Identification of Rare Loss-of-Function Genetic Variation Regulating Body Fat Distribution

Abstract

Context: Biological and translational insights from large-scale, array-based genetic studies of fat distribution, a key determinant of metabolic health, have been limited by the difficulty in linking predominantly noncoding variants to specific gene targets. Rare coding variant analyses provide greater confidence that a specific gene is involved, but do not necessarily indicate whether gain or loss of function (LoF) would be of most therapeutic benefit.

Objective: This work aimed to identify genes/proteins involved in determining fat distribution.

Methods: We combined the power of genome-wide analysis of array-based rare, nonsynonymous variants in 450 562 individuals in the UK Biobank with exome-sequence-based rare LoF gene burden testing in 184 246 individuals.

Results: The data indicate that the LoF of 4 genes (PLIN1 [LoF variants, P = 5.86 × 10-7], INSR [LoF variants, P = 6.21 × 10-7], ACVR1C [LoF + moderate impact variants, P = 1.68 × 10-7; moderate impact variants, P = 4.57 × 10-7], and PDE3B [LoF variants, P = 1.41 × 10-6]) is associated with a beneficial effect on body mass index-adjusted waist-to-hip ratio and increased gluteofemoral fat mass, whereas LoF of PLIN4 (LoF variants, P = 5.86 × 10-7 adversely affects these parameters. Phenotypic follow-up suggests that LoF of PLIN1, PDE3B, and ACVR1C favorably affects metabolic phenotypes (eg, triglycerides [TGs] and high-density lipoprotein [HDL] cholesterol concentrations) and reduces the risk of cardiovascular disease, whereas PLIN4 LoF has adverse health consequences. INSR LoF is associated with lower TG and HDL levels but may increase the risk of type 2 diabetes.

Conclusion: This study robustly implicates these genes in the regulation of fat distribution, providing new and in some cases somewhat counterintuitive insight into the potential consequences of targeting these molecules therapeutically.

Keywords: UK Biobank; cardiometabolic risk; fat distribution; genetic variants; loss of function.

Figure 1.

Miami plot for sex-combined and sex-specific single-marker association results for body mass index–adjusted waist-to-hip ratio (WHR_adjBMI). Top: Manhattan plot representing results from the main, sex-combined genome-wide association studies for WHR_adjBMI for genotyped, rare nonsynonymous variants (MAF = 0.1%-0.5%, correlation and rare allele concordance > 0.9 when compared to the exome sequencing data). Gene annotations for the genome-wide significant variants from the main, sex-combined analyses are shown in black; gene annotations and significance from the main, sex-combined analyses for variants that were genome-wide significant in sex-specific analyses (women) only are shown in red. Bottom: Sex-specific significance of the variants highlighted previously.

Figure 2.

Gene-based association results. Gene-based exome-wide discovery results for body mass index–adjusted waist-to-hip ratio (WHR_adjBMI). The horizontal dashed line represents the exome-wide significance threshold (P = 2.53 × 10^–6).

Figure 3.

Forest plot of phenotypic associations for significant variant-gene categories. Black represents the sex-combined, blue represents the men-only, and red represents the women-only analysis. Horizontal lines represent 95% CIs. Waist-to-hip ratio adjusted for body mass index (BMI) (n = 184 246), gynoid fat adjusted for total body fat (n = 178 143), trunk fat adjusted for total body fat (n = 178 143), triglyceride levels (n = 175 271), high-density lipoprotein (HDL) cholesterol (n = 161 239), and triglyceride-to-HDL ratio (n = 161 102) were all driven from UK Biobank (See Supplementary Table 12 for details) (14).

Figure 4.

Functional impact of ACVR1C I195T variant on Smad signaling. A, HEK293 cells were transiently transfected with ACVR1C expression constructs and their receptor components, along with firefly and Renilla luciferase expression plasmids. Firefly luciferase activity was normalized to Renilla activity and the luciferase activity in nonstimulated cells transfected with empty vector (EV) was set to 1. A constitutively active (CA) ACVR1C variant T194D and a kinase-deficient (KD) variant K222R were included for comparison. Results from 3 independent experiments are presented as mean ± SD. Statistical significance was evaluated by one-way analysis of variance (ANOVA) with Tukey post hoc test for multiple comparisons between pairs. WT, wild-type ACVR1C. B to D, HEK293T cells transfected with plasmids containing EV, WT ACVR1C, KD ACVR1C, or I195T ACVR1C were treated for 30 minutes with activin B (25 ng/mL) or growth and differentiation factor 3 (GDF3; 500 ng/mL), and B, SMAD signaling was determined via Western blotting. Black triangle indicates pSMAD3 band. SMAD activation was quantified by the ratio of phospho-SMAD2 to total SMAD2 protein and phospho-SMAD3 to total SMAD3 protein for each condition: C, untreated; D, activin B; and E, GDF3. Data are expressed as relative pSMAD:SMAD ratios normalized to untreated EV controls. Three or 4 independent experiments were combined for analysis, and differences between ACVR1C transfection groups were determined by one-way ANOVA with Tukey post hoc test for multiple comparisons between pairs. ***P less than .001; **P less than .01; *P less than .05; ns, not significant.

Figure 5.

Graphical summary of the findings in this study—impact of predicted loss-of-function (LoF) of ACVR1C, INSR, PDE3B, PLIN1, and PLIN4.