Distinct genetic architectures for syndromic and nonsyndromic congenital heart defects identified by exome sequencing

Abstract

Congenital heart defects (CHDs) have a neonatal incidence of 0.8-1% (refs. 1,2). Despite abundant examples of monogenic CHD in humans and mice, CHD has a low absolute sibling recurrence risk (∼2.7%), suggesting a considerable role for de novo mutations (DNMs) and/or incomplete penetrance. De novo protein-truncating variants (PTVs) have been shown to be enriched among the 10% of 'syndromic' patients with extra-cardiac manifestations. We exome sequenced 1,891 probands, including both syndromic CHD (S-CHD, n = 610) and nonsyndromic CHD (NS-CHD, n = 1,281). In S-CHD, we confirmed a significant enrichment of de novo PTVs but not inherited PTVs in known CHD-associated genes, consistent with recent findings. Conversely, in NS-CHD we observed significant enrichment of PTVs inherited from unaffected parents in CHD-associated genes. We identified three genome-wide significant S-CHD disorders caused by DNMs in CHD4, CDK13 and PRKD1. Our study finds evidence for distinct genetic architectures underlying the low sibling recurrence risk in S-CHD and NS-CHD.

Figure 1. Burden of de novo and inherited variants in NS-CHD compared to S-CHD

(A) Excess of DNMs compared to null mutation model. Excess of DNMs was computed as the ratio of the observed number of DNMs over the expectation given random mutation using a null gene-wise mutation rate model. P-values were computed using a Poisson model parameterized by the cumulative mutation rate across the gene set for the same number of probands (n_S-CHD= 518, n_NS-CHD= 847). We stratify by variant consequence and within known autosomal dominant CHD genes (n=78), autosomal dominant developmental disorder genes excluding autosomal dominant CHD genes (n=203) and all autosomal protein coding genes excluding autosomal dominant developmental disorder and CHD genes (n=17,404). No data is shown for silent variants in CHD genes for syndromic probands as no variants were detected. Error bars represent the 95% confidence interval. (B) Comparison of exome-wide excess of DNMs across different diseases stratified by variant consequence. (C) Excess of rare inherited variants (n_S-CHD= 471, n_NS-CHD= 663) compared to 12,031 controls of matched ancestry: Excess of DNMs was computed as the ratio of the observed number of rare inherited variants over the expected numbers as seen in controls. (D) Counts of de novo PTVs in S-CHD probands and rare inherited (INH) PTVs in NS-CHD probands in known monoallelic CHD-associated genes.

Figure 2. Gene-wise enrichment of de novo mutations

Gene-wise DNM enrichment was computed for A) the complete S-CHD cohort (n=518), B) ‘unresolved’ S-CHD trios without a plausible pathogenic DNM in known developmental disorder and CHD-associated genes (n=398). The probability of enrichment was computed given a Poisson distribution with the rate given by the gene-specific mutation rate multiplied by the number of chromosomes considered. This was performed for de novo PTVs and de novo missense variants independently. The de novo missense-enrichment probability was further combined with the probability of non-random clustering of de novo mutations using Fisher’s method and the minimum was taken between the combined and the original p-value. The minimum probability (considering either de novo PTVs or de novo missense mutations) was plotted. The dashed horizontal line represents genome-wide significance (p<1.31x10^–6, Bonferronni corrected P-value of 0.05 corrected for 2x19,252 protein coding genes).

Figure 3. Overview of CDK13 mutations in S-CHD cases

A) Phenotype summary of probands carrying missense mutations in CDK13. Colors indicate the number of times a certain phenotype was observed in individuals carrying a de novo mutation in CDK13. Photographs of affected probands are shown for which consent could be obtained for publication. B) clustering of DNMs in Serine-Threonine kinase domain. Density plot displays a sliding window (±10 amino acids) missense variant count in the Non-Finnish European population of the Exome Aggregation Consortium data, showing a marked reduction of missense variants in the kinase domain. C) 3D protein structure of CDK13 by homology modelling adapted from CDK12. Mutated residues are marked in bright green. Catalysing Magnesium ion is highlighted in magenta, and the co-crystallized AMP ligand is portrayed in orange.

Figure 4. Integrated analysis of de novo and inherited variant enrichment using Hierarchical Bayesian modelling

Scatter plots representing Bayes factors (ratio of the evidence given the alternative model of the gene being associated with CHD over the evidence given the null model of the gene not being associated with CHD) for the de novo and inherited components of the model for PTVs and missense variants. The diagonal solid line represents the identity line, where equal signal is obtained from de novo variation compared to inherited variation. Genes at an FDR < 10% are labelled and colors represent different confidence thresholds.