Genome sequencing reveals the impact of pseudoexons in rare genetic disease

Abstract

Purpose: Advancements in sequencing technologies have significantly improved clinical genetic testing; yet, the diagnostic yield remains around 30% to 40%. Emerging technologies are now being deployed to address the remaining diagnostic gap.

Methods: We tested whether short-read genome sequencing could increase the diagnostic yield in individuals enrolled into the UCI-GREGoR research study, who had suspected Mendelian conditions and prior inconclusive testing. Two other collaborative research cohorts, focused on aortopathy and dilated cardiomyopathy, consisted of individuals who were undiagnosed but had not undergone harmonized prior testing.

Results: We sequenced 353 families (754 participants) and found a molecular diagnosis in 54 (15.3%) of them. Of these diagnoses, 55.5% were previously missed because the causative variants were in regions not originally interrogated. In 5 cases, they were deep intronic variants, all of which led to abnormal splicing and pseudoexons, as directly shown by RNA sequencing. All 5 of these variants had inconclusive spliceAI scores. In 26% of newly diagnosed cases, the causal variant could have been detected by exome sequencing reanalysis.

Conclusion: Genome sequencing can overcome limitations of clinical genetic testing, such as the inability to call intronic variants. Our findings highlight pseudoexons as a common mechanism via which deep intronic variants cause Mendelian disease.

Keywords: Genome sequencing; Intronic variants; Pseudoexon; RNA sequencing; Rare disease.

Figure 1.

(A), (B) Sex and race distribution of probands that underwent genome sequencing.

Figure 2.

(A) Family structures corresponding to the participants (B). Trio/Trio+ had the highest diagnostic yield (21%) whereas Duo/Duo+ and Proband only families reached a diagnostic yield of 13% and 10% respectively (p = 0.03, chi-square test).

Figure 3.

(A) Types of prior testing undergone by participants that were deemed solved or probably solved (*1 patient had methylation studies in addition to panel testing; ** 3 patients had mitochondrial testing in addition to ES/GS and panel testing). (B) Reasons diagnosis was previously missed on clinical genetic testing. Limited original testing includes gene panels that lacked the causative gene or did not detect the causative variant due to its location in a genomic region not covered by the test. New information/Reclassification refers to variants reclassified due to new papers, or inclusion of ClinGen updates to the ACMG guidelines such as increasing strength for high revel scores, or further phenotypic evidence which allowed further variant criteria to be applied. (C) Pie chart showing the percentage of probands that had undergone exome or genome sequencing; percentages are calculate across the entire cohort (not only solved cases).

Figure 4.

(A) RNA sequencing reads visualized with sashimi plot to indicate splicing events showing the pseudoexon inclusion in DBT in the proband (PMGRC-658–658-0) and the affected sibling as a result of a deep (−4218) homozygous intronic single nucleotide variant. (B-E) Mini-gene splicing assay for the DBT c.1282–4218G>A variant. B) Reference and c.1282–4218G>A variant DBT intron 10 fragments were cloned into a mini-gene split GFP construct, in which N and C-terminal parts of the GFP gene were separated by SMN1 introns 7 and 8 (NM_000344.4). The construct was expressed in HEK293 cells for 48 hours followed by mRNA extraction, cDNA generation and NGS amplicon sequencing. Black arrows on the construct image indicate primer placement. C) RT-PCR gel electrophoresis demonstrates inclusion of the pseudoexon in both the wild type and reference transcript. D) Sashimi plots of reference and variant hybrid DBT-GFP constructs demonstrating a preferential pseudoexon (10B) inclusion in the sequences containing the c.1282–4218G>A variant. E) Inclusion of pseudoexon 10B leads to a premature stop codon at the end of the exon, which most likely escapes nonsense-mediated decay and leads to a truncated protein with a C-terminus different by 16-residues.

Figure 5.

(A) The solve rate for syndromic versus non-syndromic cases. (B) The solve rate stratified by phenotypic category for the non-syndromic cases.