Identification of multiple regulatory variants at the GALNT2 human high-density lipoprotein cholesterol locus. T. S. Roman1, A. F. Marvelle1, M. P. Fogarty1, S. Vadlamudi1, M. L. Buchkovich1, J. R. Huyghe2, C. Fuchsberger2, A. U. Jackson2, K. J. Gaulton1,3, A. J. Gonzalez1, P. Soininen4,5, A. J. Kangas4, J. Kuusisto6, M. Ala-Korpela4,5, M. Laakso6, M. Boehnke2, K. L. Mohlke1 1) Department of Genetics, The University of North Carolina at Chapel Hill, Chapel Hill, NC; 2) Department of Biostatistics and Center for Statistical Genetics, School of Public Health, University of Michigan, Ann Arbor, MI; 3) Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK; 4) Computational Medicine, Institute of Health Sciences, University of Oulu, Oulu, Finland; 5) Nuclear Magnetic Resonance Metabolomics Laboratory, School of Pharmacy, University of Eastern Finland, Kuopio, Finland; 6) Department of Medicine, University of Eastern Finland and Kuopio University Hospital, Kuopio, Finland.
Genome-wide association studies have identified 150 loci associated with blood lipid and cholesterol levels; however, many of the underlying molecular and biological mechanisms are unknown. We sought to identify functional variants at the HDL-C-associated locus in GALNT2 intron 1, and we evaluated the use of ENCODE open chromatin, histone modification and transcription factor ChIP-seq data to detect variants with allele-specific regulatory activity. To characterize the locus, we performed fine-mapping and conditional analyses using 9,797 individuals from the Metabolic Syndrome in Men (METSIM) study. We confirmed a single signal and observed the strongest evidence of association with total cholesterol in medium HDL (P = 1.7x10-12) among 228 analyzed traits. To identify a target gene, we performed allelic expression imbalance experiments in 36 human hepatocyte samples. Alleles associated with increased HDL-C showed a 7% increase in GALNT2 expression (P = 5.4x10-7). All 23 intronic HDL-C-associated variants (index + r2.8) overlap 1 mark of HepG2 or liver open chromatin, histone modification, or transcription factor (TF) ChIP-seq data from ENCODE or the Epigenomics Roadmap. We tested all 23 for allelic differences in HepG2 transcriptional reporter assays and observed strong allele-specific enhancer activity for the SNP rs2281721 (75- vs 26-fold) and for a 3-variant haplotype of rs4846913, rs2144300, and rs6143660 (49- vs 14-fold). Of these 4 variants, 3 overlap 22 open chromatin, histone ChIP-seq, and/or TF ChIP-seq peaks. No allelic differences were observed for 19 variants in histone ChIP-seq peaks (4 of the 19 also overlap open chromatin or TF ChIP-seq peaks). Reporter assays using site-directed alterations of the haplotype showed that rs4846913 and rs2144300 act additively to increase transcriptional activity. To examine differential transcription factor binding, we performed electrophoretic mobility shift assays. Supershifts identified allele-specific USF-1 binding to rs2281721 and FOXO3 binding to rs6143660. We also observed differential C/EBP binding to rs4846913, which we confirmed by allele-specific ChIP assays in cell lines of differing genotypes. These data show evidence that this HDL-C GWAS association signal is driven by at least three intronic variants in GALNT2. Open chromatin, histone modification, and TF ChIP-seq data aided in the detection of, but did not perfectly predict, the multiple functional regulatory variants.
You may contact the first author (during and after the meeting) at