De novo and recessive forms of congenital heart disease have distinct genetic and phenotypic landscapes

Abstract

The genetic architecture of sporadic congenital heart disease (CHD) is characterized by enrichment in damaging de novo variants in chromatin-modifying genes. To test the hypothesis that gene pathways contributing to de novo forms of CHD are distinct from those for recessive forms, we analyze 2391 whole-exome trios from the Pediatric Cardiac Genomics Consortium. We deploy a permutation-based gene-burden analysis to identify damaging recessive and compound heterozygous genotypes and disease genes, controlling for confounding effects, such as background mutation rate and ancestry. Cilia-related genes are significantly enriched for damaging rare recessive genotypes, but comparatively depleted for de novo variants. The opposite trend is observed for chromatin-modifying genes. Other cardiac developmental gene classes have less stratification by mode of inheritance than cilia and chromatin-modifying gene classes. Our analyses reveal dominant and recessive CHD are associated with distinct gene functions, with cilia-related genes providing a reservoir of rare segregating variation leading to CHD.

Fig. 1

Candidate gene discovery pipeline. PCGC trios were joint-genotyped with the Genome Analysis Tool Kit (GATK). Variant calls were annotated with the Ensembl Variant Effect Predictor (VEP version 90). Each trio VCF file was processed with the Variant Annotation, Analysis and Search Tool (VAAST version 3). Every candidate gene identified by VAAST in each trio was re-ranked with the Phenotype Driven Variant Ontological Re-ranking Tool (PHEVOR) using human phenotype ontology (HPO) terms matching the proband’s phenotype. To assess variant quality and remove potential false positives, each variant was adjudicated with the graph-based alignment tool, GRAPHITE. A final list of all recessive and de novo candidate genes was assembled for 2391 probands

Fig. 2

Cilia, chromatin, and CHD gene lists and burden analysis. a The relationships among five major gene sets used for enrichment analysis. Structural cilia (SysCilia) genes are a subset of all Cilia list genes. The CHD gene list has no overlap with cilia genes, and chromatin genes have only 3% overlap with cilia genes. b Gene burden by gene list. GnomAD-based gene-burden estimates (gray circles) are plotted as a function of transcript size. Regression lines for the genes contained in each major gene set are shown. The positive slopes indicate that, in general, large genes have higher burden than small genes. Cilia and FoxJ1-responsive genes show a higher rate (steeper slope) of increase in burden with gene size than do CHD or chromatin genes. Housekeeping genes have lower overall burden. c VAAST p values estimates are not affected by gene length or gene-burden. VAAST p values for damaged genes (gray dots) are plotted as a function of gene burden. The regression line (blue) shows the relationship between the VAAST p value estimates for damaged genes found in CHD probands and gene burden normalized by transcript size. No significant relationship between the number of damaged genes discovered nor magnitude of the p value as a function of gene burden is observed (coefficient of variation (R²_adj) = −0.00231, p value ≥ 0.66, linear-model F-test)

Fig. 3

Enrichment profiles for damaged cilia and chromatin genes in CHD probands. Five candidate gene lists, SysCilia (302), Cilia (669), FoxJ1 (116), CHD (402), and chromatin-modifying (163) genes were tested for enrichment in damaging genotypes using 2391 congenital heart disease trios. The number of damaged genes discovered in the 2391 probands for each candidate gene list (red arrows) is compared to the distribution of damaged genes found using random gene lists of equal size (blue distributions, 100,000 independent random gene lists per distribution). a, c SysCilia and Cilia genes are highly enriched for damaged recessively inherited genotypes. b, d SysCilia and Cilia genes show only modest enrichment in de novo mutations. e, f FoxJ1-responsive genes are also modestly enriched for recessive variation but not for de novo variation. g, i CHD and chromatin-modifying genes are only modestly enriched for damaging recessive genotypes. h, j In contrast to cilia genes, known CHD and chromatin-modifying genes are highly enriched for damaging de novo mutations. Burden-matched control genes (pink arrows) are not significantly enriched for any gene set. Housekeeping genes (green arrows) are depleted for damaging recessive variation and have a typical amount of damaging de novo variation. All p values are obtained by empirical permutation

Fig. 4

Enrichment profiles for additional genes and gene pathways. Several additional pathways implicated in congenital heart disease were tested for enrichment in damaging recessive genotypes and de novo mutations. a, b The Notch signaling pathway is enriched for de novo mutations (p value < 0.0012) but is not enriched for damaging recessive genotypes. c–f Genes involved in TGF-β signaling and non-ciliary cytoskeletal genes are moderately enriched in damaging recessive and de novo genotypes. g, h In contrast, receptor serine-threonine kinases show no enrichment for either damaging de novo or recessive genotypes. All p values are obtained by emperical permutation

Fig. 5

Relative enrichment of damaging recessive and de novo variation in CHD patients. To compare the findings among the distributions shown in Figs. 3 and 4, each distribution was normalized by Z-score transformation. Recessively inherited damaged genotypes found in SysCilia and Cilia genes are highly enriched in CHD probands (+7.7 SDs, p value < 1e-5). These genes show only marginal enrichment for de novo mutations as compared to the enrichment seen for damaging recessive genotypes. In contrast, there is strong enrichment for de novo mutations in chromatin-modifying and known CHD genes (+14.0 and +17.1 SDs, respectively, p value < 1e-5) but relatively moderate enrichment for damaging recessive genotypes. Other gene lists representing the Notch pathway,  the TGF-β signaling pathway, and cytoskeletal genes are also moderately enriched (+4.0 to +5.2 SDs), suggesting an important but more limited contribution to CHD as compared to cilia and chromatin genes. P values are obtained by permutation of the transformed distributions

Fig. 6

Genetic and phenotypic landscapes for 2391 patients with congenital heart defects nodes in these two Bayesian networks correspond to genotypes, gene functions, tissue of expression, phenotypes, and potentially confounding variables such as ancestry and gender. Connections, or edges, between nodes denote conditional dependencies. The width of an edge is proportional to the relative strength of the dependency, calculated as the average pairwise fold change in risk. Blue denotes positive dependencies; red denotes negative dependencies. Relationships between multistate nodes (gender, ancestry and capture method) are shown in gray, because the nature of an association (positive or negative) can vary by state, e.g. male or female for gender. Unconnected nodes indicate conditional independence. See text for additional details. a Genotypes and gene functions. Relationships between recessive and de novo damaging genotypes, and gene lists. Cilia: SysCilia and cilia-related genes; HighHeart: genes highly expressed in the embryonic heart; TGF-βː TGF-β pathway associated genes; Notch: Notch-pathway associated genes; Chromatin: chromatin-modifying and related genes; Cytoskeletal: cytoskeleton-related genes. b Genotype classes and phenotypes. Relationships between proband phenotypes, and recessive and de novo damaging genotypes in cilia and chromatin-related genes, respectively. HTX: heterotaxy; CTD: conotruncal defects; LVO: left ventricular outflow defects; OTH: other phenotypes. Dotted lines denote mutually exclusive categories