Non-coding de novo variants (DNVs) contribute to congenital heart disease (CHD) through transcriptional and post-transcriptional regulatory effects during cardiac development, according to a new study. Moreover, the proportion of individuals with CHD ascribed to non-coding DNVs might be at least as high as that with CHD attributed to coding DNVs.

CHD is the most common congenital disorder in humans, occurring in 1% of live births. A genetic cause is identified in 33% of patients with CHD but only 8% are attributed to coding DNVs. Thus, Felix Richter and colleagues hypothesized that additional variants causing CHD might be located in non-coding elements that are active during cardiac development. To examine this hypothesis, the research team compared genome sequences from 749 probands with CHD without identified probable causal genetic variants and their unaffected parents with those from 1,611 child–parent trios without CHD. The researchers used three strategies: two transcription-based approaches centred on cardiac gene regulatory elements and an analysis of post-transcriptional regulation. “We ensured cardiac relevance with a large corpus of publicly available and newly generated cardiac epigenomic data,” explains Richter.

Credit: Kiyoshi Takahase Segundo/Alamy Stock Photo

One strategy involved a neural network algorithm that could predict functional effect differences with variant-level resolution. A second approach involved the analysis of non-coding DNVs on enhancer regions that had been implicated in human cardiac development gene expression regulation in experimental studies. The neural network identified a significant enrichment of non-coding DNVs in patients with CHD compared with controls. The enhancer analysis showed that 27 genes were marginally enriched for DNVs among patients with CHD, whereas no gene was enriched for DNVs in controls. Both approaches showed a significant overlap between results. Of note, of the CHD-associated genes identified by both approaches, only COL1A2 had been previously implicated in heart development. Functional validation assays showed that five of 31 DNVs analysed significantly altered transcription levels in the associated genes, which included JPH2 (encoding a membrane protein necessary for transverse-tubule formation) and SEMA4B (which is in the top quartile for gene expression in the developing heart).

The third approach focused on RNA processing and showed DNV enrichment in RNA-binding-protein regulatory sites in individuals with CHD compared with controls. Finally, Richter and colleagues assessed whether the non-coding DNVs were associated with phenotypic CHD subgroups. The analysis revealed potentially contributory non-coding DNVs in probands with isolated CHD and in those with neurodevelopmental delays or extracardiac anomalies, suggesting varying degrees of cardiac specificity of the DNVs.

cardiac regulatory non-coding DNVs contribute to CHD pathogenesis at the transcriptional and post-transcriptional regulatory levels

Taken together, these analyses indicate that cardiac regulatory non-coding DNVs contribute to CHD pathogenesis at the transcriptional and post-transcriptional regulatory levels. “This work highlights a continued need to perform whole-genome sequencing in larger cohorts, obtain more robust and diverse cardiac epigenomic data, and develop algorithms to understand non-coding genetics,” says Richter.