Studying Isoform-Specific mRNA Recruitment to Polyribosomes with Frac-seq

Abstract

Gene expression profiling is widely used as a measure of the protein output of cells. However, it is becoming more evident that there are multiple layers of post-transcriptional gene regulation that greatly impact protein output (Battle et al., Science 347:664–667, 2014; Khan et al., Science 342:1100–1104, 2013; Vogel et al., Mol Syst Biol 6:400, 2010). Alternative splicing (AS) impacts the expression of protein coding genes in several ways. Firstly, AS increases exponentially the coding-capacity of genes generating multiple transcripts from the same genomic sequence. Secondly, alternatively spliced mRNAs are subjected differentially to RNA-degradation via pathways such as nonsense mediated decay (AS-NMD) or microRNAs (Shyu et al., EMBO J 27:471–481, 2008). And thirdly, cytoplasmic export from the nucleus and translation are regulated in an isoform-specific manner, adding an extra layer of regulation that impacts the protein output of the cell (Martin and Ephrussi, Cell 136:719–730, 2009; Sterne-Weiler et al., Genome Res 23:1615–1623, 2013). These data highlight the need of a method that allows analyzing both the nuclear events (AS) and the cytoplasmic fate (polyribosome-binding) of individual mRNA isoforms.

In order to determine how alternative splicing determines the polyribosome association of mRNA isoforms we developed Frac-seq. Frac-seq combines subcellular fractionation and high throughput RNA sequencing (RNA-seq). Frac-seq gives a window onto the translational fate of specific alternatively spliced isoforms on a genome-wide scale. There is evidence of preferential translation of specific mRNA isoforms (Coldwell and Morley, Mol Cell Biol 26:8448–8460, 2006; Sanford et al., Genes Dev 18:755–768; Zhong et al., Mol Cell 35:1–10, 2009; Michlewski et al., Mol Cell 30:179–189, 2008); the advantage of Frac-seq is that it allows analyzing the binding of alternatively spliced isoforms to polyribosomes and comparing their relative abundance to the cytosolic fraction. Polyribosomes are resolved by sucrose gradient centrifugation of cytoplasmic extracts, subsequent reading and extraction. The total mRNA fraction is taken prior ultracentrifugation as a measure of all mRNAs present in the sample. Both populations of RNAs are then isolated using phenol–chloroform precipitation; polyadenylated RNAs are selected and converted into libraries and sequenced. Bioinformatics analysis is then performed to measure alternatively spliced isoforms; several tools can be used such as MISO, RSEM, or Cufflinks (Katz et al., Nat Methods 7:1009–1015, 2010; Li and Dewey, BMC Bioinformatics 12:323, 2011; Trapnell et al., Nat Protoc 7:562–578, 2012). Comparison of total mRNAs and polyribosome-bound mRNAs can be used as a measure of the polyribosome association of specific isoforms based on the presence/absence of specific alternative splicing events in each fraction. Frac-seq shows that not all isoforms from a gene are equally loaded into polyribosomes, that mRNA preferential loading does not always correlate to its expression in the cytoplasm and that the presence of specific events such as microRNA binding sites or Premature Termination Codons determine the loading of specific isoforms into polyribosomes.