Using Rare Variants, Animal Models and Mendelian Randomization to Pinpoint Causative Genes

Background: Although accumulating evidence suggests that rare CNVs are a common health problem in the population, their phenotype spectrum have largely only been investigated in clinical (often pediatric) patients.

Objectives: Our aim is to pinpoint causative genes within the 16p11.2 region influencing sexual development in general population.

Methods: We combined data from a comprehensive set of clinically ascertained 16p11.2 families (n=660 affected individuals) with unselected adult population cohorts (>470,000 individuals recruited to the UK and EGCUT biobanks) to assess adult phenotypes.

Results: We uncovered that the 16p11.2 BP4-BP5 dosage, one of the most frequent genetic causes of mental disorders, was oppositely associated with age at menarche (AaM) in the UK Biobank (p=7.8e^-05; ANOVA, corrected for BMI and birth year). Compared to controls AaM was decreased in deletion (Δ=-1.5 years, p=0.01) and increased in duplication carriers (Δ=+1.5; p=7.8x10^-5). We replicated these associations in EGCUT (p=2.4e^-05) and in two unrelated cohorts of 16p11.2 clinical patients (p=7.7e^-05; p=3e^-03). We observed a directionally consistent trend for pubertal onset in male 16p11.2 CNV carriers. These features were accompanied by various diagnoses related to function of reproductive organs and fertility, with up to half of deletion and two-third of duplication female carriers presenting at least one diagnosis related to hormonal problems or function of genital organs. We validated the human results by detecting changes in timing of first ovulation, estrous cyclicity and uterine size in females and reduced anogenital distance in males 16p11.2 mice. Corroboratively, genes differentially-expressed in 16p11.2 patients cells and mice model cortices were enriched for urogenital disease genes. We unraveled a negative correlation between the 16p11.2 dosage and hypothalamus volume both in human (p_FWE<0.05) and mice (p<1x10^-04), suggesting that perturbation of the gonadotropin-releasing hormone (GnRH) axis could contribute to the observed phenotypes. Whereas Mendelian Randomization (MR) suggested INO80E and KCTD13 as potential causal genes for AaM, an agnostic modulation of all 16p11.2 genes dosage in gnrh3:egfp transgenic zebrafish larvae pinpointed to ASPHD1 a gene specifically expressed in brain and pituitary gland. Pairwise interaction experiments in zebrafish revealed epistasis between the ASPHD1 "driver" and the INO80E and KCTD13 “modifiers”.

Conclusions: Our findings highlight how identification of traits associated with rare variants, such as CNVs, in the general population provides valuable unbiased insight into disease etiologies. Our data also demonstrate how rare variants combined with MR can be utilized to pinpoint causal genes within GWAS loci as the common SNPs within the 16p11.2 interval were previously associated with AaM.