Identification of candidate gene FAM183A and novel pathogenic variants in known genes: High genetic heterogeneity for autosomal recessive intellectual disability

The etiology of intellectual disability (ID) is heterogeneous including a variety of genetic and environmental causes. Historically, most research has not focused on autosomal recessive ID (ARID), which is a significant cause of ID, particularly in areas where parental consanguinity is common. Identification of genetic causes allows for precision diagnosis and improved genetic counseling. We performed whole exome sequencing to 21 Turkish families, seven multiplex and 14 simplex, with nonsyndromic ID. Based on the presence of multiple affected siblings born to unaffected parents and/or shared ancestry, we consider all families as ARID. We revealed the underlying causative variants in seven families in MCPH1 (c.427dupA, p.T143Nfs*5), WDR62 (c.3406C>T, p.R1136*), ASPM (c.5219_5225delGAGGATA, p.R1740Tfs*7), RARS (c.1588A>G, p.T530A), CC2D1A (c.811delG, p.A271Pfs*30), TUSC3 (c.793C>T, p.Q265*) and ZNF335 (c.808C>T, p.R270C and c.3715C>A, p.Q1239K) previously linked with ARID. Besides ARID genes, in one family, affected male siblings were hemizygous for PQBP1 (c.459_462delAGAG, p.R153Sfs*41) and in one family the proband was female and heterozygous for X-chromosomal SLC9A6 (c.1631+1G>A) variant. Each of these variants, except for those in MCPH1 and PQBP1, have not been previously published. Additionally in one family, two affected children were homozygous for the c.377G>A (p.W126*) variant in the FAM183A, a gene not previously associated with ARID. No causative variants were found in the remaining 11 families. A wide variety of variants explain half of families with ARID. FAM183A is a promising novel candidate gene for ARID.


Introduction
Intellectual disability (ID) is an early-onset neurodevelopmental disorder affecting 1% of the general population [1]. ID is characterized by a significant impairment in cognitive ability and adaptive behavior affecting memory, language, problem solving, and visual comprehension, which can lead to impairments in activities of daily living such as self-care and interpersonal communication.
Environmental factors, such as teratogens, infections, malnutrition, and neurological trauma as well as genetic conditions can cause ID. The literature supports a strong genetic etiology for ID, with a varying proportion of cases (ranging from 17% to 50%) being attributable to genetic causes [2][3][4][5]. Variants in more than 1,000 genes have been connected to ID [6]. Support for rare de novo variants as a major cause of ID in simplex cases have been reported [7]. Microarray and exome sequencing have demonstrated the importance of de novo copy number variations (CNVs) and single-nucleotide variations (SNVs) in ID. While research elucidating chromosomal aberrations, CNVs, autosomal dominant, and X-linked variants as causes for ID has been well established, it was not until recently that studies focusing on autosomal recessive forms of ID (ARID) have gained attention [6,8]. ARID either presents as the sole clinical feature (nonsyndromic) or with additional features (syndromic), and it is extremely heterogeneous [9,10]. To date, there are fifty-five genes in the phenotypic series [PS249500] for non-syndromic ARID in the Online Mendelian Inheritance in Man (OMIM) database [6]. ARID appears to be a common form of ID, an unsolved healthcare problem creating an enormous socioeconomic burden on society, especially in the underdeveloped countries where there is a high rate of consanguinity [8,11]. To reveal the causative variants in ARID, we performed Whole Exome Sequencing (WES) in 21 families affected by non-syndromic ID; these families either had multiple affected family members or pedigrees suggestive of consanguinity.
For those families without causative variants in known ID genes, we obtained runs of homozygosity (>2 MB) shared by affected members of each family (S3 Table). Only homozygous INDELs, single nucleotide variants, and CNVs mapping to runs of homozygosity were included in the analysis.

Results
On average, 99%, 94% and 89% of the captured regions were covered by 1X, 5X, and 10X reads, respectively. Average read depth was 59X in our cohort.

Known variants associated with intellectual disability
We identified two variants that have previously been reported to cause ID in two genes ( Table 1). In family S223, one male proband (II:1) was homozygous for c.427dupA (p. T143Nfs � 5) in MCPH1. In family S25, both affected brothers (II:1 and II:2) were found to be hemizygous for c.459_462delAGAG (p.R153Sfs � 41) in PQBP1.

Novel variants in known genes associated with intellectual disability
Whole exome sequencing of the probands resulted in identification of seven novel variants in seven genes known to cause ID (Table 1 and S4 Table). These variants were associated with varying phenotypes (Table 2). Variant segregation was confirmed with Sanger sequencing (Fig  1). Homozygous variants in CC2D1A and ASPM led to frameshift mutations. Additionally, homozygous nonsense variants in TUSC3 and WDR62 and a heterozygous variant in SLC9A6 occurring at a splice site were identified. Lastly, a homozygous substitution in RARS and two heterozygous variants in ZNF335 were found. These results are described in more detail in Table 1.

Mutated FAM183A as a causative candidate for intellectual disability
Whole exome sequencing revealed a nonsense variant (c.377G>A; p.W126 � ) in FAM183A, a novel candidate gene for ARID. Two siblings, MR25-II:1 and MR25-II:2, were homozygous for the variant, and in the family, the variant segregated in an autosomal recessive fashion (Fig 1).

Discussion
We have identified causative variants in 10 out of 21 Turkish families with nonsyndromic ARID. 11 families remained unsolved in our cohort (S1 Fig and S5 Table).

Known variants associated with intellectual disability
The two families with previously reported causative variants for ID are phenotypically consistent with the previously described findings. Both brothers (II:1 and II:2) in family S25 show findings indicative of Renpenning syndrome [MIM 309500], an X-linked condition [17]. Zhang et al (2017) discusses the molecular pathogenesis of mutations in PQBP1, pointing toward the promotion of ubiquitin-mediated degradation of fragile X mental retardation Genetic heterogeneity for autosomal recessive intellectual disability protein (FMRP) resulting in synaptic dysfunction [19]. WES, elucidating the presence of a known causative variant, accompanied with phenotypic data confirmed a diagnosis of Microcephaly 1 in proband S223-II:1 [18].

Novel variants in known genes associated with intellectual disability
CC2D1A variants are associated with ARID (MRT3) [MIM 608443]. This disorder is characterized by significantly below average general intellectual functioning associated with impairments in adaptive behavior which manifest during the developmental period. Analysis in Drosophila links CC2D1A, a member of the mammal lethal giant discs (lgd) protein family, to endosomal trafficking of Notch proteins, well known transmembrane receptors that regulate cell fate during development [20]. Basel-Vanagaite et al. (2006) reported homozygosity for a protein truncating mutation in CC2D1A in affected members of 9 consanguineous Israeli-Arab families with nonsyndromic ID [21]. The homozygous deletion at c.811delG (p.A271Pfs � 30) of CC2D1A found in family AU10 is the fifth variant associated with ID reported in this gene. TUSC3 variants are known to cause Mental Retardation, Autosomal Recessive 7 (MRT7) [MIM 611093]. Including the finding reported here, four independent autosomal recessive variants in TUSC3 are known to cause ARID [22][23][24]. The exact molecular pathogenesis is unknown. One study suggests the involvement of N-glycosylation in higher brain functions [24], while another postulates disturbed magnesium levels due to TUSC3 impairment may play a role in the pathogenesis of intellectual disability [22]. This novel nonsense variant at c.793C>T (p.Q265 � ) in TUSC3 in family MR32 further demonstrates the significance of this gene's association with ARID.
Seventeen variants in numerous domains of WDR62 have been reported to be associated with Microcephaly 2, primary, autosomal recessive, with or without cortical malformations (MCPH2) [MIM 604317] [25][26][27][28]. Cellular studies indicate that WDR62 is a crucial protein in enabling spindle poles to position the cytokinetic furrow and prolong neural precursor generation, a process that is uniquely vital to the proper growth of the human cerebral cortex [25]. The attributes of the proband, S228-II:1, described in Table 2, are consistent with this phenotype. This novel homozygous nonsense variant in WDR62, c.3406C>T (p.R1136 � ) along with the phenotype of the proband further implicate this gene's involvement in ARID.
Variants in SLC9A6 are associated with Mental retardation, X-linked syndromic (MRXSCH), Christianson type [MIM 300243]. SLC9A6 encodes a monovalent sodium-selective sodium/hydrogen exchanger (NHE) found in the membranes of intracellular organelles such as mitochondria and endosomes; NHEs participate in a wide array of essential cellular processes [29]. MRXSCH is characterized by profound ID, epilepsy, ataxia, and microcephaly. The phenotype of the female proband (S234-II:1), described in Table 2, is consistent with the presentation of MRXSCH. Linkage analysis and DNA sequencing of families with MRXSCH have identified multiple variants in the SLC9A6, the majority of which are truncating and/or splice mutations [30][31][32][33][34]. The novel heterozygous splice-site variant, c.1631+1G>A in SLC9A6, adds to the body of literature on pathogenic splice-site variants reported in this gene.
ZNF335 mutations are associated with Microcephaly 10, primary, autosomal recessive (MCPH10) [MIM 615095]. One homozygous variant in ZNF335 (S236-II:1), causing both a missense change and a splice site defect, is linked to this syndrome [35]. This same study demonstrated that ZNF335 deficiency causes disrupted neuronal proliferation and differentiation in vitro and in vivo mouse models [35]. Both discovered variants in our study are predicted to be disease causing and are likely to result in a hypomorphic variant of ZNF335. This, in combination with the gene's association with MCPH10 and altered neuronal growth in vitro and in vivo, make it likely that these mutations are the cause of this family's ID.
Variants in ASPM are associated with Microcephaly 5, primary, autosomal recessive (MCPH5) [MIM 608716]. ASPM has been implicated in the determination of human cerebral cortical size via maintenance of a cleavage plane orientation allowing for symmetric, proliferative division of neuroepithelial cells during brain development [36]. Several aberrations in ASPM have been reported in the literature with a clear majority resulting in premature termination [37][38][39][40][41]. Similarly to previously reported variants, the novel homozygous deletion c.5219_5225delGAGGATA (p.R1740Tfs � 7) found in ASPM (S243-II:1) leads to a premature stop codon, suggesting that this variant is involved in ARID.
Lastly, variants in RARS are associated with Leukodystrophy, hypomyelinating, 9 (HLD9) [MIM 616140]. RARS encodes the cytoplasmic arginyl-tRNA synthetase, an enzyme essential for RNA translation and a key player in myelination, among the subunits of the multisynthase complex [42]. Thus, mutations in RARS cause a hypomylenating disorder of the central nervous system. Previously identified variants in RARS causing the HLD9 phenotype are compound heterozygous mutations [42]. In contrast to previously identified variants in RARS, the novel variant found in family S244 is a homozygous missense mutation. After exome sequencing and segregation analysis, these mutations are the only variants that segregate with the family's phenotype. Therefore, we propose that this homozygous missense variant has a role in ARID and is the most likely cause of ID this family.

Mutated FAM183A as a causative candidate for intellectual disability
In family, MR 25, there were two siblings with ID, low set ears, and microcephaly along with other correlating phenotypic features ( Table 2, Fig 1). The two affected children are homozygous for a FAM183A variant (c.377G>A; p.W126 � ). This variant, found in MR25-II:1 and MR25-II:2, is located in the longest shared homozygous genomic region in these siblings (S6 Table). FAM183A is expressed in human brain, including the hippocampus, caudate nucleus, and medulla oblongata [43]. Five heterozygous large deletions including FAM183A are reported in the DECIPHER database, (https://decipher.sanger.ac.uk/; accessed 07/08/2018). Three out of five variants are found in patients with intellectual disability, developmental delay, or microcephaly; however, the variants' pathogenicities are not known [44].
The pathogenicity models, expression in brain tissue, and validated autosomal recessive segregation of this variant with the phenotype in family MR 25 support a role for this homozygous nonsense variant as contributing to the inheritance of ID. More research, including functional studies, should be done to further support the role of FAM183A in the pathogenesis of ID.
In 10 out of 21 families investigated (47.6%) we identified likely disease-causing DNA variants. Although CNVs are important cause of ID, we did not detect any CNV in our cohort. As expected for affected offspring of healthy consanguineous parents, the clear majority of these were autosomal recessive defects. This yield is comparable to other recent investigations in highly consanguineous populations [7,[45][46][47][48][49][50].
Supporting information S1