Epitranscriptomic maps of 5-methylcytosine reveal substrate diversity of NSUN methyltransferases and links to mRNA translation and turnover

Multiple epitranscriptomic maps of the RNA modification 5-methylcytosine (m5C) have been prepared, often diverging markedly from each other in terms of site abundance and identity. Differences in detection methods, data depth and analysis pipelines, but also biological factors underly much of this disparity. To address this, we re-analysed available datasets from five human cell lines and seven tissues, generated by the prevailing bisulfite RNA sequencing method, with a coherent m5C site calling pipeline. We used the resulting union list of 6,393 m5C sites called across the broader transcriptome to study site distribution, enzymology, interaction with RNA-binding proteins and molecular function. We confirmed prior observations such as predominance of m5C sites within mRNA exons, enrichment around start codons, and that the tRNA:m5C methyltransferases (MTases) NSUN2 and 6 are the main m5C ‘writers’ for mRNAs. Each NSUN enzyme recognises mRNA features that strongly resemble their respective canonical substrates. Assessing proximity between mRNA m5C sites and footprints of RNA-binding proteins reconfirmed the mRNA export factor ALYREF as an m5C ‘reader’ and identified new candidates for functional interactions, including the RNA helicases DDX3X, involved in mRNA translation, and UPF1, an mRNA decay factor. Experiments in HeLa cells lacking NSUN2 showed that this affected both, steady-state level and UPF1-binding to target mRNAs. Our studies emphasise the emerging diversity of both, m5C writers and readers, affecting mRNA function.