Figure 1.
Principle of High-Throughput Analysis of Open Chromatin Using FAIRE-gen.
In this example, two potentially functional GWAS SNPs which are in complete LD are illustrated: SNP 1, a T>C, where the C allele occurs in a region of open chromatin, relative to allele T; and SNP 2, a G>A, where both SNPs occur within open chromatin. Following formaldehyde-fixing and sonication, the T allele from SNP 1 remains tightly bound within the nucleosome. Upon phenol:chloroform extraction, this DNA-bound nucleosome transfers to the solvent layer, whilst the C allele within open chromatin remains in the aqueous layer and is purified. Upon genotyping with a gene chip, the C allele is enriched compared to the T allele. For SNP 2, the polymorphism does not affect chromatin structure; both alleles are equally enriched following FAIRE. This would suggest that SNP 1 was the more likely causal SNP for the GWAS association, conferring a greater allele-specific regulatory potential.
Figure 2.
Correlation of FAIRE LogR Ratio with FAIRE-seq Peaks.
The graph shows the correlation of FAIRE-enriched SNP intensity using Illumina CVD BeadChip with FAIRE-seq peak intensity from the GM12878 lymphoblast cell line. The (mean log R ratio from three FAIRE-enriched DNAs) – (mean log R ratio from their respective control DNAs) were compared with known FAIRE-seq intensities (0 = no FAIRE-seq enrichment; 1–200 = lowest level of FAIRE-seq enrichment; 800–1000 = highest FAIRE-seq enrichment). There is a strong correlation between SNP intensity and FAIRE-seq peak intensity (p = 2.34×10−82).
Figure 3.
Allele-Specific Open Chromatin Signals from Heterozygous Lymphoblast Cell Lines.
Manhattan plot showing allele-specific signals of open chromatin using the Human CVD beadchip. The BAF of chromosome 11 SNP, rs7120118 (C allele), is significantly enriched following FAIRE-gen, in an examination of 3,129 SNP heterozygous SNPs in 3 lymphoblast cell lines. No other SNP showed significant allele-specific effects for open chromatin.
Figure 4.
UCSC Genome Browser Annotation of rs7120118 Locus on Human Mar. 2009 (NCBI37/hg19) Assembly.
The annotated region surrounding rs7120118 (SNP highlighted in red box) reveals the location of a putative enhancer, with typical features including H3K4me3 signatures, DNase I hypersensitivity and FAIRE-seq enrichment in a number of tissues. The SNP lies between two regions of transcription factor binding sites, including a c-Fos/c-Jun (AP-1 heterodimers), p300 (a transcriptional co-activator), YY1, SRF, GATA-2 complex, and a USF-1 binding site.
Figure 5.
Replication of Allele-Specific Effect of rs7120118 from 9 Heterozygous IL-1β Stimulated Lymphoblast Cell Lines Using Illumina Metabochip.
The boxplots indicate the effect size of allele-specific differences in open chromatin. Included are the 7 SNPs in high LD. The B allele (rs7120118 C) is enriched in open chromatin, as is the adjacent SNP, rs2279239 with less statistical significance.
Table 1.
Examination of rs7120118 and SNPs in complete LD with this using Metabochip FAIRE-gen in 9 heterozygous lymphoblastoid cell lines.
Figure 6.
Replication of the Allele-Specific Effect of rs7120118 from 9 Heterozygous Lymphoblast Cell Lines Using the TaqMan Platform.
The effect of C allele-enrichment from the Illumina Metabochip is confirmed using an alternative method of allele-quantification from the TaqMan platform.
Table 2.
Baseline characteristics of the Whitehall II study (including all individuals examined on both CVD BeadChip and Metabochip) and the Copenhagen City Heart Study (CCHS).
Table 3.
Associations of SNPs in LD with rs7120118 and HDL-C levels in 3,413 individuals from the WHII study.
Table 4.
Meta-analysis using fixed-effect model of associations of rs7120118 and HDL-C in CCHS and WHII, stratified by gender.
Figure 7.
Effect of rs7120118 on NR1H3 Gene Expression in Tissue.
A) Effect of rs7120118 genotype on NR1H3 gene expression in heart samples (n = 127). The minor allele (C) is associated with increased expression at p = 0.0127. B) Effect of rs7120118 genotype on NR1H3 gene expression in aortic adventitia samples (n = 133). There is a trend towards increased expression of NR1H3 with the C allele (p = 0.154).
Figure 8.
Association of rs7120118 with HDL-C in More Than 100,000 Individuals.
Association of rs7120118 with HDL-C levels was examined using a published dataset from a study in more than 100,000 individuals [14] using LocusZoom [27] to plot the SNPs examined and imputed from 1000 Genomes Project dataset. A high level of LD is present within the locus, with at least 29 genes implicated with HDL. rs7120118 is indicated in purple (p = 1.297×10−14), with the SNPs in strongest LD marked in red. Although not statistically the lead SNP at this region, with the additional effects of this SNP on open chromatin, NR1H3 gene expression, and proximity to NR1H3, rs7120118 is represents a good functional candidate.