Compound heterozygous mutations in the noncoding RNU4ATAC cause Roifman Syndrome by disrupting minor intron splicing

Abstract

Roifman Syndrome is a rare congenital disorder characterized by growth retardation, cognitive delay, spondyloepiphyseal dysplasia and antibody deficiency. Here we utilize whole-genome sequencing of Roifman Syndrome patients to reveal compound heterozygous rare variants that disrupt highly conserved positions of the RNU4ATAC small nuclear RNA gene, a minor spliceosome component that is essential for minor intron splicing. Targeted sequencing confirms allele segregation in six cases from four unrelated families. RNU4ATAC rare variants have been recently reported to cause microcephalic osteodysplastic primordial dwarfism, type I (MOPD1), whose phenotype is distinct from Roifman Syndrome. Strikingly, all six of the Roifman Syndrome cases have one variant that overlaps MOPD1-implicated structural elements, while the other variant overlaps a highly conserved structural element not previously implicated in disease. RNA-seq analysis confirms extensive and specific defects of minor intron splicing. Available allele frequency data suggest that recessive genetic disorders caused by RNU4ATAC rare variants may be more prevalent than previously reported.

Figure 1. Pedigree of the six Roifman Syndrome cases (kindred 1–4).

The pedigrees show the RNU4ATAC compound heterozygous SNVs in the six genotyped cases of Roifman Syndrome; [=] indicates no variant detected.

Figure 2. Facial, retinal and skeletal features of Roifman Syndrome.

Facial dysmorphic features (a–c) include a markedly long philtrum, thin upper lip and down-turned corners of the mouth. Retinal features are displayed for patient 3 at age 4 years: arteriolar attenuation (d, black arrow), wrinkling of the inner limiting membrane (d, short white arrow and e, black arrows) and pigmentary changes (d, long white arrow). Skeletal features are displayed for patients 1, 3 and 5, respectively: the radiographs show the proximal epiphyses of the femora with symmetric delayed ossification, as well as flattening and irregularity.

Figure 3. U4atac snRNA secondary structure elements, positional conservation, MOPD1 and Roifman Syndrome causal variants.

Elements of limited or null importance for splicing (dispensable region of 3′ stem–loop) have mutagenesis experiments showing no splicing alteration, are enriched in low conservation and do not present any known disease-causing variant. Elements of variable importance for splicing (3′ stem–loop, except dispensable region; single-strand region; 5′ stem–loop, except critical region) have mutagenesis experiments showing modest or no splicing alteration, or have not been probed by mutagenesis, but meet at least one of these criteria: (a) they present at least one MOPD1 causal variant (typically with reduced severity); (b) they are proximal to a MOPD1 causal variant cluster; (c) structural studies suggest they may have a functional role; these elements have mixed conservation, and only a few variants at more conserved positions may cause splicing alterations. Elements of major importance for splicing (stem II, critical region of the 5′ stem–loop, stem I, Sm protein-binding site) have mutagenesis experiments producing splicing alterations and/or overlap the MOPD1 variant cluster; in addition, they are expected to have a major functional role based on structural studies; finally, they are enriched in highly conserved positions, the majority of which are expected to cause splicing alterations in presence of variation. Positions are labelled as ‘high conservation' if placental mammal or 100-vertebrate PhyloP ≥1.75, as ‘diverged' if placental mammal and 100-vertebrate PhyloP are negative, and ‘moderate or low conservation' otherwise. Parts of U6atac are displayed only in correspondence of U4atac–U6atac duplex structures. U4atac snRNA coordinate 1 corresponds to hg19 coordinate 122,288,456. Classification of MOPD1 causal variants as ‘full severity' or ‘reduced severity' is based on a thorough review of MOPD1 literature (for more details, see Supplementary Table 17) and biochemical assays of variant effect; the latter is particularly important for less severe MOPD1 forms presenting compound heterozygosity.

Figure 4. Summary of RNA-seq findings.

(a) PIR for major (grey) and minor (orange) introns, for kindred 1 and kindred 2 subjects, showing specific minor intron retention in affected subjects compared with unaffected subjects. (b) Affected–unaffected average difference in percentage alternative splicing (dPSI) for different alternative splicing categories (ALTA, alternative 5′ splice site; ALTD, alternative 3′ splice site; EX, cassette exon; INT, major intron retention; INTmin: minor intron retention; MIC, micro-exon); significant deviation from 0 is observed only for major introns (small yet significant reduction in affected) and minor introns (significant increase in affected). (c) Log2 of the mean expression ratio between affected and unaffected, for minor intron genes and all other genes, displaying a slight shift of minor intron genes towards higher expression in affected subjects. (d) Scatterplot showing that increased expression in affected relative to unaffected subjects (y axis, log2 ratio of the mean expression in the two groups) is insufficient to compensate for the increased minor intron retention (x axis, log2 ratio of the mean correctly spliced fraction in the two groups), with 30% of the genes over-compensated and 70% of the genes under-compensated (separated by the red dashed line); the linear correlation between the expression ratio and correctly spliced ratio is negative but modest (grey dashed line).