Noncoding variants are a rare cause of recessive developmental disorders in trans with coding variants

Abstract

Purpose: Identifying pathogenic noncoding variants is challenging. A single protein-altering variant is often identified in a recessive gene in individuals with developmental disorders (DD), but the prevalence of pathogenic noncoding "second hits" in trans with these is unknown.

Methods: In 4073 genetically undiagnosed rare-disease trio probands from the 100,000 Genomes project, we identified rare heterozygous protein-altering variants in recessive DD-associated genes. We identified rare noncoding variants on the other haplotype in introns, untranslated regions, promoters, and candidate enhancer regions. We clinically evaluated the top candidates for phenotypic fit and performed functional testing where possible.

Results: We identified 3761 rare heterozygous loss-of-function or ClinVar pathogenic variants in recessive DD-associated genes in 2430 probands. For 1366 (36.3%) of these, we identified at least 1 rare noncoding variant in trans. Bioinformatic filtering and clinical review, revealed 7 to be a good clinical fit. After detailed characterization, we identified likely diagnoses for 3 probands (in GAA, NPHP3, and PKHD1) and candidate diagnoses in a further 3 (PAH, LAMA2, and IGHMBP2).

Conclusion: We developed a systematic approach to uncover new diagnoses involving compound heterozygous coding/noncoding variants and conclude that this mechanism is likely to be a rare cause of DDs.

Keywords: Clinical genetic testing; Genomics; Non-coding variants; Rare disorders; Recessive disorders.

Figure 1. The number of candidate variants identified per proband.

A. Bar plot of the number of single pLoF and/or ClinVar (likely) pathogenic variants per individual. B. Bar plot of the number of candidate non-coding second hit variants per proband-variant pair. The x-axis is truncated at 15. Four proband-variant pairs had >15 noncoding second hit variants with counts of 16, 20, 25, and 38.

Figure 2. An overview of our approach and the distribution of 52 candidate noncoding second hits across different regions.

A. Details of our search focusing on 793 genes with a single pLoF or ClinVar (likely) pathogenic variant in 4073 undiagnosed trio probands. The different candidate noncoding regions analyzed for a second hit are shown in color. B. Count of candidate variants clinically curated as “probable,” “possible,” or “unlikely,” split by region and annotations. The indicated genes are those that were assessed to be a “probable” fit. C. Upset plot of region-specific annotations used to prioritize candidate variants. This plot does not include 2 additional probands who had 2 compound heterozygous ClinVar pathogenic/likely pathogenic variants, 1 protein altering and 1 noncoding.

Figure 3. RNA-seq to interrogate candidate variants in GAA and NPHP3.

A. RNAseq reads covering nonsense variant ENST00000302262.8:c.2577G>A (p.Trp859Ter) in GAA. Forty-six reads carry the reference allele, whereas 19 carry the alternative allele. B. Sashimi plot showing splicing in the GAA proband plus 2 controls from the same sequencing batch. Skipping of exon 3 (ENST0000390015.7) is observed in the proband but not in controls (dashed line, black arrow) due to the variant 13 bp upstream of the exon 3 splice acceptor site (black “x”). C. Normalized expression level per intron for the proband (red) plus 29 controls (gray) sequenced in the same batch. Proband has higher intronic coverage than all controls for intron 2, suggesting that intron retention may be occurring. D. RNAseq reads covering near-splice variant ENST00000337331:c.3570+5G>A (chr3:132684549:C:T) in NPHP3, 5 bp from the splice donor site of exon 24, in a different proband identified as a heterozygote with the same noncoding variant as our proband. Six reads carry the reference allele, whereas 9 carry the alternative allele. E. Sashimi plot showing splicing in the individual heterozygous for the ENST00000337331:c.3570+5G>A (chr3:132684549:C:T;NPHP3) variant and 2 control individuals without the variant. Skipping of exon 24 (ENST00000337331) is observed (highlighted by black arrow) in the individual with the variant but not the controls (relative location of variant indicated by black “x”).