Genetic underpinnings of adiposity: from GWAS discovery to functional characterisation
Item statusRestricted Access
Embargo end date30/11/2021
Obesity affects more than 600 million people worldwide and causes 4 million excess deaths every year through its related co-morbidities, such as type 2 diabetes and heart disease. Detrimental anatomical distribution of body fat, specifically visceral fat accumulation, is a major driver of this mortality risk. There is clear evidence that underlying genetic predisposition plays a major role in determining body fat mass and distribution. Most genome-wide association studies to date have focused on anthropometric, rather than direct, measures of adiposity, such as BMI and WHR, and have successfully identified over a thousand candidate loci. However, the mechanisms via which these loci exert their effect is most often unknown. Thus, our understanding of the full impact and mechanisms underpinning the genetic contribution to human adiposity is incomplete. Here, we aim to further understand how human genomic variation contributes to body composition by using refined measures of adiposity and by gaining functional insight into the role of the resulting candidate loci.We worked with non-invasive imaging phenotypes in UKBiobank, a population cohort that includes 500,000 participants from across the UK. Of these, 5,000 have undergone DXA scans that allow for the direct measurement of fat and lean mass throughout the body. We first established a way to impute the DXA adiposity patterns in the rest of the UKBiobank cohort, using anthropometric measures of body size. These imputed phenotypes provided accurate predictions of the underlying adiposity traits (R~0.8) and were used to increase our novel discovery power in a genome-wide association framework. Out of the hundreds of associated loci, we sought to replicate the 27 strongest associations (P-value<6x10-13) in other population cohorts for the underlying DXA measures. To do so, we worked with genetic and DXA data from the ORCADES, EPIC-Norfolk and Fenland studies (collective DXA n=18,000). Six of the submitted SNPs were replicated successfully (FDR 10%), while the vast majority (19/27) were directionally consistent between discovery and replication. Two biologically promising candidate loci were followed-up to establish causal genes. The first locus, around gene PLA2G6, had previously been associated with various metabolic and obesity measures. However, computational approaches indicated several likely causal genes at the locus. We therefore used chromatin conformation capture technologies to pinpoint which of these genes interact with the associated interval. In doing so, we observed long-range interactions with three other genes Pick1, Dmc1 and Sox10. The second locus was composed of eQTLs that changed the expression of gene ADAMTS14 and was thus, further investigated it in a mouse model to ascertain the effect of Adamts14-null mutations on body composition in an in vivo setting. In the GWAS data, we saw a reduction of 18g in leg fat mass per copy of the effect allele and also a reduction in the expression of the gene. The Adamts14-/- mouse phenocopied this effect, as it showed reduced adiposity and weight gain under obesogenic conditions. It also showed increased energy expenditure and food intake, indicating an altered energy homeostasis.Large scale biobanks offer great gene discovery potential. Here, we harnessed the phenotypic breadth of UKBiobank to impute direct body composition measurements and identify novel loci that may affect adiposity. Using one such locus, we successfully replicated the human adiposity phenotype in a mouse model, thus validating this approach and highlighting its potential to offer biological insights into the genetics of obesity.