Transcript cleavage and polyadenylation in plants

Abstract

Eukaryotic gene expression is finely regulated at the post-transcriptional level by the untranslated regions of mRNA. The coding sequence (CDS) of mRNA is flanked by 5’- and 3’-untranslated regions (UTRs). The end boundary of the 3’ UTR is defined by transcript cleavage and polyadenylation. The genic region that determine where the cleavage and polyadenylation complex (CPMC) binds and cleaves is called the terminator. Terminators overlap significantly with 3’UTRs but also include the sequences after the 3’ UTR boundary. Elements in the resulting 3’ UTR modulate stability, nuclear export, localization, and translation. In this body of work, I will provide an overview of the historical exploration of terminator cleavage and polyadenylation, emphasizing the biotechnology that aided these discoveries. I will focus on how advances in DNA sequencing technologies expanded our understanding of terminator genetics and functionality across eukaryotes, with a particular emphasis on plants. Apart from transcriptome wide maps of cleavage and polyadenylation signaling, sequencing empowered functional genomics by enabling massively parallel reporter assays (MPRAs). These tools, in conjunction with computational machine learning, will allow the engineering of specific terminators for diverse applications in plant synthetic biology. Due to the limitations of plant systems, however, little work has been done to characterize plant terminator sequences on a genome wide basis for their strength in directing cleavage and fine tuning expression. Following upon recent developments optimizing massively parallel reporter assays in transient tobacco leaves and maize protoplasts, I characterized nearly all Arabidopsis thaliana and maize terminator sequences for their strength in conferring expression and cleavage. The resulting data helped train a deep learning model to predict terminator strength, aiding in the in silico evolution of synthetic and species-specific terminators. In the final chapter, I will address existing limitations in the field and propose new experiments to fill in the gaps. Finally, I will turn to the elephant in the room. Do all these high throughput sequencing technologies and protocols help us get any closer to accurately predicting gene expression? Are we even capturing the data in a meaningful way if we lose higher order information among all the layers of gene regulation?