MFB: Cracking the codes: understanding the rules of mRNA localization and translation

Tight control of gene expression is key to normal cell function. A complete understanding of the components of protein production would be transformative for biotechnology and biomedical research. However, our ability to predict protein output based on a set of conditions is poor, due to ineffective models that account for the impact of the genetic code, its conversion to protein output, and the influence of cellular location on protein production. This proposal will apply recent developments in RNA sequencing technology to produce improved data that captures new aspects of the RNA lifecycle, and then build and test new machine learning models to determine whether they are predictive of protein output. The project will also provide interdisciplinary undergraduate, graduate, and postdoc training in RNA biology, and conduct outreach about modern RNA sequencing techniques to local high school students. Predicting the protein output from a messenger RNA (mRNA) is a major challenge for the RNA field. Recent developments in RNA sequencing provide high resolution views of tissue and single-cell transcriptomes and insights into which mRNAs are actively translated, leading to an emerging appreciation that tRNA abundances and mRNA subcellular location influence protein synthesis for a given mRNA, and may be tissue- and even cell type-specific. However, the mechanisms underlying these two parameters remain unclear due to the lack of methods for their interrogation. This project combines ribosome profiling with novel tRNA sequencing techniques to advance our understanding of these key variables that control protein production. In addition, the project seeks to develop an “RNA passport” method leveraging sequential modification of localized RNAs and long-read sequencing to better define the impact of localization on translation. The overall hypothesis is that a refined understanding of the relationship between mRNA content, its subcellular location, and tRNA abundance will better predict protein production. A machine learning approach will identify organism-level rules and tissue-specific patterns of translational optimization, validate the predicted impact on protein synthesis using public proteomic datasets, and test the rules on synthetic gene constructs in multiple biological contexts. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.