Cultivo y medios utilizados en el estudio

Although a central regulator of cellular processes, RNA rarely acts alone in the cell [388–390]. A vast majority of transcribed RNAs including messenger RNAs (mRNAs) interact with a host of exogenous molecules and proteins, and in some cases may even resemble RNPs (Ribonucleoproteins) [391–394]. Understanding the molecular determinants of RNA/Protein interaction specificity and sensitivity is therefore central to accurately predicting regulatory processes in the cell [303, 389, 395, 396]. Although sequence motifs are important determinants of RNA binding protein’s (RBPs) specificity, RNA structure (and in particular the accessibility of the binding site) is essential to fine-tuning the interaction [153, 397–400].

Mutations have the potential to disrupt specific RNA structural elements by destabilizing existing base-pairs or favoring alternative conformations [179]. In certain cases, these can even result in human disease phenotypes [124, 402]. When considering RNA/protein interactions, two scenarios for mutations in proximity to RBP binding sites are potentially deleterious. For an RBP interacting with RNA as illustrated in Figure 5.1A, the most direct mechanism by which a mutation will disrupt the interaction is shown in Figure 5.1B. In this manuscript, we are particularly interested in the scenario illustrated in Figure 5.1C; a mutation alters the structure of the RNA and thus changes the accessibility of the binding site. In this case, the mutation occurs in a proximal region to the binding site, indicated in Figure 1C (light pink shading).

We distinguish two major classes of mutation induced structure change in non-coding regions of mRNAs including 3’ UTRs. These regions are generally not evolved to adopt a single conformation, but rather adopt multiple structures weighted by the Boltzmann ensemble [179, 203, 229, 277, 403, 404]. If a majority of the structures are similar in structure, the ensemble is considered low-entropy, while if they all differ significantly from each other, it is considered high-entropy [405, 406]. As a result a mutation may alter structure by favoring

A) B) C) D) E) 5’ NUP62 STK17B IGF2BP1-3 AGO1-4 FMR1 (isoform 7) FXR2 FMR1 (isoform 1) PUM2 0.858 0.717 0.717 -0.433 -0.217 -0.115 5’ 3’ 3’ rs11547267 _rs1044276 rs9523 rs8429 _rs11083988 rs4850664 rs7581626 rs12467988 rs6760556rs6760432 rs6745041 rs6741307 E_d 500 nt 500 nt Ed

Figure 5.1: Possible mechanisms by which a mutation (or SNP) can affect RNA/protein interactions in non-coding regions of transcripts. A) The protein (blue oval) recognizes a specific sequence motif (green nucleotides) in a highly accessible region of the RNA. B) If a SNP (red nucleotide) is present within the highly accessible recognition motif it can disrupt the protein/RNA interaction. C) Alternatively, a proximal SNP can alter the structure of the transcript and reduce the accessibility of the nucleotides for protein recognition. In this chapter we analyze direct and proximal overlap between functional SNPs and experimentally determined RNA binding sites to characterize the structural consequences of mutation induced RNA structure change on RBP binding. Proximal regions to RBP sites are indicated in light pink. D) 3’ UTR of the NUP62 (nucleoprotein 62, GeneID 23636) which is 1412 nucleotides long. Two eQTL SNPs map to the UTR (rs9523 and rs8429, indicated with red inverted triangles), and are in LD (indicated with a large red triangle). In addition three SNPs (rs11547267, rs1044276 and rs11083988) are in LD with eQTL SNPs that map to the NUP62 gene and are associated with changes in NUP62 mRNA levels [401]. Several SNPs overlap with regions proximal to the RBP binding sites (indicated in light pink) for IGF2BP1-3, AGO1-4 and FMR1 (isoform 7), indicative of enrichment as measured by theEdmetric (Equation 2). E.) Seven eQTL SNPs (four of which are in LD

with each other) map to the 3’ UTR of STK17B (3,864 nucleotides long, Serine/Threonine Kinase 17b, GeneID 9262), however none overlap with proximal regions to the PAR-CLIP determined RBP sites. Proximal eQTL SNP overlap is therefore depleted in this 3’ UTR as is evidenced by the negativeEdvalues.

an alternative low-entropy conformation or by increasing (or decreasing) the entropy of the ensemble [407–409]. We therefore decided to characterize the structural changes induced by eQTL SNPs proximal to RBP sites (Figure 5.1C) to determine the role of the entropy of the Boltzmann ensemble in RBP/RNA interactions.

To investigate the role of mutation induced RNA structure change (or RiboSNitches [124, 179]) on RBP binding we require transcriptome-wide maps of both RBP sites and ”functional” mutations. Recent high-throughput datasets obtained by cross-linking RNA and subsequent

immunoprecipitation of the RBP, specifically Photoactivatable-Ribonucleoside-Enhanced Crosslinking and Immunoprecipitation (PAR-CLIP), provide the necessary transcriptome wide coordinates for such an analysis [303, 304]. We utilized publicly available PAR-CLIP data for 13 different RBPs [410]. eQTL SNPs (expressed Quantitative Trait Loci Single Nucleotide Polymorphisms) are particularly relevant to this study as they are correlated with changes (up or down) in mRNA expression [249, 312, 411]. Many eQTL SNPs affect mRNA levels by altering transcription factor binding and thus transcriptional efficiency [412]. In this study we are interested in those eQTLs that are not near or in linkage disequilibrium (LD) with transcription factor binding sites. Instead we focus on those eQTL SNPs that map to 3’ UTRs (Untranslated Regions) of genes. These regions are not only far from transcription start sites, but are also the main binding sites of many RBPs [179, 187, 241]. The driving hypothesis of this study is that characterizing the RNA structure change that results from eQTL SNPs near RBP sites can provide insight into the role of RNA structure change (Figure 5.1C) in affecting RBP binding specificity and sensitivity.