A novel hypomorphic splice variant in EIF2B5 gene is associated with mild ovarioleukodystrophy

Abstract Objective To identify the genetic cause in an adult ovarioleukodystrophy patient resistant to diagnosis. Methods We applied whole‐exome sequencing (WES) to a vanishing white matter disease patient associated with premature ovarian failure at 26 years of age. We functionally tested an intronic variant by RT‐PCR on patient’s peripheral blood mononuclear cells (PBMC) and by minigene splicing assay. Results WES analysis identified two novel variants in the EIF2B5 gene: c.725A > G (p.Tyr242Cys) and an intronic noncanonical mutation (c.1156 + 13G>A). This intronic mutation resulted into generation of various isoforms both in patient’s PBMC and in the minigene splicing assay, showing that ~20% residual wild‐type isoform is still expressed by the intronic‐mutated allele alone, concordant with an hypomorphic effect of this variant. Conclusion We report two novel variants in EIF2B5, one of them a noncanonical intronic splice variant, located at a +13 intronic position. This position is mutated only in 0.05% of ClinVar intronic mutations described so far. Furthermore, we illustrate how minigene splicing assay may be advantageous when validating splice‐altering variants, in this case highlighting the coexistence of wild‐type and mutated forms, probably explaining this patient’s milder, late‐onset phenotype.


Introduction
Vanishing white matter disease (VWMD; OMIM #603896) is a leukodystrophy caused by recessive mutations in any of the five genes encoding subunits of translation initiation factor EIF2B. Manifestations usually start between late infancy and early childhood, and mainly consist of pyramidal and cerebellar signs with mental decline and episodes of acute deterioration following stressors. However, 15% of cases are adult-onset forms associated with a more benign course and fewer decompensations. 1 Infrequently, VWMD can appear associated with premature ovarian failure (POF), a clinical syndrome that has been called ovarioleukodystrophy. 2 Although EIF2B1-5 mutations are the main cause of leukodystrophy with POF, other genes have been described. 3 More than 120 mutations have been reported in >250 patients with an EIF2B-related disorder, mostly in EIF2B5 and EIF2B2. 4 No mutational hotspots have been found, although some recurrent mutations seem to occur in paired cytosine/guanine (CpG) dinucleotides. The vast majority of pathogenic mutations are missense, whereas truncating mutations (frameshifts, nonsense, splice site mutations) are rare and have been reported only in compound-heterozygous state, 5 indicating that total loss of function may be incompatible with life.
Here, we report a patient with an adult-onset ovarioleukodystrophy carrying two novel disease-causing variants in EIF2B5 identified by WES, one of which is an intronic mutation leading to activation of a cryptic splice donor site. The impact of this rare, noncanonical splice variant on EIF2B5 pre-mRNA processing was evaluated both in cDNA from patient's PBMC, and using minigene splicing reporter assays. These experiments revealed residual wild-type splicing, probably accounting for the mild late-onset clinical phenotype in this patient.

Participant and ethics
Blood was processed by centrifugation within 2 h of collection using a gradient of Histopaque to separate plasma, erythrocytes, and PBMC. Plasma and PBMC were stored at À80°C. The use of all samples was approved by the Clinical Research Ethics Committee of the Bellvitge University Hospital (PR076/14). Informed written consent was obtained from all patients and control individuals.

Exome sequencing and variant calling
In-solution exome capture was performed using the Seq-Cap EZ Human Exome Kit v3.0 (Roche Nimblegen, USA) with 100-bp paired-end read sequences generated on a HiSeq2000 (Illumina, Inc. USA) in the Centro Nacional de An alisis Gen omico in Barcelona (CNAG). Sequence processing was carried out with BWA aligner, the Genome Analysis Toolkit (GATK), SAMtools, and Picard Tools as previously described. 6 More information regarding methods are detailed in the Data S1.

Clinical findings
A 26-year-old woman was referred to the neurology department because of a leukoencephalopathy detected during the study of amenorrhea with hyperprolactinemia (108 ng/mL). She was born from nonconsanguineous parents and she had one asymptomatic sister. Physical exam showed right hand clumsiness, dystonic left foot postures, and generalized hyperreflexia in absence of cognitive, behavior or psychiatric symptoms. MRI evidenced T2 periventricular and pontine white matter hyperintensities with malacic areas in the frontal and atrial horns. Cortical-subcortical, spinal, and corpus callosum atrophy were also reported (Fig. 1A). Magnetic resonance spectroscopy was unremarkable. In the next 5 years, she developed unsteady gait, more evident pyramidal signs and lower limb paresthesia and cramps. Her cognitive level remained normal and she did not develop seizures. MRI performed four years after the initial one did not show significant changes. Routine hematology and clinical chemistry tests, as well as thyroid and adrenal function, were normal. Nerve conduction tests and electromyography were normal, whereas visual-evoked potentials showed bilateral latency enlargement of the P100 wave and somatosensory-evoked potentials exhibited bilaterally prolonged latencies with normal amplitudes.

WES analysis
Whole-exome sequencing (WES) revealed two suspicious novel variants in EIF2B5. The first one is a missense variant (NM_003907:c.725A > G; p.Tyr242Cys) referenced in dbSNP database as rs750767613, and has a frequency of 3,98.10e-6 in gnomAD database, with 0 homozygotes. This ultrarare missense mutation alters a highly conserved residue, and is predicted to be damaging by SIFT and PolyPhen-2. The second variant is intronic (NM_003907: c.1156 + 13G>A) and is predicted by different algorithms to increase the use of a cryptic donor splice site located at positions + 11/+12 (NNSplice, MaxEntScan, FSplice). Conversely, Human Splicing Finder and SpliceView did not predict a significant impact on splicing (Table S1).   This variant is absent from gnomAD database. Cosegregation analysis revealed that the patient inherited these variants in trans (Fig. 1B). Despite their rare frequency, cosegregation, conservation of affected amino acids, in silico predictions for missense and splicing variants, and the patient's highly concordant phenotype, these variants were classified as variants of unknown significance (VUS) according to the ACMG/AMP (American College of Medical Genetics and Genomics) guidelines for variant evaluation. 7,8 In silico analysis of previously reported splicing variants So far, 11998 single-nucleotide variants probably affecting splicing and considered as pathogenic or likely pathogenic are listed in the ClinVar database. Splice-site mutations are most commonly detected at the G(+1), T(+2), G(+5), A (À2), G(À1) canonical sequences of introns, as shown in Figure S1: +1 (4273 hits), À1 (2484), À2 (2181), +2 (1429), and +5 (551). Frequencies then decrease dramatically as distance from the acceptor/donor splice site increases. Only two variants have been reported in position +13: the A > G transition in intron 10 of the MAPT gene, associated with familial frontotemporal dementia with parkinsonism, 9 and the C > T transition in FGB gene's intron 6, associated with congenital afibrinogenemia. 10 Functional assays Given c.1156 + 13G>A variant's absence from databases and the discordant results of splicing predictors, we evaluated the functional significance of this variant in vitro. Reverse-transcription PCR encompassing exons 7-8 of the EIF2B5 mRNA produced a single 193 bp product from the control samples, whereas two bands (193 and 203 bp) were observed in the patient's sample. Sanger sequencing showed that the longest product included the first 10 nucleotides of intron 7. This 10bp insertion creates two adjacent premature termination codons (TGA and TAA), leading to a truncated protein ( Fig. 1C and D). To confirm the association between the c.1156 + 13G>A EIF2B5 variant and partial retention of intron 7, we performed a minigene splicing assay. This technique consists of the construction of an expression vector containing a minimal gene fragment encompassing the variant sequence of interest along with flanking intronic sequences, and then is transfected into cultured cells in order to evaluate splicing patterns. 11,12 This strategy allowed us to study the monoallelic effect of this EIF2B5 variant compared to a wildtype situation, and to evaluate more precisely the influence of the G > A change on the use of the identified donor cryptic splicing site. Two splicing reporter vectors were constructed in the background of the pSplice-POLR2G plasmid 13 : a small construction containing EIF2B5's exon 7 and flanking intronic sequences (143 bp of intron 6202 bp of intron 7), and a large construction encompassing EIF2B5's exons 7 and 8 and the whole intron 7 sequence. Both vectors, with either the wild-type or the mutated sequence, were transfected into HEK293T/17, U-251MG, and COS-7 cell lines. Results obtained from the smaller pSplicePOLR2G-EIF2B5 construct in HEK293T cells show that the 10 intronic bp insertion was mainly, but not only, observed with the mutated c.1156 + 13G>A allele (Fig. 1E). Similar results were observed in U-251MG and COS7 cells (Fig. S2A-C). Semi-quantitative fluorescent RT-PCR revealed a normal/abnormal-splicing ratio of 0.9 for the wild-type allele and 0.2 for the mutated allele (547 bp peak area/557 bp peak area) (Fig. 1F). Similar results were obtained from the larger pSplice-POLR2G-EIF2B5 construct. These results indicate that: (1) a cryptic donor splicing site is active in EIF2B5 intron 7 and is moderately used by the splicing machinery in different cell types; and (2) this cryptic splicing site is significantly more active in the c.1156 + 13G>A EIF2B5 pre-mRNA where it overcomes use of natural donor's splicing site.
In conclusion, we have identified a new intronic variant in the EIF2B5 gene which activates acryptic 5' donor splice site of intron 7, probably leading to synthesis of a truncated protein (if not degraded by nonsense-mediated mRNA decay). This splicing variant is hypomorphic, as it leads to a residual 20% of WT splicing, and should be considered pathogenic after applying the ACMG criteria. 7,8 In consequence, as VUS missense variant p.Tyr242Cys is in trans with the pathogenic intronic variant, this missense is reclassified as Likely Pathogenic 7,8 and our case is solved.

Discussion
Splicing mutations represent approximately one-third of disease-causing mutations, most of them occurring in conserved consensus splice sites. 14 However, the widespread use of high throughput technologies has increased the detection of variants in cryptic splice sites, which represent around 10% of total splicing mutations according to ClinVar data. In these cases, functional analysis is mandatory to confirm variant's pathogenicity since, as in the case here reported, splicing predictors may provide discordant results. Our case also underscores the importance of thorough clinical evaluation guiding the bioinformatic analysis, to uncover variants in intronic regions. As incomplete coverage of intron-exon boundaries in WES ª 2020 The Authors. Annals of Clinical and Translational Neurology published by Wiley Periodicals LLC on behalf of American Neurological Association studies may hamper detection of intronic variants, WGS may still be required to unravel some cases.
To the best of our knowledge, only four splicing variants have been reported in EIF2B genes so far, [15][16][17][18] of which only one is noncanonical, although it was not functionally validated. 15 Thus, our work is the first functional validation report of a novel, noncanonical splicing mutation in EIF2B5 associated with ovarioleukodystrophy. In this case, the c.1156 + 13G>A hypomorphic mutation generates a novel splice site 10 bp beyond, that competes with the wild-type splicing donor site, therefore generating two coexisting populations of transcripts, the mutated, and the WT form. Our work has benefited from the capacity of the minigene assay to quantify the mRNA production in an allele-specific manner. Indeed, minigene splicing assays have several advantages, such as: (1) circumventing use of patient's RNA if sample is not available; (2) analysis and quantification of the splicing outcome of mutant alleles without interference of the wild-type allele; (3) high reproducibility of results; and (4) testing of variants located in any human disease gene, independently of gene expression. This approach allowed us to observe that the cryptic splice site activated by c.1156 + 13G>A is moderately functional in the wild-type allele, and also that the mutated allele expresses a residual amount of wild-type mRNA. Competition between the wild-type and the cryptic splice donor site could determine the expression of some degree of functional EIF2B5 protein and consequently, the mild phenotype exhibited by the patient. Our finding of a partial mis-splicing is important in view of potential therapies since modulation of splice processes represents a therapeutic approach for some genetic diseases. 19 Even if ovarioleukodystrophy seems to be an infrequent cause of pure premature ovarian failure (POF), 20 identification of hyperprolactinemia or amenorrhea should prompt both an exhaustive neurological examination and, possibly, the performance of a cranial MRI. If clinical and neuroimaging studies are suggestive, WES or even WGS are recommended to detect intronic mutations in EIF2B genes that could be responsible for mild phenotypes.