Whole exome sequencing reveals a mutation in ARMC9 as a cause of mental retardation, ptosis, and polydactyly

2.1 Clinical report

We described the clinical features in three affected siblings born out of consanguineous union, characterized as mental retardation, ptosis, and polydactyly phenotype (Panigrahi et al., 2002). The proband showed severe intellectual disability (ID), bilateral ptosis, downslanting palpebral fissures, hypertelorism, round face, high arched palate, clinodactyly, tapering of fingers, and hypermobile patellae (Figure 1c). He also had joint laxity in the metacarpophalangeal and interphalangeal joints in hands and short stature. Female sibling was also proportionately short with stocky build. She had moderate ID, ptosis in right eye, pointed chin, bilateral post axial polydactyly with joint laxity. The third male sibling had marked joint laxity similar to proband, moderate ID, bilateral ptosis, and bilateral post axial polydactyly (Table 1). We collected blood samples from all affected siblings and both parents after informed consent. This study was approved by the Institutional Ethics Committee.

2.2 Genetic analysis

Genomic DNA from all affected and unaffected family members was extracted from whole blood. SNP array analysis in all three affected individuals was performed by using Illumina Human-CytoSNP-12 v.2.1 DNA Analysis BeadChip Kit (Illumina Inc., San Diego, CA). Analysis of array data was performed using Genome studio software. Size threshold for analysis was kept as 200 kb for deletion, 500 kb for duplication, and 1 Mb for homozygous regions. Genomic Oligoarray and SNP evaluation tool (v3.0) (http://firefly.ccs.miami.edu/cgi-bin/ROH/ROH_analysis_tool.cgi) was used for detection of known disease causing OMIM genes in the common homozygous regions.

Exome sequencing was performed in all three affected sibs. Agilent SureSelectXT V5 exome capture kit (Agilent Technologies, Santa Clara, CA) was used for preparation of exome library by using ∼3 μg of genomic DNA. Exome library was sequenced to mean 100× coverage on Illumina HiSeq2000 sequencing platform. Alignment to human reference genome (GRCh37/hg19) and variant calling were performed using NEXTGENe v.2.3.4.4 (SoftGenetics, LLC, State College, PA) followed by variant annotation by ANNOVAR (Wang, Li, & Hakonarson, 2010). Variants with minor allele frequency of >0.01 in 1000 Genome project (http://www.1000genomes.org/), EVS (http://evs.gs.washington.edu/EVS/), ExAC databases (http://exac.broadinstitue.org/), and in-house database were excluded from the study. Only nonsense, non-synonymous, splice site variants, synonymous splice site variants, insertion, and deletion variants that affect coding regions of gene were used for interpretation. PCR was performed using gene specific primer set (listed in Supplementary Table S1) followed by Sanger sequencing on ABI3130 Genetic analyzer (Life Technologies, Carlsbad, CA) following the manufacturer's protocol in control, parents, unaffected sibs, and affected sibs to validate variants identified on exome sequencing.

2.3 In vivo analysis of variant at splice site

An observed variation was evaluated for splicing defect using pCAS2 minigene construct (Tournier et al., 2008). Wild-Type (ARMC9_Wt) and mutant (ARMC9_Mt) constructs were created by PCR amplification from control and patient sample, respectively, followed by cloning them in BamH I and Mlu I restriction site of pCAS2 vector. Recombinant constructs as well as pCAS2 (as control) were transfected into COS7 cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Cells were harvested after 18 hr of transfection and total RNA was extracted from each transfectant using RNeasy Mini Kit (Qiagen, Hilden, Germany), as per the manufacturer's instruction. Total RNA was subjected to RT PCR using SuperScript III (Invitrogen) followed by plasmid specific amplification (P1 and P2 primer) and evaluated on 2% agarose gel. Appropriate bands were purified using QIAquick Gel Extraction Kit (Qiagen) and subjected to Sanger sequencing using on ABI3130 Genetic analyzer (Life Technologies).

2.4 Molecular modelling

To analyze conservation pattern of ARMC9 (NP_001278585), protein sequences from various species were aligned. Multiple sequence alignment was performed for the region of ARM domain I, which is expected to be affected during splicing event using ClustalW software (http://embnet.vital-it.ch/software/ClustalW.html). An ARMC9 wild-type (1–665 aa) and ARMC9 mutant structural model was generated using I-TASSER software (http://zhanglab.ccmb.med.umich.edu/I-TASSER/) for comparison of predicted molecular structures (Roy, Kucukural, & Zhang, 2010; Yang et al., 2015; Zhang, 2008).

3.1 Homozygosity mapping and exome sequencing

We did not identify any significant copy number variations in any of the affected siblings. Regions of loss of heterozygosity (LOH) greater than 1 MB shared by affected siblings were identified on chromosome 2, 7, 10, and 14 (Supplementary Table S2), harbouring 228 Refseq genes and including 81 OMIM genes. The largest region of shared homozygosity was on chromosome 2q of 11 Mb, harbouring 141 genes. None of the genes in the shared regions were found to be associated with intellectual disability, ptosis, and polydactyly occurring together. Whole exome sequencing was performed for all three affected siblings. Mapping and alignment to human reference genome was performed using NEXTGENe v.2.3.4.4 software. Variants were filtered for minor allele frequency (MAF) <0.01 in 1000 Genome project, EVS, ExAC, and in-house database of 50 Indian exomes. Mutation databases such as HGMD (http://www.biobase-international.com/products/hgmd) and Clinvar (http://www.ncbi.nlm.nih.gov/clinvar/) were also checked for presence of reported pathogenic variants in affected individuals. After filtering of variants, two candidate variants (c.654A>C in UGT1A7 and c.879G>A in ARMC9) present in shared homozygous region of 11 Mb on Chromosome 2 were studied further (Table 2 and Supplementary Table S3).

The variant in UGT1A7 was predicted to be pathogenic but the function of UGT1A7 gene in glucuronidation pathway did not appear to be related with the phenotype in the affected siblings. The variant in ZFPM1 was not present in shared region of Homozygosity in the patients and thus were not opted for further study. The variant ARMC9 c.879G>A, on chromosome 2q37.1, was found to be homozygous in all affected siblings, present in shared homozygous region, absent in 1000 Genome database, EVS, ExAC, in-house database of 50 exomes, and predicted to be damaging by various pathogenicity prediction software. This variant is a synonymous variant (p.T293T) in the last base of exon 9 of ARMC9 gene and was predicted to affect splicing by splice site prediction tools such as Human Splice Finder and MutPred Splice (Desmet et al., 2009; Mort et al., 2014). Sanger sequencing confirmed presence of homozygous c.879G>A variant in all three affected siblings (Figure 1b). The variant was found to be heterozygous in both parents and absent in unaffected sibling and was consistent with autosomal recessive inheritance (Figure 1a).

3.2 Characterization of variant by minigene assay for splicing defect

Since ARMC9 is not expressed in blood, in vivo characterization of this variant was done to determine whether it causes a defect in mRNA splicing using pCAS2 minigene. PCR amplification using pCAS2 minigene specific primers (P1 and P2) from cDNA prepared using all transfectants were checked on 2% agarose gel, which revealed presence of 255 bp band in PCAS (control plasmid) and ARMC9_Mt whereas ARMC9_Wt had 354 bp fragment (Figure 2).

These results indicate that the variant is likely to lead to skipping of exon during splicing event. To confirm the results from RNA assay, sequencing of specific bands form RT-PCR for ARMC9_pCAS_Wt and ARMC9_pCAS_Mt was done by Sanger sequencing, which revealed deletion of 99 base pairs of exon 9 of ARMC9 gene (Supplementary Figure S1). This result confirms that synonymous variant c.879G>A (p.T293T) leads to abolition of splice site and leads to skipping of exon 9 of ARMC9 which may result in in-frame deletion of 33 amino acids (exon 9) form ARMC9 in patients.

Multiple sequence alignment of ARMC9 across vertebrates and invertebrates shows that ARM domains are highly conserved across species. Structural modelling of ARMC9 predicts similar structure to other ARM domain containing proteins such as beta-catanin, which shows presence of superheilx of alpha helices with three helices per unit, this forms a grove, which presumably may interact with its partners. ARMC9 mutant structure indicates deletion of 261–293 aa residues from wild-type ARMC9 critically disrupts its structure and hence may abolish native function of ARMC9 (Figure 3).