Genetic variants associated with risk of atrial fibrillation regulate expression of PITX2, CAV1, MYOZ1, C9orf3 and FANCC.

Genome-wide association studies (GWAS) have identified genetic variants in a number of chromosomal regions that are associated with atrial fibrillation (AF). The mechanisms underlying these associations are unknown, but are likely to involve effects of the risk haplotypes on expression of neighbouring genes. To investigate the association between genetic variants at AF-associated loci and expression of nearby candidate genes in human atrial tissue and peripheral blood. Right atrial appendage (RAA) samples were collected from 122 patients undergoing cardiac surgery, of these, 12 patients also had left atrial appendage samples taken. 22 patients had a history of AF. Peripheral blood samples were collected from 405 patients undergoing diagnostic cardiac catheterisation. In order to tag genetic variation at each of nine loci, a total of 367 single nucleotide polymorphisms (SNPs) were genotyped using the Sequenom platform. Total expression of 16 candidate genes in the nine AF-associated regions was measured by quantitative PCR. The relative expression of each allele of the candidate genes was measured on the Sequenom platform using one or more transcribed SNPs to distinguish between alleles in heterozygotes. We tested association between the SNPs of interest and gene expression using total gene expression (integrating cis and trans acting sources of variation), and allelic expression ratios (specific for cis acting influences), in atrial tissue and peripheral blood. We adjusted for multiple comparisons using a Bonferroni approach. In subsidiary analyses, we compared the expression of candidate genes between patients with and without a history of AF. Total expression of 15 transcripts of 14 genes and allelic expression ratio of 14 transcripts of 14 genes in genomic regions associated with AF were measured in right atrial appendage tissue. 8 of these transcripts were also expressed in peripheral blood. Risk alleles at AF-associated SNPs were associated in cis with an increased expression of PITX2a (2.01-fold, p=6.5×10(-4)); and with decreased expression of MYOZ1 (0.39 fold; p=5.5×10(-15)), CAV1 (0.89 fold; p=5.9×10(-8)), C9orf3 (0.91 fold; 1.5×10(-5)), and FANCC (0.94-fold; p=8.9×10(-8)) in right atrial appendage. Of these five genes, only CAV1 was expressed in peripheral blood; association between the same AF risk alleles and lower expression of CAV1 was confirmed (0.91 fold decrease; p=4.2×10(-5)). A history of AF was also associated with a decrease in expression of CAV1 in both right and left atria (0.84 and 0.85 fold, respectively; p=0.03), congruent with the magnitude of the effect of the risk SNP on expression, and independent of genotype. The analyses in peripheral blood showed association between AF risk SNPs and decreased expression of KCNN3 (0.85-fold; p=2.1×10(-4)); and increased expression of SYNE2 (1.12-fold; p=7.5×10(-24)); however, these associations were not detectable in atrial tissue. We identified novel cis-acting associations in atrial tissue between AF risk SNPs and increased expression of PITX2a/b; and decreased expression of CAV1 (an association also seen in peripheral blood), C9orf3 and FANCC. We also confirmed a previously described association between AF risk variants and MYOZ1 expression. Analyses of peripheral blood illustrated tissue-specificity of cardiac eQTLs and highlight the need for larger-scale genome-wide eQTL studies in cardiac tissue. Our results suggest novel aetiological roles for genes in four AF-associated genomic regions.