Background:
Elevated levels of adiponectin, an adipose-tissue derived hormone, are associated with a decreased risk for development of obesity, cardiovascular disease, and type 2 diabetes. We sought to fine-map and characterize loci from an adiponectin genome-wide association study (GWAS) to better understand the genes, variants, and mechanisms that contribute to adiponectin levels.
Methods:
We performed a GWAS of plasma adiponectin levels in 9,262 nondiabetic Finnish men from the Metabolic Syndrome in Men (METSIM) study using an efficient mixed model (EPACTS) to account for cryptic relatedness among the subjects. To identify multiple association signals within 1 Mb of each other, we used stepwise conditional analyses and Genome-wide Complex Trait Analysis (GCTA). We annotated association signals using regulatory elements based on chromatin marks from adipocyte nuclei (Epigenomic Roadmap) and ATAC-seq data from adipose tissue (METSIM) and SGBS preadipocyte cells. We also evaluated expression quantitative trait loci (eQTL) in subcutaneous adipose RNA-seq data from 387 METSIM samples. To test for allele-specific effects on transcriptional activity, we performed transcriptional reporter assays in HeLa cells.
Results:
We identified 5 loci associated with adiponectin (
P
<5x10
-8
):
CDH13, ADIPOQ
,
IRS1, PBRM1,
and
EPHA3.
Two loci (
CDH13
and
ADIPOQ
) contained 2 and 7 association signals (
P
<1x10
-5
), respectively. At
CDH13
, the first signal contained the lead adipose eQTL variant for
CDH13
. At the novel second signal at
CDH13
, rs4782722 is located in a regulatory element and the G-allele showed increased transcriptional activity compared to the T-allele, suggesting a functional role for this variant. At
ADIPOQ
, the first association signal also contained the lead adipose eQTL variant for
ADIPOQ.
All signals at
ADIPOQ
contained ≥1 variant in a putative enhancer, and the 7th signal includes rs62625753, a coding variant (G90S;
P
init
=3x10
-3
,
P
cond
=6x10
-4
) predicted to be deleterious (SIFT) and probably damaging (PolyPhen). Accounting for multiple signals resulted in a 1.6-fold increase in variance explained over the lead signals alone (5.9 vs 9.4%).
Conclusions:
Fine-mapping, annotation, and experimental validation of GWAS signals and variants provide novel insight into the molecular mechanisms underlying genetic association signals, leading to a clearer biological basis for disease.