Rare genomic copy number variants implicate new candidate genes for bicuspid aortic valve

Abstract

Bicuspid aortic valve (BAV), the most common congenital heart defect, is a major cause of aortic valve disease requiring valve interventions and thoracic aortic aneurysms predisposing to acute aortic dissections. The spectrum of BAV ranges from early onset valve and aortic complications (EBAV) to sporadic late onset disease. Rare genomic copy number variants (CNVs) have previously been implicated in the development of BAV and thoracic aortic aneurysms. We determined the frequency and gene content of rare CNVs in EBAV probands (n = 272) using genome-wide SNP microarray analysis and three complementary CNV detection algorithms (cnvPartition, PennCNV, and QuantiSNP). Unselected control genotypes from the Database of Genotypes and Phenotypes were analyzed using identical methods. We filtered the data to select large genic CNVs that were detected by multiple algorithms. Findings were replicated in a BAV cohort with late onset sporadic disease (n = 5040). We identified 3 large and rare (< 1,1000 in controls) CNVs in EBAV probands. The burden of CNVs intersecting with genes known to cause BAV when mutated was increased in case-control analysis. CNVs intersecting with GATA4 and DSCAM were enriched in cases, recurrent in other datasets, and segregated with disease in families. In total, we identified potentially pathogenic CNVs in 9% of EBAV cases, implicating alterations of candidate genes at these loci in the pathogenesis of BAV.

Fig 1. Specific rare genomic copy number variants may influence BAV disease severity.

Aortic events, thoracic aortic aneurysm, thoracic aortic dissection, or aortic valve stenosis or regurgitation requiring aortic valve repair or replacement.

Fig 2. Overview of pipeline for CNV identification and validation.

SNP, single nucleotide polymorphism; QC, Quality control; CNV, copy number variant. Illumina B-allele frequency and signal intensity data was trimmed and exported using GenomeStudio. Three different algorithms (PennCNV [27], cnvPartition, and QuantiSNP [28]) were used to generate initial CNV calls and sample-level statistics. Sample-level quality control analysis was performed using PennCNV. PLINK [29] was used to define CNV regions for subsequent burden, enrichment, and replication tests. Raw CNV calls were individually screened for CNVs intersecting with genes implicated in BAV and enriched in case-control tests. CNVs were validated by examining the raw data in GenomeStudio.

Fig 3. UCSC genome browser plots of GATA4 and DSCAM variants.

Each bar represents a copy number variant (CNV). Blue, EBAV CNVs; Red: CNVs BAVGWAS CNVs. (a) Ideogram of Chromosome 8 with CNV region highlighted red; (b) Plot of GATA4 CNVs; (c) Ideogram of Chromosome 21 with CNV region highlighted red; (d) Plot of DSCAM CNVs. Figures were constructed using the UCSC Genome Browser (http://genome.ucsc.edu) [35].