DeepSAGE Reveals Genetic Variants Associated with Alternative Polyadenylation and Expression of Coding and Non-coding Transcripts

Abstract

Many disease-associated variants affect gene expression levels (expression quantitative trait loci, eQTLs) and expression profiling using next generation sequencing (NGS) technology is a powerful way to detect these eQTLs. We analyzed 94 total blood samples from healthy volunteers with DeepSAGE to gain specific insight into how genetic variants affect the expression of genes and lengths of 3′-untranslated regions (3′-UTRs). We detected previously unknown cis-eQTL effects for GWAS hits in disease- and physiology-associated traits. Apart from cis-eQTLs that are typically easily identifiable using microarrays or RNA-sequencing, DeepSAGE also revealed many cis-eQTLs for antisense and other non-coding transcripts, often in genomic regions containing retrotransposon-derived elements. We also identified and confirmed SNPs that affect the usage of alternative polyadenylation sites, thereby potentially influencing the stability of messenger RNAs (mRNA). We then combined the power of RNA-sequencing with DeepSAGE by performing a meta-analysis of three datasets, leading to the identification of many more cis-eQTLs. Our results indicate that DeepSAGE data is useful for eQTL mapping of known and unknown transcripts, and for identifying SNPs that affect alternative polyadenylation. Because of the inherent differences between DeepSAGE and RNA-sequencing, our complementary, integrative approach leads to greater insight into the molecular consequences of many disease-associated variants.

The roles of RNA processing in translating genotype to phenotype

“…A detailed analysis of blood samples from 94 individuals to find SNVs in linkage disequilibrium with a change in polyadenylation site usage identified 5 top candidates, all of which were in the AATAAA motif. This analysis also identified seven APA-associated SNVs with disease associations in the GWAS database 46 .…”

The Persistent Contributions of RNA to Eukaryotic Gen(om)e Architecture and Cellular Function

“…For that reason, some investigators took measures to stay clear of transcripts that are located in a certain proximity to annotated protein-coding genes (Managadze et al 2013). Nevertheless, splicing does occur in 3 0 -UTRs as well (Bicknell et al 2012;Camacho-Vanegas et al 2012;Zhernakova et al 2013) and, hence, the corresponding exons could map to corresponding protein-coding genes at a much greater distance to the ORFs. Furthermore, apart from spurious initiation of transcription anywhere in the genome, a certain level of readthrough into gene distal regions could account for a significant portion of these long RNA candidates (Finta and Zaphiropoulos 2000;Akiva et al 2006;Parra et al 2006;Richard and Manley 2009).…”

The length of the expressed 3’ UTR is an intermediate molecular phenotype linking genetic variants to complex diseases

“…We used a new efficient strategy to study how human genetic variants influence the expression of alternative 3'UTR isoforms. This issue has been previously investigated with different approaches [26,25,27,8,9]. The method we propose combines wide applicability, being based on standard RNA-Seq data, with the high sensitivity allowed by limiting the analysis to a single type of transcript structure variant, namely 3' UTR length.…”

“…A recent analysis of whole genome sequencing (WGS) data from [8] found hundreds of common single nucleotide polymorphisms (SNPs) causing the alteration or degradation of motifs that are similar to the canonical PAS [23], but did not extend the analysis to other possible mechanisms. Other studies found strong associations between genetic variants and APA regulation [24,25,26,8,9,27], but a systematic investigation based on a large number of samples and variants, specifically targeted to APA rather than generically to transcript structure, and unbiased in the choice of variants to examine, is not yet available.…”