How do DNA sequence variants in non-coding sequence contribute to MI risk?

Most of the polymorphisms discovered through GWAS are in non-coding sequence (i.e., they do not affect protein structure) and as such, it was unclear if and how these non-coding variants affect human phenotype. At the chromosome 1p13 gene region that we discovered for LDL cholesterol and MI risk, we demonstrated that a non-coding variant disrupts a transcription factor-binding site and thereby alters expression of a nearby gene, sortilin, in a tissue-specific manner (Nature 2010). In mouse models, we showed that sortilin regulates LDL cholesterol by decreasing VLDL secretion. At every GWAS discovery, four considerations emerge: 1) causal variant; 2) mechanism linking variant to gene; 3) the causal gene; and 4) mechanism linking gene to phenotype. By addressing each of these questions, our work at 1p13 has set the standard for functional follow-up of GWAS discoveries.