Constraint and conservation of paired-type homeodomains predicts the clinical outcome of missense variants of uncertain significance.

The need to interpret the pathogenicity of novel missense variants of unknown significance identified in the homeodomain of X-chromosome aristaless-related homeobox (ARX) gene prompted us to assess the utility of conservation and constraint across these domains in multiple genes compared to conventional in vitro functional analysis. Pathogenic missense variants clustered in the homeodomain of ARX contribute to intellectual disability (ID) and epilepsy, with and without brain malformation in affected males. Here we report novel c.1112G>A, p.Arg371Gln and c.1150C>T, p.Arg384Cys variants in male patients with ID and severe seizures. The third case of a male patient with a c.1109C>T, p.Ala370Val variant is perhaps the first example of ID and autism spectrum disorder (ASD), without seizures or brain malformation. We compiled data sets of pathogenic variants from ClinVar and presumed benign variation from gnomAD and demonstrated that the high levels of sequence conservation and constraint of benign variation within the homeodomain impacts upon the ability of publicly available in silico prediction tools to accurately discern likely benign from likely pathogenic variants in these data sets. Despite this, considering the inheritance patterns of the genes and disease variants with the conservation and constraint of disease variants affecting the homeodomain in conjunction with current clinical assessments may assist in predicting the pathogenicity of missense variants, particularly for genes with autosomal recessive and X-linked patterns of disease inheritance, such as ARX. In vitro functional analysis demonstrates that the transcriptional activity of all three variants was diminished compared to ARX-Wt. We review the associated phenotypes of the published cases of patients with ARX homeodomain variants and propose expansion of the ARX-related phenotype to include severe ID and ASD without brain malformations or seizures. We propose that the use of the constraint and conservation data in conjunction with consideration of the patient phenotype and inheritance pattern may negate the need for the experimental functional validation currently required to achieve a diagnosis.