genetic aberrations, extensive effort has been put forth to understand the genomic landscape of cancer. Recently, large multi-institutional consortia have genomically profiled thousands of tumors from a myriad of different cancer types. Extensive investigation of the cancer transcriptome bears the potential to expand our understanding of the potential players in oncogenesis, and thus the goal of this dissertation was to leverage large-scale RNA-seq datasets to discovery novel transcripts, and to subsequently investigate the function of two lncRNA candidates.

In chapter two, we present the MiTranscriptome, the most comprehensive annotation of the human transcriptome to date. Through a large bioinformatics effort, we curated over 6,500 RNA-seq samples from a myriad of tumor types and tissues, and employed a meta-assembly approach to generate a consensus transcriptome we termed MiTranscriptome. Many of these lncRNAs exhibited cancer and tissue specific expression patterns, and these findings have been made available in an online resource (mitranscriptome.org).

In chapter three, we describe TACO, a novel tool designed to produce highly accurate and robust transcriptome meta-assemblies from large numbers of RNA-seq libraries. This tool dramatically outperforms its competitors in its ability to accurately recapitulate the splicing patterns and transcription start and end sites of meta-assembled transcripts. TACO enables generation of a robust consensus transcriptome from large quantities of RNA-seq samples, which is critical for the identification of novel transcripts from RNA-seq data.

In chapter four, we present a study interrogating the transcriptional landscape of breast cancer, specifically focusing on discovery of novel transcripts associated with estrogen receptor (ER) driven disease. We identify DSCAM-AS1 as an ER-regulated lncRNA, and investigate its role in cancer progression of ER-positive breast tumors, implicating it with the development of tamoxifen resistance.

Finally, in chapter five, we present a study functionally investigating the lncRNA THOR, which has a cancer/testis expression pattern, and is also highly conserved with analogous isoforms in the mouse and zebrafish. We show that THOR functions through an interaction with the protein IGF2BP1, and generate zebrafish animal models to show that THOR plays a critical role in fertility and oncogenesis.