As clinical genetic testing is becoming widely deployed for patients with suspected Mendelian diseases as well as in the broad population, a major emerging challenge is the accurate prediction of pathogenicity of DNA variants. Multiple features, including allele frequencies across populations, family history, and functional studies, are currently being used to assign known or new variants to three broad categories: benign, pathogenic, or (most commonly) variant of uncertain significance (VUS). We propose here to test the hypothesis that deploying novel phenotyping methods will improve variant classification, and we will focus here on SCN5A, encoding the cardiac sodium channel. Despite the fact that this channel plays a critical role in normal heart function, SCN5A variants are surprisingly common and have been associated with serious and occasionally life-threatening phenotypes including type 1 Brugada Syndrome (BrS1), type 3 Long QT Syndrome (LQT3), conduction system disease, heart failure, and atrial fibrillation.
In aim 1, we will determine the utility of phenotyping in the electronic health record (EHR) coupled to functional studies to assess SCN5A VUS pathogenicity. The Electronic Medical Records and Genomics (eMERGE) network in which we are participants is completing sequencing of 109 Mendelian disease genes, including SCN5A, in 25,000 subjects with EHRs. We will assess VUS pathogenicity by analyzing SCN5A-related EHR phenotypes in subjects with and without SCN5A rare variants and establishing in vitro function for newly-detected variants using multiplexed semi-automated electrophysiologic methods.
In aim 2, we will build on preliminary data using deep mutational scanning (DMS) ? a high-throughput method to mutagenize each nucleotide in a target genomic sequence and establish its functional consequences ? to identify SCN5A coding variants with BrS1 or LQT3 features. In a pilot experiment, we developed a drug challenge that biases survival toward cells that do not express sodium current at their surface (the BrS phenotype) and against cells that display enhanced sustained sodium current (LQT3). We then generated all 252 possible non-synonymous or nonsense single amino acid (aa) variants across a 12aa regon of SCN5A, exposed a pool of cells transfected with one mutant/cell to the drug challenge, used next-generation sequencing pre- and post-drug challenge to identify variants with BrS1 and LQT3 features, and validated predictions with conventional patch clamp methods. We will now scale up to scan larger regions, starting with the 253 aa encoding transmembrane domain IV, known to harbor dozens of pathogenic variants. Further, we will map DMS results onto a model of the channel to probe the structural basis of loss and gain of function variants. We will incorporate DMS data into available and new statistical models to build an improved model of SCN5A variant disease risk. The result of this work will be new approaches to establish functional consequences of rare variants in SCN5A (and ultimately other genes), and thus enable genomic medicine.
/Public Health Relevance Statement Genome sequencing is entering the mainstream of medicine both to care for patients with suspected genetic conditions and to detect such conditions in apparently healthy individuals. We now know that this sequencing frequently detects genetic changes that may or may not affect a person?s health. This proposal seeks to understand how best to classify a very large number of these changes in a key gene controlling the normal heartbeat, and to thereby ultimately better care for patients at risk for serious heart rhythm problems.