Systematic analysis of microRNA targeting impacted by small insertions and deletions in human genome

Abstract

MicroRNAs (miRNAs) are small noncoding RNA that play an important role in posttranscriptional regulation of mRNA. Genetic variations in miRNAs or their target sites have been shown to alter miRNA function and have been associated with risk for several diseases. Previous studies have focused on the most abundant type of genetic variations, single nucleotide polymorphisms (SNPs) that affect miRNA-mRNA interactions. Here, we systematically identified small insertions and deletions (indels) in miRNAs and their target sites, and investigated the effects of indels on miRNA targeting. We studied the distribution of indels in miRNAs and their target sites and found that indels in mature miRNAs, experimentally supported miRNA target sites and PAR-CLIP footprints have significantly lower density compared to flanking regions. We identified over 20 indels in the seed regions of miRNAs, which may disrupt the interactions between these miRNAs and their target genes. We also identified hundreds of indels that alter experimentally supported miRNA target sites. We mapped these genes to human disease pathways to identify indels that affect miRNA targeting in these pathways. We also used the results of genome-wide association studies (GWAS) to identify potential links between miRNA-related indels and diseases.

Figure 1. Density of all genetic variants (a) and indels (b) in dbSNP 135 as well as indels (c) from the GATK resource bundle in pre-miRNAs, mature miRNAs, miRNA seed regions, and flanking regions.

Flanking regions 1 and 2 represent successive sequences adjacent to pre-miRNAs that were equal to the length of the pre-miRNA (∼100 bp). Error bars indicate the standard error. The density of all genetic variants (a) in pre-miRNAs was significantly different from the density in flanking regions, mature miRNAs, and seed regions (*p<0.01, **0.01

Figure 2. Density of all genetic variants (a) and indels (b) in dbSNP 135 as well as indels (c) from the GATK resource bundle in PAR-CLIP footprints and flanking regions, entire 3′ UTR and experimentally validated target sites.

Flanking regions 1 and 2 represent successive sequences adjacent to PAR-CLIP footprints that were equal to the length of the footprints (∼41 bp). Error bars indicate the standard error. The density of all genetic variants (a) in PAR-CLIP footprints was significantly different from the density in flanking regions (**10⁻⁵<p<10⁻³, ***p = 0.04). The density of indels in PAR-CLIP footprints (b) and (c) was significantly different from the density in flanking regions (*p<10⁻¹²). The density of all genetic variants (a) in experimentally validated targets was significantly different from the density in entire 3′ UTR regions (**p = 5.7×10⁻⁷). The density of indels in experimentally validated targets (b) and (c) was significantly different from the density in 3′ UTR regions (*p<10⁻¹²).

Figure 3. Genes in the pancreatic cancer pathway containing SNPs and indels that altered experimentally supported target sites.

Genes containing only indels (pink), only SNPs (yellow), and both SNPs and indels (green) in target sites are within colored rectangles. The miRNAs that have been shown to target these genes are shown with red text for disrupted sites and blue text for created sites.

Figure 4. Disruption of a target site of miR-378g in the 3′ UTR of DNAJC27 by an indel rs34922018 that is in linkage disequilibrium with a high-scoring marker rs713586 from a body mass index study.