Background— The mechanism whereby the 9p21.3 locus confers risk for coronary artery disease remains incompletely understood. Risk alleles are associated with reduced expression of the cell cycle suppressor genes CDKN2A (p16 and p14) and CDKN2B (p15) and increased vascular smooth muscle cell proliferation. We asked whether risk alleles disrupt transcription factor binding to account for this effect. Methods and Results— A bioinformatic screen was used to predict which of 59 single nucleotide polymorphisms at the 9p21.3 locus disrupt (or create) transcription factor binding sites. Electrophoretic mobility shift and luciferase reporter assays examined the binding and functionality of the predicted regulatory sequences. Primary human aortic smooth muscle cells (HAoSMCs) were genotyped for 9p21.3, and HAoSMCs homozygous for the risk allele showed reduced p15 and p16 levels and increased proliferation. rs10811656 and rs4977757 disrupted functional TEF-1 TEC1 AbaA domain (TEAD) transcription factor binding sites. TEAD3 and TEAD4 overexpression induced p16 in HAoSMCs homozygous for the nonrisk allele, but not for the risk allele. Transforming growth factor β, known to activate p16 and also to interact with TEAD factors, failed to induce p16 or to inhibit proliferation of HAoSMCs homozygous for the risk allele. Knockdown of TEAD3 blocked transforming growth factor β–induced p16 mRNA and protein expression, and dual knockdown of TEAD3 and TEAD4 markedly reduced p16 expression in heterozygous HAoSMCs. Conclusions— Here, we identify a novel mechanism whereby sequences at the 9p21.3 risk locus disrupt TEAD factor binding and TEAD3-dependent transforming growth factor β induction of p16 in HAoSMCs. This mechanism accounts, in part, for the 9p21.3 coronary artery disease risk.

The 9p21.3 locus was the first to yield to genome-wide association studies (GWAS) seeking common genetic variants predisposing to increased risk of coronary artery atherosclerotic disease (CAD). The 59 single nucleotide polymorphisms that show highest association with CAD are clustered in a region 100,000 to 150,000 base pairs 5' to the cyclin-dependent kinase inhibitors CDKN2B (coding for p15(ink4b)) and CDKN2A (coding for p16(ink4a) and p14(ARF)). This region also covers the 3' end of a long noncoding RNA transcribed antisense to CDKN2B (CDKN2BAS, aka ANRIL for antisense noncoding RNA at the ink4 locus) whose expression has been linked to chromatin remodeling at the locus. Despite intensive investigation over the past 7 years, the functional significance of the 9p21.3 locus remains elusive. Other variants at this locus have been associated with glaucoma, glioma, and type 2 diabetes mellitus, diseases that implicate tissue-resident macrophages. Here, we review the evidence that genetic variants at 9p21.3 disrupt tissue-specific enhancers and propose new insights to guide future studies.

No abstract