Decoding ribosomal RNA modification dynamics at single molecule resolution

A broad diversity of modifications decorate RNA molecules. Originally conceived as static components, evidence is accumulating that some RNA modifications may be dynamic, contributing to cellular responses to external signals and environmental circumstances. A major difficulty in studying these modifications, however, is the need of tailored protocols to map each modification individually. Here, we present a new approach that uses direct RNA nanopore sequencing to identify diverse RNA modification types present in native RNA molecules, using rRNA as the exemplar, and show that each RNA modification type results in distinct and characteristic base-calling ‘error’ signatures. We demonstrate the value of these signatures for prediction of pseudouridine (Y) modifications transcriptome-wide, confirming known Y modifications in rRNAs, snRNAs and mRNAs, and uncovering a novel Pus4-dependent Y modification in yeast mitochondrial rRNA. Using a machine learning classifier, we show that the stoichiometry of modified sites can be quantified by identifying current intensity alterations in individual RNA reads. Finally, we explore the dynamics of pseudouridylation across a battery of environmental stresses, revealing novel heat-sensitive Y-modified sites in both snRNAs and snoRNAs. Altogether, our work demonstrates that Y RNA modifications can be predicted and in a quantitative manner using native RNA nanopore sequencing.