Very low-depth sequencing in a founder population identifies a cardioprotective APOC3 signal missed by genome-wide imputation

Human Molecular Genetics, 2016, Vol. 25, No. 11 2360–2365 doi: 10.1093/hmg/ddw088 Advance Access Publication Date: 4 May 2016 Association Studies Article ASSOCIATION STUDIES ARTICLE Very low-depth sequencing in a founder population identifies a cardioprotective APOC3 signal missed by genome-wide imputation 1 Wellcome Trust Sanger Institute, Wellcome Genome Campus, Hinxton, Cambridgeshire CB10 1SA, UK, European Molecular Biology Laboratory, European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SD, UK, 3Wellcome Trust Centre for Human Genetics, Oxford OX3 7BN, UK, 4 Department of Nutrition and Dietetics, School of Health Science and Education, Harokopio University, Athens 17671, Greece and 5Anogia Medical Centre, Anogia 740 51, Greece 2 *To whom correspondence should be addressed. Tel: þ44 (0)1223 834244; Fax: þ44 (0)1223 496802; Email: Abstract Cohort-wide very low-depth whole-genome sequencing (WGS) can comprehensively capture low-frequency sequence variation for the cost of a dense genome-wide genotyping array. Here, we analyse 1x sequence data across the APOC3 gene in a founder population from the island of Crete in Greece (n ¼ 1239) and find significant evidence for association with blood triglyceride levels with the previously reported R19X cardioprotective null mutation (b ¼ 1.09,r ¼ 0.163, P ¼ 8.2  1011) and a second loss of function mutation, rs138326449 (b ¼ 1.17,r ¼ 0.188, P ¼ 1.14  109). The signal cannot be recapitulated by imputing genome-wide genotype data on a large reference panel of 5122 individuals including 249 with 4x WGS data from the same population. Gene-level meta-analysis with other studies reporting burden signals at APOC3 provides robust evidence for a replicable cardioprotective rare variant aggregation (P ¼ 3.2  1031, n ¼ 13 480). Introduction Dyslipidaemia is a well-established risk factor for cardiovascular disease, the leading cause of death worldwide. Blood lipid levels have a heritable component, and the underlying common-frequency genetic determinants have been studied in large-scale genome-wide association studies (GWAS) (1,2). Apolipoprotein CIII plays an important role in regulating triglyceride (TG) levels (3). Common-frequency variants upstream of the APOC3 gene, coding for apolipoprotein CIII, have been associated with plasma TG levels at genome-wide significance in studies of 100 000 individuals (2). More recently, a rare splice variant in APOC3 was found to be associated with blood TG levels in the UK10K study, replicating across a total of 15 000 European individuals (4). Power to detect genetic associations can be considerably higher in isolated populations as rare variants may have drifted up in frequency following the bottleneck event (5,6). In 2008, a low-frequency APOC3 null mutation (R19X) was found to have a cardioprotective effect in the Amish founder population (n  1800) (7), and the same variant was Received: December 23, 2015. Revised: March 4, 2016. Accepted: March 14, 2016 C The Author 2016. Published by Oxford University Press. V This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2360 Arthur Gilly1, Graham Rs Ritchie1,2, Lorraine Southam1,3, Aliki-Eleni Farmaki4, Emmanouil Tsafantakis5, George Dedoussis4 and Eleftheria Zeggini1,* Human Molecular Genetics, 2016, Vol. 25, No. 11 subsequently found to be associated with reduced TG levels in an isolated Greek population (n  1000) (8). R19X has independently risen in frequency to over 1% in both isolates, and is very rare (0.05%) in the general European population. A burden of rare loss of function (LoF) variants in APOC3 was found to be associated with coronary heart disease and TG levels in the Exome Sequencing Project study across 110 000 individuals from cosmopolitan populations (9). Recently, exome sequencing of 8500 European American and African American individuals identified a rare LoF variant burden in APOC3, also associated with TGs (10). Here, we use very low-depth whole-genome sequencing (WGS) data in a Greek isolated population to describe an APOC3 cardioprotective signal missed by genomewide imputation and to provide empirical proof-of-principle of how very low-depth sequencing can leverage the power advantages afforded by founder populations in catalysing these discoveries. A total of 990 individuals from the Hellenic 20 Isolated Cohorts Minoan Isolates (HELIC-MANOLIS) study were sequenced at 1x depth and 249 at 4x depth using Illumina HiSeq (total 1239 samples). Following variant calling and imputation-based genotype refinement, we identified 57 single nucleotide variants (SNVs) in the APOC3 gene (Supplementary Material, Table S1). We performed single-point association analysis with TG levels (n ¼ 1192), using a threshold of 1  108 to define genome-wide significance. Two variants exceeded this threshold, the null mutation R19X (rs76353203, b ¼ 1.09,r ¼ 0.163, P ¼ 8.2  1011), which is a C/T substitution in exon 2 that changes codon 19 into a premature stop codon, and the splice donor variant rs138326449 (b ¼ 1.17,r ¼ 0.188, P ¼ 1.14  109), located 1 base pair downstream, which disrupts the donor splice site in intron 2. These two variants are in very low linkage disequilibrium (LD) (r2 < 0.0001) (Fig. 1). To confirm genotype calling and imputation accuracy, we genotyped both R19X and rs138326449 in a subset of 1087 individuals using Sequenom massARRAY technology. In total, 98.9% of all genotypes were concordant for R19X and 99.1% for rs138326449. Minor allele concordance reached 72.2 and 80%, respectively. The fraction of true positives among non-reference calls, or positive predictive value (PPV) was high for both variants (96.3 and 100%), indicating that most mismatches were caused by false negatives rather than overconfidence in calling the alternate allele. We repeated the association analysis using the directly genotyped samples (n ¼ 1087), and found both variants to remain significantly associated with TG levels (b ¼ 1.19,r ¼ 0.165, P ¼ 3.24  1012 for R19X; b ¼ 1.10,r ¼ 0.190, P ¼ 1.63  108 for rs138326449), further confirming the validity of this signal. For burden testing, we restricted our focus on the four potentially functional rare or low-frequency [minor allele frequency (MAF) < 5%] variants that reside in exons or the essential splice sites in the consensus splice variant of APOC3 (APOC3-001) (Table 1). These included the two LoF variants R19X and rs138326449. We additionally identified a single carrier of a novel missense variant (11:116701489) also in codon 19 but in exon 3 as the intron falls between the first and second bases of the codon (Supplementary Material, Fig. S1). The resulting amino acid substitution (R19L) is predicted (...truncated)