Quantification of subcellular mRNA kinetics in mouse embryonic stem cells

In this work, we quantified mRNA flow rates between subcellular compartments in mouse embryonic stem cells. Combining metabolic RNA labeling, biochemical cell fractionation and RNA sequencing with mathematical modeling we were able to determine the kinetic rates of nuclear pre-, nuclear mature, cytosolic and membrane mRNA derived from more than 9000 protein-coding genes. For more than 5000 genes, we additionally estimated the transcript elongation rate. For most of the transcripts, mature nuclear half-lives are the longest, suggesting nuclear retention to be the rate-limiting step in the mRNA life cycle. Genes encoding transcription factors and immediate early genes possess fast kinetic rates, specifically a short nuclear half-life. Differentially localized mRNAs exhibit distinct combinations of rate constants, suggesting modular control within subcellular compartments. We show that membrane stability is high for membrane-localized mRNA and that cytosolic stability is high for cytosol-localized mRNA. Genes encoding target signals, such as signal peptides or transmembrane domains, have low cytosolic and high membrane half-lives with only slight differences between target signals. Nuclear-encoded mitochondrial proteins show long nuclear mature half-lives and otherwise similar features of cytoplasmic kinetics that do not resemble co-translational targeting to the mitochondria. Overall design: To obtain global insights into the nucleocytoplasmic kinetics of mRNA we conducted a time-resolved SLAM-seq experiment with subcellular fractionation in mouse embryonic stem cells (mESCs). mESCs were exposed to 4-thiouridine (4sU) for various time points to label newly synthesized RNA. Subsequently, we performed subcellular biochemical fractionation to generate cytosolic, membrane and nuclear fractions (Fig. 1A). 4sU concentrations were carefully chosen such that they allowed for sufficient labeling rates while minimizing toxicity elicited by 4sU. Consequently, we used higher concentrations of 4sU (500 µM) for labeling periods less than one hour, and lower concentrations (100 µM) thereafter. To quantify the kinetics of mRNA globally, we next extracted RNA from each fraction and from the unfractionated cells, followed by iodoacetamide (IAA) alkylation and poly(A)-selected strand-specific RNA library preparation Based on a binomial mixture model by Juerges et al. (2018) we determined the conversion rate per sample. With the conversion rates, T and T2C conversion counts the share of new to total RNA was calculated on intron/exon level, which was used later on as input for the kinetic model.To integrate the subcellular transcriptome and metabolic labeling data and estimate rates of subcellular mRNA kinetics, we developed transcript-wise mathematical models.