Single nucleotide polymorphisms with cis-regulatory effects on long non-coding transcripts in human primary monocytes

Abstract

We applied genome-wide allele-specific expression analysis of monocytes from 188 samples. Monocytes were purified from white blood cells of healthy blood donors to detect cis-acting genetic variation that regulates the expression of long non-coding RNAs. We analysed 8929 regions harboring genes for potential long non-coding RNA that were retrieved from data from the ENCODE project. Of these regions, 60% were annotated as intergenic, which implies that they do not overlap with protein-coding genes. Focusing on the intergenic regions, and using stringent analysis of the allele-specific expression data, we detected robust cis-regulatory SNPs in 258 out of 489 informative intergenic regions included in the analysis. The cis-regulatory SNPs that were significantly associated with allele-specific expression of long non-coding RNAs were enriched to enhancer regions marked for active or bivalent, poised chromatin by histone modifications. Out of the lncRNA regions regulated by cis-acting regulatory SNPs, 20% (n = 52) were co-regulated with the closest protein coding gene. We compared the identified cis-regulatory SNPs with those in the catalog of SNPs identified by genome-wide association studies of human diseases and traits. This comparison identified 32 SNPs in loci from genome-wide association studies that displayed a strong association signal with allele-specific expression of non-coding RNAs in monocytes, with p-values ranging from 6.7×10(-7) to 9.5×10(-89). The identified cis-regulatory SNPs are associated with diseases of the immune system, like multiple sclerosis and rheumatoid arthritis.

Figure 1. Comparison of total expression levels of regions annotated to intergenic lncRNAs, exons, introns, and intergenic regions.

The horizontal axes in the panels show bins of fluorescence signals from the genotyping data, summed for both alleles to give a measure for total expression. The average expression levels of annotated transcripts were 4900 fluorescence units in exons, 2100 in introns, 590 in intergenic regions, compared to 3300 in the intergenic lncRNA regions that were used in the ASE analysis. The vertical axes show the number of observations in each bin.

Figure 2. Manhattan plot.

Manhattan plot with the p-values from ASE association tests between SNPs and lncRNAs on the vertical axis and the genomic lncRNA regions analysed in the study on the horizontal axis. The p-value cut-off of 10⁻⁶ is shown as a grey line.

Figure 3. Illustration of a region with a SNP from genome wide association studies (GWAS) which is associated with ASE of lncRNAs.

The tracks are from top to bottom in each panel: Horizontal red bars represent lncRNA transcript windows (with genomic coordinates) used for determination of ASE levels; grey lines show p-values for the association of GWAS SNPs with ASE levels in the transcript window; a grey line overlayed with a red dotted line indicates that a cis-rSNP overlaps with the reported SNP in the GWAS catalog; red vertical lines are median ASE-levels for each SNP.