1. Computational and Systems Biology

Predicting mechanisms of action at genetic loci associated with discordant effects on type 2 diabetes and abdominal fat accumulation

  1. Yonathan Tamrat Aberra  Is a corresponding author
  2. Lijiang Ma
  3. Johan LM Björkegren
  4. Mete Civelek  Is a corresponding author
  1. Department of Biomedical Engineering, University of Virginia, United States
  2. Center for Public Health Genomics, University of Virginia, United States
  3. Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, United States
  4. Department of Medicine, Karolinska Institutet, Huddinge, Sweden
https://doi.org/10.7554/eLife.79834
Share this article
Cite this article
  1. Yonathan Tamrat Aberra
  2. Lijiang Ma
  3. Johan LM Björkegren
  4. Mete Civelek
(2023)
Predicting mechanisms of action at genetic loci associated with discordant effects on type 2 diabetes and abdominal fat accumulation
eLife 12:e79834.
https://doi.org/10.7554/eLife.79834
  2. Download BibTeX
  3. Download .RIS

Abstract

Obesity is a major risk factor for cardiovascular disease, stroke, and type 2 diabetes (T2D). Excessive accumulation of fat in the abdomen further increases T2D risk. Abdominal obesity is measured by calculating the ratio of waist-to-hip circumference adjusted for the body-mass index (WHRadjBMI), a trait with a significant genetic inheritance. Genetic loci associated with WHRadjBMI identified in genome-wide association studies are predicted to act through adipose tissues, but many of the exact molecular mechanisms underlying fat distribution and its consequences for T2D risk are poorly understood. Further, mechanisms that uncouple the genetic inheritance of abdominal obesity from T2D risk have not yet been described. Here we utilize multi-omic data to predict mechanisms of action at loci associated with discordant effects on abdominal obesity and T2D risk. We find six genetic signals in five loci associated with protection from T2D but also with increased abdominal obesity. We predict the tissues of action at these discordant loci and the likely effector Genes (eGenes) at three discordant loci, from which we predict significant involvement of adipose biology. We then evaluate the relationship between adipose gene expression of eGenes with adipogenesis, obesity, and diabetic physiological phenotypes. By integrating these analyses with prior literature, we propose models that resolve the discordant associations at two of the five loci. While experimental validation is required to validate predictions, these hypotheses provide potential mechanisms underlying T2D risk stratification within abdominal obesity.

Editor's evaluation

This study presents a valuable finding on five genome-wide association study loci displaying discordant effects on T2D and abdominal obesity. The evidence supporting the claims of the authors is solid, although further experiments are required to describe the mechanisms through which a genetic variant can confer increased abdominal obesity while offering protection from T2D risk. The work will be of interest to researchers working within the fields of variant-to-function analysis and endocrinology.

https://doi.org/10.7554/eLife.79834.sa0

Introduction

Metabolic syndrome (MetSyn) is a cluster of dysregulated metabolic conditions that tend to occur together to increase the risk for cardiometabolic disorders such as type 2 diabetes (T2D) (Lusis, 2006). This cluster includes insulin resistance (IR), abdominal obesity, elevated serum triglycerides (TG) levels, low high-density lipoprotein cholesterol levels, as well as elevated systolic and diastolic blood pressure. Obesity, or the excessive accumulation of fat that presents a risk to health, is a major contributor to MetSyn (Lusis, 2006; Feero et al., 2010). Obesity, which is typically defined as a body mass index (BMI) above 30, has reached unprecedented levels of prevalence, and its role as a central regulator of disease risk makes it an appealing therapeutic target (Feero et al., 2010). Several recently developed T2D therapeutics have even successfully targeted obesity; SGLT2 inhibitors and GLP-1 agonists have been reported to result in a 2–6 kg reduction of body weight and reduced IR (Brown et al., 2021).

Despite the promise of these obesity-centered therapeutic strategies, there has also been a growing body of evidence describing a rare phenotype known as metabolically healthy obesity (MHO) (Blüher, 2020). MHO describes a group of phenotypes in which individuals with obesity are protected from adverse metabolic effects (Smith et al., 2019). While no formal definition of MHO exists, it is often described as either obesity with less than three components of MetSyn, or obesity without IR as computed by the homeostasis model assessment of insulin resistance (HOMA-IR) (Smith et al., 2019). Mechanisms proposed to mediate this include depressed ectopic fat accumulation, subcutaneous adipose tissue (SAT) expansion plasticity, and shifts in fat storage from the abdomen to the legs (Blüher, 2020; Smith et al., 2019; Loos and Kilpeläinen, 2018). In recent years the ability of abdominal obesity to mediate cardiometabolic disease risk has gained attention. People with MHO have less intra-abdominal fat accumulation compared to people with metabolically unhealthy obesity (Klöting et al., 2010; Karelis et al., 2005; Chen et al., 2015; Jennings et al., 2008; Hayes et al., 2010; Koster et al., 2010). Intra-abdominal fat accumulation can be approximated through the ratio of waist-to-hip circumference (WHR) adjusted for BMI (WHRadjBMI). WHRadjBMI is a causal factor that increases susceptibility for T2D, but the genetic and molecular mechanisms underlying fat distribution remain largely unknown (Emdin et al., 2017; Gill et al., 2021; Li et al., 2021). Understanding the mechanisms mediating WHRadjBMI, MHO, and T2D is critical to our understanding of disease pathogenesis and to clinical strategies to treat MetSyn.

Most of the genetic mechanisms of MHO described have been associated with increased BMI without increased disease risk. For example, the missense variant rs373863828 in CREB3 regulatory factor has been shown to increase BMI without a corresponding increase in HOMA-IR and circulating TG, or a decrease in circulating adiponectin (Minster et al., 2016). Ob/ob mice with overexpression of adiponectin but lacking in leptin are shown to accumulate considerable fat mass without a corresponding increase in insulin sensitivity (Kim et al., 2007). In contrast, the genetic loci associated with increased WHRadjBMI but without increased disease risk have not yet been described. To date, all genes that have been shown to increase fat accumulation into abdominal fat depots have also been shown to increase the risk for T2D (Fathzadeh et al., 2020; Small et al., 2018; Yang et al., 2022; Gesta et al., 2011; Loh et al., 2020; Loh et al., 2015).

As complex traits with both environmental and genetic risk factors, abdominal obesity and T2D have been the subject of multiple genome-wide association studies (GWAS). While GWAS have identified hundreds of genetic loci associated with abdominal obesity and T2D, moving from association to mechanism at a locus is not trivial. The use of colocalization analysis (COLOC), which identifies loci that contain shared genetic architecture for multiple traits of interest, can inform mechanistic hypotheses moving from association to function by integrating data from multiple studies (Hukku et al., 2021; Wallace, 2013; Wallace et al., 2012; Wallace, 2020). For example, the colocalization of a GWAS signal with genetic regulation of genes at quantitative trait loci (QTLs) implies a mechanistic relationship between the regulated gene and GWAS trait (Hormozdiari et al., 2016). Another recently developed approach named Tissue of ACTion scores for Investigating Complex trait-Associated Loci (TACTICAL) (Torres et al., 2020) incorporates gene expression data, and epigenetic annotations with GWAS associations to predict the causal eGenes and tissues of action at GWAS loci. These methods have been used to inform data-driven mechanistic predictions at GWAS loci that have been experimentally validated and can recall previously validated loci as positive controls.

To advance the understanding of mechanisms linking body fat distribution to T2D risk, independent of overall obesity, we used COLOC and TACTICAL to predict the mechanisms of action at genetic loci associated with both T2D and WHR, both adjusted for the BMI. Using the most recent GWAS summary statistics, QTL summary statistics, tissue-specific gene expression data, and high-resolution epigenetic annotations, we predicted the shared genetic architecture of T2DadjBMI and WHRadjBMI at 79 genetic loci. Here, we present the identification of five loci that contained association signals with discordant effects on abdominal fat and T2D risk, meaning that the allele of the lead variant associated with protection from T2D was associated with increased abdominal fat accumulation. We predicted the eGenes and tissues of action at these five loci and explored the relationship between adipose eGene expression with cellular and physiological phenotypes. Here, we provide data-driven hypotheses about predicted candidate causal eGenes at GWAS loci with associations that recall metabolically healthy abdominal obesity.

Results

Colocalization analysis of genetic loci associated with T2D and body fat distribution predicts colocalization of discordant T2DadjBMI and WHRadjBMI association signals at six association signals

To identify genetic loci which contained pleiotropic association signals for both T2DadjBMI and WHRadjBMI, we performed colocalization analysis (Figure 1A). This analysis yielded 79 genetic loci where a single variant was significantly associated with both T2DadjBMI and WHRadjBMI. We obtained the 99% credible set of variants in colocalized loci (Supplementary file 1) and discovered the presence of 143 variants in five loci associated with discordant effects on T2DadjBMI and WHRadjBMI. We also discovered 851 SNPs in 73 loci with the expected concordant effects on both traits. Although almost all of the representative lead discordant variants reached genome-wide significance, two associations reached nominal significance (p<5e-05). Because of recent work demonstrating that even variants with only nominal and local significance in GWAS can also have functional relevance to GWAS traits, we included variants prioritized in the 99% credible set but with only nominal significance (Li et al., 2020a). We then performed fine-mapping of the causal variants in each locus containing a discordant association signal while relaxing the assumption of a single causal variant per locus. In four of the five loci, this fine-mapping recalled only one likely candidate causal signal. In the 5q21.1 locus, SuSiE identified a secondary association signal that was also associated with discordant effects on T2DadjBMI and WHRadjBMI (Figure 1B and Supplementary file 2). To parse the associations between specific components of WHRadjBMI, including WC, HC, WHR, and BMI, with both T2D and T2DadjBMI, we performed multi-trait colocalization analysis with Hyprcoloc of the associations at discordant loci (Supplementary file 3). At three of the five discordant loci, the discordant association signals were also colocalized with WHRadjBMI component traits waist circumference (WC) and WHR.

Figure 1
Download asset Open asset
Analysis summary and discordant variant characteristics.

(A) Summary of analysis pipeline and generated results. Details of data sources are available in Supplementary file 1. (B) Effect size (WHRadjBMI) and odds ratio (T2DadjBMI) of lead genetic variant at discordant association signals. (C) Phenome-wide association study (PheWAS) of lead discordant genetic variant effect sizes on glycemic and anthropometric traits. From left to right: random glucose (RG), fasting glucose (FG), FG adjusted for body mass index (BMI) (FGadjBMI), fasting insulin adjusted for BMI (FIadjBMI), glycated hemoglobin (HBA1C), pancreatic fat percentage (PF), trunk fat ratio (TFR), visceral adipose tissue (VAT), VAT adjusted for BMI and height (VATadjBMIHeight), VAT to abdominal subcutaneous adipose tissue (VATtoASAT), VAT to gluteofemoral fat (VATtoFGAT), waist circumference, waist circumference adjusted for BMI (WCadjBMI), waist-to-hip ratio (WHR), and BMI. (D) Variant effect prediction of 99% credible set variants in discordant genetic loci.

Figure 1—source data 1

Genetic, transcriptomic, and epigenomic data sources in Figure 1.

https://cdn.elifesciences.org/articles/79834/elife-79834-fig1-data1-v1.xlsx

We next investigated the genetic and physiological consequences of discordant variants. We performed a phenome-wide association study (PheWAS) for anthropometric and glycemic traits with the most highly powered GWAS available (Figure 1—source data 1; Costanzo et al., 2023). We used the most highly powered GWAS or GWAS meta-analysis for each trait included in our PheWAS and queried the summary statistics for the associations of each lead discordant variant (Figure 1C). This query revealed consistent significant associations with discordance across anthropometric and glycemic traits in each locus. At the association signal in the 5q11.2 region, association signals exemplified this metabolic discordance. Represented by genetic variant rs459193, the association signal was associated with increased abdominal obesity in nearly every metric, but also with protection from T2D in nearly every metric. At all lead discordant variants, effects were consistent with a phenotype of increased abdominal obesity but protection from T2D.

We then queried the variant effect predictor (VEP) to discover genetic variant annotations (McLaren et al., 2016; Figure 1D and Supplementary file 4). VEP predicted that discordant variants overwhelmingly lie in noncoding regions of the genome, with only one missense variant in a coding region. Because the vast majority of discordant variants lie in noncoding regions, it is likely their function lies in altering genetic regulation of proximal genes (Civelek and Lusis, 2014). Therefore, we investigated the coincidence of these discordant variants with the genetic regulation of proximal genes with functional prediction methods.

Integration of molecular QTLs and genomic annotations to predict functional genes in tissues of action at discordant genetic loci

To investigate the role of eGenes in physiological phenotypes and cellular phenotypes, we evaluated the correlation of adipose tissue eGene expression and T2D-relevant phenotypes since these correlations can reveal biologically relevant functional relationships (Civelek et al., 2017). To predict the genes and tissues of function at discordant loci, we used publicly available multi-omic data from metabolically relevant tissue-specific resources to predict functional mechanisms underlying associations. We first interrogated where the 143 discordant variants in the credible set were located in relation to tissue-specific chromatin state data in pancreatic islet, adipose, liver, and skeletal muscle tissues (Kim et al., 2007). We computed the enrichment of colocalized association signals in various chromatin state annotations in each of these tissues (Figure 2A). We noted the specific enrichment of adipose tissue chromatin states of high activity, such as active transcription start sites, enhancer regions, and areas of transcriptional activity. For every other tissue, the leading annotations represented areas of decreased transcriptional activity. We additionally queried 3D chromatin data for discordant variant enhancer/promoter contact but did not find any significant interactions (Figure 2—source data 1). We then used these enrichment scores, chromatin states, and gene expression data to predict the functional tissues at each colocalized locus (Supplementary file 5). We predicted that adipose tissue was classified as the candidate tissue of action (TOA) at three loci, and skeletal muscle and liver tissue shared classification with adipose tissue at the remaining two discordant loci (Figure 2B).

Figure 2
Download asset Open asset
Predicting functional tissues and effector genes at discordant loci.

(A) Tissue-specific enrichment of chromatin states of variants in the 99% credible set of colocalized variants. (B) Tissue of action scores for association signals in the five discordant loci. Orange coloration indicates predicted adipose tissue of action at the locus, and blue coloration indicates shared tissue of action assignment at the locus. (C) Summary table of the expression quantitative trait loci (eQTL) and splicing QTL (sQTL) colocalizations with waist-to-hip circumference adjusted for the body mass index (WHRadjBMI) and T2DadjBMI for discordant loci. The expression effect direction is with respect to the protective type 2 diabetes allele. (D) Expression of predicted effector genes in discordant loci across cell types. From left to right: adipocyte progenitor stem cells (ASPC), lymphatic endothelial cells (LEC), smooth muscle cells (SMC), and natural killer cells (nk). Data was obtained from Emont et al., 2022.

Figure 2—source data 1

Genetic, transcriptomic, and epigenomic data sources used in Figure 2.

https://cdn.elifesciences.org/articles/79834/elife-79834-fig2-data1-v1.xlsx

To predict effector genes (eGenes) regulated by discordant variants, we next predicted the colocalization of QTLs with the WHRadjBMI and T2DadjBMI GWAS. Colocalization of a GWAS association signal with a genetic regulatory association signal can be used to prioritize mechanisms underlying association. We obtained expression QTL (eQTL) and splicing QTL (sQTL) summary statistics from multiple cohorts and tissue groups (Figure 2—source data 1). We extracted eQTL summary statistics for all genes within 1 Mb of the lead variant of all discordant colocalized loci from adipose, pancreatic, skeletal muscle, and liver tissues. We extracted sQTL summary statistics for all genes within 1 Mb of the lead variant of all discordant colocalized loci for adipose tissue data that was available. We used Summary-based Mendelian Randomization (SMR) and Coloc.abf to perform GWAS-QTL colocalization and used the framework developed by Hukku et al. to reconcile the results of SMR and Coloc.abf. In this framework, colocalization found using Coloc.abf but not with SMR potentially represents signals with horizontal pleiotropy, whereas colocalization found through SMR but not through Coloc.abf potentially represents locus-level colocalization (Hukku et al., 2021). Colocalization found using both methods represents the identification of candidate causal effector transcripts. Our colocalization analysis revealed seven candidate causal effector transcripts at three of the five discordant loci (Figure 2C). With Coloc.abf, we predicted four putative eGenes in these two loci. At the 2p21 locus, we predicted THADA-AS (SAT, VAT) to be the sole eGene. At the 5q21.1 locus, we predicted GIN1 (SAT), PAM (SAT & SKM), and PPIP5K2 (SAT) to be the eGenes. The association signal at rs6860588 was associated with a novel alternative splicing isoform of PAM in SAT, which skips the 14th exon. Using SMR, we predicted four eGenes at two discordant loci. At the 5q21.1 locus, we predicted that the genetic association signal represented by rs6860588 was also associated with the regulation of EIF3KP1 (SAT, VAT, SKM, PANC), PPIP5K2 (SAT), and GIN1 (SAT, SKM, PANC). At the discordant association signal in the 19q13.11 locus, we predicted that the genetic association signal represented by variant rs3786897 was also associated with the regulation of PEPD (SAT, VAT). As the colocalization transcripts GIN1 and PPIP5K2 were replicated with both methods (Supplementary files 6 and 7), these represent high-confidence predictions of potentially causal effector transcripts underlying the genetic association with discordance in the 5q21.1 locus. We queried white adipose tissue single-cell RNA sequencing data (Emont et al., 2022) for discordant association signal eGenes and found that eGenes were expressed in adipocytes and adipocyte progenitor stem cells (ASPCs) (Figure 2D). Because body fat distribution associations are driven by ASPCs and adipocytes in adipose tissues (Lu et al., 2016; Locke et al., 2015; Hansen et al., 2023), we reasoned that exploring adipose expression data could help to explain discordant associations. This multi-omic data enabled us to make high-confidence consensus predictions of tissues and eGenes of action at discordant loci.

Adipose gene expression analysis of discordant loci eGenes reveals dynamic expression in adipogenesis and relationships with metabolic physiology

To investigate the role of eGenes in physiological phenotypes and cellular phenotypes, we then evaluated the gene expression dynamics of eGenes in adipose tissue. Correlations between relevant tissue gene expression and metabolic phenotypes can reveal biologically relevant functional relationships (Civelek et al., 2017). We used SAT transcriptomic data from the 426 men of the METSIM cohort to investigate how adipose tissue expression of discordant locus eGenes was related to 23 metabolic phenotypes underlying T2D and abdominal fat accumulation (Figure 3—source data 1; Brotman et al., 2022; Laakso et al., 2017). We extracted adipose tissue gene expression data for eGenes. Gene expression data were available for six of the seven eGenes. We additionally extracted splice junction expression data for the only gene with a colocalized splice junction, PAM. We then computed the biweight midcorrelation of transcript counts or splice junction counts with 23 metabolic phenotypes. We found significant (false discovery rate [FDR] <0.05) correlations of adipose tissue gene expression of three genes with 13 phenotypes (Figure 3A). We found that adipose tissue expression of THADA-AS, PEPD, and GIN1 was significantly correlated with inflammatory, glycemic, and anthropometric phenotypes. SAT THADA-AS expression was positively correlated with IR, abdominal fat accumulation, and serum triglyceride levels, but with higher levels of plasma interleukin-1 receptor antagonist (IL-1RA) and C-reactive protein (CRP). IL-1RA plays a protective role in resolving inflammation (Volarevic et al., 2010), and elevated levels have been linked to prediabetes (Luotola, 2022; Grossmann et al., 2015). CRP has been used as a biomarker of increased inflammation in chronic diseases (Herwald and Egesten, 2021). The eQTL and GWAS data are associated with decreased expression of THADA-AS, which is consistent with the protection from IR in the correlation data but not with the increased abdominal obesity and inflammation. We are unable to resolve this correlation evidence with the discordance, but because the METSIM cohort was collected using single-end RNA sequencing, parsing the correlations of THADA and THADA-AS is difficult (Li et al., 2013). SAT expression of GIN1 was correlated with higher plasma adiponectin. Adiponectin, secreted by adipocytes, increases insulin sensitivity, and this provides a mechanism for protection from T2D (Achari and Jain, 2017). This expression is consistent with the QTL and GWAS data, providing a direct potential mechanism linking the eQTL to protection from T2D. SAT PEPD expression was also positively correlated with plasma IL-1RA levels. The QTL at this locus is associated with decreased expression of PEPD, providing another direct potential mechanism linking the eQTL to protection from T2D. Through this correlation analysis, we were able to predict the physiological consequences of eGenes at three discordant loci.

Figure 3
Download asset Open asset
Predicted physiological and cellular effects of effector genes (eGenes) on metabolic phenotypes and adipogenesis.

(A) Biweight midcorrelation of adipose tissue eGenes expression with metabolic phenotypes (false discovery rate [FDR] <5%). From left to right: Homeostatic model assessment of insulin resistance (HOMA-IR), high-density lipoproteins (HDL), low-density lipoproteins (LDL), interleukin-1 receptor agonist (IL1RA), C-reactive protein (CRP). (B) Dynamic expression of adipose tissue eGenes over 16-day adipogenesis time course in Simpson-Golabi-Behmel syndrome (SGBS) cells. We performed the likelihood ratio test (LRT) to evaluate if each gene was dynamically expressed over the time course. The p-value of the LRT is included.

Figure 3—source data 1

Genetic, transcriptomic, and epigenomic data sources in Figure 3.

https://cdn.elifesciences.org/articles/79834/elife-79834-fig3-data1-v1.xlsx

We next evaluated if eGenes identified in adipose tissues were dynamically expressed in adipogenesis. Dynamic gene expression in adipogenesis could point to the regulatory and structural roles of eGenes in adipogenesis (Nassiri et al., 2016; Anderson et al., 2020). We obtained time series ASPC adipogenesis time course data and evaluated eGenes for dynamic expression. Gene expression data were available for five of the seven eGenes. Because the expression data was single-stranded and unable to resolve forward or reverse-strand sequences, we included the probe for THADA to represent THADA-AS. We found that all eGenes except PPIP5K2 were dynamically expressed over a 16-day adipogenesis time course (Figure 3B), implying potential functional roles for these genes in regulating preadipocyte fate.

Integration of analysis to predict the functional genes and tissues of action at the discordant 5q21.1 locus

By predicting the mechanisms of action at discordant loci, we were able to generate specific hypotheses about the genes at each locus that underlie GWAS associations. We predicted that the causal discordant signal at the 5q21.1 locus was represented by variant rs6860588. The T allele of rs6860588 is associated with protection from T2D, increased abdominal obesity, decreased SAT expression of GIN1, increased SKM expression of PAM, decreased SAT expression of PPIP5K2, increased SAT expression of a PAM splice variant with a skipped exon 14, and decreased SAT expression of the canonical PAM splice junctions, exon 13:14 and exon 14:15 (Figure 4A; Figure 4—figure supplement 1 and Figure 4—source data 1). While the eGenes, PAM, GIN1, and PPIP5K2, have not been studied in the context of obesity and metabolism, they have been studied for their function in other cell types. We found that GIN1 and PAM were dynamically expressed over the course of adipogenesis (Figure 3B). GIN1 has been hypothesized to be a key regulator of energy metabolism in atria (Li et al., 2020b), but little is known about gypsy integrases and their molecular function. PAM facilitates C-terminus glycine residue amidation, which can catalyze protein potency (Thomsen et al., 2018; Merkler, 1994). PAM additionally has been linked to metabolic phenotypes in multiple model organisms, where its deficiency is associated with decreased peptide secretion and potency critical to insulin release, but not with increased diabetes (Chen et al., 2020a; Zieliński et al., 2016). PAM loss of function likely results in deficient peptide synthesis and secretion in adipocytes as well, and its increase of function likely results in increased myokine signaling from skeletal muscle. Knockdown of PPIP5Ks results in decreased proliferation, increased mitochondrial mass, decreased inositol metabolism, and accelerated glycolysis in tumor cell lines (Gu et al., 2021; Gu et al., 2017; Badodi et al., 2021). We did not observe significant interactions between adipose PPIP5K2 expression and adipogenesis or metabolic phenotypes, but this does not rule out a role for PPIP5K2 in the metabolic discordance at 5q21.1. Thus, we propose that the T allele at rs6860588 regulates a group of genes that promotes adipogenesis, glycolysis, and inflammation in white adipose tissue while simultaneously decreasing preadipocyte expansion and increasing skeletal muscle peptide secretion and potency (Figure 4B). This model is consistent with the TOA score and QTL analysis, which both predict skeletal muscle and adipose tissue contribution to the associations at the locus and reconcile the associations with abdominal obesity but protection from T2D associated with the T allele of rs6860588 (Figure 4C).

Figure 4 with 2 supplements see all
Download asset Open asset
Predicted model of effects associated with T allele at rs6860588.

(A) β of the T allele of discordant variant rs6860588 with respect to waist-to-hip circumference adjusted for the body mass index (WHRadjBMI), T2DadjBMI, and colocalized effector genes (eGenes). (B) Summary of associations with T allele at rs6860588. (C) Integrated model reconciling metabolic discordance with eGene-associated phenotypes in two tissues of action. Created with BioRender.com.

Figure 4—source data 1

Genetic, transcriptomic, and epigenomic data sources in Figure 4 and Figure 4—figure supplements 1 and 2 .

https://cdn.elifesciences.org/articles/79834/elife-79834-fig4-data1-v1.xlsx

Discussion

We report here the integration of multi-omic data spanning the genome, transcriptome, and epigenome to predict functional genes and tissues underlying genetic signals associated with abdominal obesity but protection from T2D. We predicted the colocalization of T2DadjBMI and WHRadjBMI association signals at 79 genetic loci. The protective allele of six association signals was associated with lower T2D risk but higher abdominal fat accumulation, independent of overall obesity (Figure 1). By analyzing colocalization with molecular QTLs, computing the enrichment of variants in epigenomic and genomic annotations, and comparing tissue-specific gene expression, we predicted the eGenes and tissues of action at discordant association signals (Figure 2). We found significant evidence that adipose tissue biology is a significant contributor at colocalized loci. We then explored the effects of eGenes expression in adipose tissue and preadipocytes on adipogenesis metabolic phenotypes (Figure 3) and proposed a model by which the genetic variant rs6860588 might confer protection from T2D but increased abdominal obesity (Figure 4).

The six genetic association signals associated with discordant metabolic phenotypes offer potential insight into the genetic mechanisms underlying risk stratification of T2D risk within abdominal obesity. While mechanisms promoting MHO have been described, most have focused on body fat distribution. Defining more mechanisms that promote MHO is critical as rates of obesity rise globally. Complicating the study of MHO is the lack of precision in its definition. Some definitions include obesity with less than three components of MetSyn, obesity with healthy HOMA-IR, or even obesity with the lack of a metabolic and cardiovascular disorder (Blüher, 2020). MHO has been controversial and termed an intermediate state (Caleyachetty et al., 2017; Rey-López et al., 2015; Blüher, 2017), but a growing body of evidence has accumulated providing evidence that genetic mechanisms influence predisposition to it. In Samoans, the common CREBRF coding variant rs12513649 increases BMI and overall adiposity but protects from IR (Li et al., 2020a). Additionally, IRS1, COBLL1, PLA2G6, and TOMM40 have been associated with higher BMI but with protective lipidemic and glycemic traits (Loos and Kilpeläinen, 2018). The physiological functions of these genes have been proposed to involve adipose tissue caloric load capacity and body fat distribution (Loos and Kilpeläinen, 2018; Lu et al., 2016; Kilpeläinen et al., 2011; Lotta et al., 2017).

While abdominal fat accumulation is known to be one of the strongest predictors of obesity-related complications (Emdin et al., 2017; Censin et al., 2019; Dale et al., 2017), our findings point to mechanisms that contradict this trend. Each locus must be functionally annotated before translating the association results to the clinic. If these discordant variants are functionally annotated and fully characterized, they might have clinical utility to T2D risk allele carriers and inform personalized therapeutic strategies. Discovering mechanisms uncoupling abdominal obesity from T2D can aid in personalized therapeutic strategies and in understanding personalized risk stratification. Risk-stratified personalized obesity treatment could prioritize patients that would or would not benefit significantly from weight-loss interventions, and use genotype as a biomarker for patients who would benefit from other therapeutic strategies (Klonoff, 2008; Williams et al., 2022). Thus, the importance of personalized risk stratification for T2D will only increase as abdominal obesity becomes more prevalent. Personalized risk stratification with an understanding of specific molecular, cellular, and physiological mechanisms will aid in the prioritization of effective therapies. This investigation provided specific hypotheses linking functional genes at discordant loci to tissues of action for experimental follow-up in vitro and in vivo. Functional characterization of the effect of these genes on insulin uptake, preadipocyte proliferation, and adipogenesis, as well as secretome characterization, will elucidate precise mechanisms through which these eGenes might contribute to the discordant association signals.

We predicted tissues and mechanisms of action at five loci containing six discordant association signals with increased abdominal obesity and protection for T2D. A particular example of a peculiar metabolic discordance was revealed at the 2p21 locus containing THADA and THADA-AS, represented by variant rs6752964 (Figure 4—figure supplement 2). The associations have been replicated multiple times (Mahajan et al., 2018; Zeggini et al., 2008; Grarup et al., 2008), but the exact mechanisms underlying this association are unknown. THADA plays an evolutionarily conserved role in intracellular calcium signaling and consequently non-shivering thermogenesis. In Drosophila melanogaster, thada knockout flies developed obesity and hyperphagia without altered circulating glucose levels (Moraru et al., 2017). In mice, pancreatic Thada knockout resulted in protection from T2D through the preservation of β-cell mass and improvement of β-cell function (Zhang et al., 2023). Mendelian randomization studies in humans have likewise found consistent links between THADA and adiposity, but have not yet been able to link it to diabetic phenotypes such as insulin secretion (Grarup et al., 2008; Simonis-Bik et al., 2010). Our investigation revealed relationships between THADA and THADA-AS expression with diabetic and obesity-abdominal obesity phenotypes, as well as dynamic expression in adipogenesis (Figure 3). Regulatory interactions whereby THADA-AS expression interferes with THADA transcription could provide a basis by which variant rs6752964 might confer abdominal obesity, but protection from T2D (Brantl, 2002; Faghihi and Wahlestedt, 2009; Wight and Werner, 2013). Further, we also found colocalization of genetic regulation of PEPD in adipose tissue with the discordant association signal represented by variant rs3786897. Depletion of PEPD in preadipocytes has been shown to reduce adipogenic potential, decrease triglyceride accumulation, and phospho-Akt signaling, which is critical to insulin sensitivity (Chen et al., 2020b). Notably, a secondary signal represented by variant rs731839 was apparent in this locus but was not significant for WHRadjBMI. This signal has been associated with sex-specific effects on serum lipid levels in Han and Mulao populations (Lin et al., 2014). Further in vivo and in vitro work must be done to resolve this multi-tissue, multi-eGene locus.

Although our analysis incorporated genome, transcriptome, epigenome, and phenome data in multiple cohorts, and used the consensus of orthogonal methods to predict the mechanisms of action at discordant loci, follow-up is required to validate each prediction. Additionally, our genetic expression data used single-strand sequencing, and therefore parsing out the associated effects of sense and antisense transcripts is difficult. Finally, it is critical to discover to diversify ancestry and sex in genetic association studies to identify more genetic loci underlying MHO. Without experimental follow-up and extensive clinical studies, genotype should not be used as a diagnostic metric. CRISPR editing of alleles in relevant cell types to study cis-regulatory effects on genes and phenotypic effects on cells, and work in animal models is necessary to fully annotate these loci. In addition, it is important to identify the indirect and direct effects of discordant variants, as these endocrine tissues are major contributors to peptide and hormone secretion. Further experimental characterization is critical to placing these results in the proper context and providing the basis for personalized interventions for T2D. The predictions at these six signals provide specific hypotheses to be tested, and should they be validated experimentally provide knowledge of the precise mechanisms of uncoupling obesity from T2D risk.

Methods

GWAS-GWAS colocalization analysis

GWAS results for T2DadjBMI and WHRadjBMI were obtained from Mahajan et al., 2018, and Pulit et al., 2019. The set of single nucleotide polymorphisms (SNPs) within 500 kb of a genome-wide significant SNP in either GWAS was included in the colocalization test. Rare variants, defined as SNPs reported to have effect allele frequencies of less than 1% in either GWAS, were excluded. Proximal analysis windows (>250 kb) were merged, and the colocalization test was performed on these genetic loci with three methods: Coloc.abf (Wallace, 2020), Hyprcoloc (Foley et al., 2021), and visual inspection of LocusCompare plots (Liu et al., 2019).

The default parameters were used for Hyprcoloc. In Coloc.abf, the default parameters for p1 and p2 prior probabilities were used for the individual GWAS hypotheses. The parameter p12, the prior for single variant colocalization, was set to 5e-06 as prescribed by Wallace, 2020, to balance false negative and positive results. Loci were considered colocalized if the regional probability of colocalization was greater than 0.70. In Coloc.abf, this was the sum of the PPH3 and PPH4 statistics, and in Hyprcoloc this was the regional probability statistic. Loci that met colocalization criteria in either method were plotted using LocusCompare with the default European ancestry linkage disequilibrium (LD) data from 1000 Genomes (Fairley et al., 2020) and with genome build hg19. This resulted in 121 LocusCompare plots on which visual inspection was performed to verify colocalized genetic association signals. If genetic loci were considered colocalized by at least two of the three colocalization analysis methods, we considered these consensus colocalized loci. We termed this consensus analysis ‘COLOC’.

Discordant locus identification

We obtained the 99% credible set of SNPs from the results of Bayesian factor analysis implemented through Coloc.abf at each locus. We calculated the Z-scores for the association test of each genetic variant and the GWAS trait. If the Z-score associated with SNP had the opposite sign for association with WHRadjBMI and T2DadjBMI with respect to the same allele and the p-value for the association with both traits was less than 1e-05, we considered the variant discordant. We then identified in which loci the SNPs were located, and queried haploReg (Ward and Kellis, 2016) LD data with the haploR package in R (Zhbannikov et al., 2017) to separate signals in the same loci using LD clumping (R2 > 0.50) on the discordant variants.

Phenome-wide association study

We queried the GWAS meta-analysis associations of glycemic and anthropometric traits for each lead discordant variant in the Type 2 Diabetes Knowledge Portal (T2DKP) (Costanzo et al., 2023). We additionally obtained the summary statistics of abdominal fat MRI scans in the UK Biobank and queried these summary statistics for discordant variants (Liu et al., 2021).

Multi-trait colocalization analysis

We obtained GWAS summary statistics for WC, hip circumference (HC), WHR, WHRadjBMI, T2D, and T2DadjBMI. We extracted summary statistics of variants within genetic loci containing a discordant association signal (Mahajan et al., 2018; Pulit et al., 2019) and performed multi-trait colocalization with Hyprcoloc (Foley et al., 2021). We considered an association signal colocalized for multiple traits if Hyprcoloc computed a posterior probability for both body fat distribution traits (WC, HC, WHR, and WHRadjBMI) and T2D or T2DadjBMI.

Fine-mapping analysis

We performed variable selection in multiple regression as implemented in the R package SuSiE (Wang et al., 2020). This method implements the sum of single-effects models to fine-map the causal variant(s) in a locus. Using the T2DadjBMI and WHRadjBMI GWAS summary statistics and the 1000 Genomes LD data, we performed fine-mapping of loci containing a genetic variant associated with discordant effects on T2DadjBMI and WHRadjBMI. We used the default flag options in SuSiE and performed a sensitivity analysis of the results to a range of priors. We selected causal variants with a PPH4 greater than 0.70.

GWAS-QTL colocalization analysis

We obtained eQTL data from the Genotype-Tissue Expression (GTEx) for 49 tissues (Battle et al., 2017), the Stockholm-Tartu Atherosclerosis Reverse Networks Engineering Task (STARNET) cohort for 6 tissues (Franzén et al., 2016), and the Metabolic Syndrome in Men (METSIM) for SAT (Brotman et al., 2022). We also obtained SAT sQTL results from the METSIM cohort. Data sources and further information are detailed in Figure 2—source data 1. We extracted the QTL data for each gene or transcript within 1 Mb of a discordant locus start or end site and independently colocalized with the T2DadjBMI and WHRadjBMI GWAS using Coloc.abf and SMR. When implementing Coloc.abf, we considered a signal to be colocalized if PPH4 was greater than 0.50 (a threshold used for GWAS-QTL colocalization in admixed populations; Gay et al., 2020). We repeated the analysis in SMR and used an FDR threshold of 5% to control for false positives. We then performed a visual inspection of GWAS-QTL colocalization of plots generated by LocusCompare. If a GWAS-QTL colocalization met these criteria, the proximal gene was termed an eGene.

fGWAS annotation enrichment analysis

We used the functional GWAS (fGWAS) (Pickrell, 2014) command-line tool to compute the enrichment of associations, in particular genomic and epigenomic regions. We first obtained the chromosome and base-pair position of each variant in the 99% credible set from each of the 79 colocalized loci. We mapped the SNPs to their placement in genomic regions using bed files. We used bed files from tissue-specific chromatin state data (adipose, liver, pancreatic islet, and skeletal muscle) and genome-level coding region annotations, and mapped SNPs to their presence in these regions. From these maps, we performed enrichment analysis with the complete model of all annotations with the -fine and -xv flags on fGWAS. We used the natural log of the Bayes factor of the colocalization test and computed the enrichment of SNPs for presence in coding regions to genetic and epigenetic annotations.

TOA analysis

We conducted TOA score analysis using the credible set of SNPs from each of the 79 colocalized loci. TACTICAL computes the TOA score with the SNP-level Bayesian probabilities, the SNP annotation maps, and the annotation enrichment scores. We used the Coloc.abf PPH4 scores for the SNP-level Bayesian probability, the fGWAS annotation enrichment scores, and the SNP annotation maps to compute the TOA score at all colocalized loci. We separated independent association signals in the same loci (LD <0.5) with HaploReg (Ward and Kellis, 2016). With TACTICAL (Torres et al., 2020), we integrated the credible set of SNPs with the enrichment for genome-level and tissue-specific annotations. We used the default tissue classification thresholds of 0.20 to classify signals as belonging to a particular TOA and less than 0.10 difference to classify signals as sharing TOA assignments between multiple tissues.

Gene expression and phenotype correlation analysis

For each eGene, we computed the biweight midcorrelation and its significance, as implemented by the Weighted Genetic Coexpression Network Analysis (WGCNA) package (Langfelder and Horvath, 2008), between gene expression with metabolic phenotypes measured in the METSIM cohort (Laakso et al., 2017). We controlled for false positives with a 5% FDR threshold as implemented by the q-values package in R (John, 2002).

Adipogenesis gene expression dynamics analysis

We obtained Simpson-Golabi-Behmel syndrome preadipocyte adipogenesis time series gene expression data from GEO (accession number GSE76131) (Nassiri et al., 2016). We evaluated the dynamic expression of each adipose tissue eGene by fitting the gene expression over time to a linear model and applying the likelihood ratio test (LRT) to compare the time-dependent models to time-independent null models. We considered an eGene to be dynamically expressed in adipogenesis if the p-value of the LRT was less than 0.05.

Data availability

The current manuscript is a computational investigation using publicly available data, so no data have been generated for this manuscript. All publicly obtained data sets are included in Supplementary Table 1. All analysis and figure-generating code uploaded to the following Github repository: https://github.com/aberrations/predicting-functional-mechanisms-discordant-loci, (copy archived at swh:1:rev:e457dc5b8ecdc6abb119730bc64014876e9d852e).

The following previously published data sets were used
    1. Pulit SL
    2. Stoneman C
    3. Morris AP
    (2018) Zenodo
    Meta-analysis of Body Fat Distribution GWAS.
    https://doi.org/10.5281/zenodo.1251813
    1. Koplev S
    (2022) STARNET
    ID 10.1038/s44161-021-00009-1. Stockholm-Tartu Atherosclerosis Reverse Networks Engineering Task.
    http://starnet.mssm.edu/
    1. Nassiri I
    (2016) NCBI Gene Expression Omnibus
    ID GSE76131. Network Activity Score Finder analysis of SGBS cells.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE76131

References

    1. John DS
    (2002) A direct approach to false discovery rates
    Journal of the Royal Statistical Society Series B 64:479–498.
    https://doi.org/10.1111/1467-9868.00346
    1. Lin Q-Z
    2. Yin R-X
    3. Wu J
    4. Guo T
    5. Wang W
    6. Sun J-Q
    7. Shi G-Y
    8. Shen S-W
    9. Wu J-Z
    10. Pan S-L
    (2014)
    Sex-specific Association of the Peptidase D Gene Rs731839 polymorphism and serum lipid levels in the Mulao and Han populations
    International Journal of Clinical and Experimental Pathology 7:4156–4172.
    1. Locke AE
    2. Kahali B
    3. Berndt SI
    4. Justice AE
    5. Pers TH
    6. Day FR
    7. Powell C
    8. Vedantam S
    9. Buchkovich ML
    10. Yang J
    11. Croteau-Chonka DC
    12. Esko T
    13. Fall T
    14. Ferreira T
    15. Gustafsson S
    16. Kutalik Z
    17. Luan J
    18. Mägi R
    19. Randall JC
    20. Winkler TW
    21. Wood AR
    22. Workalemahu T
    23. Faul JD
    24. Smith JA
    25. Zhao JH
    26. Zhao W
    27. Chen J
    28. Fehrmann R
    29. Hedman ÅK
    30. Karjalainen J
    31. Schmidt EM
    32. Absher D
    33. Amin N
    34. Anderson D
    35. Beekman M
    36. Bolton JL
    37. Bragg-Gresham JL
    38. Buyske S
    39. Demirkan A
    40. Deng G
    41. Ehret GB
    42. Feenstra B
    43. Feitosa MF
    44. Fischer K
    45. Goel A
    46. Gong J
    47. Jackson AU
    48. Kanoni S
    49. Kleber ME
    50. Kristiansson K
    51. Lim U
    52. Lotay V
    53. Mangino M
    54. Leach IM
    55. Medina-Gomez C
    56. Medland SE
    57. Nalls MA
    58. Palmer CD
    59. Pasko D
    60. Pechlivanis S
    61. Peters MJ
    62. Prokopenko I
    63. Shungin D
    64. Stančáková A
    65. Strawbridge RJ
    66. Sung YJ
    67. Tanaka T
    68. Teumer A
    69. Trompet S
    70. van der Laan SW
    71. van Setten J
    72. Van Vliet-Ostaptchouk JV
    73. Wang Z
    74. Yengo L
    75. Zhang W
    76. Isaacs A
    77. Albrecht E
    78. Ärnlöv J
    79. Arscott GM
    80. Attwood AP
    81. Bandinelli S
    82. Barrett A
    83. Bas IN
    84. Bellis C
    85. Bennett AJ
    86. Berne C
    87. Blagieva R
    88. Blüher M
    89. Böhringer S
    90. Bonnycastle LL
    91. Böttcher Y
    92. Boyd HA
    93. Bruinenberg M
    94. Caspersen IH
    95. Chen Y-DI
    96. Clarke R
    97. Daw EW
    98. de Craen AJM
    99. Delgado G
    100. Dimitriou M
    101. Doney ASF
    102. Eklund N
    103. Estrada K
    104. Eury E
    105. Folkersen L
    106. Fraser RM
    107. Garcia ME
    108. Geller F
    109. Giedraitis V
    110. Gigante B
    111. Go AS
    112. Golay A
    113. Goodall AH
    114. Gordon SD
    115. Gorski M
    116. Grabe H-J
    117. Grallert H
    118. Grammer TB
    119. Gräßler J
    120. Grönberg H
    121. Groves CJ
    122. Gusto G
    123. Haessler J
    124. Hall P
    125. Haller T
    126. Hallmans G
    127. Hartman CA
    128. Hassinen M
    129. Hayward C
    130. Heard-Costa NL
    131. Helmer Q
    132. Hengstenberg C
    133. Holmen O
    134. Hottenga J-J
    135. James AL
    136. Jeff JM
    137. Johansson Å
    138. Jolley J
    139. Juliusdottir T
    140. Kinnunen L
    141. Koenig W
    142. Koskenvuo M
    143. Kratzer W
    144. Laitinen J
    145. Lamina C
    146. Leander K
    147. Lee NR
    148. Lichtner P
    149. Lind L
    150. Lindström J
    151. Lo KS
    152. Lobbens S
    153. Lorbeer R
    154. Lu Y
    155. Mach F
    156. Magnusson PKE
    157. Mahajan A
    158. McArdle WL
    159. McLachlan S
    160. Menni C
    161. Merger S
    162. Mihailov E
    163. Milani L
    164. Moayyeri A
    165. Monda KL
    166. Morken MA
    167. Mulas A
    168. Müller G
    169. Müller-Nurasyid M
    170. Musk AW
    171. Nagaraja R
    172. Nöthen MM
    173. Nolte IM
    174. Pilz S
    175. Rayner NW
    176. Renstrom F
    177. Rettig R
    178. Ried JS
    179. Ripke S
    180. Robertson NR
    181. Rose LM
    182. Sanna S
    183. Scharnagl H
    184. Scholtens S
    185. Schumacher FR
    186. Scott WR
    187. Seufferlein T
    188. Shi J
    189. Smith AV
    190. Smolonska J
    191. Stanton AV
    192. Steinthorsdottir V
    193. Stirrups K
    194. Stringham HM
    195. Sundström J
    196. Swertz MA
    197. Swift AJ
    198. Syvänen A-C
    199. Tan S-T
    200. Tayo BO
    201. Thorand B
    202. Thorleifsson G
    203. Tyrer JP
    204. Uh H-W
    205. Vandenput L
    206. Verhulst FC
    207. Vermeulen SH
    208. Verweij N
    209. Vonk JM
    210. Waite LL
    211. Warren HR
    212. Waterworth D
    213. Weedon MN
    214. Wilkens LR
    215. Willenborg C
    216. Wilsgaard T
    217. Wojczynski MK
    218. Wong A
    219. Wright AF
    220. Zhang Q
    221. LifeLines Cohort Study
    222. Brennan EP
    223. Choi M
    224. Dastani Z
    225. Drong AW
    226. Eriksson P
    227. Franco-Cereceda A
    228. Gådin JR
    229. Gharavi AG
    230. Goddard ME
    231. Handsaker RE
    232. Huang J
    233. Karpe F
    234. Kathiresan S
    235. Keildson S
    236. Kiryluk K
    237. Kubo M
    238. Lee J-Y
    239. Liang L
    240. Lifton RP
    241. Ma B
    242. McCarroll SA
    243. McKnight AJ
    244. Min JL
    245. Moffatt MF
    246. Montgomery GW
    247. Murabito JM
    248. Nicholson G
    249. Nyholt DR
    250. Okada Y
    251. Perry JRB
    252. Dorajoo R
    253. Reinmaa E
    254. Salem RM
    255. Sandholm N
    256. Scott RA
    257. Stolk L
    258. Takahashi A
    259. Tanaka T
    260. van ’t Hooft FM
    261. Vinkhuyzen AAE
    262. Westra H-J
    263. Zheng W
    264. Zondervan KT
    265. ADIPOGen Consortium
    266. AGEN-BMI Working Group
    267. CARDIOGRAMplusC4D Consortium
    268. CKDGen Consortium
    269. GLGC
    270. ICBP
    271. MAGIC Investigators
    272. MuTHER Consortium
    273. MIGen Consortium
    274. PAGE Consortium
    275. ReproGen Consortium
    276. GENIE Consortium
    277. International Endogene Consortium
    278. Heath AC
    279. Arveiler D
    280. Bakker SJL
    281. Beilby J
    282. Bergman RN
    283. Blangero J
    284. Bovet P
    285. Campbell H
    286. Caulfield MJ
    287. Cesana G
    288. Chakravarti A
    289. Chasman DI
    290. Chines PS
    291. Collins FS
    292. Crawford DC
    293. Cupples LA
    294. Cusi D
    295. Danesh J
    296. de Faire U
    297. den Ruijter HM
    298. Dominiczak AF
    299. Erbel R
    300. Erdmann J
    301. Eriksson JG
    302. Farrall M
    303. Felix SB
    304. Ferrannini E
    305. Ferrières J
    306. Ford I
    307. Forouhi NG
    308. Forrester T
    309. Franco OH
    310. Gansevoort RT
    311. Gejman PV
    312. Gieger C
    313. Gottesman O
    314. Gudnason V
    315. Gyllensten U
    316. Hall AS
    317. Harris TB
    318. Hattersley AT
    319. Hicks AA
    320. Hindorff LA
    321. Hingorani AD
    322. Hofman A
    323. Homuth G
    324. Hovingh GK
    325. Humphries SE
    326. Hunt SC
    327. Hyppönen E
    328. Illig T
    329. Jacobs KB
    330. Jarvelin M-R
    331. Jöckel K-H
    332. Johansen B
    333. Jousilahti P
    334. Jukema JW
    335. Jula AM
    336. Kaprio J
    337. Kastelein JJP
    338. Keinanen-Kiukaanniemi SM
    339. Kiemeney LA
    340. Knekt P
    341. Kooner JS
    342. Kooperberg C
    343. Kovacs P
    344. Kraja AT
    345. Kumari M
    346. Kuusisto J
    347. Lakka TA
    348. Langenberg C
    349. Marchand LL
    350. Lehtimäki T
    351. Lyssenko V
    352. Männistö S
    353. Marette A
    354. Matise TC
    355. McKenzie CA
    356. McKnight B
    357. Moll FL
    358. Morris AD
    359. Morris AP
    360. Murray JC
    361. Nelis M
    362. Ohlsson C
    363. Oldehinkel AJ
    364. Ong KK
    365. Madden PAF
    366. Pasterkamp G
    367. Peden JF
    368. Peters A
    369. Postma DS
    370. Pramstaller PP
    371. Price JF
    372. Qi L
    373. Raitakari OT
    374. Rankinen T
    375. Rao DC
    376. Rice TK
    377. Ridker PM
    378. Rioux JD
    379. Ritchie MD
    380. Rudan I
    381. Salomaa V
    382. Samani NJ
    383. Saramies J
    384. Sarzynski MA
    385. Schunkert H
    386. Schwarz PEH
    387. Sever P
    388. Shuldiner AR
    389. Sinisalo J
    390. Stolk RP
    391. Strauch K
    392. Tönjes A
    393. Trégouët D-A
    394. Tremblay A
    395. Tremoli E
    396. Virtamo J
    397. Vohl M-C
    398. Völker U
    399. Waeber G
    400. Willemsen G
    401. Witteman JC
    402. Zillikens MC
    403. Adair LS
    404. Amouyel P
    405. Asselbergs FW
    406. Assimes TL
    407. Bochud M
    408. Boehm BO
    409. Boerwinkle E
    410. Bornstein SR
    411. Bottinger EP
    412. Bouchard C
    413. Cauchi S
    414. Chambers JC
    415. Chanock SJ
    416. Cooper RS
    417. de Bakker PIW
    418. Dedoussis G
    419. Ferrucci L
    420. Franks PW
    421. Froguel P
    422. Groop LC
    423. Haiman CA
    424. Hamsten A
    425. Hui J
    426. Hunter DJ
    427. Hveem K
    428. Kaplan RC
    429. Kivimaki M
    430. Kuh D
    431. Laakso M
    432. Liu Y
    433. Martin NG
    434. März W
    435. Melbye M
    436. Metspalu A
    437. Moebus S
    438. Munroe PB
    439. Njølstad I
    440. Oostra BA
    441. Palmer CNA
    442. Pedersen NL
    443. Perola M
    444. Pérusse L
    445. Peters U
    446. Power C
    447. Quertermous T
    448. Rauramaa R
    449. Rivadeneira F
    450. Saaristo TE
    451. Saleheen D
    452. Sattar N
    453. Schadt EE
    454. Schlessinger D
    455. Slagboom PE
    456. Snieder H
    457. Spector TD
    458. Thorsteinsdottir U
    459. Stumvoll M
    460. Tuomilehto J
    461. Uitterlinden AG
    462. Uusitupa M
    463. van der Harst P
    464. Walker M
    465. Wallaschofski H
    466. Wareham NJ
    467. Watkins H
    468. Weir DR
    469. Wichmann H-E
    470. Wilson JF
    471. Zanen P
    472. Borecki IB
    473. Deloukas P
    474. Fox CS
    475. Heid IM
    476. O’Connell JR
    477. Strachan DP
    478. Stefansson K
    479. van Duijn CM
    480. Abecasis GR
    481. Franke L
    482. Frayling TM
    483. McCarthy MI
    484. Visscher PM
    485. Scherag A
    486. Willer CJ
    487. Boehnke M
    488. Mohlke KL
    489. Lindgren CM
    490. Beckmann JS
    491. Barroso I
    492. North KE
    493. Ingelsson E
    494. Hirschhorn JN
    495. Loos RJF
    496. Speliotes EK
    (2015) Genetic studies of body mass index yield new insights for obesity biology
    Nature 518:197–206.
    https://doi.org/10.1038/nature14177
    1. Lu Y
    2. Day FR
    3. Gustafsson S
    4. Buchkovich ML
    5. Na J
    6. Bataille V
    7. Cousminer DL
    8. Dastani Z
    9. Drong AW
    10. Esko T
    11. Evans DM
    12. Falchi M
    13. Feitosa MF
    14. Ferreira T
    15. Hedman ÅK
    16. Haring R
    17. Hysi PG
    18. Iles MM
    19. Justice AE
    20. Kanoni S
    21. Lagou V
    22. Li R
    23. Li X
    24. Locke A
    25. Lu C
    26. Mägi R
    27. Perry JRB
    28. Pers TH
    29. Qi Q
    30. Sanna M
    31. Schmidt EM
    32. Scott WR
    33. Shungin D
    34. Teumer A
    35. Vinkhuyzen AAE
    36. Walker RW
    37. Westra H-J
    38. Zhang M
    39. Zhang W
    40. Zhao JH
    41. Zhu Z
    42. Afzal U
    43. Ahluwalia TS
    44. Bakker SJL
    45. Bellis C
    46. Bonnefond A
    47. Borodulin K
    48. Buchman AS
    49. Cederholm T
    50. Choh AC
    51. Choi HJ
    52. Curran JE
    53. de Groot L
    54. De Jager PL
    55. Dhonukshe-Rutten RAM
    56. Enneman AW
    57. Eury E
    58. Evans DS
    59. Forsen T
    60. Friedrich N
    61. Fumeron F
    62. Garcia ME
    63. Gärtner S
    64. Han B-G
    65. Havulinna AS
    66. Hayward C
    67. Hernandez D
    68. Hillege H
    69. Ittermann T
    70. Kent JW
    71. Kolcic I
    72. Laatikainen T
    73. Lahti J
    74. Leach IM
    75. Lee CG
    76. Lee J-Y
    77. Liu T
    78. Liu Y
    79. Lobbens S
    80. Loh M
    81. Lyytikäinen L-P
    82. Medina-Gomez C
    83. Michaëlsson K
    84. Nalls MA
    85. Nielson CM
    86. Oozageer L
    87. Pascoe L
    88. Paternoster L
    89. Polašek O
    90. Ripatti S
    91. Sarzynski MA
    92. Shin CS
    93. Narančić NS
    94. Spira D
    95. Srikanth P
    96. Steinhagen-Thiessen E
    97. Sung YJ
    98. Swart KMA
    99. Taittonen L
    100. Tanaka T
    101. Tikkanen E
    102. van der Velde N
    103. van Schoor NM
    104. Verweij N
    105. Wright AF
    106. Yu L
    107. Zmuda JM
    108. Eklund N
    109. Forrester T
    110. Grarup N
    111. Jackson AU
    112. Kristiansson K
    113. Kuulasmaa T
    114. Kuusisto J
    115. Lichtner P
    116. Luan J
    117. Mahajan A
    118. Männistö S
    119. Palmer CD
    120. Ried JS
    121. Scott RA
    122. Stancáková A
    123. Wagner PJ
    124. Demirkan A
    125. Döring A
    126. Gudnason V
    127. Kiel DP
    128. Kühnel B
    129. Mangino M
    130. Mcknight B
    131. Menni C
    132. O’Connell JR
    133. Oostra BA
    134. Shuldiner AR
    135. Song K
    136. Vandenput L
    137. van Duijn CM
    138. Vollenweider P
    139. White CC
    140. Boehnke M
    141. Boettcher Y
    142. Cooper RS
    143. Forouhi NG
    144. Gieger C
    145. Grallert H
    146. Hingorani A
    147. Jørgensen T
    148. Jousilahti P
    149. Kivimaki M
    150. Kumari M
    151. Laakso M
    152. Langenberg C
    153. Linneberg A
    154. Luke A
    155. Mckenzie CA
    156. Palotie A
    157. Pedersen O
    158. Peters A
    159. Strauch K
    160. Tayo BO
    161. Wareham NJ
    162. Bennett DA
    163. Bertram L
    164. Blangero J
    165. Blüher M
    166. Bouchard C
    167. Campbell H
    168. Cho NH
    169. Cummings SR
    170. Czerwinski SA
    171. Demuth I
    172. Eckardt R
    173. Eriksson JG
    174. Ferrucci L
    175. Franco OH
    176. Froguel P
    177. Gansevoort RT
    178. Hansen T
    179. Harris TB
    180. Hastie N
    181. Heliövaara M
    182. Hofman A
    183. Jordan JM
    184. Jula A
    185. Kähönen M
    186. Kajantie E
    187. Knekt PB
    188. Koskinen S
    189. Kovacs P
    190. Lehtimäki T
    191. Lind L
    192. Liu Y
    193. Orwoll ES
    194. Osmond C
    195. Perola M
    196. Pérusse L
    197. Raitakari OT
    198. Rankinen T
    199. Rao DC
    200. Rice TK
    201. Rivadeneira F
    202. Rudan I
    203. Salomaa V
    204. Sørensen TIA
    205. Stumvoll M
    206. Tönjes A
    207. Towne B
    208. Tranah GJ
    209. Tremblay A
    210. Uitterlinden AG
    211. van der Harst P
    212. Vartiainen E
    213. Viikari JS
    214. Vitart V
    215. Vohl M-C
    216. Völzke H
    217. Walker M
    218. Wallaschofski H
    219. Wild S
    220. Wilson JF
    221. Yengo L
    222. Bishop DT
    223. Borecki IB
    224. Chambers JC
    225. Cupples LA
    226. Dehghan A
    227. Deloukas P
    228. Fatemifar G
    229. Fox C
    230. Furey TS
    231. Franke L
    232. Han J
    233. Hunter DJ
    234. Karjalainen J
    235. Karpe F
    236. Kaplan RC
    237. Kooner JS
    238. McCarthy MI
    239. Murabito JM
    240. Morris AP
    241. Bishop JAN
    242. North KE
    243. Ohlsson C
    244. Ong KK
    245. Prokopenko I
    246. Richards JB
    247. Schadt EE
    248. Spector TD
    249. Widén E
    250. Willer CJ
    251. Yang J
    252. Ingelsson E
    253. Mohlke KL
    254. Hirschhorn JN
    255. Pospisilik JA
    256. Zillikens MC
    257. Lindgren C
    258. Kilpeläinen TO
    259. Loos RJF
    (2016) New Loci for body fat percentage reveal link between Adiposity and Cardiometabolic disease risk
    Nature Communications 7:10495.
    https://doi.org/10.1038/ncomms10495

Article and author information

Author details

  1. Yonathan Tamrat Aberra

    1. Department of Biomedical Engineering, University of Virginia, Charlottesville, United States
    2. Center for Public Health Genomics, University of Virginia, Charlottesville, United States
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    ya8eb@virginia.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6055-2291

  2. Lijiang Ma

    Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, United States
    Contribution
    Resources, Data curation, Software, Formal analysis, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared

  3. Johan LM Björkegren

    1. Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, United States
    2. Department of Medicine, Karolinska Institutet, Huddinge, Stockholm, Sweden
    Contribution
    Resources, Data curation, Supervision, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared

  4. Mete Civelek

    1. Department of Biomedical Engineering, University of Virginia, Charlottesville, United States
    2. Center for Public Health Genomics, University of Virginia, Charlottesville, United States
    Contribution
    Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Methodology, Project administration, Writing - review and editing
    For correspondence
    mete@virginia.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8141-0284

Funding

National Heart, Lung, and Blood Institute (2T32HL007284-46)

  • Yonathan Tamrat Aberra

National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK118287)

  • Mete Civelek

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Copyright

© 2023, Aberra et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,118
    views
  • 208
    downloads
  • 2
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Yonathan Tamrat Aberra
  2. Lijiang Ma
  3. Johan LM Björkegren
  4. Mete Civelek
(2023)
Predicting mechanisms of action at genetic loci associated with discordant effects on type 2 diabetes and abdominal fat accumulation
eLife 12:e79834.
https://doi.org/10.7554/eLife.79834

Share this article

https://doi.org/10.7554/eLife.79834

Categories and tags

Research organism

Further reading

    1. Genetics and Genomics
    2. Computational and Systems Biology

    Special Issue: Systems Genetics

    Edited by David James et al.
    Collection

    In this Special Issue we present a range of studies that showcase novel approaches that researchers are exploring to better decipher the link between genotype and phenotype.

    1. Computational and Systems Biology
    2. Physics of Living Systems

    Emergence of planar cell polarity from the interplay of local interactions and global gradients

    Divyoj Singh, Sriram Ramaswamy ... Mohd Suhail Rizvi
    Research Article Updated

    Planar cell polarity (PCP) – tissue-scale alignment of the direction of asymmetric localization of proteins at the cell-cell interface – is essential for embryonic development and physiological functions. Abnormalities in PCP can result in developmental imperfections, including neural tube closure defects and misaligned hair follicles. Decoding the mechanisms responsible for PCP establishment and maintenance remains a fundamental open question. While the roles of various molecules – broadly classified into ‘global’ and ‘local’ modules – have been well-studied, their necessity and sufficiency in explaining PCP and connecting their perturbations to experimentally observed patterns have not been examined. Here, we develop a minimal model that captures the proposed features of PCP establishment – a global tissue-level gradient and local asymmetric distribution of protein complexes. The proposed model suggests that while polarity can emerge without a gradient, the gradient not only acts as a global cue but also increases the robustness of PCP against stochastic perturbations. We also recapitulated and quantified the experimentally observed features of swirling patterns and domineering non-autonomy, using only three free model parameters - rate of protein binding to membrane, the concentration of PCP proteins, and the gradient steepness. We explain how self-stabilizing asymmetric protein localizations in the presence of tissue-level gradient can lead to robust PCP patterns and reveal minimal design principles for a polarized system.

    1. Computational and Systems Biology
    2. Neuroscience

    Basolateral amygdala oscillations enable fear learning in a biophysical model

    Anna Cattani, Don B Arnold ... Nancy Kopell
    Research Article

    The basolateral amygdala (BLA) is a key site where fear learning takes place through synaptic plasticity. Rodent research shows prominent low theta (~3–6 Hz), high theta (~6–12 Hz), and gamma (>30 Hz) rhythms in the BLA local field potential recordings. However, it is not understood what role these rhythms play in supporting the plasticity. Here, we create a biophysically detailed model of the BLA circuit to show that several classes of interneurons (PV, SOM, and VIP) in the BLA can be critically involved in producing the rhythms; these rhythms promote the formation of a dedicated fear circuit shaped through spike-timing-dependent plasticity. Each class of interneurons is necessary for the plasticity. We find that the low theta rhythm is a biomarker of successful fear conditioning. The model makes use of interneurons commonly found in the cortex and, hence, may apply to a wide variety of associative learning situations.