The Regulation and Evolution of long, noncoding RNA Transcription

Abstract

Advances in genomics techniques, such as high-resolution tiled microarrays and deep sequencing, have revealed that transcription is highly pervasive. This suggests that transcription occurs not only at protein coding genes (open reading frames, or ORFs), but also at unannotated portions of the genome, leading to the production of thousands of long, noncoding RNAs (lncRNAs, > 200 basepairs), which are distinct from more traditional noncoding RNAs such as transfer RNAs or ribosomal RNAs. This striking finding is observed across all eukaryotes, from humans to budding yeast. It was once thought that these lncRNAs were largely the product of stochastic transcription. However, through detailed functional analyses, it is clear that some lncRNAs play key regulatory roles. For instance Xist is a lncRNA necessary for X-chromosome inactivation in higher eukaryotes. In budding yeast, lncRNAs produced at the GAL10 and PHO84 locus are needed for attenuating transcription of the overlapping mRNA. Though these examples illustrate that we understand the functions of some lncRNAs, the vast majority of lncRNAs have as-of-yet unidentified biological roles. Moreover, even less is known about how lncRNAs are regulated or the conditions underlying their evolution. This dissertation describes the use of budding yeast as a model to elucidate fundamental principles of lncRNA regulation and evolution. I first describe the development of a high-throughput genetic screen in the budding yeast, Saccharomyces cerevisiae, that takes advantage of synthetic growth defects of mutants when RNA interference is reconstituted to identify 408 putative repressors of lncRNAs. Among these putative hits were four highly-conserved chromatin remodeling factors: Swr1, Isw2, Rsc, Ino80. I then use strand-specific, high-throughput RNA sequencing (ssRNAseq) to identify the lncRNA targets of these complexes. Further, I go on to show that these factors largely regulate distinct populations of lncRNAs genome-wide, that a subset of these lncRNAs are directly regulated by these remodeling factors (termed chromatin remodeling regulated antisense transcripts, CRRATs), and that some CRRATs might function to regulate the mRNA that they overlap with. Next, I describe the use of comparative genomics in five species of budding yeast to show that, since the loss of RNAi in the budding yeast lineage, levels of antisense, long noncoding RNAs (ASlncRNAs) have gradually risen genome-wide. I then identified a subset of ASlncRNAs that are highly conserved at the level of expression among the sensu stricto lineage of budding yeasts, and I assign putative biological roles for these ASlncRNAs using gene ontology. Finally, I show, using S. cerevisiae and S. castellii as models, that RNAi likely attenuates ASlncRNAs across the genome. Using genetic data, I suggest that this is likely due to deleterious effects when ASlncRNAs are elevated in the presence of an active RNAi pathway. lncRNAs have been shown to play key regulatory roles across many different eukaryotes. Understanding how lncRNAs are regulated and how they might have evolved will provide insights into their biological functions and their roles in disease states such as cancer.