Novel missense variants in the RNF213 gene from a European family with Moyamoya disease

In this report, we present a European family with six individuals affected with Moyamoya disease (MMD). We detected two novel missense variants in the Moyamoya susceptibility gene RNF213, c.12553A>G (p.(Lys4185Glu)) and c.12562G>A (p.(Ala4188Thr)). Cosegregation of the variants with MMD, as well as a previous report of a variant affecting the same amino acid residue in unrelated MMD patients, supports the role of RNF213 in the pathogenesis of MMD.


Clinical presentation
The female proband (III:9, Figure S5) presented at the age of 4 years with two episodes of transient left sided weakness and aphasia, brought on by excitement. Both episodes resolved completely. She also had a long-standing history of alternate day headache, without other associated features. She underwent brain magnetic resonance imaging, and subsequently cerebral angiography (Supplementary Figure S1) that confirmed the diagnosis of bilateral Moyamoya disease (MMD). She was commenced on aspirin, underwent right sided pial synangiosis and has subsequently remained clinically and radiologically stable for over 5 years. Of note, there was a positive family history of MMD in at least 2 generations on her paternal side ( Figure S5). Her paternal grandmother (I:2) was diagnosed with MMD and died aged 65 years. Her father (II:8) had brain imaging for investigation of epilepsy that did not show evidence of MMD. Her paternal aunt (II:1) had 3 children; she was unaffected as was her youngest son (III:3). However her other 2 children (III:1 and III:2) had a diagnosis of MMD, having presented with stroke aged 5 and a movement disorder, respectively. The paternal uncle (II:5, clinically unaffected) had a child who died from a brain haemorrhage aged 7 years although no formal diagnosis of MMD is recorded. Another paternal aunt (II:7) has a diagnosis of MMD, having suffered a stroke at the age of 21. Finally the other paternal aunt, (II:3) had 3 sons. One of the sons (III:4) was under investigation for learning difficulties, without current diagnosis, and the remaining sons (III:5 and III:6) were in good health.

Methods
Diagnosis of Moyamoya disease: Diagnosic criteria of MMD by the Research Committee on the Pathology and Treatment of Spontaneous Occlusion of the Circle of Willis (Moyamoya disease) in Japan (Fukui, 1997;Hashimoto et al., 2012) were used for diagnosis of MMD in the affected members of the family and include all of the following items based on the conventional angiographic findings: (i) stenosis or occlusion of the terminal portion of the intracranial internal carotid artery or proximal portions of the anterior cerebral artery and/or the middle cerebral artery, (ii) development of abnormal vascular networks near the occlusive or stenotic lesions in the arterial phase, (iii) bilaterality of findings (i) and (ii).

Variant interpretation:
The modified criteria from the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) guidelines were used for variant classification (Richards et al., 2015;Gelb et al., 2018). Pathogenicity and benign evidence tags was input into ClinGen Pathogenicity Calculator (Patel et al., 2017) for assessing pathogenicity of genetic variants.
Sanger sequencing: PCR amplicon sequencing was used to verify the presence of each mutation and to perform segregation analysis. The target region was amplified using primers (forward 5 -TCCACAGTCACATTGGCGACT-3 and reverse 5 -AGCCCAGTAGTAGTCTATTCG-3 ). PCR products were purified with ExoSap and sequenced using Big Dye Terminator Cycle Sequencing Kit v3.1 (Life Technologies, Foster City, CA, USA). Result electropherograms were analyzed with Geneious software (Biomatters Ltd., Auckland, New Zealand).

Variant analysis
To identify a causative variant in the family affected with MMD over three generations, we performed whole exome sequencing of five family members: II:8 (father), II:9 (mother), III:1 (cousin), III:8 (sister) and III:9 (proband). We focused our analysis on rare (allele frequency <0.5% or absent in public databases) variants (SNVs and indels) within RefSeqGene exonic and splice regions, as annotated by ANNOVAR (Wang et al., 2010). We promptly excluded the possibility of MMD being caused in this family by de novo mutations because its occurence in three generations. We also ruled out the possibility of autosomal recessive mode of inheritance because parental consanguinity was absent in the family and MMD occurred in three first cousins from two different marriages. X-linked recessive, Y-linked and mitochondrial modes of inheritance were excluded as well because both sexes were affected (with more affected females than males) and the disease appeared to be transmitted by either sex, including male-to-female. The most compatible mode of inheritance was found to be either X-linked dominant (XLD) or autosomal dominant (AD) since both sexes in subsequenct generations were affected with the same disease.
Because the pedigree was not fully consistent with AD or XLD mode, we suspected incomplete penetrance of the condition, supported by previous studies on MMD inheritance (Mineharu et al., 2006(Mineharu et al., , 2008. Since the affected proband (III:9) most likely inherited the condition from her father (II:8) with a family history of MMD, we assumed that the father was an asymptomatic carrier of a dominant defective allele. Moreover, although the father did not have MMD per se, he had epilepsy which can be one of the clinical presentations of MMD (Cho and Tominaga, 2010;Koizumi et al., 2017). We applied the same assumption to the unaffected sister (III:8) that could have either 0/0 or 0/1 genotype. Assuming AD or XLD mode of inheritance with incomplete penetrance, the unaffected mother (II:9) was the only individual not related to the index case (I:2) and served as a control (genotype 0/0), while the father (II:8), the affected cousin (III:1), the sister (III:8) and the affected proband (III:9) were all case samples.
No variants following XLD mode of inheritance were identified ( Figure S2). The search for the causative genotype under AD mode of inheritance in the affected family members resulted in 20 variants, with 16 of them being deleterious (missense and frameshift variants) ( Figure S2, Table S1). Out of the sixteen variants, two missense variants (Table S1) were detected in the RNF213 gene, previously associated with MMD (Kamada et al., 2011;. Both variants were private to the family (with no presence in public databases) and were also found in the unaffected sister (III:8) ( Figure S3, Table S2) supporting incomplete penetrance of the condition. Beside the RNF213 variants, no other candidates that could explain the clinical phenotype were identified in our analysis.

Variant interpretation
The c.12553A>G (p.(Lys4185Glu)) and c.12562G>A (p.(Ala4185Thr)) variants in affected family members and obligate carriers were observed in the RNF213 gene, which has previously been reported as a major susceptibility gene for MMD (PP4-Supporting) (Kamada et al., 2011;Koizumi et al., 2017), and have not been published to our knowledge. Both variants were absent from NHLBI Exome Sequencing Project (evs.gs. washington.edu/EVS/), 1000 Genomes Project (1000Genomes Project Consortium et al., 2015, or ExAC/gnomAD (Lek et al., 2016) databases (PM2-Moderate) and were private to the family. The RNF213 variants cosegregate with MMD in multiple affected family members (PP1-Supporting) and obligate carriers. Since there are 5 segregations in the family (Supplementary Figure S5), we upgraded PP1 evidence from default supporting to moderate strength (PP1-Moderate) according to the modified ACMG/AMP criteria for RASopathies (Gelb et al., 2018). Multiple lines of computational evidence support a deleterious effect of the c.12553A>G variant on the gene product (PP3-Supporting) (Supplementary Table S6) Table S6) and is listed as "disease causing" in the HGMD database although no classification according to the ACMG/AMP guidelines is provided. Given that the p.(Lys4185Thr) variant does not meet "likely pathogenic"/"pathogenic" criteria, we downgraded PM5 from default moderate to supporting strength (PM5-Supporting). Since there are 2 moderate and 2 supporting pathogenicity evidence tags, p.(Lys4185Glu) variant in the RNF213 gene is therefore interpreted to be likely pathogenic for MMD and acts in a dominant manner (Supplementary Figure S6).  Table S6). We interpret p.(Ala4188Thr) variant in the RNF213 gene as a variant of uncertain significance with respect to MMD due to conflicting/insufficient evidence ( 1 supporting benign evidence tag, 1 moderate and 1 supporting pathogenicity evidence tags) (Supplementary Figure S6). a b Figure S1. Catheter cerebral angiograms (frontal projection) of left (a) and right (b) internal carotid arteries showing severe occlusive disease of both middle and anterior cerebral arteries and typical basal "moyamoya" collaterals in the proband undertaken at the age of 6 years.
Total number of identified variants:

148,682
Number of RefSeqGene exonic and splice region variants:

34,863
Number of variants not present or rare (allele frequencies <0.5%) in public databases:

1,936
Number of variants retained based on genotype requirements for AD mode of inheritance:

46
Number of variants retained based on genotype requirements for XLD mode of inheritance: 0 Figure S2. Variant filtering flowchart. In total 148,682 variants were identified in joint (multi-sample) calling for the sequenced family members. 34,863 variants lied in RefSeqGene exonic and splice regions. Out of these variants, 1,936 were not present or were rare (allele frequencies <0.5%) in public databases. Twenty variants were further retained based on genotype requirements for autosomal dominant (AD) mode of inheritance (see Supplementary Text S1), with 16 variants being deleterious (missense and frameshift variants). Out of the sixteen variants, two were found in the gene (RNF213 ) previously associated with the disease. No variants following X-linked dominant (XLD) mode of inheritance were identified. on a same sequencing read indicates that they are located on the same chromosome.   of other family members were obtained with Sanger sequencing. There are 2 segregations from individual II:7 to II:1 (obligate carrier) and II:8 (obligate carrier), 2 segregations from II:1 to III:1 and III:2 and 1 segregation from II:8 to III:9. Individuals I:2 and III:7 were affected but were not genotyped, so it was not counted as a segregation. Individual III:8 is genotype positive but is not affected probably due to a delayed onset of the disease and was not counted as a segregation either.

PM5-Moderate
Novel missense change at an amino acid residue where a different missense change determined to be pathogenic has been seen before

PM5-Supporting
Downgraded to supporting given that other variant does not meet "likely pathogenic"/"pathogenic" criteria

PP1-Supporting
Cosegregation with disease in multiple affected family members in a gene definitely known to cause a disease

PP1-Moderate
Upgraded to moderate because of 5 segregations in the family

PP3-Supporting
Multiple lines of computation evidence support a deleterious effect on the gene or gene product (conservation, evolutionary, splicing impact, ect.)

PP4-Supporting
Patient's phenotype or family history is highly specific for a disease with a single genetic etiology Likely pathogenic Figure S6. Interpretation of the RNF213 variants according to the ACMG/AMP guidelines (Richards et al., 2015;Gelb et al., 2018).
Absence from controls (or at extremely low frequency if recessive) in Exome Sequencing Project, 1000 Genomes, or ExAC databases

PP1-Supporting
Cosegregation with disease in multiple affected family members in a gene definitely known to cause a disease

PP1-Moderate
Upgraded to moderate because of 5 segregations in the family

PP4-Supporting
Patient's phenotype or family history is highly specific for a disease with a single genetic etiology Uncertain significance

BP4-Supporting
Multiple lines of computational evidence suggest no impact on gene or gene product (conservation, evolutionary, splicing impact, etc.)     Dong et al., 2014;Liu et al., 2016) predictions for the RNF213 (NM_001256071.2) missense variants reported in this study and by Smith and coworkers (Smith et al., 2014) Jagadeesh et al., 2016 k The larger the score (Phred-like), the more likely the variant is damaging (Kircher et al., 2014;Rentzsch et al., 2018) l Dong et al., 2014 The larger the score, the more conserved the site (maximum scores 1.312 for phyloP30way and 10.003 for phyloP100way) (Pollard et al., 2010) n The larger the score, the more conserved the site (maximum score 1 for both phastCons30way and phastCons100way (Siepel et al., 2005) o The larger the score, the more conserved the site (maximum score 6.17) (Cooper et al., 2005(Cooper et al., , 2010