Compound heterozygous splicing variants expand the genotypic spectrum of EMC1-related disorders

Abstract

EMC1 encodes subunit 1 of the endoplasmic reticulum (ER) membrane protein complex (EMC), a transmembrane domain insertase involved in membrane protein biosynthesis. Variants in EMC1 are described as a cause of global developmental delay, hypotonia, cortical visual impairment, and commonly, cerebral atrophy on MRI scan. We report an individual with severe global developmental delay and progressive cerebellar atrophy in whom exome sequencing identified a heterozygous essential splice-site variant in intron-3 of EMC1 (NM_015047.3:c.287-1G>A). Whole genome sequencing (WGS) identified a deep intronic variant in intron-20 of EMC1 (NM_015047.3:c.2588-771C>G) that was poorly predicted by in silico programs to disrupt pre-mRNA splicing. RT-PCR revealed stochastic activation of a pseudo-exon associated with the c.2588-771C>G variant and mis-splicing arising from the c.287-1G>A variant. This case highlights the utility of WGS and RNA studies to identify and assess likely pathogenicity of deep intronic variants and expands the genotypic and phenotypic spectrum of EMC1-related disorders.

Graphical Abstract

Introduction

EMC1 encodes subunit 1 of the highly conserved, multi-functional endoplasmic reticulum (ER) membrane protein complex (EMC)¹. Comprised of ten subunits, the EMC plays an important role in the biosynthesis of integral membrane proteins by facilitating insertion of transmembrane domains into the ER membrane². The EMC is suggested to have chaperone activity acting as a quality control checkpoint for transmembrane proteins transiting the secretory pathway, with additional roles related to ER-associated degradation, sterol regulation and autophagy³. Variants in EMC1^4–7 are associated with syndromic neurodevelopmental disorders.

Variants in EMC1 have been described in 12 individuals with early-onset severe global developmental delay, hypotonia, diminished deep tendon reflexes and ophthalmological findings^4–7 (summarized in Figure 1, Table S1), including cortical visual impairment, myopia, optic atrophy and strabismus^4–7.

Figure 1. EMC1 variants associated with global developmental delay, hypotonia, diminished deep tendon reflexes and ophthalmological findings.

Schematic of EMC1 exons (black rectangles) and introns (black line) with exons encoding the quinoprotein alcohol dehydrogenase-like domain (PQQ_2) and domain of unknown function 1620 (DUF1620) labelled. A) i) EMC1 biallelic variants reported by Harel et al. 2016⁴ (orange), Geetha et al. 2018⁵ (green) and Cabet et al. 2020⁷ (purple). Ii) EMC1 variants identified in our proband. c.1034G>C variant is common, with a gnomAD allele frequency of 0.44. B) EMC1 monoallelic variants reported by Harel et al. 2016⁴ (orange) and Chung et al. 2022⁶ (teal). C) Pedigree showing segregation of EMC1 variants in our proband. D) MRI at 6 years of age: i) Sagittal T1-weighted MRI shows mild atrophy of the superior cerebellar vermis ii) Coronal T1-weighted MRI shows mild atrophy of the superior cerebellar hemispheres. E) Muscle histopathology from biopsy at 2 years old; i) NADH-TR (dark fibres type-1) shows mild fibre type variation and mild type-1 hypotrophy, scale 50 μm; ii) electron micrograph of minicore, scale 10 μm F) Fibroblast Western blot showing i) Reduced EMC1 protein expression in proband compared to controls, standard curve of controls to show antibody is in linear range ii) No protective effect of MG132 treatment on EMC1 levels in proband. D: DMSO treated, M: MG-132 treated. EMC1 antibody: A305-604A-M. β-tubulin as loading control, see Figure S2 for total protein.

Variants were biallelic for 8/12 individuals, including missense, splice-site, frameshift and nonsense variants (Figure 1A). The described recessive missense variants resided within two conserved domains of EMC1: quinoprotein alcohol dehydrogenase-like domain (PQQ_2) and domain of unknown function 1620 (DUF1620)⁷ (Figure 1A). All 4 individuals^4,6 with de novo heterozygous missense variants are located within the eight-bladed β-propeller structure of EMC1⁶ (Figure 1B).

We describe a child with biallelic variants in EMC1, including a novel deep intronic single nucleotide variant that activates inclusion of a pseudoexon, to expand the genotypic and phenotypic spectrum of EMC1-related disorders.

Materials and Methods

See supplementary information.

Results

Clinical History

The proband was the first child of NZ Māori/Samoan parents (Figure 1C). She presented at 5 months of age with hypotonia, feeding difficulties and global developmental delay without history of regression. Rolling was achieved at 4 years and sitting at 6 years of age. Speech was severely delayed with only five spoken words.

Cortical visual impairment was present with structurally normal ophthalmic examination. Bilateral cataracts developed at 6 years of age, possibly related to self-injurious behaviour. Severe bulbar dysfunction and aspiration necessitated nasogastric feeding from 5 months, followed by gastrostomy insertion. Severe obstructive sleep apnoea was documented at 3 years of age. She had recurrent chest infections and died at 7 years of age from a severe respiratory illness. She had no history of seizures.

Examination findings included mild brachycephaly and mid-temporal narrowing. Head circumference was on the 50^th centile. She had a high palate and intermittent esotropia but was not otherwise dysmorphic. She had profound hypotonia, decreased muscle bulk and absent deep tendon reflexes. She had mild pectus excavatum, positional kyphosis, but no scoliosis.

Extensive neurometabolic investigation did not yield a clinical diagnosis. MRI at 7 months was normal with repeat MRI at 6 years of age showing mild atrophy of the superior vermis and symmetrical atrophy of the superior aspect of the cerebellar hemispheres (Figure 1D). Nerve conduction studies were normal.

Skeletal muscle histopathology

Collectively, muscle biopsy findings were interpreted as non-specific (Figure 1E, supplementary results, quantified in Table S2).

Genetic investigations and findings

Exome sequencing identified a maternal acceptor splice-site variant in EMC1 intron-3 (Chr1(GRCh37):g.19570202C>T;NM_015047.3:c.287-1G>A), previously reported in ClinVar⁸ as likely pathogenic (VCV000445564.1). With no other likely candidates identified, the trio progressed to WGS which identified a paternal deep intronic variant in EMC1 intron-20 (Chr1(GRCh37):g.19548113G>C;NM_015047.3:c.2588-771C>G). Both EMC1 variants are rare in the Genome Aggregation Database (gnomAD); c.287-1G>A at a frequency of 0.00000398 (1/251444 alleles) and c.2588-771C>G at 0.0000360 (1/27810 alleles)⁹. Further, in 1285 whole-genome sequenced Samoan individuals from the Soifua Manuia study¹⁰, the EMC1 c.287-1G>A variant was absent, and the c.2588-771C>G variant was rare with a frequency of 0.00157 (4/2552 alleles).

One alternate plausible variant was identified by WGS: a homozygous, deep intronic variant in ARV1 (chr1(GRCh37):g.231128933C>A;NM_022786.3:c.449-2573C>A). RNA testing showed no splicing impact (Figure S1). While NM_022786.3:c.449-2573C>A is rare in gnomAD (1/152098 alleles, allele frequency 0.000006575), it is relatively frequent in the Samoan population (127/2570 alleles, 8 homozygotes, allele frequency 0.049416342), with collective evidence indicating this variant is benign (Figure S1/Table S3).

Western blot of EMC1 protein

Western blot of primary fibroblasts shows a ~50% reduction in EMC1 protein expression in the proband compared to two age- and sex-matched controls (Figure 1Fi/Figure S2), with no rescue of EMC1 protein levels in fibroblasts treated with 5μM MG132 for 5 hours to inhibit proteosomal degradation (Figure 1Fii). Validation of three EMC1 antibodies (Table S4) revealed A305-604A-M (Thermo Fisher Scientific) as the only specific antibody (Figure S3). Western blot of skeletal muscle using all EMC1 antibodies was uninformative, with numerous non-specific bands complicating confident interpretation of results (Figure S4).

EMC1 RT-PCR

RT-PCR was performed on mRNA extracted from proband skeletal muscle (Figures 2–3) and fibroblasts (not shown). The maternal c.287-1G>A variant elicits intron-3 retention (r.286_287ins286+1_287-1;p.Asp96Glyfs*45), use of an intron-3 cryptic acceptor (r.286_287ins287-91_287-1;p.Asp96Glufs*22), use of an exon-4 cryptic acceptor (r.287_310del;p.Asp96_Gly103del) and exon-4 skipping (r.287_380del;p.Asp96Valfs*14) (Figure 2Ai–v). The r.287_310del transcript removes 8 amino acids from the PQQ_2 domain of the EMC1 subunit. All other abnormal splicing outcomes encode a premature termination codon (PTC) resulting in nonsense mediated decay (NMD). A benign, maternal variant in exon-10 (c.1034G>C) was used to phase splicing events with Sanger sequencing of correctly spliced amplicons spanning exons 2–11 showing they arise exclusively from the paternal allele (only bear the reference c.1034G, Figure 2Bvi), indicating all maternal transcripts are disrupted by the c.287-1G>A variant.

Figure 2. RT-PCR of the EMC1 c.287-1G>A variant.

A) Schematic of EMC1 exons 2–6 (black rectangles) and sequence context of the NM_015047.3:c.287-1G>A variant. SpliceAI¹⁶ delta scores for annotated and cryptic acceptor splice-sites (green). Branchpoints (red) were predicted using SpliceSiteFinder-like¹⁷. B) RT-PCR using muscle RNA from the proband. i) Primers in exons-2 and −5 demonstrate four mis-splicing events in the proband absent in controls. Asterisks: heteroduplex of intron-3 retention, use of intronic cryptic acceptor and normal splicing. ii) Primers in intron-3 and exon-5 confirm c.287-1G>A variant-associated intron-3 retention arises exclusively from the maternal allele. iii) Use of the intronic cryptic acceptor in the proband. iv) Use of the exonic cryptic acceptor in the proband. v) Exon-4 skipping. Sequencing chromatograms are of exon skipping band at 220 base pairs in i). vi) Reduced levels of normal splicing for the proband relative to controls. Sanger sequencing demonstrates absence of the maternal c.1034G>C variant, indicating no normally spliced maternal transcripts. P: Proband, C1: Control 1, C2: Control 2, C3: Control 3.

Figure 3. RT-PCR of the EMC1 c.2588-771C>G variant.

A) Schematic of EMC1 exons 19–22 (black rectangles) showing NM_015047.3:c.2588-771C>G and the activated pseudo-exon sequence. SpliceAI¹⁶ delta scores are shown for cryptic acceptor (green) and donor splice-sites (dark blue). Branchpoint (red) was predicted using SpliceSiteFinder-like¹⁷ and the variant created SRp55 binding motif (yellow) was predicted using ESEFinder¹¹. B) RT-PCR of c.2588-771C>G using muscle RNA. i) Activation of the pseudo-exon in the proband, absent in controls. Normal splicing is reduced in the proband relative to controls. A band corresponding to a heteroduplex of pseudo-exon inclusion and normal splicing can be seen in the proband. ii) A primer in the pseudo-exon confirms pseudo-exon activation in the proband. iii) Reduced levels of normal splicing for the proband relative to controls. Sanger sequencing demonstrate heterozygosity of the maternal c.1034G>C variant, indicating that normally spliced paternal transcripts were present. iv) GAPDH primers demonstrate equal loading. P: Proband, C1: Control 1, C2: Control 2, C3: Control 3.

ESEFinder¹¹ predicts the paternal c.2588-771C>G variant creates a strong exonic splicing enhancer (ESE) that recruits the splicing factor SRp55 (ESEFinder score: 4.59). In addition, SpliceAI¹² predicts c.2588-771C>G enhances likelihood of use of flanking cryptic acceptor and donor splice-sites (delta-scores of 0.10 and 0.01 respectively, below the high sensitivity 0.2 threshold¹², Figure 3A). RT-PCR confirms c.2588-771C>G activates inclusion of a 118 nt pseudo-exon from EMC1 intron-20 (r.2587-2588ins2588-780_2588-663;p.Ile863Lysfs*27) (Figure 3Bi–ii), encoding a PTC and predicted to result in NMD. RT-PCR to amplify transcripts with normal exon 20–21 splicing (exon-9 forward primer, reverse primer bridging exon 20–21 splice junction) detects paternal and maternal transcripts, showing the pseudo-exon is not incorporated into all paternal transcripts (Figure 3Biii).

Discussion

Variants in EMC1 have recently been implicated as a cause of severe global developmental delay, with additional features including cortical visual impairment, scoliosis, seizures, and cerebellar atrophy^4,5,7. We expand on these findings by characterising compound heterozygous EMC1 splicing variants identified in a child with severe global developmental delay, hypotonia, cortical visual impairment, cataracts, bulbar dysfunction, severe respiratory disease, and progressive cerebellar atrophy. The maternal EMC1 intron-3 acceptor splice-site variant (c.287-1G>A) causes mis-splicing of all transcripts, with the majority of mis-spliced transcripts encoding a PTC. The paternal intron-20 variant (c.2588-771C>G) is a hyomorphic allele associated with pseudo-exon inclusion (encoding a PTC) in most but not all EMC1 transcripts. Application of the ACMG-AMP guidelines¹³ classifies the c.287-1G>A variant as pathogenic (PVS1_Strong, PS3_Strong, PM2_Supporting, PP4) and the c.2588-771C>G variant as likely pathogenic (PS3_Moderate, PM3, PM2_Supporting, PP4), expanding the genotypic spectrum of EMC1-related disorders.

The c.2588-771C>G deep intronic variant likely activates pseudo-exon inclusion via creation of an ESE that recruits SRp55: a splicing factor known to regulate alternative splicing¹⁴ and previously implicated in pseudo-exon activation in DMD¹⁵. SRp55 has variable expression in different human tissues¹⁴, therefore it is plausible that levels of alternative splicing of the pseudo-exon may vary in different tissues of the proband. Western blot of fibroblasts reveals reduced levels of apparently full length EMC1 protein, while expression levels in skeletal muscle were inconclusive. CRISPR/Cas9 knockout studies demonstrate 2/3 commercially available EMC1 antibodies are not specific, making it difficult to conclusively interpret the effects of mis-spliced transcripts on protein expression. However, residual levels of normally spliced EMC1 transcripts and EMC1 protein may be a contributing factor to the phenotypic differences observed in the proband compared to previously reported individuals. For example, normally spliced transcripts arising from the paternal allele may have spared the proband from seizures, which were reported in individuals harbouring two EMC1 variants that result in PTCs^4,5. Additional tissue samples to test this hypothesis were unavailable.

To our knowledge, this is the first, pathogenic, deep intronic EMC1 variant. Upon identification of the maternal c.287-1G>A by exome sequencing, clinical opinion asserted EMC1 as a phenotypically concordant gene. Our study highlights benefits of using WGS to identify the second variant for ‘single hit’ cases, and also highlights the importance of interrogating candidate genetic variants in population-relevant databases, which was vital to confidently assert the homozygous ARV1 variant NM_022786.3:c.449-2573C>A as benign.

Data Availability Statement

Genotypes calls from the whole-genome sequencing for the Soifua Manuia study, labeled “NHLBI TOPMed: Genome-wide Association Study of Adiposity in Samoans” are available in dbGaP (phs000972.v4.p1).