Protocol to distinguish pre-mRNA from mRNA in RNA-protein interaction studies

Abstract

Transcriptome-wide studies on interactions between RNA-binding proteins (RBPs) and protein-coding RNAs in general preclude interpretations regarding RBP preference for binding to the more abundant mRNA over the less abundant pre-mRNA. Here, we present a protocol to determine the binding preference of the RBP tristetraprolin (TTP, Zfp36) for pre-mRNA versus mRNA. We describe steps for the identification and quantitation of intronic and exonic fragments in RNA bound to TTP. This protocol can potentially be applied to any RBP. For complete details on the use and execution of this protocol, please refer to Bestehorn et al.¹.

Keywords: Bioinformatics; Gene Expression; Genomics; Molecular Biology; RNA-seq; Sequence analysis; Sequencing.

Graphical abstract

Figure 1

Flowchart of reference genome processing (steps 1–5) A schematic track of RNA-seq coverage is depicted in gray with annotations in black below. Exons are shown as boxes, introns as lines and arrows indicate the orientation of the gene. Prior to processing, reference genomes include exon annotations for each known transcript variant of each gene. In this scheme, one transcript variant includes a more upstream exon which does not have RNA-seq fragment coverage, and some exons and introns of different transcript variants partially overlap. Steps 1 and 2 of the reference genome processing filter for protein-coding and fully expressed variants, thereby getting rid of the variant with the more upstream exon. In step 3, intron annotations are added to the reference genome, so that in step 4, any overlapping parts of exon and intron annotations can be removed. This leaves several redundant annotations of unambiguous exons and introns for each gene, which are finally collapsed into a non-redundant and unambiguous reference genome in step 5.

Figure 2

TTP prefers binding to pre-mRNA Mean InEx for each TTP pull-down (PD)-enriched gene in Input and PD is depicted together with boxplots indicating the quantiles of either group. The mean InEx is significantly increased by the PD (t-test, ∗∗∗ = p<0.001), revealing a higher percentage of unspliced RNA in the TTP-bound fraction. The conclusion is that TTP binds preferentially pre-mRNA.

Figure 3

Annotation inaccuracy affects the InEx value RNA-seq fragment coverage tracks for three replicates of Input and PD are shown with annotations from the original reference genome shown in black (before adaption described in this protocol). Boxes in annotations are exons and connecting lines are introns, arrows on lines indicate the directionality of the gene. Red arrows indicate incongruencies between annotations and RNA-seq coverage and corrected putative annotations based on RNA-seq coverage are shown in red below the gene name. (A) Rsrp1 is shown as an example for retained introns that are not annotated as such. Coverage is in general much lower in introns than in exons. Intron 3, however, has equal coverage to exons, indicating it is retained in these cells. This falsely increases the quantity of intronic-covering fragments. (B and C) Rab24 and Med28 are shown as examples of genes with annotations where an exon extends further than the RNA-seq coverage, indicating that a more upstream polyadenylation signal is used in these cells. This can occur either in combination with (Rab24, in B) or without (Med28, in C) a more congruent variant annotated. Due to the averaging of coverage over exons and introns, a region annotated as exon without coverage decreases the exon quantification, thereby increasing the InEx. The discrepancies between annotations and RNA-seq coverage mentioned in A–C affect the InEx of either Input or PD samples in the same manner and are therefore negligible when comparing the InEx between Input and PD. (D) Uqcc4 is another example of an exon annotation extending further than RNA-seq coverage in Input. However, the pull-down enriched a longer transcript variant that covers the entire annotated exon. As in the depicted example, this can happen if the RBP binding site is only present in the longer variant. CLIP data used to infer the TTP binding site in this gene is from Sedlyarov et al. For this gene specifically, the InEx of Input and PD is therefore subject to different biases. Inferring a higher binding preference of TTP for pre-mRNA than mature mRNA from such InEx differences would therefore be invalid.