Endocrine DOI 10.1007/s12020-013-0088-7

ORIGINAL ARTICLE

Common genetic variation in sFRP5 is associated with fat distribution in men J. K. Van Camp • S. Beckers • D. Zegers • A. Verrijken • L. F. Van Gaal • W. Van Hul

Received: 23 July 2013 / Accepted: 14 October 2013  Springer Science+Business Media New York 2013

Abstract Considering the role of sFRP5 in Wnt signalling, an important group of pathways regulating adipogenesis and inflammation, we performed a genetic association study on sFRP5 polymorphisms in a population of obese and lean individuals. Using information from the HapMap, two tagSNPs were identified in the sFRP5 gene region and genotyped on a population of 1,014 obese, non-diabetic individuals and 606 lean controls. We performed logistic and linear regression analysis including a wide variety of obesity parameters (BMI, waist circumference, height, WHR, fat mass, fat mass percentage and visceral, subcutaneous and total abdominal fat), in addition to OGTT and HOMA-IR values. We were able to show a significant association of sFRP5 with both total abdominal and subcutaneous fat. The association signal was only seen in obese males, and in this population, the minor allele of rs7072751 explains 1.8 % of variance in total abdominal fat. In addition, we saw a trend towards an association of rs10748709 with glucose metabolism. Although further research is necessary, we can conclude that sFRP5 is a significant regulator of fat development and distribution in obese males. We postulate that altered transcription factor binding on the rs7072751 surrounding sequence might play a role in the associations we found with both total abdominal and subcutaneous fat. In addition, although no conclusive evidence was found, our results indicate that sFRP5 genetic J. K. Van Camp  S. Beckers  D. Zegers  W. Van Hul (&) Department of Medical Genetics, University of Antwerp, Prins Boudewijnlaan 43, Edegem, 2650 Antwerp, Belgium e-mail: [email protected] A. Verrijken  L. F. Van Gaal Department of Endocrinology, Diabetology and Metabolism, Antwerp University Hospital, Wilrijkstraat 10, 2650 Antwerp, Belgium

variation may affect glucose metabolism and it would be interesting to investigate this further. Keywords Wnt  sFRP  Obesity  Glucose metabolism  Genetics  Association study

Introduction Obesity is a chronic disease with a continuously increasing prevalence worldwide. In 2008, more than one in ten of the world’s adult population was obese and each year at least 2.8 million adults die as a result of being obese or overweight (http://www.who.int/mediacentre/factsheets/fs311/ en/index.html, February 2013). Obesity is defined as an excessive accumulation of body fat and can lead to the development of cardiovascular disease, cancer, and metabolic disorders such as type 2 diabetes mellitus (T2DM) [1–3]. These metabolic comorbidities have often been associated with the low-grade inflammatory state that is seen in obese fat tissue. Adipose tissue secretes a variety of cytokines, referred to as adipokines, which are often antior pro-inflammatory [4]. Sfrp5 (secreted frizzled related-protein 5) has recently been identified as an anti-inflammatory cytokine, which binds and antagonizes Wnt (wingless-type MMTV integration site) ligands such as Wnt5a [5]. Wnts are a family of secreted glycoproteins that activate or repress the canonical or noncanonical Wnt pathways [6, 7]. The canonical Wnt pathway has been most widely studied and was shown to regulate proliferation, migration and differentiation of mesenchymal stem cells (MCSs) [8]. By binding of Wnts to the frizzled (Fz) receptor and low density lipoprotein receptor-related protein (LRP) coreceptor on the cell surface, b-catenin is released from a

123

Endocrine

degradation complex consisting of proteins such as Axin, Casein kinase 1a (CK1a) and Glycogen synthase kinase 3b (GSK3b) [9, 10]. As a consequence, b-catenin accumulates and migrates to the nucleus where it binds transcription factors of the T cell factor (TCF)-lymphoid enhancer factor (LEF) family [11, 12]. The formation of a b-catenin-TCF/ LEF complex leads to inhibition of adipogenesis by blocking the expression and activity of Peroxisome proliferator-activated receptor gamma (PPARc) and CCAAT/ enhancer-binding protein alpha (C/EBPa) [8, 13–16]. Several noncanonical Wnt-pathways have also been identified, including at least 3 calcium-mediated pathways [17–20] and a Dishevelled (Dvl)-c-Jun N-terminal kinase (JNK) pathway [21]. The latter was shown to be an important facilitator of adipose tissue inflammation and disrupts insulin signalling [22, 23]. Ouchi et al. compared the gene expression profile of adipose tissue from lean mice with that from obese mice fed a high-fat, high-sucrose (HF-HS) diet, and found secreted frizzled-related protein 5 (sfrp5) to be expressed at substantially higher levels in white adipose tissue than in other tissues. After studying a sfrp5-/- mouse model, fed a normal chow or HF-HS diet, they saw that sfrp5-/- mice fed a HFHS diet showed an increase in body weight and an impaired glucose clearance and insulin sensitivity compared to WT mice. In addition, they observed a reduced sfrp5 expression in obese leptin-deficient (ob/ob) mice and Zucker diabetic fatty rats [5]. In humans, plasma sFRP5 levels were lower in obese compared to lean subjects and in type 2 diabetes (T2DM) patients compared to normal glucose tolerant subjects [24, 25]. Conflicting results were however found by Mori et al., who reported resistance to diet-induced obesity and a mild improvement in glucose tolerance in sfrp5-/(sfrp5Q27Stop) mice. These conflicting results might have been caused by a difference in specific diets of the mice used in the different studies, or the age of the mice at the time measurements were performed [26]. The majority of evidence, however, indicates that sFRP5 counteracts the state of low-grade inflammation and insulin resistance, by inhibiting Wnt5a activation of noncanonical Dvl-JNK Wnt signalling and by promoting pre-adipocyte differentiation. In this study, we performed a genetic association analysis on common polymorphisms in the sFRP5 gene region. We analysed the association with obesity in a population of Belgian obese subjects and controls and in addition, several parameters of obesity and insulin signalling were examined as quantitative traits in an obese subpopulation. Although currently 50 loci have been found to be associated with BMI or adiposity parameters such as waist-to-hip ratio, waist circumference and body fat percentage in several waves of genome-wide association studies (GWAS), the sFRP5 locus is not one of them [27].

123

However, taking into account the above-mentioned in vitro and in vivo proof, identifying sFRP5 as a regulator of adipogenesis and adipocyte inflammation, we believe it is appropriate to perform a candidate-gene-based association study including the genetic variation in sFRP5. To our knowledge, this is the first study describing common genetic variation in sFRP5 and its impact on both obesity and glucose tolerance in humans.

Materials and methods Subjects We recruited a population of one thousand and fourteen obese individuals (466 men and 548 women, BMI C 30 kg/ m2; Table 1) from patients consulting the outpatient obesity clinic at the Antwerp University Hospital (a tertiary referral facility). Inclusion criteria were obesity (BMI C 30 kg/m2) and age 21–69 years. Exclusion criteria were pregnancy, diabetes or impaired glucose tolerance.

Table 1 Description of the study population Parameter

Obese cases

Controls

N

1,014

606

Male/female

466/548

223/383

Age (years) BMI (kg/m2)

42 ± 12 38.2 ± 6.2

35 ± 7 22.0 ± 1.6

Weight (kg)

111.5 ± 21.8

65.6 ± 9.1

Height (m)

1.7 ± 0.1

1.7 ± 0.1

Waist circumference (cm)

116.2 ± 14.6

–

WHR

0.99 ± 0.15

–

Fat free mass (kg)

57.3 ± 11.7

–

Fat mass (kg)

51.4 ± 13.2

–

Fat mass percentage

47.1 ± 7.5

–

Total fat (cm2)

751.5 ± 156.5

–

Visceral fat (cm2)

180.7 ± 88.4

–

Subcutaneous fat (cm2)

571.6 ± 140.9

–

Glucose 0 min (mg/dl)

88.0 ± 18.3

–

Glucose 120 min (mg/dl)

124.0 ± 46.0

–

Insulin 0 min (mg/dl)

20.7 ± 15.8

–

Insulin 120 min (mg/dl)

104.4 ± 90.2

–

HOMA-IR

4.6 ± 3.7

–

Mean values of general and obesity-related parameters ± standard deviation are given. Parameters WHR, fat free mass, fat mass and total, visceral and subcutaneous were only available for 899 (369 males and 530 females) out of 1,014 obese patients, and not for control samples. HOMA-IR, plasma glucose and insulin values were only available for 834 obese cases (308 males and 526 females). BMI body mass index, WHR waist-to-hip ratio; Plasma glucose 0.120 min and insulin 0.120 min describe values of the oral glucose tolerance test

Endocrine

Six hundred and six lean control individuals (223 men and 383 women, BMI 18.5–25 kg/m2; Table 1) were recruited among the university and hospital personnel and among couples seeking prenatal counselling at the Department of Medical Genetics (due to high maternal age or increased triple test). Couples seeking prenatal genetic counselling because of familial disease history were excluded. All subjects were Caucasian and at enrolment none were involved in an ongoing weight management programme. Newly diagnosed or treated diabetics and individuals with impaired glucose tolerance were excluded. All subjects had given their written informed consent before participation and the study protocol was approved by the ethics committee of the Antwerp University Hospital.

Anthropometry Height was measured to the nearest 0.5 cm; body weight was measured with a digital scale to the nearest 0.2 kg. Body mass index (BMI) was calculated as weight (in kg) over height (in m) squared. Waist circumference was measured at mid-level between the lower rib margin and the iliac crest, hip circumference at the level of the trochanter major and the waist-to-hip ratio (WHR) was calculated. Visceral (VFA), subcutaneous (SFA) and total abdominal (TFA, visceral ? subcutaneous) fat areas were determined with a computerized tomography (CT) scan that was performed at the L4–L5 level, according to the technique described by Van der Kooy and Seidell [28]. Body composition was determined by bio-impedance analysis as described by Lukaski et al. [29] and fat mass percentage was calculated, using the formula of Deurenberg et al. [30]. Average values of all obesity parameters are summarized in Table 1 for both the obese and the control population.

Laboratory analyses Venous blood samples were collected in the fasting state and during the course of an oral glucose tolerance test (OGTT) at 0 and 120 min after the ingestion of 75 grams of glucose. Plasma glucose was measured with the glucose oxidase method (on Vitros 750 XRC; Ortho Clinical Diagnostics Inc, Rochester, NY). Insulin was measured by a radioimmunoassay with the use of Pharmacia Insulin RIA (Pharmacia Diagnostics, Uppsala, Sweden). Insulin resistance was estimated using homeostasis model assessment (HOMA-IR) as described by Matthews et al. [31] and was calculated as (insulin (mU/L) 9 glucose (mmol/L))/22.5, with 1 as a reference value for normal insulin sensitivity. Table 1 summarizes OGTT and HOMA-IR values.

Genotyping Genomic DNA was extracted from whole blood by a method adapted from Miller et al. [32]. Using information from the international HapMap project [33, 34], two tagSNPs were identified in the sFRP5 gene region, including 10 kb upstream and 5,8 kb downstream of sFRP5 (HapMap phase II ? III, Feb09, NCBI Build 36 assembly, Chr10:99531760…99510642). Data from the CEPH population were selected, containing SNPs identified in Utah residents with ancestry from Northern and Western Europe (CEU). Only SNPs with a minor allele frequency (MAF) C 0.05 were included in the Tagger analysis, using aggressive tagging of 2- and 3-marker haplotypes and with r2 and LOD thresholds at 0.8 and 3.0, respectively (Haploview version 4.2) [35]. Two tagSNPs (rs7072751 and rs10748709) and one multimarker prediction test were needed to capture the information on all eight identified SNPs in the 21.1 kb sFRP5 genetic region in the CEU population. The GA haplotype of rs7072751 and rs10748709 captures the minor allele of rs1556971. Rs7072751 and rs10748709 are localized upstream of the sFRP5 gene, while rs1556971 is located in intron 2. TaqMan Pre-Designed Genotyping Assays (Applied Biosystems Inc., Foster City, CA, USA) were used to genotype the two selected SNPs, according to the manufacturer’s protocol, on a Lightcycler 480 Real-Time PCR System (Roche, Penzberg, Germany). Genotypes of rs1556971 were imputed using PLINK software [36]. Blank samples were included as negative controls and a minimum of 28 samples were placed in duplicate on each run, to confirm correct genotyping. The minimal genotyping success rate was 95.8 %. Statistical analysis Hardy–Weinberg equilibrium (HWE) was calculated for all the selected SNPs by use of the LINKUTIL package [37] with the significance level set at 0.01. Allele frequencies in lean and obese individuals were compared by means of univariate logistic regression under an additive model to calculate odds ratios (OR), corrected for age and sex. Linear regression was used to evaluate the association with various obesity parameters, i.e. BMI, waist circumference, height, WHR, fat mass, fat mass percentage and visceral, subcutaneous and total abdominal fat, corrected for age and sex. In addition, we analysed parameters indicating the level of insulin resistance, i.e. OGTT and HOMA-IR values, in obese individuals by linear regression analysis. Conditional analyses were performed using a likelihood ratio test to see whether associated SNPs have independent effects. All analyses were performed under an additive mode

123

Endocrine

of inheritance and the significance level was set at p = 0.05. Probability of type I errors due to multiple testing was contained by applying a Bonferroni correction for number of SNPs genotyped. All statistical analyses were performed for the total, male and female population separately using IBM SPSS version 20.0 (SPSS, Chicago, IL, USA). Power calculations were performed using Quanto and in our case–control population (males ? females), we have 80 % power to detect a risk of 1.55–1.24 with a SNP with MAF of 5–50 %. In the male and female case–control population separately, we have, for both populations, 80 % power to detect a risk of 1.55–1.30 with a SNP with MAF of 10–50 %. In the obese part of our population (males ? females), used for linear regression analysis, we have [80 % power to detect an effect of 1–1.5 % for the selected parameters. In the obese, male population, the power is [80 % to detect effects of 2.6 % or higher. In the obese, female population, the smallest effect size that we can detect with a power of [80 % is 1.5 % [38]. In silico prediction In silico prediction of transcription factor binding sites in the sFRP5 gene region was performed by use of the TFSEARCH online program, which searches highly correlated sequence fragments against data from the ‘TRANSFAC’ databases by GBF-Braunschweig [39].

of the obese subpopulation after Bonferroni correction (data for the total population are shown in Table 3, data for the female population are not shown (all p values [ 0.05)). However, when we performed linear regression analysis on the male part of the obese subpopulation, we detected several significant p values for adiposity parameters (p \ 0.05, Table 4). Rs10748709 and rs1556971 were significantly associated with both waist circumference and fat mass percentage. Conditional analysis in the male population showed that for both parameters separately, the two associated SNPs represent the same association signal (rs10748709 allelic heterogeneity p value = 0.692 for waist circumference and p value = 0.736 for fat mass percentage; rs1556971 allelic heterogeneity p value = 0.779 for waist circumference and p value = 0.728 for fat mass percentage). In addition, all SNPs were significantly associated with total abdominal fat and similarly, these association signals were dependent on each other as shown by conditional analysis (rs7072751 allelic heterogeneity p value = 0.101; rs10748709 allelic heterogeneity p value = 0.507; rs1556971 allelic heterogeneity p value = 0.593). rs7072751 was also significantly associated with the amount of subcutaneous fat. Finally, rs10748709 was associated with glucose values after 120 min, as measured in the OGTT test. After correcting for multiple testing using the Bonferroni method, only the associations of rs7072751 with total abdominal fat and subcutaneous fat remain (Table 4). Patients who are homozygous for the A allele of rs7072751 show a reduction of 108.24 cm2 in total abdominal fat, of which a reduction of

Results Statistical analysis of two common polymorphisms (rs7072751 and rs10748709) and one multimarker prediction test, covering most of the genetic variation in sFRP5 and its 50 and 30 flanking regions, was performed on 1,014 obese patients and 606 control subjects. In our Caucasian population, minor allele frequencies for rs7072751, rs10748709 and the imputed SNP rs1556971 were 11.2, 18.8 and 21.5 %, respectively, which is similar to the frequencies found in the HapMap CEU population. HWE was present for all SNPs analysed (p [ 0.01; data not shown). Genotype frequencies in the obese and lean population are represented in Table 2. Odds ratios were calculated by logistic regression in a case– control setting but showed no significant results in the total population (Table 2), nor in the male or female part of the population. Furthermore, we performed linear regression analysis on the obese subpopulation. For none of the investigated obesity parameters (BMI, weight, waist circumference, height, WHR, fat mass, fat mass percentage and total, visceral and subcutaneous fat) nor the glucose and insulin levels during OGTT and HOMA-IR parameters, corrected for age and sex, a significant association with sFRP5 polymorphisms could be found within the total or the female part

123

Table 2 Genotype frequencies and logistic regression results of TagSNPs in sFRP5 for the total population Cases (%)

Controls (%)

MAF (%)

OR

95 % CI

p value

11.2

0.92

0.73–1.16

0.47

18.8

0.94

0.77–1.13

0.49

21.5

0.89

0.74–1.07

0.22

rs7072751 GG

802 (79.1)

466 (76.9)

GA AA

200 (19.7) 12 (1.2)

132 (21.8) 8 (1.3)

rs10748709 AA

671 (66.2)

386 (63.7)

AG

303 (29.9)

199(32.8)

GG

40 (3.9)

21 (3.5)

rs1556971 TT

635 (62.6)

354 (58.4)

TA

337 (33.3)

229 (37.8)

AA

42 (4.1)

23 (3.8)

Genotype frequencies are shown as absolute numbers (% in parentheses). MAF is shown as a percentage for the complete population (cases and controls). Odds ratios, 95 % CI and nominal p values were calculated by logistic regression with age and sex as covariates (SPSS 20.0) and are shown for the total population (males ? females). MAF minor allele frequency, CI confidence interval, OR odds ratio

Endocrine Table 3 Results of the linear regression analysis for the total obese subpopulation (males ? females) rs7072751 p value

rs10748709

rs1556971

BMI (kg/m2)

0.545

0.312

0.221

Weight (kg)

0.129

0.211

0.118

Height (m) Waist circumference (cm)

0.054 0.539

0.383 0.521

0.248 0.450

WHR

0.972

0.958

0.881

Fat free mass (kg)

0.092

0.421

0.317

Fat mass (kg)

0.183

0.152

0.075

Fat mass percentage

0.440

0.136

0.085

Total fat (cm2)

0.327

0.520

0.406

Visceral fat (cm2)

0.604

0.551

0.484

Subcutaneous fat (cm2)

0.355

0.597

0.475

Glucose 0 min (mg/dl)

0.504

0.421

0.648

Glucose 120 min (mg/dl)

0.775

0.040

0.177

Insulin 0 min (mg/dl)

0.092

0.542

0.486

Insulin 120 min (mg/dl)

0.259

0.679

0.470

HOMA-IR

0.161

0.344

0.440

In silico prediction indicates that the rs7072751 minor allele (A) and surrounding sequence resembles a binding site for the RUNX1 (AML1a) transcription factor, whereas the sequence containing the rs7072751 wild-type allele (G) does not [39].

Discussion

Linear regression was performed with age and sex as a covariates (SPSS 20.0). Significant p values are shown in bold. No p values withstand Bonferroni correction for multiple testing

84.78 cm2 in subcutaneous fat. The minor allele (A allele) of rs7072751 thus explains 1.8 % of variance in total abdominal fat in males.

sFRP5 was shown to be an important modulator of both adipose tissue development and glucose signalling, through its ability to counteract Wnt signalling. Sfrp5-/- mice, fed a HF-HS diet, presented with a higher body weight and impaired glucose clearance and insulin sensitivity compared to WT mice. Acute sfrp5 administration in these mice, as well as in ob/ob mice, reversed the state of hyperglycaemia and improved insulin sensitivity [5]. It was shown that sFRP5 neutralizes noncanonical JNK activation and the resulting state of inflammation by Wnt5a in adipocytes [5, 40]. Furthermore, it was hypothesized that sFRP5 stimulates adipocyte differentiation by counteracting the canonical Wnt pathway and thereby promotes lipid storage in fat cells, which further counters inflammation [41]. The above-mentioned evidence thus identifies sFRP5 as a pro-adipogenic, anti-inflammatory adipokine with an insulin-sensitizing action. We therefore not only chose to investigate the association of sFRP5 genetic variation with obesity in a case–control population and with body

Table 4 Results of the linear regression analysis for the male part of the obese subpopulation rs7072751 p value

rs10748709 B

95 % CI

p value

B

rs1556971 95 % CI

p value

2

BMI (kg/m )

0.226

0.344

Weight (kg)

0.141

0.181

0.137

Height (m)

0.397

0.263

0.205

B

95 % CI

-1.99

-3.94 to -0.03

0.304

Waist circumference (cm)

0.051

0.045

WHR

0.432

0.091

0.165

Fat free mass (kg)

0.410

0.683

0.566

Fat mass (kg)

0.153

0.119

Fat mass percentage

0.093

0.046

-1.10

-2.19 to -0.02

0.045

-1.08

-2.13 to -0.02

Total fat (cm2)

0.008*

0.045

-32.42

-64.09 to -0.75

0.031

-33.91

-64.69 to -3.20

Visceral fat (cm2)

0.239

-54.12

-93.88 to -14.36

0.047

0.094

0.120

0.060

0.060

0.010*

Glucose 0 min (mg/dl)

0.558

0.359

Glucose 120 min (mg/dl)

0.568

0.021

-74.38 to -10.40

-4.08 to -0.05

0.227

Subcutaneous fat (cm2)

-42.39

-2.06

0.484 -11.23

-20.76 to -1.70

0.139

Insulin 0 min (mg/dl)

0.180

0.612

0.608

Insulin 120 min (mg/dl)

0.055

0.333

0.205

HOMA-IR

0.291

0.374

0.525

P value, B and 95 % CI were calculated by linear regression with age as a covariate (SPSS 20.0). B and 95 % CI are given only for parameters with p values lower than 0.05. Significant p values are shown in bold. Asterisks indicate p values that withstand Bonferroni correction for multiple testing CI Confidence interval

123

Endocrine

composition parameters in an obese population, but also the association with HOMA-IR and glucose and insulin levels during OGTT in a population of obese, non-diabetic subjects. Several large-scale GWAS have currently identified at least 50 loci associated with BMI or adiposity parameters such as waist-to-hip ratio, waist circumference and body fat. Despite the identification of these robustly associated (p \ 0.5 9 10-8) loci, and the repeatability of the results, the effect sizes are small and the variants explain only a small proportion of the total heritability for obesity [27]. It is therefore likely that many more variants remain to be found, and we believe that by performing this candidate gene association study, we can contribute to this search. In addition, among the 50 loci associated with obesity parameters, there are several loci near or in genes that play a role in the Wnt pathway (LRP1B [42], LGR4 [43], RSPO3 [44], TNKS [45] and KREMEN [44]), confirming the validity of sFRP5 as a candidate gene. When performing logistic regression analysis in a case– control setting, we were unable to find a significant association between the three investigated tagSNPs, representing most of the common genetic variation in and around sFRP5, and obesity. However, linear regression analysis in the male (but not in the total or female), obese subpopulation, resulted in several significant association signals between sFRP5 tagSNPS and adiposity parameters and one significant association with glucose levels. Of the nine significant p values found, only the association of rs7072751 with both total abdominal and subcutaneous fat remained significant after performing a Bonferroni correction for multiple testing (Table 4). Conditional analysis however indicated that the significantly associated signals of both rs10748709 and rs1556971 with total abdominal fat were dependent on the rs7072751 association, and thus represent the same association signal. rs7072751 is a SNP that is not in linkage disequilibrium (r2 [ 0.80) with any other SNP (MAF [ 0.01) in the sFRP5 gene region in the CEU population of the HapMap [33]. Nevertheless, this does not automatically mean that rs7072751 is the causal SNP for the observed association and functional studies will be necessary to determine the actual source of the association signal. We can however conclude, without knowing the actual causative SNP for the association signals, that sFRP5 is important for the determination of the volume and distribution of both total abdominal and subcutaneous fat in obese men. Moreover, our data show that male patients homozygous for the rs7072751 minor allele have an average reduction of 84.78 cm2 in subcutaneous fat which contributes to an overall reduction of 108.24 cm2 in total abdominal fat. This polymorphism thus explains 1.8 % of variance in total abdominal fat in the investigated population of obese males.

123

Because rs7072751 is located upstream of sFRP5, it is possible that this polymorphism affects transcription efficiency. The rs7072751 minor allele and surrounding sequence is predicted to resemble a binding site for the RUNX1 (AML1a) transcription factor, whereas the sequence containing the rs7072751 wild-type allele (G) does not [39]. Although more research on the effect of rs7072751 on sFRP5 transcription is clearly required, we can hypothesize that binding of RUNX1 to the sFRP5 regulatory region can suppress the expression of sFRP5. In the past, RUNX1 has already been identified as a negative regulator of GATA3 expression in helper T cells, so it is plausible that it would also function as a transcriptional repressor for other genes [46]. However, when looking at the Wnt5a-inhibiting action of sFRP5, a decrease in sFRP5 transcription would lead to an increase in the Wnt5a vs. sFRP5 ratio, which in the past has been associated with an impaired insulin and glucose signalling [40]. It is therefore unclear why we did not observe a clear genetic association between rs7072751, or any of the other sFRP5 tagSNPS, and OGTT and HOMAIR values. This could be due to the fact that we only included normoglycemic subjects in our population. However, we did observe a significant association between rs10748709 and glucose values at 2 h after initiation of the OGTT test, in both our total, obese population and male, obese subpopulation (Tables 3, 4). The association signals did not withstand the strict Bonferroni correction for multiple testing, but they do indicate a possible effect of genetic variation in sFRP5 on glucose metabolism. It was striking that we were only able to find significant association signals in the male part of our obese population. One reason could be that differences in genetic factors contributing to fat development, leading to a disparity in fat deposition patterns and fat metabolism between males and females, result in deviating association signals [47]. Sexual dimorphism has been observed before in other genetic association studies investigating fat distribution parameters [44, 48]. Furthermore, in a study by Hu et al. [24], it was seen that circulating sFRP5 levels were significantly lower in men than in women indicating sex-specific activity or regulation, which could also contribute to the difference in genetic association that was found in our study. The fact that we did not include a follow-up population for replication of our results is a clear limitation of our study. As such, we were not able to confirm the associations of rs7072751 with total abdominal and subcutaneous fat and the possible association of sFRP5 polymorphisms with glucose metabolism in an independent population. Replication of these results will thus be required in the future, and in addition it will be helpful to include data on adiposity and glucose/insulin parameters in the control sample, as these were missing in our study, and limited our

Endocrine

use of statistical tests. Only BMI could be included as a parameter in logistic regression analysis, and as it appears that sFRP5 genetic variation is associated with waist and fat distribution, rather than with overall adiposity, no significant odds ratios could be found. In conclusion, despite numerous indications of the importance of sFRP5 in insulin sensitivity and glucose signalling, we were unable to find a significant association between sFRP5 polymorphisms and HOMA-IR or glucose and insulin levels during OGTT. We did however see an indication that genetic variation in sFRP5 affects glucose metabolism, and it would be interesting to investigate this association further. In addition, we have shown that sFRP5 is an important regulator of fat development and distribution in males, by identifying a polymorphism in the sFRP5 genetic region that leads to a decrease in the amount of both total abdominal and subcutaneous fat, possibly due to a decrease in sFRP5 expression. Additional research on the rs7072751 polymorphism, or one of its proxies, and its effect on sFRP5 protein function is, however, imperative to understand the true role of sFRP5 in the fat metabolism of lean and obese humans. Acknowledgments This work was supported by a TOP-research grant from the University of Antwerp to WVH and by funding from the Belgian Science Policy Office Interuniversity Attraction Poles (BELSPO-IAP) programme through the project IAP P7/43-BeMGI. SB holds a postdoctoral fellowship obtained from the Flemish Fund for Scientific Research (F.W.O. Vlaanderen). Conflict of interest

The authors declare no conflicts of interest.

References 1. L.F. Van Gaal, I.L. Mertens, C.E. De Block, Mechanisms linking obesity with cardiovascular disease. Nature 444(7121), 875–880 (2006) 2. G.A. Colditz et al., Weight gain as a risk factor for clinical diabetes mellitus in women. Ann. Intern. Med. 122(7), 481–486 (1995) 3. A. Bergstrom et al., Overweight as an avoidable cause of cancer in Europe. Int. J. Cancer 91(3), 421–430 (2001) 4. E.E. Kershaw, J.S. Flier, Adipose tissue as an endocrine organ. J. Clin. Endocrinol. Metab. 89(6), 2548–2556 (2004) 5. N. Ouchi et al., Sfrp5 is an anti-inflammatory adipokine that modulates metabolic dysfunction in obesity. Science 329(5990), 454–457 (2010) 6. Y. Kawano, R. Kypta, Secreted antagonists of the Wnt signalling pathway. J. Cell Sci. 116(Pt 13), 2627–2634 (2003) 7. P.W. Finch et al., Purification and molecular cloning of a secreted, Frizzled-related antagonist of Wnt action. Proc. Natl. Acad. Sci. USA 94(13), 6770–6775 (1997) 8. S. Kang et al., Wnt signaling stimulates osteoblastogenesis of mesenchymal precursors by suppressing CCAAT/enhancerbinding protein alpha and peroxisome proliferator-activated receptor gamma. J. Biol. Chem. 282(19), 14515–14524 (2007) 9. P. Bhanot et al., A new member of the frizzled family from Drosophila functions as a Wingless receptor. Nature 382(6588), 225–230 (1996)

10. A. Bejsovec, Wnt signaling: an embarrassment of receptors. Curr. Biol. 10(24), R919–R922 (2000) 11. J. Papkoff et al., Wnt-1 regulates free pools of catenins and stabilizes APC-catenin complexes. Mol. Cell. Biol. 16(5), 2128–2134 (1996) 12. H. Clevers, M. van de Wetering, TCF/LEF factor earn their wings. Trends Genet. 13(12), 485–489 (1997) 13. M. Kawai et al., Wnt/Lrp/beta-catenin signaling suppresses adipogenesis by inhibiting mutual activation of PPARgamma and C/EBPalpha. Biochem. Biophys. Res. Commun. 363(2), 276–282 (2007) 14. K.A. Longo et al., Wnt10b inhibits development of white and brown adipose tissues. J. Biol. Chem. 279(34), 35503–35509 (2004) 15. C.N. Bennett et al., Regulation of Wnt signaling during adipogenesis. J. Biol. Chem. 277(34), 30998–31004 (2002) 16. J.A. Kennell, O.A. MacDougald, Wnt signaling inhibits adipogenesis through beta-catenin-dependent and -independent mechanisms. J. Biol. Chem. 280(25), 24004–24010 (2005) 17. A. Ahumada et al., Signaling of rat Frizzled-2 through phosphodiesterase and cyclic GMP. Science 298(5600), 2006–2010 (2002) 18. A.D. Kohn, R.T. Moon, Wnt and calcium signaling: beta-cateninindependent pathways. Cell Calcium 38(3–4), 439–446 (2005) 19. L. Ma, H.Y. Wang, Mitogen-activated protein kinase p38 regulates the Wnt/cyclic GMP/Ca2? non-canonical pathway. J. Biol. Chem. 282(39), 28980–28990 (2007) 20. K. Schlessinger, E.J. McManus, A. Hall, Cdc42 and noncanonical Wnt signal transduction pathways cooperate to promote cell polarity. J. Cell Biol. 178(3), 355–361 (2007) 21. R. Sugimura, L. Li, Noncanonical Wnt signaling in vertebrate development, stem cells, and diseases. Birth Defects Res. C Embryo Today 90(4), 243–256 (2010) 22. J. Hirosumi et al., A central role for JNK in obesity and insulin resistance. Nature 420(6913), 333–336 (2002) 23. G. Solinas et al., JNK1 in hematopoietically derived cells contributes to diet-induced inflammation and insulin resistance without affecting obesity. Cell Metab. 6(5), 386–397 (2007) 24. W. Hu et al., Circulating Sfrp5 is a signature of obesity-related metabolic disorders and is regulated by glucose and liraglutide in humans. J. Clin. Endocrinol. Metab. 98(1), 290–298 (2013) 25. Z. Hu, H. Deng, H. Qu, Plasma SFRP5 levels are decreased in Chinese subjects with obesity and type 2 diabetes and negatively correlated with parameters of insulin resistance. Diabetes Res. Clin. Pract. 99, 391–395 (2013) 26. H. Mori et al., Secreted frizzled-related protein 5 suppresses adipocyte mitochondrial metabolism through WNT inhibition. J. Clin. Invest. 122(7), 2405–2416 (2012) 27. F.R. Day, R.J. Loos, Developments in obesity genetics in the era of genome-wide association studies. J. Nutrigenet. Nutrigenom. 4(4), 222–238 (2011) 28. K. van der Kooy, J.C. Seidell, Techniques for the measurement of visceral fat: a practical guide. Int. J. Obes. Relat. Metab. Disord. 17(4), 187–196 (1993) 29. H.C. Lukaski et al., Assessment of fat-free mass using bioelectrical impedance measurements of the human body. Am. J. Clin. Nutr. 41(4), 810–817 (1985) 30. P. Deurenberg, J.A. Weststrate, J.G. Hautvast, Changes in fat-free mass during weight loss measured by bioelectrical impedance and by densitometry. Am. J. Clin. Nutr. 49(1), 33–36 (1989) 31. D.R. Matthews et al., Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28(7), 412–419 (1985) 32. S.A. Miller, D.D. Dykes, H.F. Polesky, A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 16(3), 1215 (1988)

123

Endocrine 33. International HapMap Consorium, The International HapMap Project. Nature 426(6968), 789–796 (2003) 34. The International HapMap Project (2011), http://hapmap.ncbi. nlm.nih.gov/thehapmap.html.en. Accessed 06 Jun 2013 35. P.I. de Bakker et al., Efficiency and power in genetic association studies. Nat. Genet. 37(11), 1217–1223 (2005) 36. S. Purcell et al., PLINK: a tool set for whole-genome association and population-based linkage analyses. Am. J. Hum. Genet. 81(3), 559–575 (2007) 37. J. Ott, Documentation to LINKAGE UTILITY programs (2011), http://linkage.rockefeller.edu/ott/linkutil.htm. Accessed 10 Feb 2011 38. W. Gauderman, J. Morrison, Gene 9 environment, gene 9 gene interaction home page (2009), http://hydra.usc.edu/gxe/. Accessed 10 Feb 2011 39. T. Heinemeyer et al., Databases on transcriptional regulation: TRANSFAC, TRRD and COMPEL. Nucleic Acids Res. 26(1), 362–367 (1998) 40. D.M. Schulte et al., Pro-inflammatory wnt5a and anti-inflammatory sFRP5 are differentially regulated by nutritional factors in obese human subjects. PLoS ONE 7(2), e32437 (2012) 41. C. Lv et al., Sfrp5 expression and secretion in adipocyte are upregulated during the differentiation and are negatively correlated with insulin resistance. Cell Biol. Int. 36, 851–855 (2012)

123

42. E.K. Speliotes et al., Association analyses of 249, 796 individuals reveal 18 new loci associated with body mass index. Nat. Genet. 42(11), 937–948 (2010) 43. G. Thorleifsson et al., Genome-wide association yields new sequence variants at seven loci that associate with measures of obesity. Nat. Genet. 41(1), 18–24 (2009) 44. I.M. Heid et al., Meta-analysis identifies 13 new loci associated with waist-hip ratio and reveals sexual dimorphism in the genetic basis of fat distribution. Nat. Genet. 42(11), 949–960 (2010) 45. A. Scherag et al., Two new Loci for body-weight regulation identified in a joint analysis of genome-wide association studies for early-onset extreme obesity in French and german study groups. PLoS Genet. 6(4), e1000916 (2010) 46. O. Komine et al., The Runx1 transcription factor inhibits the differentiation of naive CD4? T cells into the Th2 lineage by repressing GATA3 expression. J. Exp. Med. 198(1), 51–61 (2003) 47. S. Lemieux et al., Sex differences in the relation of visceral adipose tissue accumulation to total body fatness. Am. J. Clin. Nutr. 58(4), 463–467 (1993) 48. S. Cha, H. Yu, J.Y. Kim, Bone mineral density-associated polymorphisms are associated with obesity-related traits in Korean adults in a sex-dependent manner. PLoS ONE 7(12), e53013 (2012)