Effects of Genetic Loci Associated with Central Obesity on Adipocyte Lipolysis

Objectives Numerous genetic loci have been associated with measures of central fat accumulation, such as waist-to-hip ratio adjusted for body mass index (WHRadjBMI). However the mechanisms by which genetic variations influence obesity remain largely elusive. Lipolysis is a key process for regulation of lipid storage in adipocytes, thus is implicated in obesity and its metabolic complications. Here, genetic variants at 36 WHRadjBMI-associated loci were examined for their influence on abdominal subcutaneous adipocyte lipolysis. Subjects and Methods Fasting subcutaneous adipose tissue biopsies were collected from 789 volunteers (587 women and 202 men, body mass index (BMI) range 17.7–62.3 kg/m2). We quantified subcutaneous adipocyte lipolysis, both spontaneous and stimulated by the catecholamine isoprenaline or a cyclic AMP analogue. DNA was extracted from peripheral blood mononuclear cells and genotyping of SNPs associated with WHRadjBMI conducted. The effects on adipocyte lipolysis measures were assessed for SNPs individually and combined in a SNP score. Results The WHRadjBMI-associated loci CMIP, PLXND1, VEGFA and ZNRF3-KREMEN1 demonstrated nominal associations with spontaneous and/or stimulated lipolysis. Candidate genes in these loci have been reported to influence NFκB-signaling, fat cell size and Wnt signalling, all of which may influence lipolysis. Significance This report provides evidence for specific WHRadjBMI-associated loci as candidates to modulate adipocyte lipolysis. Additionally, our data suggests that genetically increased central fat accumulation is unlikely to be a major cause of altered lipolysis in abdominal adipocytes.


Subjects and Methods
Fasting subcutaneous adipose tissue biopsies were collected from 789 volunteers (587 women and 202 men, body mass index (BMI) range 17.7-62.3 kg/m 2 ). We quantified subcutaneous adipocyte lipolysis, both spontaneous and stimulated by the catecholamine isoprenaline or a cyclic AMP analogue. DNA was extracted from peripheral blood mononuclear cells and genotyping of SNPs associated with WHRadjBMI conducted. The effects on adipocyte lipolysis measures were assessed for SNPs individually and combined in a SNP score.

Results
The WHRadjBMI-associated loci CMIP, PLXND1, VEGFA and ZNRF3-KREMEN1 demonstrated nominal associations with spontaneous and/or stimulated lipolysis. Candidate genes in these loci have been reported to influence NFκB-signaling, fat cell size and Wnt signalling, all of which may influence lipolysis.

Introduction
With an almost epidemic global increase in obesity [1], related complications including type 2 diabetes and cardiovascular disease are a growing burden on healthcare systems. However, it has become clear that the distribution of fat tissue and the function of adipose tissue, rather than the amount of fat per se, exert an important impact [2,3]. In particular, central fat accumulation, reflected by high waist-to-hip ratio (WHR), is strongly associated with risk for type 2 diabetes and other complications [3].
There is a strong genetic component influencing fat distribution. WHR adjusted for body mass index (WHRadjBMI) can be considered as a measure of fat distribution independent of overall obesity [4]. Recent genome wide association studies (GWAS) have identified several genetic loci associated with WHRadjBMI [4][5][6]. These findings notwithstanding, the identity of both the culprit genes in the loci and the mechanisms by which they influence adipose function or clinical phenotypes remain unclear. Importantly, whereas central regulation of food intake is believed to be a major determinant of overall obesity and BMI [7], it seems likely that fat distribution, as assessed by WHR, is primarily regulated locally in adipose tissue. In support of this, many genes at WHRadjBMI-associated loci are expressed in adipose tissue [8]. In addition, a few genes at these loci have recently been shown to be differentially expressed between subcutaneous and visceral adipose tissue compatible with a depot specific function [8].
A major function of adipose tissue is to store and release adipocyte fatty acids through esterification to triglycerides and hydrolysis of these triglycerides (lipolysis) according to the body's energy demands. Disturbances in adipocyte lipolysis have been associated with abdominal obesity and WHR [9,10]. Increased spontaneous lipolysis provides excess amounts of fatty acids which, in turn, induce insulin resistance [11,12]. It is well established that the ability of catecholamines to stimulate lipolysis is blunted in obese individuals [13]. These subjects may retain lipids in adipocytes and thereby expand adipose tissue since the triglycerides within adipocytes constitute >90% of adipocyte volume and is the major volume component of the adipose mass.
Herein, we tested the hypothesis that genetic variation in WHRadjBMI-associated loci influence fat cell lipolysis. We utilized a unique large patient cohort in which both spontaneous and catecholamine-stimulated subcutaneous abdominal adipocyte lipolysis was measured.

Cohort description
Characteristics of the study population are presented in Table 1. Adult subjects with a large inter-individual variation in body mass index (BMI, range 17.7-62.3 kg/m 2 , 587 women and 202 men) were recruited by local advertisement. All participants were examined at 8 a.m. after an overnight fast. A trained research nurse collected information about body weight, height, waist, and hip circumferences. A venous blood sample was obtained for routine laboratory measurements and extraction of DNA. Finally, a subcutaneous adipose tissue biopsy was obtained by needle aspiration under local anesthesia from the abdominal region. All subjects were of European origin, were healthy except for obesity, and were not taking any medication. For consistency with the previous reports [4][5][6], WHRadjBMI was calculated in a sex-specific manner by inverse-normal transformation of the residuals of the linear regression model: WHR adjusted for age, age 2 and BMI [4]. The study was approved by the regional Ethics Committee in Stockholm and all subjects gave their written informed consent to participation.

Lipolysis experiments
The adipose tissue specimens were brought to the laboratory, rinsed repeatedly in saline and visual blood vessels and cell debris were removed. Spontaneous or unstimulated lipolytic activity was determined in explants of adipose tissue as described [12]. In brief, pieces of adipose tissue (200 or 300 mg) were incubated for 2 h as described below (100 mg/ml). Glycerol release into the medium was measured using a sensitive bioluminescence method and expressed as amount of glycerol release per 2 h and 10 7 adipocytes [12]. Adipocytes are the only source of glycerol which is an end product of lipolysis.
The remaining tissue (about 1 gram) was subjected to collagenase treatment as described [14] to obtain isolated adipocytes. Measurement of stimulated lipolysis in isolated adipocytes was investigated as previously described [15]. Diluted cell suspensions (2% v/v) were incubated in duplicate for 2 h at 37°C with air as the gas phase in Krebs-Ringer phosphate buffer (pH 7.4) supplemented with glucose (8.6 mmol/l), ascorbic acid (0.1 mg/ml) and bovine serum albumin (20 mg/ml), without (basal) or with increasing concentrations (10 −9 -10 −5 mol/l) of the synthetic non-selective β1, β2 and β3-adrenoreceptor agonist isoprenaline (Hässle, Mölndal, Sweden), or 10 −5 to 10 −3 mol/l of dibutyryl cyclic AMP (dcAMP, Sigma-Aldrich, Stockholm, Sweden), which is a phosphodiesterase-resistant cyclic AMP analogue that activates the cyclic AMP-dependent protein kinase A (S1 Fig). Variations in the lipolytic action of the nonmetabolizing dcAMP can only be explained by variations in lipolysis activation at, or beyond, the level of cAMP-dependent protein kinase A, whereas lipolysis variation of isoprenaline can be explained at any step in catecholamine stimulated lipolysis, from beta-adrenoreceptors to hormone sensitive lipase [13]. Stimulated lipolysis was expressed as the ratio of isoprenalineinduced or dcAMP-induced amount of glycerol release per 2 h at maximum effective concentration divided by basal glycerol release. We recently demonstrated that this measurement correlates better with clinical phenotypes than expression of absolute rates of glycerol release per adipocyte [16,17].

Statistical Methods
Correlations between raw phenotypes were assessed by Spearman rank coefficients. All variables were assessed for normality and where a skewed distribution was observed, natural log transformed before further analysis. Statistical analyses were conducted in STATA (STATAcorp LP, Texas, USA). For genetic analysis, linear regression was used to assess the effect of SNPs on the traits of interest using additive models in PLINK [19]. As a sex effect was reported for most tagging SNPs, sex-specific analyses were conducted, as well as a sex-combined analysis including a SNP x sex interaction. Unweighted SNP scores were constructed by addition of the number of reported WHRadjBMI-increasing alleles ( Table 2). Analyses of lipolysis measures were adjusted for sex, age, age 2 , and BMI as above for consistency and to enable comparison between phenotypes. Under the assumption that the heritability of lipolysis is similar to that of fat distribution [6], we have 99% power to detect an effect with nominal p value 0.05. False discovery rate (FDR) was calculated according to Benjamini-Hochberg. Because the selection of SNPs is based on known associations with related biological traits adjustment for multiple testing could be too conservative and a p value 0.05 is reported together with FDRs.

Relationships between measures of obesity and lipolysis
In order to test our hypothesis that central obesity influences lipolysis we first examined whether measures of adipocyte lipolysis correlated with obesity measures. We investigated spontaneous, isoprenaline-(activating β-adrenoceptors) and dcAMP-(activating protein kinase A) stimulated lipolysis for correlations with the obesity phenotypes WHR, BMI and WHRadjBMI. All assessed obesity phenotypes correlated positively with spontaneous lipolysis but inversely with isoprenaline or dcAMP induced lipolysis ( Table 3). As the tested variable most accurately reflecting the accumulation of metabolically detrimental fat independent of overall obesity [4], we focused on WHRadjBMI for further analyses.

Effect of WHRadjBMI-associated SNPs on lipolysis
Whether altered lipolysis causes or results from central fat accumulation remains unclear. Here, a Mendelian randomisation experiment was performed to test causality by using SNPs tagging 40 loci robustly associated with WHRadjBMI [4][5][6]. If central obesity regulates lipolysis (increased spontaneous lipolysis and/or reduced stimulated lipolysis) then it would be expected that SNPs robustly associated with increased central obesity would demonstrate consistent effects on lipolysis phenotypes. The WHRadjBMI-increasing allele previously reported [4][5][6] is taken as the true WHRadjBMI-increasing allele, and is referred to as such in the further analyses. The SNP scores were calculated by summing the number of WHRadjBMI-increasing alleles, and thus represent the genetic burden or predisposition to increased central obesity. Only subjects successfully genotyped for all SNPs included in the score were considered in the analysis. The SNP scores including all 40 tagging SNPs demonstrated no associations with all tested measures of lipolysis (Table 4). To exclude pleiotropic effects, a SNP score was calculated excluding SNPs (rs17451107, rs1195252, rs2925979, rs303084, rs4765219, rs4846567, rs9687846 and rs998584) with known effects on related traits, namely levels of HDL, LDL, TG or adiponectin and birthweight, which also revealed no associations with lipolysis measures (Table 4). It should be noted that for a larger cohort of 331-556 subjects (depending on phenotype) in which we assessed scores for the 14 loci known [4][5] prior to the recent Shungin et al report [6], we found a similar lack of association with the adipose lipolysis measures (S2 Table). It is possible that some SNPs influence mechanisms leading to central fat accumulation but that their effects might be diluted when all SNPs are combined into the SNP score. Therefore  (Table 5). Nominally significant associations with spontaneous lipolysis were detected for two loci in which the WHRadjBMI-increasing allele was associated with an increased rate of lipolysis (PLXND1 rs10804591 and CMIP rs2925979). Moreover, two loci demonstrated effects on isoprenaline-stimulated lipolysis, with the WHRadjBMI-increasing allele being associated with increased lipolysis (VEGFA rs6905288 and ZNRF3-KREMEN1 rs4823006). Finally, the WHRadjBMI-increasing allele of rs4823006 (ZNRF3-KREMEN1) was associated with increased dcAMP-stimulated lipolysis, whilst the WHRadjBMI-increasing allele of rs2925979 (CMIP) was associated with decreased dcAMP-stimulated lipolysis. None of these genetic associations were significant with FDR <5% after adjustment for the analysis of multiple SNPs ( Table 5).

Influence of individual WHR-adjBMI-associated genetic variants on central obesity
To determine whether the SNP effects on WHRadjBMI in our study concurred with those reported previously, linear regression analyses was conducted. Of 40 SNPs analysed here, 24 of the reported WHRadjBMI-increasing alleles demonstrated the expected positive effect on WHRadjBMI in this cohort, and 7 SNPs demonstrated the expected effect size (Table 6) [4][5][6].
For most complex quantitative traits studied to date, maximum effect sizes of~0.05 have been reported. Thus, the effect sizes of 32 SNPs, even if differing slightly from those previously reported, are unsurprising whilst the 8 SNPs with effect sizes larger than 0.05 should be interpreted with caution (Table 6) [4][5][6]. These inconsistencies might be due to the relative small sample size of subjects available here (n = approximately 750 compared to the reported n = 210,000 [6]), the enrichment for obese subjects in the here analysed study population, or the relative numbers of men and women (n = 202 and n = 587 respectively). The male-specific analysis shows effect sizes rather larger than those previously observed [6], but the relatively small sample size (n = 202) means that this is likely a spurious result.

Discussion
Whilst a plethora of genetic variants associated with measures of obesity have been identified, the modulated molecular mechanisms underpinning poor metabolic regulation remain elusive.  Where: EA, effect (WHRadjBMI-increasing) allele; L95, lower boundry of 95% confidence interval; U95, upper boundry of 95% confidence interval; analyses adjusted for sex, age, age 2 and BMI; FDR, false discovery rate doi:10.1371/journal.pone.0153990.t005 WHR, when adjusted for BMI (WHRadjBMI [4]), has been proposed as representative of central fat accumulation independent of total obesity. Here we used WHRadjBMI-associated loci to investigate whether genetically determined central fat accumulation controls lipolysis   The nominal associations between specific tagging SNPs identified by WHRadjBMI GWAS and lipolysis measures reported here, although becoming nonsignificant after adjustment for multiple testing. highlights potential novel biology as none of the candidate genes suggested by Shungin et al (6) for these loci are established regulators of lipolysis [20].
Firstly, effects on NFκB signalling. CMIP has been reported to influence NFκB signalling [21] which in turn can modulate adipocyte lipolysis [22]. Recently our group has demonstrated that CMIP expression levels were negatively associated with insulin-stimulated lipogenesis in subcutaneous adipocytes [23]. The observation here that the rs2925979 CMIP locus is also associated with spontaneous and dcAMP-stimulated lipolysis extends mechanistic understanding of this locus.
Secondly, effects on adipocyte morphology. PLXND1 has been shown to regulate adipose morphology, i.e. the relationship between adipocyte size and number (24), with a genetic deficiency demonstrating reduced fat mass [24]. In light of the fact that spontaneous lipolysis is strongly correlated with adipocyte size, as previously discussed [10], the WHRadjBMI-increasing allele of rs10804591 in the PLXND1 locus demonstrating a positive association with increased spontaneous lipolysis is intriguing. Furthermore, the WHRadjBMI-increasing allele of rs6905288 in the VEGFA locus revealed a positive correlation with isoprenaline-stimulated lipolysis. Interestingly, expression of VEGFA in adipose tissue has been shown to have beneficial metabolic effects via proangiogenic activity during adipose tissue expansion [25].
As for KREMEN1 and ZNRF3, associated here with both dcAMP-and isoprenaline-stimulated lipolysis, both are involved in Wnt signalling, which inhibits adipogenesis [26][27][28]. Considering that that WHRadjBMI is inversely associated with stimulated lipolysis, it is surprising that the WHRadjBMI increasing allele at the VEGFA and KREMEN1-ZNRF3 loci are associated with increased stimulated lipolysis. This raises the question whether these are spurious associations.
Whilst the cohort studied here is unique regarding the relatively large number of subjects with specific adipocyte lipolysis measures, there are some limitations to this study. Incomplete genotyping, finally including 40 of 49 reported WHRadjBMI-associated loci, means that the SNP score does not fully represent genetically-determined fat distribution. Also, for some SNPs we cannot confirm reported GWAS associations with WHRadjBMI [6], adding some uncertainty to our conclusion that WHRadjBMI does not regulate adipose lipolysis. The nominal genetic associations reported here were became nonsignificant after adjustment for multiple testing and would therefore benefit form replication in an independent cohort. Unfortunately there is as far as we know no replication cohort available. Despite the shortfalls of this study, we believe that the findings are relevant since we used established genetic associations to select the SNPs and based the study on a hypothesis that was experimentally validated, i.e. the association reported here between measures of obesity and lipolysis.
In summary, this study shows nominal significant associations between specific central obesity loci and measures of lipolysis functions in abdominal subcutaneous adipose tissue but refutes genetically increased central fat accumulation as a main cause of altered lipolysis.
Whilst generally weak associations were observed, our findings suggest that the mechanisms by which genetic variants influence central fat accumulation are diverse and locus-specific.
Supporting Information S1