Population-level deficit of homozygosity unveils CPSF3 as an intellectual disability syndrome gene

Abstract

Predicting the pathogenicity of biallelic missense variants can be challenging. Here, we use a deficit of observed homozygous carriers of missense variants, versus an expected number in a set of 153,054 chip-genotyped Icelanders, to identify potentially pathogenic genotypes. We follow three missense variants with a complete deficit of homozygosity and find that their pathogenic effect in homozygous state ranges from severe childhood disease to early embryonic lethality. One of these variants is in CPSF3, a gene not previously linked to disease. From a set of clinically sequenced Icelanders, and by sequencing archival samples targeted through the Icelandic genealogy, we find four homozygous carriers. Additionally, we find two homozygous carriers of Mexican descent of another missense variant in CPSF3. All six homozygous carriers of missense variants in CPSF3 show severe intellectual disability, seizures, microcephaly, and abnormal muscle tone. Here, we show how the absence of certain homozygous genotypes from a large population set can elucidate causes of previously unexplained recessive diseases and early miscarriage.

Fig. 1. Structure of CPSF3 (RefSeq transcript NM_016207.3) and location of the two missense variants found in homozygous state in patients with a severe intellectual disability syndrome.

a The p.Ile354Thr and p.Gly468Glu variants are located in exon 9 and 12 of CPSF3, respectively. Dark gray sections represent exons and untranslated regions, light gray lines represent introns. b The p.Ile354Thr variant is in the β-CASP domain of CPSF3 (amino acids 246–367, shown in green). The p.Gly468Glu variant is located within an uncharacterized linker domain (amino acids 448–479) between the critical second metallo-β-lactamase domain (shown in pink) and the highly conserved C-terminal domain (amino acids 479–682, shown in yellow) of CPSF3. The first metallo-β-lactamase domain is shown in blue, uncharacterized domains are shown in gray.