CELSR2 is a candidate susceptibility gene in idiopathic scoliosis

Abstract

A Swedish pedigree with an autosomal dominant inheritance of idiopathic scoliosis was initially studied by genetic linkage analysis, prioritising genomic regions for further analysis. This revealed a locus on chromosome 1 with a putative risk haplotype shared by all affected individuals. Two affected individuals were subsequently exome-sequenced, identifying a rare, non-synonymous variant in the CELSR2 gene. This variant is rs141489111, a c.G6859A change in exon 21 (NM_001408), leading to a predicted p.V2287I (NP_001399.1) change. This variant was found in all affected members of the pedigree, but showed reduced penetrance. Analysis of tagging variants in CELSR1-3 in a set of 1739 Swedish-Danish scoliosis cases and 1812 controls revealed significant association (p = 0.0001) to rs2281894, a common synonymous variant in CELSR2. This association was not replicated in case-control cohorts from Japan and the US. No association was found to variants in CELSR1 or CELSR3. Our findings suggest a rare variant in CELSR2 as causative for idiopathic scoliosis in a family with dominant segregation and further highlight common variation in CELSR2 in general susceptibility to idiopathic scoliosis in the Swedish-Danish population. Both variants are located in the highly conserved GAIN protein domain, which is necessary for the auto-proteolysis of CELSR2, suggesting its functional importance.

Fig 1. The pedigree included in the current study.

Affected individuals are indicated in black, unaffected in white. An obligate carrier of the chromosome 1 putative scoliosis risk haplotype (denoted by blue bars) is marked with a black dot. All numbered individuals have been genotyped and included in the linkage analysis. All putative non-risk haplotypes are denoted by white bars. The two exome-sequenced individuals are marked with asterisks. Carriers of the rare CELSR2 variant identified by exome sequencing are marked by a green box.

Fig 2. Genetic linkage peak on chromosome 1, indicating sharing in all affected individuals.

The maximum K&C non-parametric LOD (NPL) score was attained through the exponential model. The x-axis shows the position on the chromosome (in cM), the y-axis shows the NPL score.

Fig 3. Schematic of the structure of the CELSR2 protein.

Shown are the cadherin, EGF-like, and laminin-G-like domains. Also the transmembrane (TM) 7-pass domain and the evolutionarily conserved GAIN domain. The GAIN domain contains within it a GPS domain and is the site of autoproteolytic cleavage of CELSR2. The rare V2287I variant and more common R2060R tagging variant are both located within the GAIN domain.

Fig 4. Visualisation of the predicted structure of the GAIN domain of CELSR2.

The left panel shows the wildtype 2287V form, the right panel shows the mutated 2287I form. The location of the V2287I and R2060R variants are shown, as well as the site of autoproteolytic cleavage.